Second International Symposium
FIELD SCREENING METHODS FOR
    HAZARDOUS WASTES AND
       TOXIC CHEMICALS
           February 12-14, 1991
      Symposium Proceedings

-------
      SECOND INTERNATIONAL SYMPOSIUM
FIELD SCREENING METHODS FOR
     HAZARDOUS WASTES AND
         TOXIC CHEMICALS
              Februaiy 12-14,1991
            CO-SPONSORS
          U.S. Environmental Protection Agency
             U.S. Department of Energy
      U.S. Army Toxic and Hazardous Materials Agency
 U.S. Army Chemical Research, Development and Engineering Center
                U.S. Air Force
              Florida State University
   National Environmental Technology Applications Corporation
      National Institute for Occupational Safety and Health

-------
                          DISCLAIMER

Although this Proceedings Document reports the oral and poster presentations and
discussions that occurred during this Symposium funded by the United States
Environmental Protection Agency, the contents represent views independent of
Agency Policy. This Document has not been subjected to the Agency's peer review
process and does not necessarily reflect the Agency views. No official endorse-
ment should be inferred.

-------
SYMPOSIUM ORGANIZATION

Symposium Chairman - Llewellyn Williams, EPA/EMSL-Las Vegas, NV
     Vice-Chairman - Eric Koglin, EPA/EMSL-Las Vegas, NV
  Executive Secretary - John Koutsandreas, Florida State University
     ACKNOWLEDGEMENTS

     This symposium has been arranged through a contract with
 ICAIR, Life Systems, Inc. The following personnel were involved in
               coordinating this symposium:
           Program Manager - Ms. Jo Ann Duchene
           Presentation Coordinators - Mr. Ron Polhill
                               Ms. Donna Studniarz
           Exhibit Coordinator - Mr. Charles Tanner
        Registration Coordinator - Ms. Linda Hashlamoun

-------
                                         FOREWORD
The role of and need for field screening methods for the identification and quantification of contaminants in
environmental media is growing rapidly. This nation and its European neighbors are faced with the tremendous task
of remediating thousands of  hazardous  waste  sites  -- the legacy of our  much less environmentally aware
predecessors. Field screening methods that generate real-time information on the nature and extent of contamina-
tion improve the cost-effectiveness of remediation. Many of these same methods can, and in some cases are already
being used to improve our capability to measure exposure, at the point of exposure, thereby improving our ability
to assess risks to human health and the environment.
The U.S. EPA is not the only viable  user of field screening methods; that fact is reflected in the list of this
Symposium's co-sponsors. Other agencies are discovering  applications for these same technologies to address
issues such as worker safety, drug interdiction, and chemical  warfare defense. The research activities supported by
these same agencies are advancing innovative technologies that may have application in environmental monitoring
and field screening.
To present a global view of technological developments, this Symposium featured over 120 platform and poster
presentations from the United States and around the world. The papers and discussions that follow represent three
days of intense communication and cooperation among a variety of communities—regulatory, academic, industrial
and users. It is my hope that the products of this Symposium will find many uses and will provide the impetus for
new initiatives in field screening methods.
Llewellyn R. Williams
U.S. Environmental Protection Agency
Environmental Monitoring Systems Laboratory
Las Vegas, Nevada

-------
                                               CONTENTS
OPENING PLENARY SESSION
Opening Remarks — Dr. Llewellyn Williams, U.S. EPA, Environmental Monitoring Systems Laboratory, Las Vegas	1
Keynote Address — Analytical Issues in the U.S. EPA Superfund Program
        Larry Reed, U.S. EPA, Director Hazardous Site Evaluation Division, Office of Emergency and Remedial Response ....3
Overview ofDOE's Field Screening Technology Development Activities
        C.W. Frank, T.D. Anderson, C.R. Cooley, K.E. Hain, S.C.T. Lien, U.S. Department of Energy; R.L. Snipes,
        Martin Marietta Energy Systems; M.D. Erickson, Argonne National Laboratory	5
Department of Defense Field Screening Methods Requirements in the Installation Restoration Program
        Dennis J. Wynne, U.S. Army Toxic and Hazardous Materials Agency	15
An Overview of Army Sensor Technology Applicable to Field Screening of Environmental Pollutants
        Raymond A. Mackay, U.S. Army Chemical Research, Development and Engineering Center	17
Field Analytical Methods for Superfund
        Howard M. Fribush and Joan F. Fisk, U.S. EPA	25

Field Delineation of Soils Contamination on Hazardous Waste Sites Regulated Under New Jersey's Hazardous Waste Program
        Frederick W. Cornell, New Jersey Department of Environmental Protection	31
Plenary Session Discussion	40

SESSION 1:
Chemical Sensors
Chairperson: Dr. Ed Poziomek, University of Nevada Environmental Research Center

A FiberOptic Sensor for the Continuous Monitoring of Chlorinated Hydrocarbons
        P.P. Milanovich, P.P. Daley, K. Langry, B.W. Colston, S.B. Brown and S.M. Angel, Lawrence
        Livermore National Laboratory	43

Chemical Sensors for Hazardous Waste Monitoring
        M.B. Tabacco, Q. Zhou, K.  Rosenblum, Geo-Centers, Inc.; M.R. Shahriari, Rutgers University	49
Rapid, Subsurface, In Situ Field Screening of Petroleum Hydrocarbon Contamination Using Laser Induced
Fluorescence Over Optical Fibers
        S.H. Lieberman, G.A. Theriault, Naval Ocean Systems Center; S.S. Cooper, P.O. Malone and R.S. Olsen, U.S. Army
        Waterways Experiment Station, Vicksburg; P.W. Lurk, U.S. Army Toxic and Hazardous Materials Agency	57
Chemical Sensors Panel Discussion	64

Spectroelectrochemical Sensing of Chlorinated Hydrocarbons for Field Screening and In Situ Monitoring Applications
        Michael M. Carrabba, Robert B. Edmonds and R. David Rauh, EIC Laboratories, Inc.; John W. Haas, III,
        Oak Ridge National Laboratories	67

Surface Acoustic Wave (SAW) Personal Monitor for Toxic Gases
        N.L. Jarvis, H. Wohltjen and J.R. Lint, Microsensor Systems, Inc	73
Arrays of Sensors and Microsensors for Field Screening of Unknown Chemical Wastes
        W.R. Penrose, J.R. Stetter and W.J. Buttner, Transducer Research, Inc.; Z. Cao, Illinois Institute of Technology	85

SESSION 2:
Ion Mobility Spectrometry
Chairperson: Dr. Steve Harden, U.S. Army Chemical Research, Development and Engineering Center

Real-Time Detection of Aniline in Hexane By Flow Injection Ion Mobility Spectrometry
        G.E. Burroughs, National Institute for Occupational  Safety and Health; G.A. Eiceman and L. Garcia-Gonzalez,
        New  Mexico State University	95

Detection of Microorganisms by Ion Mobility Spectrometry
        A.P. Snyder, M. Miller and D.B. Shoff, U.S. Army Chemical Research, Development and Engineering Center;
        Gary A. Eiceman, New Mexico State University; D.  A. Blyth, J. A Parsons, Geo-Centers, Inc	103
Data Analysis Techniques for Ion Mobility Spectrometry
        Dennis M. Davis, U.S. Army Chemical Research, Development and Engineering Center	113

-------
Ion Mobility Spectrometry as a Field Screening Technique
        Lynn D. Hoffland and Donald B. Shoff, U.S. Army Chemical Research, Development and Engineering Center	137

Hand-Held GC-Ion Mobility Spectrometry for On-Site Analysis of Complex Organic Mixtures in Air or Vapors Over Waste Sites
        Suzanne Ehart Bell, Los Alamos National Laboratory; G.A. Eiceman, New Mexico State University	153
Remote and In Situ Sensing of Hazardous Materials by Infrared Laser Absorption, Ion Mobility Spectrometry and Fluorescence
        Peter Richter, Technical University of Budapest	167

SESSION 3:
Robotics
Chairperson: Dr. Carolyn Esposito, U.S. EPA Risk Reduction Engineering Laboratory

The Department of Energy's Robotics Technology Development Program for Environmental Restoration and
Waste Management
        A.C. Heywood, Science Applications International Corporation; S.A. Meacham, Oak Ridge National Laboratory;
        P.J. Eicker, Sandia National Laboratories	173
Field Robots for Waste Characterization and Remediation
        William L. Whittaker, David M. Pahnos; Field Robotics Center, Carnegie Mellon Institute	181

Space Technology for Application to Terrestrial Hazardous Materials Analysis and Acquisition
        Brian Muirhead, Susan Eberlein, James  Bradley and William Kaiser, NASA/Jet Propulsion Laboratory	187

Development of a Remote Tank Inspection (RTI) Robotic System
        Chris Fromme, Barbara P. Knape, Bruce Thompson, RedZone Robotics, Inc	197

Automated Subsurface Mapping
        Jim Osbom, Field Robotics Center, Carnegie Mellon Institute	205

SESSION 4:
QA and Study Design
Chairperson: Dr. Janine Jessup Arvizu

A Quality Assurance Sampling Plan for Emergency Response (QASPER)
        John M. Mateo, Christine M. Andreas, Roy F. Weston; William Coakley, U.S. EPA	217

A Rationale for the Assessment of Errors in Soil Sampling
        Jeffrey van Ee, U.S. EPA; Clare L. Gerlach, Lockheed Engineering & Sciences Company	227

A Review of Existing Soil Quality Assurance Materials
        K. Zarrabi, A.J. Cross-Smiecinski and T. Starks, University of Nevada	235

SESSION 5:
Air Pathway Monitoring at Superfund Sites
Chairperson: Dr. William McClenny

Evaluation of Emission Sources and Hazardous Waste Sites Using Portable Chromatographs
        R.E. Berkley, U.S. EPA	253

High Speed Gas Chromatographyfor Air Monitoring
        S.P, Levine, H.Q. Ke and R.F. Mouradian, University of Michigan; R. Berkley, U.S. EPA; J. Marshall, HNU Systems.... 265

Screening Volatile Organics By Direct Sampling Ion Trap and Glow Discharge Mass Spectrometry
        Marcus B. Wise, G.B. Hurst, C.V. Thompson, Michelle V. Buchanan and Michael R. Guerin, Oak Ridge National
        Laboratory	273
Development and Testing of a Man-Portable Gas Chromatography/Mass Spectrometry System for Air Monitoring
        Henk L.C. Meuzelaar, Dale T. Urban and Neil S. Arnold, University of Utah	289

On-Site Multimedia Analyzers: Advanced Sample Processing with On-Line Analysis
        S. Liebman, Geo-Centers, Inc.; M.B. Wasserman, U.S. Army Chemical Research, Development and Engineering
        Center, E.J. Levy and S. Lurcott, Computer Chemical Systems, Inc	299

Using a FID-Based Organic Vapor Analyzer in Conjunction with GC/MS Summa Canister Analyses to Assess the Impact of
Landfill Gassesfrom a Superfund Site on the Indoor Air Quality of an Adjacent  Commercial Property
        T.H. Pritchett, U.S. EPA; D. Mickunas and S. Schuetz, IT Corporation	307

-------
SESSION 6:
Field Mobile GC/MS Techniques
Chairperson: Dr. Stephen Billets, U.S. EPA Environmental Monitoring Systems Laboratory, Las Vegas

Field Analytical Support Project (FASP) Use to Provide Data for Characterization of Hazardous Waste Sites for Nomination
to the National Priorities List (NPL): Analysis ofPolycyclic Aromatic Hydrocarbons (PAHs) and Pentachlorophenol (PCP)
        Lila AccraTransue, Andrew Hafferty and Tracy Yerian, Ecology and Environment	309

Thermal Desorption Gas Chromatograph-Mass Spectrometry Field Methods for the Detection of Organic Compounds
        A. Robbat, Jr. T-Y Liu, B. Abraham and C-J- Liu, Tufts University	319
Rapid Determination of Semivolatile Pollutants by Thermal Extraction/Gas Chromatography/Mass Spectrometry
        T. Junk, V. Shirley, C.B. Henry, T.R. Irvin and E.B. Overton, Louisiana State University; J.E. Zumberge,
        C. Sutton and R.D. Worden, Ruska Laboratories, Inc	327

The Application of a Mobile Ion Trap Mass Spectrometer System to Environmental Screening and Monitoring
        William H. McClennen, Neil, S. Arnold, Henk L.C. Meuzelaar, JoAnn A. Lighty, University of Utah;
        Erich Ludwig, GSF Munchen, Institut fur Okologische Chemie	339

Field Measurement of Volatile Organic Compounds by Ion Trap Mass Spectrometry
        M.E. Cisper, J.E. Alarid, P.H. Hemberger, E.P. Vanderveer, Los Alamos National Laboratory	351
Transportable GC/lon Trap Mass Spectrometry for Trace Field Analysis of Organic Compounds
        Chris P. Leibman and David Dogruel, Eric P. Vanderveer, Los Alamos National Laboratory	367

SESSION 7
Portable Gas Chromatography
Chairperson: Dr. Thomas Spittler, U.S. EPA New England Regional Laboratory

The Use of Field Gas Chromatography to Protect Gmundwater Supplies
        Thomas M. Spittler, U.S. EPA	377
Field Screening Procedures for Determining the Presence of Volatile Organic Compounds in Soil
        Alan B. Crockett and Mark S. DeHaan, EG&G Idaho, Inc	383

Comparison of Field Headspace Vs. Field Soil Gas Analysis Vs. Standard Method Analysis of Volatile Petroleum
Hydrocarbons in Water and Soil
        Randy D. Golding, Marty Favero, Glen Thompson, Tracer Research Corporation	395

Field Screening ofBTEX in Gasoline-Contaminated Groundwater and Soil Samples by a Manual, Static Headspace GC Method
        James D. Stuart, Suya Wang and Gary A. Robbins, University of Connecticut; Clayton Wood, HNU Systems, Inc. ...407
Comparison of Aqueous Headspace Air Standard Vs. SUMMA Canister Air Standard for Volatile Organic
Compound Field Screening
        H. Wang, Roy  F. Weston, Inc.; W.S. Clifford, U.S. EPA	415

Quantitative Soil Gas Sampler Implant for Monitoring Dump Site Subsurface Hazardous Fluids
        Kenneth T. Lang, Douglas T. Scarborough, U.S. Army Toxic and Hazardous Materials Agency; Mark Glover,
        D.P. Lucero, IIT Research Institute	423

SESSION 8
Field Screening Methods for Worker Safety
Chairperson: Dr. Judd Posner, National  Institute for Occupational Safety and Health

Tunable COi Laser-Based Photo-Optical Systems for Surveillance of Indoor Workplace Pollutants
        Harley V. Piltingsrud, National Institute for Occupational Safety and Health	433

Immuno-Based Personal Exposure Monitors
        Arbor Drinkwine, Stan Spurlin, Midwest Research Institute; Jeanette Van Emon, U.S. EPA;
        Viorica Lopez-Avila, Mid-Pacific Environmental  Laboratory, Inc	449

A Remote Sensing Infrared Air Monitoring System for Gases and Vapors
        S.P. Levine, H.K. Xiao, University of Michigan; W. Herget, Nicolet Analytical; R. Spear, University of California;
        T. Pritchett, U.S. EPA	461

Adriamycin Exposure Study Among Hospital Personnel
        R.L. Stephenson, Thomson Consumer Electronics Inc.; C.H. Rice, J. Dimos, University of Cincinnati	465

-------
Real-Time Personal Monitoring in the Workplace Using Radio Telemetry
        Ronald J. Kovein and Paul Hentz, National Institute for Occupational Safety and Health	473

Improvements in the Monitoring of PPM Level Organic Vapors with Field Portable Instruments
        Gerald Moore, GMD Systems, Inc	483

SESSION 9
X-Ray Fluorescence
Chairperson: Dr. John Barich, U.S. EPA Region X

Rapid Assessment of Superfund Sites for Hazardous Materials with X-Ray Fluorescence Spectrometry
        W.H. Cole  III, R.E. Enwall, G.A. Raab, C.A. Kuharic, Lockheed Engineering and Sciences Co.;
        W.H. Engelmann, L.A. Eccles, U.S. EPA	497

A High Resolution Portable XRF Hgh Spectrometer for Field Screening of Hazardous Wastes
        J.B. Ashe, Ashe Analytics; P.P. Berry and G.R. Voots, TN Technologies, Inc.; M. Bemick, Roy F. Weston, Inc.;
        G.  Prince, U.S. EPA	507
Low Concentration Soil Contaminant Characterization Using EDXRF Analysis
        A.R. Harding, Spectrace Instruments, Inc	517

Data Quality Assurance/Quality Control for Field X-Ray Fluorescence Spectrometry
        Clark D. Carlson, John R. Alexander, The Bionetics Corporation	525

A Study of the Calibration of a Portable Energy Dispersive X-Ray Fluorescence Spectrometer
        C.A. Ramsey, D.J. Smith and E.L. Bour, U.S. EPA	535

SESSION 10
Fourier Transform Infrared Spectrometry and Other Spectroscopy Methods
Chairperson: Dr. Donald Gurka, U.S. EPA Environmental Monitoring Systems Laboratory, Las Vegas

Use of Long-Path FTIR Spectrometry in Conjunction with Scintillometry to Measure Gas Fluxes
        Douglas  I. Moore, Clifford N. Dahrn, James R. Gosz, University of New Mexico; Reginald J. Hill, NOAA	541

Pattern Recognition Methods for FTIR Remote Sensing
        Gary W.  Small, University of Iowa; Robert T. Kroutil, U.S. Army Chemical Research,
        Development and Engineering Center	549

Remote Vapor Sensing Using a Mobile FTIR Sensor
        R.T. Kroutil, J.T. Ditillo, R.L. Gross, R.J. Combs, W.R. Loerop, U.S. Army Chemical Research,
        Development and Engineering Center; G.W. Small, University of Iowa	559

Use of Wind Data to Compare Point-Sample Ambient Air VOC Concentrations with Those Obtained by Open-Path FT-IR
        Ray E. Carter, Jr. and Dennis D. Lane, Glen A. Marotz, University of Kansas;
        Mark J. Thomas, Jody L. Hudson, U.S. EPA	571

Remote Detection ofOrganics Using Fourier Transform Infrared Spectroscopy
        Jack C. Demirgian and Sandra M. Spurgash, Argonne National Laboratory	583

Intrepretation ofPPM-Meter Data from Long-Path Optical Monitoring Systems as They Would be Used at
Superfund Hazardous Waste Sites
        Thomas H. Pritchett, U.S. EPA; Timothy R. Minnich, Robert L. Scotto and Margaret R. Leo,
        Blasland, Bouck &  Lee	591

CLOSING PLENARY SESSION
        Awards Ceremony	593
        Closing Remarks	595

POSTERS

Calibration of Fiber Optic Chemical Sensors
        W.F. Arendale and Richard Hatcher, University of Alabama; Bruce Nielsen, Hq. AFESC/RDVW	597

Gas-Chromatographic Analysis of Soil-Gas Samples at a Gasoline Spill
        R.J. Baker, J.M. Ficher, N.P. Smith, S.A. Koehnlein, A.L. Baehr, U.S. Geological Survey	599

-------
Significant Physical Effects on Surface Acoustic Wave (SAW) Sensors
        David L. Bartley, National Institute for Occupational Safety and Health	601
An Evaluation of Field Portable XRF Soil Preparation Methods
        Mark Bemick, Donna Idler, Lawrence Kaelin, Dave Miller, Jayanti Patel, Roy F. Weston; George Prince and
        Mark Sprenger, U.S. EPA	603

Development of a Field Screening Technique for Dimethyl Mercury in Air
        Brian E. Brass, Lawrence P. Kaelin, Roy F. Weston; Thomas H. Pritchett, U.S. EPA	609
Applicability of Thin-Layer Chromatography to Field Screening of Nitrogen-Containing Aromatic Compounds
        William C. Brumley, Cynthia M. Brownrigg, U.S. EPA	615
Assessing the Air Emissions from a Contaminated Aquifer at a Superfund Site
        S. Burchette and T.H. Prichett, U.S. EPA; S. Schuetz, IT Corporation; K. Harvey, Roy F. Weston, Inc	619
Calculation and Use of Retention Indices for Identification of Volatile Organic Compounds with a Microchip
Gas Chromatograph
        K.R. Carney, E.B. Overton and R.L. Wong, Louisiana State University	621
Determination ofPCBs by Enzyme Immunoassy
        Mary Anne Chamerlik-Cooper, Robert E. Carlson, ECOCHEM Research, Inc; Robert O. Harrison,
        ImmunoSystems, Inc	625
Practical Limits in Field Determination of Fluorescence Using Fiber Optic Sensors
        Wayne Chudyk, Kenneth Pohlig, Carol Botteron and Rose Najjar, Tufts University	629
The Colloidal Borescope—A Means of Assessing Local Colloidal Flux and Groundwater Velocity in Porous Media
        T.A. Cronk, P.M.  Kearl, Oak Ridge National Laboratory	631

Fieldable Enzyme Immunoassay Kits for Drugs and Environmental Chemicals
        Peter H. Duquette, Patrick E. Guire, Melvin J. Swanson, Martha,  J. Hamilton, Stephen J. Chudzik and
        Ralph A. Chappa, Bio-Metric Systems, Inc	633
Xuma Expert System for Support of Investigation and Evaluation of Contaminated Sites
        W. Eitel, R. Hahn Landesanstalt f. Umweltschutz B.; W.W. Geiger and R. Weidemann,
        Institut f. Datenverarbeitung	645

A Rapid Response SAW-GC Chemical Monitor for Low-Level Vapor Detection
        John A. Elton, James F. Houle, Eastman Kodak Company	649

Passive Cryogenic Whole Air Field Sampler
        Steven J. Fernandez, Bill G. Motes, Joseph P. Dugan Jr., Susan K. Bird, Gary J. McManus, Westinghouse Idaho
        Nuclear Company	653

Effectiveness of Porous Glass Elements for Suction Lysimeters to Monitor Soil Water for Organic Contaminants
        Stanley M. Finger, Hamid Hojaji, Morad Boroomand and Pedro B.  Macedo, Catholic University of America	657
Comparison of Mobile Laboratory XRF and CLP Split Sample Lead Results from a Superfund Site Remediation in New Jersey
        Jon C. Gabry, Ebasco Environmental	671
Screening of Groundwater for Aromatics by Synchronous Fluorescence
        R.B. Gammage, J.W. Haas, III and T.M. Allen, Oak Ridge National Laboratory	673
In Situ Detection of Toxic Aromatic Compounds in Groundwater Using Fiberoptic UV Spectroscopy
        J.W. Haas III, T.G. Matthews and R.B. Gammage, Oak Ridge National Laboratory	677
Development of Field Screening  Methods for TNT and RDX in Soil and Ground Water
        Thomas F. Jenkins and Marianne E. Walsh, U.S. Army Cold Regions Research and Engineering Laboratory;
        Martin H. Stutzand Kenneth T. Lang, U.S. Army Toxic and Hazardous Materials Agency	683
Quantification of Pesticides on Soils by Thermal Extraction-GC/MS
        T. Junk, T.R. Irvin, Louisiana State University; K.C. Donnelly and D. Marek, Texas A&M University	687

A Portable Gas Chromatograph  with an Argon lonization Detector for the Field Analysis of Volatile Organics
        Lawrence P. Kaelin, Roy F. Weston, Thomas H. Pritchett, U.S. EPA	689
Sea Mist—A Technique for Rapid and Effective Screening of Contaminated Waste Sites
        Carl Keller and Bill Lowry, Science and Engineering Associates,  Inc	693
                                                          xi

-------
Portable Gas Chromatograph Field Monitoring ofPCB Levels in Soil at the El:a Gate Property
        Marty R. Keller and Gomes Ganapathi, Bechtel National, Inc	697

Real Time Monitoring of the Flue of a Chemical Demilitarization Incinerator
        S.N. Ketkar and S.M. Penn, Extrel Corporation	701

Field Evaluation of the Bruker Mobile Mass Spectrometer Under the U.S. EPA SITE Program
        S.M. Klainer, M.E. Silverstein, V.A. Ecker, D.J. Chaloud, Lockheed Engineering and Sciences Company and
        S. Billets, U.S. EPA	705

The DITAM Assay A - Fast, Fieldable Method to Detect Hazardous Wastes, Toxic Chemicals, and Drugs
        Cynthia Ladouceur, U.S. Army Chemical Research, Development and Engineering Center	709
Rapid Screening of Ground Water Contaminants Using Innovative Field Instrumentation
        Amos Linenberg and David Robinson, Sentex Sensing Technology, Inc	711

Improved Detection of Volatile Organic Compounds in a Microchip Gas Chromatograph
        Aaron M. Mainga and Edward B. Overton, Louisiana State University	713
On-Line Screening Analyzers for Trace Organics Utilizing a Membrane Extraction Interface
        Richard G. Melcherand Paul L. Morabito, The Dow Chemical Company	717

Candidate Protocols for Sampling and Analysis of Chemicals from the Clean Air Act List
        R.G. Merrill, J.T. Bursey, D.L. Jones, T.K. Moody, C.R. Blackley, Radian Corporation;
        W.B. Kuykendal, U.S. EPA	721

The Investigation of Soil Sampling Devices and Shipping and Holding Time Effects on Soil Volatile Organic Compounds
        J.R. Parolini, V.G. King, T.W. Nail and T.E. Lewis, Lockheed Engineering and Sciences Company	725
Developmental Logic for Robotic Sampling Operations
        Michael D. Pavelek II, Micren Associates, Chris C. Fromme, RedZone Robotics, Inc	729
Practical Problems Encountered in Remote Sensing of Atmospheric Contaminants
        Kirkman R. Phelps and Michael S. DeSha, U.S. Army Chemical Research, Development and Engineering Center....733
A SI/LI Based High Resolution Portable X-Ray Analyzer for Field Screening of Hazardous Waste
        Stanislaw Piorek and James R. Pasmore, Outokumpu Electronics, Inc	737
Measurement and Analysis ofAdsistor and Figaro Gas Sensor Used for Underground Storage Tank Leak Detection
        Marc A. Portnoff, Richard Grace, Alberto M. Guzman, Jeff Hibner, Carnegie Mellon University	741

Extraction Disks for Spectroscopic Field Screening Applications
        Edward J. Poziomek, University of Nevada; DeLyle Eastwood, Russell L. Lidberg,
        Gail Gibson, Lockheed Engineering and Sciences Co	747

Field Analytical Support Project (FASP) Development of High-Peiformance Liquid Chromatography (HPCL) Techniques
for On-Site Analysis ofPolycyclic Aromatic Hydrocarbons (PAHs) at PreRemedial Superfund Sites
        Andrew Riddell, Andrew Hafferty and Tracy Yerian, Ecology and Environment, Inc	751

A Field Comparison of Monitoring Methods for Waste Anesthetic Gases and Ethylene Oxide
        Stanley A. Salisbury, G.E. Burroughs, William  J. Daniels, Charles McCammon and Steven A. Lee,
        National Institute for Occupational Safety and Health	755

On-Site and On-Line Spectroscopic Monitoring of Toxic Metal Ions Using Fiber Optic Ultraviolet Absorption Spectrometry
        Kenneth J. Schlager, Biotronics Technology, Inc.; Bernard J. Beemster, Beemster and Associates	759

Rapid Screening of Soil Samples for Chlorinated Organic Compounds
        H. Schlesing, N. Darskus, C. Von Hoist and R.  Wallon, Biocontrol  Institute for Chemische Und Biologische
        Untersuchungen Ingelheim	763
Development of a Microbore Capillaiy Column GC-Focal Plane Mass Spectrograph with an Array Detector
for Field Measurements
        M.P. Sinha, California Institute of Technology	765

Application of a Retention Index Approach Using Internal Standards to a Linear Regression Model for Retention Time
Windows in Volatile Organic Analysis
        Russell Sloboda, NUS Corporation	775

-------
Detection of Airborne Microorganisms Using a Hand-Held Ion Mobility Spectrometer
        A. Peter Snyder, U.S. Army Chemical Research, Development & Engineering Center; David A. Blyth,
        John A. Parsons, Geo-Centers, Inc; Gary A. Eiceman, New Mexico State University	783
Field Analysis for Hexavalent Chrome in Soil
        Robert L. Stamnes, U.S. EPA; Greg D. DeYong, HACH Company; Clark D. Carlson, Bionetics Corp	785
Transportable Tunable Dye Laser for Field Analysis of Aromatic Hydrocarbons in Groundwater
        Randy W. St. Germain and Gregory D. Gillispie, North Dakota State University	789
Real Time Detection of Biological Aerosols
        Peter J. Stopa, Michael T. Goode, Alan W. Zulich, David W. Sickenberger, E. William Sarver and
        Raymond A. Mackay, U.S. Army Chemical Research, Development and Engineering Center	793
Laser Fluorescence EEM Instrument for In-Situ Groundwater Screening
        Todd A. Taylor, Hong Xu and Jonathan E. Kenny, Tufts University	797
Analysis of Total Poly aromatic Hydrocarbon Using Ultraviolet-Fluorescence Spectrometry
        T.L. Theis, A.G. Collins, P.J. Monsour, S.G. Pavlostathis and C.D. Theis, Clarkson University	805

On-Site Analysis of Chlorinated Solvents in Groundwater by Purge and Trap GC
        Stephen A. Turner, Daniel Twomey, Jr., Thomas L. Francoeur and Brian K. Butler, ABB
        Environmental Services, Inc	811
U.S. EPA Evaluation of Two Pentachlorophenol Immunoassay Systems
        J.M. Van Emon, U.S. EPA; R.W. Gerlach, R.J. White and M.E. Silverstein, Lockheed Engineering and
        Sciences Company	815

Rapid Screening Technique for Polychlorinated Biphenyis (PCBs)  Using Room Temperature Phosphorescence
        T. Vo-Dinh, G.H. Miller, A. Pal, W. Watts and M. Uziel, Oak Ridge National Laboratory;
        D. Eastwood and R. Lidberg, Lockheed Engineering & Management Services Co	819
Rapid Determination of Drugs and Semivolatile Organics by Direct Thermal Desorption Ion Trap Mass Spectrometry
        Marcus B. Wise, Ralph H. Ilgner, Michelle V. Buchanan and Michael R. Guerin, Oak Ridge National Laboratory ....823

A New Approach for On-Site Monitoring of Organic Vapors at Low PPB Levels
        H. Wohltjen, N.L. Jarvis and J.R. Lint, Microsensor Systems	829
A Rapid Screening Procedure for Determining Tritium in Soil
        K.M. Wong and T.M. Carsen, Lawrence Livermore National Laboratory	835

Field Preparation and Stabilization of Volatile Organic Constituents of Water Samples by Off-Line Purge and Trap
        Elizabeth Woolfenden, Perkin-Elmer Limited, James Ryan, The Perkin-Elmer Corporation	837

A Field-Portable Supercritical Fluid Extractor for Characterizing Semivolatile Organic Compounds in Waste and Soil Samples
        Bob W. Wright, Cherylyn W. Wright and Jonathan S. Fruchter, Battelle, Pacific Northwest Laboratories	841

Detection of Mercuric Ion in Water with a Mercury-Specific Antibody
        Dwane E. Wylie, Larry D. Carlson, Randy Carlson, Fred W. Wagner, Sheldon M. Schuster, BioNebraska	845
The Effects of Preservatives on Recovery and Analysis of Volatile Organic Compounds
        Kaveh Zarrabi, Steven Ward, Thomas  Starks and Charles Fitzsimmons, University of Nevada	849

Participants' List	851
                                                          XIII

-------
                                        OPENING REMARKS

Welcome to the Second International Symposium on Field Screening Methods for Hazardous Waste and Toxic Chemicals.

Twenty-eight months ago, the first of these symposia was held here in Las Vegas, here at the Sahara, and the response to
that Symposium clearly indicated that the time was right. There was really a need for a forum to exchange information
about the emerging technologies that can be and have been applied to environmental monitoring in the field.

As you can see from the list of the Symposium sponsors, EPA is certainly not alone in its appreciation for these technolo-
gies and their potential for the future. I believe that we have assembled a powerful program for you this next two and a
half days.

The team responsible for this program was made up of, Mr. John Koutsandreas, the Executive Secretary from Florida
State University, Mr. Eric Koglin, the Matrix Manager here at EPA-Las Vegas for the Advanced Field Monitoring Meth-
ods Program and the coordinators at Life Systems, Inc. But for all of the efforts of the Symposium team, it's really the
interest, the enthusiasm, and the participation, over the next couple of days, of all the attendees, that will really set this
Symposium apart.

We are already planning for the Third International Symposium. We try to stagger them in such a way that enough time
elapses — that the papers  aren't the same and the technologies have had an opportunity to advance. We're looking at just
about two years from now.

We are very interested in getting your feedback on what you like and what you don't like about the way the Symposium goes
this year, and any recommendations you can make to help us  strengthen the next Symposium will be greatly appreciated.

This year, we have added a Scientific Awards Committee. Some of you have had an opportunity to see the certificates and
a couple of dramatic eagle trophies as you came in.

We're privileged to have  a number of leaders in the area of environmental measurement here at this Symposium. We'll
share their views and the  views of their organizations about current and future applications of field screening and field
analytical technologies.

Someone once said, Llew, why don't you write a poem? It was a long time ago, but it was never quite forgotten, so if
you'll bear with me:

     The Second International has finally arrived.
     Your program will suggest to you just how hard we have strived.
     To bring to you the latest scoops and field technology
     That's based on engineering, chemistry, biology.
     Besides the platform papers that I know you'll want to hear.
     The poster session entrees will just knock you on your ear.
     And for the technophilic crowd, exhibitors galore
     Will tell you all about their products, and a wee bit more.
     We'll try to slake your appetite for the newest and the best.
     And give you opportunities to mingle with the rest,
     To share and learn,  to see and show our efforts, may they yield
    Accelerated products we can take into the field.
     The future of our measurements, if any bets I'd hedge.
    Resides in these technologies, we're on the leading edge.
     So welcome to this overview of all those things to come.
    And welcome to Las Vegas, where the Rebs are number one.

                                                                   Llewellyn R. Williams
                                                                   Symposium Chairperson

-------
                                                   KEYNOTE ADDRESS
                          ANALYTICAL ISSUES IN THE U.S. EPA SUPERFUND PROGRAM
             Larry Reed, U.S. Environmental Protection Agency, Director Hazardous Site Evaluation Division, Office of Emergency and Remedial Response
I am glad to be invited to this Field Screening Symposium because our
Superfund office in EPA is such a primary user and booster of the
technology. We want to get more and more use out of the technology
and the field analytic methods. It's always good to  be here and
participate. We've been very strong participants and boosters of the
EMSL-Las Vegas operation in field methods, and we'll continue to do
so for years to come.
I wanted to begin by setting the backdrop of where we are in the
Hazardous Waste Superfund Program, and then discuss the vital role
that field methods plays in that program.
We now have a Superfund Program that encompasses a full pipeline:
from discovery of sites (1,500) to two thousand new potential sites
identified to us each year, to the listing of approximately about one
hundred sites a year on the National  Priorities  List, through  the
remedial action and remedial design process. The whole pipeline is in
complete use,  and now more than ever, we're putting higher and
higher emphasis on focusing on the worst sites first throughout the
program. This obviously puts a premium on having the best and the
quickest environmental data available to evaluate and clean up sites
in the program.
The second aspect of where we are now in the Superfund Program that
bears on this Symposium, is we have just, (in December) promulgated
revisions to the Hazard Ranking System (HRS) which will become
effective March 12, 1991. This will expand the types of sites that we
will be looking at and screening. We have added new concerns, a
greater emphasis on ecological concerns, incorporated direct expo-
sure to soils and more emphasis on sediments. We are very proud of
this rule, and we will be gathering a lot more information for screening
sites for future National Priorities List updates.
Also, we have finalized the last of our proposed sites. In the Federal
Register, we proposed ten sites under  the old HRS. All sites have
therefore been finalized.  Eleven hundred and eighty-nine (1,189)
final sites are now on the National  Priorities List. We will be hitting
the ground running, listing new sites as quickly as possible under the
new Hazard Ranking System, for the rest of the Superfund Program.
The focus of our Superfund Program has been on enforcement first,
integrating the use of the fund with the use of  our enforcement
authorities. This is focusing more and more on a  consistent use of
analytic methods, including both field methods and fixed lab meth-
ods, across the program, on appropriate QA procedures across all the
different types of sites, regardless of whether they are enforcement
lead, state lead or fund lead.
The final background point as far as field methods is concerned is our
adoption and phased incorporation of the principles of Total Quality
Management (TQM) into the Superfund Program. We began with
pilot projects last year, designed to embrace the principles of TQM.
The basic concepts of this program include:
   • continual improvement in the process
   • identifying our clients (ensuring that you know who they are, and
    since there are various levels of clients and different relationships
    with those clients)
   • working with our clients
   • identifying and addressing the worst problems first
   • gathering data for informed decision making
These are the kinds of principles we're trying to address in all the
aspects of our program. So more and more, we'll be working with you,
and participating in this kind of audience where, at various times,
either you're our clients or we're your clients. This type of gathering
enforces  that interaction among the various communities that deal
with field screening.
I now want to discuss some specific points on field analysis.
Howard Fribush of my staff went out and visited all 10 of our regional
offices to determine what is the state of the use of field screening in
our very decentralized program. We found field screening has a lot of
purposes including determining worker safety requirements, particu-
larly for our removal  program and for the site assessment program,
which lists  sites.
Field screening obviously provides immediate feedback to the site
assessors, to the samplers and to our clean-up contractors. That, again,
is a strong benefit that we see in encouraging the use of field methods
to continually improve and streamline our Superfund process.
An important application of field screening methods is how they can
be used to shorten the time that it takes to evaluate the risk posed at a
site. This can also be used to generate data to determine the appropri-
ate technologies to be used for clean-up and what levels of clean-up
are appropriate. These applications are evident looking at Regional
history—field screening technologies have been used in the Superfund
Program, basically from its inception. We have seen advances in field
instruments, and this is making on-site analysis at Superfund sites
much more desirable.
As part of Howard's Regional visits,  the different  aspects of our
Superfund program, were polled. The arms of the Program can be
divided into three functional aspects 1) the Site Assessment Program,
the front end of the Program that generates data needed to evaluate the
site and whether it needs to be included on the National Priorities List,
2) the Remedial Program, where once a site is on National Priorities
List  the  actual  clean-up process is initiated, and 3) the Removal
Program, which can be called out at any time to clean up immediate
health threats at sites. We found a split among those different parts of
the program. About ten percent of the data being gathered for the Site
Assessment Program was from field screening. Similarly for the
Remedial Program, about ten percent of the data gathered was with
field screening methods, field  analytic methods. The  biggest user
proportionally was our Removal Program, slightly over a third of the
data gathered from the Removal Program is related to field screening
methods. What we'd like to do is, working through symposia such as
this, try to encourage and increase that use to even higher levels as
appropriate.
The  role that we  play  in the  Office of Emergency and Remedial
Response, and my division, the Hazardous Site Evaluation Division,
is basically providing guidance for this on-site analysis. As I men-
tioned before, when you're dealing with a decentralized program, you
always have to encourage consistency of methods among sites, but
you also have to deal with the uniqueness of each site. We are bridging
the gap by coming up with guidance to provide the Regional offices
on the use of methods.

-------
As follow-up to the Regional review we are evaluating the advantages
of field analysis in the Superfund Program and building on that to
expand its uses  as appropriate in the future. Our future guidance
documents will address evaluating when to use it and then how to use
it. We are also trying to get consistent terminology in our guidance.
Screening technology, portable methods, fieldable methods, mobile
methods, all of these terms have been used. We've been trying in our
field methods catalog to come up with some consistency so even those
unfamiliar with these various technologies, can become familiar with
the basic terminology.
Several major efforts are underway in the Superfund Program. Within
the last year we established our first field methods management
forum. The focus of that field methods management forum was to get
managers involved, not just those that have to go out in the field and
implement the technology, but the managers who would be the ones
to determine what proportion of overall analytic support is necessary
for field methods versus fixed labs. The first meeting of this gpoup was
in June, 1990. We had seven regions, headquarter offices, and the
EMSL-Las Vegas group at this session. The objective of this effort
was  to get management  involved and to focus on the blockages
preventing us from getting field methods used to a greater extent.
Future topics for meetings include: 1) regional administration of field
screening (where does it go, who is  in charge;) 2) collecting the
method and instrument performance information, and 3) trying to get
this data out to the field in the best usable form to those familiar with
the technologies, their usage, their limitations and their strengths.
There's another effort underway — the Field Methods Work Group.
This group contains the worker bees, the people that have to go out and
get the job done. This group has been meeting since 1987. Their initial
focus was looking at things at the very basic level of data quality
objectives—how to define them in order to get them in a more useful
format understood by both the chemists and the field engineers. In
July, 1990, they met and focused on the catalog of field methods and
the need for a  new version. The Field Methods Screening Catalog
User's Guide came out three or four years ago, and we realized the
limitations of it. At the time we wanted to get  information on some 30
different  screening methods. Obviously the next stage is updating
this, adding more methods and more data that will be useful to our
field offices. We expect to release this update of the Field Methods
Catalog some time this year. It will triple the  number of methods that
are contained in the original catalog to about 100.
We are obviously going to be looking at both QA and QC of field
methods. The basic question is the need for Regional consistency.
What is appropriate Q A for a field lab? How can we get that guidance
out? What are the appropriate QC requirements for field methods? We
need to get that information out to the field again, by bringing in the
user community.

Another issue obviously encouraging consistency and appropriate
use of the technology is training. We have been working to come up
with a training program on field methods with the regions and EMSL-
Las Vegas. We've even gotten one of our regional offices to hopefully
loan some of their field equipment in a true bureaucratic gesture to
EMSL-Las Vegas to use as a basis for training programs. We hope to
have this training program developed this year. Obviously the level
we'll have to look at then is  how much and what level of training do
we need to provide out there? How much training should be done for
the people using the field  methods? At what levels should  it be
presented, and how much should be mandatory to ensure and promote
consistency?

There are several basic field method issues that I haven't mentioned,
but that I'd like to touch on before closing: how do we capture
performance information on methods and the instruments? The state-
of-the-art is obviously rapidly changing. How do we capture that
information given, among other things, federal regulations about how
much we can provide in working with industry. How do we capture
that performance data and get it to the field for use in the most useable
form? We have 100 methods that we have looked at for the upcoming
catalog update. What type of data do the people want, and what type
of format? How much? Do they want extensive data, shortened data
or very abstract data. What type of data will encourage the use in the
field?

The final point, and one that I know this Symposium will be working
on, is introducing improved methods, particularly to the Superfund
Program. How do we get the new methods out? What are the incentive
systems? How do we call out and identify the best methods so that they
are being selected for use in the field?

In closing, there are a lot of efforts we have underway to encourage
a maximal, appropriate use of field screening methods. This sympo-
sium is a key one. I mentioned the Field Methods Management Forum
and the Field Methods Work Group, two continuing efforts to provide
direction and recommendations for additional guidance for consis-
tency and use of technology in the field. Field screening methods is a
big field. It is a continuing, emerging field that will continue to
command national attention. We in the Superfund Program are great
boosters and great users of it. I speak as both a provider, working with
EMSL-Las Vegas and  their services, and a user, working on risk
assessments and the site assessment program. I encourage you in your
pursuits to increase the use of field screening methods.

-------
                               OVERVIEW OF DOE'S FIELD SCREENING
                             TECHNOLOGY DEVELOPMENT ACTIVITIES
                                                  by
                    C.W. Frank, T.D. Anderson, C.R. Cooley, KJE. Hain, and S.C.T. Lien
                                    Office of Technology Development
                                       U.S. Department of Energy
                                         Washington, DC  20874

                                              R.L. Snipes
                                        Support Contractor Office
                                     Martin Marietta Energy Systems
                                      Oak Ridge, Tennessee 37831

                                             M.D. Erickson
                          Research and Development Program Coordination Office
                                      Chemical Technical Division
                                      Argonne National Laboratory
                                         Argonne, Illinois  60439
ABSTRACT

The Department of Energy (DOE) has recently created
the Office of Environmental Restoration and  Waste
Management, into which it consolidated those activities.
Within this new organization, the Office of Technology
Development  (OTD)  is  responsible  for  research,
development, demonstration,  testing,  and evaluation
(RDDT&E) activities aimed at meeting DOE cleanup
goals,  while  minimizing   cost  and  risk.     Site
characterization using traditional drilling, sampling, and
analytical  methods comprises a significant part of the
environmental restoration efforts in terms of both cost
and time to accomplish.  It can also be invasive and
create additional pathways for spread of contaminants.
Consequently, DOE is focusing on site characterization
as one of the areas in which significant  technological
Work supported  by the U.S.  Department  of Energy,
under contract  W-31-109-Eng 38.
advances are possible which will decrease cost, reduce
risk, and shorten schedules for achieving  restoration
goals.   DOE is investing considerably in R&D and
demonstration activities which will improve the abilities
to screen chemical, radiological, and physical parameters
in the field.  This paper presents an overview of the
program objectives and status and reviews some of the
projects which are currently underway in the area.
INTRODUCTION

The  Department of  Energy  (DOE)  has  recently
consolidated its  environmental restoration  and waste
management activities into the Office of Environmental
Restoration  and Waste  Management,  formed  by
Secretary James  Watkins in  early 1989.  Within that
Office  of  Technology  Development,  in  part
                     The_submitted manuscript has been authored
                     by a contractor of the U. S. Government
                     under  contract  No. W-3M09-ENG-38.
                     Accordingly, the U. S Government retains a
                     nonexclusive, royalty-free license  to publish
                     or reproduce the published form of this
                     contribution, or allow others to  do so, for
                     U. S. Government purposes.

-------
new   organization,   the   Office   of   Technology
Development  (OTD)  oversees  DOE's  Technology
Development Program, whose objective is to establish
and maintain a national program for applied research,
development,  demonstration,  testing,  and evaluation
(RDDT&E).  These activities will pursue technologies
that will enable DOE to meet its 30-year compliance
and cleanup goals safely, efficiently, and effectively.(1)

The first step in environmental restoration is site and
contaminant characterization.   Characterization of the
current  distribution   of   contaminants  and   the
geohydrological factors that promote and control their
spread will provide the starting point for determining
what  must  be remediated  and for  selecting  and
designing remediation methods.
STATUS OF OTD ACTIVITIES

A cross section of the technology development activities
which have been or are being conducted are described
below.   Space  limitations  preclude  describing  all
activities in this area.  Some  of these activities will be
described in more detail by the principal investigators at
this conference.

DUVAS   Fiberscope   for   in   Situ  Groundwater
Monitoring.   Because of its  proven ability to detect
compounds such as benzene and its derivatives, which
are  common  solvents  and  components   of fuels,
derivative ultraviolet absorption spectrometry (DUVAS)
is being developed as a rapid and reliable method for in
situ  detection of  aromatic  pollutants.   To  date,  a
prototype DUVAS fiberscope has been constructed and
tested for measuring spatial and temporal distribution of
organics  in groundwater.  An important component of
the fiberscope is a rugged,  down-well probe with a
unique "detector-in-head"  design  that  increases the
maximum depth  of   subsurface  detection.    Results
comparable  to  those  obtained  with  a conventional
laboratory spectrometer have been achieved with optical
fiber lengths up  to 50 meters. The portable DUVAS
fiberscope will provide faster, more reliable, and less
expensive measurement of  subsurface  groundwater
contamination.  For  further  information,  contact the
Principal  Investigators, J.W.  Haas   in  and  R.B.
Gammage, Oak Ridge National Laboratory, P.O. Box
2008, Oak Ridge, TN 37831-6113.  Phone: (615) 574-
5042 (Haas), (615) 574-6256  (Gammage).

Advances in Surface-Enhanced Raman Spectroscopv for
Applications  in  Real-Time  Subsurface  Monitoring.
Because  of its excellent selectivity, surface-enhanced
Raman scattering  (SERS)  has attracted considerable
attention  as a potentially powerful  analytical tool for
detecting and  screening trace-level contaminants in
groundwater.  The narrow Raman bands hold promise
for  simplifying  the  identification   of  individual
components  in  complex mixtures.   An  inexpensive
computer-controlled   portable  spectrometer   system
coupled to a fiber-optic probe is  being developed for
rapid  on-site and in situ determination  of organic
groundwater contamination. Critical issues pertaining to
durability, repeatability, sensitivity, selectivity,  and
universality  are  being examined,  while  means for
improvement  in these  areas are being tested.  The
feasibility of utilizing SERS under harsh conditions has
been demonstrated.  Substrates have been tailored for
maximum efficiency at particular excitation wavelengths
as  a  means  for  increasing  the  sensitivity  of the
technique. Ongoing efforts have refined the  state-of-the-
art Raman optrode design and have shown the feasibility
of producing a simple, inexpensive instrument for field
applications.   As the  technique approaches maturity,
SERS will provide powerful  screening  capabilities for
numerous organic and inorganic materials.  It promises
rapid, reproducible, quantitative detection of trace-level
contaminants  in  aqueous  solutions.    For  further
information, contact the Principal Investigator, Eric A.
Wachter, Oak Ridge  National Laboratory, Health and
Safety Research Division, P.O. Box 2008,  Oak Ridge,
TN 37831.  Phone:  (615) 574-6248 (FTS 624-6248).

Fiber   Optic   Raman  Spectrograph   for  in  Situ
Environmental Monitoring.   A small   (suitcase-sized)
surface-enhanced Raman spectrometer (SERS) is being
developed to use in field screening for a  wide variety of

-------
organic and metallic pollutants in ground and surface
waters. The focus of this contract is twofold:  (1)  to
demonstrate a small spectrograph  with high resolution
(3500 cm"1) and (2)
to demonstrate a micro-optical SERS probe head with
substrates engineered to detect certain critical pollutants
at ppm to ppb levels. The spectrograph will have no
moving parts and will employ fiber-optic sampling, an
ultracompact  solid-state  diode  laser  for  Raman
excitation, a high-order diffraction  grating, holographic
optical  filters,  and a state-of-the-art charge-coupled
device  (CCD) detector.   The probe  head will be
contained at the sampling end of a fiber-optic  probe
over 50 meters long inserted into a well less  than 5
centimeters in diameter.  The system will identify trace
contaminants in groundwater in real time.

This  technique  will   increase   the  efficiency  of
environmental characterization and  mapping,  reduce
costs of field sampling and ex situ laboratory analysis,
reduce   personnel   exposure,    and   provide  site
characterization information.  For  further information,
contact the Principal Investigator, Michael Carraba, EIC
Laboratories, Inc., Ill  Downey  St., Norwood, MA
02062, Phone: (617) 769-9540.

In Sim Detection of Organics. The long-term objective
of this research is to develop a fiber-optic-based system
for monitoring contaminant species in groundwater and
to demonstrate  it on contaminated groundwater  at
Lawrence Livermore National Laboratory  (LLNL).
These  efforts require  the  development  of optical
indicator reagents that are compatible with fiber-optic
chemical sensors (optrodes).  Development of optrodes
for ppb-level detection of trichlorethylene  (TCE) and
chloroform (CHC13) is complete and has moved into the
demonstration phase.  Carbon tetrachloride (CCl^ and
perchloroethylene (PCE)  optrodes are currently  being
developed.

The fiber-optic approach has the potential of providing
less   expensive   measurements    of   groundwater
contaminants.   Also, the reagent indicators and the
chemistry developed in the process  of developing the
optrodes  will "spin  off into  other applications.  For
example, one chemistry that was developed serves as the
basis for a proposed TCE remediation technique, the
"TCE sponge".  Finally, it should be pointed out that
these simple indicators are new and could  be used in
other  types  of contaminant  assays.   For  further
information, contact the Principal  Investigator,  Mike
Angel,  Lawrence   Livermore  National  Laboratory,
Environmental Sciences Division, P.O. Box 808, L-524,
Livermore, CA  94550. Phone:  (415) 423-0375 (FTS
543-0375).

Optical Fiber Photothertnal Spectroscopies  for in Situ
Monitoring and  Characterization.  Optical fiber sensors
using thermal lens and photoacoustic spectroscopies for
remote,   on-site,    real-time  optical   absorption
measurements  of chemical  species  in groundwater
environments are being developed. Optical fiber sensors
based on photothermal  spectroscopies are ideal  for
ultrasensitive optical  absorption  measurements   of
actinides  and  other  chemical  species in  aqueous
environments. An optical absorption spectrum provides
qualitative  and  quantitative  analysis  of the  species
present in the aqueous environment.  The spectra can
also provide complexation information for actinides,
which  is  important  for migration behavior.   These
photothermal sensors rely on tunable  wavelength for
selectivity and  therefore do  not require immobilized
agents at the distal fiber end (in the sample  area).

Research   has   demonstrated  two   optical   fiber
photothermal sensors with excellent sensitivity for rare
earth and actinide ions in aqueous solutions.  A remote
photoacoustic sensor was demonstrated using a  100-
meter fiber to deliver the tunable laser beam to a glove
box located in a separate room from the laser. Acoustic
signals were returned to the instrument lab  via coaxial
cables.    An  all-fiber  thermal   lens  sensor  was
demonstrated using a fiber to deliver the laser light to a
remote sample  solution and  a  second fiber,  with a
photodiode attached to the distal end, to measure optical
absorption; electrical cables were  not  required at the
sample  area.   For further information, contact  the
Principal  Investigators,   Richard   Russo,   Lawrence

-------
Berkeley Laboratory, Applied Science Division, M.S.90-
2024, Berkeley, CA 94720, Phone: (415) 486-4258 (FTS
452-4258);  and Robert  Silva,  Lawrence Livermore
National Laboratory, Nuclear Chemistry  Division, L-
396, Livermore, CA 94550.  Phone:  (415) 423-9798
(FTS 543-9798).

Field Measurement of Groundwater Contamination by
Ion Trap Mass Spectrometrv.  A transportable ion trap
mass spectrometer for the in situ characterization of soil,
air, or water at chemical waste sites is being developed
and demonstrated.  The instrument will have a turnkey
operating   system   for  use  by minimally   trained
personnel.   The approach  uses modular design  to
produce an instrument that can be readily modified and
repaired in  the  field.  Specifically, this project will
develop a daughter microprocessor  system to  control
ancillary hardware for sampling and separation and will
develop  new  software,  write  macros,  and  modify
existing software for semi-automated computer  control
of the instrument.

The instrument consists of specialized sampling modules
for air,  soil, or  water samples; a separations module
containing  sorbent  traps  and a megabore  capillary
chromatography  column; and a detection module, the
Finnigan Ion Trap Detector.  Soil or water samples are
purged  with helium  and  the  evolved  organics  are
collected on  sorbent  traps.   A sampling pump  is
incorporated for air samples. The full analysis sequence
required 10 minutes.  The Finnigan software was
modified through the addition  of macros and Forth
routines. The analytical procedure can be selected from
a menu  from the instrument's data system. Sampling,
calibration, analysis, and data reduction proceed under
computer control

The  detection  limit for TCE in water is approximately
20  picograms.  Mass  spectral  identification  of 50
picograms of TCE is possible by library comparison of
spectra.  A linear calibration curve can be obtained from
10 ppt to 10 ppm organics in water.
Although   transportable   mass   spectrometers   are
commercially available for environmental analyses in
the field, the transportable ion trap technology described
here provides several additional benefits, including low
cost.  The instrument can be assembled for a parts cost
of about $75K.  For further  information, contact the
Principal Investigator, Philip H. Hemberger, Los Alamos
National Laboratory, Analytical Chemistry Group, Mail
Stop G740, Los Alamos, NM 87545.  Phone:  (505) 667-
7736 (FTS 843-7236).

Direct Sampling  Mass Spectrometrv.  Rapid analytical
technology  based   upon   direct   sampling  mass
spectrometry is  being  developed to determine trace
organic  pollutants in  the environment.  This project is
jointly sponsored by DOE, the  Department of the Army,
and EPA.  Closely related work is  sponsored by the
National  Cancer Institute (NCI)  for analyses  of
physiological fluids.  Oak Ridge National Laboratory
(ORNL) has developed sampling, sample interface, and
ionization chemistry  techniques  that  are first  being
combined with  commercial  mass  spectrometers  to
provide  rapid laboratory-based methods.  Knowledge
gained is used to develop instrumentation optimized for
on-site  analysis.   Field-sampling  and field-sample-
processing methods are being  developed to support the
mass spectrometric technologies. The general approach
involves a systematic  comparison of  the  developed
methods using  accepted  EPA methods to  analyze
organics in water, soil, air, and waste.  Ion  trap mass
spectrometry (TTMS) and glow  discharge  ionization
quadrupole  mass spectrometry  (GDMS)   are  being
investigated.  Both GDMS and ITMS are applicable to
the quantitative determination  of ppb concentrations of
organics in water and in soil with analysis times of five
minutes or less.  This is achieved by purging the water
or soil-water slurry with air or helium and routing the
purge stream directly  into the mass spectrometer. Less
volatile  organics may be similarly  determined by
collection  on  a  suitable solid sorbent followed by
thermal  desorption.   The method has  thus far been
demonstrated for the  quantitative  determination  of
benzene,  trichloroethylene,  and  tetrachloroethylene.
Applicability to semivolatiles has been demonstrated by

-------
the successful determination of nicotine and cotinine in
urine for the NCI and for die determination of military
chemical agents in air for the Army. A method is under
development for the simultaneous collection of samples
for subsequent confirmatory  analysis in those cases
where  interferences cannot be distinguished by mass
spectrometry   or   by   mass    spectrometry/mass
spectrometry alone.

Successful development and validation can reduce costs
and  increase sample throughput by  up to 90%  as
compared to current regulatory  analytical  methods.
Field-versions of the  technology will allow real-time
monitoring of remedial action progress, monitoring of
associated  occupational exposure, and  screening  of
samples  prior  to  shipment  to  die laboratory  for
regulatory analyses.  For further information, contact the
Principal Investigators, M.B. Wise, M.R. Guerin, and
M.V. Buchanan, Oak Ridge National Laboratory, P.O.
Box 2008, Bldg. 4500-S, MS-6120, Oak Ridge, TN
37831-6120.  Phone (615) 574-4862 (FTS  624-4862)
(Mike Guerin).

Assessment of Subsurface Volatile Organic Compounds
(VOCs) Using Chemical Microsensor Arrays.  A new
monitoring instrument that utilizes an array of coated
surface-acoustic-wave (SAW)  microsensors  is  being
developed.   Pattern  recognition  analysis  of  the
multidimensional sensor output permits determination of
the  identity  and quantity  of  target  vapors from
difference  chemical   classes  typically  found   in
contaminated soils and groundwater.   The small size,
low cost, low power requirements, high sensitivity, and
large dynamic range of the instrument will facilitate its
use in a variety of applications related to site assessment
and process and control.

The project addresses some fundamental questions:  (1)
what is the performance of the SAW microsensor array
instrument in applications relevant to site assessment
and  restoration,  namely, monitoring volatile organic
chemicals (VOCs) in high humidity environments,  (2)
how are the measurements provided by this instrument
related to soil contaminant levels, and (3) how can they
best be  utilized  in  site assessment and restoration
activities? A series of controlled laboratory experiments
will be performed to address diese questions.

The results of this  research  will  demonstrate that
microsensor array instruments  can provide rapid and
reliable  compound-specific  concentrations of volatile
organics  in soil vapor.   The  low projected cost  of
manufacture (less than $1000 in production quantities),
the capabilities of continuous, unattended operation, and
the ability to transmit data from remote locations make
the SAW sensor-based monitors a cost-effective and
desirable monitoring approach.  For further information,
contact  the Principal Investigator, Stuart Batterman,
University of Michigan, Department of Environment &
Industrial Health,  2505 School of Public Health, Ann
Arbor, MI 48109-2029. Phone: (313) 763-2417.

Thin-Layer Detectors: NO2 Detection  with Polystyrene
Thin Layers. A solid-state  sensor that can be used to
detect NO2 without  interference by  other species is
being  developed.     The  device   incorporates   an
interdigitated electrode with a polystyrene thin layer and
operates  by  simply  monitoring   the  change   in
conductance of this  thin film  as a function of NO2
exposure.  Although the film  is an  insulator in  the
absence of  NO2 , showing conductance of less than
10    S,  upon exposure to NO2 gas, an increase  in
conductivity of this highly  insulating material occurs
over several orders of magnitude to  10"8-10"9 S.  No
interference from ambient gases or water vapor has been
observed, and the effect is very specific to NO2.  Upon
elimination  of the  NO2  gas, the  device  becomes
completely  insulating again, all  effects occurring  at
ambient temperature and pressure.

The mechanism of  the  conduction  within  the  film
remains unclear, although the level of conductivity is
related to the amount of residual benzene solvent within
the film.  Thus, as  the  benzene  evaporates from  the
film, the change in conductivity of the film upon NO2
exposure diminishes dramatically. This effect appears
to be related to a stabilization of NO2 dimer by benzene
within the film. The increased conductivity of the film

-------
in the presence of benzene is attributed to the well-
known self-ionization of N2O4 to NO+ + NO3".   For
further information contact the Principal  Investigator
Stephen F. Agnew, Los Alamos National Laboratory,
Los Alamos, NM 87545. Phone: (505) 665-1764 (FTS
843-1764).

Antibody-Based  Fiberoptics  Sensors  For  in  Situ
Monitoring.  Sensitive and selective chemical sensors
for in  situ  monitoring of hazardous compounds in
complex samples are being developed. Special focus is
on a unique fluoroimmuno-sensor (FIS) which derives
its  analytical  selectivity  through  the  specificity  of
antibody-antigen reactions.  Antibodies are imirtobilized
at the terminus of a fiberoptic within the FIS for use in
in situ fluorescence assays under field conditions. High
sensitivity is provided by  laser excitation and optical
detection  techniques.   The  technique  can detect
femtomoles (10  M)  of the carcinogen benzo(a)pyrene
and other chemicals  of environmental interest.   For
further information, contact the Principal Investigators,
T. Vo-Dinh  and  G.D. Griffin, Oak Ridge  National
Laboratory, P.O. Box 1008, MS-6101, Oak Ridge, TN
37831-6101.  Phone: (615) 574-6249  (Vo-Dinh)  and
(615) 576-2713 (Griffin).

Underground Imaging  for Site Characterization  and
Clean  Up   Monitoring.     State-of-the-art  image
reconstruction techniques (tomography) can be used to
characterize the geology and  hydrology of hazardous
waste sites.  These methods extend spatial information
of geologic structure and hydrology between boreholes.
Both two- and three-dimensional imaging can be done
using these techniques. High-frequency electromagnetic
(HFEM) tomography is a proven technology for imaging
water content with high spatial resolution, (i.e., submeter
scale for small geologic scale applications (ten meters).
Electrical  resistance  tomography  (ERT)  is a newer
technology which has  been  used in  the field  with
moderate-scale resolution on larger scale images (meters
on tens to hundreds of meters).

Characterization of  the  subsurface  geology   and
hydrology  is needed to select the most appropriate
remediation alternative and to demonstrate regulatory
compliance.  Design of remedial actions must be based
upon   knowledge  of  the  often   anisotropic  and
heterogenous nature of the subsurface environment and
the natural processes that act upon the waste, as well as
upon protective barriers.  Groundwater flow  strongly
influences contaminant mobilization and transport and
geologic  structure  affects  the  flow of groundwater.
Current  subsurface characterization  techniques  for
addressing these above problems depend heavily upon
drilled  boreholes.   Drilling is expensive and  time
consuming and also creates conduits for contaminant
spread.   A special need exists for three-dimensional
noninvasive  subsurface  characterization technologies.
For more information, contact the Principal Investigator,
William   Daily,   Lawrence   Livermore   National
Laboratory, P.O.  Box  808, L-156, Livermore, CA
94550. Phone: (415) 422-8623  (FTS 532-8623).

Development  of  the  SEAMIST Concept  for Site
Characterization  and Monitoring.   This  project  is
developing the Science and Engineering  Associates'
Membrane Instrumentation  and Sampling  Technique
(SEAMIST). The technique permits rapid emplacement
of instrumentation and sampling apparatus in a punched
or drilled hole.  The objective  of the technique is to
pneumatically emplace an impermeable membrane liner
carrying  many  instruments into a  hole  to provide
simultaneous access to the entire hole wall (e.g., many
measurement   horizons   per hole),  elimination  of
circulation of fluids within  the hole, and isolation of
instruments at discrete locations between the hole wall
and the membrane.  The membrane is emplaced  by
eversion—it is rolled inside out and then everted using
air pressure.  This  causes minimal disturbance to the
hole because the assembly does  not slide down as with
traditional rigid casings.  Instruments such as fiber-optic
sensors, thermocouple psychrometers, gas- and liquid-
sampling systems, and other small instruments are easily
attached to the membrane and carried into the hole with
it.

Using this technique will save  50%-90% of the field
costs, as compared to current monitoring well practices.
                                                  10

-------
In addition, the technique is applicable to both vertical
and horizontal wells.  For further information, contact
the Principal Investigator,  Carl Keller, Science  and
Engineering Associates, 612 Old Santa Fe Trail, Same
Fe, NM 87501. Phone: (505) 646-5188.

Site Characterization and Analysis Penetrometer System
(SCAPS).  DOE  is working  with the Department of
Defense on the further development and demonstration
of the SCAPS for use on DOE facilities. The SCAPS,
as  developed  by  the Army   Corps  of  Engineers
Waterways Experiment Station for the Army Toxic and
Hazardous  Materials  Agency,  includes  surface
geophysical equipment, survey and mapping equipment,
sensors  for contaminant detection, and soil sampling
equipment.  Computer systems  have  been integrated
with the SCAPS in order  to  provide data  acquisition,
data processing, and 3-D visualization of site conditions.
The system is mounted on a uniquely-engineered truck
that provides protective work spaces to minimize worker
exposure to toxic chemicals.  The truck also provides
equipment to seal each penetrometer hole with grout.

Real-time  sensors  that are  currently  available  for
characterization work include those which can determine
the strength, electrical resistivity, and spectral properties
of soils.  Two sensors successfully  demonstrated to
detect contaminant plumes at DOD facilities are the soil
resistivity unit  and a fiber optic contaminant sensor.
The primary advantage of the fiber-optic sensor over
resistivity measurements is  based on  laser-induced
fluorescence, which presents a problem for contaminants
such as TCE that do not fluoresce; however, colorimetry
and absorption techniques such as the sensors which are
being developed   by Lawrence  Livermore National
Laboratory and by Fiberchem  are tentatively planned to
be demonstrated in conjunction with the penetrometer at
the Savannah River integrated demonstration in FY-91.
Additionally,  samplers such as  the  "Terra  Trog"
developed by the  Army Corps  of Engineers may be
tested in FY-91 at the Savannah River Site.  For further
information, contact the Principal Investigator, Stafford
Cooper, Waterways Experiment Station, P.O. Box 631,
Vicksburg, MS 39181-0631.  Phone: 601-634-2477.
Design. Manufacture, and Evaluation of a Hvdraulicallv
Installed. Multi-Sampling Lvsimeter.  A new lysimeter
sampling  device  design,  approximately  1 inch  in
diameter, having multiple sampling zones and capable
of being hydraulically installed at a desired depth in the
vadose zone without drilling will be developed.  This
lysimeter will be readily retrievable  for reuse and will
provide  an  inexpensive  monitoring   technique  in
comparison to installation of lysimeters into predrilled
holes.   In  this  project,  the  hydraulically  inserted
lysimeter will be designed and constructed.  The effect
of hydraulic insertion on the operation of the lysimeter
will be investigated by comparing hydraulic insertion
with standard boring procedures. The lysimeter should
be commercialized within three years. This new design
is  less disruptive  to  the subsurface, both  during
installation and after removal, requiring only a  1-inch-
diameter hole vs. the 4-inch holes commonly drilled for
monitoring wells.  Costs are estimated to be under 50%
of that to  drill monitoring wells.   This project is  a
collaborative effort  among Bladon International, Inc.,
Institute for Gas Technology, and Timco Manufacturing.
For   further  information   contact   the  Principal
Investigator, Joe Scroppo, Bladon  International, Inc.,
880 Lee Street, Des Plaines, IL 60018.  Phone: (505)
883-3636.

Minimally   Invasive   Three-Dimensional   Site
Characterization.  Hardware  and software are being
developed  to   permit  data  acquisition from   three
minimally  invasive  measurement   techniques—cone
penetrometer,  synergistic   electromagnetic  mapping
technology and reflection seismology.  The software will
permit rapid feedback, comparison, co-calibration, and
data   analysis   from   the   combined  technology.
Simultaneous application of these  three technologies
permits physical and electrical property measurements
to be  used to cross-calibrate each data set.  The early
acquisition of preliminary  data  allows field personnel
quickly to adapt their field study strategy to changes in
the  perceived  site  conditions   or  contamination
distribution.
                                                  11

-------
 Costs will be saved by rapid feedback of the data to
 field personnel, the improved informational quality, and
 the lower cost of an integrated system.  The minimally
 invasive  system reduces environmental impact  and
 reduces risk to field personnel. For further information
 contact Principal Investigator, John Gibbons, Applied
 Research Associates, Inc., 4300 San Mateo Blvd., N.E.,
 Suite A220, Alburquerque, NM 87110.  Phone: (505)
 883-3636.

 High  Resolution  Shear  Wave  Seismic  Reflection
 Surveying for Hydrogeological  Investigation.    This
 technology will enhance the ability to directly determine
 aquifers  in the characterization and sensing of geologic
 and hydrogeologic features. The project will extend the
 state-of-the-art of  shallow subsurface hydro-geological
 characterizations by means of high resolution shear (S)
 wave  seismic reflection  profiling.   High  resolution
 seismic   reflection   profiling   using   conventional
 compressional (P) wave technology has evolved over the
 past ten  years to  the point where this technique  has
 become a major component of numerous environmental
 investigations. Extension of the existing technology to
 include S-wave reflections has the potential for greatly
 enhancing  the data which can be extracted from  the
 subsurface. Unlike a P-wave, an S-wave will not travel
 through  a purely liquid medium,  hence  its  advantage
 over current P-wave techniques.

 Conventional high-resolution seismic reflection profiling
 has proven cost-effective for environmental assessment
 by reducing the number of holes and the cost of boring.
 S-wave  reflection  technology   will  enhance   the
 information content of the seismic reflection technique
 and improve the cost-effectiveness of the technique. For
 further information contact the Principal Investigator,
William  Johnson, Paul C. Rizzo Associates,  Inc., 300
Oxford Dr., Monroeville, PA 15146. Phone: (412) 856-
9700.  .

Field Measurements for the Hydrology and Radionuclide
Migration Program (HRMP) at the Nevada  Test Site.
The  HRMP was begun in  1974  for the purpose of
determining the potential for migration of radionuclides
 from underground test areas. HRMP is a multi-agency
 research project and is  coordinated by  the  Nevada
 Operations  Office  of DOE.   The  participants  are
 Lawrence Livermore National Laboratory, Los Alamos
 National Laboratory, Desert Research Institute, and the
 U.S.  Geological Survey.   The present  goals  of the
 program are to learn more about the groundwater rates
 and directions of flow on the Nevada test site  (NTS),
 which is located approximately 80 miles northwest of
 Las Vegas,  in regional and local  systems, to develop
 mathematical models of the flow systems,  to determine
 the effects  of nuclear tests on the systems,  and to
 measure the migration rates of selected radionuclides
 under various conditions.

 Transport   mechanisms   for  radionuclides   from
 underground  nuclear  detonations  are   studied  by
 sampling  both the  contaminated  cavity  water and
 groundwater pumped from the surrounding formation.
 Radioactivity  in  water  greater  than  9-cavity-radii
 distance from the detonation point has been measured
 without stressing or pumping the aquifer.  A plume of
 radioactivity which is being rapidly transported by the
 local  groundwater has been intercepted.   Micro- and
 ultrafiltration studies on this groundwater  have  shown
 that radionuclides   can  be  present  and  mobile  in
 groundwater systems in colloidal form.  Water pumped
 from a tritium contaminated satellite well over a 20-year
 period drains into a mile-long ditch and has created a
 secondary  site emphasizing  the   unsaturated   zone.
 Current  studies  along   the   discharge   ditch   are
 investigating the moisture and  tritium front through
 shallow alluvium. This project  is  developing systems
 which can measure  contaminants  such  as organics,
 tritium, and long-lived radionuclides in wells in  depths
 from  1400  to 3300 feet.   For further  information,
contact the  Principal Investigator, Jo Ann Rego  at
 Nuclear  Chemistry   Division,   Lawrence  Livermore
 National Laboratory, P.O. Box 808, L-234, Livermore,
 CA 94551. Phone: (415) 422-5516 (FTS 532-5516).

Depth Profiling in the Water Table Region of a Sandy
Aquifer. The feasibility of using a new  multilayered
sampler  to  investigate   organic  contaminants  in
                                                  12

-------
groundwater is being explored.  The device passively
collects  simultaneous   groundwater  samples  from
multiple  levels in the subsurface.   In  addition,  the
project will develop a new device based on experience
with existing sampler.

The  sampler, developed at  the Weizmann Institute of
Sciences,  Rehovot, Israel,  was  used  to detect  the
presence of several inorganic and organic species at a
contaminated  Brookhaven   site.   The   presence  of
microscale heterogeneities  in  concentration  gradients
over a vertical interval of  200 cm  was observed for
eight solutes, including  metals, organics, and anions. A
planned remediation was  modified based on results of
this short sampling event.  It is believed that the new
plan will be  more cost effective  than  the  original
because the  contamination  was  better defined in  the
vertical plane and because an oxygen-depleted zone was
found  where it was previously thought to be fully
saturated.  For further information, contact the Principal
Investigator, Edward   Kaplan,  Brookhaven  National
Laboratory, Radiological  Sciences Division, Building
703M, Upton, NY 11973-5000. Phone: (516) 282-2007
(FTS 666-2007).

Kr81  Counting for Nuclear Waste  Sites.   A  new
technology to date groundwater is being developed.  By
combining resonance ionization spectroscopy and mass
spectroscopy, ultralow levels of Kr81 in groundwater can
be detected.  From the  quantity of Kr8 , the age of the
groundwater can be determined. This information helps
find  suitable locations to store nuclear wastes or highly
toxic chemical wastes in groundwater. Several samples
from  Europe  have been tested and the results  are
adequate to search  for new waste sites.  It is beneficial
to the Department  of Energy waste program to find a
geologically  safe place to  store  nuclear wastes  and
highly toxic chemical wastes. For further information,
contact the Principal Investigators, C.H. Chen and M.G.
Payne, Oak  Ridge National Laboratory, Photophysics
Group, Building 5500,  MS-6378, P.O. Box 2008, Oak
Ridge, TN 37831-6378. Phone:  (615) 574-5895 (FTS
574-5895).
FUTURE   TECHNOLOGY   DEVELOPMENT
NEEDS

The OTD activities described here address some, but by
no means all, of the key needs which DOE foresees in
the area of in situ monitoring.

Present site characterization  methods  are  imprecise,
costly, time-consuming, and overly invasive. Improved
site  characterization  methods  will  require   better
technologies for accurately describing  the subsurface
geohydrologic features of a site.  For example, more
efficient nonintrusive sampling strategies and practical
models are necessary for understanding and predicting
subsurface transport.  Also needed are  more reliable
procedures for interpreting characterization data, such as
how clean is "clean".

Traditional hydrologic characterization of the subsurface
environment  is  highly  dependent  on  data  from
groundwater   monitoring   wells.      A   thorough
understanding of the subsurface environment requires a
series of hydraulic wells. Interpretation depends greatly
on proficiency of the scientific staff, making subsurface
characterization highly subjective and at times uncertain.
Research is needed to make hydrologic characterization
more precise and more cost effective.

Currently accepted analytical procedures such as those
in the Environmental Protection Agency's (EPA's) SW-
846 do not cover all materials that need to be measured
at DOE sites.  DOE is working with the EPA and others
to alleviate such problems with sampling and analyses.
Close  coordination  with EPA and  other regulatory
agencies is needed not only to identify, develop, and
validate appropriate methods,  but  also to ensure  the
acceptance of data generated using these methods.

Intrusive exercises,  such as sampling and excavation
during remediation of a site, often involve immediate
hazards to  workers  in  the  form  of exposure  to
radioactive  and/or toxic materials.   Remote  real-time
analyses of ambient levels of potential hazards in the
air, water, and soil during characterization, as well as in
                                                   13

-------
the remedial action phase, would help ensure worker
safety and allow continuous operation. Instrumentation
capable  of  detecting  broad  classes of hazardous
materials and specific compounds is needed to indicate
cleanup status.  Better characterization methods based
on real-time analyses are especially important to confirm
the most effective use of certain in  situ remediation
technologies.  In the absence of real-time monitoring,
excessive volumes  of soil and water must be treated to
guarantee   compliance;   otherwise,   pockets   of
contamination may be missed.

Special characterization technologies are necessary for
inactive  facilities,  underground  storage  tanks, and
wastewater  lagoons.  These facilities often  contain
significant  quantities of radioactive wastes, in certain
cases mixed with heavy metals and/or hazardous organic
compounds  that make  personnel entry unacceptable.
Thus, the development of advanced robotic samplers,
smart probes, mobile and in situ fiber-optic devices, and
nonintrusive characterization instrumentation (based on
electromagnetic, thermographic, and acoustic principles)
is needed for sampling and chemically characterizing
these sites.  The development of such techniques will
significantly reduce radiological exposure to  workers
and provide more  assurance  that the  correct remedial
technology has been selected.

Clearly, there are more technology development needs
and more good ideas than there are resources to devote
to these investigations. Priorities must be set to support
those activities deemed most urgent.
OPPORTUNITIES FOR PARTICIPATION

OTD is interested in eliciting broad participation from
qualified  organizations  who  can  contribute  to  its
RDDT&E activities.  We are  becoming increasingly
aware of the wealth of technological talent and good
ideas in all sectors.  OTD has initiated steps during the
past year to  increase participation of the private sector
(academia   and   industry)   through  competitive
solicitations   and  through funding  of  unsolicited
proposals.    We  have  also  worked  to  increase
participation   by  academia   through  interagency
agreements for cooperative funding of research and
through establishment of  DOE educational consortia.
Several significant technology development activities are
being  conducted at   DOE sites  such  as  national
laboratories. DOE is  funding technology development
activities  beyond the United  States  through  direct
contracts,   international   agreements,   and   other
mechanisms.

DOE  plans to continue  this  type of  support  for
technology  development   in  the  coming   years.
Organizations interested in responding  to solicitations
should contact John Beller (for Innovative Technology)
at Innovative Technology Program Coordination Office,
EG&G Idaho, P.O.  Box 1625,  Idaho Falls, ID 83405-
6902.  Dr. Erickson (for  applied R&D) at the  above
address or Mr.  Snipes (for DT&E) at the above address
to be placed on distribution lists. Organizations wishing
to submit unsolicited proposals should contact  Larry
Harmon, Director, Division of Program Support (EM-
53), Department of Energy,  12800 Middlebrook  Road,
Trevion II Building,  Germantown,  MD 20874,  for
information on submission format and procedures prior
to preparation of a proposal.
REFERENCES

1.     United   States   Department   of   Energy
      Environmental   Restoration   and   Waste
      Management. Five-Year Plan, Fiscal Years 1992-
      1996, June  1990, DOE/S-0078P.
                                                  14

-------
           DEPARTMENT OF DEFENSE FIELD SCREENING METHODS REQUIREMENTS IN THE
                                 INSTALLATION RESTORATION PROGRAM
                                              Mr. Dennis J. Wynne
                                 U.S. Army Toxic and Hazardous Materials Agency
The Superfund Amendments and Reauthorization Act
(SARA) and the implementing executive orders under this
legislation require that contamination resulting from Depart-
ment of Defense (DOD) past operations be remediated. In
response to this legislation, the DOD has undertaken a com-
prehensive program to comply with these mandates. Over the
years this program has expanded from a $150 millon effort in
FY 1984 to a $1 billion effort in FY 1991. Some 17000 sites
have been identified at 1808 DOD Installations. Ninety DOD
Installations have been identified on the National Priorities
list by the Environmental Protection Agency. The detection
and remediation of contamination is a long term and resource
intensive effort. Research that allows us to proceed more
quickly in locating contaminants and in pin pointing key soil
and water samples for analysis, assessment, and remediation
purposes can provide a tremendous resource savings to the
ITR Program and, ultimately, the taxpayer. It is noted that
over 30% of the budget is estimated to be totally dedicated to
drilling, sampling and sample testing. Any improvement in
Field Sampling and Analysis will quickly repay the cost of
its associated research and development.
DOD Field Sampling and Analysis accomplishments include
the fielding of a truck-mounted cone penetrometer for more
efficient contaminant plume identification, tracking and
reducing well drilling requirements. Also completed was the
development of a field Analytical Method for the explosives
TNT and RDX in soil and water. Current program efforts
include the development of various contaminant sensors to
be employed in the cone penetrometer system to define
concentrations of contaminants in soil and groundwater as
the penetrometer is advanced through the soil. Future plans
include the concept of placing sampling devices into the
ground with the penetrometer which can be sampled and
analyzed with field instrumentation at regular intervals
thereafter.  All these efforts have significant cost reduction
implications and have the interest and funding support of not
only DOD  but also DOE.
                                                       15

-------
                      AN OVERVIEW OF ARMY SENSOR TECHNOLOGY APPLICABLE
                       TO FIELD SCREENING OF ENVIRONMENTAL POLLUTANTS
                                     RAYMOND A. MACKAY
                                U.S. Army Chemical  Researchf
                              Development and Engineering Center
                                    Detection Directorate
                                      ATTN:  SMCCR-DDT
                           Aberdeen Proving Ground, MD  21010-5423
                  ABSTRACT

    The Army has under development a number
of technologies directed toward the field
detection and Identification of chemical  and
biological  (CB) agents.  This Includes not
only specific sensors, but the technology
required to Integrate these sensors Into
effective field detection systems.  Much of
this technology can be adapted to materials
of environmental concern.  In particular,
there are technologies 1n various stages of
development which are applicable to vapor
and aerosol clouds, as well as to
contaminated surface water and terrain.
These Include both point sampling and
monitoring systems, as well as remote sensing
systems capable of providing rapid wide area
coverage.  This paper will provide an
overview of Army programs applicable to
field screening methods, with particular
emphasis on mass spectrometrlc. Infra red,
and aerosol sampling technologies.
the form of vapors or aerosols.  The two
main areas which will be covered are
standoff detection and point detection.
Standoff detection has sometimes been
referred to as remote detection.  However,
remote detection 1s defined here as the use
of point detectors which are located at the
site to be monitored, which may be of some
distance from the main monitoring station or
base, and connected to 1t by hard wire or
telemetry.  Standoff detection refers to the
use of equipment located at the monitoring
base which can sense chemicals at a distant
location.  The point detection technology to
be discussed 1n this paper Is pyrolysls-mass
spectrometry.  There will also be some
discussion of aerosol sampling, since this
is pertinent to point detection of
aerosolized particulates, liquid or solid.
It 1s not the aim of this paper to present
detailed experimental results but rather to
provide an overview of the technology and
Its range of applicability.
                INTRODUCTION

    Technologies which can be utilized for
the detection of chemical  warfare agents In
the field may also be applicable to the field
detection, classification  and identification
of various substances of environmental
Interest.  Although Army detection programs,
particularly those 1n the  early stages of
development, focus on biological as well as
chemical  detection, and much of the
technology 1s applicable to both.  In this
paper, the emphasis will be on chemicals 1n
                DISCUSSION

STANDOFF DETECTION:  The U.S. Army Chemical
Research, Development and Engineering Center
(CRDEC) 1s currently engaged in an extensive
multi-year exploratory development program
to exploit laser radar for Chemical
Biological (CB) Standoff Detection.  At
present, the only near term capability for
the detection of chemical agents at a
distance Is the use of passive infrared
sensors.  These sensors can detect only
chemical vapors.  Active (laser) infrared
                                                  17

-------
(IR) systems employing  Differential
Scattering and Absorption  L1dar (DISC/DIAL)
are being developed  for the  detection of
chemical agents  1n all  physical  forms:
vapors, aerosols, and rains,  as well  as
liquid surface contamination.   In  addition,
an ultraviolet (UV)  system employing  laser-
Induced fluorescence 1s being  developed for
the detection of biological  clouds
consisting of organisms, toxins and  related
materials.  The  principles of  operation of
these systems and the background of  their
development will be  briefly  discussed.   The
IR and UV breadboard systems  have  recently
been used 1n an extensive  field test
employing various non-toxic  chemicals and
Interferents with excellent  results.   These
data will be discussed  along with  the
necessary development efforts  required  to
adapt the DISC/DIAL  technology to  practical
field use.

    The Army is making  a significant
investment in standoff  technology  because it
1s the only technology  known that  can provide
rapid wide area surveillance capability
while simultaneously reducing  the  total
number of detectors  required.   At  CRDEC
         there are three phases to the  Standoff
         Detection program; the XM21 Passive  Remote
         Sensing Chemical Agent Alarm,  along  with
         technology upgrades; the Laser Radar (LIDAR)
         CB Standoff Detection System;  and, for  the
         future, combining these technologies with
         other electro-optic systems 1n integrated
         sensor suites.

             First to be discussed is chemical
         detection portion of laser radar  project
         called IR DISC/DIAL.  The objective  is  to
         provide chemical laser Standoff detection
         systems for CB defense applications.  The
         planned systems capabilities are  to  scan
         surrounding atmosphere and terrain,  operate
         in fixed or mobile mode, detect chemical
         contamination in all its physical forms, and
         range resolve, quantify and map data.   The
         purposes of the current program are  to
         demonstrate concept feasibility,  establish
         capabilities and limits, complete science
         base, determine effectiveness  in  field
         situations and establish basis for rapid
         transition to mature development.  The  IR
         DISC/DIAL system can develop data in  four
         ways (as shown 1n Figure 1):
                                          FIGURE 1
                                             AGENT VAPOR
                        TOPOGRAPHICAL REFLECTION
                               (VAPOR)
                         DIFFERENTIAL ABSORPTION
                               (VAPOR)
AGENT VAPOR .
       V.
NATURAL
AEROSOLS
                         DIFFERENTIAL SCATTERING
                             (AEROSOL/RAIN)
             AGENT/RAIN/
              AEROSOLS -
                          DIFFERENTIAL SCATTERING
                          (SURFACE CONTAMINATION)
                SURFACE
             CONTAMINATION

                                                                         A0332-EE6 23400"
                                                     18

-------
Topographic reflection DIAL;  By transmit-
ting different IR frequencies and detecting
their topographic return* chemical vapor
clouds can be Identified by their selective
absorption of some of the IR frequencies.
This measurement detects the presence of the
cloud and Its total concentration times path
length (CD; however, It does not tell how
far away the cloud is or Its density
(concentration).

Aerosol backscatter DIAL;  By the same
technique, but with higher laser power, the
normally occurring atmospheric aerosol
begins to reflection IR energy back to the
detector.  This distributed reflector can be
"range resolved" by gate timing the
returning signal just as radar systems do.
In this way» average concentrations and
ranges can be developed for many cells
(range lines) down the LIDAR path.  By
scanning the system spatially, a map can
then be made of vapor chemical agents.

Agent backscatter DISC:  In the same manner,
chemical agent aerosols and agent rains can
be detected by the selective frequencies
that they directly backscatter to the
detector.

Surface reflection;  The fourth mode of
detection is the detection of selective IR
frequencies backscattered from agents on
surfaces.  This measurement is dependent on
the amount of material located on the
surface of dirt, grass, trees or equipment.
                                          FIGURE 2
    Figure 2 shows that, for each of the
detection modes, the return signals are
different so that all measurements can be
made simultaneously.  This is important
because there are no significant hardware
design constraints to add aerosol rain and
surface detection to an aerosol backscatter
DIAL system.  The first objective of the
DISC/DIAL project was to build a Ground
Mobile Breadboard (GMB) system to demon-
strate the feasibility of DISC/DIAL chemical
detection.  The system was mounted in a van
and tested.  Based on these tests, the GMB
was upgraded.  The current specifications of
the Ground Mobile Breadboard Upgrade (GMBU)
are given in table 1.

    The GMBU along with other devices was
then exposed to extensive U.S. Army Dugway
Proving Ground (DPG) field testing.  The
goals of these tests were:

    (1)  Investigate effects of reducing
system size, weight and power on detection
performances.  This was because the Army's
near term use was a ground mobile vehicle
application for reconnaissance.

    (2)  Obtain quantifiable data on vapors,
aerosols, and liquid detection and on
interferences to prove feasibility.

    (3)  Use more realistic field scenarios
to develop workable use concepts.
                                          BACKSCATTER FROM
                                          "NORMAL" CLEAR AIR
                                                 AEROSOL CLOUD

                                                        VAPOR CLOUD
                                                                   HARD TARGET
                                                                     RETURN
                                                      TIME
                                                  19

-------
Four C02 TEA Laser
      TABLE I.   DISC/DIAL Specifications

Transmitter

Lasers

TunablHty           Line-Tunable by Grating

Wavelengths          9.2 to 10.8 Microns

Energy (on 10P20)    2.0 J/Pulse

Pulse-to-Pulse       ±.3 Percent
  Power Stability

Pulsewldth (3dB)     90 ns

Repetition Rate      20 Hz

Beam Divergence      3.5x4.0 MRAD

Mode                 Multlmode or TEMoo

Timing Jitter        2 NS Pulse-to-Pulse


Receiver

Telescope Diameter   16 Inches

Detector             HgCdTe Quadrant

Size                 1x1 mm Per Element

Detectivity          4x10 cm/Hz1/2w

Field of View        8 MRAD

Overall Electronic   10 Hz to 7 MHz
  Bandwidth

    These tests Involved large scale
simulant clouds created by a special 100
meter long spray system as well as aircraft
spray.  Also, aerosols were generated by
spray from a high ranger boom, and surfaces
(such as dirt, grass, concrete, trees, or
vehicles) were coated with simulants.  The
many accomplishments of these large scale
tests were:

    -  Demonstrated feasibility of ISC/DIAL
       technology

    -  Demonstrated high sensitivity

    -  Demonstrated operation 1n motion,
       scanning and mapping

    -  Detected cloud through a cloud
-  Detected collocated DMMP and SFg

-  Detected DMMP (dimethyl
   methyl phosphorate)

     -  up to 5 Km (range resolved)

     -  up to 10 Km (column-content)

     -  1n presence of all Interferents
        (fog, rain, dust and military
        smokes)

     -  on ground by secondary vapor

     -  at night and 1n reduced
        visibility

     -  1n calibrated chamber

-  Detected SF96 - as an aerosol

                 - as ground
                   contamination
                   on six surfaces

-  Detected other volatile and non-
   volatile simulants

-  Validate emulation and simulation
   models
                                 Figure 3  shows a typical GMBU map of a
                                 simulant  vapor cloud.  Although not evident
                                 in  this black and white  Illustration, the
                                 range cells are colored  to show the average
                                 concentration from 0.1 to 2.0 mg/nrr.

                                     Additionally, this field work was backed
                                 up  with an extensive emulation and simula-
                                 tion program which was able to show excel-
                                 lent correlation between predicted and
                                 actual performance.  For example, the DMMP
                                 and SFg 1 Km range resolved predicted and
                                 measured  values are Identical.  Using this
                                 excellent agreement, one can Infer the
                                 following  sensitivities to chemical vapors
                                 with strong absorptions  1n the 9-10 micron
                                 region of the Infrared.
                                     Column Content

                                    2 Km        10 Km

                                  10 mg/m      12 mg/m2
                     Range Resolved

                          1 Km

                        0.5 mg/m3
                                The minimum detectable concentration of
                                liquid  simulants on the ground were measured
                                at 0.5-5.0 g/m  depending on the porosity of
                               the surface.  Also very encouraging  1s  the
                             20

-------
                                    FIGURE 3  GMBU MAP
fact that one four wavelength set (1/20 sec
data)  can provide a high amount of informa-
tion about the situation.   An example:
                            between biological simulants and inter-
                            ferents/backgrounds of UV/LIF are below:
Accuracy of Prediction
(Range  Over All  Data)

97.2-100 percent
87.4-87.8 percent
66.2-74.1 percent
   Information
1 simulant on any
1 of 5 surfaces
                 Scattering
                Signal  Level
                   248 nm
            Fluorescence
            Signal Level
             280-410 nm
1 simulant on
of 5 surfaces
any 5
3 simulants on any
6 of 6 surfaces
This demonstrates that a real time surface
detection algorithm can be developed.

    The UV LIF based laser radar was also
successfully tested at DPG for detection of
biological and toxin materials.  While not
nearly as far along in development as the IR
system, this system demonstrated significant
detections at ranges up to 1.2 Km.  The
system, which measures the laser induced
fluorescence of tryptophane, a compound
occurring in all living material, can sense
the presence of biological/toxin clouds but
cannot as yet uniquely identify the
material.  Relative optical discrimination
                                                            None
                                                            None
                                                            None
                                  Small
                                  Small
                                  Strong
                                  Strong
Tryptophane


    EG


Egg Albumen


Diesel Exhaust


 Auto Exhaust


  Road Dust


   Trees
Strong


Strong


Strong


Strong


 Weak


 None


Strong
                            Other optical concepts based on Mueller
                            Matrix scattering are currently being
                            investigated to  add additional identifica-
                            tion capabilities to UV/LIF system.
                                                  21

-------
Passive IR.  The standoff detection and
Identification of chemical vapor clouds 1s
currently achieved by recording the IR
spectrum 1n the 8-12 micron wavelength
region by means of an Interferometer.  This
Is the XM21 Remote Chemical Agent Sensing
Alarm.  It Is a tripod-mounted device
weighing approximately 55 pounds, exclusive
of the source power.  It scans a 1.5° field
of view (FOV) for 2 seconds, co-adding eight
scans.  If the cloud fills the entire FOV,
the sensitivity Is on the order of a
concentration-path length product of 150
mg/m  , the precise value depending upon the
strength of the absorption bands.  The
interferogram, taken 1n the time domain, 1s
converted to a frequency domain spectrum In
the microprocessor by means of a fast
Fourier transform.  A background spectrum of
the FOV must be obtained and stored, and
then  subtracted from the sample scan prior
to further signal processing.  Because of
the relatively slow scan speed, and the
requirement of the current algorithm for a
background subtract, 1t cannot be operated
from  a moving platform.

    A lightweight  (20 Ibs), fast scan
interferometer  is under development.   In
addition,  recent developments  1n direct
signal processing  In the time  domain have
both  reduced demands on the microprocessor
and  relieved the  requirement for a
background scan.   Since  results equivalent
to those  on the XM21 can  be achieved  in  a
single scan without  a pre-determined back-
ground spectrum, this device can be  operated
from  a moving  platform  such as a ground
vehicle or alrframe.  Thus, if only  vapor
detection 1s  required,  passive technology
 represents an  attractive  method  for  rapid
survey of an  area,  particularly by air.

     In summary,  CRDEC  has demonstrated the
 feasibility  of IR DISC/DIAL technology for
 the detection of chemical  agents  in  all
 forms, as well  as passive IR  for chemical
 vapor detection.   Prototypes  for ground
 mobile,  fixed site and  test facility appli-
 cation are beginning to be developed.   The
 potential exists for modifying these systems
 to mount on helicopters, RPVs, and even
 satellites,  and to add the capability of
 detecting biologicals and toxins,  as well
 as chemicals.

 POINT DETECTION:  There are two specific
 technologies which form the basis of
 recently fielded and developmental Army
 point detectors; namely, Ion mobility and
 mass spectrometery.
Ion Mobility Spectrometrv.  This 1s a
technology which operates at atmospheric
pressure.  The air sample containing the
vapor(s) to be detected are drawn through a
permselective membrane Into an ionizatlon
region where reagent gas Ions react with the
(polar) compounds to be detected and form
cluster ion species.  These are gated Into a
drift tube where the ions migrate under an
applied electric field, and are separated
according to their mobility as measured by
their time of arrival at the collection at
the end of the drift tube.  These may be
operated in both a positive and negative
mode.  The U.S. Army currently has fielded a
hand-held monitor, the Chemical Agent
Monitor (CAM), and has a point alarm system
(XM22) under development.  These relative
low weight, man portable, field hardened
devices are quite sensitive and should be
quite useful for field screening and
monitoring of a wide variety of
environmentally hazardous vapors.  Since
this technology and  Its applications will be
discussed extensively  in the symposium,  1t
will not be considered further here.

Mass Spectrometrv.   A  mass spectrometer
system which can provide  sensitive,
effectively  real time  detection and
identification of chemicals  in the  form  of
vapors,  aerosols, and  ground surface
contamination,  is currently  under
development  by CRDEC.  Since this system
also has the potential to detect materials
of biological origin,  it  is  referred to  as
the Chemical Biological Mass Spectrometer
 (CBMS).

    The CBMS consists  of  two major
components,  the  biological  probe and the
mass analyzer chassis.  An  artist's concept
 is shown in  figure  4.  The  biological
 sampling probe  contains  the virtual  impactor
 and infrared  pyrolyzer.   The mass  analyzer
 chassis contains the mass analyzer,
 instrument computer, data processing
 computer and display,  alarm and
 communication  modules.

     The virtual  Impactor block of  the
 biological  sampling probe consists of  a 1000
 l/m1n  pump and  a four stage virtual Impacter
 concentrator.   This device separates the
 aerosol particles from the air by virtue of
 their inertia and directs them onto a  quartz
 wool  matrix.  The quartz wool  1s mounted
 Inside of the infrared pyrolyzer assembly.
 Periodically this assembly 1s heated to
                                                    22

-------
                               • 10 SAMPLER
CHEMICAL/BIOLOGICAL

MASS SPECTROMETER
                               Figure 4
temperatures  near 600  C.   As  a  result,  any
biological  material  collected on  the quartz
wool  is  pyrolyzed.   Although  the  focus  is on
biological  aerosols, any  aerosol  particle in
the applicable  size  range will  also be
collected  and analyzed in the same way.
This includes liquid or solid chemical
aerosols,  or  chemicals adsorbed on or
attached to other aerosol  particles of
network  or anthromorphic  origin.   These
pyrolysis  products are then  drawn into  a
heated 3 meter  long, 1mm  O.D. capillary
column and pulled to the  mass analyzer
chassis.   Any chemical vapors in  the air are
also drawn into this capillary  and pulled to
the mass analyzer.

    The  pyrolysis products and/or chemical
vapors enter  the mass  analyzer  by permeating
through  a  silicone membrane.  This membrane
separated  the high vacuum mass  analyzer from
the ambient pressure sample.  After the
sample enters the mass analyzer,  it is
ionized  using an electron gun and the mass
spectra  taken of the ionized  components.

    The  instrument control computer controls
the mass analyzer, the pyrolysis  event, and
all other  instrument related  functions
including  temperature  settings, electron gun
current, and  rf/dc  voltages and frequencies.
The data processing  computer interprets the
mass spectra  and generates the  necessary
system  responses.  The display, alarm and
communications  modules are the  primary
interfaces to the  operator.  A  block diagram
is shown 1n figure 5.
       A QUISTOR  (Quadrupole Ion Storage
   Device) mass analyzer  is used in the CBMS.
   (Figure 6)  This mass  analyzer consists of
   two end caps and a  ring electrode.  An ion
   getter pump or molecular drag pump can be
   used to produce the  required vacuum.  An
   electron gun is mounted on the sample inlet
   side.  Selected masses are either trapped
   within the QUISTOR  or  expelled out through
   the end caps depending on the voltages and
   frequencies applied  to the caps and ring.
   The masses of  the  ions that are expelled are
   directly correlated  to the voltages and
   frequencies applied  to the rings and caps.

       In principle,  a  mass analysis is made as
   follows.  First a  vapor sample enters the
   QUISTOR.  This sample  is then ionized using
   the electron gun.   The voltages and
   frequencies applied  to the rings and end
   caps cause these  ions  to become trapped
   within the QUISTOR's Internal electric
   fields.  The dc voltage applied to the
   QUISTOR 1s then changed at a controlled
   rate.  At specific  voltages, certain masses
   become unstable and  are expelled from the
   QUISTOR and are detected at the electron
   multiplier.  A plot is made of the signal
   from the electron  multiplier as a function
   of the applied voltage.  This voltage  is
   increased until all  ions are expelled.  The
   final mass record  is then obtained by
   correlating the applied and plotted voltage
   to the corresponding masses that should be
   expelled.
                                                    23

-------
                 Figure 5
                                            DISPLAY ALARM
                                                AND
                                           COMMUNICATIONS
            Figure  6.   QUISTOR Schematic
   RING
ELECTRODE
  IONIZATION
   REGION
       \ \
          V
                0
                                 ION STORAGE
                                    REGION       ELECTRON
                                               MULTIPLIER
                             24

-------
                          FIELD ANALYTICAL METHODS FOR SUPERFUND
                         Howard M.  Fribush,  Ph.D.  and Joan F.  Fisk
                           U.S. Environmental Protection Agency
                          Analytical Operations Branch   (OS-230)
                                 Washington,  D.C.  20460
Abstract

      The Analytical Operations Branch  (AOB)
of   the  U.S.   EPA   is   responsible   for
coordinating    field    analytical   methods
information transfer for Superfund.  With the
assistance  of  the  Environmental  Monitoring
Systems Laboratory in Las Vegas (EMSL-LV) , AOB
has initiated a series of projects designed to
facilitate  the  appropriate  use  of   field
analytical methods throughout Superfund.  This
paper  will  summarize   the  use  of   field
analytical methods  in  the various phases of
Superfund activities,  and will describe  AOB
efforts  in  coordinating   field   analytical
methods information transfer throughout  EPA.
In addition,  this paper  will  summarize  the
field  analytical   methods  currently   used
throughout   EPA's  Superfund   program   and
describe  the development  of   comprehensive
document that will  compile field  analytical
methods and  provide guidance  to  the use of
field  analytical  methods  for  environmental
samples.

Introduction

     Field analytical methods have been widely
used for the past eight-to-ten years by  EPA
organizations   under    various   Superfund
contracts, such as Field  Investigation  Teams
(FIT),  Technical  Assistance   Teams  (TAT),
Emergency   Response   Cleanup   Services
(ERGS),    and    Ranedial    Engineering
Management  (REM) contracts.   As efforts
to   streamline   the   Superfund   site
assessment,  site  characterization,  and
site clean-up processes  have developed,
the  need  to  assess  field  analytical
technologies for their appropriate use in
Superfund decision-making have increased.
The Analytical Operations Branch  (AOB) of
the Hazardous  Site Evaluation  Division
(HSED) has been involved in coordinating
information on  field analytical methods
used in support of Superfund.  The AOB's
first  efforts  at   coordinating  field
analytical methods resulted in  a  paper
entitled 'Field Monitoring Methods in Use
for Superfund Analyses'  (1) and the Field
Screening Methods Catalog (2).

     Field  analytical  methods are  used
throughout  the  Superfund  process.    In
EPA's Site Assessment,  or  Pre-remedial
Program, FIT teams under  the direction of
EPA  Site   Assessment   Managers  (SAMs)
analyze  samples in  the  field  for  Site
Inspections (SI).   The results of the SI
determine whether a site should be added
to the National Priorities List (NPL) of
Superfund  hazardous  waste   sites.    In
EPA's Removal Program, a TAT team, under
the   direction   of   an  EPA   On-Scene
Coordinator (OSC) , will conduct a Removal
                                                25

-------
Assessment,  often   using   field  analytical
methods, to determine if an emergency response
(a  removal  action)   is  necessary.   When  a
Removal Action  is initiated, an  ERGS cleanup
contractor,  under the direction of  the  OSC,
may  be dispatched  to the  site  for  further
analysis  and cleanup.    The result of  the
removal  action  is  typically a  short-term
stabilization of a site, and field analytical
methods are  often used to monitor  the extent
of the cleanup and determine when to stop the
removal action.   In  EPA's  Remedial  Program,
REM  contractors,  under  the  direction  of
Remedial   Project   Managers   (RPMs),   have
conducted  field analyses to characterize the
extent  of   contamination   at  a  site  (the
Remedial  Investigation),  to  test  remedial
treatment technologies (the Remedial Design),
and  for  site  cleanup  activities  (Remedial
Actions).   In all  of these programs, field
analytical methods are often used to identify
critical   samples   for   CLP   confirmatory
analyses.

     Field  analytical methods are typically
not as rigorous  as chemical analyses conducted
in a  "fixed" laboratory - a laboratory  in  a
permanent location.   Field  methods are often
used  for  screening   sites  to determine  if
contamination  is present,  and  to obtain  a
general idea of the  extent  of contamination.
Further,  field  analytical  methods are  most
useful when  the contaminants of  concern  have
already   been   identified,  so  that   the
appropriate  methods, dilutions,  calibration
ranges, etc., can be employed.   In addition,
field analytical methods are usually designed
to identify only a limited number of analytes.
Recently, however, more sophisticated and  more
rugged instrumentation have allowed  for  more
rigorous analyses in the field; consequently,
field analytical chemistry does not have to be
limited  to  screening.     Even  so,  it  is
generally believed that field analyses provide
less  precision  and   accuracy  than  analyses
conducted  in fixed laboratories.  (It should
be   noted,  however,  that   despite   this
perception,  a  focused  gas chromatographic
analysis is likely to be better than a heavily
quality-controlled QC/MS screen.)   In all of
the  Superfund  activities  described  in  the
previous  paragraph,  field analyses  are  used
for  the  rapid  turnaround of sample  results.
These results are,  in turn, used to expedite
site  assessments  for NPL  listings  or  for
emergency    removal    actions,    site
characterizations, and ultimate cleanup.  Data
quality   is  not  compromised,   since field
analyses are usually conducted in conjunction
with confirmatory analyses, such as GC/MS
or  ICP/MS  analyses  using EPA  Contract
Laboratory   Program   (CLP)  protocols.
Consequently,  field analyses are often
used  to   identify  samples  for  more
rigorous, CLP-type analyses.

Site Assessment Program

      As part  of determining whether a
site should be added to the NPL,  the Site
Inspection   (SI)  attempts   to  make  a
determination  of   "observed release".
This  determination  indicates  that the
site is discharging  contaminants into the
environment.

     The Site Assessment Program  conducts
up to ten percent of its  analyses  in the
field,  and  about  75 percent  of  the
samples are sent to  the CLP for full scan
analysis.     In  the  Site  Assessment
Program, very little  is usually known
about the site  and  its  contaminants;
consequently, it is  more cost effective
to use the CLP as a screen  rather than
conduct extensive field analyses  designed
for analyzing a limited number of  target
compounds.   Nevertheless,  FIT,  the Site
Assessment Program's primary contractors,
conduct  a   limited  number  of  field
analyses  to  obtain real-time  data  to
determine worker safety requirements, the
extent of contamination,  the presence or
absence   of   contamination,   for  the
placement of monitoring  wells, and  to
select   samples  for   subsequent  CLP
confirmatory analysis.

     To accomplish these  analyses, EPA's
Site Assessment Branch has developed the
Field Analytical Support  Project (FASP).
This  project  has,  at   this   writing,
developed 31 field analytical  methods,
called FASP Standard Operating Guidelines
(SOGs),  and are designed  to  be  modified
as   needed   to    meet    site-specific
conditions (3).  These rapid  turnaround,
FASP SOGs have been developed by FIT for
water,   soil,   or   oil   analyses  for
volatiles,    polynuclear   aromatic
hydrocarbons,   pesticides,   PCBs,  and
metals.

     Some EPA Regions  have used FASP to
perform  preliminary evaluations of new
instrumentation.     For   example,  two
Regions are evaluating Long Path Fourier
Transform  Infrared   (FTIR)  Spectroscopy
                                                 26

-------
for the analysis of air samples remote from a
site, and one Region has evaluated the Thermal
Chromatography/Mass   Spectrometry    (TC/MS)
system  for  the  analysis of  solid  samples.
According to these latter studies, TC/MS shows
promise as  a rapid screen for solid  samples
since there is minimal sample preparation.

Ranedial Program

      The purpose of the Remedial  Program is
to clean, or remediate, a site.  This process
can be rather complex, and usually consists of
a  Remedial Investigation  (RI)  phase,   a
Feasibility Study  (FS), a Record of  Decision
(ROD), a treatability study, a Remedial Design
(RD) phase, and a Remedial  Action (RA) phase.
The RI consists of data collection activities
undertaken to determine the degree and extent
of contamination  within all media.    The  RI
supports  the FS, which  determines the  risk
that the  site  poses  to human health  and  the
environment,   and   identifies    the  most
appropriate remedial alternatives that can be
used to remediate the site.  The ROD is issued
by EPA as the  final  remedial action  plan  for
a site.  If  necessary, a treatability study is
performed to determine the most  appropriate
conditions  for treatment, the  remedy is then
designed  (RD),  and the site is cleaned  (RA).

      During   all   of   these   phases,   the
potential   exists   for  the  use  of  field
analyses.   For  example, during the  three-
dimensional characterization of the extent of
contamination  (the RI),
rapid  turnaround of  sample results  may  be
necessary to focus subsequent analyses to the
determination of the extent of  contamination.
Here,  the analyses may be  used to  optimize
sampling  grids  for  three-dimensional  site
characterizations, to determine  the  location
of monitoring wells and well screen depths, or
to  determine  the  direction  and  speed  of
groundwater  plumes.    During  treatability
studies,  rapid  turnaround of data  may  be
necessary to avoid shutting down a treatment
operation to wait for sample results.  In the
Remedial  Design phase of  the  remediation,
rapid  turnaround of sample results may  be
necessary  to evaluate  the efficiency of  a
design.   These data may then be  used to make
improvements on  the design,  the net result
being  more  rapid  development  of  remedial
designs.   In  removal  and remedial  actions,
rapid  turnaround of  data may be required to
determine cleanup levels and to  minimize  the
costs associated with using expensive cleanup
equipment such as bulldozers.   When the field
analyses suggest that a regulatory level
has  been   reached,  CLP   confirmatory
analyses can then be performed to confirm
the cleanup level reached.

      To accomplish these analyses, EPA's
Hazardous Site Control Division developed
the  Close   Support  Laboratory   (CSL)
Program.  Because  site  remediations  are
often  very  complex  and typically take
several  years  to   complete,   the  REM
contractors found  it more  convenient to
construct   temporary,   "close-support"
laboratories at the site rather than  use
mobile    laboratories    or    portable
instruments    for    the     analytical
investigations.      This   program   has
resulted in the development of 15 field
analytical methods  for metals, volatiles,
semivolatiles, and  polynuclear aromatic
hydrocarbons  in water and  soil matrices
(4) .   In addition, the CSL program  has
developed  16  field protocols for  the
determination of physical measurements to
be used during treatability studies.

     The Remedial Program conducts about
ten percent of its  analyses in the field.
Once EPA has placed the site on the NPL,
Potentially  Responsible Parties  (PRPs)
are  finding  that   it   is  more  cost-
effective to assume  the  costs of site
characterizations.   Consequently,  there
are a growing number of these "PRP-Lead"
sites,  requiring  fewer  analyses by  the
EPA.   As a  result,  in many  Regions  the
Remedial Program is placing increasingly
more   resources   on   overseeing   the
analytical activities of the PRPs. This
shifting of  focus  from  "Superfund-Lead"
sites to PRP oversight has also coincided
with the phasing out of the REM contracts
and  phasing  in of the new  Alternative
Remedial   Contracts   Strategy   (ARCS)
contracts. Nevertheless, there are still
many   Superfund-Lead  remediations   in
progress, and the  Remedial Program  is
planning  to  use  ARCS  contractors   to
perform analyses in the field.

Removal Program

       In  addition  to  the   long-term
remedial actions,  Superfund legislation
provides for  short-term, removal actions.
Removals are performed in emergency-type
situations on unstable sites.  A removal
is  the cleanup or removal of released
hazardous substances which may present an
                                                27

-------
imminent    and    substantial    danger.
Consequently, removals may be necessary in the
event of a release of hazardous substances, or
to monitor, assess,  and evaluate  the threat of
release  of hazardous substances  to  prevent,
minimize,  or mitigate  damage to  human health
or the environment.
      Due  to the nature  of  these activities,
removals often  require a rapid turnaround of
analytical data; consequently, field analyses
are  used quite  often.   The Removal  Program
conducts about  30 percent of its analyses in
the  field.  Under the direction  of  the  OSC,
TAT - the Removal Program's primary technical
contractor - may use field analytical methods
for  purposes similar  to  those   of  the  FIT
teams.  If a more in-depth study is required,
the  OSC   may   require  the   use of field
analytical methods  to  determine  an estimated
extent of contamination.   If drums are present
and the contents within the drums are unknown,
TAT may use a Hazard Categorization field kit
to categorize the potential hazard associated
with the contents of the  drums.  TAT uses  this
field kit to perform simple qualitative tests
to  determine gross characteristics  of  the
waste -  the compound class, flash point  and
other properties, and consequently, determine
the disposal options for the waste.

  The Removal Program uses field  analyses for
Classic Emergencies  (for example,  for fires,
spills, train derailments, and explosions), to
determine  worker  safety  requirements,   for
designing   sampling   grids,   to   estimate
exposure,  for monitoring well  placement,  and
to  determine cleanup  levels.    Across   all
programs, the reasons for using field analyses
are  for  time savings, cost savings, and  to
identify  critical  samples  for  confirmatory
analyses.  Other reasons  include being able to
take more  samples,  ease of acquisition,  and
minimal paperwork requirements.

      To   accomplish   these   analyses,   the
Removal Program established the Environmental
Response   Team   (ERT).    The  ERT  provides
expertise   to   the   OSCs  in  the  area   of
performing field analyses and field analytical
methods development.  The ERT has developed a
number of field analytical methods, including
portable  gas chromatography  methods, x-ray
fluorescence methods for metals,  and  methods
for the screening and analysis of air samples
(5).
EMSL-LV
      The  Environmental  Monitoring  Systems
Laboratory   in   Las   Vegas   (EMSL-LV)
supports the  Superfund  field  analytical
programs   through  both  research   and
development and through technical support
to  the EPA  regions.    In the  Advanced
Field Monitoring Methods Program (AFMMP),
EMSL-LV  is  developing  and  validating
field  analytical   methods.      In   its
Technical   Support   Program,   EMSL-LV
dispatches  field  analytical   teams  to
hazardous     waste    sites    for
characterization studies.

      EMSL-LV   is   working  under   its
Advanced Field Monitoring Methods Program
(AFMMP)   in   coordination   with   the
Analytical  Operations Branch   (AOB)  to
identify, develop, and validate new and
existing  field analytical  methods  and
instrumentation.     In  addition,   the
objectives of AFMMP include the transfer
to and exchange of  information  with the
EPA  regions.    EMSL-LV  has   performed
studies involving immunochemical methods,
soil   gas  techniques,   portable   gas
chrcmatographs and associated  analytical
methods, X-ray fluorescence,  and  fiber
optic sensors.  In addition, EMSL-LV has
identified a number of new techniques for
study, including Fourier Transform Infra-
Red (FT-IR) , portable supercritical fluid
extractor  and solid  phase extraction,
field  test kits,  portable GC/MS,  ion
mobility spectrometers, and luminescence
methods.

Development   of   a   Superfund   Field
Analytical Methods Catalog

      The Analytical  Operations  Branch
(AOB) is the focal point for coordinating
field  analytical   method  information
transfer for Superfund.  In 1988, the AOB
coordinated an effort to compile some of
the  field analytical  methods  used  in
Superfund into a document entitled "Field
Screening Methods Catalog".

      The AOB is currently designing and
developing  a  comprehensive  compendium
that  will contain  many  of  the  field
analytical  methods  described  in  this
paper  for  use by all persons  involved
with  Superfund field  analyses.    This
compendium will contain developed field
analytical  methods,  it  will   contain
instrumentation    requirements,
requirements  for  quality assurance  and
quality   control,   analytical   method
                                                 28

-------
performance,   guidelines    for    effective
coinnunication, health and safety guidelines,
and evidentiary guidelines.  This  compendium
is being prepared with the assistance of  the
Field Analytical Methods Workgroup, which  had
its first meeting on July 19-20, 1990 and  the
Field  Analytical Methods  Management Forum.
The forum is a group of  EPA  Headquarters  and
Regional management representatives who met on
June  27-28  to determine Superfund  policies
regarding field analyses in Superfund.

      The field analytical methods that will
be a part of  the catalog will come from  the
sources described in this paper.  The methods
will be  presented in chapters  structured by
fraction,  analyte  group,  and  media.     In
addition, the methods  will be  restyled into
SW-846   format   for  consistency,   ease   of
reading,  and  to   allow  for   variations.
Instrumentation requirements will be provided
for each type of method based on available
information and research by EMSL-LV. Quality
Assurance and quality control information will
be designed to facilitate a  rapid  turnaround
of  data appropriate  for  the  generation of
field analytical data, and will be tiered to
allow a variation of requirements for quality.
The compendium will contain a user's guide and
will  stress "interactive  management"   -  the
communication between  the site manager,  the
field analyst, and the sampler.  In addition,
an   electronic   bulletin  board   will   be
established   to   house  the   methods   for
downloading, facilitate the quick transfer of
technology,  information, and ideas.   Health
and  safety  guidelines will  be  established
based on recent OSHA regulations, and evidence
guidelines for samples and analyses will also
be addressed.

References

1.  Fisk, Joan F. Field Monitoring Methods in
Use   for Superfund Analyses.     Pittsburgh
Conference.  February, 1987.

2.       Office   of  Emergency  and  Remedial
Response, Hazardous Site Evaluation Division.
Field Screening Methods Catalog. Users Guide.
EPA/540/2-88/005.   U.S. EPA.  Washington, DC.
  September,  1988.

3.     Site  Assessment  Branch.    U.S.E.P.A.
Hazardous  Site  Evaluation Division.   Field
Analytical Support  Project Standard Operating
Guidelines.   Unpublished.   Washington,  DC.
July, 1990.
4.   Hazardous  Site  Control Division.
Compilation of  CSL Analytical Methods.
Unpublished.  U.S. EPA.  Washington, DC.

5.  Environmental  Response Team.  Quality
Assurance Technical Information Bulletin,
Standard    Operating     Procedures.
Unpublished.  U.S. EPA.  Edison, NJ.
                                                 29

-------
       FIELD DELINEATION OF SOILS CONTAMINATION ON HAZARDOUS WASTE SITES
              REGULATED UNDER NEW JERSEY'S HAZARDOUS WASTE PROGRAM
                               Frederick W.  Cornell

               New  Jersey Department of Environmental Protection
                      Division of Hazardous  Site Mitigation
              Bureau of Environmental Evaluation and Risk Assessment
                         401 East State Street,  Floor 6W
                             Trenton, NJ  08625-0413
ABSTRACT

    The   New  Jersey  Hazardous  Waste
Management   Program   (HWMP)  recognizes
the  potential   for   field  analysis
techniques     to     expedite     site
delineation   while   decreasing   site
characterization    costs.    Although,
field     analysis    methods   produce
accurate, real-time  data at a low cost
per    sample,     the    absence    of
standardized data   quality  objectives
and method   specific quality assurance
and    quality      control     (QA/QC)
requirements has  prevented widespread
use of these  technologies.   The HWMP
has defined data   quality  objectives
for each  phase of  site  investigation,
and  outlined  QA/QC   procedures   for
several    widely    available    field
analysis   methods,   including   field
x-ray  fluorescence spectrometry, field
gas    chromatography,      colormetric
analysis,      and      photoionization
surveying.   The  development  of  these
method-specific     and    use-specific
procedures   has  allowed  the  HWMP  to
routinely recommend  the use  of field
analysis  methods   to   expedite  site
evaluation.
INTRODUCTION

    The    New   Jersey   Environmental
Cleanup   Responsibility .   Act   (ECRA)
program  requires industrial  facilities
that  handle   hazardous  materials   to
conduct  a site  evaluation  and  develop
a site remediation plan (if  necessary)
prior to any  real estate  transfer  or
cessation  of  industrial  operations.
Given the real estate  and stock market
activity  of  recent years  it  is  not
surprising that ECRA subject  sites  are
often  operational  facilities.    Since
ECRA's enactment  in 1984,  thousands of
sites  have  been  processed  by  ECRA.
For larger  industrial  facilities,  site
evaluation has proven  to  be costly  and
time   consuming,   frequently   taking
several years to complete.

    Site    characterization    efforts
typically  involve  a  historical  site
survey,  site  screening,  and  several
phases   of   site   delineation    (1).
Although,  initial  site  screening  is
usually    conducted    using    survey
instruments,  the  remaining delineation
phases  generally involve  collecting a
limited    number   of    samples   for
laboratory  analysis and evaluating the
lab results to determine  the need for
additional    sampling   phases.    This
typical  investigation  scheme  is  time
consuming,   requiring   months  between
phases  to  allow  for sample collection,
data    analysis,    delineation   plan
development,  and regulatory review and
interface;     however,    the   phased
approach    is   necessary   to   limit
analytical   costs.    The  unfortunate
result  of  phased investigation is  that
remedial    investigations    frequently
last   years   and  cost   hundreds  of
thousands  of dollars.

    These  delays  in  site remediation
may    not    only   render    industrial
operations   or    property   transfers
difficult   or impossible  to  conduct,
but    also   may   cause   unnecessary
contaminant migration  and exposure to
                                          31

-------
human or  environmental receptors.   In
these  situations  it   is  desirable  to
implement analytical  methods that  can
provide the necessary  data  in a  timely
and   cost-effective   manner.     Field
analysis  is  ideally suited  for  rapid,
cost-effective   site   characterization
as it can provide  real-time data which
is reliable  and inexpensive on a  per
sample basis.
FIELD SCREENING AND ANALYSIS METHODS

    To develop  the standard  operating
procedures  (SOPs)  included   in   this
paper, efforts were initially directed
at   determining   the   minimum   data
quality necessary to  make  appropriate
technical   and    regulatory   decisions
(this  is  described in  further  detail
below).   Subsequently,   a   literature
search was  used  to   identify  reliable
methods   from  the   vast  number   of
commercially  available   technologies.
Using    this    information,    method
specific SOPs were developed to detail
the   minimum  requirements   a   field
delineation plan  must meet  to  receive
agency   approval.    These   SOPs   are
designed  to  encourage  the  generation
of  consistent and  reliable  data  from
user  to  user  and site  to  site.   The
quality assurance and quality  control
(QA/QC) requirements  in  each  SOP  were
formulated  in  consistency   with   the
reliability,  accuracy,  and limitations
of   each   method   (particularly   when
considering    field     use),    while
considering  the  ultimate use  of  the
resulting data.

    Several   instruments  and  methods
have  been   evaluated  and  determined
effective   (or  potentially  effective)
at  detecting  site   contamination  at
milligram    per    kilogram    (mg/kg)
concentrations.    Although,   a  single
instrument   or  method   may  only  be
useful   for  analyzing   one   or   two
classes   of  compounds,  the  use  of
several  field  analysis  procedures in
tandem   enables   site   investigation
teams    to   detect    most   priority
pollutant    compounds   at   or    near
background     concentrations.      For
example,  ambient  temperature headspace
analysis   is   extremely  effective  in
analyzing  volatile organic  compounds,
but   not   polyaromatic  hydrocarbons
 (PAHs)  or metals.  Color-metric tests,
on  the  other hand,  are  effective at
analyzing       aromatic      compounds
 (including  the PAHs), and  a field XRF
will  detect  PCBs and  most  metals at
concentrations    as    low   as   20-100
milligrams  per  kilogram.   Thus,  by
using  several  field   instruments  or
methods  in  tandem a  broader suite of
contaminant  compounds  may  be   field
analyzed.  It should be noted that  the
methods  cited  in  this paper are  by no
means a  comprehensive list  of suitable
or    potentially     suitable     field
methodologies.  Initial selection  for
these  SOPs  was  based  on  instrument
availability,   amenability  to   field
use, and in-house experience.
DATA QUALITY OBJECTIVES

    The  New  Jersey   Hazardous   Waste
Management program  (HWMP)  data  quality
designations   are   based   on   those
developed  by  the EPA  (2-3) .   The  EPA
has  established  five  levels  of  data
quality  objectives   (DQOs).    Two  of
these,    Level   1    (Field    Survey
Instruments)   and   Level   2   (Field
Portable  Instruments), generate  real-
time,  field data.  Level  3 and  4  are
laboratory   methods   with   differing
QA/QC  requirements,  and  level  5  is
laboratory  special  services.   The  EPA
has  clearly  stated the  minimum  data
quality  level  required for each  stage
of   site   investigation.    Additional
explanation   of  these  data   quality
levels  may  be  found  in   any  of  the
EPA's  Data  Quality  Objectives manuals,
cited above.

    The  HWMP  data  quality  standards
have  been developed  to encourage  the
use   of   real-time  analysis   methods
during site characterization  (4).  The
HWMP  field  data DQOs are:   Level  1
(Field  Survey  Instruments),  Level  1A
(Field Analytical Methods),  and  Level
2    (Field   Portable   Instruments).
However,  unlike the EPA  designations,
minimum  QA/QC   and  support   documen-
tation  (deliverables)  requirements are
defined   to  assure   that   the   data
generated   by   these   methods  can  be
validated  based on  technical  criteria.
A  detailed  description   of   all  DQO
levels is provided below.

   Data  Quality Level   1  instrumen-
tation   are  intended  primarily  for
health   and  safety  or  initial  site
screening.     Quality   control    and
deliverable  requirements   are  limited
to   a   continuing   calibration   for
site-specific    compounds    and    the
reporting  of  values  on  field/boring
logs.    Level  one   (1)   methods  are
real-time   and   at   times,   erratic.
These  methods   can  be  described  as
pseudo-qualitative      and     pseudo-
                                           32

-------
quantitative   as  the  end   user  can
easily  be  led to  believe  that these
instruments    are    reporting   "true
values"  or providing selectivity, when
indeed they  are not.  For example, the
photoionization detector  (PID)  survey
instrument is  commonly thought  to be
selective  and not sensitive to  species
whose  ionization potentials  (IPs)  are
higher   than  that   of  the  internal
ionization    lamp.     In    practice,
however,   species  with  IPs  above  the
lamp  energy are  routinely  detected by
PID survey instruments.   With  respect
to   quant itat ion,    a   PID   survey
instrument  reports   a  value   often
expressed   in  mg/kg;  however,  since
detector  response  is  highly variable
among  chemical  species  this reported
value    may    not   represent    site
conditions  or  correlate  with  other
site  data.  For  these  reasons level  1
data   should  generally  be  used  to
indicate   contaminant    presence   or
absence,  rather than compound identity
or    total     concentration.     The
application   of  level  1  data  should
therefore  be   limited  to  health  and
safety   screening  or  to   guide  the
placement  of samples being analyzed by
higher     DQO    methods.      Level     1
instruments    include    field   x-ray
fluorescence  spectrometers  (XRF)  with
a  remote  probe    and   PID  survey
instruments.

   Data  Quality   Level   1A   methods
produce  fairly precise data;  however,
a reduced quality  control  program is
employed  to   allow  high   frequency,
low-cost  sampling.   Level  1A  methods
are suitable  for  site  screening and
site  delineation  when  proper QA/QC
practices     are    employed.     When
delineating  using   level  1A methods,
minimum     deliverable    requirements
typically  include:   calibration data
for  site-specific   compounds,   check
standards   data,    a  non-conformance
summary,   a   certification   statement
signed    by    the    analyst,    sample
calculations,   isopleth  maps,   tables
indicating     results      (raw     and
"corrected"  based on lab confirmation
data),   and   chain-of-custody documen-
tation.   In  addition,  lab confirmation
data  (10-30% of all  samples  collected)
must  provide  "calibration"  throughout
the   entire    analysis    range   and
confirmation   of  the   "clean"   zone.
Level   1A  methods   include   headspace
analysis  of  volatile  compounds  and
analysis using colormetric techniques.

   Data  Quality   Level   2   methods
produce   precise  data  when required
QA/QC    procedures    are    employed.
Quality  assurance and quality control
requirements  are  sufficient  to  allow
rigorous  data   interpretation,   while
providing  reasonable  field  operation
requirements.    Level  2   methods  are
ideally  suited  for low-cost,  one phase
delineation.     Minimum    deliverables
requirements     will     include:     an
instrument  log,  calibration  data  for
site   specific   compounds,   standards
data,  split  sample data,  raw  sample
data,  blank  data,   a   certification
statement  signed  by  the  analyst,  a
non-conformance     summary,     sample
calculations,   isopleth   maps,  tables
indicating     results      (raw     and
"corrected"  based on  lab confirmation
data), and  custody documentation.  Lab
confirmation   data    (5-15%   of   all
samples    collected)    must    provide
"calibration"   throughout  the  entire
analysis range  and confirmation of the
"clean"  zone.   Level  2 methods include
field  gas   chromatography   (GC)   and
field    XRF     analysis     using    a
silicon-lithium  detector.

    Data  Quality  Levels  3  and  4  are
"Standard  Lab  Methods"  with  varying
deliverable    requirements.    Methods
which provide  these data  qualities may
be   used    for    conventional    site
characterization  activities   or   to
confirm   field   instrument   results
obtained    during   site    delineation
activities.   It  should  be  noted that
the specific  QA/QC procedures required
will  be  dictated  by  the  applicable
regulatory   program.      Data   quality
level  3  methodologies include  SW-846
(5)  methods and NJ ECRA Deliverables
(1).   Data  quality  level  4  methods
include  CLP methods and  Scope of Work
(SOW) requirements  (6).

    Data Quality  Level  5  methods  are
generally   state-of-the-art   or   non-
approved  methods  chosen  specifically
for  a   particular   site.    Level   5
methods  are  required  when "Standards
Lab Methods"  are either  unavailable or
impractical.    Level   5   data  may  be
accepted to confirm  field  results  or
define a "clean  zone".

    The  goal  of any site  investigation
is  to   assure   that   the  information
obtained  is sufficient  to  select  and
design    an    appropriate    remedial
technology.   Ideally,  site character-
ization     will    provide    complete
definition    of   contamination   with
respect  to  both concentration  trends
and  actual  contaminant   load.    The
advantages  of   levels  1,  1A,  and  2
                                         33

-------
analysis  are  rapid  site  delineation
and low per sample  costs  allowing high
frequency   sampling   and   a   rapid
estimation of  concentration gradients;
however,   the   concentration   results
must be  assumed to have  up to a 150%
error.   Level  3 and 4  analysis  methods
are  not  real-time   and   are  more
expensive,      limiting       sampling
frequency, but reported results can  be
assumed  to  be quite  accurate  and  a
good indicator of  actual  contamination
present.  In summary,  the trade-off  is
rapid,      less     expensive     site
characterization  verses  data  quality
and accuracy.

    At  first  glance  it  may appear  as
if  HWMP has  chosen  to   expedite site
characterization  at   the  expense   of
data quality by encouraging the use  of
level  1A and  level  2 methods.  Upon
closer  examination,  it  can  be  seen
that although  the  raw  data obtained  by
field  instruments  are  less  accurate
and  less  precise,  the  data  set   is
highly   consistent   within   itself,
clearly    indicating     trends    and
contamination   zones.    Also,   since
field analysis costs  are  generally  per
diem  rather  than  per  sample,  field
samples may  be collected  at a  greater
frequency, providing  the  project  team
with better  site definition and  fewer
data gaps.   Lastly, all  field data  are
supported     by     an     independent
calibration    or    correction   factor
provided    by   the    required    lab
confirmation samples,  discussed above.
Thus,  the  end  product  generated  is
actually  a  hybrid of field  analysis
data   and   lab   data   which,   when
combined,  may not  only  be equivalent
in  data  quality  to  that  obtained  by
standard  methods,  but   may  actually
provide  a more  reliable  and complete
characterization of site  conditions.
SITE INVESTIGATION STRATEGY

    The  newly developed  HWMP DQOs use
a  combination of high  and low quality
data  to produce  a  data  set  which is
moderate in both  quality and quantity.
These  DQOs   rely  on  the  ability  of
users  to calibrate  field analysis data
to   laboratory  confirmation  samples,
providing   superior   site   character-
ization  at  a reduced  cost.   The net
effect    is     that     most    site
investigations may  be completed  in a
maximum of 1-2 phases or less than one
(1)  year.    To  accomplish   this,  the
following  site investigatory procedure
is   recommended  (where  site  contam-
ination is  known,  step 1A  may not be
required).

    1.   Obtain historical  information
         (i.e.   past  or  present   site
         activities).

    1A.  If  the  contamination source
         is    unknown,    a    sampling
         program   incorporating   site
         screening tools  (level 1)  and
         laboratory   sample   analysis
         (level    3/4)     should    be
         implemented.    The  goal  of
         this  effort  is   to  identify
         all  contaminants  present by
         documenting  worst-case   site
         conditions.

    2.  The  information  above should
        then  be  used  to  develop an
        open     ended,      contaminant
        delineation   plan,   including
        the  use  of  real-time  (Level
        1A/2   quality  data)   methods.
        The  plan   should  incorporate
        sampling    contingencies    to
        assure   site    delineation  is
        completed  during  this sampling
        phase.   To provide additional
        data     reliability,     field
        instruments      should      be
        calibrated    to   site-specific
        compounds   of    interest   as
        defined  by previously obtained
        information.

    3.  Upon receipt  of the laboratory
        confirmation  data,  the   need
        for a  revised delineation plan
        should   be    assessed.     If
        required,   a  phase II  delin-
        eation plan should incorporate
        field    analysis    methods  to
        complete site delineation.

    4.  The complete  database  should
        then be used  to develop a  site
        remediation plan.   If  in  situ
        remedial  measures   are  to  be
        used  and  system  design limits
        are   being   approached,   an
        increased     percentage     of
        laboratory   data    may    be
        required.
 DEVELOPMENT OF FIELD SOPS

     The development  of  field  SOPs  is
 considered the most efficient means  of
 assuring that data collected from  site
 to  site  is  consistent.   These  SOPs
 were   developed   by   consulting   the
 literature,  instrument  manufacturers,
 and  personnel  with   extensive   field
                                           34

-------
and/or  instrumental  experience.   Each
SOP has  5  technical   sections,  i.e.
method  overview,  method  requirements
(including     QA/QC     requirements),
interferences   and  limitations,  data
interpretation  and  reporting require-
ments,    and    health    and   safety
considerations.

   The   method  overview  or  general
guidance   section    is    intended   to
provide   the   reader  with   a  basic
understanding   of  the  method.   This
section   details  method  applications,
including    applicable     matrices,
detectable   compounds,    and  minimum
detection limits   (MDLs).    Additional
information  is  provided for use by the
project   manager,   including  estimated
cost  per sample,  level  of training
required   to   effectively   use   the
method,   lab method  equivalent,  and
theory    of   operation.   The   theory
section   contains   instrumental  and/or
chemical      details      aimed     at
familiarizing   the   reader   with  the
actual science  of operation.  The last
section   of  each SOP also  provides  a
list      of     references    directing
interested  readers to  a more detailed
explanation  of  instrumental  theory and
use.

   The   method  requirements   section
provides  four  types  of  information:
sampling     considerations,     sampling
requirements,  field operation  require-
ments,    and    QA/QC     requirements.
Sampling     considerations      include
general   information  applicable when  a
sampling  program  is  being  developed.
This  section   provides  guidance  with
respect  to sample frequency, selection
of lab  confirmation samples,  and  any
other    useful    information   gained
through  field experience.  As would  be
expected, this section is  continually
evolving  as the experience base grows.
The  sampling    requirements   section
details    proper   sample   collection
procedures    when    standard    field
sampling   methods   (7)   are   inappro-
priate.   This  section  also  includes
sample handling requirements when past
experience      has     shown     sample
preparation  to  significantly   impact
final results,  as is the  case  with  XRF
analysis.  The  field operation  section
contains     actual    method    guidance
intended   to   supplement  or   replace
manufacturer's  recommendations.  This
guidance  customizes  method  procedures
in an effort to  meet the goals of  the
HWMP   regulatory  program.   The   last
section,   QA/QC,   states  all   quality
assurance recommendations and  require-
ments.    The    requirements   include
analyst "competence"  tests,  submission
of   all   raw    data,    and   support
documentation.

    The  interferences and  limitations
section  discusses  problems  which  may
be   encountered   during  field   use.
These   comments    are    intended   to
supplement   manufacturer's   recommen-
dations   by    highlighting   problems
encountered   during    previous   site
operations.   It  is  likely  that  this
section will be in constant transition
until  a  comprehensive   database  has
been established.

    The    data    interpretation    and
submission     requirements     section
details  data  manipulation  procedures
and   regulatory   submission  require-
ments.   Data   interpretation  require-
ments  vary by  method  and  DQO  level;
however,    all    SOPs    require    the
calculation  of  "corrected"  results,
accounting  for  discrepancies  between
laboratory  and field data.   Reporting
requirements  are standardized  for all
field  methods   and  include:   scaled
site  maps with plotted  data,  summary
tables  indicating  all  field  results
(raw  and corrected)  and  lab reported
values,   a  calibration  plot  of  lab
split  sample data verses  field  data,
and   quality    assurance  and  quality
control  documentation (consistent with
the  QA/QC requirements  stated above).
These  requirements  are  intended  to
expedite  the required   review  time by
standardizing   report   contents   and
format,  while  facilitating validation
of both lab and field data.
CURRENT AND  PENDING SOPs

    Standard operating procedures have
been    completed    for    four   field
instruments   and  two  field  analysis
methodologies.   Additionally,  several
other  SOPs  are under  development.   A
listing of all  SOPs is provided below.

Level  1 Data Quality

    Field Screening  Using  a   Photo-
     ionization Survey Instrument.
    Field Screening  Using  an  X-ray
     Fluorescence          Spectrometer
     Equipped with a Remote Probe.
    *Field  Screening  Using  a  Flame
     lonization Survey Instrument.
    *Field Screening Using  a Portable
     Infrared Instrument.
                                          35

-------
Level 1A Data Quality

    Field    Delineation    Using     a
     Colonnetric Test Kit.
    Field  Delineation  Using   Ambient
     Temperature Headspace Analysis.
   *Field Delineation Using a  Portable
     Infrared Instrument.
   *Field Delineation Using a  Portable
     Ultraviolet Spectrometer.
Level 2 Data Quality
    Field   Delineation
     Fluorescence.
    Field Analysis Using  a
     Chromatograph.
     Attachments:  1
                   2
                   3
                   4
                   *5
    Using   X-ray
       Field  Gas
                  *6,
                  *7,
PID Detector.
FID Detector.
AID Detector.
ECD Detector.
Analyzing
 Extractables
 (BNs/PCBs).
Analyzing  Water
Samples.
Analyzing  Air  or
Headspace
Samples.
    * - under development
FUTURE DIRECTIONS

    Currently,     the    field    SOPs
described  above are  in  widespread use
throughout  the  HWMP  program.   Since
these  instruments  and  methods are  a
small    subset   of    all   currently
available   field   analysis   methods,
similar  SOPs  will  be  developed  for
several  additional methods,  including
FID    survey    instruments,    several
spectrometers,   and  additional  field
gas chromatography  applications.

    The  performance  of  each  of  these
methods  (on NJ regulated  sites)  will
be   monitored   using   an    in-house
database.      Upon    collection    of
sufficient   data,   the  SOPs  will  be
revised    as   appropriate.    It   is
expected   that   additional    field
experience     and    the    associated
understanding   of  method  limitations
and accuracy will lead to wider use of
field  analysis methods,   making  site
evaluation  a  much  less time-consuming
and costly  process.
5.

6.
          REFERENCES

New     Jersey    Department     of
Environmental  Protection  (NJDEP) ,
Division    of   Hazardous    Waste
Management   (DHWM).   March   1990.
Remedial Investigation Guide.

U.S.   EPA.    March   1987.    Data
Quality  Objectives   for  Remedial
Response                Activities.
EPA/540/G-87/003,  EPA/540/G-87/004
and OSWER Directive 9335.0-7A&B.

U.S. EPA.   October  1988.  Guidance
for       Conducting       Remedial
Investigations    and    Feasibility
Studies        Under        CERCLA.
EPA/540/G-89/004      and     OSWER
Directive 9355.3-01.

NJDEP,  Division of Hazardous Site
Mitigation      (DHSM),     Standard
Operating     Procedures:     Field
Delineation  Series  (4.25), 1990.

U.S. EPA.  SW-846, Third Edition.

             CLP-IFB,   most  recent
                                                 U.S.   EPA.
                                                 version.
                           NJDEP/DHWM.  February  1988.  Field
                           Sampling Procedures Manual.
                                    BIBLIOGRAPHY

                       General Field Sampling and Analysis

                       Myers,  J.C.  "Converging  Technologies",
                       Hazmat World, June 1989,   p24-27.

                       U.S.    EPA.    Environmental  Response
                       Team.     1989.    Standard    Operating
                       Procedure:    Photoionization  Detector
                       (12056).

                       Siegrist,     R.L.;    Jenssen,     P.O.
                       "Evaluation of Sampling  Method  Effects
                       on     Volatile    Organic    Compound
                       Measurements  in  Contaminated  Soils",
                       Environmental  Science and  Technology,
                       1990,  24,  1387-1392.
                                                1988.    Field
                                                      Catalog.
U.S.   EPA.    September
Screening        Methods
EPA/540/2-88/0055.

Keith,  L.H.  "Environmental  Sampling:
A  Summary",  Environmental  Science  and
Technology, 1990, 24, p610-617.

Gretsky,  P.;   Barbour,  R.;  Asimenios,
G.S.  Pollution  Engineering,  June 1990,
p!02-108.
                                           36

-------
Organic  Survey  Instruments

Nyguist,  J.E.;  Wilson, D.L.  "Decreased
Sensitivity      of     Photoionization
Detector Total  Organic Vapor Detectors
in the  Presence of  Methane",  Journal
of  the   American  Industrial  Hygiene
Association,  1990,  51(6),  p326-330.

Gervasio,  R. ;  Davis,  N.O.  "Monitoring
in  Reduced   Oxygen  Atmosphere   Using
Portable   Survey     Direct   Reading
Instruments      (PID     and     FID)",
Proceedings HMCRI,  1989-90.

Tillman,  N.;  Ranlet,  K.;  Meyer, T.J.
"Soil  Gas Surveys:   Part  I", Pollution
Engineering,  July 1989,  p86-89.

Tillman,  N.;  Ranlet,  K. ;  Meyer, T.J.
"Soil    Gas     Surveys:     Part   II",
Pollution Engineering,   August   1989,
p79-84.
Headspace Analysis

Holbrook, Tim "Hydrocarbon Vapor  Plume
Definition  Using  Ambient  Temperature
Headspace   Analysis",  Proceedings   of
the NWWA/API  Conference  on  Petroleum
Hydrocarbons  and Organic  Chemicals  in
Ground Water -  Prevention, Detection,
and Restoration, November,  1987.

Roe,  V.D.;  Lacy,   M.J.;  Stuart,  J.D.
"Manual  Headspace  Method  to  Analyze
for the Volatile Aromatics of Gasoline
in  Groundwater  and  Soil  Samples",
Analytical    Chemistry,    1989,     61,
P2584-5.
Colormetric Analysis

Roberts,     R.M.,     Khalaf,     A.A.,
Friedel-Crafts  Alkylation Chemistry;  A
Century  of  Discovery,  Macel   Dekker,
Inc.,  New  York,  1984.

Rohriner,  R.L.,  Fuson,  R.C.  et  al.,
The   Systematic   Identification   of
Organic   Compounds,   John   Wiley  and
Sons,  New  York,  1980.

Hanby,  J.D.,  "A  New  Method  for the
Determination    and   Measurement   of
Aromatic  Compounds  in Water",   Written
Communication,      Hanby      Analytical
Laboratories,   Inc.,  Houston,   Texas,
1989.
X-ray Fluorescence

Piorek, Stanislaw "XRF Technique  as a
Method of  Choice for  On-site Analysis
of   Soil    Contaminants    And   Waste
Material",   Proceedings   38th  Annual
Conference  on  Applications  of  X-ray
Analysis, Denver, Vol. 33,  1988.

Grupp,  D.J.;   Everitt,   D.A.;   Beth,
R.J.;    Spear,     R.    "Use    of    a
Transportable   XRF  Spectrometer   for
On-Site Analysis  of Hg in Soils",  AFI,
November 1989, p33-40.

J.R.   Rhodes,    J.A.    Stout,    J.S.
Schindler,  and  Piorek  "Portable  X-Ray
Survey  Meters    for   In    Situ   Trace
Element      Monitoring     of      Air
Particulate",   American   Society   for
Testing    and    Materials,    Special
Technical    Publication    786,    1982,
p70-82.

Piorek,  S.;  Rhodes,   J.R.  "Hazardous
Waste Screening Using  A Portable X-ray
Analyzer",  Presented  at  the  Symposium
on   ' Waste     Minimization      and
Environmental  Programs within  D.O.D.,
American  Defense Preparedness  Assoc.,
April 1987.

Piorek,   S.;  Rhodes,  J.R.    "A   New
Calibration    Technique    for    X-ray
Analyzers   Used   in   Hazardous   Waste
Screening",  Proceedings   5th  National
RCRA/Superfund Conference,  April 1988.

Piorek,  S.;  Rhodes,   J.R.  "In  Situ
Analysis of  Waste Water Using Portable
Preconcentration   Techniques   and   a
Portable  XRF Analyzer",   Presented  at
Electron     Microscopy    and    X-ray
Applications   to   Environmental   and
Occupational      Health       Analysis
Symposium,      Pennsylvania      State
University, October 1980.

Barish, J.J.; Jones,  R.R.; Raab,  G.A.;
Pasmore,    J.R.   "The  Application  of
X-ray  Fluorescence Technology in  the
Creation  of  Site  Comparison  Samples
and  in  the  Design  of Hazardous  Waste
Treatability      Studies",      First
International     Symposium:      Field
Screening  Methods for Hazardous  Waste
Site Investigations, October  1988.

Piorek, S.  "XRF Technique  as a Method
of Choice  for On-Site  Analysis of Soil
Contaminants and  Waste Material",  38th
Annual Denver X-Ray Conference, 1989.
Watson,  W.;  Walsh,  J.P.;  Glynn,  B.
"On-Site       X-Ray       Fluorescence
Spectrometry    Mapping     of    Metal
Contaminants  in  Soils   at  Superfund
                                          37

-------
Sites",   American   Laboratory,    July
1989, p60-68.

Freiburg,  C.;   Molepo,   J.M.;  Sansoni
"Comparative Determination  of Lead  in
Soils  by  X-Ray  Fluorescence,   Atomic
Absorption  Spectrometry,   and   Atomic
Emission  Spectrometry",  Fresenius   Z
Anal Chem, 1987, 327, p304-308.

Smith, G.H.; Lloyd,  O.L. "Patterns  of
Metals    Pollution   In    Soils:    A
Comparison  of  the Values  Obtained  By
Atomic   Absorption   Spectrophotometry
and X-Ray  Fluorescence", Environmental
Toxicology and  Chemistry, 1986,  Vol.5,
P117-127.

Jenkins,     R    "X-Ray   Fluorescence
Analysis", Analytical Chemistry,  1984,
56(9), p!099A.
Field Gas Chromatography

U.S.    EPA.      Standard     Operating
Procedure:   Sentex   Scentograph   G.C.
Field Use (SOP #1702), December 1988.

U.S.    EPA.      Standard     Operating
Procedure:  Photovac  10S50,  10S55,  and
10S70  Gas   Chromatography   Operation
(SOP #2108), January 1989.

Wylie, Philip, L.  "Comparing Headspace
with  Purge  and  Trap  for  Analysis  of
Volatile     Priority      Pollutants",
Research  &   Technology,   1988,   80(8),
p65-72.
                                          38

-------
   TABLE I.  NJDEP/HWMP DATA QUALITY CLASSIFICATIONS
DATA
QUALITY
LEVEL
PURPOSE
  OF
SAMPLE
EXAMPLE METHODS
      OR
  INSTRUMENTS
        Health & Safety

        Site Screening

        Field Use when
         excavating.
                   Portable PID (HNU).
                   Colormetric Analysis.
                   Portable FID (OVA).
                   XRF with a remote probe
                    (X-met).
1A      Site Screening.

        Field Use when
         excavating.

        Site Delineation.
                   ATH Analysis.
                   Colormetric Analysis.
        Field use when
         excavating.

        Site Delineation.
                   Portable GC.
                   Portable XRF with SiLi
                    detector.
                   Mobile Lab  (limited QA/QC)
        Site Delineation.

        Lab Confirmation of
          field delineation
          samples.

        Traditional Site
          Characterization.
                   Laboratory Analyzed
                    Samples, without
                    QA/QC documentation,
                    i.e. 600 Series.
                   Mobile Lab.
        Traditional  Site
          Characterization.

        Lab  Confirmation of
          field  delineation
          samples.
                   Laboratory Analyzed
                    Samples, with full
                    QA/QC documentation,
                    i.e. CLP-IFB.
         Traditional  Site
          Characterization.

         Lab  Confirmation of
          field delineation
          samples.
                   Laboratory Special
                    Services.
                   Mobile  Lab.
                            39

-------
                                           PLENARY SESSION DISCUSSION
LLEWELLYN WILLIAMS: There was a reference in the first or second paper
to our concern about the acceptance of field screening and field analytical data
by a regulatory group. How do we deal with, or how do we encourage the
acceptance of field screening and field analytical method data in the regulatory'
arena?

DENNIS WYNNE: Pan of it. I think, is encouraging the risk-taking among our
managers. What we have dealt with in the past is a tendency to rely almost
exclusively on the tried and true methods of the contract lab program (CLP).
What we are trying to  do in the Superfund Program is to wean people off
inordinate use of the CLP by saying that that is for a specific intended purpose.
Il's not intended for all uses. If you focus on thedataquality objectives approach,
there's not a need lo over rely on the CLP. because you often have for some uses
a gold plated version that isn't needed for some of the basic uses. The field
screening methods would be more appropriate. Some of the ways to do that is by
the work groupapproach and trying things toencourage managers to use it. We're
trying  to focus on things like  streamlining  the remedial investigations and
feasibility studies, and you really can't tell what you find if you're using fixed
labs exclusively. There's downtime while the data are being sent out. analyzed
and rev icwed. In some cases I think what we're trying to do is look where the time
being spent on the program. Trying to shorten those times where we can. trying
to encourage the user community to come together in work groups, being able to
provide guidance through training programs are ways we can gel more people
familiar with field screening and thereby limiting some of the conservativism
that we deal with in some of the managers who tend to rely on contract lab
programs.

Another pan of it. I think, would be toemphasize that some of the field analytic
methods can provide you with as much accuracy as you get through fixed labs.
We need to emphasize  those points, so people aren't always assuming field
methods are sort of the poor cousin of the fixed labs.

1 guess from a PRPperspective. which the Army is. our approach has been to use
field screening as a powerful tool to guide the traditional quality control in lab
data. We've been forced that way because of our negotiations with the regions.
because of the requirement for a lot of this data to eventually stand up in court.
So we see it as minimizing the requirement for that extreme case of chain of
custody and total reliability of the data because of extreme quality control. We'll
minimize the number of samples we really have to take, because of this powerful
tool, the field screen.
HOWARD FRIBt'SH: 1 think that your continued use of field analytical
methods and analyses is going to force it to be accepted, for one thing. Another
thine, the acceptance seems to be more fragmented. That is. it seems to be more
accepted to say in  the Removal Program, and less accepted  but somewhat
accepted in other programs. I think that without field analytical methods and
analyses, we're really sampling blind, and there is no reason what 90^ of the
samples that get sent to the CLP should be nonhits. when 909t could be hits.

Another way is to document all this just like we're trying to do with the catalog
and the user's guides. I  think it will be accepted much more than it is now.

N ABIL YACOUB: My question has two pans: 1) would that manual encompass
methods developed by the Army and other entities? 2) w ould the methods incl ude
those for matrices other than w ater. because in the real w orld, y ou have a problem
with soils and sludges and such.

HOWARD FRIBUSH: Yes. it will include other matrices. As long as the
methods have been used for Superfund activities, and they have been shown to
work, and they've been field tested, there is no reason why they can't be included.
That's why channeling performance information one of these methods, back to
EMSL-Las Vegas for an ongoing evaluation is so important. In  the  future
updates, we can either delete some of the methods, or it might help us to combine
some of the methods. And as far as your first question: 1 would say that we're
definitely open to including methods developed by the Army, especially if
they're users  in Superfund activities, for example in the Federal Facilities
Program. If there is performance information, we would like to know that.

MICHAEL C ARR ABB A: I have more of a comment or a suggestion directed
at both the  Environmental Protection Agency, as well as the Department of
Energy.

If you look at the Chemical Sensors session, there are six talks: two representa-
tives from the Federal Government, and four representatives from small busi-
ness. My comment is that the  Environmental Protection Agency, as well the
Department of Energy, is grossly underutilizing the small business innovative
research program to bring forth  some of these field screening technologies, such
as in the area of chemical sensors or optical spectroscopy. If you look at the
current solicitations for 1991 for the Department of Energy, we've been hearing
about this great problem in environmental restoration and field screening. There
arc no topics in there for small  business, and a lot of the innovation that we're
going to need in the future, particularly for the DOE and EPA, is going to come
from small business with new and innovative ideas. This is not the case for the
Department of Defense, who is actually doing a pretty good job in using the SBIR
program lo fulfill these needs.

EDGAR SHL'LMAN: I noticed in the user's guide that is presently out, that
there is a heavy emphasis on  fieldable methods, and very little on the man-
portable type of instruments or methodology. Could you comment on what the
future direction is relative to the man-portable type of instruments for  field
screening? And also perhaps to other panelists in terms of their judgment as to
the value of smaller dev ices for field screening?

HOWARD FRIBUSH: 1 think that the catalog in the user's guide is intended to
be comprehensive, and there is no reason that the smaller survey instruments.
such as organic vapor analyzers, or portable radionuclide analyzers couldn't be
included. In fact, since they are used a lot, especially in the Site Assessment
Program, and the Removal Program, they should be  included and will  be
included.

Up a stage to the man portable instruments, we now have portable GC/MS. Those
certainly will  be included. I think the short answer  to your question is that we
w ant everything that is used in Superfund typically to be included into the catalog
and user's guide.

EDGAR SHULMAN: I guess I was looking toward your judgment in terms of
the v alue of small devices. Would the priority in the future be toward encouraging
people to actually get much smaller devices? I know you are talking about man-
portable GC/MS.  but  they really are not  man portable right now. They're
fieldable. you still need a truck or something similar.

HOWARD FRIBUSH: Are you talking about field kits, or are you talking about
survey instruments?

EDGAR SHULMAN: I'm talking about survey instruments, try-ing to encour-
age their research and  development community, in terms of an agenda for
research. Maybe that's what I'm looking for. Where  should the priorities be put,
from the R&D community, relative to the kinds of methods that are envisioned
for the future.

LARRY REED: You've made a good point. I think looking at the present catalog
we have out. There is a bias that was introduced when we were gathering existing
information, a large pile which was available as part of the Field Analytical
Support Program. This program was developed in  part for field investigation
teams, and the Site Assessment Program nationwide. There had been a focus to
look at the bigger equipment and the more refined type of equipment. I think that
was done just to get the catalog out, what information is in use — was available
and useful.  I think what we are going to try and do is balance it now by more of
the technologies, try to focus more on portable kinds of instruments, also. I know
                                                                         40

-------
                                                        DISCUSSION
in particular when I'm looking at the future of the Site Assessment Program, as
the field investigation team contracts start to expire this year, we're going to be
looking at two phases of the shifting of the equipment, the larger field analytic
support equipment, and then the portable equipment. We want to make sure that
that equipment will be transferred to the people who are going to be doing the Site
Assessment work and in looking at the next generation of it. That's a good point
you make. I think we'll try to balance it out.

HOWARD FRIBUSH: I just wanted to say that the survey instruments have a
definite use in Superfund. They are used quite often to determine the health and
safety requirements for workers, and also to identify hot spots. So since they have
a definite use in Superfund, they will be included in the next update.

LLEWELLYN WILLIAMS: I was just reminded thai the EPA is not 100%
Superfund. There are a fair number of other programs out there for which field
screening technologies will have a place, and in many of those applications, I
think some truly portable measurement instruments are going to be very, very
important.

CHRIS LIEBMAN: I though that the key to the compendium and the success
of the compendium was really dependent on people submitting their methods to
the working group, so that we can see that they are included in the compendium.
I think it's important to point out that if survey instruments are not  currently in
the compendium, that largely reflects the fact  that people had not submitted
methods. I think if you are unhappy with what is in the compendium, to change
that, make your submissions.

DOUG PEERY: In putting the catalog together, of course you're addressing
purely programs that the EPA is addressing, in dealing with private clients who
rely on these things for their own information, you get locked in. We  also have
to respond to that. Is there going to be a flexibility in this catalog whereby we.
as the person developing the procedure, can go through steps and prove that the
procedures are applicable and usable, and not be locked in or having to reply.
Maybe taking the USATHAMA Procedure and Methodology Proof  Program,
making it simpler, and integrating the two, so that it can be done very quickly and
easily and economically would be one way. Is there a procedure or  a thought to
adding something along that line?

HOWARD FRIBUSH: I think that is a really good  idea, and a  really good
statement. This is something that the work group has not yet addressed. I think
that is a good topic  for a future item at our next work group.
Originally, we had talked about EMSL-Las Vegas doing some of that validation.
When we look at all the methods that  we have, I think  it might be more
appropriate to have EMSL-Las Vegas look at the performance information. But
for new methods, I think that that is an area for future consideration and I
appreciate the comment.

COLLEEN PETULLO: I notice that DOD, DOE and EPA are all developing
innovative technologies, or supporting innovative technologies to be developed.
We all march to a different drummer in terms of QA. How is that all being
coordinated?

LLEWELLYN WILLIAMS: There are a number of ways in which attempts are
being made to  harmonize Quality Assurance, not the least of which is the
interagency ad hoc committee on QA for environmental measurements that's just
been established. We are looking very hard at both QC and QA requirements,
both  from a process standpoint and from an operations standpoint, to see if we
can get more uniform application of QA/QC procedures, agency wide, as well as
across the agencies. We're well aware that there have been concerns  in the past
with  respect to dealing with each  of  our individual Regions, as separate
autonomies, and that a DOE or a DOD may have a difficult time in getting the
same kind of response to the same situation going from Region to Region. This
is part of what we're hoping can come out of the interagency work is to get more
uniform application and uses.

COLLEEN PETULLO: Is there one form of QA program plan that you're kind
of leaning to at this point?

LLEWELLYN WILLIAMS: When you say a form  of QA program plan, I'm
not quite sure  what you mean.

COLLEEN PETULLO:  EPA tends to be more a laboratory type QA versus
field operational, and DOE tends  to be more field operational, and I'm just
curious as to how you're going to get all this all melted together.

LLEWELLYN WILLIAMS: I think there is much we can learn from the
approaches of other agencies. We will attempt to accommodate and  utilize the
best  that the other  agencies can offer,  and provide a focused program that
everyone can buy in on and live with.
                                                                     41

-------
                         A FiberOptic Sensor for the Continuous Monitoring of
                                     Chlorinated Hydrocarbons
                         P.P. Milanovich1, P.P. Daley2, K. Langry1, B.W. Colston1 Jr.,
                                       S.B. Brown1, and S.M. Angel1
                                    Environmental Sciences Division,
                                   Environmental Restoration Division
                                 Lawrence Livermore National Laboratory
                                          Livermore, CA 94550
Abstract

   We have developed a fiber optic chemical sensor for use
in groundwater and vadose zone monitoring.  The sensor is
a result of modification of previous work in which we dem-
onstrated a fluorescence based sensor for the non-specific
determination of various volatile hydrocarbons.  The prin-
ciple of detection is a quantitative, irreversible chemical
reaction that forms visible light absorbing products. Modifi-
cations in the measurement scheme have lowered the detec-
tion limits significantly for several priority pollutants. The
sensor has been evaluated against gas chromatographic
standard measurements and has demonstrated accuracy and
sensitivity sufficient for the environmental monitoring of
trace levels of the contaminants trichloroethylene (TCE) and
chloroform.
   In this paper we describe the principles of the existing
single measurement sensor technology and show field test
results. We also present the design of a sensor which is
intended for continuous, sustained measurements and give
preliminary results of this sensor in laboratory experiments.

Background

   This sensor technology is an outgrowth of research
initially sponsored by the U.S. Environmental Protec-
tion Agency. Here, a fluorescence based probe for the
remote detection of chloroform was conceived, devel-
oped and demonstrated in the mid-1980's.1  The sensi-
tivity and accuracy of the probe proved insufficient for
many monitoring applications and research was dis-
continued. However, in DOE sponsored research one
of us (SMA) invented a new concept sensor that has
demonstrated significantly improved sensitivity and
accuracy for both TCE and chloroform.2 This sensor is
currently under evaluation in monitoring well and
vadose zone applications.

Principles of Operation

   The basic components of the sensor technology are
the chemical reagent, electro-optic measurement de-
vice, and the sensors. For the latter, we have developed
two versions, one for single and one for continuous
measurements. A brief desrciption of the components
follows.

   Chemistry. The chemical basis of this technology is
the irreversible development of color in specific re-
agents upon their exposure to various target molecules.
The primary reagent is an outgrowth of the work of
Fujiwara3 who first demonstrated that basic
pyridine,when exposed  to certain chlorinated com-
pounds, developed an intense red color. This red color
is due to the formation of highly conjugated molecules
as shown below. We and others have since demon-
                                                    43

-------
strated that this and closely related reactions can be
used to detect trace amounts of these same com-
pounds.4


                          H   H
                                 H   R  H
                                    (Red)
   Sensors The single measurement sensor (Fig 1) is
comprised of the terminus of two optical fibers and an
aliquot (20 ul) of reagent in a small capillary tube. The
fibers are sealed into one end of the capillary tube and
reagent is placed into this capillary to a length of ap-
proximately 5 mm.  A porous teflon membrane is
placed over the open end of the capillary to prevent
loss of the reagent. Target molecules, TCE for example,
readily pass through the membrane and produce color
in the reagent. This color results in decreased transmis-
sion of light at 540 nm.  The measurement of the time
history of the color development provides a quantita-
tive measure of the target molecule concentration.
Since the reaction is non-reversible, the reagent must be
replenished for every measurement. This is readily
accomplished through the use of easily replaceable,
disposable capillaries.
   Electro-optics. The readout device is shown highly
schematically in Fig 3. Here the emission of a minia-
ture tungsten-halogen lamp is collected by suitable
optics, chopped with a tuning fork and directed into an
optical fiber. The fiber transmits this light with high
efficiency to the sensor where it passes through the
chemical reagent, reflects off the teflon membrane, and
is collected by a second optical fiber. This latter fiber
transmits the reflected light to an optical block where it
is divided into two beams by a long pass dichroic
mirror. These resulting beams are optically filtered at
540 nm and 640 nm, respectively, and their intensity is
ultimately measured  with silicon photodiodes using
phase sensitive detection techniques.
                   Figure 3. Sensor readout device
     Computer
  -: M :-•
                                   -.J---_TCE Chloroform
Figure 1. Schematic of the single-measurement sensor
   Figure 2 shows a sensor that has been designed for
continuous operation." It is essentially identical to the
single measurement version with the exception of the
addition of two micro-capillary tubes.  These are used
to supply new reagent to the sensor either continuously
or on demand.
 n«»a»m
                                       -*-TCE Chloroform
Figure 2. Schematic of the continuous-measurement
             sensor
   Since the colored product absorbs strongly at 540
nm and is virtually transparent at 640 nm, the ratio of
540 to 640 gives a nearly drift free measure of 540 nm
absorption. The sensors are calibrated in two ways (1)
in the headspace above standard TCE solutions of
known w/w concentration or (2) in vapor phase using
calibrated dilutions (v/v) of dry TCE vapor. Figure 4
shows the time dependent transmission of sensors
exposed to TCE standard  solutions and a resulting
calibration curve.

Results and Discussion

   Groundwater monitoring. The sensor has been
evaluated against contractor sample and analysis of 40
monitoring wells located within the boundary of LLNL.
These wells are sampled quarterly with subsequent
chemical analysis performed by EPA standard 624
purge and trap gas chromatography (GC). We ob-
tained concurrent samples during the quarterly con-
tractor sampling and used our fiber sensor to make

-------
   0.90
   0.70
 o
 |  0.50
 a>
 I
 5  0.30
   0.10
          3.0
                   9.0
                            15.0
                        Time (min)
                                     21.0
                                               0 ppb
                                              27.0
                                               0)
                                               O
0.60

0.50

0.40

0.30

0.20

0.10

0.00
                                                                     100
                                                                            200
                                                                                   300
                                                                                         400
                                                                                                500
                                                                                               600
                                                         [TCE] (equilibrium vapor phase over given
                                                            ppb in stirred water solution at 25°C)
Figure 4.
Standard
Sample transmission ratio curve, and working standard curve for dual-wavelength absorption sensor.
curve obtained from % transmission at a fixed time following iniation of exposure
duplicate TCE concentration determinations. Figure 5
shows a diagram of the laboratory measurement appa-
ratus.  Samples were sequestered with no head space
into 250 ml Pyrex bottles. These were immediately
returned to the laboratory and divided in half. The
fiber sensor was then introduced into the resulting
headspace through a gas tight valve and a measure-
ment was initiated after stiring the sample for 5 min-
utes.
                                Optical fiber
                                (to spectrometer)

                                Capillary to pump
                                (when operating in
                                continuous mode)

                                Gas-tight valve

                                Sensor

_JL^
^ <=> -4
^
t 	

**"»',
•

-" Water sample
	 Magnetic stirrer

                                                 Table 1 below shows the comparison of some of the
                                              contractor measurements with the fiber sensor. All
                                              fiber sensor values are the average of the duplicate
                                              samples. There is excellent agreement between the GC
                                              and fiber sensor determinations with nearly all values
                                              within the variance of the GC.

                                                 Vadose zone monitoring. LLNL site 300 was chosen
                                              as the location for initial vadose zone evaluation of the
                                              fiber sensor. The vadose zone was accesssed at several
                                              locations through existing dedicated soil vapor moni-
                                              toring points.  The samples were drawn at nominally
                                              450 cc/min through copper tubing to a remote mobile
                                              laboratory. The lab contained both the fiber sensor
                                              apparatus and a portable GC. The instruments were
                                              connected to the sample stream in series as depicted in
                                              Fig 6 below. Both devices were calibrated for TCE
Figure 5. Schematic of vessel used for laboratory
headspace measurements
                                              Figure 6. Schematic of vadose zone sampling and
                                              calibration apparatus. Sample air is drawn with a pump
                                              on board the GC
                                                    45

-------
          Table 1. Representative data from field calibration study, compiled from TCE measurements
          from monitoring wells and piezometers at LLNL.
Well
MW352
P418
MW271
MW217
MW365
Date
2/13/90
2/13/90
3/7/90
3/5/90
3/6/90
[TCEKppb)
Fiber GC
44
54
86
106
27
58
72
160
86
22
Well
MW357
P419
MW364
MW458
MW142
Date
2/13/90
2/13/90
3/7/90
3/6/90
3/6/90
[TCEKppb)
Fiber GC
78
61
59
33
94
84
66
74
20
140
  15-H
              10
20        30
 Elution time (min)
                                       40
measurements with precision gas mixtures prior to
sampling. The fiber sensor tracked the GC very well
through a wide range of concentrations. Figure 7 is a
particularly interesting result.  Here both instruments
were compared in a nearly contamination free location.
It is clear that the GC was at its limit of detection,
whereas the fiber sensor readily made a successful
measurement. Estimates of TCE concentration for this
location was <10 ppb.

   Continuous measuring sensor. The above described
sensor has demonstrated adequate sensitivity and
accuracy to represent a viable new environmental
monitoring technology. However, the current design,
                          Time (min)
Figure 7. Results of (above) GC (SRI Instruments 8610,
      PID detector, 6' x 1 /8" silica gel column), and
      (below) fiber sensor measurement of extremely
      low TCE levels in soil gas (estimated to be -150
      ppbv/v, i.e.: 150 umoles TCE per mole air).
                                                         1.0-1
                                                       E
                                                       M
                                                       | 0.6-
                                                         0.4-
                                                         0.2-
                                                                   '0
                                                                          20
                                                            30
                                                        Time (min)
                                                                                                        60
                                Figure 8. On-demand measurement of 10 ppm TCE
                                (i.e.: headspace measurement over water containing 10
                                ppm TCE) with continuous sensor system

-------
which incorporates  an irreversible  chemical
reaction,  requires  the sensor to be  refurbished
subsequent to each  measurement.  This  liabil-
ity limits its application somewhat in envi-
ronmental monitoring.
  The sensor shown in figure 2 represents the lowest
risk mitigation of this liability. Preliminary resits with
prototypes of this sensor are very promising. Figure 8
shows typical on-demand measurements obtained with
this sensor in laboratory testing. We anticipate that this
sensor will become an integral component in a down-
well monitoring instrument currently being developed
atLLNL.

Acknowledgements

    This work is supported by the DOE Office  of
Technology Development (OTD) and  performed
under the auspices  of DOE contract W-7405-
Eng-48 and the Center for  Process  Analytical
Chemistry. The authors are indebted to Dr.
Lloyd Burgess and the Center for Process Ana-
lytical Chemistry, Univ of  Washington for col-
laboration that led to the design  and demon-
stration of the continuous sensor.  The authors
also wish to thank Dr. F. Hoffman of LLNL for
many helpful discussions.
References

   1. F.P. Milanovich, D.G. Garvis, S.M. Angel, S.K.
      Klainer, and L. Eccles, Anal. Inst, 15,
      137(1986).
   2. S.M. Angel, M.N. Ridley, K. Langry, T.J. Kulp
      and M.L. Myrick, "New Developments and
      Applications of Fiber-Optic Sensors," in
      American Chemical Societry Symposium
      Series 403, R.W. Murray, R.E. Dessey, W.R.
      Heineman, J. Janata and W.R. Seitz,
      Eds.,(American Chemical Society, Washing
      ton, D.C.,1989) pp 345-363.
   3. K. Fujiwara, Sitzungsber. Abh. Naturforsch.
      Ges. Rostock, 6,33(1916).
   4. S. M. Angel, P. F. Daley, K. C. Langry, R.
      Albert, T. J. Kulp, and I. Camins, LLNL UCID
      19774, "The feasibility of Using Fiber Optics
      for Monitoring Groundwater Contaminants
      VI. Mechanistic Evaluation of the Fujiwara
      Reaction for the Detection of Organic Chlo
      rides", June, 1987.
   5. R. J. Berman, G. C. Christian and L. W. Bur
      gess, Anal. Chem, 62,2066(1990).
                                                  47

-------
                     Chemical Sensors for Hazardous Waste  Monitoring
M.B.  Tabaccoj  Q.  Zhou,  and K.  Rosenblum
           GEO-CENTERS, INC.
            7  Wells Avenue
       Newton  Centre, MA  02159
            M.R. Shahriari
          RUTGERS UNIVERSITY
Fiber Optic Materials Research Program
            Piscataway,  NJ
 ABSTRACT:

       A family of novel fiber optic
 sensors is being developed for on-
 line monitoring of chemical species
 in gases and liquids.  The sensors
 utilize porous polymer or glass op-
 tical fibers in which selective che-
 mical reagents have been immobiliz-
 ed.  These reagents react with the
 analyte of interest resulting in a
 change in the optical properties of
 the sensor (absorption, transmis-
 sion, fluorescence).   Using this ap-
 proach, low parts per billion level
 detection of the aromatic fuel va-
 pors, benzene, toluene and xylene,
 and hydrazines has been demonstrat-
 ed, as have sensors for ethylene
 vapor.  Also relevant to groundwater
 monitoring is the development of a
 pH Optrode System for the pH range
 4-8, with additional optrodes for
 lower pH ranges.

 INTRODUCTION

       The  functional operation of
 optical fiber chemical1 sensors in-
 volves the interaction of light
 which propagates through the fiber,
 with a reagent that in turn selec-
 tively interacts with the environ-
 ment to be sensed.   Typical optical
 properties including evanescent ab-
 sorption and fluorescence,  and che—
 miluminescence can be exploited in
 these sensors.   The reagents are
 normally immobilized into a membrane
 or porous  polymer matrix and then
 coated either on the  tip or side of
 the fiber.
        One of the problems encounter-
  ed with fiber optic chemical sensors
  based on evanescent absorption is
  their characteristic low sensitivi-
  ty.  This results from the limited
  depth of penetration of the evanes-
  cent field of the light into the
  reagent cladding as well as the ef-
  fect of internal reflections [1-4].

        Figure 1 illustrates the  prin-
  ciple of detection used in fiber
  optic chemical sensors.   In the fig-
  ure,  porous glass and porous polymer
  approaches are compared to conven-
  tional evanescent chemical sensors.
  In the porous fiber,  the analyte
  penetrates into the pores and inter-
  acts  with the reagent which is  pre-
  viously cast (immobilized) into the
  pores.  The porous fiber has a large
  interactive surface area (due to the
  large surface area provided by the
  pores),  resulting in dramatically
  enhanced sensitivity in the optrode.
  Another advantage of a porous glass
  fiber is the small sensing region
  (about 0.5 cm in length and 250 mi-
  crons in diameter).  Additionally
  the sensor is an integral part of
  the fiber waveguide.   This latter
  feature minimizes the complications
  associated with the physical and
  optical coupling of the sensor probe
  to data transmission fibers.  In
  addition, multiple fiber sensors can
  be deployed using a single analyti-
  cal interface unit.  These sensors
  are expected to be less expensive
  than conventional fiber optic chemi-
  cal sensors based on materials cost
  and ease of fabrication.  Porous
                                               49

-------
fiber sensors for the measurement of
humidity, pH, ammonia, ethylene, CO,
hydrazines, and the aromatic fuel
constituents benzene, xylene and
toluene have been successfully dem-
onstrated by GEO-CENTERS, and by
Rutgers University [5-12].

Fabrication of Porous Glass Optical
Fiber

      Porous glass optical fibers
are fabricated by the Fiber Optic
Materials Research Program at
Rutgers University, using the meth-
odology described below  [5].

      The  material used  in the  fiber
is  an alkali borosilicate glass with
the components  Si02> B203 and alkali
oxides.  This  type of glass  is  a
well characterized system, produc-
ible at  a  low  cost.   Most  important-
ly  it  exhibits  the phenomenon of
liquid/liquid  immiscibility  within a
certain  temperature  range.   The
above  composition is melted in an
electrical furnace  at 1400°C and
cast into  rods with a 20 mm diameter
 and 0.5  m in length.  The rods are
 drawn into fibers at about 700 "C by
 a draw tower equipped with an elec-
 trical furnace.  Fibers with a 250-
 300 micron diameter with a 5-10 cm
 length are then heat treated in a
 tube furnace at 600°C for about 3
 hours.  The heat treated glass be-
 comes phase separated, with one
 phase silica rich and the other bo-
 ron rich.   The boron rich phase is
 leached out of the glass by placing
 the fiber in a bath of hydrochloric
 acid.  The fibers are subsequently
 washed with distilled water and
 rinsed with alcohol.  Figure 2  il-
 lustrates the processing steps  for
 fabricating porous  fibers.

       Subsequent to  fiber prepara-
 tion, the porous segment is  cast
 with the  sensing reagent  (indicat-
 or) .  This  is  done  by dissolving  the
 reagent in  a  solvent at a predeter-
 mined concentration and soaking the
 porous  fiber  in the solution.   The
 reagent is  then dried into  the pores
 by air  drying or in a low  tempera-
 ture  oven.  Alternatively,  the glass
 surface can be treated with a silan-
  izing reagent to facilitate chemical
 coupling  to the sensing reagent.
Fabrication of Porous Polymer
Optical Fiber

     As an alternative to chemical
immobilization or physical adsorp-
tion in porous glass, porous polymer
optial fibers can also be used to
create fiber optic chemical sensors.
Sensors using these fibers have been
demonstrated for ethylene, CO, NH3,
pH, and humidity detection.  The
principle of porous polymer fiber
sensors has the same basis as porous
glass sensors.  Consequently high
sensitivity is achieved.  In this
approach the indicator is dissolved
directly into the monomer solution
before forming the polymer fiber;
therefore, the indicator is strongly
bonded to the polymer network.  In
fact, the porous polymer approach
provides the advantage of both chem-
ical bonding and physical entrapping
of the indicator.  Also, the pore
size and the amount of indicator can
be precisely controlled by changing
the composition of the monomer solu-
tion, resulting in very good sensor-
to-sensor reproducibility.  This
fabrication process is additionally
quite suitable for mass production.
This reduces the cost of optrodes.

      The porous polymer fibers are
prepared by a heterogeneous copoly-
merization technique.  The basic
principle behind this technique is
the polymerization of a mixture of
monomers which can be crosslinked in
the presence of an inert and soluble
component solvent.  Subsequent to
polymerization, the inert solvent
which is not chemically bound to a
polymer network, is easily removed
from the polymer leaving an inter-
connected porous structure.

      Monomer starting solutions are
prepared which contain the cross-
linker, initiator, inert solvent and
chemical indicator.  The mixture,
including the indicator, is injected
into a length of glass capillary,
(typically 500 microns in diameter).
The filled glass capillaries are
sealed such that they are virtually
free of air, and polymerization is
initiated and completed in a low
temperature oven.  After polymeriza-
tion, the uniform and transparent
polymer fibers are pulled out of the
capillaries.  Finally, the fibers
                                               50

-------
are washed in an organic solution to
remove any remaining inert solvent.

      A combination of parameters
determines the final physical prop-
erties of the cross—linked polymer
network.  These include the solvent
properties, amount and type of inert
solvent, as well as the quantity of
cross-linking agent employed.

Results and Discussion

      Porous glass and porous poly-
mer optrodes have been designed and
demonstrated for aromatic fuel va-
pors (benzene, toluene, xylene),
hypergol vapors (hydrazine and
UDMH), for NH3,  CO and ethylene.
Similarly, optrodes have been demon-
strated for the chemical parameters
pH, humidity and moisture content.

      A pH Optrode System is cur-
rently under development which is
applicable to a variety of field
screening and contamination monitor-
ing tasks.  Porous glass pH optrodes
have been fabricated which are oper-
ational in the pH 4-8 range.  A
unique co—immobilization technique
was developed to tailor the sensor
pH sensing range to a specific ap-
plication.  Optrodes are fabricated
by first silanizing the porous fiber
surface to facilitate the attachment
of the sensitive indicator material.
Spectral transmission scans are con-
ducted in order to identify the
wavelength region of maximum sens-
itivity to pH.  The sensor interro-
gation wavelength is selected based
on these spectral scans.

      Optical intensity versus time
measurements  as a function of pH,
have been made for each optrode at
the  interrogation wavelength.  The
sensitivity and linearity is deter-
mined by plotting optical intensity
at equilibrium, versus pH.   Figure 3
shows  the  response of  the optrode
with an immobilized indicator.  The
sensor  is  operational between  pH  4
and  pH  6.5, with  greatest sensitivi-
 ty and linearity between  pH  4.5  and
pH 6.   Saturation of  the  sensor  re-
 sponse  occurs at  pH values  above  7
and less  than 4.
      A second indicator, which is
structurally very similar to the
first indicator, has been tested
with the intent of increasing sensi-
tivity at higher pH values.  The
response of this indicator is pre-
sented in Figure 4.  The data indi-
cates good linearity and sensitivity
above pH 7.

      A mixture of the two indica-
tors was immobilized in a porous
glass fiber.  The results with this
sensor are shown in Figure 5.  The
data indicates both excellent sensi-
tivity and linearity across a pH
range extending from 4 to 8.   The
co—immobilization of these two indi-
cators represents a unique approach
to sensor design and demonstrates
that sensing range can be tailored
to meet specific requirements.

      The reversibility of these
sensors has been evaluated.  This is
accomplished by cycling a test solu-
tion, into which the pH optrodes
have been immersed, between pH val-
ues of 4.5 and 7.

    Figure 6 depicts the variation in
 optical  transmission of the  pH opt-
 rode as  a function of time.  The
 data indicate that the sensor  is
 fully reversible  and peak to peak
 reproducibility is better than 90%.
 The  spikes  in the response curves
 are  artifacts associated with  the
 test setup.   Similar results have
 been obtained using porous polymer
 optical  fiber.

 Fuel Vapor  Optrodes

       GEO-CENTERS, INC. has  design-
 ed,  fabricated and evaluated porous
 fiber optrodes for detection of aro-
 matic fuel  constituent vapors.  A
 xylene optrode with sensitivity <50
 ppb  has  been demonstrated.  Response
 time, reproducibility, linearity,
 and  selectivity have been determin-
 ed.   Benzene and toluene optrodes
 have also been demonstrated.   Labo-
 ratory results indicate that there
 are  highly  sensitive optrodes, with
 near real time response.  They are
 additionally capable of selective
 detection of target species.
                                               51

-------
      With these  optrodes  (as well
 as  the hypergol,  ethylene,  and  CO
 optrodes)  the  rate  of change of the
 optical  transmission is  directly
 proportional to analyte  concentra-
 tion.  An  example of  xylene optrode
 response to different xylene concen-
 trations is presented in Figure 7.
 Each curve corresponds to  a differ-
 ent xylene concentration.   A plot of
 the slopes of  the data in  Figure 7
 versus xylene  concentration is  shown
 in  Figure  8.   This  data  demonstrates
 good sensor linearity from low  part
 per billion to low  part  per million
 concentrations.

      Hypergolic  fuel optrodes  have
 been developed to detect vapors for
 NASA and U.S.  Air Force  operation
 applications.
The principle of operation and  sen-
sor response is similar  to that of
the xylene optrodes.  The hypergolic
fuel optrodes can be configured as
personal dosimeters for  industrial
hygiene applications or  as portable
detection  instruments.   Figure  9
shows a typical optrode  response as
a function of time for different
concentrations of hydrazine.  The
slope of the optical intensity ver-
sus time curve may be correlated to
the hydrazine vapor concentration.

Conclusions

      Sensors utilizing  optical
waveguides offer many advantages for
hazardous waste monitoring applica-
 tions including size, near real  time
response,  and  low manning and exper-
 tise requirements.  Additionally,
porous glass and polymer optical
 fibers offer significant advantages
 in  these applications because their
 large interactive surface area  dra-
matically  improves sensitivity.
They also  provide a continuous  opti-
cal path.  This minimizes mechanical
and optical coupling  losses.  Addi-
 tionally,  sensor  interfaces can be
developed  that allow multi-sensor
operation.  These chemical optrodes
can be applied in a variety of  envi-
 ronmental  monitoring  scenarios,  as
well as  to developmental bioreac-
 tors, control  of process streams,
and industrial hygiene.  A  family of
 fiber optic optrodes  offers the pos-
 sibility of effectively  having  a wet
 chemistry  laboratory  that can be
brought  to the field.
References

 1.   J.F. Giuliani, H.  Wohltjen,
      and N.L. Jarvis, Opt.  Lett.  8,
      54 (1983).

 2.   A.P. Russell and K.S.  Fletch-
      er,  Anal. Chem.  Actal.  170.
      209 (1985).

 3.   D. S.  Ballantine and H.
      Wohltjen, Anal.  Chem.  58,  883
      (1986).

 4.   C. Zhu and G.  M. Hiefttse,
      Abstract 606,  paper  presented
      at the Pittsburgh  Conference
      and Exposition on  Analytical
      Chemistry and  Applied  Spectro-
      scopy,  Atlantic  City,  N.J.,
      (1987).

 5,   M.R.  Shahriari,  Q. Zhou, G.H.
      Sigel,  Jr.,  and  G.H. Stokes,
      First  International  Symposium
      on Field Screening Methods for
      Hazardous Waste  Site Investi-
      gations,  Las Vegas, NV (1988).

 6.    M.R. Shahriari, G.H. Sigel,
      Jr., and Q. Zhou,  Proc. of
      Fifth  International Conference
      on Optical Fiber Sensors, Vol.
      2  Part 2,  373, (January 1988).

 7.   M.R. Shahriari, Q. Zhou, and
     G.H. Sigel, Jr. Opt. Lett. 13,
     407  (1988).

 8.   M.R. Shahriari, Q.  Zhou and
     G.H. Sigel, Jr., "Detection of
     CO Based on Porous Polymer
     Optical Fibers", Chemical,
     Biochemical and Environmental
     Fiber Sensors,  V. 1172, SPIE
     Sept. 6-7, 1989.

9.   M.B.  Tabacco  and K. Rosenblum,
     "Aromatic Hydrocarbon Optrodes
     for Groundwater Monitoring
     Applications",  GEO-CENTERS,
     INC.  Technical  Report GC-TR-
     89-1912.  April  1989.
                                               52

-------
10.   M.B. Tabacco, K. Rosenblum,
     and Q.  Zhou,  "Optrode Develop-
     ment for  Environmental pH Mon-
     itoring" , GEO-CENTERS, INC.
     Technical Report GC-TR-89-
     1989, August  1989.

11.   M.B. Tabacco, K. Rosenblum,
     and Q.  Zhou,  "Personal Hydra—
     zine Vapor Dosimeter", GEO-
     CENTERS,  INC. Technical Report
     GC-TR-90-2071,  February 1990.

12.   M.B. Tabacco, Q. Zhou, and K.
     Rosenblum, "Development of
     Trace Contaminant Vapor Moni-
     tors",  GEO-CENTERS,  INC. Tech-
     nical Report  GC-TR-90-2138,
     August  1990.
                                              53

-------
     a) Evanescent (Internal Reflection), RFS

                       Chemical Reagent
b) Evanescent (Internal Reflection), Side Coated FOCS
c) Porous Fiber (In-Line Absorption or Luminescence)


                Figure 1.
 Schematic Diagram Comparing Basic
            Sensor Designs
100
80
>
1 60
1 40
Q.
0
20
0
4
Sei







x
^




\
\




v
^*»




^•^




— ••
5 6
PH
Figure 3.
isor Response With Bromocresol Green
Indicator As A Function Of pH
7
                                                           Composition Design
                                                                                     Na2BeO,3   Si02
 Melting And Casting
                                                           Fiber Drawing
                                                           Heat Treatment
                                                           Leaching
                                                           Surface Treatment
                 Figure 2.
Processing Steps For Producing Porous
               Glass Fibers
                                      CH2
                                       ii z
                                     (CH3)
                                    o-c

                                       S;
360
300
£" 240
I
Z 180
0" 120
60
0








(





**• -.





•^--i





' 	 ,





<
\
I







\
\





56789
pH
Figure 4.
Sensor Response With Bromocresol Purple
Indicator As A Function Of pH

-------
             Figure 5.
Sensor Response With Co-immobilized
   Indicators As A Function Of pH
                                                        50    100    150    200    250

                                                                Time (seconds)
                                                                                      300
               Figure 6.
Optrode response time as a function of pH
105
X
«
E
c
i
o
in
0
(






• ^"^"^







^
^r
* i
)


^s>
S




^^



^,
^





2
Xylene Concentration in ppm
Figure 8.
Calibration Curve for Xylene Optrode
Based onPorous Glass Fiber
                                                              234

                                                               Time in Minutes
                                                               Figure 7.
                                                Response Curves for Porous Glass Xylene
                                               Sensor. Xylene Concentrations Range from
                                                           2 ppm to -40 ppb
                                          55

-------
x 10
Average Slope (sec
20
15
5
0
Y = 2.55 + 0.1


^

^^

S
13X R = 0.99
„,--
^

*^



I^*3*"


; 20 40 60 80 10
         Hydrazlne Concentration (ppb)
              Figure 9.
Optrode Response to Hydrazine Vapor
  at 32% relative Humidity and 24 °C

-------
                                 Rapid, Subsurface, In Situ Field Screening
                                 of Petroleum Hydrocarbon Contamination
                          Using Laser Induced Fluorescence Over Optical Fibers
    S H Lieberman and G A Theriault
     Naval Ocean Systems Center
              Code 522
         San Diego. CA 92152
            (619)553-2778
     S S Cooper P G Malone and
             RSCMsen
US Army Waterways Experiment Station
         Vicksburg, MS39180
           (601)634-2477
              PWLurk
U S Army Toxic and Hazardous Materials
               Agency
 Aberdeen Proving Ground, MD 21010
            (301)671-2054
ABSTRACT

A new field screening method is described that couples a fiber
optic-based chemical sensor system to a truck mounted cone
penetrometer. The system provides the capability for real-time,
simultaneous measurement of chemical contaminants and sol
type to depths of 50 meters. Standard sampling rates yield a
vertical spatial  resolution of approximately 2-cm  as the
penetrometer probe is pushed into the ground at a rate of 1-m
mirf1

The system employs a hydraulic ram mounted in a truck with a
20-ton reaction mass to push 1 meter long, threaded, steel pipes
into the ground.  The first section of pipe is terminated  in a
60-degree cone and includes strain gauges for measurement of
tip resistance and sleeve friction. A sapphire window mounted
in the side of the pipe, approximately 60-cm above the probe
tip, provides a view port for a fiber optic-based ftuorometer
system. The soil sample is excited through the sapphire window
by light transmitted down the probe over a 500 micron d iameter,
60 meter long fiber coupled to a pulsed nitrogen laser located
at the surface.  Fluorescence generated in the sol  sample is
carried  back to  the  surface by  a second fiber where  it is
dispersed using  a spectrograph  and quantified with a time-
gated, one-dimensional photodiode array. Readout of a fluores-
cence  emission spectrum  requires approximately 16
mail-seconds.  A micro-computer based data acquisition and
processing system controls the fluorometer system, acquires
and stores sensor data once a second, and plots the data In
real-time as vertical profiles on a CRT display.

Results are presented from the first field tests of the system at a
POL (Petroieum-OH-Lubricant) contaminated hazardous waste
site.  Initial results from  a series of more than thirty pushes
indicate that the system is useful for rapid characterization, in
three-dimensions, of the boundaries of a POL contaminant
plume at concentrations equivalent to sub-parts-per-thousand
of diesel fuel marine.  Vertical fluorescence profiles show sig-
nificant small scale vertical structure on spatial scales of a few
cm. This vertical micro-structure appears to correlate with sol
characteristics estimated from point resistance and sleeve fric-
tion. Field ami laboratory calibration of the fiber optic sensor
system using different fuel products is presented and discussed.
Sensor performance is characterized as a function of sol mois-
ture content
                     Introduction

                     Defining the location and extent of subsurface chemical con-
                     tamination is a difficult task. Detailed site investigations require
                     installation of many monitoring wells and subsequent analysis
                     of discrete sol and groundwater samples. Effective site char-
                     acterization is often limited by the ability to select optimal
                     locations for monitoring wells.  Furthermore, the ability to
                     resolve horizontal and vertical features in the distribution of
                     chemical contaminates is a function of limitations imposed by
                     the spacing between wells and the vertical  spacing between
                     samples.

                     At present, locations for monitoring wells are usually based on
                     information gleaned from site historical data,  ground water
                     hydrology, and/or indirect chemical screening such as soil gas
                     measurements.   Because of uncertainties in the information
                     available, well placement is at best an inexact science. Histori-
                     cal data is often incomplete or inaccurate.  Knowledge of
                     groundwater hydrology at the site may not provide the level of
                     detal required to understand site characteristics.  Interpreta-
                     tions of sol gas measurements may be complicated by erratic
                     movement of vapor in the sol due to impervious layers and
                     changes in atmospheric temperature and pressure.   Conse-
                     quently, many wells are not property positioned and, therefore,
                     yield information of marginal utlity.

                     Accurate delineation of the boundaries of contaminant plumes
                     and defining small scale vertical structure in the distribution of
                     contamination has important implication with respect  to site
                     remediation. The more precisely the area of contamination is
                     defined, the less likely ft is that "dean" material will be unneces-
                     sarily removed or subjected to costly remediation procedures.
                     Improved techniques for in situ, subsurface, field screening
                     would have several benefits. Knowledge of the distribution of
                     chemical contamination in sols and groundwater could be used
                     to more effectively guide the placement of monitoring wells and
                     thereby, greatly reduce the number of wells required.  Field
                     screening methods that provide real-time chemical information
                     at closely spaced intervals could be used to rapidly delineate
                     small scale horizontal and vertical structure in contaminant
                     plumes. In addition to increasing the effectiveness of site char-
                     acterization there should also  be a significant cost savings
                                                         57

-------
   Figure 1.   Photograph ofpenetrometer truck developed for use with the fiber optic fluorometer system. The data acquisition system
   and fluoromcter system are located in the rear compartment.  The hydraulic system used to push the penetrometer probe into the soil is
   in thefoward compartment.
associated with the reduced requirement for monitoring wells
and associated analytics.

Towards this goal of improving capabilities for rapid site char-
acterization,  we  have equipped a  truck-mounted cone
penetrometer system (Fig. 1) with a fiber optic based, laser-in-
duced fluorometer system.   Cone penetrometers have been
widely used for determining soH strength and soil type from
measurements of tip resistance and sJeeve friction on an instru-
mented probe (1).  The  probe is normally pushed  into the
ground at a rate of approximately 2-cm sec  using hydraulic
rams working against the reaction mass of the truck. For a 20
ton vehicle, the standard (35-mm diameter) peoetrometer rod
can be pushed to a depth of approximately 50-m in normally
compacted soils.  In order to extend the measurement
capabilities of the penetrometer system to chemical con-
taminants of environmental  concern, it is possible to use the
penetrometer system as a platform for insertion of other sensors
into the soil. To date, use of penetrometers for direct sensing
of chemical constituents in soils has been limited to resistivity
measurements (2) and sensors for measuring radioactivity (3).
This report describes the  development of an optical based
sensor for direct in situ screening of chemical contaminants.
The system employs optical fibers to  make remote  laser-in-
duced fluorescence measurements through a  window in the
probe tip. The system can be used to characterize contaminant
plumes that contain compounds that fluoresce  when exposed
to ultra-violet light. In its present configuration, which uses a
nitrogen (N2) laser (337 nm) excitation source, the system is
selective for polycyclic  aromatic hydrocarbon compounds
which are components of POL products. Coupling the optical
fiber sensor with the cone penetrometer provides a capability
for direct,  real-time sensing of petroleum  hydrocarbon com-
pounds in soils that has not previously existed.
System Description

A schematic diagram of the fiber optic fluorometer system is
shown in Fig. 2. The system was adapted from a design original-
ly developed for in situ fluorescence measurements in seawater
(4-5).  The penetrometer system  uses two silica clad silica
UV/visible transmitting optical fibers. One fiber is used to carry
excitation radiation down through the center of penetrometer
pipe and a second fiber collects the fluorescence generated in
the soil sample and carries  it back to the detector system.
Excitation and emission fibers are isolated from the sample at
the probe tip by a 6.35-mm diameter sapphire window mounted
flush with the outside of the probe approximately 60-cm from
the tip. Although different fibers from several sources have been
employed, the fibers used in studies reported here were 500-^m
in diameter and 60-m in length, unless otherwise noted.  At-
tenuation was specified  by the supplier to be about 100 dB/km
at 337 nm (this corresponds to 25% transmission at 337 nm for
a 60 m fiber.

Excitation radiation  is provided by a pulsed N2 laser  (Model
PL2300, Photon Technology, Inc) that operates at 337 nm with
a pulse width of 0.8 nsec and a pulse energy of 1.4  mJ.  The
beam is coupled into the excitation fiber using a 2.5-cm quartz
lens. Because of asymmetry in the beam dimensions, 6-mm x
9-mm at the laser aperture, coupling losses into the fiber are
somewhat greater than what would be expected for a conven-
tional Gaussian resonator type laser.  No attempt has been
made to reshape the beam to improve coupling.  Instead, we
take advantage of the non-symmetrical beam shape by using a
separate length of optical fiber to intercept a portion of the laser
beam that would not normally be coupled into the excitation
fiber. This auxiliary fiber is coupled to a photodiode that is used
to provide an optical trigger for time gating the detector.  Opti-
cal triggering  of the detector eliminates problems associated
with laser jitter that are experienced with electronic triggering of
                                                           58

-------
                                         • ELECTRICAL SIGNAL

                                         ' FIBER OPTIC CABLE
                           Figure 2.   Schematic of laser induced fiber optic fluorometer system.
the detector.

A photodiode array detector system is used to quantify the
fluorescence emission spectrum brought back to the surface
over the second 60-m fiber. The detector system consists of a
Model 1420 Intensified Photodiode Array Detector (EG&G
PARC) coupled to a quarter-meter spectrograph which houses
a 300 line/mm diffraction grating. The 1024 element array con-
sists of 25 micron wide diodes centered at 25 micron incre-
ments.   For the 300 line/mm grating the dispersion of the
spectrograph translates to a spectral resolution of 0.45 nm per
pixel at the array surface when a 25 micron input slit is used.
The resolution may be increased to 0.075 nm per pixel by using
an 1800 line/mm grating.  Readout of an emission spectra
requires approximately 16 msec. Because the detector can be
readout quickly it is possible to add spectra from multiple laser
shots in order to improve the signal to noise ratio of the meas-
urement. Typically, 10 laser shots are used per sample interval.
Control and readout of the detector is performed by a Mode)
1460 optical multichannel analyzer (OMA) (EG&G  PARC).
Measurements are initiated by an electronic signal from the OMA
that fires the laser.  The laser pulse then triggers an optical
trigger (Model 1303, EG&G PARC) which sends an electronic
signal to a fast pulser (Model 1302,  EG&G PARC).  The fast
pulser implements an appropriate delay and gates the detector
"on" fora period of 20 nanoseconds. Fast-gating of the detector
activates it only during the time period when the fluorescence
signal is  present, thereby minimizing any contribution to the
signal from background light and detector noise.

Incrementing the delay of the detector gate for successive laser
pulses also permits determination of fluorescence decay times.
Other studies have shown that differences in fluorescence
decay times are useful for discriminating compounds of environ-
mental interest (eg., pdycyclic  aromatic hydrocarbons)  that
cannot be resolved based on differences in their fluorescence
emission spectra (5).  At present, fluorescence lifetime meas-
urements are not performed routinely with the penetrometer
system because additional measurement and processing time
would be required. In the future, however, fluorescence decay
measurements could easily be implemented via software con-
trol to take advantage of "dead time" that is currently not utilized
when the push is halted every meter in order to install the next
section of pipe.

 An Intel 386 based microprocessor host computer is used to
automate the overall measurement process.  The host computer
controls the OMA system and stores fluorescence emission data
received from the  OMA and data from strain gauges on the
probe tip.   A representative fluorescence spectrum obtained
                                                         59

-------
    16000

    14000-

    12000-

    10000-

     8000

     6000-

     4000

     2000-
         300   350  400   450   500   550   600  650  700
                         Wavelength (nm)
                                             hle=OUA .iai soil

   Figure 3.   Fluorescnece emission spectum measured for
   contaminated soil using fiber optic fluorometer system.

from contaminated soil at the first test site is shown in Figure 3.
The host computer is also used to generate real-time depth plots
on a CRT of the chemical fluorescence measurements and soil
characteristics as interpreted from the strain gauge data. Under
normal operating conditions, fluorescence measurements are
made at a rate of approximately once a second. For the stand-
ard push rate of 2-cm sec"1 this corresponds to a vertical spacial
resolution between measurements  of 2-cm.  Because each
fluorescence measurement consists of intensities measured at
1024 wavelength points, a push to a depth of only 10 meters will
generate more than 500,000 data points.  In order to simplify
data presentation a window (approximately 50 nm wide) is set
in  the spectral region  anticipated to contain  the maximum
fluorescence intensity.  The average fluorescence intensity in
the spectral window is then plotted as a function of depth, in
real-time, as the probe is pushed into the soil.  A typical data plot
is  shown in  Figure  4.  The entire fluorescence emission
spectrum is stored on a fixed disk to facilitate post-processing
of the data
Characterization and Calibration of Sensor Response

Initially, there were several practical concerns about the viability
of using an optical fiber system to make in situ measurements
in soil  in conjunction with the cone penetrometer.  Issues of
concern included: (1 )Would the sapphire viewing window retain
contaminant after exposure and thereby exhibit a memory ef-
fect?  (2)Could the optical fiber withstand the necessary han-
dling required to thread it through the penetrometer pipe during
insertion and removal? (3)Would the constant flexing of the fiber
during measurement significantly alter the attenuation charac-
teristics of the fiber and thus, invalidate quantitative measure-
ments? Experience gained to date, suggests that none of these
issues appears to be a problem.  Inspection of data in Fig. 4
shows that when the probe was pushed through layers of soil
containing relatively  high concentrations of contaminant,
fluorescence intensities rapidly approached background levels
as soon as the probe moves out of the contaminant zone. This
suggests that the high pressures acting on the window as the
probe is forced through the soil are effective in removing any
residual contamination that might be adsorbed on the window.
Field  experience to date demonstrates that  the fibers can
withstand the normal handling required for operations with the
penetrometer. No fiber failures have occurred during the more
than 80 cone penetrometer tests  (CRTs) that have been made
so far.  Finally, measurements in the field showed that there was
no measurable difference in the amount of laser energy trans-
mitted through the 60-m excitation fiber depending on whether
the fiber was laid out on the ground with no bends or threaded
through 50 meters of penetrometer rod with a 180 degree bend
approximately every meter (as was normally the case).  It ap-
pears that as long as the minimum fiber bend radius, for which
total internal reflection is maintained for all modes,  is not ex-
ceeded there is no significant variation in throughput loss.

Response  of the fiber optic fluorescence sensor  has been
calibrated both in the laboratory and in the field using different
fuel products added to soils.  We have  elected to use fuel
                                          Sleeve      Cone     Soil Class Fluorescence
                                          Friction    Resistance          s    (relative)
                                        (tons sq ft)  (tons'sq ft)          &
                                                               2  ?  -O T3
                                                               I  I  I I
                                        012345  0   100 200   012345   0 10002000
                                                          C-3O-90
   Figure 4.   Example of real-time display showing vertical profiles of soil characteristics and chemical fluorescence measurements.
                                                            60

-------
          2   4    6    8   10   12   14    16
         Diesel Fuel Marine (parts-per-thousand)
18
                       20   40   60  80   100  120  140 160  180  200
                        DFM CONCENTRATION (parts per thousand)
   Figure 5.   Laboratory calibration curves for DFM in soil as
   a function of soil moisture content.
                 Figure 6.   Field calibration of penelrometer fiber optic sen-
                 sor using dieslefuel marine in sand. Inset shows response is
                 linear below Wppt.
products rather than pure compounds because fuel products
contain a representative mixture of the compounds that may
fluoresce in environmental samples. Obviously, there is no way
to be sure that the distribution of compounds that respond to
our measurement system in the field is an exact match to the
product we select for calibration.  In fact, in many cases there
will undoubtedly be a mismatch between the distribution of
compounds in the  product  used for standardization and the
mixture of compounds present at environmental "dump" sites.
These sites may contain a potpourri of products that have had
time to undergo degradation and  loss of more volatile com-
ponents.  However, at sites such  as tank farms that contain
recent or ongoing fuel leaks, it may be  possible to get a good
match between the product used to calibrate the sensor and the
product in the ground. Therefore, it should be stressed that the
utility of the system, in its present form, is for rapid delineation
of hydrocarbon contaminant plumes in order to guide the place-
ment of monitoring wells. With these qualifications with respect
to calibration in mind, data is presented which shows that the
fluorescence sensor appears to be at least a semi-quantitative
sensor for in situ screening of petroleum hydrocarbons.

Laboratory results  (Fig. 5) show that measured fluorescence
intensities increased linearly as a function of diesel fuel marine
(DFM) added to uncleaned beach sand.  Added quantities of
DFM ranged from 500 to 20,000 parts-per-million (ppm) for this
experiment.   Standards were generated by adding known
quantities of fuel product to weighed samples of "clean" soil and
tumbling the mixture overnight in tightly sealed glass containers.
Figure 5 also shows that the measured response did not change
significantly when the water content of the soil was varied from
0 to 10%. Other calibrations using jet fuel (JP-5) in sand also
showed that the fluorescence response did not change when
the water content of the soil sample was varied from 0 to 25%.
This suggests that  the response of the fluorescence sensor
should be  relatively insensitive to changes in soil moisture
content as the probe moves through the vadose zone into the
saturated zone.

The penetrometer fluorescence sensor was also calibrated in
the field by placing a cylinder over the sapphire window and
filling it with "clean" beach sand (Fisher Scientific) containing
               added quantities of DFM.  Results (Fig. 6, inset)  show linear
               response (^ = 0.99) for concentrations in the range of 1000 -
               10000 ppm.  This is similar to laboratory results discussed
               above.  Figure 6 shows that for higher concentrations,  fluores-
               cence intensities appear to approach a saturation val ue at about
               10% DFM in sand (weight/weight). This appears to set an upper
               limit on  the  concentrations that can be quantified with this
               system.  We believe this saturation effect arises because the
               fluorescence response of the sample  is to a large extent a
               surface phenomena. At high concentrations of fluorophore, the
               surface of the soil particles become saturated with product and
               therefore, the fluorescence approaches a limiting value. The
               lower limit of detection for the system configuration described
               in this report is approximately 100 ppm (two times noise) using
               10 laser shots. Detection limits can be improved, at the expense
               of analysis time, by increasing the number of laser shots that are
               stacked  for  each sample interval.  Efforts are currently in
               progress to determine the effect of soil type on fluorescence
               response and to evaluate the "depth of view" of the fluorescence
               measurement (ie., how far into the sediment adjacent to the
               sapphire window does the measurement penetrate).
               Results of initial field tests

               Initial  field tests  of the  fiber optic fluorometer equipped
               penetrometer were conducted at a hazardous waste site in the
               southeastern United States. The site, which dates back to the
               1940's, had been used for several decades as a disposal area
               for mixed petroleum wastes. In the mid-1980's a ditch was dug
               around the site and a recovery system installed. A map of the
               site showing locations of the CRTs is given in Figure 7. Figure
               8 shows representative results from a transect paralleling the
               recovery ditch (CRTs 30-37). The depth of sampling in this study
               was limited to 30 ft by a hard limestone layer.  Inspection of the
               fluorescence profiles ind icates that hydrocarbon related fluores-
               cence was detected at locations 30,32,33,35 and 36 but not at
               location 34 or 37.  These results illustrate how it is possible to
               rapidly delineate the horizontal extent of the contaminated area
               by making a series of CRTs at the  site.  Each CRT required
                                                          61

-------
   Figure 7.   Map showing locations of cone penetrometer
   tests (CPTs) at test site.
approximately 20 minutes to complete. Detailed inspection of
the vertical structure in fluorescence profiles at the locations with
the highest fluorescence intensities (CPTs 30,32 and 36) shows
marked similarities.  Highest intensities were observed at a
depth of approximately 15 feet with a secondary maximum at
about 10 feet and background levels at the surface and at the
                                                               bottom of each profile.  Similarity in the vertical structure ex-
                                                               hibited by the fluorescence profiles at the three locations and
                                                               the covariance with measured soil characteristics supports the
                                                               hydrogeological consistency of the data. The observation that
                                                               CPTs 34 and 37 showed no measurable fluorescence suggests
                                                               that at this  site naturally occurring organic material did not
                                                               contribute to measured fluorescence signals.   In order to
                                                               facilitate interpretation, fluorescence and soil property data from
                                                               individual CPTs can be combined with position information and
                                                               transformed (Dynamic  Graphics,  Inc) into a 3-dimensional
                                                               gridded file for visualization on a minicomputer system. Figure
                                                               9 shows an example of. a 3-dimensional representation of the
                                                               fluorescence data from the CPTs at the sites indicated on the
                                                               map in Figure 7. For this example, fluorescence intensities have
                                                               been converted into  diesel fuel equivalents using the  linear
                                                               portion of the calibration curve presented in Figure 6.
Conclusions and future efforts

Efforts to date suggest that use of a fiber optic based fluorometer
system in conjunction with a cone penetrometer may be useful
for rapid delineation of subsurface petroleum hydrocarbon con-
tamination  at  hazardous waste sites.   Laboratory and field
calibration of the fluorometer system using fuel products (diesel
fuel marine and JP-5) indicates that the fluorometer system is
quantitative for direct determination of these products in soil
             Fluorescence  Soil Class Fluorescence  Soil Class Fluorescence  Soil Class Fluorescence  Soil Class Fluorescence  Soil Class Fluorescence  Soil Class Fluorescence
           |   (relative)        |  (relative)        |  (relative)        |  (relative)        |  (relative)        |  (relative)        ;  (relative)

      its!           mi          nil          nil          ml          ml          ml
       012345   0 10002000   012345  0 10002000   012345  0 10002000  ^ 0123*5  0 10002000  ( 01 2345  0 10002000  ^12345  0 10002000  ( Q1 2345  0 10002000
. -

•
j;::
u-
.

u-
il.il


NO
DATA












P1 — i~t~







HO
DATA

r . i — i—
]_ HOLE 2 _ |
C-3S-90
                                                         L--J        L--J        |_»°-J
   Figure 8.   Test data showing the use of the fiber optic fluorescence sensor for locating the boundaries of a hydrocarbon plume.
                                                              62

-------
  Figure 9.   Eample of 3-dimensional visulalion of soil contamination based on CPTdata. The volume shown represents areas that
  had fluorescence intensities equivalent to 1000 ppm or more dieselfuel marine. The lines on the upper surface represent cultural fea-
  tures (ditches and roads) present at the site.
(sands) for concentrations in the range of 100 ppm to 10000
ppm. At present, the greatest utility of the system is for rapid
screening for POL contamination in order to more precisely
locate contaminated zones, and thus significantly reduce the
number of monitoring wells required for site characterization.
The accuracy of converting measured fluorescence intensities
to concentration units will depend on how closely the product
used for sensor calibration emulates the product in the soil.
Experience in the field indicates that the optical fiber system is
rugged enough to withstand normal deployment procedures
with the penetrometer system and that the sapphire viewing
window appears to be self-cleaning, thereby avoiding memory
effects.

Efforts currently planned, or in progress, include: (1) rigorous
intercomparison of penetrometer field measurements with con-
ventional sampling and standard analytical methods, (2) char-
acterization of  the effect of different soil types  and
characteristics on system calibration,  (3)  enhancing the
capabilities of the sensor system for measuring compounds that
are excited at higher energies by replacing the N£ excitation
source with  a Nd-YAG operating at the third and fourth har-
monics (355 and 266 nm).
References

1.  Olsen, R.S. and J.V. Farr. "Site Characterization Using the
Cone Penetrometer Test."  Proceedings of  the ASCE Con-
ference on Use of In-situ Testing in Geotechnical Engineering.
Amer. See of Civil Eng., New York, N.Y. 1986.

2.  Cooper,  S.S., P.G.  Malone,  R.S. Olsen and D.H. Douglas.
"Development of a computerized penetrometer system for Haz-
ardous waste Site Soils Investigations." Rept. No. AMXTH-TR-
TE-882452, U.S. Army Toxic and Hazardous Materials Agency,
Aberdeen Proving Ground, MD. (1988), 58 pp.

3.  Campanula, R. G. and P. K. Robertson,  "State-of-the-art in
in-situ testing of Soils:  Developments since 1978," Department
of  Civil Engineering, University of British Columbia, Vancouver,
Canada, 1982.

4.  LJeberman,  S.H., S.M.  Inman and  G.A. Theriault.  "Use of
Time-Resolved Ruorometry for Improving Specificity of Fiber
Optic Based Chemical Sensors."  In: Proceedings SPIE Op-
toelectronics & Fiber Optic Devices & Applications, Environ-
ment and Pollution Measurement Systems.  Vol 1172. (1989),
p.  94-98.

5. Inman, S.M., P.J. Thibado, G.A. Theriault and S.H. LJeberman,
Development of a pulsed-laser, fiber-optic-based fluorimeter:
determination of  fluorescence decay  times of  polycyclic
aromatic hydrocarbons in sea water," Anal. Chim. Acta, 239,
(1990), p. 45-51.
                                                          63

-------
                                                             DISCUSSION
 The following is a panel discussion in which questions were posed to the first three
 authors of papers in the Chemical Sensors Session.

 DICK GAMMAGE: Most of the data you showed was for sand. You're going
 to  have  different quenching problems, different degrees  of quenching  for
 different soils. Can that throw you out at all? Also, I thought the original intent
 of this device was to be able to lower it directly into groundwater and take in water
 measurements. And I'm wondering  why your focus seems to be totally on the
 headspace ai this stage?

 STEPHEN LIEBERMAN: I'll talk about this soil type question. That's a good
 question. It's something thai has been on our mind. We actually have a laboratory
 study going right now where we're going to evaluate the effect of soil type on the
 response of the sensor. One of the other considerations with soil type, and this was
 something we visually observed, is if you have a sand the sample volume is going
 lo be different than if you have a very fine grain clay or something like that.  We
 have not parameterized that or really documented what that effect is yet. but we
 are looking  at that. That's kind of one of the drawbacks of rushing some of the
 stuff out in  the field, just to see if you can get that fiber down there without
 breaking in and some of ihose very basic questions. But we haven't ignored that.

 FRED MILANOVICH: The answer to the second pan is a quite complicated
 answer. The experience we've had is that headspace measurement is far and away
 more reproducible. And since this  is  a result-driven technology, we want
 something that  works. When we designed the continuous probe the reagent is
 now in contact with a membrane. When we wet it on the other side with water,
 we have problems. In the original probe there was an air space, and you could
 stick that probe into the water. With the membrane being teflon, the wetting
 phenomenon was different than what has been exposed to the pyridine. So some
 work would have to be done there. But I don't see there's a great liability lo stay
 with headspace.

 JOHN SCHABRON: How often do you have to recalibrate the probe? I guess
 now that you can introduce solution into it. you can calibrate it more frequently.
 Could you also address the issue that, with the two diodes, the red and the green,
 you're not compensating for the  difference in output of the two diodes as you
 would if you had a single lamp and  a monochromalor with two different
 wavelengths.

 FRED MILANOVICH: The calibration issue is a function again of a lot of
 factors. If you make enough reagent and it's stored  cold, you can go with  the
 calibration.  We've gone months with the calibration. But if you mix  a new
 reagent, open a new bottle of pyridine. chemistries are different. So you'd have
 to recalibrate.

 MARY BETH TABACCO: Basically we found that you can adjust the output
 from those two diodes to make them match, make one greater or one less. The
 ability  for the ratio to remain constant isn't  dependent on the output from  the
 ditxles. In the graph that  I showed you. the green output was lower. In fact,  the
 system electronics that we've built, the green is just about the same output value.
 By  adjusting the current lo the LED. you adjust the output value.

 DeLYLE EASTWOOD: As some  of you  may know, there is a fiber optic
 committee chaired by  Dr. Tuan  Vo-Dinh of Oakridge which is working  on
 developing the calibration standards, fluorescence standards and standards  for
 terminology. and collecting a data base for fiber optic chemical sensors. We use
 the  term fiber optic chemical sensors because as some of you know. Optrode is
a registered  trademark. Dr. Vo-Dinh is giving a presentation on that  at the
 Pittsburgh conference in Chicago, Monday. March 4.1 will also chair a meeting
on luminescence at that conference.

There's been a lot of previous work on classification and identification of oils.
some of which is in the literature, and  is the basis for a couple of ASTM methods.
My question is.  do you plan to use another laser and fiber to measure BTX?
STEPHEN LIEBERMAN: Yes, but I'm not sure we're going to get down to
BTX. We did have plans to use  a different excitation source. That should be
coming on line, should at least be available to us about the end of this month. That
will give us the 266 and 355 excitation. But I think benzene and others are even
excited at lower wavelengths. The thing we're bucking there is the transmission
down the fiber. As you know, the attenuation dramatically increases as you go
down in the UV. So right now the 337 is kind of a nice compromise between what
we can get down there and a wavelength that will excite some of the 2,3,4-type
ring compounds. But if we could get the energy down there, it would be real nice
to try to go 200 or so. But I don't see that happening right now. I think 260 is going
to be pushing it. Even at that, we're going to be brute forcing the energy down
there. So I think we may be approaching the damage threshold of the fiber, versus
what we can get out the other end.

GORMAN BAYKUT: I have a  question about telling compounds apart in  a
mixture. You gave an example of a mixture of three compounds. If you have a
high concentration of some compounds with a very low concentration of another
compound, do you have any problems with determining them just using the
slopes?

STEPHEN LIEBERMAN: We have not actually done experiments where
we'vejuggled concentrations of these different compounds and really determined
what the range of concentrations  we're able to discriminate. Obviously that's a
concern. We've done a little bit of work using Lifetimes as a way to discriminate
different metal ions that complex w ith a particular indicator molecule. We've had
some success fining biexponential curves to those compounds. But again, we
haven't really pushed the limit by having tremendous differences in concentration.
Our current thinking is it's going to take a combination of techniques and maybe
a smart pattern recognition-type techniques. We  may be  looking a neural
networks as a way.  But obviously,  there's going to be  some  point in  the
differences in concentration that you're going to be able to determine.

FRED MILANOVICH: In these experiments we actually prepared the solution
so that they'd give a similar initial intensities.

BRIAN PIERCE:  I have four questions: (1) These indicators in your porous
meter, are these reactions reversible? (2) What are the polymers you're using in
your porous polymer monitor or sensors? (3) Have you considered waveguide
configurations? (4) How is it possible to construct these 3-D visualizations from
the finite number of points that you've sampled? What kind of assumptions go
into that?

MARY BETH TABACCO: We're working with both reversible and irrevers-
ible systems. The pH Optrodes. the ammonia sensors are all fully reversible.
Right now. for some of the other vapors sensors forhydrazines.carbon monoxide.
we have  irreversible indicator systems. But as 1 mentioned, in e case of the
irreversible systems we've demonstrated that by monitoring the slope you can
look at real time changes in concentration. For example, with ethylene, we've
cycled concentrations from  100 ppb to 100 ppm and you  basically can monitor
the change in the slope to pull out real time information.

Your third question was about waveguiding. And no. we've not considered that
approach here.

Concerning the actual polymers we're working with, we're using a variety of
polymer  systems,  both  hydrophilic and  hydrophobic.  These are
methylmethacrylate  systems  with  bis-acrylamide  cross-linkers. The actual
formulation varies depending on the sensor. We have applied for a patent for the
pH Optrode under development. But as I mentioned, it is kind of a witch's brew
at this point.

STEPHEN LIEBERMAN: By the 3-D visualization I assume you mean  the
fancy three-dimensional figure of field data. I'm not quite sure if I understand the
question. There's actually a lot of data points here that represents about 30
                                                                         64

-------
                                                           DISCUSSION
pushes. We're firing that laser about once a second as we're pushing it into the
ground. So we're getting a point in the vertical about every two centimeters. Now
obviously you have to be careful in any kind of three-dimensional visualization
— it only represents reality as good as those contouring algorithms. I think the
proper way is to first plot out your raw data in cross-section or by profile. You
have to make sure that the visualization you generate by the more sophisticated
computer program reflects the reality of what you saw in those individual
profiles.

MARTY HARSHBARGER-KELLY: What is the software package you're
using on that  Macintosh  for data manipulation and who's  the software
manufacturer?

FREDMILANOVICH: The software package is Lab View. It's all icon driven.
so no words  are typed to do all  that interfacing, just moving icons around. I
believe the software manufacturer is National Instruments.

BERT FISHER: Your instrument is measuring polyaromatic hydrocarbons, so
it's a bit misleading to say that you're measuring product,  because you're
measuring some chunk of that. Also, this really shou Id be able to look at historical
spills. Have you looked at weathered materials, because the PAH's will hang
around. And my comment on the three-dimensional visualization is, it's a lot like
doing geology.  You have great resolution  in the vertical and you  accept the
horizontal on faith. So it's like doing stratigraphy.

STEPHEN LIEBERMAN: As to your question regarding weathered product,
I showed you data from a Jacksonville site that has a rather checkered past. Those
deposits go back 30 or 40 years. Now in geological terms that may not be your
idea of a weathered product, but  it's not a fresh product. Actually there's some
work Iknow out of the petroleum people that shows that those PAH spectra don't
seem to change very much as  a function of time, at least with the PAH
components, but we don't have any real evidence. This is also sort of a brute force
method here. We're taking this thing out on the field and we're sticking it in the
ground. We don't know very well  what's down there or what we're even looking
at. Personally I think it would be much nicer to go to some sites where we have
some more recent leaks from a tank farm or something like that where we could
put ourselves to a better test of whether wecan discriminate for instance JP5 from
diesel fuel. Hopefully we would also have information on how old the product
is and how long it's been in the ground.

BERT FISHER: That really was my concern, in that you would be seeing stuff
where there in fact was no product, but you were looking at a tremendous amount
of PAHs that had been hanging around for many years.

STEPHEN LIEBERMAN: That may be the case in that example.

PETER KESNERS: As I  understand your apparatus, there's a  membrane
permeation front on it. What sort of membrane types have you investigated? Do
you think it's feasible to measure pyridine in water with other membranes with
the sensor working the other way around?
FRED MILANOVICH: That's a real interesting plot. Our concern with the
membrane is to keep pyridine out of the water, so we have solicited help
anywhere we can. The current membrane that works the best is plumber's tape,
simple expandable teflon plumber's tape. And that's a result of trial and error
from attempts too numerous to mention. Probably 40 or 50 membranes have been
tried and plumber's tape is the best. We do have some proprietary technology
from companies that we aren't able to speak about yet that could exceed the
plumber's tape.

TODD TAYLOR: It seems to me that the cal ibration curve that you showed on
the screen is going to depend on quite a few things in addition to the soil type. It
seems to me it's going to depend on the water content, because water is going to
affect the amount of oxygen quenching going on in the soil. It's going to depend
on the oxygen concentration.  Surface soils are known to contain a lot of humic
materials, and those materials naturally fluoresce. Their fluorescence, in fact,
depends on metal concentration in the soil. So it seems to me there are quite a few
factors which may be involved in looking at the fluorescence of the soil. And the
last question is not really a question. It's more the fact that I think that you have
a lot more work to do in characterizing your system.

STEPHEN LIEBERMAN:  In the previous graph, I did show that we have
looked at varying the water content over from all the way dry to up to 10% in the
data I showed you and 25% with JP5. And seeing, somewhat surprisingly to me,
no real significant change in the response of the sensor. And so I think at least as
a first cut we have addressed that. As to  the question of humics, we've also
considered that question. In the case of the Jacksonville data, we showed the fact
that we could leave the area that historically was the site where the contamination
was and get down the background fluorescence, at least at that site. I don't think
we have a problem with background fluorescence due to the humic substances,
although we have done some other tests where we've measured humic substances.
We' ve looked at their spectra characteristics and also looked at their decay times.
The decay times for the humic substances appear to be much shorter than what
we're seeing for the petroleum products. So if we do run into a case where we are
getting background fluorescence due to naturally occurring organics, there's at
least some hope that we may be able to resolve that based on their emission
curves.

I agree with you, there's tons of problems out there that need to be addressed and
looked at in more detail. Our approach has been one to let's push this thing out
in the field and see what happens. Let's fill in some of these questions later, when
we get some handle on what we are seeing. But I think that the true proof of this
thing, and this is where we stand right now, is going to be to do some of these
profiles and then rigorous validation of it: to collect samples and analyze them
by the more conventional methods. Obviously that needs to be done. And  that's
going to be the thrust of our effort now.
                                                                     65

-------
          SPECTROELECTROCHEMICAL SENSING OF CHLORINATED HYDROCARBONS
               FOR FIELD SCREENING AND JN. SHU MONITORING APPLICATIONS
                        Michael M. Carrabba, Robert B. Edmonds and R. David Rauh

                                         EIC Laboratories, Inc.
                                 111 Downey Street, Norwood, MA 02062

                                                 and

                                           John W. Haas, m

                        Oak Ridge National Laboratories, Health and Safety Division
                                P.O. Box 2008, Oak Ridge, TN 37831-6383
ABSTRACT

The detection and identification of chlorinated hydrocarbon
solvents (CHS) have been demonstrated by combining the
principles of spectroscopy and electrochemistry.   The
successful observation of die CHS is highly dependent on
the analysis procedure. The procedure is based on a photon
induced electrochemical reaction which is detected by
surface enhanced  Raman   spectroscopy  (SERS)  on
electrodes. The results and methodology of the technique
will be discussed.

INTRODUCTION

The importance of  techniques to sense and  monitor
chlorinated hydrocarbon solvents  (CHS) are becoming
increasingly more important with the intensifying presence
of groundwater contaminations.   Our research and
development effort is aimed at  producing a commercial,
low cost, field portable instrument for the field screening/in
situ monitoring of contamination from chlorinated organic
solvents  based on  spectroelectrochemical fiber  optic
probes.  Some  of  the advantages of  this technique for
monitoring a contamination site are cost, small size of
sampling probe, real-time analysis, the capability of sensing
in adverse environments, and the ability of using a central
detection facility.   The technique  has  an advantage over
current fiber   optic chemical  sensing  methods  for
chlorinated organics in that the sensing only takes place
when the electrochemical device is turned on. This should
enable long term monitoring of a well to be accomplished
with only one probe.

Our monitoring system for chlorinated organic solvents is
based on the  principle  of combining spectroscopic,
electrochemical and  fiber optic techniques  (Spectro-
electrochemical Fiber Optic Sensing (SEFOS)). SEFOS is,
in principle, a generic technique which can be adapted to
many different sensing applications.   With the SEFOS
technique, we use electrochemical methods to reduce the
chlorinated organic solvents into reactive intermediates.
The  reactive intermediates can  then react  with the
"trapping" reagent  and spectroscopic  changes, such as
surface enhanced Raman  spectra, are  used to sense the
chlorinated organics at levels  far below their detection
means by electrochemical methods alone. Previous work
(1) has shown the usefulness of using surface enhanced
Raman  spectroscopy  (SERS)  for  the  detection  of
groundwater contaminations and the technique has also
been successfully applied  to fiber optics (2).  However,
these past experiments have mainly been restricted to
aromatic hydrocarbons.

In this manuscript we will discuss some of the fundamental
aspects of using SERS for the examination of the following
chlorinated hydrocarbons  or organochlorides:  carbon
tetrachloride,  1,2-dichloroethane (DCE), chloroform and
trichloroethylene (TCE). Our interest in these compounds
stems from their existence in the groundwater  at the
Department of Energy hazardous waste sites.

EXPERIMENTAL

The Raman spectroscopy system for conducting the SERS
experiments at EIC has been previously described (2). The
system used at Oak Ridge National Laboratory (ORNL) is
shown in Figure 1 and, with the use of an optical fiber for
excitation, represents a first step toward a remote fieldable
Raman system. Of note in  the optical system is placement
of the laser line pass filter (BP) after the optical fiber to
remove interfering Raman scattering from the fiber itself
                                                  67

-------
(3).   Both research groups employed high-resolution
spectrometers and diode array detectors for measuring
Raman  scattering from  similar spectroelectrochemical
cells. As shown in Figure 2A, each cell was fabricated from
a 3 x 6 x 3 cm quartz cuvette with O-ring joints fused into
three sides and the top. Electrodes were fed into the cell
through O-ring joints and consisted of Pt counter, Ag/AgCl
reference, and copper working electrode.   The working
electrode was placed about 2 mm from the Oarge) face of
the cell between the two electrodes.   This orientation
minimized the path length of incident and scattered light
through the sample solution and simplified alignment of the
electrode in the optical system. For transport/concentration
studies,  a membrane could be  sandwiched between  die
spectroelectrochemical cell  and a second  cuvette with
matching O-ring joint fused into the bottom (Figure 2B).

The   spectroelectrochemical  procedures   were  first
developed at EIC and then used at ORNL. Electrochemical
roughening of polished copper electrodes,  consisting of
high purity 1.0 mm  copper  wire, was achieved  with an
oxidation/reduction cycle (ORC) from -0.6 to +0.2V in a
0.1M KC1 electrolyte at 25 m V/sec. Saturated solutions of
the chlorohydrocarbon solvents (CHS) in distilled water or
100  ug/ml solutions of CHS in 0.1 M KC1 were cycled
several times under the same conditions and optimum SERS
spectra were acquired at -0.2V on the cathodic sweep. All
cycling occurred under laser illumination at 625 nm at EIC
or 647 nm Krypton illumination at ORNL. The use of the
slightly different wavelengths for illumination and Raman
spectroscopy did not produce significantly different results
at the two labs.

RESULTS AND DISCUSSION

Our results confirmed previous experiments (1)  which
indicated that carbon tetrachloride was not observable on
Ag substrates. In addition, we were unable to observe the
chlorinated hydrocarbons on  Ag  or Au  substrates.
However, when we examined the chlorinated hydrocarbons
with a Cu electrode, we were able to observe the SERS
spectra of carbon tetrachloride (Figure 3) as well as the
SERS spectra of TCE, DCE and chloroform  (Figure 4).

The best SERS spectra were obtained when the ORC cycle
was stopped during the reduction step at the potential of
zerocharge for Cu(-0.2V) (4). The observation of the SERS
spectra was also highly dependent on illumination during
the cycling. Previous work by Thierry and Leygraf (5) has
indicated the  importance of illumination during the
electrochemical roughening of Cu electrodes to produce
Raman active sites.

The vibrational features in Figure 4 indicate that a reaction
is occurring on the  electrode  surface (see Table 1  for
vibrational assignments).  From the spectra, it appears that
ring formation is occurring due to an electrochemical and/or
photochemical process. However, in our experiments no
SERS spectra of the CHS  were observed  unless the
electrode was illuminated during the reduction step and thus
a strictly electrochemical reaction can be ruled out
This "photo" induced result indicates the possibility of a
photoelectrochemical process. Copper oxides are known
to be p-rype semiconductors which eject electrons under
illumination (Equation 1) (6). The band gaps for the two
possible copper oxides are 2.0-2.6 eV (620-477 nm) for
Cu2O and 1.7 eV (730 nm) for CuO. These electrons can
then electrochemically reduce the chlorinated hydrocarbon
solvents.

               Cu2O + X -> Cu2O(hV)

This electrochemical reduction  is similar to a reaction
scheme  for the  electrochemical reduction of chloroform
which has been determined by Fritz and Kornrumpf (6) to
be:
              CHC13 + 2e -> CHC12 + Cl

          CHC12- + CHC13 -> CH2C12 + CC13

                  CC13   -> :CC12 + Cl
The  formation  of the  dichlorocarbene  during  the
electrochemical reduction process would tend to form a ring
type structure (6).  This ring type structure is indicated in
our SERS spectra with the strong band at 1380 cm'1.

A preliminary observation  has indicated that the SERS
spectrum is only observable for a finite amount of time. The
result is either due to the degradation of the electrode or the
sample. If the electrode was replaced with a new SERS
surface and then placed in the same solution, the spectrum
was still not observable. This indicates that the chlorinated
hydrocarbons were being consumed during the experiments
in the small volume (10 ml) of analyte. Confirmation of this
result would indicate that the SERS  on Cu surfaces is a
method which is capable of both sensing and removing the
chlorinated hydrocarbons from the solution.

To determine the cause of the disappearing SERS signal, a
series of SERS/GC experiments which determined the TCE
concentration before and after the SERS experiments were
performed. Saturated samples of trichloroethylene (TCE)
in 0.1M KC1 and distilled H20 were cycled in a sealed glass
SERS cell to prevent the possibility of outgassing of the
TCE.   Samples of the saturated TCE  solutions were
collected both before and after the electrochemical cycling.
These samples were analyzed on a Hewlett-Packard Model
HP 5730A Gas Chromatograph.  Chromatograms were
recorded  and the  magnitudes of  retention  peaks were
examined for the TCE peak in  the experiments.  Large
spikes at the 45 second retention time were due to impurities
in the distilled water. The chromatograms showed that a
large amount of TCE was consumed during electrochemical
cycling. Figure 5 represents a typical "before" and "after"
chromatogram.
                                                    68

-------
 Analysis of "before" and "after" chromatograms showed an
 average consumption of 66% of the trichloroethylene
 during the electrochemical cycling and SERS experiments.
 This is consistent with our observation that a film was being
 formed on the roughened copper  surface of our working
 electrode. The formation of a film also indicated the carbene
 may be  originating  a radical  induced polymerization.
 Methods for determining the exact structure of the products
 formed during electrochemical cycling are currently under
 investigation.

 CONCLUSION

 The observation of a "photo" induced SERS process in the
 analysis of the chlorinated hydrocarbon solvents has future
 implications for environmental sensors. Previous to this
 work it was thought that the CHS type compounds were not
 observable by the SERS technique. Upon completetion of
 our fundamental experiments, future work will concentrate
 on the analytical applications  of the process and  the
 development of field portable Raman instrumentation.

 ACKNOWLEDGMENT

 This work was conducted in part under a collaborative
 research agreement (CR-90-003) between EIC and ORNL
 (Martin Marietta Energy  Systems). Financial support for
 this work was derived in part from the Office of Health and
 Environmental Research  Division of the Department of
 Energy under the Small Business Innovative Research
 program.

 REFERENCES

 1. Carrabba,  M.M.,  R.B.  Edmonds  and R.D.  Rauh,
   "Feasibility Studies for the Detection of Organic Surface
   and   Subsurface    Water    Contaminants    by
   Surface-Enhanced  Raman  Spectroscopy  on  Silver
   Electrodes", Anal. Chem., 52,2259 (1987).

 2. Carrabba, M.M., R.B. Edmonds, PJ. Marren and R.D.
   Rauh, Proceedings of the First International Symposium
   on Field Screening  Methods for Hazardous Waste Site
   Investigations, Las Vegas, Nevada, October 1988, p33.

 3. Carrabba, M.M. and  R.D.  Rauh, "Apparatus for
   Measuring Raman Spectra Over Optical Fibers", U.S.
   Patent Application 07/442,235 (1989).

4. Bunding, K., J. Gordon, and H. Seki, "Surface-Enhanced
   Raman Scattering by Pyridine on a Copper Electrode",
   J. Electroanal. Chem., 184.405 (1985).

5. Thierry,  D.  and  C.  Leygraf, "The  Influence  of
   Photoalteration on Surface-Enhanced Raman Scattering
   from Copper Electrodes", Surf. Sci, 149.592 (1985).

6.  Fritz, H. and W. Kornrumpf,  "An Improved Cathodic
   Generation of Dichlorocarbene", Liebigs Ann. Chem ,
  2,1416(1978).
 Figure 1.   Experimental setup for "photo" induced SERS
 experiments at ORNL. F = optical fiber, O - microscope
 objective, CL- collimating lens, FL = focusing lens, P =
 right angle prism, BP = laser line pass filter, BR = laser line
 rejection filter, C = spectroelectrochemical cell.
                             (A)
              0=0=0
Figure 2.   Diagram of spectroelectrochemical cell.  (A)
Top view showing 3  electrode ports and O-ring joint
opening in the top of the cell. (B) side view showing sample
reservoir attached  to  the  top  for  membrane concen-
tration/transport studies. Only 2 of the 3 electrode ports are
visible. In both diagrams the arrows point along the optical
axis as shown in Figure 1.
                                                 69

-------
          4OO     60O    8OO
                 WAVENUMBER
10OO   12OO
                                                              TCE "before"
                                                                                            TCE "after"
                                                                                      UL
                                                         Time (minute*)
                                                    Time (minutes)
Figure 3.  The SER spectnim of a saturated solution in
water on a Cu electrode of carbon tetrachloride.  The
spectrum has been smoothed for clarity.
                   Figure 5.   Gas chromatograms of TCE solution before
                   and after the SERS experiment. Retention time for the TCE
                   peak was 2 minutes.
                                        B
         BOO    BOO   1OOO   12OO
                    WAVENUMBER
                                     14OO
 Figure 4.   The SER spectra of saturated solutions in water
 on  a  Cu  electrode  of (A)  trichloroethylene,  (B)
 1,2,-dichloroethane and (C) chloroform.
                                                    70

-------
                                                 Table 1
  Major Raman/SERS Peak Positions (cm"1) and Vibrational Assignment for the Chlorinated Hydrocarbon Solvents
CCl,
Raman
227s


319s
462s




762 w
787 w
















SERS
220 w
261 w
288 w


521 w




791m



1051 w
1089 w











CHC13
Raman







689s


760m






1218 w









SERS





526m

670 w


783s


1021 w
1056m

1151 w

1234 w
1313m
1352m
1381s

1465 w

1550 w
1581 w
DCE
Raman







656s
674m

755s
882 w
944 w

1055 w


1209 w

1306 w


1433 w




SERS





521m




782s

965 w
1024 w
1058m
1101 w
1148w

1239 w
1312m

1379s

1464 w
1509 w

1582m
TCE
Raman






628s



780m
842 w
930 w





1247m







1585s
SERS





524m




781s
862 w
963 w
1018 w
1055m
1105w
1167w

1237 w
1312m
1358m
1379s

1463 w
1505 w

1580s
Vibrational Assignment
Cu-C?
Cu-C?
Cu-C?
"chain expansion"
symmetric CC14 str.
CCl str., Cu-C stretch?
CCl str. - secondary CA
CCl str. - primary CA
symmetric CC13 str.
CCl str. - primary CA
CCl str. - primary CA
CCl str. - primary CA
CC skeletal str.
CC skeletal str., ring
"breathing"
in-plane CH deformation, CC
str., ring "breathing"
CC str., ring "breathing"
CC str., ring "breathing"
ring "breathing" - cyclopropane
type
CH2 twist and rock
CH2 twist and rock, in-plane
CH deformation
CH2 in-phase twist, CH2 twist
and rock, in-plane CH
deformation
CH deformation
ring str.,
CH2 deformation
CH2 deformation
symmetric C=C str. - cyclo
C=C str. - cyclobutene
C=C str. CA, 3 or C=C couple
str. - polyene
s - strong intensity, m - moderate intensity, w - weak intensity
CA = Chloroalkane, str. = stretch
                                                    71

-------
                                                         DISCUSSION
ARTHUR D'SILVA: In the E.I.C. experiments at what wavelength did you
measure the fluorescence?

MICHAEL CARRABBA: We're looking at the complete spectrum, in this case
a very simple proof of concept. We weren't trying to develop a highly  skilled
system as the Livermore people have developed, or as the people at GEO-
Centers. We're proving the concept here. We just monitored the intensity under
the total fluorescence band.

ARTHUR D'SILVA: What is the excitation wavelength?

MICHAEL CARRABBA: The excitation wavelength was 514 nanometers. We
added an argon-ion laser. We believe  we could  use just about any  of the
wavelengths from 488 up to about possibly 600, but we really didn't try the 600.

EDWARD POZIOMEK: In the experiment where you described the photon
induced reaction, did you utilize a base?
MICHAEL CARRABBA: In the electrochemical experiment you don't need
the base. We use it as our bench mark, and then put the electrodes in. I believe we
don't need the base, and that's probably the important point.

EDWARD POZIOMEK: If you had the opportunity to solve a technology
barrier, which one would you go after first in this area to move it faster?

MICHAEL CARRABBA: The implication of the dichlorocarbene, going after
a double bond, could be quite lucrative in the future. And we believe we can make
probe systems that have been coded right onto an optical fiber and a very simple
sensor. That's where I think we'd pursue it at this point. Basically we'd use some
particular dyes that when the dichlorocarbene attacks the double bond it breaks
the conjugation  and the fluorescence disappears or new fluorescence appears.
That's the direction that we're working on right now.
                                                                     72

-------
                    SURFACE ACOUSTIC WAVE (SAW) PERSONAL MONITOR FOR TOXIC GASES
                                     N. L Jarvis, H. Wohltjen, and J. R. Lint
                                          Microsensor Systems, Inc.
                                             6800 Versar Center
                                            Springfield,  VA 22151
ABSTRACT

A demonstration  model 4-sensor Surface Acoustic Wave
(SAW) Personal Monitor for Toxic Gases was designed and
built, with emphasis on  minimizing  the overall  system
size, weight, power consumption, and complexity.   The
completed demonstration unit contained four 158 MHz SAW
delay lines,  supporting RF  electronics, microcomputer
(microcontroller),  a miniature pump,  valve, gas transfer
lines, and  a small  scrubber  to provide  a clean, dry, air
source  to  establish sensor  baseline  frequencies.   The
demonstration unit weighs  approximately 2 pounds.  The
projected size of the follow-on unit is  expected to be  6" x
3" x  1".  Unlike previous SAW vapor sensor arrays, which
utilized coatings that  interact  reversibly with  specific
classes of toxic organic vapors, this SAW Personal  Monitor
takes advantage of sensor coatings that react irreversibly
with  toxic chemicals.   Thus  it can  more  easily  and
effectively determine total exposure to a given toxic  gas.
The following toxic inorganic gases were selected for study
with the demonstration system: HCI, NO2, SO2, NO2,  H2S
and  NH3.   Coating materials  were  selected  that react
irreversibly with each gas.  The coatings were applied to
the  SAW sensors  and their performance evaluated for
exposure to a single gas.   The results show that suitable
materials are available for  use  as dosimeter coatings for
SAW sensors. Thus the potential exists for developing an
effective  SAW   Personal Monitor  for  detecting  and
monitoring each  of the above gases,  except NOg, at
concentrations well below the OSHA "action levels".
INTRODUCTION

In  all  areas of environmental  monitoring,  as well as
industrial  hygiene,  there  is  a need for smaller,  more
sensitive,  and  inexpensive  personal  monitors  (e.g.,
dosimeters) for toxic  gases and  vapors.  For example,
personnel involved in  field  screening must be concerned
with their personal health  and safety when  working at a
field site, and may often require accumulated exposure data
for various toxic gases. SAW sensor technology, however,
is not limited to use in a Personal Monitor (e.g., a toxic gas
monitor that can be worn on clothing).   The same sensor
technology could be extended to the development of small,
hand-held or  in-situ  monitors  for  a  variety of  field
screening applications.

There are a number of techniques currently being used to
acquire toxic  exposure data, however,  each have their
limitations.  In the future large  numbers of more effective
monitors  will  be  required for  the  rapid  and  reliable
detection  and/or monitoring of toxic gases and vapors at
ever lower  concentrations, in  response  to  increasingly
stringent  state  and  federal health  and  environmental
regulations.  Chemical microsensors have demonstrated the
sensitivities and physical  properties needed  to meet the
size,  cost,  and  performance   requirements of a new
generation of  personal  monitors,  and  should ultimately
find  a wide range of applications within  the  industrial,
medical, and environmental communities (1  -  13).

Of the chemical microsensors that have been investigated to
date, SAW devices, which measure changes in mass when a
chemically specific surface coating adsorbs or reacts with
an appropriate  gas, are the best  characterized and the most
promising for rapid development.  SAW devices have been
shown to respond in  just  seconds to selected vapors at
concentrations  down  to the  parts per billion  range for
specific organic chemicals.   Because of their solid state
construction and compatibility with integrated electronics,
they  can  be  easily  incorporated   into  very  small,
lightweight instruments, small  enough  to  be worn on
clothing.   The  primary   challenge  remaining  in  the
development of  SAW  based  microinstruments  is  the
development of more selective and sensitive SAW coatings
for specific gases and vapors.  Other technical areas to be
addressed are the miniaturization of supporting electronic
components and the development of computer software to
facilitate  sensor  operation, data analysis, and  data
reporting.
                                                       73

-------
OBJECTIVE
                                                            2.     SAW Sensitivity and Selectivity
The objective of the present study was to demonstrate the
feasibility of developing  a miniaturized Surface Acoustic
Wave  (SAW) Personal Monitor with the size, sensitivity,
selectivity,   reliability,   and  low  power  consumption
appropriate  for wearing on clothing.  To achieve this
objective it was necessary demonstrate that:  (1) the SAW
sensors and  necessary  support  electronics can  be
sufficiently  miniaturized;  (2)  chemically  selective SAW
coating materials are available or can be developed for the
detection of  a wide range of toxic gases; and  (3) the SAW
sensors and their coatings can be sufficiently sensitive to
specific toxic  gases  to  meet the  requirements of  field
screening,  personal  safety, and  related  monitoring
applications.
SAW SENSOR INSTRUMENTATION
1.
SAW Sensor Operating Principles
SAW devices are mechanically resonant structures whose
resonance  frequency is perturbed by the mass or elastic
properties of materials  in contact with the device  surface.
Rayleigh surface waves can  be generated  on very small
polished chips  of piezoelectric materials (e.g. quartz) on
which  an  interdigital  electrode array  is lithographically
patterned.   When the  electrode is excited with  a radio
frequency  voltage,  a  Rayleigh wave  is generated that
travels across the device surface until it is "received" by  a
second electrode. The Rayleigh wave has most of its energy
constrained to the surface of the device and thus interacts
very strongly with any  material that is  in contact  with  the
surface.  Changes  in mass or mechanical  modulus of  a
surface coating  applied to the device produce corresponding
changes in wave velocity.  The most common configuration
for  a SAW  vapor/gas sensor  is that of a  delay line
oscillator in which the RF voltage output of one electrode is
amplified and fed to the  other.  In  this way  the device
resonates at a frequency determined by the  Rayleigh wave
velocity and  the electrode spacing.   If the mass of  the
coating  is  altered, the  resulting change in  wave  velocity
can be  measured as a shift  in resonant frequency. SAW
vapor/gas  sensors are similar to bulk  wave piezoelectric
crystal sensors, except they have the distinct advantages of
substantially  higher sensitivity,  smaller  size, greater ease
of coating,  uniform surface mass sensitivity, and improved
ruggedness.  Practical  SAW sensors  currently have active
surface areas of a few square millimeters and resonance
frequencies in  the range of hundreds of MHz.  However,
SAW devices having total surface  areas significantly less
than a square  millimeter and  resonant frequencies in  the
gigahertz    range   are   possible   using    modern
microlithographic techniques.    Such devices would
ultimately increase device sensitivity as well as decrease
size.   Most  of  the SAW vapor sensors reported in  the
literature employ two delay line oscillators  fabricated side
by  side  on the same  chip, with one  delay line  used to
monitor  the  toxic chemical  and the  other to act as  a
reference  to  compensate   for  changes  in   ambient
temperature and pressure.
A 158  MHz SAW device having an active area of 8 mm2'
will  give a resonant frequency shift of about 365 Hz when
perturbed by a surface mass change of 1  nanogram. This
sensitivity  is  predicted  theoretically   and  has  been
confirmed experimentally.   The same device  exhibits a
typical  frequency "noise" of less than 15 Hz RMS over a 1
second measurement interval (i.e.  1  part  in 107).   Thus,
the  1 nanogram mass change gives a signal to noise ratio of
about 24 to 1.  For vapor or gas sensing  applications, the
objective is to have the chemical selectively adsorb onto
the  mass sensitive surface of the device.   Chemically
selective coatings are used for this critical operation.


3.     Selective Coatings

The operational behavior of a Surface Acoustic Wave device
can be  very  sensitive  to changes  in  density, elastic
modulus, and  viscosity of the  surrounding  medium;
however, SAW  devices are not inherently sensitive to the
chemical properties of the medium surrounding the device.
When coated with a chemically selective thin film they can
exhibit remarkable sensitivity  to  small  quantities  of a
chemical vapor or gas. The development  of such selective
coatings  for toxic chemicals can take two directions, (1)
coatings that will  selectively  and reversibly  adsorb a
selected  vapor   or  gas   by  matching  "solubility"
characteristics; and (2) coatings that react chemically and
irreversibly  with   a  selected vapor or gas.    SAW
selectivities  in  excess  of 10,000  to  1  for certain  toxic
chemical  agents  have  been  demonstrated   using  the
"solubility" approach.  Much greater  selectivities should
be  possible using chemically reactive coating/vapor (gas)
combinations.
                                                     SAW INSTRUMENTATION DEVELOPMENT

                                                     1.     Miniaturization of SAW Sensor Array and RF
                                                            Electronics

                                                     Ultimate  miniaturization would  be achieved by  going to
                                                     hybrid circuitry,  where  the sensors and support  RF
                                                     electronics could be reduced in  size to a few cm2 or less.
                                                     Hybridization,  however, will require  a major engineering
                                                     effort and was beyond the scope  of this study. The emphasis
                                                     was therefore on the  selection and  arrangement of  the
                                                     discrete components and electronic packages to  minimize
                                                     the size of the demonstration unit. The basic design of the
                                                     system is essentially the  same  as used in  previous SAW
                                                     Vapor  Monitors.  The  four coated  SAW dual delay  line
                                                     devices were mounted in small, gold 1C packages.  The  lids
                                                     of each package  were  modified with short,  1/16"  ID, gold
                                                     plated gas inlet and outlet tubes to provide the toxic gases
                                                     access to the sensors.  A fifth SAW dual delay line, sealed to
                                                     prevent exposure to the ambient environment, was place in
                                                     a separate package.  In the demonstration unit, this  fifth
                                                     device was  used as a reference for  all other sensors to
                                                     compensate for changes in temperature and pressure.  The
                                                     output of the 4 SAW Sensor Array was integrated  with  a 4
                                                     channel frequency interface card to generate the measured
                                                         74

-------
frequency   differences,  Af,  and   with   an  onboard
microcomputer  (microcontroller) for data  analysis.

2.     Instrument  Configuration

 The system was designed with three circuit cards: a sensor
card, a  four channel frequency  interface card,  and  a
microcomputer  card.  The entire instrument will fit in an
enclosure  4-3/4"  x  8"  x  3", allowing  room  for  the
necessary  pumps,  valves and  gas  transfer lines.   The
system was designed for either battery operation or with a
120 VAC  50-60  Hz power  supply.    1/8"  Swagelok
bulkhead fittings on the enclosure provided gas inlet  and
outlet  to the  system.   Except for  the  stainless steel
Swagelok finings on the front of the enclosure,  all surfaces
in contact with  the gas  up to the SAW devices are either
Teflon or gold.

The four channel  microcomputer controlled  frequency
counter measures and reports the frequency of each SAW
sensor every two  seconds while controlling the solenoid
valves by  means  of  a  solid state relay.   For laboratory
evaluation  of the demonstration  model  SAW  Personal
Monitor for Toxic Gases, the counter  output is provided on
a 9600 baud RS-232C  serial communications line.   For
better control and  monitoring of the demonstration model,
and it's subsystems,  all communication with the unit  was
through the FtS-232  line and a personal computer with a
serial  communication port.  In a follow-on program,  a
different  communication  scheme will be devised so  that the
user will have the  option  of entering  all instructions
directly on the instrument.   Also, all concentration  data
and/or signals will be presented on visual  (LCD)  displays
or by audio alarms mounted  on the instrument enclosure.
There will  still  be the option of communicating  with the
SAW Personal  Monitor via a  personal computer to retrieve
data stored in memory.
 In the demonstration  unit, the onboard Octagon SB S-150
 microcomputer was programmed to control operation of
 the system, but not for analysis  of the  sensor array data.
 Development  of a sensor array data analysis program is
 planned for the follow-on effort.   With  the demonstration
 unit, the performance  of each SAW sensor,  and it's coating,
 was evaluated individually  against a specific toxic  gas.
 There are a number of experimental variables that  also
 require computer control and or analysis.  For example,
 due to the possible adsorption/desorption of ambient gases
 (especially water  vapor) on the coatings, the computer
 must continually determine the  actual  baseline  for each
 sensor,  by intermittently providing  clean, dry  (filtered)
 air to  the sensors.   The computer  must  also  store
 calibration data for each sensor and provide total exposure
 values on demand and/or activate  an alarm when certain
 values are exceeded.   Figure 1  provides a pictorial layout
 of a SAW Array Personal Exposure  Monitor.
SAW COATING SELECTION

 1 .     Selection of Candidate Coatings

A series of candidate  materials was selected for screening
as coatings for the SAW devices.  They were selected on the
basis  of their known reactivity with the toxic gases chosen
for evaluation.  The coatings selected for screening against
the reactive gases are given in Table 1 .
Table  1. Candidate  Coating  Materials  for
          SAW Sensors
       Candidate Coating
       Diphenylbenzidine
       2,4,  Dinitrophenylhydrazine
       o-Toluidine
       Triethylenediamine   (TEDA)
       Na[HgCl2] (hydrate)
       Pb(C2H302)2 '  5H20
       CuSO4 • 5H2O
       K[Ag(CN)2]
       Ninhydrin
                                         Reactive Gas
                                            NO2
                                            NO2
                                            NO2
                                            S02
                                            SQ2
                                            H2S
                                            H2S
                                            H2S
2 .
       Polyvinylpyridine    (PVP)
       Coating of SAW Devices
                                            HCI
 Each of the above coatings was applied to two 158 MHz Saw
 devices. Each SAW device to be coated was inserted into a
 suitable connector  mounted on  a  circuit  board  that
 contained the necessary electronics to operate the device
 and provide frequency signals to an external data aquisition
 system.  Prior to coating, each dual 158 MHz SAW device
 was  ultrasonically  cleaned  in isopropanol or chloroform,
 dried  in  a stream  of  compressed dry, zero  air, and
 positioned  in the  coating  apparatus. In all  but a few
 instances,  the coatings  were applied by a spray deposition
 technique developed by Microsensor Systems. The primary
 requirement is that the coating material must be soluble in
 a volatile solvent.  Zero air was used to generate a fine mist
 of the specific coating solution.  A mask was placed over the
 SAW device so that only the interdigitated delay lines were
 coated.

 The quantity of coating  material deposited on  each delay
 line was closely monitored by the computer data system
 which reported the  mass  of material deposited as  an
 increase in frequency, Af.  The amount of coating material
 applied was  held  closely to 250  KHz + 50  KHz.  The
 frequency shift, Af,  corresponds  to  coating thickness,
 assuming  uniform  surface coverage.  Once the coatings
 were applied, the SAW devices were covered and stored in a
 low  humidity (<  10%  RH)  environment  until  ready  for
 testing. As the candidate coating materials given in Table 1
 are generally  hygroscopic, it can be assumed that a certain
 amount of water will  be associated  with each coating and
 must be considered in subsequent gas interactions.
                                                        75

-------
                SAW ARRAY
        PERSONAL EXPOSURE MONITOR
             PICTORIAL LAYOUT
                    (PHASE
     REPLACEABLE
     SAW SENSORS
     (4 UNDERNEATH
     SCREW-ON LID)
             MICROCOMPUTER
             BOARD
   6VIS I 2 Ah
   4x2 X I 807.
RECHARGEABLE
BATTERY PACK
                                                      AMBIENT
                                                      VAPOR
                                                      INLET
   Figure 1.  Pictorial Layout of SAW Array Personal Exposure Monitor
                        76

-------
3.    Screening and Selection of Coatings for SAW Test
and Evaluation

The following  criteria were  established to  define  a
successful candidate  material:  (1) that a  coating give a
frequency shift equivalent  to  a  100:1 signal to noise ratio
when exposed to the toxic gas at a concentration  of
approximately 100  ppm for  1  minute or  less;  and   (2)
that the coating  react irreversibly with the test gas.  With
a baseline noise level of approximately 15 Hz, a 100:1
signal to noise ratio would be  equivalent to a frequency
shift on the order of 1500  Hz.  Thin film coatings showing
less response would not have sufficient  sensitivity nor
capacity to be useful in field monitoring applications.

A calibrated cylinder of each of  the test gases (NO2,  SO2,
HCI, H2S, NHa)  in  air  was  obtained from  the Scott
Specialty Gas Co.   The concentration of each gas source
was:

         Toxic Gas     Source Concentration
            HCI           103.3 ppm
            NHs          106.5 ppm
            H2S          100.6 ppm
            NQ2          108.0 ppm
            SQ2          102.5 ppm

 By simple dilution of  the compressed gas with clean, dry,
zero air, a steady state concentration at any  value less than
100 ppm could be easily  prepared.  A constant gas flow
rate of 200 cc/min was maintained.  A valve was arranged
so that  clean air, or a known concentration of the specific
test gas, could be alternately delivered to the sensor.  A lid
with 1/8" gold gas inlet and outlet tubes was placed over
the device and  was connected to the output of the gas
dilution  chamber. The frequency output of the dual delay
lines could be monitored using a small frequency counter.

In the tests, a coated SAW device was first exposed to clean,
dry air at 200 cc/min  to  obtain  a  steady  baseline
frequency. The valve was then turned to expose the sensor
to a known concentration of the toxic gas, at the same flow
rate, for a pre-determined period of time.  The sensor was
then exposed once again to clean, dry air to establish a new
baseline.  If  the  clean air  baseline, after exposure to the
toxic gas, was significantly different from the initial clean
air baseline,  it was assumed the change in  frequency was
due to  an increase  in coating mass resulting  from  the
irreversible reaction  with the challenge gas.  If there was
no significant change  in SAW frequency, the device was
exposed to higher gas concentrations for longer periods of
times.  If there was still no permanent change in  baseline,
it was assumed there was no reaction and that the coating,
in its present  form at least, was ineffective.   All tests were
performed with dry air, unless  otherwise specified in the
text.

The results of the initial screening tests are given in Table
2.  They show that for each toxic gas there was at  least one
coating  that  gave an  acceptable  response.  However,  in
several  instances there were rather unexpected results.
For example,  NC-2 did not appear to react  at all with 2,4
Dinitrophenyl  hydrazine unless there was a relatively high
moisture content (= 80% RH) in the carrier gas.    It was
also surprising that H2S did not react readily  with the lead
acetate coating, even though we have observed this surface
reaction in a previous study.   Copper  sulfate seemed
unreactive initially, however, after repeated cycling  it did
react to give a very large and permanent frequency shift.
The reaction, or lack of it, in each case may depend to a
large extent upon the amount of water present in the  film.

Table  2.  Results of Initial  Coating Screening
           Test
      (Thickness of all coatings approx. 250 Hz)
Coating
Diphenylbenzidine
2,4,  Dinitrophenyl
   hydrazine
o-Toluidine
TEDA
Na[HgCl2]
Pb(C2Hs02)2*'
CuS04"**
Ninhydrin
CoCl2
PVP
                                 Af    Stable
                  Conc./Time     (Hz) Reaction
            NC-2 50 ppm/60 s.   900    No
            NO2 50 ppm/60 s. 2,800   Yes

            NC>2 50 ppm/60 s. <100
            SO2 50 ppm/60    1,000   Yes
            SO2 50 ppm/60 s.
            H2S
            H2S 50 ppm/60 s. 2,000   Yes
                 50 ppm/60 s.   100
                 50 ppm/ 20 s 2,700   Yes
            HCI  (known to react)
       Reacted only  in presence of high RH
* *     Reacted in a previous study, but now
       Reaction occurred after repeated H2S exposure

 Based on the results of Table 2, the following coatings were
 selected for more careful  evaluation.  2,4 Dinitrophenyl-
 hydrazine was not used for NO2-  Rather TEDA was used for
 both SO2 and NO2-
                             Toxic Gas
                               HCI
                           NO2andS02
                               H2S
       Coating Material
       Polyvinylpyridine  (PVP)
       Triethylenediamine (TEDA)
       Copper sulfate (CuS04)
       Cobaltous chloride (CoCl2)
TEST AND EVALUATION OF SAW SENSORS AS MONITORS FOR
TOXIC GASES
1.
Coating of SAW Sensors
The coating procedure  used was  the same as  described
above.  Both SAW delay lines on each device were coated
simultaneously, and the amount deposited was  measured
and recorded.  The identification number of each device and
the  coating mass  (in terms  of frequency shift,  Af) are
given in Table 3. The coatings applied are very thin, on the
order of a micron or so in thickness, on the average.

2.     Evaluation of SAW Sensors  as Monitors for Toxic
       Gases

The frequency difference, Af, of each SAW device being
tested was input to  a Apple  Macintosh computer  where the
data was collected and displayed. The test system evaluated
                                                       77

-------
only one sensor at a time against a single toxic gas.  Even
though each of the coating materials being tested could very
likely react with  more than one gas, binary gas mixtures
and interference  studies were  not  included   in this
preliminary investigation.  Interference studies will  be a
part of  the follow-on study, using  multiple sensor arrays
and other techniques to address the problem of sensor
specificity.

The gas dilution chamber was again used to deliver known
concentrations of each  test gas to  the SAW sensors at a
constant flow rate of  200 cc/min at ambient pressure, and
a constant "baseline" frequency established for each  SAW
device by exposing it to a clean, dry air stream.   Once a
constant baseline frequency was established, the sensor
was exposed to a predetermined "dose" of the selected toxic
gas.  The size of the dose could be varied from 10 to 100
ppm over any selected time interval. After exposure to the
toxic gas, the sensor was again exposed to clean,  zero air
until a  new baseline frequency was established.  The
difference  between  the  initial  baseline  and the  final
baseline was  taken  as the  frequency  shift due to the
irreversible  reaction  of the  toxic  gas with  the  coating
material.  The magnitude of this frequency shift could  be
correlated with the amount of  toxic gas interacting with the
sensor.

The intent  of  the tests  was  to quickly look for  order of
magnitude  changes in  frequency and general  reproduci-
bility of  performance when exposed to moderate changes in
gas concentrations; i.e., to identify  coatings that could  be
used in a more comprehensive follow-on development
program.  This   study  did  not  include  a   careful
characterization of each coating reaction.  In any event an
accurate characterization of the surface reactions would be
difficult without  a more careful control  of trace water,
both in the hygroscopic coating materials and  the gas
delivery system.
 3.
Exposure of NHs to CoCl2 Coated SAW Sensor
The SAW devices were at ambient temperature and  thus
subject to the room temperature  fluctuations (=  25° +/-
1°C).   Although  a reference  SAW  device was used  to
compensate for both temperature  and  pressure changes,
the compensation is not exact,  and  may have caused some
small, random drift in device background frequency.  These
slow  changes  occurred in cycles of many minutes and  thus
did not adversely effect the measurements.  Even  though a
number of the coating materials have a  small  volatility,
the signal drift reflected  "apparent" increases as well  as
decreases in  weight.    Thus volatility did  not  have a
measurable effect on the measurements.  Once a device was
equilibrated  with  the  laboratory  environment  (temper-
ature and pressure) the slow baseline drift was usually  on
the order  of ± 50 Hz.  In addition to temperature changes
and the possibility of volatility, the baseline drift may  also
be due in  pan to changes  in gas flow rate (due to changes in
flow through the non-precision needle valve used to set the
flow  rate).   Even with the small  observed  background
drift,  the following data show that system performance was
excellent and clearly able to detect and  monitor changes in
                                                    SAW frequency  upon exposure to the challenge  gases.
                                                    Sensor  drift will  be  corrected  for in  the  follow-on
                                                    Personal  Monitor development program.

                                                    An example of data  for the exposure of ammonia to the
                                                    CoCl2 coated SAW devices is shown in Figures  1.  An
                                                    exposure of 20 ppm NHs for 20 seconds was selected for
                                                    testing the  CoCl2  coated sensors.  When the NHs  was
                                                    introduced,  there was a large initial decrease in  SAW
                                                    frequency followed by a rapid increase.  Each point on the
                                                    curve corresponds  to a 2 second time interval. After 20
                                                    seconds,  when  the  clean air  at  200 cc/min  was again
                                                    introduced,  Af  continued to increase through a small
                                                    maximum and then  level off to  a new, higher, baseline
                                                    value.  The initial negative "spike" in the  Af vs time plot
                                                    may be due in part to disruption and  re-establishment of a
                                                    constant  gas flow rate, while the subsequent increase  in Af
                                                    most probably results from both adsorption  and reaction of
                                                    the NHs  witn tne CoCl2 coating.  The maximum may  result
                                                    from a more gradual  desorption of non-reacted NHs  from
                                                    the coating.  The equilibrium frequency shift values for all
                                                    devices are shown in Table 4.
                                                          160000
                                                      0)
                                                      -
                                                      CT
                                                      o
                                                          150000 -
                                                          140000 -
                                                          130000
                                                                              .
                                                                      V
                                                                            200  400  600  800  1000 1200 1400

                                                                                     Time (seconds)
                                                      Figure 1.  Frequency Shift (Hz) vs. Time for Repeat
                                                                Exposure of CoCIa Coated SAW Device
                                                                (9024-11) to  20 ppm  NHs for 20 Sec.
                                                      Table 3. Thickness of SAW Device  Coatings
                                                       Coating Material
                                                           PVP
                                                           CuSO4



                                                           CoCl2


                                                           TEDA
   Coating Thickness (KHz)
Device Number      Side "A"
   9024-1
   9024-2
   9024-3
   9024-7
   9024-8
   9024-9
   9024-10
   9024-1 1
   9024-12
   9024-4
   9024-5
   9024-6
255
198
198
149
150
196
136
1 12
106
149
178
300
                                                         76

-------
Table  4.  Frequency Shifts for CoCl2  Coated  SAW
         Devices  Upon  Repeated Exposure to
         20 ppm  NH3  for 20 seconds
                                           158000
  Device Number
     9024-10
   (Coating 112 KHz)

     9024-11
   (Coating 136 KHz)

     9024-12
   (Coating 106 KHz)
Exposure
Frequency Shift
 a. - d.   (dose optimization test)
   e,
   f.
   a.
   b.
   c.
   a.
   b.
   c.
   d.
   e.
   f.
 1,200  Hz
      OHz
 4,000  Hz
 4,000
 1,000
 2,600
 2,000
 1,200
 1,600
 2,000
Hz
Hz
Hz
Hz
Hz
Hz
Kz
                                        OHz
From the data in Table 4  it  is evident that  CoCl2 coated
SAW devices show  large (Kilohertz),  irreversible shifts
in frequency when exposed to small doses of  ammonia, and
that with continued  exposure the coatings saturate as
expected.   Even allowing for the variation  is response of
the different sensors,  the sensitivity of the CoCl2 coatings,
i.e., those with some  residual capacity, is on  the order of 5
to 10 Hz/ppm/sec.  Considering that the background noise
level  of the SAW sensors  is  on the order of 15 Hz, a ten
seconds exposure of  a sensor to  1 ppm NHs would give a
signal of  better than 50 Hz, at least three  times the
background noise. Thus the CoCl2 coatings have more  than
enough sensitivity  to detect  ammonia  at concentrations
below the  OSHA Exposure Limit  of 50 ppm  NHs  for an 8
hour weighted average.

4.     Exposure of CuSCU Coated SAW Sensor to H2S Gas

The test procedure was essentially the same as described
above.  Typical results are shown  in Figure 2 for device
9024-7.  H2S shows a decrease in SAW frequency  with
exposure rather than an increase in Af as  observed  with
the reaction of NHs  with  the CoCl2  Also,  there  was no
initial "spike" in Af when the challenge gas was introduced.
Upon repeated exposure, the frequency shifts  became
progressively  smaller, due to saturation of the  reactive
sites of the CuSO4 coating.

The Af  values for the CuSCM coated  sensors  9024-7 and
9024-8  are  given in  Table  5.   SAW device  9024-9
apparently  became defective during the coating  process.
SAW device 9024-7 was exposed five times to 20 ppm of
H2S  for 20 seconds.  With  the initial dose  of  H2S, Af
decreased by 1,400 Hz.  The second exposure decreased Af
by only 400 Hz.  Subsequent doses caused  essentially no
further  change in Af.  Thus  the CuSC>4   coatings  were
essentially  saturated  by a  single  20 ppm dose of H2S for
20 seconds.
                  5
                  ~
                  —
en
                                           157000 -
                       156000 -
                                                                155000
                       ]?
                      -,
           0   200  400  600  800  1000 1200 1400

                         Time (seconds)

Figure 2.  Frequency Shift  (Hz) vs. Time for Repeat
          Exposure of CuSO4 Coated SAW Device
          (9024-7) to 20 ppm H2S  for 20 Sec.
                                     Table  5.  Frequency  Shirts for CuS04 Coated  SAW
                                                Devices Upon Exposures  to 20 ppm  H2S
                                                for 20  seconds
                                        Device Number
                                            9024-7
                                         (Coating 149 KHz)
                                            9024-8
                                         (Coating 150 KHz)
                                            9024-9
                                         (Coating 196 KHz)
                                               Exposure     Fjeguencv Shift
                                                  a.
                                                  b.
                                                  c.
                                                  i'..
                                                  e
                                                  a.
                                                      1,400 Hz
                                                       400 Hz
                                                       100 Hz
                                                          0 Hz
                                                          OHz
                                                      1,400 Hz
                                                (device defective after coating)
                                     Thus the CuSCvj coated SAW devices, like the CoCl2 coated
                                     devices,  do  give  large (KHz),  irreversible  shifts in
                                     frequency when exposed to small doses of an appropriately
                                     reactive gas, and that with continued exposure the coatings
                                     saturate as expected.  The sensitivity of a newly prepared
                                     CuSO4 coating is on the order of 3 to 4 Hz/ppm/sec.  With
                                     background  noise  on  the order of  15  Hz,  a  ten  second
                                     exposure to  1 ppm  H2S would give a signal of around  30 to
                                     40  Hz, equivalent  to a signal to noise ratio of 2:1.   The
                                     detection limit  of this coating is thus also is well below the
                                     OSHA Exposure  Limit  of 20  ppm  H2S for an  8  hour
                                     weighted average.

                                     5.     Exposure of TEDA Coated SAW Sensor to SO2 Gas

                                     The procedure used to test the TEDA coated SAW sensors
                                     with S02 was the same as described above.  Typical results
                                     are shown in Figure 3 for device 9024-6.  The results for
                                     device 9024-5 were similar.   SAW device 9024-4  was
                                     reserved for  testing with  N02, which was expected to  react
                                     with TEDA  in  much the  same way as SO2.  A rather
                                     unexpected behavior was observed  when the TEDA coated
                                     devices were initially exposed to SO2-   For the first few
                                                      79

-------
exposures of 20 ppm  SO2 (20 seconds),  the coatings did
not  respond  significantly.    After  several  repetitions,
however, the coatings did begin to respond with positive
shifts in Af with the continuing exposure.  Thus it appears
there was a "conditioning" period, after which the coatings
began to respond.   The  "conditioning" must be associated
with some chemical change in the coatings upon exposure to
the test  gas, or to the zero  air,  most likely involving
associated water. As each device,  after being coated, was
covered  with a close  fitting  lid  (but  not hermetically
sealed) and  stored in a = 10% RH  environment, they must
have  adsorbed  some water  vapor (or  perhaps another
ambient gas) which was subsequently desorbed from  the
coatings  by the dry (<  1%  RH) zero air and/or the  dry
sample  (S02) air. This "conditioning" or  "ageing" effect
was  not further explored at this time, but  will of necessity
be investigated in  the follow-on  study in  order to provide
coatings that behave predictably and reproducibly.
        280000
  3
        270000 -
        260000 -
        250000
              0   200   400  600  800 1000 1200  1400

                               Time (seconds)

 Figure  3(a).  Frequency Shift (Hz) vs.  Time for Repeat
              Exposure of TEDA Coated  SAW Device
              (9024-6) to 20 ppm SO2 for 20 Sec.
              (First  exposure, a)
        268500
  £    267500 -
        266500 -
        265500 -
        264500
              0   200  400   600  800  1000 1200 1400

                              Time (seconds)
 Figure  3(b).  Frequency Shift (Hz) vs. Time for Repeat
               Exposure of TEDA Coated SAW Device
               (9024-5) to 20 ppm  SOa for 20 Sec.
               (Exposures d to h)
After the  initial  induction period, the frequency  shift vs
time  plot  in both  Figure 3(b) shows an increase in  the
SAW baseline with each 20 second dose of SO2, after the
initial "spike" in  Af.  Device 9024-5 was allowed to stand
in the test  apparatus for  approximately two  hours with
continuous exposure to zero air, before the  run.  Even so,
it wasn't until exposure f that the device began  to respond.
Somewhat  similar behavior  was  observed  for  device
9024-6,  however  the  conditioning  period  was  much
shorter.   For  both  device  9024-5 and 9024-6, once  the
coatings became  reactive, the shifts in frequency were
regular and  irreversible.

The frequency shifts are given in Table 6.  The data clearly
show the induction period during which there was no effect
of SC-2 exposure, and the subsequent increases in Af when
reaction began to occur. If we assume an average response
of 1,200 Hz for  device 9024-5 and  1,800  Hz for  device
9024-6, the  sensitivities  are  approximately  3  and  4.5
Hz/ppm/sec, respectively.  The coating on  device 9024-6
was  a third again the mass of  the coating on 9024-5 (300
KHz  to 178  KHz), thus one would  expect the sensitivity to
SO2  to be a third again as high, which was observed. Thus
the  two  coated  devices had  essentially  equivalent
sensitivities.
Table  6. Frequency Shifts for TEDA  Coated SAW
          Devices  Upon Repeated Exposure to
          20  ppm  SO2 for 20 seconds
   Device Number
      9024-5
   (Coating 178 KHz)
       9024-6
    (Coating 300 KHz)
Exposure
   a.
   b.
   c.
   d.
   e.
   f.
   g.
   h.
   a.
   b.
   c.
   d.
   e.
   f.
Frequency Shift
     OHz
     OHz
     OHz
     OHz
     OHz
   800 Hz
 1,400 Hz
 1,000 Hz
     OHz
     OHz
   200 Hz
 1,600 Hz
 2,000 Hz
 1,800 Hz
With sensitivities  of about 3 to 4 Hz/ppm/sec, depending
upon coating thickness, and a background noise level of 15
Hz  for the SAW  devices, the sensors  should  ultimately
detect  concentrations  of S02 as low as  1 ppm within 10
seconds  at a signal to noise ratio of about 2:1.  With this
sensitivity,  these  coatings should easily detect SC>2 at or
below the OSHA Exposure Limit of 5 ppm SO2 for an 8 hour
weighted average.

-------
6.     Exposure of TEDA Coaled SAW Sensor to NO2 Gas

 It was anticipated that TEDA would respond to NO2 in much
the same manner as to SO2; however, the data for the one
available sensor  showed quite different behavior.  First, no
conditioning period was observed. The first 20 second dose
of  20 ppm  NO2 gave  a relatively  small but definite
increase in SAW frequency which apparently saturated the
sensor, as no further  increase in  Af was  observed  with
additional  exposure to N02-  The frequency shift data are
given in Table 7.  The  baseline shift of approximately 350
Hz for an exposure  of 20 ppm NO2  for 20  seconds,  is
equivalent to about  1  Hz/ppm/sec,  well  below the
sensitivity to  S02-  With a sensitivity of approximately 1
Hz/ppm/sec,  and  a background noise level of 15 Hz, the
TEDA coated sensors would have to be exposed to 1 ppm NO2
for over 30 seconds to  give a 2:1 signal to noise ratio.  In
addition, the  film apparently has a very  low capacity for
N02  (i.e.,   saturating  at  a   very  low  exposure
concentration).  TEDA  is therefore  of only marginal utility
as a dosimeter coating for NO2-
Table 7. Frequency Shins for TEDA  Coated
         SAW  Devices Upon  Repeated  Exposure
         to 20 ppm  NO2 for  20 seconds
   Device Number      Exposure
     9024-4             a.
   (Coating 149 KHz)    b. - g.
Frequency Shift
   350 Hz
     OHz
7.     Exposure of PVP Coated SAW Sensors to HCI Gas

Device 9024-1  was given 5 separate exposures to 20 ppm
of HCI for 20  seconds, over approximately a 30 minute
period, with no apparent reaction of the  HCI with the PVP.
We know from previous studies that surface films of PVP
do react with  HCI,  thus the lack of response  must  be
similar to the "conditioning" period observed for S02 gas
on TEDA. To accelerate the reaction, the  PVP coated device
9024-1 was exposed to a higher concentration of  HCI
(100 ppm) for 2 minutes.   The result  was a very  large
increase in Af,  over 30,000 Hz in  the 2 minute period, as
shown in Table 8.  A second large dose (100  ppm over a
60 second period) further increased Af  by  only 4,800 Hz,
indicating that the PVP coating was approaching saturation.
The estimated  sensitivity, based on the  30,000 Hz shirt is
about 3 Hz/ppm/sec.

Device 9024-2  was exposed to repetitive doses of HCI at a
concentration of 25 ppm  for 20 seconds.   The results given
in Table 8 indicate no conditioning period was needed. The
very first exposure gave an  increase of  about  900 Hz and
appeared to be stable with time.  Subsequent exposures also
increased Af, until the film began  to saturate.   Sensitivity
based on the initial  exposure is about 2 Hz/ppm/sec.
Device 9024-3 did  require  a conditioning period when
exposed to 25 ppm HCI  for 20  seconds.    HCI exposures
were  increase to 50  ppm  for 30,  60  and 90 seconds,
before an increase  in  Af was observed.  With  the final
exposure, a frequency increase of  approximately 6,400  Hz
was observed.
Table  8. Frequency Shifts for PVP Coated SAW
          Devices  Upon Repeated Exposure to HCI
                                           Frequency
   Device Number           Exposure           Shift
      9024-1          a.(20 ppm  20 sec)         0  Hz
    (Coating 255 KHz)  b.(20 ppm  20 sec)         0  Hz
                      c.(20 ppm  20 sec)         0  Hz
                      d.(20 ppm 20 sec)          0  Hz
                      e.(20 ppm  20 sec)         0  Hz
                      f.(100 ppm120 sec) 30,000  Hz
                      g.(100 ppm 60 sec)  4,800  Hz
      9024-2          a.(25 ppm  20 sec)      900  Hz
    (Coating 198 KHz)  b.(25 ppm  20 sec)      600  Hz
                      C.(25 ppm  20 sec)      400  Hz
                      d.(25 ppm  20 sec)      600  Hz
                      e.(25 ppm  20 sec)      400  Hz
                      f.(25  ppm 20 sec)      200  Hz
      9024-3          a. (25 ppm  20 sec)         0  Hz
    (Coating 198 KHz)  b.(25 ppm  20 sec)         0  Hz
                      c.(25 ppm  20 sec)         0  Hz
                      d.(50 ppm  30 sec)         0  Hz
                      e.(50 ppm  60 sec)         0  Hz
                      f.(50  ppm 90 sec)    6.400  Hz


The sensitivities of the PVP coated SAW devices were in the
range of  1  to  3  Hz/ppm/sec.   Device 9024-1,  with the
greatest  apparent sensitivity (3 Hz/ppm/sec),  had  the
highest, coating mass, as would be  expected.  Thus the
results for the three devices  are consistent.   With  a
sensitivity of 1  to 3 Hz/ppm/sec, a sensor would have to
be exposed to 1 ppm HCI  for 10 to 30  seconds to give a 2:1
signal to  noise  ratio.  The PVP  films  do appear to have a
high capacity for HC1, as  evidenced by the 30,000 Hz  shift
for device 9024-1.   Considering  that  the OSHA Exposure
Limit is 5 ppm  HCI for an 8 hour  weighted average, the
PVP coating should  be considered a good candidate for
further development as a coating for monitoring acid gases.
                       CONCLUSION

                       In the evaluation of the various SAW coatings it was found
                       that for each toxic gas, except NO2, a relatively  large,
                       easily measured  SAW response was observed  when an
                       appropriate coating was  exposed  small concentrations.
                       The measured sensitivities show that each toxic gas studied
                       (except NO2) could be detected by a SAW sensor well below
                       the  "action  level"   set by OHSA,  when monitored for a
                       period of one minute or less.  The candidate coatings, toxic
                       gases, and the respective OSHSA exposure limits, are:

                                                             OS'HA Exposure
                                                              Limit -  8  hour
                       Candidate Coating             Toxic Gas   Weighted Ave.
                       polyvinylpyridine  (PVP)       HCI           5 ppm
                       triethylenediamine (TEDA)  NO2  and  SO2       5 ppm
                       copper sulfate (CuSO4)         H2S         20 ppm
                       colbaltous  chloride (CoCl2)     NHs         50 ppm

                       The  study  thus  successfully achieved it's  objective  of
                       demonstrating that: (1) the  SAW sensors and necessary
                       support  electronics can  be appropriately  miniaturized;
                                                      81

-------
(2) a number of successful coatings are readily  available
and  others can certainly be identified in the literature,  or
developed, for additional toxic gases; and (3) SAW sensors
are sufficiently  sensitive to meet OHSA requirements,  at
least for the toxic gases selected for this demonstration
study.  A number of  technical problems and/or  potential
limitations  of  the   technology  were identified  and
approaches suggested  for  their solution.  Based on the
results  of  this  program, we  conclude  that a prototype
Surface Acoustic Wave  Personal Monitor for Toxic Agents
could be  readily  developed in a follow-on program.   In
addition  to  use as a Personal Monitor, such  a  small,
sensitive and rugged solid state instrument could possibly
find  other  applications in  the field screening  for toxic
chemicals.   In all applications however,  the usefulness  of
SAW sensors will increase with the continued development
of more sensitive and selective device coatings.
ACKNOWLEDGEMENT

This research was supported by the Department of Health
and Human Services, Public Heath Service, Small Business
Innovation Research (SBIR) Program, under Phase I Grant
No.  1R43 ES5039-01A1.

REFERENCES

1.   H. Wohltjen and R.E. Dessy, "Surface Acoustic Wave
    Probe for Chemical Analysis I. Introduction and
    Instrument  Design",  Anal. Chem., 51(9), 1458-
    1464  (1979).

2.   H. Wohltjen and R.E. Dessy, "Surface Acoustic Wave
    Probe for Chemical Analysis II.  Gas Chromatography
    Detector",  Anal.  Chem.,  51(9),  1465-1470  (1979).

3.   H. Wohltjen and R.E. Dessy, "Surface Acoustic Wave
    Probe for Chemical Analysis III. Thermomechanical
    Polymer Analyzer", Anal.  Chem., 51(9),  1  470-
    1475  (1979).

4.   H. Wohltjen and H. Ravner, "The Determination of the
    Oxidative Stability of Several Deuterated Lubricants
    by an Electronic Gas Sensor", Lubrication
    Engineering,  39(11),  701-705  (1983).

5.   (Invited) H. Wohltjen, "Chemical Microsensors and
    Microinstrumentation",  Analytical  Chemistry,
    56(1),   87A-103(1984).

6.   H. Wohltjen, "Mechanism of Operation and Design
    Considerations for Surface Acoustic Wave Vapor
    Sensors", Sensors and  Actuators, 5 (4), 307-325
    (1984).

7.   H. Wohltjen, W. R. Barger, A. W. Snow, and N. L.
    Jarvis, "A   Vapor Sensitive Chemiresistor Fabricated
    with  Planar Microelectrodes and a Langmuir-Blodgett
    Organic Semiconductor Film", IEEE Trans, on Electron
    Devices. ED-32,  No.  7,  1170-1174  (1985).
8.   W.R. Barger, J.F. Giuliani, N.L. Jarvis, A.Snow, and H.
    Wohltjen, "Chemical Microsensors- A New Approach
    for the Detection of Agro Chemicals", Environ. Sci.
    Health,  B20(4),  359-371   (1985).

9.   W.R. Barger, A.W. Snow, H.  Wohltjen, and  N. L.
    Jarvis, "Derivatives  of Phthalocyanine  Prepared for
    Deposition as Thin Films by the Langmuir-Blodgett
    Technique",  Thin  Solid  Films, 133, 197 206  (1985).

10. A.W. Snow, W.R. Barger, M. Klusty, H. Wohltjen, and
    N.L. Jarvis, "Simultaneous Electrical Conductivity and
    Piezoelectric Mass Measurements on Iodine-Doped
    Phthalocyanine Langmuir-Blodgett  Films", Langmuir,
    2,   513-519  (1986).

11. D.S. Ballantine, S.L. Rose, J.W. Grate, and H.
    Wohltjen, "Correlation of SAW Coating Responses with
    Solubility Properties and Chemical Structure Using
    Pattern Recognition",  Anal. Chem.  58,  3058 (1986).

12. G. S. Calbrese,  H. Wohltjen, and M.K. Roy, "Surface
    Acoustic Wave  Devices as Chemical Sensors in
    Liquids", Anal.  Chem. 59,  833  (1987).

13. J. W. Grate, A. W. Snow.  D. S. Ballantine, Jr., H.
    Wohltjen, M. H. Abraham, R. A. McGill, and P. Sasson,
    "Determination of  Partition  Coefficients  from Surface
    Acoustic Wave Vapor Sensor  Responses and
    Con-elation with Gas-Liquid  Chromatographic
    Partition  Coefficients", Anal.  Chem. 60, 869  (1986).
                                                         82

-------
                                                           DISCUSSION
WILLIAM BOWERS: You showed some data on individual sensor responses
for single exposures. Have you done any interference effects on some of these?
I am glad to see you're going to resonators now.

N. LYNN JARVIS: We did no interference studies in this particular program.
You could probably tell that many of the coatings used would respond to more
than one vapor. These were not selective coatings in that sense. Selectivity is
much more difficult to get. That's why we end up using an array of sensors to get
the selectivity. Resonators are much, much nicer.

MICHAEL CARRABBA: When you put the coating on these SAW devices,
and the coating goes over electrodes, is the area on the whole surface sensing the
weight or is it just the area between the electrodes, or the area on the electrodes'?

N. LYNN JARVIS: The whole area surface senses the weight. The wave will
cover most of the surface. Most of the surface is sensitive and you get a response.

PHILLIP GREENBALM: Have you tried attaching antibodies to these? And
if not, do you think that would be a problem?

N. LYNN JARVIS: We have not and you could certainly attach them. The
problem is that antibodies are very large, and you're trying to attack very small
molecules with the antibody. You may get a very small signal i.e., the change in
weight is very small. Sensitivity might be fairly low in this case. It would not be
a way we would probably choose to go with these particular sensors. There are
probably better sensors for that.

MAHADEVA SINHA: Are these things disposable once you use them? After a
certain while do you throw them out?
N. LYNN JARVIS: Yes. In this system, once a sensor is used up. we propose to
it throw it away and plug in a new one.

MAHADEVA SINHA:  You talked about the reversibility of some of the
reactions. What did you mean by that?

N. LYNN JARVIS: There are two ways you can go with a coating on a SAW
dev ice. You can use coat ings where the vapors absorb onto the coating, depending
on solubility characteristics and other factors. They will absorb when the vapor
is present. When the vapor challenge is removed, it desorbs again from this
polymer and is removed. So it's a completely reversible system with certain
vapor coating combinations. You can use a coating where there is no chemical
reaction. However, if  you have a  chemical reaction, then it is completely
irreversible, which is what we're looking for in this particular application. In
some applications you want reversibility; in some you don't, depending on the
intended use.

EDWARD POZIOMEK: In your last viewgraph and also in your comments
you  mentioned the  possibility of the  wide applications to  environmental
measurements, and you said something about putting a SAW down a well.
Perhaps you could comment on the state of this SAW technology for use in
liquids, because the applications presented here were for vapors or for gases.

N. LYNN JARVIS: If we put a sensor in a well, it would have to be within the
well  headspace to be monitored, not the liquid. The technology for SAWs in
liquid is very poorly developed, and is just barely beginning. We know of no
really effective way to monitor using a SAW in solution.
                                                                     83

-------
                               ARRAYS OF SENSORS AND  MICROSENSORS
                         FOR FIELD SCREENING  OF UNKNOWN  CHEMICAL WASTES
                   W.R. Penrose, J.R. Stetter, M.W.  Findlay,  and W.J.  Buttner
                         Transducer Research, Inc., Naperville,  IL 60540

                            Z.  Cao, Illinois Institute of Technology,
                           Department of Chemistry,  Chicago,  IL 60616
Abstract

The high cost of laboratory-based analysis  has
driven  the  development  of  rapid  screening
methods  for  hazardous  chemicals  in  unknown
wastes.  Screening methods permit the  "triage"
of  samples  into those that  (a)  contain   no
regulated   wastes,    (b)   definitely  contain
regulated  chemicals,  or  (c)   are  ambiguous.
Only  the   last  category  requires   detailed
analysis.

The requirements of portability and ease  of  use
place extraordinary demands on the designers of
analytical instruments.  In this paper, we will
discuss   several   approaches   to   obtaining
qualitative   analytical  data  from   multiple
sensors  or  highly-selective  sensors.   These
are:  (a) a  sensor with  a  selectivity 1000-
10000  times   greater  for   chlorinated   or
brominated  compounds  than for unsubstituted
ones;  and   (b)   pyrolysis-EC,   which   uses
catalytic pyrolysis,  arrays of electrochemical
sensors,  and pattern recognition  methods  to
identify  pure  chemicals and  mixtures.    Two
applications  of the latter are described,  the
rapid identification  of  chemical  vapors,  and
the grading  of  grain according to "odor".
Introduction

The high cost of laboratory-based analysis  has
driven the  development  of   rapid  screening
methods for  hazardous  chemicals  in   unknown
wastes. A screening method is one that can be
done on-site,  by  non-chemists, inexpensively
and safely.   On  the other hand,  a screening
method is less likely to provide the definitive
data that  a  full  laboratory analysis,  perhaps
requiring  GC/MS or  ICP,  might  give.    In  the
case where no information is available, however,
even  limited information  can  be of  value,
especially if it  is used to  supplement  data
gathered from other  sources.   For example,  a
suite of simple screening methods may be used for
the "triage"  of unknown samples into positive,
negative, and ambiguous groups.   Often,  the
nature of the chlorinated compounds may be known
from purchase or production records, so that only
the ambiguous category may require  detailed
analysis. Screening methods may also be useful
for confirming conclusions that  have already been
drawn from independent data, for example, that a
collection of similar barrels do indeed contain
the same materials.

The will ingness to accept reduced certainty for
the sake of economy and practicality opens the
door to a wide variety of useful techniques that
can be used in the field.  In this paper, we will
describe two such methods.

A unique semiconductor sensor has been found that
is very sensitive to chlorinated and brominated
organic compounds (1-3).  It shows no detectable
response to hydrocarbons, oxygen- or nitrogen-
containing  organic compounds, or fluorocarbons.

A  second method  that has given  us promising
results has been catalytic pyrolysis of chemical
vapors combined with electrochemical detection.
Compounds that are not normally thought of as
electrochemical analytes, such as chloroform or
cyclohexane, can be partially oxidized on a hot
platinum surface  (4).   The  volatile products
always include some that give a response on a
porous-electrode electrochemical sensor. We have
confirmed over several years that the products of
the pyrolysis are reproducible for most organic
and some inorganic compounds when the conditions
are kept reasonably constant (5).  We have also
                                               85

-------
confirmed  the critical  requirement that  the
products    are    independent    of    analyte
concentration, at  least at concentrations  of
below 200 ppm.  We call this method pyrolysis-
EC.

The present  embodiment of pyrolysis-EC  is  an
instrument we  call  the CPS-100.   This device
uses an  array of electrochemical  gas  sensors
with different, but overlapping, selectivities.
The incoming gases  are  pyrolyzed over  noble
metal     catalysts    heated    at    controlled
temperatures.  The operation of the instrument
is orchestrated by a fairly  powerful  computer
which  can  perform  pattern  analysis  on  the
resulting data.  In  this paper,  we report  the
results  of  a study on pattern  recognition  of
odors in spoiled grain.  The unique properties
of  neural  networks  have been  shown   to  have
significant potential for handling low-quality
information.    On   reflection,   this   unique
application  is  not  so  different  from  the
problems   encountered   in   classifying   and
handling hazardous wastes.

A simplified implementation of pyrolysis-EC has
also been tested that uses  a single sensor  and
a single catalytic filament.  This drastically
simplified   system  was   still    capable   of
distinguishing many  organic chemicals.  With
fewer parts  and lower power  consumption,  this
simplified configuration  may be  suitable  for
selective hand-held  vapor monitors.
Experimental Methods

Organochlorine sensor.  The sensor was made by
mounting a coil of platinum wire on a threaded
base.  A separate platinum wire is also mounted
on the base  and  located  axially within  the
coil.  A mixture of lanthanum oxide,  lanthanum
fluoride, and a binder was applied to the coil.
The  coil  was slowly  heated with  an  electric
current until a reaction occurred, forming the
active material. The  sensor is used by heating
it  to  550   °C  with   an   electric   current;
conductivity  is  measured between  the  heating
coil  and  the  separate  platinum electrode.  When
the sensor contacts the vapor of a chlorinated
organic compound, the conductivity increases.
A  simple  circuit  can  be  used  to provide  a
voltage output  which is  proportional  to  the
concentration.

Permeation  device.    The  permeation  sampler
consisted  of  a   bundle  of   0.025"   o.d.
dimethyl si licone tubing  (Silastic, Oow-Corning)
(Figure 1).  The bundle could be placed in an
aqueous sample containing dissolved organic or
organochlorine compounds. A continuous flow of
air was circulated through the  lumen of the
tubing, and organic material diffusing inward
through the silicone membrane entered the gas
phase.  In a typical experiment, two permeators
were  used  to provide separate  reference and
sample signals  (Figure  2).

Pyrolysis-EC. The CPS-100 Toxic Gas Monitor has
been described in several earlier publications
(5-11);  its configuration is  diagrammed in Figure
3. The four sensors had platinum or gold working
electrodes. For the grain odor experiments, the
sensors were biased  at differing  oxidizing
potentials, since reducing potentials gave very
low or poor signal s. A single rhodiurnpyrolysis
filament was operated at 25,  450, 750, and 850 °C.
The  combination  of  four   sensors  and  four
temperatures gave an array of sixteen data points
per analysis.

The  apparatus  for  simplified  pyrolysis-EC
consisted of a single platinum filament and a
single platinum-electrode gas sensor. A control
circuit maintained the catalyst at any one of
four presel ected temperatures.  The f i 1 ament was
enclosed in  a Teflon-lined chamber of  small
volume through which the analyte gas was pumped
at about 50 cc/min.  The gas  then passed through
a short tube to the sensor. The experiments were
controlled, and data gathered, by a commercial
datalogger (Onset Computer  Corp., N.  Falmouth,
MA).

Gas samples.  Accurate  samples of test compounds
in vapor form were made by  injecting measured
volumes of  the 1 iquids into  40-1 iter Tedlar gas
bags  and  filling with  air pumped through  a
charcoal/Purafil filter. A flowmeter together
with a stopwatch was used to determine the volume
of air being pumped into the bag.  Samples of
permanent gases were made from standard mixtures
obtained from commercial  sources. Volumes of the
standard mixtures and air were calculated and
pumped into a sample bag,  using the flowmeter and
stopwatch to determine  the volumes.

Samples from grain odors  were generated  by
heating a sample of grain to 60 °C and flushing
with a measured volume of air. The effluent air
was passed through an ice trap to collect a "non-
volatile" fraction and a  liquid nitrogen trap to
collect the "volatiles".  The two fractions were
run separately and in duplicate.  Grain samples
were obtained from Drs. L. Seitz, and 0. Saur of
the USDA Grain Marketing Research Laboratory.
                                                86

-------
 Results and Discussion

 Organochlorine sensor.   Typical responses  of
 the sensor to different vapors in air are shown
 in Figure 4.  The sensor was exposed to 100 ppm
 concentrations of chlorobenzene, benzene,  and
 n-hexane.     Only  chlorobenzene   caused   a
 response.     Of   a   series   of    compounds
 investigated,    only   HC1,   and    compounds
 containing carbon-chlorine and  carbon-bromine
 bonds,  gave a response (Table I). The response
 to concentration  is  essentially  linear over at
 least four orders of magnitude.

 Combined  with the permeator device,  the highly-
 selective organochlorine  sensor was shown  to
 respond rapidly to dissolved material.   Figure
 5 shows the response to chloroform  in water  at
 concentrations  that  dip  below  the part-per-
 million level.   This sensor  can  be  used  to
 measure an organochlorine in groundwater, for
 example,  without  any sample preparation.  Many
 sites,  especially military  bases,  and areas
 such  as Rockford, Illinois, where  there is  a
 large concentration  of  machine shops,  have
 serious problems with chlorinated C2 compounds
 in the groundwater.  In these cases,  the nature
 of the  compounds  is  generally  known,  and
 selectivity is  not a concern.   Nevertheless,
 the sampling procedure, sample preparation, and
 gas chromatography to determine these compounds
 is involved and expensive. The availability of
 a  simple  probe that  can just  be  inserted into
 a  groundwater  sample will greatly  reduce the
 number  of  laboratory  analyses  that  need to be
 done.   The silicone material   is  chemically
 resistant,  and can be left in place for years.
 Particulates cannot enter the system.  Lastly,
 and   importantly,   the   permeator   is   very
 inexpensive.

 Pyrolysis-EC: Grain Odors  Only  a few organic
 compounds  will  react directly with amperometric
 sensors under field conditions.   On  a typical,
 platinum-electrode   sensor,   we  can   detect
 alcohols,  epoxides, and  formaldehyde.  We also
 detect  many permanent gases,  such   as  carbon
 monoxide  and  hydrogen sulfide.   Among  these
 gases  that do  react,  there  is  no  inherent
 selectivity.  The  use of different sensors and
 controlled pyrolysis, however, gives  us extra
 degrees of freedom that can be  used  to achieve
 selectivity.

 The grain  odor problem is very  instructive,
 even  to an audience  that is  concerned  with
 identifying  individual  hazardous  compounds.
 Sensor-array-based  methods,   including   the
 pattern-analysis   methodologies   used,   treat
mixtures no differently than single  compounds;
 both give  characteristic patterns which  can  be
 identified against a pattern made from the same
 mixture. The individual components of a mixture
 need not be identified.

 Grains  are presently  classified by odor by a
 panel of trained inspectors.  The results are
 necessarily subjective.  More importantly, the
 subjective opinion is the standard; there is no
 point in telling  a customer that a sample of
 grain is acceptable because a machine says so.
 If it smells bad,  it smells bad.  On the other
 hand, trained inspectors frequently disagree to a
 greater or lesser extent on both the category and
 degree  of  an  odor (Table  II).    Attempts  to
 identify specific compounds  associated with the
 odors, using GC or GC/MS, have produced masses of
 data, but limited results  (12,  13).

 The data obtained on the CPS-100 was subjected to
 two different kinds of analysis.  The first was
 an established method called k-nearest neighbor
 (KNN, ref.  5).  The 16 data points acquired by
 the CPS-100 were  treated  as a vector  in 16-
 dimensional space.  Each known sample of grain
 produced a vector which could be associated with
 a  particular odor category.  The vectors from the
 unknown samples were tested against this library
 of known vectors  by  calculating the  scalar
 distance between the  unknown vector  and each
 known vector in the library.  All vectors were
 first normalized to constant length, to  remove
 the  concentration-dependent  part  of   the
 information.   The shortest  distance is  the
 identification (Figure 6).

 The second  method  is  the neural  network (for
 general  references, see 14,  15).   This  is a
 recently-developed method that has received so
 much "hype" that  we were at first  suspicious of
 it.     However,   its   performance  has   been
 outstanding in this application,  the moreso
 because we used a  commercially-available packaged
 method   (NeuroShell,   Ward  Systems  Group,
 Frederick, MD), without really understanding the
 internal mechanics of the method.  This is a very
 important feature of a method which may be used
 in  the  field  by  operatives with  differing
 technical backgrounds.

 Figure 7 shows the CPS-100 data,  in histogram
 form,  for "good" wheat samples. The patterns are
 very similar, in contrast with data showing some
 extreme samples (one "sour" (S3) and one "insect"
 (13) odor) (Figure 8).  A experiment using the
 older KNN method was run using a dataset derived
 from three grades  of wheat samples.  A library of
 vectors was prepared by averaging the signals for
 all runs on each  sample of wheat.  The scalar
distances were  calculated between all possible
pairs of the original data set and each of the
averaged  vectors.      A   summary  of   the
 identifications is  shown in Table III.  We were
very (pleasantly) surprised  to find that those
samples that are "misclassified" by  the KNN
                                               87

-------
algorithm  are  also  those  that   the   human
inspectors did not  agree on!   Sample 42,  for
example, was voted "good" by two inspectors and
"musty"  and  "COFO"  by the  other two.   (COFO
means   "commercially   objectionable  foreign
odor".)

Although KNN has shown good performance in past
applications (5,  6,  8-11),  it  has some serious
practical disadvantages.  The greatest is that,
when the sensors become aged or drift for other
reasons,  the complete  training set  must  be
remeasured.

A larger data set had been gathered by the time
the work was begun with  the neural  nets.   This
data set had a peculiarity built into it:  one
of the  sensors in the array went  bad halfway
through the measurements  and was replaced.  The
data taken  after that  point  gave  noticeably
different histograms.

The data set was arbitrarily divided  into  two
groups.   One group  was  used  to  "train"  the
neural  network, a process requiring up  to  150
hours  on a  386-type computer.    The  actual
classification process took seconds.  Two tests
were run on  the  optimized  neural  net.   First
was a  test to confirm  that the optimization
process was complete.  This was done  by  using
the training set  itself  as  unknowns.   The rate
of correct  classification  was 100%.  Second,
random, linearly-distributed errors were added
to the data, followed by classification.   The
net  tolerated 5%   error  without   missing  a
correct classification.  Added error of 10% and
15% caused a small amount of degradation  (Table
IV).

Having confirmed the robustness of the  neural
net,  it was  challenged  using  the  reserved
dataset.  The  net had not  seen these numbers
before;  nevertheless  the  rate   of  correct
classification was  65%  (Table  IV).   This  is
low, although substantially better than random.
Because the test  conditions had  changed during
the measurements, we added another element to
the  data    vectors   to   differentiate   the
measurements made before and after the  sensor
was changed.  The numbers were arbitrary,  100
for the old sensor and 200  for the  new.   Using
these 17-element vectors,  the neural  net  was
retrained.      Now,    the  rate   of  correct
classification of the reserved dataset jumped
to 83%.

Pyrolysis-EC: Simplified Version This work is
the result of a project to  determine whether a
greatly-simplified  form of pyrolysis-EC  would
be  useful   for  situations  requiring  limited
selectivity.   Figure 9  is a diagram of  the
patterns obtained for  representative compounds
in a typical experiment.   The temperature of
   the catalyst is programmed for two minutes at
   room   temperature,   two   minutes  each   at
   temperatures of 500, 600,  700,  and 800 °C,  and
   finally two minutes  at room temperature again.
   The patterns that are obtained are distinct for
   many compounds.  If your field problem is simply
   confirming  the  identity of the contents of a
   number of similar barrels of an unknown chemical,
   the  pyrolysis-EC  approach  may in  itself  be
   sufficient, although most practitioners would
   feel more comfortable if it supplemented other
   field  screening methods.

   A   table   of  distances  for   this  limited
   configuration is shown in Table V.  The smaller
   the number, the more similar the two compounds
   will appear for a given configuration  of  the
   experimental apparatus.  This configuration gives
   very good identification of ethylene oxide in the
   presence of all but  alcohols.

   The pyrolysis-EC method has several advantages
   that are especially conducive to field work.  It
   is  suitable for portable  instrument use;  the
   components are shock-resistant and will  operate
   in   any   orientation.     They  compact   and
   lightweight,  and  the power  requirements  are
   small.  They  are also inexpensive.
   Conclusions

         1.  A  sensor  has  been  developed  and
   characterized that can identify chlorinated or
   brominated compounds in the vapor phase or, with
   the use of a permeable membrane, in dissolved
   form.

         2. A combination of catalytic pyrolysis and
   electrochemical detection (pyrolysis-EC) can be
   used to distinguish  unknown  compounds  with a
   modest degree of selectivity that may be adequate
   for many field applications.

         3. Pyrolysis-EC data,  combined with k-
   nearest    neighbor   and    neural   network
   classification methods, has been used effectively
   for such varied tasks as the classification of
   stored grains by odor, or the classification of
   waste chemicals by functional group (11).

         4. The  neural  net can  be  made to adapt
   dynamically to instrument drift.  In effect, it
   learns from experience.

         4.  Errors  made by  the classification
   methods correspond in a  general  way to  errors
   made  by  human  experts  faced  with similar
   ambiguities in the data.
88

-------
Bibliography
      Stetter,  J.R., and Cao, Z.   "Gas  sensor
      and   permeation   apparatus   for   the
      determination    of    chlorinated
      hydrocarbons in water". Anal.  Chem.  62,
      (1990),  182-185.

      Stetter,  J.R., and Cao, Z.   "A  real-time
      monitor   for  chlorinated   organics   in
      water".     Proc.   1990  EPA/AWMA   Int'l.
      Symposium on  "Measurement  of Toxic  and
      Related  Air  Pollutants",  Raleigh,  NC,
      April  3  -  May 4, 1990.

      Cao,  Z.,  and Stetter, J.R.  "A  selective
      solid-state   sensor   for    halogenated
      hydrocarbons".    Case  Western Reserve
      University,   Edison   Sensor  Technology
      Center,  Proc.  Third  Int'l.  Meeting  on
      Chemical    Sensors,    Cleveland,     OH,
      September  24-26, 1990.

      Stetter,  J.R., Zaromb, S.,  and Findlay,
      M.W.   "Monitoring of  electrochemically
      inactive  compounds  by amperometric  gas
      sensors".     Sensors  and  Actuators   6,
      (1984),  269-288.

      Stetter, J.R.  Penrose, W.R.,  Zaromb,  S.,
      Christian,  D.,  Hampton, D.M., Nolan,  M.,
      Billings,    M.W.,    Steinke,    C.,    and
      Otagawa,       T.       "Evaluating    the
      effectiveness  of   chemical   parameter
      spectrometry  in  analyzing  vapors   of
      industrial   chemicals". Proc.     Second
      Annual  Technical   Seminar   on  Chemical
      Spills,    Environmental      Protection
      Service,  Environment  Canada,  Toronto,
      Canada,  February 5-7, 1985.

      Stetter,  J.R.,  Jurs,  P.C., and   Rose,
      S.L.   "Detection of Hazardous Gases  and
      Vapors:  Pattern Recognition Analysis  of
      Data   from   an  Electrochemical   Sensor
      Arrays,"   Anal.  Chem.  58,  (1986) 860-
      866.

      Stetter,  J.R., Zaromb,  S.  and  Penrose,
      W.R.   "Sensor   array  for   toxic   gas
      detection".   U.S.  Patent no. 4,670,405,
      1987.

      Stetter,  J.R.,  Penrose, W.R.,  Zaromb,
      S., Nolan, M., Christian, D.M., Hampton,
      D.M., Billings,  M.W.,  and Steinke,  C.  "A
      portable   toxic   vapor  detector   and
      analyzer using an electrochemical  sensor
      array".   Proc.  DIGITECH/85  Conference,
      Instrument  Society  of America, Boston,
      MA, May 14-16,  1985.
9.    Stetter,  J.R., Zaromb, S.,  Penrose, W.R.,
      Otagawa, T., Sincali, A.J.,  and Stull, J.O.
      "Selective   monitoring   of   hazardous
      chemicals in emergency situations". Proc.
      1984 JANNAF  Safety and   Environmental
      Subcommittee Meeting, Laurel, Maryland.

10.   Stetter,  J.R., Zaromb, S.,  Penrose, W.R.,
      Findlay,  M.W., Otagawa, T., and Sincali,
      A.J. "Portable device for detecting  and
      identifying hazardous vapors". Hazardous
      Materials Spills  Conference,  April 9-12,
      1984, Nashville, TN.

11.   Findlay,    M.W.,   Stetter,   J.R.,   and
      Pritchett, T. "Sensor array based monitor
      for hazardous waste site screening". Proc.
      HAZMAT 90 Central Conference, Environmental
      Hazards Management Institute,  Durham, NC,
      March 13-15, 1990.

12.   Weinberg, D.S. "Development  of an Effective
      Method of Detecting and Identifying  Foreign
      Odors in  Grain Samples,"  Final  Report,
      Volume I, USDA  Contract # 53-6395-5-59,
      SoRI-EAS-86-1208, Dec., 15,  1986.

13.   Ponder,   M.  C.   and  Weinberg,   D.S.
      "Development of an  Effective Method of
      Detecting and Identifying Foreign Odors in
      Grain Samples," Literature and Equipment
      Survey USDA, Contract # 53-6395-5-59, SoRI-
      EAS-85-727, Aug., 5, 1985.

14.   Nelson,  M.M.,   Illingworth,  W.T.    A
      Practical Guide to Neural Nets. Addison-
      Wesley Publishing Company, Reading, Mass.,
      1990.

15.   Caudill,  M.,  "Neural Network Primer", AI
      Expert, Miller Freeman Publications, 1990.
                                              89

-------
Table  I.  Sensitivities  of the organochlorine
sensor to several  halogenated compounds.
Vapors
C«HTCI
dH?Br
C,H,I
CJfcF
C.H.C1
C.H.BF
C.H.I
C.C1F.
Concitntration
(ppn)
125
US
125
62.5
61. S
62.5
125
12.5
R4t«pon*«
( X 10'*«hO/PPB)
0.024
0.016
0.003
0.005
0.029
0.020
0.00]
0.022
Table IV. Summary of the accuracy of the neural
network  algorithm for  identifying vapors drawn
from the wheat samples.
Sorghum Accuricy of
Dati Set Identification
1. Orlglnjl Diti
2. 5X Error tdded
3. 10% Error Added
4. 15X Error Added
100X
100X
98%
92%
Wheit Simples Accuricy of
Qltl Sfit JfJepf 1f1cit1on
1. Totil Oiti
2. Tnln on S5X of
Diti set
3. Add cliinnel for
Test Conditions
100%
65X
83X
Table II. Subjective odor characterization of the
grain  samples used in our  study.
               OKRL INSPECTORS
SAMPLE '
NO.
F41
F42
F67
F78
F128
F30
F39
F69
FB9
N53
N166
N168
OS
CI
QUO
OKO
OHO
OKO
13
12
11
13
12
S3
82
US
OKO
OKO
HI
OKO
OKO
13
C3
12
13
S3
S3
S3
ICF
OKO
HI
OKO
M2
OKO
13
12
12
12
32
S3
SI
KM
OKO
C2
OKO
OKO
OKO
13
Cl
H3
S3
S3
S3
32
FUIS
CONSENSUS
OK
OK
OK
OK
OX
INSECT
INSECT
INSECT
INSECT
S3
S3
S3
nve.
INTENSITY
0.5
0.7
0.2
0.5
0.0
3.0
2.0
1.0
2.7
2.8'
J.91
2V
 Table  V.  Distance  matrices  for  a  series  of
 organic  compounds.    Table   V-A  is  several
 concentrations   of   ethylene    oxide;    the
 concentrations are  shown as the numbers  in  the
 symbols, e.g., ET0100 - 100 ppm.  Table V-B shows
 the distances  among  the  series  of thirteen
 compounds.   The Abbreviations are:
                      ISO - Isopropanol
                      KER - kerosene
                      STY - styrene
                      ETG - ethylene glycol
                                                         CHX - cyclohexane
                                                         ETE - ether
                                                         CLO - chloroform
                                                         FORM - Formaldehyde
                                                         ETO - Ethylene Oxide
ACE - acetone
XYL - xylene
WL- hilothne
ETA - ethanol
                                                         TABLE V-A
                                                         Distance for Ethylene Oxide
rroioo ET040 rroio
ET0100
FT040
ET020
ETO:
ETOS
ETOl
0.00
0.31
0.28
0.22
0.25
1.02
0.31
0.00
0.07
0.21
0.18
0.80
0.2>
0.07
0.00
0.21
0.16
0.82
                                                                                      ET05
                                                                                         0.22
                                                                                         0.21
                                                                                         0.31
                                                                                         0.00
                                                                                         0.09
                                                                                         0.85
                                                                                             ETO5
                                        0.2S
                                        0.18
                                        0.16
                                        0.09
                                        0.00
                                        0.80
     1.02
     0.60
     0.62
     0.85
     0.80
     0.00
                                                         TABLE V-B
 Table III. KNN classification of the USDA grain
 samples.

                  Average of Known Vectors
           Good           Insect          Sour
         (128. 42. 67, 41)       (30, 39, 89)       (53, 166, 168)

CHX
ISO
ACE
JTI
XYL
KER
CLO
STY
FORM
HAL
CT3
ETO
ETA
CHX
0
1.57
0.19
1.76
1.03
1.07
0.69
1.44
1.74
2.09
1.52
1.73
1.95
ISO ;
1.57
0
1.43
0.44
0.76
0.63
1.49
0.46
0.11
0.73
0.4
0.63 3
0.81 ]
kCE m XYL
1.19 1.76 1.02
L.42 0.44 0.76
0 1.59 0.85
.59 0 0.82
.85 0.82 0
.91 0.74 0.34
.55 1.53 0.98
.27 0.41 0.45
.56 0.2 0.85
.93 0.59 1.37
.35 0.3 0.55
.55 0.35 0.75
.77 0.36 0.98
KIR
1.07
0.62
0.91
0.74
0.34
0
1
0.49
0.76
1.06
0.54
0.75
0.9S
CLO
0.69
1.49
0.55
1.53
0.88
1
0
1.26
1.55
1.93
1.33
1.44
1.62
STY
1.44
0.46
1.27
0.41
0.45
0.49
1.26
0
0.45
0.9]
0.13
0.37
0.63
FORM
1.74
0.31
1.56
0.3
0.85
0.76
1.55
0.45
0
0.61
0.34
0.41
0.56
HAL
3.09
0.73
1.93
0.59
1.37
1.09
1.93
0.93
0.61
0
0.64
0.79
0.73
ETC
1.52
0.4
1.35
0.3
O.SS
0.54
1.33
0.12
0.34
0.84
0
0.3
0.55
ETO
1.73
0.62
1.55
0.25
0.75
0.75
1.44
0.37
0.41
0.79
0.3
0
0.27
ETA
1.9»
0.61
1.77
O.Ji
0.96
0.99
I.*'
0.61
0.5«
0.73
o.ss
0.27
0
128,128.42,
67,67,41,41,
41,41
89
168
42
30, 30, 30, 39,
39, 89, 89, 89
168, 168
42
30
53, 166, 166,
166, 168, 168
                                                      90

-------
                                                                             Vent
                                                                                  Sensors
 Figure 1. Permeation  apparatus used to extract       Figures. Configuration of the CPS-100 Toxic Gas
 °i"ganochlorines  from water.                          Analyzer,  fitted  with  four electrochemical
                                                       sensors and two  catalyst filaments.


Permeation
apparatus


3-way
valve


Permeation
apparatus


Ci"rier a
                 In blank water
                                           Exhaust
 Fl9ure 2. Experimental  apparatus for selective
 analysis  of aqueous  chlorinated  hydrocarbons
 Using a separate reference permeator.
                                                                RESPONSE OF C6H5CI, C6H6 and C6H14
                                                                         100 PPM, SEH5O» *OO t 10-M
                                                                        40      60      80
                                                                            TIME (WIN)
                                                                                              100
                                                                                                      120
Figure 4. Response of the organochlorine sensor
to chlorobenzene,  benzene, and  hexane.
                                                  91

-------
                  Low cone. CHCI3 in Water
                      Oun.100. FFI-UOcc/mnl
  Figure 5. Response of the organochlorine sensor
  to decreasing  concentrations  of chloroform.
                                                                                                 ,
                                                                   It  12 13  14 21  22  23 24 31  32 33  34 41  42 43  44
                                                                                     CHANNEL
                                                                            I  IB
                                                                                          JI | B^. 4
Figure  7.  Histogram of normalized  responses of
the CPS-100 to  four  samples of "good"  grain.
                                                                             5*6»oigwq1 900d/tOHout/123co(o
Data vectors are normalized
to vectors of unit length.
U, is unknown compound,
P, and P2are
known pattern vectors.

Scalar distance between
vectors U, (unknown)
andn P( and U, and
P3are calculated and
compared (D, and D,)
  Figure 6.  Schematic representation of the KNN
  pattern  recognition  method  in  3-dimensional
  space.   PI  and  P2  are  library patterns  for
  known compounds,  and  Ul  is  the  vector  for an
  unknown.   The  distances  from  Ul to PI  and P2
  are  calculated and  compared.

100
60
20
-9D-














~H



[





"H






n


1

1


UH





                         CHANNEL

            Q] OK (Conlrol) B§ 101(53)  (
                                                                                            l 1?3 ICOFO1
Figure 8.  Normalized responses of the CPS-100 to
"good"  (OK),  sour  (S3),  and COFO grain.

-------

                        Chloroform
Isopropanol
                                                                                                 Elliylene Oxide
                         Acetone
                                                            Elhanol
                                                                                                   Halothane
                                    Figure    9.    Responses    of    the    simplified
                                    pyrolysis-EC    apparatus    to    six    different
                                    chemicals.    In  this  experiment,   the  catalyst
                                    filament was  programmed  in  2  minute  steps  at
                                    room   temperature,   500,    600,   700,   and   800
                                    degrees, and  room  temperature again.
                                                         DISCUSSION
GORMAN BAYKUT: My question is about the chemical analysis with these
sensors. I'm not talking right now about the wheat vapor. But in terms of real
chemical analysis, you must know the compounds you are going to analyze,
otherwise you can't do the analysis because you need training. You can't analyze
the unexpected compounds, am I right?

WILLIAM BUTTNER: The way  the  CPS  100 Program was originally
envisioned, you had to install the library vectors of potential compounds. If you
were going to look at TCE, there had to be a library vector associated with the
TCE. On the other hand, these arrays are not totally selective in response. The
response to TCE was similar to  PCE, that is,  tetrachloroethane.  You could
therefore identify classes of compounds. But you are right. You have to have
some ideaof the type of vapors present. Atotally unknown situation will still give
some ambiguity in your analyses.

GORMAN BAYKUT: But I think even though your software is powerful, you
need a training period for every compound. How about the mixtures? If you
analyze the mixtures will there be a problem?

WILLIAM BUTTNER: Mixtures are a problem for this type of system. Certain
types of mixtures are well behaved. Gasoline, for example^ is a mixture of many
types of compounds, but it behaves as a single class.

GORMAN BAYKUT: I'm referring to the cracker. You have a thermal cracker
in front of the electrochemical sensor areas. Sometimes you have a mixture of
two or three compounds, or five, or seven and they react in the cracker. You get
different answers, and the correlation is not linear.

WILLIAM BUTTNER: What you're referring to are the reaction products of
the thermal catalysis that result from mixtures being exposed to the sensors. Yes,
you are right. There is frequently  a nonlinear response. The reaction products
frequently do react with each other. That's a comment relevant to many field
        screening techniques. In some mixtures that factor is a little less significant. If
        you do generate very reactive compounds, for example from chlorinated com-
        pounds TCE, you do get a nonlinear response. That is a problem. This instrument
        was designed to look at single vapors, maybe not necessarily positively identi-
        fied, but single vapors.

        STEVEN KARR: I wondered if you've given any thought to applying fuzzy
        logic algorithms to this problem as opposed to neural networks?

        WILLIAM BUTTNER: The neural network was a six-month program that we
        tried on the SBIR (we've just finished Phase I).  To stay within the time
        constraints, we stuck  to simple systems. We are investigating other  neural
        network software packages and other identification algorithms. We will certainly
        consider fuzzy networks.

        EDWARD POZIOMEK: Have you tried any real-world environmental samples
        with the system.

        WILLIAM BUTTNER: I had a program through Savannah River to monitor for
        TCE emissions out of their stripping tower, as part of their groundwater clean up.
        Initially the results were very encouraging. The analyses that I measured were
        compared back to groundwater samples as measured at an independent labora-
        tory. They  were  comparable in value. The unfortunate  thing is that  these
        amperometric sensors did not behave truly reversibly to chlorinated compounds,
        and  that after u period of time  their response factor, their sensitivity,  would
        degrade and ultimately their response would die completely. For that reason it
        was  determined that these types of sensor systems would not be applicable for
        the problems associated with Savannah River Laboratory. This was before this
        chlorine selective sensor was developed. It could potentially have application
        down there.
                                                                   93

-------
                    REAL-TIME DETECTION OF ANILINE IN HEXANE
                   BY FLOW INJECTION ION MOBILITY SPECTROMETRY
              G.E. BURROUGHS
    National  Institute for Occupational
 Safety and Health,  4676 Columbia Parkway,
           Cincinnati,  OH   45226
  G.A.  EICEMAN and L.  GARCIA-GONZALEZ
            Chemistry Department
        New Mexico State University
           Las Cruces, NM  88003
 DISCLAIMER:   Mention of company names or
 products  does not  constitute  endorsement
 by    the    National    Institute    for
 Occupational Safety  and  Health.
ABSTRACT

Ion  mobility spectrometry  (IMS)  with  a
conventional  "Ni  ion  source  exhibits
chemical   behavior   that    should   be
advantageous  in  detection  of molecules
with   high  proton   affinity  such  as
aromatic   amines   in   common   organic
solvents.  Since  IMS  instrumentation can
be considered a continuous-sampling point
sensor, IMS may be adapted for industrial
process monitoring  or area environmental
monitoring. However, quantitative aspects
of  IMS  are  not  well  established  and
possible   interferences   may  limit  the
usefulness   of   IMS.     In   order  to
characterize IMS  behavior as  an effluent
sensor, a  flow injection IMS device was
evaluated  in which an IMS was used as  a
detector for a heated injector port.  An
IMS drift  tube was used with an acetone
doped  reaction  region  and   a  membrane
inlet.  Five microliter replicate samples
were  introduced  and  vaporized  in  the
inlet  at  15 -  90  second intervals and
drawn  into the  IMS.    Detection   limits
were ca. 0.5 mg L-1 for 5 ul  aliquots (2
ng per sample). Sampling intervals could
be  reduced   to  15   seconds  for  all
concentrations below 40 mg L-1  above which
however  a  working   range   could   be
considered to  approximately   100 mg L-1.
Precision  was  10  -   25% RSD  and  was
largely concentration independent.  Since
the IMS alone in a vapor  stream shows ca.
 1-2% RSD, the bulk of variance was from
 the inlet and inlet-IMS interface.   Four
 solvents  (benzene,  methylene  chloride,
 ethyl   acetate,    and   acetone)    were
 evaluated as  interferences.  All solvents
 at some  concentrations affected the peak
 area for  aniline,  although the  causes
 arose through different mechanisms.  The
 use of IMS as a  flow  sensor for aniline
 in organic solvents should  presently be
 restricted to samples  free  of  compounds
 with  strong  proton   affinities   and
 solvents  which   do not  exhibit  strong
 dipoles.

 INTRODUCTION

 Ion mobility spectrometry  (IMS) with  a
 conventional  "Ni  ion  source   exhibits
 chemical  behavior   that    should   be
 advantageous  in  detection  of  molecules
 with  high  proton affinity  such   as
 aromatic  amines   in   common   organic
 solvents.  Since  IMS instrumentation can
 be considered a continuous-sampling point
 sensor,  IMS may be adapted for industrial
 process  monitoring or area environmental
 monitoring.       However,    quantitative
 aspects  of IMS  are not well  established
 and  possible  interferences may  limit the
 usefulness of IMS. Among the attributes
 of an acceptable  "field screening method
 for  hazardous waste and toxic chemicals"
 are  sensitivity,  specificity, accuracy,
 precision, speed,  and portability. Also,
 to be worthwhile,  it should be applicable
 to the screening  of analytes or classes
 of compounds  which have a reasonably high
 toxicity.     The   optimum  value  of  a
 real-time field technique would  be in the
 screening  of  substances   with  acute
 toxicity,  thereby  assisting   in   the
elimination of short term exposures.  The
purpose  of this  work  is  to investigate
                                          95

-------
such  quantitative  aspects  of  IMS  as
sensitivity,  accuracy,  and  precision;
interference is examined as a comparison
of response to solvents of varying proton
affinity;  and speed  of analysis  is an
additional experimental parameter.

In IMS, vapors are drawn into a reaction
region where  analyte  is ionized through
proton  or  electron  transfers  from  a
reservoir  of  charge,  the reactant ions.
The  reactant  ions  originate  in  beta
emission from a 63Ni radioactive foil and
the reactant  ions exhibit  near thermal
energies.   Consequently,  product  ions
usually  experience  little fragmentation
and exist principally  as M+, MH ,  or MjH"1".
lonization  in  the  reaction  region  is
based on competitive charge exchange, and
unequivocal  response  occurs  when  the
target  analyte  has  a proton  affinity
larger than that for any component in the
sample matrix.  When this  is not assured,
response  can  become confusing  even  for
simple mixture  (1) .    Thus,  the primary
basis  for selectivity  of   IMS  as  a
detector  is based  upon  differences  in
proton    affinities    of   constituents
following vaporization into a flowing air
stream.  Product ions  are  injected into a
drift  region  where   ions   acquire  a
constant  velocity  in  a  weak  electric
field.  Differences  in ion velocities are
due  to  differences  in cross-sectional
areas,  and  this serves  as  a  useful,
second   level  of  selectivity  in  IMS.
However, response in IMS is fundamentally
governed by the original step of product
ion creation;  thus,  if a product ion is
not formed in the ion  source, regardless
of  cause, a peak corresponding to that
substance will not  be observed  in the
mobility spectrum.


Flow injection analysis (FIA)  is a type
of continuous analytical technique where
discrete, reproducible aliquots of sample
are introduced  into a  flux,  allowed to
interact  with other components  of that
flux or with forces  exerted on that flux,
and  are   subsequently  monitored  by  a
detector having some inherent specificity
for  the  resultant  species.  Reviews of
flow injection analysis by Betteridge (2)
and by Ranger  (3)  date the  origins of
this technique to the  early to mid-1970's
as  an  adaptation   or  subcategory  of
"continuous flow analysis" as described
by Skeggs  (4) .  This type of analysis has
the advantages of being simple, accurate,
reliable,   reproducible,   and  can   be
accomplished  with  a  small  amount  of
simple   equipment.      All   of   these
attributes   are   desirable    in   any
real-time, field  screening method.   The
disadvantages of FIA methods come from a
dependence on detector selectivity in the
absence of any  separator  techniques,  as
will be seen later.

Chemically, the high  proton affinities of
aniline and other aromatic amines suggest
that ion  mobility spectrometry  may be a
technically   acceptable   technique  for
monitoring of these  substances  by flow
injection  technique. Development of  a
field   screening   method   for   these
compounds  would be  worthwhile  based on
toxicity,  the  primary  toxic effects of
this class of compounds on man including
methemoglobin formation and cancer of the
urniary   tract   (5).    Environmentally,
"aromatic amines  constitute a family of
serious pollutants due in part to a high
degree  of toxicity  toward  aquatic life
(6) .  Particular attention has been given
to  the   effects  of  aniline,   aniline
derivatives,  and  aromatic amines  on fish
(7,8), Daphnia  magna (9,10) and microbes
in estuarine water (11)."  (Eiceman et al)
Commercially,   they   are  important  as
intermediates  in  the  manufacture  of
dyestuffs and pigments,  but are also used
in  the chemical,  textile, rubber,  dying,
paper industries  and other  (5).
EXPERIMENTAL

Instrumentation
The  introduction  of  a  flow  injection
stream    to   an   IMS   detector   was
accomplished  using  the  instrumentation
and procedures described  below.  A block
diagram   of  the   flow  injection  IMS
apparatus is  shown  in  Figure 1 and was
comprised of a heated injector taken from
a  gas chromatograph,  an  Airborne Vapor
Monitor    (Grasby   Analytical,   Ltd.,
Watford,  UK)  as  the  IMS  detector,   a
pressurized source  of  air and supporting
electronics   to   control    injector
temperature.     Air  flow   through  the
injector  port was  ca.  5  ml/min and the
injector  temperature was  100°C.  Both the
injector   block   assembly  and  the  IMS
instrument   were   placed    inside    a
laboratory hood,  and there was a distance
of less  than  1  cm between  the injector
exhaust   and  the  IMS  inlet.    Digital
signal  averaging  was  used  to  acquire
mobility  spectra with  an  Advanced Signal
                                            96

-------
Processor (ASP) (Grasby Analytical, Ltd.)
into an  IBM  XT  microcomputer.    Also,
signal was routed  from an output voltage
on the ASP  to a  Hewlett-Packard  3380A
recording  integrator  so  peak areas  for
the aniline product ion could be recorded
versus time  and integrated.   The window
of observation for  drift times  for  the
aniline peak was ca.  0.1 - 0.2  ms wide
and was  centered on the drift  time  for
aniline,  8.74  ms.   Other  parameters  for
signal collection  through the ASP board
were: number of waveforms, 32; points per
spectrum, 512; and scale expansion, 0.25.
The   integrator     parameters     were:
attenuation and threshold,  each  9; chart
speed, 1  cm/min;  area  rejection,  10000;
and peak width 0.5.

Reagents and materials
The following  solvents were  obtained in
high commercial  purity and used without
further  treatment:     aniline  (Aldrich
Chemical  Co.,  Milwaukee,  WI,  99.5%+),
hexane  (Chromopure,  Burdick  &  Jackson
Co.,  Muskegon, MI),  acetone (Chromopure,
Burdick  &  Jackson  Co.),  benzene  (B&J
Brand,  Chromopure,  Burdick   &   Jackson
Co.), ethyl  acetate (Fisher Scientific,
Pittsburgh,  PA), and methylene  chloride
(Fisher Scientific) .

Procedures
In general,   5 ul  aliquots   of  liquid
sample  were  delivered  with  a  10   ul
syringe (Hamilton  Co.,  Reno,  NV) to  the
heated injection port during  continuous
signal  processing  with  the   IMS.    An
interval  of   15   to   90  seconds   was
permitted for  the  air to  sweep  vapors
from  the  inlet before another injection
was  made.     Several  parameters  were
examined to  determine optimum operating
conditions and access the reliability of
IMS as a  flow injection  detector.   The
particular  details  of  each  of  these
studies were:

Clearance study and response curve - Five
microliters  of  aniline   in   hexane   at
concentrations   from   0   to   100   ppm
(volume/volume liquid) were delivered in
five  replicates  at  different intervals
from  15 to 90  seconds.  Peak areas were
determined for the  aniline product ion in
the  preparation   of   a  quantitative
response curve.  The effect of injection
interval also permitted the determination
of memory effects  in  the IMS  under  a
range of concentrations.
Chemical interferences  - In the  study of
chemical   interferences   in    aniline
determinations,  5  ul of 5 ppm aniline in
hexane were co-injected with 0 to 4 ul of
pure   interfering   solvent.       These
interfering   solvents   were   methylene
chloride,  benzene,  acetone,  and  ethyl
acetate.   Five replicate determinations
were made at  60  second  intervals.

RESULTS AND DISCUSSION

General
The reactant  ion peak (RIP)  with acetone
reagent  ion  chemistry  and the  mobility
spectrum for aniline in the hand-held IMS
are  shown in Figure  2.    The  mobility
spectrum for  aniline contained  a  single
symmetrical peak at 8.74 ms drift time,
consistent with  previous  findings  for
aniline with water-based chemistry in the
ion source  (12).   Residual  amounts  of
reactant  ion  at   6.97   ms  in  aniline
mobility spectrum  demonstrated that  the
ion source  was  not  saturated  and  that
comparable behavior may be anticipated at
vapor  levels  lower  than  this.    This
mobility spectrum  was generated using  5
ul of a  5 ppm solution  (25 ng  absolute
mass)  and the peak height relative to the
RIP was reasonable considering the  high
proton    affinities    of    aniline.
Previously, aniline was shown with IMS/MS
to  yield  a  protonated   molecule,   MH"1"
product  ion   (12)  although  the ambient
temperature drift  tube  and alternate ion
chemistry  used   here  may   favor   the
existence of  a MH^S ion where  S is  an
acetone solvent  molecule,  but this  has
not been unequivocally  established.

Clearance Behavior. Standard  Deviation.
and Response Curve
The hand-held IMS used in this  work would
be suited for field use due to  its  size
(40cm x 15cm x 8cm), weight(2.6  kg),  and
ability  to   operate   continuously   in
hostile environments unattended.  The IMS
itself is  battery  powered  and could  be
interfaced with a battery powered lap top
computer for data  acquisition, providing
a  portable  system.   However,  this  IMS
could  be  expected  to  exhibit  memory
effects  from  the ambient  temperature
drift cell and membrane-equipped  inlet.
At high concentrations  of aniline,  slow
clearance from repetitive determinations
might occur.   In Table  1, peak areas and
percent  relative   standard   deviations
(%RSD) from repetitive determinations are
given for solutions between 0 and 100 ppm
at  injection  intervals  from  15  to  90
seconds.  The %RSD ranged from 13 to 125,
but showed  a  median of  21%    Previous
                                          97

-------
experience with this IMS as a detector in
FIA  methods  had yielded  reproducibility
of peak heights of  8 to 10  %RSD  and this
large variance was suspected to be due to
the  placement of  the FI-IMS  in  the fume
hood.  Turbulence in a fume hood has been
associated with position  and movement of
the user as  well  as amount and  location
of equipment in  the hood   (13).   This
turbulence likely affected yields in the
interface between the  inlet and IMS and
this  large   RSD  was   suggestive  that
mechanical   improvements   in  interface
between the  IMS and injection port are
needed.   A   straightforward  leak-tight
connection  was not employed in  these
studies due  to  the  flow characteristics
for this IMS and  the eminent rupture of
the    membrane    inlet    if    pressure
differences  developed  between the inlet
and ion source regions.
The  anticipated memory effect from slow
clearance of the aniline from the IMS was
evident in the peak areas given  in Table
1.  In general,  peak areas with 90 second
injection intervals were the lowest for a
given  concentration  level.   Injection
intervals less than 90 seconds caused an
accumulation of  aniline  in  the  IMS and
peak areas increased for  example as much
as 100% at  30 second intervals  with the
100   ppm  concentration.      This  was
manifested in the signal  for  continuous
monitoring as a rising baseline and in
the  mobility spectrum as  a  persistent
product  ion.  Memory  effects here were
dependent    upon    concentrations,   as
expected, and at  concentration  below 20
ppm,  injection  intervals  of  15 seconds
could   be   employed   with   reasonable
differences  in absolute areas.

A plot of peak area versus  concentration
of aniline in hexane for  5  ul injections
at 90 second intervals is  shown in  Figure
3  and  resembled  previous  response  or
calibration  curves  in  IMS  (14).   Such
curves  are  comprised  of  narrow  linear
ranges  (in this instance  between 5 to 20
ppm) ,   a  shallow   but  mostly   linear
response at concentrations above the main
linear  region and  a  nearly  linear plot
with  shallow  slope   below  the  linear
region.   This  behavior  is  due to  the
nature  of the kinetics  of  reactant  ion
formation  from  the  beta  emitting  ion
source   and,   thus,   to  the   limited
reservoir of charge available to analyte
vapors.
Chemical Interferences

The existence of solvents with a range of
proton  affinities  in  industrial  waste
streams    constitutes    a    potential
compromise  on   the  integrity   of  IMS
response in flow injection determinations
through  two  mechanisms.    Conceivably,
large  levels  of  such  solvents  might
compete for charge  resulting  in reduced
peak  areas for  aniline  at given  vapor
levels.  Alternately, solvents may cause,
at ambient cell temperatures, ion-solvent
clusters which  lead to shifts  in  drift
times for product ions.  This will cause
a  decline in  certainty regarding  peak
identity or may cause the peak  to fall
outside a  window of observation  in the
signal processing software.

Four solvents with low and medium proton
affinities were selected for interference
studies   and   mobility   spectra   for
individual solvents are shown  in Figure
4.     Methylene   chloride   gave  little
response  in  positive  polarity  IMS  as
expected due  to a  low proton  affinity.
For  the same  reason,  benzene  showed  a
weak  response with an acetone  reactant
ion chemistry and the  product  ion had a
drift time shorter than that for the RIP.
Acetone  formed  cluster ion, with  drift
times  longer  than  that   for  the  RIP,
through ion-molecule interactions in the
IMS drift region as described by Preston
and  Rajadhax  (15).   Only  ethyl acetate
(EtOAc)  showed  significant  competition
with  the  reactant  ion,   due  to  large
proton  affinities  of   EtOAc  relative to
acetone,  with the  obvious result  of  a
product  ion.    Of  these solvents,  only
benzene has  been mass identified  as M
(16)  though  acetates   are  known  to form
MH+ and MjH"1" product ions  (17).

The  influences  of these solvents on IMS
response  to  a 5  ul injection of  5 ppm
aniline in hexane are  shown in Figure 5
as a  plot  of  peak height  for aniline in
various  ratios  of  four  solvents  in  a
binary mixture with hexane. All solvents
affected   the  peak  area   for  aniline
although   the   causes  arose   through
different  mechanisms.     In  Figure  6,
mobility  spectra are  shown from  egual
mixtures of hexane and solvent for 5 ppm
aniline   and   these   can   be   compared
directly   to   spectra for  individual
solvents   (Figure   4)   and for  aniline
(Figure 2) .   For EtOAc,  the product ion
dominated  the ion chemistry when aniline
                                            98

-------
was present even though proton affinities
favored  aniline. Ethyl  acetate at  high
concentrations   relative   to   aniline
appropriated  virtually  all  the  charge
except that remaining with the RIP.   The
ion-molecule chemistry for acetone as an
interference  also followed  this  pattern
and  aniline was not detected  with  high
levels  of acetone.    Thus,  the rise  in
peak areas  in  Figure  5  represented  a
false positive  by  acetone for  aniline
since  acetone  product   ion  intensity
intruded upon  the  drift time window used
to monitor aniline.   In such a situation,
only inspection of the mobility spectrum
could avert  an error in  monitoring  on
analyses.   A  product ion for aniline was
evident with methylene chloride  due  to
the  low  proton affinities  of methylene
chloride.     However,  the  increase  in
response   for   aniline   in   positive
polarities  from addition  of  methylene
chloride  to   hexane   (Figure  5)   was
unprecedented  in  IMS  and  conclusions
cannot be  made pending  IMS/MS studies.
Benzene, with  proton affinities between
methylene chloride and acetone or EtOAc,
exhibited   a   type   of   intermediate
behavior.  A product  ion for aniline was
observed in the presence of benzene, but
the benzene was at a  level  sufficient to
effectively compete  for protons from the
RIP  and  a  benzene product ion was also
observer (Figure  6) .  These spectra and
trends  suggest  that an   IMS will  be
sensitive  to  common solvents  at low
levels  even  with an alternate reactant
ion  chemistry, a membrane  inlet, and low
 (<1%)  levels  of  solvents  other  than
hexane.     However,  if   the  solvent
composition   is  known  and   reasonably
constant,  calibrations presumably  could
be   prepared   in  that  matrix.     These
findings for  simple compositions  argue
for  standard  addition  techniques  with
flow injection IMS determinations.

 CONCLUSIONS
 Ion mobility spectrometry has never been
widely   regarded   as   a   quantitative
 instrument,  but as  a detector for flow
 injection  determination,  IMS  exhibited
 suitable   response    curves,    standard
 deviations, and response times.  This was
 accomplished    under   the    demanding
 situation of a fast transient vapor level
 in FIA methods.  The linear  range is  a
 weak  aspect   to  quantitative  IMS  and
 alternative     configurations    to
 conventional   63Ni    sources   should   be
 sought.  Reactant ion chemistry based on
 acetone was   not  wholly  successful  in
discriminating chemically against  common
organic  solvents.    Consequently,  until
improved   source  chemistry  is   found,
standard  addition should  be considered
the method of choice for quantitative FIA
with IMS for aromatic amines.

REFERENCES

 1.  Eiceman,  G.A.,  Blyth, D.A.,  Shoff,
D.B., Snyder, A.P.  Anal. Chem., 1990,  in
press.
 2.   Betteridge, D.   Anal. Chem. 1978,
50, 832A-845A.
 3.  Ranger,  C.B.   Anal. Chem. 1981 53,
20A-32A.
 4.   Skeggs, L.T.   Am.J.  Clin. Pathol.
1957 13, 451.
5.   Beard,  R.R.,  Noe,  J.T.    "Aromatic
Nitro and  Amino  Compounds," in  Pattv's
Industrial Hvoiene and  Toxicology.  Vol.
2A,  G.D.   Clayton  and F.E.  Clayton,
editors,   Wiley-Interscience,  New York,
1981.
 6.  National Research Council,  "Aromatic
Amines:  An Assessment  of the  Biological
and     Environmental    Effects,"     No.
PB83-133058,  Washington, DC.  1981.
 7.  Bradbury,  S.P., Henry, T.R.,  Nieme,
G.J.,   Carlson,   R.W.,  Snarski,   V.M.
Environ. Toxicol. Chem.  1989, 8, 247-261.
 8.     Newsome,  L.D.,   Johnson,   D.E.,
Cannon,  D.J.,  Lipnick,  R.L.  "Comparison
of Fish Toxicity Screening Data and QSAR
Predictions for 48 Aniline Derivatives,"
QSAR  Environ.   Toxicol.,  Proc.  Int.
Workshop,   2nd,   Kaiser,  K.L.,   Editor,
Reidel,    Dordrecht,   Netherlands,   pp.
231-250,  1987.
  9.    Kuehn,  R.,  Pattard, M.,   Pernak,
K.D.,  Winter, A.   Water Res.,  1989, 23,
 495-499.
 10.   Gersich, R.M., Milazzo,  D.P.  Bull,
 Environ.  Contam. Toxicol.,  1988, 40, 1-7.
 11.  Hwang, H.M., Hodson, R.E., Lee, R.F.
Water Res. 1987, 21, 309-316.
 12.   Karpas,  Z.  Anal. Chem.  1989,  61,
 684-689.
 13.  National Research Council, Committee
 on Hazardous Substances in the
 Laboratory,   "Prudent   Practices    for
 Handling Hazardous  Chemicals in
 Laboratories,"  Washington, DC, 1981.
 14.  Leasure, C.S., Eiceman, G.A.  Anal.
 Chem., 1985, 57, 1890-1894.
 15.  Preston, J.M., Rajahyax, L.    Anal.
 Chem., 1988, 60, 31-34.
 16.   Kim,  S.H.,  Betty,  K.R., Karasek,
 F.W.  Anal. Chem,  1978, 50, 1754-1758.
 17.  Eiceman, G.A., Shoff, D.B.,  Harden,
 C.S.,  Snyder,   A. P.    Internal.  J.   Mass
 Spectrom.   Ion  Processes,   1988,    85,
 265-275.
                                           99

-------
                fro* Plot* of AniliM Product Ion lotwwitr
          AniliM
          Concentration
          (PP»>
 PEAK AREA 
-------
                                   CH
                                   6 6
                                   EtOAc
              Drift Time (msl
4.   Ion  mobility  spectra for  solvents
    expected   to   be  encountered  in
    analysis of  non-aqueous streams for
    aniline.     Mobility  spectra  were
    obtained  in positive  polarity with
    acetone   reagent  ion   chemistry.
    Spectra were obtained with solvent
    vapors permitted  to deplete reactant
    ion    intensity   ca.    50%   from
    background  levels.
              w/CH?l
                                              I
                                                     7.77 ™   w/Acetone
                                                                         | """   w/EtOAc
                                                            I   15.01  5
                                                            Drift Time (ms)
6.   Mobility  spectra for mixtures  of 5
     ppm aniline  in 50  : 50  mixtures of
     individual   solvents  with  hexane.
     Aniline in hexane exhibited a single
     product ion  with drift time of 8.74
     ms as shown  in Figure  2.
             0.5       5        25
             Percent in Hexane
5.   Effect on peak height for aniline at
     5 ppm in binary solvent mixtures of
     hexane with  other  common solvents
     with  vol/vol percentage  from  0»  -
     50%   Curves were normalized to the
     peak  Sit   of  aniline  in  hexane
     solution.
                                          101

-------
                                                           DISCUSSION
STEVEN HARDEN: The question I have is with respect to orthonitrophenol
and the sensitivity of the IMS system to that particular kind of material. Did you
ever do a calibration run to determine what that sensitivity might be under various
conditions?

PETER SNYDER: The answer to that question is no, we have not on pure
orthonitrophenol. However, the signals—the amount of signal that we see from
the other point of view, looking at it from the organism's  point of view, and
knowing how much organism we have. It seems like there is still plenty of
analyte, given the relatively short time of detection, and knowing that the signal
is still a bit spread out. The signal is not in one, or say two, or maybe three at the
most, peaks. We see it at about seven, eight, nine 10 peaks, until it finally clears
down.

So I'm not trying to skirt the question. It's just that no, we haven't done it to see
how sensitive the CAM itself is, or the ion mobile spectrometer 20MP. However,
I suspect that it has to be very  sensitive, since 200, even 50 cells is a good
response, and the response is spread out, so if we can find ways of compacting
it, it'd be that much better.
MAHADEVA SINHA: What are the vapor pressures for the orthonitrophenol
when it gets combined with the glucose. Do you get any response?

PETER SNYDER: Yes, we've done many, many blanks. We always do a blank
before and after.

First of all, the vapor pressure of orthonitrophenol is 5.54 torr  at ambient
temperature. That doesn't should like much, but relatively speaking, that's a lot
for the CAM. And the controls — we have done ONP by itself, with buffer,
without buffer, and then just organisms themselves. Organisms do produce some
peaks, but that's just right after the reactant ion peak. But it just happens to tail
off, and there is no signal in the area that the ONP shows. So we have been pretty
lucky in that respect.

The ONP has very negligible vapor pressure by itself. Even if you get a bottle of
the dry powder, and just stick the CAM in the bottle, you see no response at all.
That should be the most amount, the dry powder, and if anything's going on it
would show. But even in the solution, there's no problem.

Orthonitrophenylacetate is a different story. There is hydrolysis going on and
over a couple of hours, you can see orthonitrophenol being produced.
                                                                       102

-------
               DETECTION OF MICROORGANISMS BY  ION MOBILITY SPECTROMETRY
A.P. Snyder, M. Miller
 and D.B.  Shoff
U.S. Army  Chemical Research,
 Development and Engineering
 Center, Aberdeen Proving
  Ground,  MD  21010-5423
G.A. Eiceman
New Mexico
 State University
 Las Cruces, NM
  88003
D.A. Blyth & J.A. Parsons
GEO-CENTERS, INC.
c/o U.S. Army Chemical
  Research, Development
  and Engineering Center
Aberdeen proving Ground,
 MD  21010-5423
ABSTRACT

A relatively  new  concept is explored
where  the  potential  for  ion mobility
spectrometry  is  investigated for the
detection  and determination of living
microorganisms.   The hand-held,
NATO-fielded  Chemical Agent Monitor
(CAM)  embodies the  analytical device.
Advantage  is  taken  of the inherent
enzymes found in  microorganisms and an
exogenous,  tailored  substrate was
provided in order to initiate the
desired biochemical  reaction.  The
substrate  was ortho-nitrophenyl-beta-
D-galactopyranoside, and the product,
ortho-nitrophenol,  can be detected in
the negative  ion  mode of the CAM and
signals the presence of  bacteria.
Detection  limits  of  approximately 10E4
E. coli bacterial cells  in 5 min. and
3300 E. coli  cells  in 15 minutes were
realized.   The results suggest a new
application of the  CAM in the
screening  of  bacterial contamination
in community  water  and wastewater
testing situations.
KEYWORDS:   ion  mobility spectrometry;
microorganisms; E.  coli; enzymes;
ortho-nitrophenol;  Chemical Agent
Monitor; ortho-nitrophenyl-galacto-
pyranoside;  fecal coliforms.
INTRODUCTION

Detection  and  identification of
microorganisms is a challenge in view
        of the required sensitivity, selec-
        tivity, and time of response of the
        detection technique.  Table 1 lists
        these requirements for a number of
        methods.   It appears that analytical
        instrumentation techniques broadly
        fall in the detection limit range of
        10E6 bacterial cells with an instru-
        mental response time of approximately
        1.5 hr.  The colorimetric and fluoro-
        metric enzyme assay procedures fare
        better and can be characterized by
        10E3-10E5 bacterial cell limits of
        detection in a 0.25-4 hr response
        time domain.

        ion mobility spectrometry (IMS) is a
        straightforward, analytical vapor
        detection technique.  Neutral analyte
        vapors enter the device and are
        ionized,  usually by a   Ni ring.  The
        ions are  electrically gated and
        "drift" through an antiparallel flow
        of buffer gas (air or nitrogen).  The
        ions are  focussed by an electrical
        field about the heated, cylindrical
        drift region and are registered by
        a  Faraday cup detector.  The entire
        process,  from vapor sampling to the
        detection event, takes place at
        ambient or near-ambient pressure, and
        thereby atmospheric pressure ioniza-
        tion chemistry characterizes the ion
        formation process.  Ions are parti-
        tioned primarily according to their
        mass and  shape and 'are characterized
        by their  corrected drift times (typi-
        cally in  msec)  or ion mobilities.  In
        the negative mode, IMS is very simi-
        lar with  respect to an electron
                                        103

-------
capture detector in terms of the
detection event and sensitivity.

The detection of bacteria by IMS
originated from the concept of aug-
menting the hand-held Chemical Agent
Monitor (CAM) with capabilities for
biological detection, more specifi-
cally, that of viable microorganisms.
The Hypothesis was that the ion-mole-
cule chemistry that characterizes the
atmospheric pressure-based IMS tech-
nique, embodied by the CAM device,
could be used to detect a targeted
volatile product of the biochemical
reaction between an rn vivo bacterial
enzyme and a tailored organic
substrate.  This proved to be an
interesting challenge because
parallels could be drawn with that of
standard, well-established microbio-
logical and clinical bacterial evalu-
ation procedures in the process of
devising the CAM detection of micro-
organisms .
EXPERIMENTAL

Ortho-nitrophenol (ONP) and ortho-
nitrophenyl-beta-D-galactopyranoside
(ONPG) were obtained from Aldrich
Chemical Co., Inc., Milwaukee, WI and
Sigma Chemical Co., St. Louis, MO,
respectively.  The beta-galactosidase
enzyme and ONP-acetate were obtained
from Sigma Chemical Co., St. Louis,
MO.  Pure E. coli suspensions  (ATCC
11303) or Bacillus globigii (ATCC
9372) were prepared by growth  in a
nutrient broth solution for 48 hr
which was supplemented with 0.5%
lactose sugar for induction of the
beta-galactosidase enzyme.  The
bacterial growth was centrifuged
and the pellet was washed three times
with a sterile 0.1M phosphate-
buffered saline solution (0.7% NaCl)
at pH 7.4 (PBS).  Approximately Ig of
human fecal matter was suspended in
10 ml of distilled water.  Strips of
Whatman 15 filter paper (Whatman
International, Ltd., Maidsgone,
England) were baked at 150 C over-
night in glass vials and used  for
bacterial determination experiments.
Two microliters of the E. coli or
fecal matter suspensions were  used
for filter paper experiments and 0.1
ml was used for bulk volume liquid
experiments.  Two microliters of a
2.0 mg/ml ONPG solution in PBS were
used for the filter paper experiments
 and 1.9 ml of the same ONPG solution
 was used for bulk volume microbial
 determinations.  The fecal bacterial
 experiments were conducted at room
 temperature (25 C)  while the pure E.
 CQp.i experiments were carried out at
 38 C.

 After selected 'incubation periods at
 the given temperatures,  the headspace
 of the bottle was sampled with the
 hand-held CAM by removing the cap and
 immediately placing the  vial opening
 at the inlet of the CAM  unit.

 The hand-held CAM (Graseby Ionics,
 Ltd., Watford, England)  device was
 used as the analytical detection
 technique which was designed speci-
 fically for air sensing  in military
 field applications  (15).   Signals
 from the CAM were processed by using
 a  Graseby Ionics, Ltd.,  advanced
 signal processing (ASP)  board and
 software with an IBM-PC/AT.   Details
 of the CAM unit are as follows:
 drift gas,  nitrogen or air;  ion
 source,  10-mCi   Ni;  drift region
 length,  7 cm;  drift field,  230 V/cm;
 drift gas flow,  300 mL/min;  reaction
 region length, 3 cm;  drift tube tem-
 perature, ambient;  shutter  width,
 0.1 msec (16).   A schematic  and de-
 tails of the  operation of  the hand-
 held CAM ion  mobility spectrometry
 unit can be  found elsewhere  (17).
 For  the  fecal  bacteria experiments
 the  data were  captured and  displayed
 by the ASP  software  while  for  the
 pure E.  coli determinations,  the  ion
 mobilTty signals were  captured and
 displayed by  a Nicolet 4094A  oscillo-
 scope  and Hewlett/Packard  7470A
 plotter.
RESULTS AND DISCUSSION

A number of constraints were
realized in that for a system such as
the CAM to be a realistic analytical
method for biological detection, only
minimal logistic burdens to the
collection, processing and introduc-
tion of the sample to the hand-held
IMS unit would be tolerated.  There-
fore the question was posed:  How can
the CAM be used as it is intended
(i.e. - a vapor detector)  in the
detection and possible identification
of extremely complex entities such as
microorganisms?  The microbiological
                                          104

-------
literature provided constructive  in-
sights  into this problem  in  the  form
of constitutive enzymes  (enzymes  that
are always present  in  a bacterium)
that are secreted at significantly
different quantitative levels  depen-
ding on the organisms. This is  a
property of living  active cells  and
not of dead or dormant microorgan-
isms.  Conventional clinical proce-
dures used in the detection  and  iden-
tification of organisms rely on
tailored substrates  (i.e. -  compounds
that mimic the enzyme's natural
substrate) to interact selectively
with the secreted enzymes of bac-
teria.  The enzyme-catalyzed products
of natural substrates  are usually
spectroscopically-silent  and as  such,
tailored compounds  substitute  a  por-
tion of the natural substrate  with  a
compound such that  when  it is  re-
leased, it becomes  spectroscopically
active  (e.g. - colorimetric  or
fluorimetric properties).  This  con-
cept was then related  to  the proposed
CAM detection of bacteria, except
that the product would have  to dis-
play a  relatively high vapor pressure
and the CAM must respond  to  the
product.

Enzyme  Substrate and  product

Previous  investigations  in this
laboratory  (13)  have  shown that
bacteria  such as Bacillus subtilis
(BG) ,  the yeast  Saccharomyces
cerevisiae,  Serratia  marcescens   (SM)
and  E.  coli  produced  at  various  rates
the 3-hydroxyindole (indoxyl)  as a
highly fluorescent  and blue colori-
metric product  from the  reaction of
indoxylacetate,  indoxylglucoside and
indoxylphosphate with their respec-
tive  esterase,  glucosidase and phos-
phatase enzymes.   4-methyl-umbelli-
feryl-beta-D-galactoside reacted with
the beta-D-galactosidase enzyme  in E.
coli  and  SM to  produce the
fluorescent  4-methylumbelliferone
product (13).   The  indoxylacetate
probe (13)  was  the  most  sensitive
where as  little as  500 BG cells/ml
could be  detected  in  under  15
minutes.   Modification of these
substrates,  with extensive biochemi-
cal IMS experimentation  underscored
the role  of the organic  substrate  as
the heart of the project.  A  number
of important requirements concerning
the substrate must be satisfied  in
order to ensure a successful
approach.  Requirements of the sub-
strate include that  it  (a) is water
soluble, (b) is recognized by a  tar-
geted enzyme,  (c) displays rapid
enzyme-substrate kinetics  (i.e.  -
favorable association constant),  (d)
has minimal/negligible  spontaneous
hydrolysis and  (e)  that  it gives  a
minimal/negligible  response  to  the
CAM.  Requirements  for  the product
include  (a)  a  low association
constant with  biological  material,
(b) a relatively low water solubil-
ity,  (c) favoring the gaseous  phase,
and  (d)  being  "CAM-active".    Alter-
nate compounds were sought.   instead
of ester compounds, established
microbiological  colorimetric indica-
tors were  analyzed.  ONPG displays an
acetal  functional group that joins
ONP  and  the  beta-D-galactopyranoside
sugar monomer  (Figure 1)  and is a
standard microbiological indicator
for  the  detection  of all (total)
fecal  coliform bacteria  (18, 19).
Fecal  coliform bacteria belong to the
Enterobacteriaceae  and are comprised
of  E.  coli  (4xlOE8  cells/g feces) ,
Klebsiella sp. (5xlOE4 cells/g) ,
EnterobacTer  (10E5 cells/g)  and
Citrobacter (10E6 cells/g)  (20) .
 These bacteria, with E. coli as the
 predominate species, are tound  in
 fecal matter, and  the latter three
 genera are  also associated with
 plants and  soils.   E. coli, however,
 can only be found  in the environment
 through fecal contamination  (21).

 Figures 1 and 2 pictorially display
 the enzyme-substrate biochemical and
 detection events of the  ONP product
 by the CAM.   Figure 3  shows a  CAM
 response of a phosphate-buffered
 saline  solution of ONP  in the
 negative  ion  mode.  The  main  peak  at
 6.2 msec,consists  largely of
 O  (H 0)     clusters and  the  shoulder
 t£ ti?e  ¥eft of  the peak  is
 characteristic  of  the  chloride ion.
 The peaks at  9.1 msec  represent the
 ONP monomer at  different concentra-
 tions and  the low  intensity peak at
 11.7 msec  represents the dimer ion
  (22).   Thus,  a favorable analytical
 situation  has been established in
 that  a  compound has been found that
 not  only  has  established roots in the
 microbiological detection and
  identification arsenal as a
 colorimetric  indicator but also
                                         105

-------
 responds to ion mobility spectrometry
 through well established ion-mole-
 cule, gas-phase reaction chemistry.

 CAM-Bacterial Trials

 A buffered solution of the ONPG
 substrate produced no response from
 the CAM unit.  When an aliquot of
 pure beta-D-galactosidase enzyme was
 added to the ONPG solution, a yellow
 color appeared within seconds and the
 CAM ASP registered this event in the
 negative ion mode in a fashion
 similar to that in Figure 3.   Bac-
 terial tests followed.  one was from
 a pure culture of E. coli and the
 other bacterial source was of fecal
 origin.  Microliter volumes of
 bacterial sample and buffered ONPG
 were spotted on a strip of sterile
 filter paper and the latter was
 inserted into a vial.  The vial was
 secured with a screw cap in order to
 contain any ONP product that  was
 released into the vial headspace.
 For the fecal  suspension, 2
 microliters of a Ig feces/10  ml
 distilled water was used.  Since an
 approximate concentration of  E. coli
 in  human fecal matter (20)  is~~about
 4xlOE8 cells/g,  the actual  applied
 amount of bacteria was approximately
 8xlOE4 cells.   Figure 4 portrays the
 results of  this study.   Position A
 in  Figure 4  represents a background
 CAM response of the bacterial
 inoculation  without ONPG substrate.
 The bacterium  does provide  distinct
 ion mobility peaks which are most
 likely due  to  inherent bacterial
 volatile  compounds.   A blank
 consisting of  the  buffered  ONPG
 solution  produced  only the  negative
 background  ion mobility  signal
 (Position B  in  Figure 4).   Position  C
 in  Figure 4  represents the  CAM
 response  of the  vial  headspace  after
 the  buffered ONPG  substrate was  added
 to  the  bacterial spot  on  the filter
 paper  and was  acquired  40 min.  after
 substrate addition.   Note that  in
 addition  to the  background  ion
mobility  signal  and  the  three peaks
 representing the bacterial  volatile
products, a new  peak  appeared at  9.1
msec which matched  that of  ONP
 (Figure 3).    Figure  5 shows a
replicate experiment  where  frame A
represents the ONP  response 15 min.
after an ONPG solution was added  to a
fecal inoculation on a filter paper
strip.  Frame B shows that at 45 min,
the ONP signal grew considerably.

An inoculated dose of 10E4 E. coli
cells from a pure suspension on a
filter paper strip produced a peak in
five minutes (Figure 6).  At the same
E. coli inoculation, Figure 6 also
shows the CAM "response to the
production of ONP after 10, 15 and 20
minutes.  The background shows
essentially no peak in the 9.1 msec
time window and the reaction
consisted of 2 ul of phosphate buffer
added to 2 ul of ONPG.  This
indicates that the spontaneous
hydrolysis of ONPG at 38 C is minimal
and intense ONP signals can be
observed over a relatively short
period of time resulting from the
bacterial enzymatic reaction at the
relatively low amount of 10E4 E. coli
cells.  Figure 7 shows similar data
except that the amount of inoculated
E. coli was 3.3xlOE3 cells.  Indeed,
within 20 min., a clear ONP signal
was observed at 9.1 msec.  This
experiment was repeated (Figure 8)
and in 15 min. a discernible ONP peak
was observed.  A bulk 2.0 ml volume
suspension consisting of ONPG and
fecal matter (a total of 4xlOE6 fecal
bacterial cells) took 2 hr for a
response from CAM while the yellow
ONP color in the suspension was
observed prior to the CAM detection
event.  The longer dwell time is to
be expected because the relatively
large volume of water had a small
surface area for the ONP to partition
into the gas phase as opposed to
microliter amounts which rapidly
diffuse across a strip of filter
paper .
Other Enzyme/Substrate Complexes

ONP-acetate can be cleaved by an
esterase and this compound was used
in the determination of the lipase
enzyme in Bacillus globigii.   Table
2 presents the amount of bacteria
used to generate an ONP ion mobility
peak after a 15 min. incubation time.
One thousand cells of B. globigii
produced an ONP signal comparable to
that of Figure 6E.  However, with the
ONPG substrate, no signal was
observed with 10E5 cells.  The
absence of an ONP signal is due to
                                        106

-------
c —
approximately  3.:
bacterial cells.
the  fact  that  B.  globigii,  as well as
most other  bacilli,  do not  contain
the  beta-galactosidase enzyme and as
such ONP  is not produced.  The
opposite  situation occurs with E.
coli.   As Table 2 indicates,  E. coli
provides  a  positive  biochemical
reaction  with  ONPG,  but not with
ONP-acetate.

Comparison  to  Other  Techniques

For  the E.  coli fecal coliform ONPG
test,  the CAM  unit was observed to
provide an  ONP signal in 15 min. with
               ,3xlOE3 E. coli
                  It is of  interest
to compare  these  response time/
inoculation figures  of merit with
that of established  and potential
microbiological,  clinical and
analytical  instrumentation
techniques. Table  1 provides a list
of a number of these methods
including total number of bacteria
and  the time needed  for a reliable
analysis  of bacterial presence.  The
CAM  concept of bacterial detection
via  inherent enzyme  biochemical
reactions which yield tailored
volatile  products appears to be a
competitive technique in the
determination  of  microbial  presence.
CONCLUSIONS

A major  step in  the  chemical detec-
tion and identification of viable
(i.e. -  living)  microorganisms was
presented  in terms of analytical
techniques.   The ion-molecule
chemistry  associated with IMS was
shown to be  a promising avenue for
the monitoring of bacterial presence
by taking  advantage  of available sub-
strate-induced accessible enzymes.
The hand-held ion mobility spectrom-
eter CAM unit displayed detection
sensitivity  levels for E. coli fecal
coliforms  and response times similar
or better  than that  of most commer-
cially-available methodologies and
analytical instrumentation techni-
ques.  This  suggests a potential
application  of IMS for screening of
bacterial  presence in community/local
water and  wastewater testing
protocols.
                                            ACKNOWLEDGEMENT

                                            The authors wish to thank Ms. Linda
                                            Jarvis for the preparation and
                                            editing of the manuscript.
                                            REFERENCES
                                            1.
                                            4.
Newman,  R.S. ,  and  O'Brien,  R.T.,
"Gas  Chromatographic Presumptive
Test  for  Coliform  Bacteria  in
Water,"  Appl .  Microbiol.  Vol.
1975,     ~
                                                                              30,
Bachrach,  u.  and  Bachrach,  z.,
"Radiometric  Method  for  the
Detection  of  Coliform  Organisms
in Water,"  Appl.  Microbiol. vol.
28,  1974,  pp.  169-171.

Wilkins, J.R.,  Young,  R.N.  and
Boykin, E.H. ,  "Multichannel
Electrochemical Microbial
Detection  Unit,"  Appl. Environ.
Microbiol.  vol. 35~T"T978, pp.
214-215.

Cady, P.,  Dufour,  S.W.,  Shaw, J. ,
and  Kraeger,  S.J., "Electrical
Impedance  Measurements :  Rapid
Method for  Detecting and
Monitoring  Microorganisms," J.
Clin. Microbiol.  vol.  7, 197?,
pp.  265-272.

Fraatz, R.J.,  Prakash, G. and
Allen, F.S.,  "A Polarization
Sensitive  Light Scattering  System
for  the Characterization of
Bacteria,"  Am.  Biotechnology Lab.
Vol. 6, 1988, pp.  24-28.

Libby, J.M., and Wada, H.G.,
"Detection  of Neisseria
meningitidis and YersTnia pestis
with a Novel Silicon-Based
Sensor," J. Clin. Microbiol . vol.
27,  1989, pp. 1456-1459.

Shelly, D.C., Quarles, J.M., and
Warner, I.M., "Preliminary
Evaluation  of Mixed Dyes for
Fingerprinting  Non-Fluorescent
Bacteria,"  Anal . Lett. , vol.
14(813), 1981,  pp. 1111-1124.

Steinkamp,  J.A. , Fulwyler ,  M.J.,
Coulter, J.R.,  Hiebert, R.D.,
Homey, J.L. and Mullaney,  P.F.,
"A New Multiparameter Separator
                                         107

-------
    for Microscopic Particles and
    Biological Cells," Rev. Sci.
    Instrum. Vol. 44, 1973, pp.
    1301-1310.

9.  Graham, K.,   Keller, K. , Ezzel,
    J. and Doyle, R., "Enzyme-Linked
    Lectinosorbent Assay (ELLA) for
    Detecting Bacillus anthracis,"
    Eur. J. Clln. MicroDio.1. vol. 3,
    1984, pp. 210-212.

10. Feng, P.C.S. and Hartman, P.A.,
    "Fluorogenic Assays for
    Immediate Confirmation of
    Escherichia  coli ," Appl.
    Environ. Microbiol. Vol. 43,
    1982, pp. 1320-1329.

11. Warren, L.S., Benoit, R.E. and
    jessee, J.A., "Rapid Enumera-
    tion of Fecal Coliforms in Water
    by a Colorimetric beta-Galacto-
    sidase Assay," Appl. Envi ron.
    Microbiol. Vol. 35, 1978, pp.
    136-141.

12. Godsey, J.H., Matteo, M.R.,
    Shen, D., Tolman, G. and Gohlke,
    J.R., "Rapid Identification of
    Enterobacteriaceae with
    Microbial Enzyme Activity
    Profiles," £. Clin. Microbiol.
    Vol. 13, 1981, pp. 483-490.

13. Snyder, A.P., Wang, T.T. and
    Greenberg, D.B., "pattern
    Recognition Analysis of In
    Vivo Enzyme-Substrate
    Fluorescence Velocities in
    Microorganism Detection and
    Identification," Appl. Environ.
    Microbiol. Vol. 51, 1986, pp.
    969-977.

14. Berg, J.D. and Fiksdal, L.,
    "Rapid Detection of Total and
    Fecal Coliforms in Water by
    Enzymatic Hydrolysis of
    4-Methylumbelliferone-beta-D-
    Galactoside," Appl. Environ.
    Microbiol. Vol. 54,
                    1988, pp.
    2118-2122.
                                        16.  Eiceman,  G.A.,  Shoff,  D.B.,
                                            Harden,  C.S.,  Snyder,  A.P.,
                                            Martinez, P.M., Fleischer,  M.E.
                                            and  Watkins,  M.L.,  "Ion  Mobil-
                                            ity  Spectrometry of Halothane,
                                            Enflurane, and  Isoflurane
                                            Anesthetics in  Air  and
                                            Respired  Gases," Anal. Chem.
                                            Vol. 61,  1989,  pp.  1093-1099.

                                        17.  Eiceman,  G.A.,  Snyder, A.P.
                                            and  Blyth, D.  A., "Monitoring
                                            of Airborne Organic Vapors
                                            using Ion Mobility  Spectrom-
                                            etry,"   Intl.  J. Environ.
                                            Anal. Chem. Vol. 38, 1990,
                                            pp 415-425.

                                        18.  Paik, G., "Reagents, Stains,
                                            and  Miscellaneous Test
                                            Procedures, in  Manual of
                                            Clinical Microbiology, Third
                                            Edition,  E.H.,  Lennette, A.
                                            Balows,  W.J.  Hausler,  Jr. and
                                            J.P. Truant,  eds.,  American
                                            Society for Microbiology,
                                            Washington, DC, 1980,  p. 1006.

                                        19.  Colilert Most Probable Number
                                            Method Product Brochure, Access
                                            Medical Systems, Inc.,
                                            Branford, CT  06405, 1989.

                                        20.  Olivieri, V.P., "Bacterial
                                            Indicators of Pollution," in
                                            Bacterial Indicators of
                                            Pollution, W.O. Pipes, ed., CRC
                                            Press, Boca Raton,  FL, Chapter
                                            2, 1982.

                                        21.  Stratman, S., "Rapid Specific
                                            Environmental Coliform
                                            Monitoring," Am. Lab, vol. 20,
                                            1988, pp. 60-64.

                                        22.  Snyder, A.P., Shoff, D.B.,
                                            Eiceman,  G.A.,  Blyth, D.A. and
                                            Parsons,  J.A.,  Anal. Chem.,
                                            1991, in  press.
15.
CAM Chemical Agent Monitor;
Commercial brochures from
Graseby Ionics, Ltd.:  Watford,
England, 1988.
                                         108

-------
TABLE 1.  COMPARISON OF MICROORGANISM  DETECTION  BY  IMS TO OTHER
          TECHNIQUES
Total number
of bacteria

   8°c
   10i
   105
   io7
   107
   10-

   10
11
1
2.7xl04
10?
IO4
IO5
5xl07
IO5
10^
0.5
3
9
4
4.25
0.5
0.25
0.25
 3.3x10-
Time
(hr)    Technique

8.5     gas chromatography
1       radiometry
1.5     electrochemical
0.5     organism growth
0.25    polarized light
        scattering
0.4     light-addressable
        potentiometric sensor
1       excitation-emission
        matrix
        3-laser flowthrough
        cytometry
        enzyme-linked
        lectinosorbent assay
        H2/CO2 evolution
        gfucuronidase enzyme
        extracellular enzyme
        aminopeptidase enzymes
        extracellular enzymes
        extracellular enzymes,
        nutrients
0.25    CAM
                                           Response
                   Reference
ethanol metabolite    1
  CO  metabolite      2
H« metabolte          3
electrical impedance  4
Mueller matrix        5

redox potential       6

fluorescence          7

fluorescence          8

lectin-conjugate      9

visual/ gas bubbles   10
fluorescence          10
colorimetric          11
fluorescence          12
fluorescence          13
fluorescence          14

vapor metabolite      this
                      study
    TABLE 2.  ENZYME/SUBSTRATE BIOCHEMICAL REACTIONS PROBED IN MICROORGANISMS
                                                        PRESENT LIMIT
ORGANISM
E. coli
E. coli
Bacillus subtilis
B. subtilis

ENZYME
PROBED
/3-galactosidase
Lipase
/3-galactosidase
Lipase
SUBSTRATE
ONPG
ONP acetate
ONPG
ONP acetate
OF DETECTION
(Bacterial Cells)*
3.3 x IO3
6 x IO5"
IO5**
IO3
     *Within 15 minutes
    **No signal observed at the  given  concentration
                                  109

-------
                                            coll  +  e«l«cto»« + ortho-nlfro(>h»nol
                                         •ruym*                    (ONP)
     FIGURE 1.  PICTORIAL REPRESENTATION OF THE £. COLI/EETA-SAIACTOSIDASE BIOCHEHICAL
               REACTION KITH THE OHP6 SUBSTRATE.
                           HAND-HELD VAPOR DETECTOR
        or0*
      C3
             -KO,
-^

                  Small lon« trav«t («al»r than larg* loni In «n •:*ctrlcal Br«dl»nl

      FIGURE 2.  PICTORIAL REPRESENTATION OF THE ONP DETECTION EVENT WITH THE CAK
                HAND-HELD MONITOR.  REFERENCE 17 PROVIDES DETAILS OF THE OPERATION
                OF THE CAM.
                                                                           IMS
                                                                           SCAN
                                                                          NUMBER
                                                                           (SEC)
                               6.2
                                                  11.7
                                         KSEC
FIGURE 3.  ION  MOBILITY SPECTRUM OF  ONP IN THE  NEGATIVE MODE.   THE PEAK  AT 6.2
             MSEC REPRESENTS THE  BACKGROUND ION SIGNAL  AND THE PEAKS THAT LIE  AT
             9.1 MSEC  REPRESENT ONP AT DIFFERENT RELATIVE CONCENTRATIONS.
                                             110

-------
                                                                  32
                         S.2
                               MSEC
                                     9.1
FIGURE 1.   ION MOBILITY SPECTRUrt  IN THE NEGATIVE  ION MODE  (A) OF AN  INOCULATION
            OF 8x10^ FECAL BACTERIAL CELLS ON A-FILTER PAPER STRIP. (B)  OF ONPfi
            SOLUTION ON A FILTER PAPER, (C) AFTER  10 KIN. FROM AN ONP6 SOLUTION
            ADDED TO AN  INOCULATION OF SxlO11 FECAL BACTERIAL CELLS  ON A  STRIP  OF
            FILTER PAPER.  A PEAK  AT 9.1 RSEC,  DUE TO ONP,  ONLY  APPEARS  VHEK BOTH
            OHP6 AND BACTERIAL CELLS ARE PRESENT.
                             J
                      w**wvrK'»»

                     MiU^uv-'
                                 	T~
                                 6.2
                                       KSEC
                                              9.1
  FIGURE 5.  (A) 15 HIK. AND (B) 15 HIM. ION MOBILITY SPECTRA OF A REPLICATE FECAL
             BACTERIA EXPERIMENT (REFER .TO FIGURE 1C FOR DETAILS).
FIGURE 6.
                               ION  MOBILITY  SPECTRA OF ONP LIBERATED FROM THE  REACTION  OF lO4 £. £flU
                               CELLS AND OKP6 KITH  AN INCUBATION AT 58°C FOR (B) 5 HIN  (SHADED AREA),
                               (C)  10 HIK,  (D> 15 «IN,  20 HIK.   FRAME A REPRESENTS  THE  ION MOBILITY
                               SPECTRUM  OF A BLANK  CONSISTING CF  TWO HICROLITERS OF  BUFFER AND OKP6
                               SOLUTIONS ON A PIECE CF FILTER PAPER,
6.2       9.1
    nstc
                                           111

-------
                             6.2     9.1
                                RSEC
FIGURE 7.  ION MOBILITY SPECTRA OF ONP LIBERATED FROM THE REACTION OF J.JxlO3
           £. COL1 CELLS AND ONP6 KITH AN  INCUBATION AT 38°C FOR  (B) 5 MIN,
           (C) 10 MIN, (D) 20 MIN.  FRAME A REPRESENTS THE ONP6 BLANK.  NOTE
           THAT ONLY FRAME D SHOWS A CLEAR ONP RESPONSE OVER BACKGROUND.
                                6.2     9.1
                                    KtC

  FIGURE  8.   REPLICATE EXPERIMENT OF FIGURE 7 EXCEPT THAT SPECTRUM D  WAS  TAKEN
             AT 15 MINUTES.   NOTE THAT ONLY FRAME  D  SHOKS A  CLEAR  ONP RESPONSE
             OVER  BACKGROUND.
                                        112

-------
                          DATA ANALYSIS TECHNIQUES FOR
                            ION MOBILITY SPECTROMETRY
                                 Dennis  M.  Davis
               Analytical Research Division, Research Directorate
        U.S. Army  Chemical Research,  Development and Engineering Center
                    Aberdeen  Proving  Ground, MD 21010-5423.
               ABSTRACT

    The past several years have  seen
 the advance of ion mobility
 spectrometry (IMS) as an analytical
 technique.  Most of these advances
 have been made in the hardware
 development end of the problem,  the
 result being that portable IMS
 devices have begun to appear  in  the
 marketplace.  The other end of the
 problem, the signal processing and
 data analysis techniques, has not
 been addressed to the same degree.
 Recent attempts at applying data
 analysis techniques to IMS data  have
 been made, and the results are
 encouraging.  Data processing
 algorithms ranging from those which
 perform simple tasks to those
 performing more difficult tasks  have
 been developed.  Among the algorithms
 which will be discussed are
 algorithms for measuring the  peak
 areas of selected peaks of interest
 in biological studies,  and linear
 discriminant analysis for detecting
 and identifying industrial chemicals
 at, or near their maximum exposure
 limits.

            INTRODUCTION

   When dealing with environmental
 issues, there are two points  of
 emphasis that must be considered.
 These  two points of emphasis  are  the
protection  of individuals  in  the
workplace,  a task regulated by the
 Occupational Safety and Health
 administration (OSHA), and the
 protection of the environment in
 which we live, a task regulated by
 the Environmental Protection Agency
 (EPA).  These two points of
 emphasis, while dealing with the
 same general problem, are typically
 at different ends of the
 concentration range of chemical or
 biological contamination or
 exposure.  The concentration ranges
 for which one must monitor an
 individuals exposure to chemical and
 biological contaminants is usually
 in the low parts-per-million, ppm,
 range to tens of thousands of ppm
 [1-3],  and is set by Federal law
 [3].   The concentration range which
 is monitored for environmental
 compliance is usually parts-per-
 billion,  ppb,  to low ppm.   A useful
 method for the monitoring both
 concentration ranges at the same
 time  is ion mobility spectrometry,
 IMS.

      Ion  mobility spectrometry is
 based upon the flow,  or drift,  of
 molecular ions through  a gas  of
 uniform temperature  and pressure.   A
 weak  electric  field  is  uniformly
 applied to the gas in the drift
 region  of  the  IMS, causing  the  ions
 to move along  the field  lines.
 These ions  continue to drift until
 their movement  is impeded by
 collisions with neutral gas
molecules.  Since the electric field
 is still being applied to the gas,
                                        113

-------
 the ions are accelerated once again
 and the process  of acceleration and
 collision is repeated until the ions
 strike the detector.   IMS is similar
 to  Time of Flight  mass spectrometry
 in  that the electric  field causes  the
 ions to drift, but it differs in that
 Time of Flight mass spectrometry is
 performed under  vacuum and there are
 few,  if any collisions to retard the
 ions.   The average velocity,  vd, of
 the ions is determined by millions of
 the accelerations  and energy-losing
 collisions.   The time required for an
 ion to traverse  a  known distance in
 the drift region of the spectrometer
 is  the drift time,  td.

       The average  velocity of the
 ions,  also called  the drift velocity,
 is  related to the  strength of the
 applied electric field through the
 equation
vd = Id /
                      =  KE
(1)
where vd  is  the drift velocity,  ld  is
the  length of the drift  region of the
spectrometer, td is the  drift time  of
the  ion,  E is the electric  field
strength, and K is a constant of
proportionality.  This constant  K is
also called  the "mobility"  of the
ion.  The mobility of the ion is
directly  dependent upon  both the
molecular ion being studied, and the
neutral gas  through which the ion
must drift.  A more useful  constant
which is  used in IMS work is the
"reduced  mobility" of the ion.   The
reduced mobility of the  ion, the
mobility  of  an ion through  a gas at
standard  temperature and pressure, is
related to the measured  mobility of
the ion through the equation
    K0 = K  (273.15/T)  (P/760)
                         (2)
where T is the absolute temperature
of the gas in the drift region, P is
the total pressure of the gas and the
ions in the drift region, and Ko is
the reduced mobility of the ion.
Because it is often difficult to
measure the temperature and pressure
within the drift region of the
spectrometer, a common practice which
is used in determining the identity
of ions is to measure the ratio of
the reduced mobility of the ion of
interest to that of a known species.
This known species is usually the
reactant ion for the study.  If the
neutral gas is air, the reactant
ions are H3O+ when dealing with
positive ions, and 02"when dealing
with negative ions.  The ratio of
the reduced mobilities are related
to measurable quantities through the
equation

(K01/K02) = (Ki/K2) = (td2/tdl) (4).

The only parameters which are needed
in the analysis is the ratio of the
drift times for the ions.

   The equation for calculating the
mobility of an ion through a gas has
been shown to dependent on the
first-order collision integral
[4,5], which is proportional to the
transport cross section.  This
implies that the mobility of an ion
is dependent on the size of the
ions, the shape of the ion, and the
distribution of charge on the ion;
this results in the possibility of
more than one ion having the same
mobility.

     In an ion mobility
spectrometer, Figure 1, the sample
is introduced through a sample inlet
probe.  This inlet probe contains a
semi-permeable membrane, which
allows only a portion of the sample
to enter the ionization chamber.
The portion of the sample which does
not enter the ionization chamber is
vented through the exhaust.  The
carrier flow gas, which is input
directly into the ionization chamber
and the sample are then exposed to
the ionizing source,  a 63Ni source
in this work.   The ions and the gas
molecules are then allowed to mix
and react in the ionizing chamber.
Typical ion reaction schemes which
take place in the ionization chamber
are shown in Table A.   A driving
pulse of known shape and duration is
then applied to the bipolar gating
grid, allowing the mixture to enter
the drift region of the
spectrometer.   While in the drift
region, the ions are subjected to an
applied electric field (200 V/cm in
our studies),  which causes the ions
to begin their acceleration and
collision process.  After the ions
have traversed the drift region,
                                          114

-------
                                      TABLE A

                           TYPICAL ION REACTION SCHEMES

           Typical Positive  Ion Reactions
                (X is the species to be detected)

           Typical Negative Ion Reactions

             02~+  AB  -> 02  + AB~
             02~+  AB  -> 02 + A + B~
             O2~+  AB  -> (AB'O2)~
               (AB is the species to be detected)
 they strike the collector  electrode
 The signal is then processed  to
 produce the ion mobility spectrum.
 For those who wish, a more detailed
 description of ion mobility
 spectrometry can be found  elsewhere
 [6].

      The past several years  have
 seen the advance of ion mobility
 spectrometry as an analytical
 technique, with the utility of IMS as
 an analytical tool for the rapid
 detection of airborne vapors  in the
 atmosphere being previously
 demonstrated [7-10], and computer
 techniques for pre-processing IMS
 signals have also been presented [11-
 12].
            EXPERIMENTAL
Equipment
      Data were collected on an IMS
spectrometer [Airborne Vapor Monitor
(AVM)  from Graseby Analytical,
Watford,  Great Britain] and stored on
an IBM Personal Computer.  The data
transfer  is accomplished using a
Graseby Analytical Advanced Signal
Processing (ASP)  board and its
associated software.   Each spectrum
consisted of 640  data points, which
was collected at  a sampling frequency
of 30  kHz.   The other operational
parameters of the AVM are shown in
Table  B.
Vapor Generation
   The vapors being used in the
linear discriminant data set are
generated with a Q5 vapor generator,
shown in Figure 2 .  The Q5 generator
has 16 component parts.  These parts
are: (1) an equilibrator assembly,
(2) an air supply  (or nitrogen
supply) stopcock,  (3) a constant
pressure regulator (stabilizer) for
the air supply, (4) two sampling
bubblers filled with solvent (the
bubbler is not shown Figure 2), (5)
a flowmeter (manometer) for the air
supply, (6) a constant pressure
regulator  (stabilizer) for the
diluent air supply, (7) stopcocks
for the stabilizers, (8) a stopcock
shut off the flow of air from the
equilibrator to the mixing chamber,
(9) a flowmeter (manometer) for the
diluent air supply, (10) a mixing
chamber, (11)  a reservoir, (12,13)
sampling stopcocks, (14) a reservoir
exhaust stopcock,  (15) a charcoal
trap on the exhaust of the reservoir
(not shown in Figure 2) , and (16)  a
charcoal canister on the sampling
line after the SAW device  (not shown
in Figure 2) .
   The equilibrator assembly is the
liquid test reagent container of the
dilution apparatus.  Dry air, under
a constant controlled pressure,
flows into the equilibrator.  This
air stream passes over the surface
of the test reagent, and becomes
                                       115

-------
                                      TABLE B

                         OPERATIONAL PARAMETERS FOR THE AVM

            Number Of Waveforms To Be Summed  -    32
            Number Of Samples Per Waveform    -   640
            Gating Pulse Repetition Rate      -    40 Hz
            Gating Pulse Width                -   i80 US
            Delay To Start Of Sampling        -     o us
            Sampling Frequency                -    30 KHz
            Gating Pulse Source Is
                                              ** External **
 saturated with the reagent vapor.
 The equilibrator is maintained at a
 constant temperature of 25 °C by
 partial immersion in a constant
 temperature water bath.  Included in
 the equilibrator is a porous alumina
 cylinder (from Thomas Scientific,
 Swedesboro, N.J.)  to produce a
 greater surface area for the liquid-
 vapor equilibration. The dry air-
 test vapor mixture flows from the
 equilibrator assembly to the mixing
 chamber where it is diluted with dry
 air to the required concentration of
 milligrams test vapor per liter of
 dry air.

      The flow of air through the
 equilibrator is controlled by an in-
 line stopcock,  a constant pressure
 regulator,  and a flowmeter.   The
 stopcock is located at the inlet of
 the equilibrator,  and acts as the
 shutoff valve for  the air supply,
 from the  flowmeter to the
 equilibrator.   The constant  pressure
 for the air supply is maintained  by
 bubbling  the dry air  through a
 constant  level  of  fluid,  e.g. water,
 in  the stabilizer.  By raising or
 lowering  the level of the  fluid in
 the  stabilizer, the air pressure
 controlled.   The level  of the fluid
 is  raised by adding fluid to the
 stabilizer,  and lowered by draining
 fluid through the stabilizer stopcock
 located on the bottom  of the
 stabilizer.  Changing the pressure of
 the air supply in this way increases
 or decreases the flow of the test
vapor through the dilution apparatus.
Excess air passing through the
 stabilizer is vented to the
 laboratory hood.   The flowmeter,  or
 manometer,  consists  of an  inner
 glass tube,  which is graduated in
 millimeters,  and  outer glass  tube
 through  which the air flows,  a glass
 capillary  tube of predetermined bore
 size,  a  cover to  seal the  capillary,
 and  a bulb type bottom filled with
 colored  water,  which is connected to
 the  constant pressure regulator.
 The  capillary is  calibrated such
 that the flowrate through  the
 capillary  is known for any water
 height.  Thus,  the flowrate is
 determined by the height of the
 water in the inner tube, and  the
 capillary  calibration data.   The
 flowmeter measures the flow rate  of
 the  dry  air-test  vapor mixture in
 milliliters per minute.  The  flow
 rate of  the diluent  air is
 controlled in the same fashion as
 the  equilibrator  air supply with  a
 larger inside diameter capillary
 tube.  The flowmeter for the  diluent
 air  is measured in liters per
 minute.  The nominal  concentration
 of the test vapor can be calculated
 using the equation
  C = {(f * p)/([F + f]*P)}
(5)
where C is the nominal concentration
of the test vapor in parts-per-
million by volume, f is flow rate of
air through the equilibrator, F is
the flow rate of the diluent air, p
is the vapor pressure of the test
reagent at the temperature of the
experiment, and P is atmospheric
pressure.  Thus, the concentration
                                         116

-------
of the test vapor may be easily
changed by varying either the flow
rate of air through the equilibrator,
or by changing the flow rate of the
diluent air.  In practice, it works
best to change the flow rate of the
diluent air, when possible, because
the efficiency of the vapor
generation in the equilibrator
decreases at higher flow rates.

    The dry air-test vapor mixture
from the equilibrator and the diluent
air are passed into the mixing
chamber located at the entrance of
the reservoir.  The dilute test vapor
is thoroughly mixed by a swirling
circular motion of the air in the
mixing chamber before entering the
reservoir.  The reservoir is the
container for the diluted test vapor,
from which samples are taken for
concentration analysis and for
testing purposes.  There is a
charcoal canister located on the
exhaust of the reservoir.  This
canister serves as a scrubber to
remove test vapors passing from the
reservoir to the atmosphere in the
laboratory hood.
 Pre-Processinq of Spectra for Linear
 Discriminant Analysis

    The pre-processing and data
 processing procedure used in the
 linear discriminant analysis is shown
 in Figure 3.  The first pre-
 processing step is to determine if
 the spectrum has been collected in
 the positive (+) or negative (-)
 mode.  This knowledge is important
 since the Graseby ASP board does not
 differentiate between the two types
 of spectra, i.e. the ASP board
 converts all spectra to positive
 values.  The determination of the
 operating mode under which the
 spectrum was collected is made by
 reading the data file header which
 includes a single character which is
 used to designate mode.  A
 preliminary discrimination is made
 based on the mode; a spectrum
 collected in the negative mode has
 no chemical semblance to a spectrum
 collected in the positive mode.  Once
the mode has been determined, it is
necessary to determine the time at
which the reactant ion peak  (RIP)
appears.  The reactant ion for the
AVM. O2~in the negative mode and
H3CT~ in the positive mode, is the
species which transfers the charge
to the chemical species being
analyzed.  The location of the RIP
must be determined for each
spectrum, if possible, because the
location is affected by changes in
temperature, pressure, and relative
humidity.  If no RIP is found, then
one must assume the RIP is located
at the same time as the RIP  for the
previous spectrum.  After
determining the time at which the
RIP appears, the spectrum is
normalized to create a dimensionless
X-axis.  To do this, each value on
the X-axis was divided by the value
position of the reactant  ion peak.
For negative ion spectra  collected
at, or near, sea level, the  peak
position, with the maximum intensity
between  6.0 and 7.0 milliseconds
drift time was used for the
identification of the reference  ion
peak.  For positive ion data, a
value between 6.5 and 7.5
millisecond drift time was used  as a
window  in which to  find the
reference ion peak.  This reference
window  is easily adjusted for
spectra  collected at other altitudes
or  pressures by multiplying  the
window values by the ratio of the
operating pressure  to atmospheric
pressure at sea level.  This new
spectrum also appears as  a pseudo-
"Reduced Mobility"  spectrum  which
has a dimensionless X-axis
corresponding to a  Ratio  of  Drift
Times, TR.  Only the data in the
range 0.5 to 3.0 along the TR axis
are used.  A cubic  spline is then
applied  to the spectra such  that
every spectrum has  the same  data
spacing  with respect to the  Ratio of
Drift Times axis.   The IMS data
files used  in this  study  have data
points  every 0.005  TR.

LINEAR  DISCRIMINANT ANALYSIS

     Traditionally, much  of  the
effort  associated with the analysis
of  the  IMS  spectra  has been  left to
the chemist.  In an effort to aid  in
the preliminary  identification,  a
                                       117

-------
  personal  computer (PC)  based spectrum
  identification package  has been
  developed.   This  package,  written  in
  Microsoft Fortran, uses a  linear
  discriminant function for  its
  identification, and  consists of three
  separate  programs.   These  programs
  are:  IMSDISC, a program which reads
  selected  data files  from the PC and
  builds a  discrimination data set;
  TRAIN, a  program  which  analyzes the
  discrimination set and  calculates  the
  linear discriminant  function that
  best  isolates the data  of  interest
  from  the  interferant data;  and
  IMSIDENT, a  program  which  reads the
  data  to be analyzed  and identified
  and calculates its linear
  discriminant value.

     Linear discriminant analysis,   one
  of the most  basic forms  of pattern
  recognition  used by  scientists,  is
 used  as a supervised learning
 technique.   In supervised learning
 techniques, the computer learns to
 classify the samples being analyzed
 based on knowledge about the samples;
 in this study, the samples either
 belong to the class of chemicals you
 wish to identify,  or they do not.
 The goal of the learning is to
 develop a classification rule, the
 linear discriminant function, which
 allows the validity of the
 classification to  be tested and
 ultimately to properly classify
 unknowns .

     The linear discriminant function
 has the general  form
               n
  g(x)  = w0 +   X    Wi  xi        (6)
where wo  is  the  threshold vector, wi
is  the weight vector,  X£  is the
response  vector,  and g(x)  is the
response  function.  The discriminant
function, g(x) is determined by
choosing  those variables  xj[ with
characteristics which  differ between
the groups being  classified.  These
variables are then linearly combined
and weighted such that the  groups are
as statistically  different  as
possible.  This linear combination of
variables is calculated using the
perceptron convergence criteria.
      The perceptron [13-15]  is a
pattern recognition procedure which
 consists updating the weight vector
 by considering only those patterns,
 or spectra in this work, which have
 been misclassified in the training
 set.  Each misclassified pattern is
 considered in turn, with a fraction
 of each misclassified spectrum being
 added to the weight vector.  This
 procedure is continued until all of
 the spectra are classified
 correctly, or until it is determined
 that the procedure fails to converge
 to a satisfactory solution.

      In this software package,  the
 three programs are run separately,
 but are still inter-related.   The
 first program,  IMSDISC,  uses a file
 called NAMES.  NAMES is  simply the
 file that contains the names of the
 individual data files to read,  and a
 value that tells the program whether
 the file is to be treated as the
 sample or as an interferant.   The
 data from the individual data files
 is then treated such that all the
 files are compatible with respect to
 time spacing between data points,
 delay to start of data sampling,  and
 number of data  points.   To
 accomplish this,  IMSDISC uses a
 spline function to interpolate  and
 fit the data.   After the data has
 been treated to fill  the
 compatibility requirement,  the
 discriminant threshold is set to
 zero by  multiplying  all interferant
 spectra by negative  1, (-1).  The
 sample spectra  are left  unaltered.
 The data is  then  stored  in a
 discriminant data file.

      The second program,  TRAIN,
 develops a  linear-discriminant  based
 on the perceptron convergence
 criteria.  TRAIN  prompts  the
 operator for the  name of  the  input
 discriminant  file  that was created
 with  the program  IMSDISC.  It reads
 the data from the  discriminant  data
 set,  accepts input for the values of
 a  scaling factor, between
 0.000000001 and 0.1, and the number
 of iterations to perform using this
 scaling  factor.  In practice, it is
 generally necessary to use a series
 of decreasing scaling factors and
 iterations to calculate the linear
discriminant function which best
differentiates the samples and the
interferants.  After the  linear
                                         118

-------
discriminant function has been
calculated,  the linear coefficients
are written to a file on the computer
disk for use by the last program.
These first two programs, IMSDISC and
TRAIN, are the time consuming
programs and are run only when a new
compound is to be added to the
database.

    The third program in this
package, IMSIDENT, uses the linear
discriminant values created with the
program TRAIN.  Thus, it is dependent
on the first two programs in the
package.  IMSIDENT can be used in one
of two possible configurations; the
first configuration is as a stand-
alone program, and the second is that
it can be incorporated into a data
collection program for real time
identification of an unknown
environment.  In the stand-alone
configuration, the program prompts
the operator for the name of the data
to analyze.   The program reads the
data, and performs a spline
interpolation to make the data
compatible with the discriminant data
sets.  Next, the program reads a file
named COEF.FIL that contains the
names of the coefficient files.  The
linear discriminant value is then
calculated.   i'f the linear
discriminant value is positive, an
alarm message is generated which
notifies the operator that the
spectrum has been identified.  No
message is generated if the
discriminant value is negative.  The
results of the identification process
are then written to a file named
ALARM.RPT for later use, and the
program then prepares to read the
next data file to be analyzed.

    In the second configuration, the
program functions as a real time
monitor.  The name of the data file
to be analyzed is passed from the
data collection program to the
IMSIDENT package rather than
prompting the operator for the name
of the data file to analyze.  The
spline interpolation is then
performed on the data, and the linear
discriminant value is calculated.  If
the discriminant value is positive,
the alarm message is generated; no
message is generated if the
discriminant value is negative.  The
  results  of  the  identification
  process  are written  to a  file  named
  ALARM.RPT for later  use.
 DISCUSSION
        The program package was
 developed for use with the Graseby
 Ionics Advanced Signal Processing
 (ASP) board, the Graseby Airborne
 Vapor Monitor  (AVM), and a Zenith
 286 PC.  Using this hardware and the
 linear discrimination package, it
 has been possible to identify and
 semi-quantitate the presence of 15
 common chemical vapors in air.
 These compounds, most of which are
 of industrial importance,  and the
 levels at which the Occupational
 Safety and Health Administration
 (OSHA) have determined them to be
 hazardous are shown in Table C, with
 the ion mobility spectra of these
 compounds shown in Figures 4 through
 21.  When the software is used in
 the stand-alone configuration (i.e.,
 separate from the data collection
 routines)  and using the Zenith 286
 PC, the presence of these compounds
 can be determined and the compound
 identified in less than ten seconds.
 This includes the time necessary to
 perform the  spline interpolation and
 the calculation of the discriminant
 value for  the data; however,  this
 does not include the  time required
 to create  the discriminant
 functions.

      The results  shown  in Table  D
 are from the  evaluation of  a  series
 of files used to determine  the
 presence of N-Methyl  Formamide.  The
 "All Clear" report indicates  that
 the IMSIDENT  program  does not find
 any similarities between  the N-
 methyl formamide test spectrum and
 the spectra of the fifteen compounds
 stored in the database.  The report
 of  an alarm indicates that the
 program did find similarities in the
 spectra, and the magnitude of the
 discriminant is a measure of the
 amount of similarity.

     It is not really surprising
 that there are a number of false
positive alarms indicating the
presence of diethyl ether.  Older
                                        119

-------
versions of the AVM used an acetone
dopant within its detection system,
whereas newer versions of the AVM use
water vapor in the atmosphere as the
dopant.  This dopant in the older
AVM's results in the presence of an
acetone reactant ion.  This reactant
ion is the ionic species which is
responsible for transferring the
ionic charge to the chemical compound
being studied.  All of the spectra
used in the discrimination functions
were recorded using water as the
reactant ion.  Thus, the discriminant
functions have not been trained to
eliminate the possibility of alarming
on a spectrum which has an acetone
reactant ion peak, and an alarm is
reported.  Examination of two
representative spectra for which an
alarm was reported, shows the
similarity of the IMS spectrum for
the diethyl ether, the lower trace
in Figure 22 (ETHER in Table D) and
N-methyl formamide background
spectrum, the upper trace in Figure
22 (\AVM\DATA\nmfoOOOO.ACQ in Table
D).  The location of the reactant
ion peak does not appear at the same
time as does the diethyl ether peak,
however the ba.nd shapes are similar.
If the discriminant function is
trained to ignore the acetone
reactant ion peak, one does not get
an alarm.  Results of identification
procedure with the acetone reactant
ion peak being ignored is shown in
Table E.
                                     TABLE E

                               File "ALARM.RPT" for
                             N-Methyl Formamide Analysis
                          with Acetone Reactant ion Ignored

               ALL CLEAR FOR FILE  \AVM\DATA\nmfoOOOO.ACQ
               ALL CLEAR FOR FILE  \AVM\DATA\nmfo0001.ACQ
               ALL CLEAR FOR FILE  \AVM\DATA\nmfo0002.ACQ
               ALL CLEAR FOR FILE  \AVM\DATA\nmfo0003.ACQ
               ALL CLEAR FOR FILE  \AVM\DATA\nmfo0004.ACQ
               ALL CLEAR FOR FILE  \AVM\DATA\nmfo0005.ACQ
               ALL CLEAR FOR FILE  \AVM\DATA\nmfo0006.ACQ
               ALL CLEAR FOR FILE  \AVM\DATA\nmfo0007.ACQ
               ALL CLEAR FOR FILE  \AVM\DATA\nmfo0008.ACQ
               ALL CLEAR FOR FILE  \AVM\DATA\nmfo0009.ACQ
               ALL CLEAR FOR FILE  \AVM\DATA\nmfo0010.ACQ
               ALL CLEAR FOR FILE  \AVM\DATA\nmfo0011.ACQ
               ALL CLEAR FOR FILE  \AVM\DATA\nmfo0012.ACQ
               ALL CLEAR FOR FILE  \AVM\DATA\nmfo0013.ACQ
               ALL CLEAR FOR FILE  \AVM\DATA\nmfo0014.ACQ
               ALL CLEAR FOR FILE  \AVM\DATA\nmfo0015.ACQ
                                         120

-------
LITERATURE CITED
1.   Threshold Limit Values and
Biological  Exposure Indices for 1989-
90,  American Conference of
Governmental Industrial Hygienists,
Cincinnati,  OH,  (1989) .
2.   NIOSH Pocket Guide to Chemical
Hazards.  U.S.  Department of Health
and Human Resources,  National
Institute of Occupational Safety and
Health, Washington,  D.C., (1985).

3.   Code  of Federal  Regulations, 29
CFR 1910,  Subpart Z  - Toxic and
Hazardous Substances, 1910.1000 - Air
Contaminants.  19 January 1989.

4.   E. W.  McDaniel and E. A. Mason,
The Mobility and Diffusion of Ions in
Gases. Chapt.  2, John Wiley and Sons,
New York,  1973.

5.   H. E.  Revercomb  and E. A. Mason,
Anal. Chem.. 1975, 47. 970.

6.   Plasma Chroroatocrraphv. Carr, T.
W., ed.,  Plenum Press, New York,
1984.

7.   G. A.  Eiceman, A. P. Snyder,  and
D.  A. Blyth, Inter.  J. of Environ.
Anal. Chem.. 1989, 38. 415.
 8.   G.  A.  Eiceman,  M. E. Fleischer,
 and C.  s.  Leasure,  Inter. J. of
 Environ. Anal.  Chem., 1987, 28. 279.

 9.   c.  s.  Leasure,  M. E. Fleischer,
 and G.  A.  Eiceman,  Anal. Chem..
 1986, 58.,  2141.

 10.  J. M.  Preston  and L. Rajadhyax,
 Anal. Chem.,  1988,  60, 31.
11.   D. M.  Davis  and R.  T.  Kroutil,
Anal. Chim. Acta.  1990,  232.  261.

12.   D. M.  Davis  and R.  T.  Kroutil,
in P. Jurs  (Ed.),  Computer-Enhanced
Analytical  Spectroscopy,  Vol.  3,
Plenum Press, New York,  1990,  (in
press).
13. F. Rosenblatt,  Principles  in
Neurodvnamics: Perceptrons  and the
Theory of Brain Mechanisms, Spartan,
New York, (1962).

14. R.O., Duda and  P.E. Hart, Pattern
Classification and Scene  Analysis,
Wiley, New  York,  (1973).

15. Y.-H. Pao, Adaptive Pattern
Recognition and Neural Networks,
Addison-Wessley, New York,  (1989).
                                        121

-------
             CARRIER FLOW C6TB-,   I	1   DRIUINC PULSE
             CfiRRIERFLOWCAIR) _|   |_ TO CRATING GRID
                                I ^-GATING
                             O .    GRID
                                                      DRIFT
                                                     REGION
                                                   o

ft
>
•
1J
* O
9 o o o
* o o







i
1 	 IONIZING



X
I






                            SAMPLE
                            IONS
                                                               OUTPUT SIGNAL
                                                                  AMPLIFIER
                                                                MICROPROCESSOR
                                                                     SVSTEM
                                    REGION
                        TO  PUMPS
  AIR  IONS

1 SAMPLE IONS
            COLLECTOR
            ELECTRODE
Figure 1.  Schematic diagram of an ion mobility spectrometer.
                                        122

-------
                                         CAPILLARY
                                                 >IR SUPPLY CONNECTION
                                                 KHNO FLOWNETEH
NITROSEN SUPPLY
                                                                      SAMPLING  LINE
                                                                      VACUUM SUPPLY
         Figure  2.   Schematic  diagram  of the Q5  vapor generator.
                                             123

-------
             LINEAR DISCRIMINANT
                       ANALYSIS
                          ACQUIRE DATA
                           DETERMINE
                          MODE(+ OR-)
                              J_
                     LOCATE
                   REACTANT ION
                    PEAK (RIP)
                   NORMALIZE
                  SPECTRUM WITH
                  RESPECT ID RIP
                IF NO PEAK
                 PRESENT,
                 ASSUME
              LOCATION FROM
              LAST SPECTRUM
                    COMPARE
                    RESULTANT
                  SPECTRUM WITH
                  THE REFERENCE
                    SPECTRA
           IF DISCRIMINANT
           VALUE < 0 ; ALL
              CLEAR
IF DISCRIMINANT
VALUE > 0 ; SET
  ALARM FOR
 THAT SPECIES
Figure 3.   Block diagram showing the steps taken when performing a linear
discriminant analysis on an ion mobility spectrum.
                              124

-------
 in
•4J
•H
 c
D

 3)
 L
 03
 L
-M
•H
I]
 L

-------
   in
   -M
   •H
   c
   D
   L
   ID
   L
   4J
   •H
   £!
   L
   
-------
    01
    -M
    •-I
    C
    D

    31
    L
    05
    L
   4J
   •H
   U
    L
   
-------
   in
  4->
  •H
   c
  D

   31
   L
   ID
   L
  4-1
  •H
  U
   L
  
   •H
   n
   L
   
-------
  III
 •p
 •H
  c
 D
  L
  ID
  L
 •M
 •H
 a
  L
 
-------
    in
    +j
    •H
    c
    D

    31
    L
    ID
    L
    -P
    •H
    12
    L
    
-------
 in
 -M
 •H
 c
 D

 31
 L
 ID
 L
 4J
 •H
 J]
 L
 
-------
   in
  -u
  •H
   c
  D

   Dl
   L
   ID
   L
  +J
  •H
  SI
   L
  
-------
 in
4J
•H
 c
D

 31
 S.
 IU
 L
4J
•H
12
L

-------
                         Arbitrary  Units
  ro.
   CJ
H- CO
in  •
ID  en
n
o  H>
2  ®
D. •
in  -4
   CO
   (0
   (0

   A)

   ro
   H*
   •
   CO
03
  Figure 22. Typical IMS spectra analyzed using linear discriminant analysis.

  Spectra show the similarities often encountered  in IMS spectra.   Spectrum A

  is diethyl ether, and spectrum B is an acetone reactant  ion  spectrum.
                                      134

-------
                                                           DISCUSSION
DREW SAUTER: Perhaps you could explain certain aspects that have hindered
adoption of ion trap mass spectroscopy, basically ion molecule reactions. One of
the things I've run into, and others have, is that in certain limited scenarios, you
can probably define your ion molecule chemistry.

PETER SNYDER: Yes

DREW SAUTER: But the truth of the matter, and correct me if I'm wrong, is
that you can have unknown reacting ions in the sample. In an unknown situation,
it would seem that you could actually get spectra that were sample dependent.
Basically, would you see IMS being more useful as a sort of screening tool on
relatively limited scenarios, as opposed to a tool that could offer more general
analysis capabilities?

PETERSNYDER: Well, I can't disagree with that when you just talk about IMS
by itself. Because of the potential complicating responses that can occur if your
environment is not controlled, anything can happen.

DREW SAUTER: What I mean though is in the real environmental world, a lot
of samples have a lot more than one compound, and not only that do they have
a lot more than one compound, recognizing that you can separate things by GC's,
they tend to  have different concentrations.

PETER SNYDER: Yes.

DREW SAUTER: Hence if they have different concentrations, and there's ion
molecule reactions going on. you have them going on with some rate constant.
They're producing different populations of ions, and hence a different sample
dependent spectra. That strikes me as a significant drawback, despite all the
grand things that you've shown.

PETER SNYDER: You have to consider what IMS is based on? IMS is based
on ion molecule reactions, and that can be broken down into proton affinity and
electron affinity by and large.  So  then you have  to  look at what kind of
compounds are responding.

DREW SAUTER: But there's al so a concentrat ion term that you showed in your
graph.

PETERSNYDER: Yes, absolutely. Concentration is very important. I guess the
difficulty in response comes then when you get to phosphonate compounds or
phosphoryl compounds that are very sensitive to proton affinity. They get that
proton very nicely, and by and large to the exclusion of many other compounds,
even in their presence, or at relatively high concentrations. Ammonia, probably
would take exception too. That might be a complicating factor.

But in most cases, phosphoryl compounds really come through, and that's one of
the strengths of the chemical agent monitor, in terms of looking for phosphoryl-
oased nerve agents.

As you go down to amines, esters,  ketones and alcohols, the relative proton
affinities are not as wide.

STEVE HARDEN: I'd like to just comment on that before we get on to the next
question, and say that yes, you have indeed hit upon one of the problems with ion
mobility spectrometry for analyzing real-world mixtures.
The reason the Army has developed it for their purposes is that the compounds
they're interested in either have  such  extreme proton affinities, or extreme
electronegativities, and that the sensitivity is very high for those compounds. So
it works for our purposes, and it may not work for some environmental purposes
that you mentioned, because of this mixture problem.

It also points out one of the needs and requirements in this unknown analysis, or
analysis of unknowns, for preparation  of sample you mentioned the GC/MS
system. We'll hear some more about that in our next paper.

But one can also point out that in some of the data (in this paper) for some
compounds that do have a high electronegativity, can be picked out using these
techniques that we were talking about, and we can then point out the fact that yes,
indeed, that material was present.

That little bump on the side of that peak was, I think, the mustard, which is an
Army compound of interest. The  bump was on the side of a peak  of phenol,
phenol being in much greater concentration.

In previous sensitivities and single processing techniques, we can bring it even
more if we used preparation of samples. However, you do separate samples at the
expense of complexity of instrumentation, and that's one reason why the Army
hasn't pursued that to this particular point.  So we have.

HERB HILL: Fora long time now we have been using ion mobility spectrometers
as a  chromatographic detector, basically because we feel that there really are
problems with interferences, except for very specific cases.

I'm  really excited to see us  beginning to talk about the use of, what 1 call
chromatographic filters on the front end of IMS, for field monitoring. We've
done studies, for example, treating IMS as a chromatographic detector, and you
can see that the interferences under conditions like that are no worse than you
would have with a flame ionization detector, an electron capture detector. The
quantitative value of IMS is acceptable in any range. It's as good as any of the
standard chromatographic detectors that we have. We've published papers in
which we've put interfering species in, compared them to an FID, and ECD, an
IMS, and you see that the quantitative value of the data is fine, it's good in  IMS.
When you add the chromatograph controls on the front end, you can do dioxins.
We do ligands in blood analysis,  we do a  variety of very small, minute  trace
compounds in very, very complex mixtures, as well or better, than you can be a
lot of techniques.

And it should apply very well to field analysis for portable, if you put a portable
GC on the front end of that.

PETER SNYDER: Yes, you' re absolutely right. And the literature that you have
published over the past decade and a half, attests to that. There's many different
sample matrices that Professor Hill has looked at with very good  resolution,
depending upon the column characteristics.  There has been a lot of  good
information coming out of that, using an IMS as a detector.

So basically the newer innovative topic we're looking at here, is using the hand-
held version of the IMS, to see how far we can go with that.
                                                                      135

-------
           ION MOBILITY SPECTROMETRY AS A FIELD SCREENING TECHNIQUE
                   Lynn D.  Hoffland and Donald B. Shoff

            Analytical  Research Division,Research Directorate,
     U.S. Army epical Research,Development  and Engineering Center,
                       Aberdeen Proving  Ground,  MD
1.   INTRODUCTION

    Ion  Mobility  Spectrometry  (IMS),
also  called Plasma  Chromatography,  is
used  to  detect   trace quantities  of
organic  va,pors  in  gaseous   mixtures.
Several  researchers over  the  past  15
gears have  demonstrated the utility  of
mobility  detection  for  a  variety  of
organic  compounds. 1~H   Quantities  as
low as
10-10  grams of  nitrosamines have  been
reported^^.
    IMS   is  a   conceptually   simple
technique that relies on the drift time,
or time  of flight,  of  molecular,  or
cluster,  ions  through  a  host  gas  as a
means  of differentiation.  This  differs
from classical mass spectrometry in that
there  is little,   if any,  fragmentation
and the  ions  are not  mass  analyzed.
Detailed    theory   can    be    found
elsewhere.13'15        The    ions    are
differentiated    by   charge   and   by
mobility.    The   reduced   mobility  KQ
(corrected   for  standard  pressure  and
temperature) is  expressed  as

              Kg  = 42.51  D

where   D   is    the  sealer   diffusion
coefficient of Pick's law.  This reduced
mobility KQ is catalogued and identified
for each ionic species present.

2.   EXPERIMENTAL

     The work was  performed on a MMS-290
Ion Mobility  Mass  Spectrometer   (POP,
Inc.).  shown  in  figure  1  and  an Air
Vapor Monitor  made by Graseby Ltd.
     The PCP,  Inc.  MMS-290 spectrometer
used in these experiments consists of an
ion mobility spectrometer  followed by a
quadrapole mass spectrometer coupled  to a
Nicolet signal  averager with a computer
interface for storage,  data manipulation
and display.
     There  are four  modes  of operation
for the -MMS-290.   In the total ion  mode
the  MMS-290  acts  as  an   ion mobility
spectrometer.   Ions  are  gated into  the
drift   region  and  detected   by   the
electrometer.    All  ions  detected   are
averaged, stored  and displayed.    In  the
integral  ion  mode  the  mass  spectrometer
is   the    detector   instead   of   the
electrometer.     Again,  all  ions   are
detected, averaged, stored and displayed.
There  is  no mass analysis  in  this mode.
It   is  used  to   check   that the   ion
distribution  is not changed by traveling
the  extra   distance  through   the  mass
spectrometer.   The  third mode  is the mass
spectrum.  The shutter  grid  is held  open
to allow a  continuous stream of ions into
the  mass   spectrometer  which  is  mass
analyzing  the  ions.   This provides a mass
spectral scan  of  the  total ion flux.   The
last  mode of operation is  the tuned ion
mode  where the MMS-290 is operated as in
 the   integral  ion  mode  but  the  mass
spectrometer is only  detecting one  mass
ion  at a  time.   This shows  which  mass
 ions   are  associated with  each  mobility
peak.

      The  Airborne  Vapor  Monitor  (AVM)
 used in these experiments consists of an
 IMS described above with a membrane inlet
 and   internal  electronics  for  signal
 processing  and alarm.   It  operates in
 both positive and  negative ion mode, has
                                        137

-------
no   internal   display   but   can   be
interfaced  to a  personal  computer for
display and storage of the IMS  spectra.
The AVM has only an  electrometer, it has
no mass spectrometer to mass  analyze the
ions,  and  it  operates as the total ion
mode of the MMS-290.

     Air,  or  the sample  gas,  is drawn
into the ionizing region and is ionized
by 60  keV  Beta  rays from a radioactive
Ni63   source.      A  potential  exists
between  the  ionizer  and  the collector
forcing the ions in  the direction of the
shutter  grid.  The  closed  shutter  grid
neutralizes all  ions  reaching  it.   The
shutter is pulsed open for approximately
0.1  millisecond  (msec)  and   a  cross
section of the ions flow into the drift
region.    The   shutter  closes  again
isolating  a  short  pulse of  ions  that
travel down  the  drift region propelled
against  the  drift  gas  flow  by  the
potential  on  the collector. The   ions
are differentiated  by their  charge  in
the electric field and their mobility in
the drift gas (velocity V*)

                    Vd = K E.

The IMS differentiates the ions because
by the time that they reach the shutter
grid  the  ion  molecule  reactions  have
equilibrated and in the drift region no
more reactions take place.

     As  the  separated  ions reach  the
collector, they  are detected by a  fast
electrometer, and a  current is generated
directly proportional  to  the number  of
ions.      The  resultant   spectrum   is
depicted in figure 2 **>.

     The  highest   KQ   ions   (C+)   are
usually smaller or more compact  followed
by the slower ions B+ and A+, in time.
     Both  positive  and  negative  ion
formation of  reactant  and product ions
are multistep processes.  Good, Durden,
and  Keburle*7    have  determined  the
mechanisms involved in  positive reactant
ion formation:
                        + e-
                            + 2 N
                               OH
                         (H20)2H+
The size of  the  resultant  reactant ion
water clusters depend upon the relative
humidity but  generally  water chemistry
dominates the  positive  ion mode.   The
water ion may cluster directly with the
sample molecule  M or,  as  is  the  case
more   often,   the   sample   molecule
abstracts  the  proton  from   the  water
cluster  and  then  may attract  more  or
less  water  molecules depending  on the
humidity.   At high  concentrations the
sample molecules may form dimers with a
proton.   whenever there is  some other
molecules present  with  a  higher proton
affinity than water they may replace the
water  in  the  above  mechanisms  i.e.
Acetone or NH-,.  So, in figure two peak
C  may  be  the  reactant  ion,  B  the
hydrated monomer,  and A the protonated
dimer.
      Negative reactant ion formation as
 summarized by  Spangler and  Collins-1**
 include the following:
   e~(thermal)  +  0,
         "here n = 1,2.
The sample molecule can cluster with the
O,- or  abstract the O2   from  the CO2-
As can be easily seen,  the chemistry can
be  quite  involved  before  any  products
are formed.
                                           138

-------
      The operating  parameters  for the
 MMS-290 were;
         Cell length

         Operating voltage

         Electric field
         Carrier gas

         Drift gas

         Cell Temperature

         Pressure

         Drift distance
            15 cm

            3000 volts

            200 volts/cm
            200 ml/min

            500 ml/min

            40 °C

             Entered  Daily

            10 cm
     The AVM  was operated  as  received
from Grase&y  Analytical,  Ltd. (Watford,
Herts,  UK) .   Signals from this IMS were
processed  with  a  Graseby  Analytical,
Ltd.,  advanced  signal averaging  (ASP)
board   installed   in  an   IBM   PC/AT
computer. Known or approximate operating
conditions were;

                   inlet  flow

                   Drift  tube temperature

                   membrane temperature

                   reaction region

                   drift  region

                   field  gradient


    The samples  were generated  using a
0-5 apparatus,  (where a saturated  vapor
stream  is  mixed  with  a  high   volume
diluent dry air stream).  By varying the
quantities    of   both   streams    the
concentration  of  sample in  the diluted
vapor  stream   was  controlled.      The
resulting  diluted  vapor   stream   was
sampled  by  either  the IMS /MS inlet  or
the AVM.   All  samples  were used  as
received  from  the manufacturer.    The
concentration  of  the  saturated   vapor
             500  ml/min.

              ambient

             70 °C

             2.2  cm

             "3.8 cm

             ~ 200 V/cm
stream was calculated from vapor pressure
data or from the Antoine  equation.

3. RESULTS AND DISCUSSION

     The data following are an  example of
the  power of  this detection  system  to
high concentration vapors of acetic acid.
The acetic acid was used "as is",  and, as
will  be  shown,   was   contaminated  with
acetic anhydride(as  is often  the  case).
The target concentration  for acetic acid
                                         139

-------
detection  with  the  AVM  was  the  Time
Weighted Average (TWA) of 10 ppmiy, the
Short Term  Exposure  Limit (STEL)  of 15
ppmi9, up  to  the Immediately Dangerous
to Life  or Health  (IDLH)  level of 1000
ppm20.  Figures 3-6 show the  response of
the AVM  for these  three concentrations.

     The  identity  of the  peaks  in the
above  data  was   determined with  the
IMS/MS in  the  following manner.  First,
the reduced mobility is calculated for
each peak. Since the  reduced mobility is
a factor of pressure  and temperature and
these  vary in  the AVM  and  between the
AVM and  IMS/MS, a drift  time ratio is
calculated   by  dividing   the  specie
mobility by the  reactant ion mobility
 (both are under the same temperature and
pressure).  Then,  the IMS/MS is operated
in  the  total  ion mode and  the integral
ion  mode  to   check   that  there  is  no
effect  between the different inlets of
 the AVM  and the IMS/MS and that the mass
spectrometer entrance of the IMS/MS does
not change  the specie (figures 7  and 8) .
The  first  thing  noticed  is  that the
pinhole  inlet  of the  IMS/MS  is much more
 sensitive than the membrane of the AVM.
 The  membrane  is  required,  however,  to
 keep  too  much  water and  contaminants
 from  spoiling  the sensitive IMS  cell.
 So,   allowing  for  the  difference   in
 sensitivity,   the  mobility  spectrum  of
 the IMS/MS is  compared with the AVM  to
 correlate the  mobility peaks between the
 two instruments.   Once  confirmed,  the
 mass  spectrum  is taken to determine what
 mass  species  are the major contributors
 to the ion  mobility spectrum (figure 9).
 Then, each mass is scanned  in the tuned
 ion mode  to  determine  to  what  peak  in
 the   mobility   spectrum   each   mass
 contributes  (figure 10).    As  can  be
 seen,  in  this  low   concentration,  the
 masses  55,73,83,101,and  129  are  all
 hydrates   and   "nydrates"   of  the   H
 "reactant  ion" and   the  masses  79,  97,
 125,  and 153 are hydrates and "nydrates"
 of  the  H+  acetic acid  monomer.    The
 concentration  is  then increased and the
 analysis  series  is  repeated.   As  the
 concentration   increases    the   mass
 spectrum  becomes  more complicated  but
assignments can be  made  bases  upon past
experience.  Since, at this  time,  we do
not  have  the   capability  there  is  no
secondary    mass    fragmentation    for
confirmation of these species.   Tables 1
indicates  the assignments  for  each mass
fragment in the mass  spectrum.   Table 2
is a list of the mobility ratios and the
assignments for each  mobility  peak seen
at the various concentrations.

CONCLUSION

This example of acetic acid  illustrates
the  potential   of  this  hand  held  ion
mobility  spectrometer to  differentiate
between   regulated   concentrations   of
hazardous  chemicals.     In  support  of
another program this work has been
extended   to  identification   of  these
regulated concentrations (TWA,  STEL, and
IDLH)  of  15  other  solvent  chemicals.
Although limited  in scope,  by  extending
this data base  the AVM could be used as a
field screening device   and  as a safety
device for field personnel.
                                          140

-------
                              TABLE 1
AMU

55
73

79

83

97

101

125

129

153
Specie

H+(H20) 3
H+(H20)4

m H+(H2O)

H+(H2O)3+^2

m H+(H2O)2

H+(H20)4+N2

m H+(H2O)2+N2

H+(H20)4+2N2

m H+(H2O)2+2N2
Comment

reactant ion
reactant ion

monomer hydrate

reactant ion

monomer hydrate

reactant ion

monomer "nydrate"

reactant ion

monomer "nydrate"
Mobility Ratio

1.00

1.08-1.09

1.18-1.24

1.34-1.35

1.47-1.48
              TABLE 2


          Assignment

          fi+(H20)x(N2)y

          m »+(H20)x(N2)y

          m2 H+(H20)X(N2) y

          m n H+(H20)x(N2)y

          n2 H+(H20)x(N2)y
     Reactant Ion

     Acid Monomer

    Acid Dimer

    Acid Anhydride

    Anhydride Dimer
                                  141

-------
References

1.   Cohen, M.  J.,  and  Karasek,F.  W, ,
"Plasma chroma tography - A new dimension
for   gas    chroma tography   and   mass
spectrometry",  J.  Chromatogr.  Sci.  £,
(1970), 331.

2.   Karasek,F.   W.,   and  Kane,D.  M. ,
"Plasma  chroma tography  of  the  n-alfcyl
alcohols",   J.   Chromatogr.   Sci.,  10,
(1972), 673.

3.   Karasek,F.  W. ,  Tatone,0.  S.,  and
Denney,D.  W. ,  "Plasma chromatography of
the n-alfcyl  halides", J. Chromatogr. 87,
(1973), 137.

4.   Karasek,F.  IV.,  Tatone,O.  S. ,  and
Kane,D. M. ,  "Study of  electron capture
behavior  of  substituted  aromatics  by
plasma chromatography",  Anal.  Chem, 45,
(1973), 1210.

5.   Karasek,F.     W . ,    "Plasma
chromatography", Anal. Chem. 46 ,  (1974),
710A and references.

6.   Karasek,F.  IV.,  Denney,D.  W. ,  and
Dedecker,B. H. , "Plasma chromatography of
normal alkanes and its  relationship to
chemical ionization  mass spectrometry",
Anal. Chem,  46_,  (1974),  970.

7.   Karasek,F.  W.f  and  Denney,D.  W. ,
"Detection   of  aliphatic  N-nitrosamine
compounds   by   plasma   chromatography",
Anal. Chem., 46,  (1974), 1312.

8.   Karasek,F.   W.,    Malcan,A.,   and
Tatone,O. S.,  "Plasma chromatography of
n-alkyl acetates",  J. Chromatogr., 110,
        295.
9.   Karasek,F.   tf.,   Kim,S.   H. ,  and
Rokushika,S. ,  "Plasma  chromatography of
alfcyl amines", Anal.  Chem., 50,  (1978),
2013.

10.  Spangler,G.  E. ,  and Lawless,?. A.,
"lonization of nitrotoluene compounds in
negative   ion   plasma  chromatography",
Anal. Chem., 50,  (1978), 884.
 11.   Shumate,C.,  St.  Louis,R.  H. ,
 Hill,  Jr.,H.  H., "Table of  reduced
 mobility   values    from    ambient
 pressure ion mobility spectrometry",
 J.  Chromotogr.,  373,  (1986), 141.

 12.   Karasek,   F.   W.  and  Denney,
 D.IV.,   "Detection  of Aliphatic  N-
 Nitrosamine  Compounds  by   Plasma
 Chromatography",  Anal.  Chem.46,No.
 9,  (August 1974),  1214-1312.

 13.   McDaniel,E.  W. ,  and  Mason,E.
 A.,  The Mobility  and  Diffusion of
 Ions in Gases,  John Wiley and Sons,
 New York,  (1973) .

 14.   McDaniel,E.    W.,    Cermak,V.,
 Dalgarno,A.,   Ferguson,E.   E.,  and
 Friedman,!.., Ion-Molecule Reactions,
 John  Wiley  and  Sons,   New  York
 (1970).

 15.   Loeb,L.  B.,  Basic Processes of
 Gaseous Electronics  (2nd  edition),
 UniversityofCalifornia   Press,
 Berkeley (1960).

 16.   Spangler,G.  E., and  Cohen,M.
 J.,     "Instrument     Design    and
 Description"in   p.    15,    Plasma
 Chromatography, Ed. Timothy W. Carr,
 Plenum Press,  New York, 1984,1-42.

 17.   Good,A.  I., Durden,D.  A.,  and
 Kebarle,P., "Ion-molecule reactions
 in   pure   nitrogen  and   nitrogen
 containing traces of water at total
 pressures   0.5-4  torr.  Kinetics of
 clustering   reactions    forming
 H+(H20) ",  J.  Chem.   Phys.,  52,
 (1970), 212.

 18.   Spangler,G.  E.,  and Collins,C.
 I.,  "Reactant Ions  in  Negative Ion
 Plasma Chromatography", Anal.  Chem.
 43,  (March 1975),  2.

 19.   Threshold  Limit   Values  and
 Biological  Exposure  Indices  for
 1988-1989,AmericanConferenceof
 Governmental  Industrial Hygienists

20.   NIOSH  Pocket  Guide   to   Chemical
Hazards, U.S.  Department' of  Health  and
Human Services, 1985.
                                           142

-------
                  FIGURE 1
             IMS/MS
ATMOSPHERIC PRESSURE (760 TORR)
      MOBILITY REGION FOR 0 NO. 1
            lOc

-------
      TYPICAL [ION ARRIVAL TIME SPECTRUM]
T 2x10-«A
                  MILLISECONDS
                 FIGURE 2
                             144

-------
 in
4J
•H
 c


 31
 L
 01
 S-
4J
•H
n


-------
                         Figure 4: AVM Spectrum (Acetic Acid 10 pom)
 in
4-»
•H
 c


 DJ
 L
 to

4-«
•H
J3
 L

-------
                              Figure 5:AVM Spectrum  (Acetic Acid  15 ppm)
m H+(H2O)x(N2)y   Acid Monomer
    in
    4J
    •H
    D
    31
    ID
    -4J
    •H
    JQ
    
-------
                          Figure €:AVM Spectrum (  Acetic flcic?  I
                                                        000 ppm)
 m
•n-m-(H20)x(N2)y  Acid Anhydride
 01
4J
•H
 c
                              \
D m2 H+(H20)x(N2)y   Acid Dimer
 3)
 L
 01
 L

•H   H
JQ
 L.

-------
     c:
     a
    UJ
    c_>
     -
        •t.13   IO.3S,
                           ST)  22. 6B  2B.81   3-t . 3S  •* 1  12
                           Tine  tnsi
Figure 7:IMS/MS Spectrum "Total Ion Mode"  ( Acetic  Acid 80 ppb)
                                    149

-------
 G
 c?
  or

  r~


r- —
r—  .
'Of-
UJ
                           1/.M
             10.35  16.SO  22.BS  28.61   3-j . 36  VI.12
                        TIME IMSI
      Figure  8:IMS/MS Spectrum "Integral  Ion Mode"
                (Acetic Acid  80 ppb)
                                150

-------
                                                   p

                                               H3C-C-OH
 a
 a
cu
r-
•2.
ru
        97
                     73
                55
                     79 83
                           101
                                  125
                               i
                                  129
                                        153
     10.0   'jQ.D
7O.D   1DO.D  13O.O  1 GO.0  13O.O
   ness
     Figure 9:IMS/MS Spectrum  "mass spectrum mode"

              (Acetic Acid  80  ppb)
                                151

-------
                              m/e 153
                               m/e 125
                              m/e 97
                              m/e 79
                               Total Ion
           ID.35  IB.SO  22.BS  26.Bl  3*1.36  'r 1  12
                      TIME  (MSI
Figure 10:IMS/MS Spectra  "Tuned  Ion  Mode" (Acetic Acid 80 ppb)
                                152

-------
 HAND-HELD GC-ION MOBILITY SPECTROMETRY  FOR ON-SITE ANALYSIS
   OP COMPLEX ORGANIC MIXTURES  IN AIR OR VAPORS OVER WASTE
                            SITES
Suzanne Ehart Bell
Los Alamos National
Laboratory
MS K484
Los Alamos, NM 87545
G.A. Eiceman
New Mexico State University
Department of Chemistry
Box 30001, Dept. 3C
Las Cruces, NM 88003
         ABSTRACT

     Ion mobility
spectrometry (IMS)  was
formally introduced
approximately 21 years
ago, and has been used as
a detector for chemical
warfare agents.  IMS
research and development
outside the military has
recently been the subject
of renewed interest.
Military IMS units are
small, rugged,  and
portable which makes them
ideal candidates for
inclusion in portable
airborne vapor monitoring
systems. The strengths of
IMS are low detection
limits, a wide range of
application, and
simplicity of design and
operation.   The gentle
ionization processes used
in IMS impart a measure of
selectivity to its
response.  However,
atmospheric pressure
chemical ionization with
compounds of comparable
proton affinities leads to
mobility spectra for which
interpretive and
predictive models do not
exist.  An alternative
approach for the analysis
of complex mixtures with
IMS is the use of a
separation device such as
a gas chromatograph (GC)
as an inlet.  The
attractions of GC-IMS over
GC-mass spectrometry (MS)
for field use include the
small size, low weight,
and low power demands of
GC-IMS.
     Parameters in GC-IMS
which required examination
before further development
or field application
included three major
concerns.  The first was
selection of an optimum
temperature of the IMS
detector and evaluation of
the effect of IMS
temperature on mobility
spectra.  The second was a
study of the  stability
and reproducibility of
chromatographic retention
and mobility behavior.
The final issue was the
                              153

-------
development of suitable
data reduction methods.
Results suggest that an
IMS cell temperature of
ca. 150° to 175°C provided
mobility spectra with
suitable spectral detail
without the complications
of ion-molecule clusters
or fragmentation.  A
commercially available,
portable IMS unit was
configured as a GC
detector to evaluate the
possibility of using the
unmodified unit as the
basis for a portable
prototype.  Significant
fluctuation in peak
heights were observed  (ca.
+/- 12%), but mobilities
varied slightly  ( ca. 1 %)
over a 30 day test period.
Neural network pattern
identification techniques
were applied to data
obtained at room
temperature and at 150°C.
Results showed that
spectral variability
within compound classes
was insufficient to
distinguish related
compounds when mobility
data was obtained using
the commercial room
temperature IMS cell.
Similar but less severe
difficulty was encountered
using the 150°C data.
Incorporation of retention
indices as a referee
parameter was useful in
eliminating false
positives.

        INTRODUCTION

Background
     The detection of
trace levels of hazardous
organic volatile compounds
in complex mixtures
represents an analytical
and sampling challenge.
Waste site sampling
requires ppb detection
limits in samples
comprised of complex
matrices and mixtures of
from ten to hundreds of
analytes.  Other
considerations include the
time of sampling and time
of analyses, delays in
analysis, labor costs,
labor training, and
cost/sample ratio.  The
time and expense of
complete laboratory
analyses can force that
fewer samples be taken
with the attending risks.
Technical aspects make the
translation of widely
accepted laboratory
instrumentation (GC-MS and
GC-FTIR) difficult or
unsatisfactory due to cost
and complexity.
Certainly, gas
chromatography with some
advanced detector will be
required for chemical
resolution of complex
mixtures of organic
compounds over waste
sites.  Proven detectors
such as mass spectrometry
and infrared spectrometry
allow necessary
specificity of detection
but represent cumbersome
and intricate
instrumentation not easily
configured for field use.
These instruments often
require highly skilled
operators as well.  The
high power consumption of
portable GC/MS and GC/IR
systems certainly limits
their use in many field
situations.  Other
detectors which have been
                               154

-------
common to portable GC
units lack specificity and
necessitate a reversion to
dual column or dual
detector methods for
confirmation of peak
assignments.   The
development of a hand-held
GC-IMS combines the
separation power of GC in
combination with a
multidimensional detector.
The release of the
civilian counterpart of
the military IMS units was
a logical starting point
for development of a
portable GC-IMS.

Ion Mobility Spectrometry
     Ion mobility
spectrometry (Figure 1) is
based on the ionization of
vapors in air at
atmospheric pressure. The
differentiation of ions
occurs by measurement of
gaseous ionic mobilities
(1).   A typical IMS
instrument is divided into
two regions.   The first is
the reaction region
containing an ion source
(typically 63Ni).  Ion
separation occurs in the
second (drift)  region of
the spectrometer, where
separation is based on the
size-to-charge ratio of
the ions.  The ion shutter
that separates the two
regions injects ions from
the reaction to the drift
region using period pulses
of the shutter field.  The
drifting ions are detected
at the end of the drift
tube by a detector plate.
     In IMS,  ionization
occurs through collisional
charge transfer between a
reservoir of charge, i.e.
the reactant ions, and
neutral analytes, M.  The
most abundant reactant
ions generated from a
beta-emitting source in
air are (H2O)n*H+ and
(H2°)n*°2~-  These ionic
clusters co-exist at near
thermal energies in the
reaction region.  Product
ions experience little or
no fragmentation and exist
commonly as M+ and MH+ or
M~ and M*C>2~ depending on
proton or electron
affinities of the neutral
species.  Ions formed in
the reaction region are
injected into the drift
region by the ion shutter.
In the drift region, ions
move at particular drift
times (td) through an
electric field, E, of ca.
200 V/cm.   For a drift
region with a given
length, L (cm) , the drift
time is related to
velocity  (vd, cm/s) and
ion mobility  (K, cm /V*s) )
through equations 1 and 2 :
d =
     (1)
      / td
     (2)
= vd /E
                         K
Ions strike a flat plate
detector and a mobility
spectrum or plot of
detector current (in pA or
nA) versus td (usually in
ms) is produced.
Consequently, the basis
for selectivity in IMS is
differences in drift times
for ions governed by ion
mobilities.  Drift times
are dependent on
temperature and pressure
and are normalized to
reduced mobility
constants, K0, that are
related to molecular
                             155

-------
properties through the
Mason-Shamp equation.  In
general, the equations for
mobility constants are
considered well-
established for small
spherical ions but
extrapolations to large
organic molecules may be
tenuous.  Practically
speaking, direct
quantitative predictions
of Ko values for organic
molecules are presently
impossible.  Mobilities
are inversely proportional
to collisional cross
sections.  Thus, IMS is an
ion separator based on
size/charge rather than
mass/charge as found in
mass spectrometers.
     Ion mobility
spectrometry offers
advantages such as low
power, simple and rugged
construction, ppb
detection limits, and
mobility spectra
representative of
individual constituents.
Disadvantages
traditionally ascribed to
IMS include significant
memory effects,
irreproducible behavior
and complex response to
mixtures (2).  These
difficulties can be
circumvented with the
addition of a GC as an
inlet and with the
reconfiguration of the
drift tube (3,4).
Furthermore, hand-held IMS
instruments are currently
available in military-
hardened form with battery
operation (5).  The
military IMS cells are
attractive for use in
portable GC units and were
used as a starting point
for the study of GC/IMS
parameters.

Objectives
     Several areas of GC-
IMS have not been
addressed and must be
understood for practical
advances in field
applications of GC/IMS.
The first area is
optimization (or
influence) of IMS
temperature on GC/IMS
performance and on the
mobility spectra obtained
from the IMS.  Second is
the  evaluation of the
effect of concentration on
reduced mobility and
mobility patterns.  Third
is the evaluation of a
commercially available
portable IMS as a GC
detector, and the final
area is the preparation of
a suitable software peak
identification program.
Each of these has served
as the basis for an
objective in the work
described below.
  RESULTS AND DISCUSSION

Effects of Temperature on
Ion Mobility

     The successful
development of a portable
GC-IMS requires that the
optimum IMS temperature be
determined.  This data had
to be determined
empirically, since little
foundational theory was
available.  Typically, low
temperature mobility
behavior shows
considerable ion
clustering and complexity,
while higher temperatures
encourage ion
                               156

-------
fragmentations.  An
intensive study was
undertaken to determine
the optimum operating
temperature for the IMS
since a wide variety of
analytes are expected to
be encountered.  A
representative set of 43
compounds was selected
from seven different
chemical classes, shown in
Table 1.  The temperature
effect study was conducted
on a Tandem Ion Mobility
Spectrometer (TIMS, PCP
Inc.,  West Palm Beach,
Fla.)  which allowed
heating of the inlet and
drift tube.
Confirmational mass
spectral studies were
conducted on an MMS-160
IMS/MS (PCP, Inc., West
Palm Beach, Fla).
     There are four basic
processes that can occur
when a compound is
introduced into the IMS.
First, there may be no
detectable reaction, such
as when a species that is
active only under positive
polarity is introduced
into an IMS operating in
negative polarity.
Second, clusters may form
between the analyte and
various ions such as
N2+, or NH4+.  Such
clusters appear as peaks
in the spectrum.  The
third possibility is the
formation of cluster ions
which subsequently undergo
equilibria reactions while
in the drift tube.  The
magnitude of the
equilibrium constant will
determine the effect on
the resulting mobility
spectrum.  If the
equilibrium is slow
relative to transit time,
no significant effects
will be seen.  If the
equilibrium is fast
relative to the transit
time, the ions arriving at
the detector can differ
significantly from the
original ions produced,
and peak broadening may
result.  Finally,
fragmentation may occur,
and the resulting spectra
may exhibit such behaviors
as a generalized increase
in the baseline or a
series or numerous small
peaks.  The exact
manifestation will depend
on the degree of
fragmentation.  The IMS
portion of a portable
GC/IMS should operate
isothermally to reduce
power consumption and
complexity.  It is thus
essential to select the
cell temperature such that
clearly resolved, sharp,
and reproducible peaks are
produced.  Peak broadening
and fragmentation patterns
will be difficult, if not
impossible, for a data
reduction system to
classify.  It is also
desirable that the cell
operating temperature be
as low as possible to
minimize power
requirements.  The other
factor that must be
considered for temperature
selection is memory
effect.  Higher
temperatures encourage
rapid clearing of the cell
and promote cleaner
operation.  Thus, 3
factors must be balanced
in selecting the optimum
IMS temperature: clearing
time, mobility behavior,
and power requirements.
                              157

-------
     The effect of IMS
cell temperature on
mobility behavior was
studied by analyzing the
43 target compounds using
nine different cell
temperatures from 50 to
250°C.  The results showed
that while all compounds
behaved differently, a
general pattern was
discernable.  At the lower
temperatures (ca. 50 to
150°C), many compounds
experienced drift tube
reactions, and peaks were
either very broad or moved
as the concentration in
the drift tube changed.
At the midrange
temperatures (ca. 100-
200°C), drift tube
equilibria decreased, and
stable ion/molecule
clusters were observed.
At the higher temperatures
(ca. 200-250°C),
fragmentation became
prevalent.  Figures 2 and
3 show two examples of
compound classes and their
behavior over the
temperature range studied.
The aromatics  (figure 3)
are not dramatically
affected by temperature
changes, although benzene
and ethylbenzene do show
evidence of drift tube
reactions at 75 through
150°C.  The alcohols
(figure 4) show greater
variability with
temperature than the
aromatics, but the general
pattern of drift tube
reactions-clustering-
fragmentation  is evident
in the ethanol and n-
propanol.
      Members of  the
chemical  classes of
ketones,  alcohols,
halocarbons, and esters
were examined by IMS/MS at
three temperatures to
confirm the data obtained
using the TIMS.  At 50°C,
ion cluster formation
dominated mobility spectra
and the formation of dimer
and solvated ions was
evident.  At elevated
temperatures (150° and
225°C), these ions were
not observed or present at
low levels.  At 225°C,
fragmentation was
prevalent rendering
mobility spectra less
informative than those
from lower temperatures.
     Compilation of the
TIMS and IMS/MS data leads
to several observations
cogent to the design of a
hand-held GC/IMS.  First,
a portable GC/IMS will
require the use of a
heated IMS cell to obtain
distinctive and
informative mobility
spectra.  If the
instrument is to be used
as a monitor for a wide
range of compounds, the
optimum temperature range
appears to be  150-200°C.
Second, the cell
temperature can be set to
optimize the response of
selected compound classes.
For example, the
halocarbons showed greater
spectral detail at higher
temperatures than did the
rest  of the target
compounds.  If the GC/IMS
is to be used  as an  in-
situ  monitor for
halocarbons, the IMS  cell
temperature could be  set
at 225°C.  Finally,  the
variations in  behaviors
with  temperature might be
useful  as  an added
discriminator  in GC/IMS
applications.  For
                               158

-------
example, acetone and
isopropanol have similar
chromatographic retention
indices on many GC
columns.  At lower IMS
cell temperatures,
isopropanol and acetone
both exhibit drift tube
equilibrium reactions, and
their spectra have many
similar features that
might confuse pattern
recognition software.  At
175°, the spectrum of
isopropanol begins to show
distinct stable peaks,
while acetone still shows
drift tube reactions up to
ca. 225°.  Thus, the
selection of cell
temperature could be used
to help discriminate
between these two
compounds.
Stability and
Reproducibility of IMS

     Graseby Analytical
(United Kingdom),
manufactures a portable
IMS that is used by
western military
establishments for
detection of chemical
warfare agents.  This IMS
(abbreviated as AVM for
airborne vapor monitor)
was coupled to a GC to
evaluate three parameters.
The GC used was a Hewlett-
Packard (Palo Alto, CA)
5730 equipped with a
Supelco (Supelco Park, PA)
SPB-5 30 meter  capillary
column.  Nitrogen was used
as the carrier gas, and
makeup gas was air.  The
AVM operated in a water
chemistry mode.  The
effect of concentration on
mobility behavior was
examined first to
determine if IMS mobility
patterns were
significantly influenced
by analyte concentration.
The stability and
reproducibility of the IMS
response over an extended
period was evaluated as
well.  These findings were
then used to determine if
it would be practical to
use an essentially
unaltered AVM as the IMS
cell for a portable
prototype GC-IMS.  These
findings were also used to
isolate and identify those
features of the AVM that
could be modified to
improve its performance as
a GC detector.
        '  The effect of
concentration on mobility
was studied, by injecting a
series of dilutions of
each of the target
compounds into the GC-AVM.
Review of the data
obtained led to several
unanticipated findings.
First, the AVM spectra of
many of the positive mode
compounds were very
similar.  The data
obtained at 50°C using the
TIMS did not show these
similarities.  As the
concentration of the
target analyte decreased,
the similarities between
the spectra generally
increased.  Product ions
were often shoulders off
the reactant ion peak as
opposed to the separate
product peaks usually
observed using the TIMS.
Finally, a clear linear
relationship between peak
height and concentration
was not obtained over the
concentration range
studied.  As a result, no
definitive statement
                              159

-------
regarding the effect of
concentration on mobility
was possible.
     The reproducibility
of AVM was evaluated over
a 1 month period.  Peak
heights, drift times, and
mobilities were monitored
for positive and negative
background spectra.  The
spectra of known amounts
of positive and negative
mode standard compounds
(ethylbenzene and CCl^,
respectively) were also
examined.  The results of
the study are shown in
Table 2.  The variability
of intensity of the
reactant and product ions
showed drift over the 30
days, but reduced
mobilities varied
slightly.   Any attempt at
quantitation using only
mobility spectra patterns
and relative abundances
would be difficult using
the AVM as configured.
Table 2 also shows that
the larger ions exhibit
more reproducible
behavior, as shown by the
decrease in relative
standard deviations with
decreases in mobility.
This fact was exploited in
neural network pattern
identification studies
which followed.

Evaluation of Neural
Networks for
Identification of
Compounds

     Neural  networks have
in the  last  10 years
become very  popular  for
pattern  recognition  in
many disciplines.  A
network consists of  a
series  of interconnected
nodes  (called neurons  or
perceptrons) in which
mathematical weighting,
summation, and submission
to a function are
performed.  The output of
each neuron is-then sent
on to another neuron where
a similar operation takes
place.  The network itself
can consist of a variable
number of neurons in a
layer, and variable
numbers of layers.  The
network is trained by
submitting to it target
vectors consisting of
input and the target
output desired.  In this
work, the factors included
in the training vector
were retention index and
mobility peak data.  The
target output was the name
of the compound possessing
these GC-IMS
characteristics.  The
network takes each
training vector and
adjusts the weights
applied in  each neuron to
get the correct value
output. The next training
vector is submitted using
the previously obtained
weighting factors, and the
resultant error is used to
adjust the  weights again.
This  repetitive process
continues until the
weights are adjusted  so
each  training vector
submitted to the network
yields the  correct output.
Training  sets may consist
of hundreds of  facts,  and
the training process
itself may  take hours.
Once  the  network  is
trained,  however, response
is rapid.   For  this
reason, neural  networks
are   well suited  for  use
in a  portable  instrument.
                               160

-------
       For this study,
neural networks were used
with both the TIMS data
(150°C) and the AVM data.
The training vectors
consisted of retention
indices, reduced
mobilities, and in some
cases, the percent
relative abundance of the
mobility peaks.  Aspects
of network structure,
training, and failures
were examined with both
data sets.  The network
was unable to train on the
AVM data for the alcohols.
Many of the alcohol
spectra were very similar,
and the network was unable
to distinguish between
them even with the
retention index included.
The network was able to
train successfully using
the TIMS alcohol data.
The difficulty with the
AVM data may arise from
operating the cell at
ambient temperature and
from using a membrane in
the inlet.
     A network was trained
using data from all the
positive mode compounds
obtained at 150°C.
Approximately 10% of the
initial test data was set
aside as a test set.  The
network was trained using
the remaining 90% of the
original data set.  The
trained network was able
to identify ca. 95% of the
test set. Failures were
associated with similar
compounds, i.e., within
compound classes.  A
typical problem was
differentiating
ethylbenzene from the
xylenes.  This problem was
successfully addressed by
using the retention index
of the test compound to
determine the correct
identification.  For
example, if the network
yielded both ethylbenzene
and o-xylene as potential
identifications, the
retention index of the
test compound was compared
to the retention index of
the standard target
compounds.  In all cases
of multiple
identifications, this
approach eliminated the
false positives.  In no
instances were false
identifications seen
across compound classes,
i.e., never was a ketone
mistakenly identified as
an alcohol when the
retention index cr iteria
was used.
        CONCLUSIONS

     The findings
demonstrate that GC-IMS is
a viable field monitoring
technique, and holds
promise of evolving into a
genuinely portable and
powerful field screening
device.  Elevated
temperature cells,
operating without
membranes, will be
required for such devices.
Commercial portable IMS
units such as the AVM
cannot, as currently
configured, be used as
detectors for GC-IMS.
While these devices work
well for specialized
applications, use of the
AVM as a generalized
detector is not possible
without modifications.
Neural networks can be
successfully used to
identify compounds when
                              161

-------
chromatographic data is
included in the training
process and mobility data
obtained at elevated
temperatures is used.
When the pattern
recognition process fails
to identify a compound,
retention index can be
used to obtain the correct
identification.  Neural
networks are system
specific.  The network can
not be trained using data
obtained on different GC-
IMS system.  Aspects of
the chromatographic and
mobility behavior  (via
temperature) can be
modified to suit specific
applications or can be set
to cover a broad range of
target compounds.  The
small size and low power
requirements of GC-IMS
combined with the  ability
to tune the instruments to
different applications
gives GC-IMS an advantage
over many other portable
techniques.
        REFERENCES
1. G.A. Eiceman, Critical
Reviews in Analytical
Chemistry 1990, in press.
2. M.M. Metro and R.A.
Keller, J. Chrom. Sci.
1973, 11, 520.
3. H.H. Hill, Jr.,
Critical Reviews in
Analytical Chemistry 1990,
21,
4. C.S. Leasure, V.J.
Vandiver, G. Rico, and
G.A. Eiceman, Analytica
Chimica Acta 1985, 175,
135.
5. D.A. Blyth,  "A Vapour
Monitor for Detection and
Contamination Control",
Proc.  Internl.  Symp.
Protection Against
Chemical Warfare Agents,
Stockholm, Sweden June 17-
19,  1983, pp.  65-69.  b)
Commercial brochures  from
Graseby Ionics,  Ltd.  and
Graseby Analytical, Ltd.,
Watford, Herts., UK.
                                                  ACKNOWLEDGEMENTS

                                                  Financial support to
                                             NMSU by KRUG Life Sciences
                                             for NASA through project
                                             no. 50,016 is gratefully
                                             acknowledged as is
                                             financial and professional
                                             assistance from Los Alamos
                                             National Laboratory to
                                             Suzanne Bell.
                               162

-------
Table I. Listing of analytes studied using GC-IMS.
                    Positive Mode
ALCOHOLS
     Methanol
     Ethanol
     n-Propanol
     i-Propanol
     n-Butanol
     i-Butanol
     s-Butanol
     t-Butanol
AROMATICS
     Benzene
     Toluene
     Ethylbenzene
     o-Xylene
     m-Xylene
     p-Xylene
     Styrene
ESTERS
     Methyl Methanoate
     Methyl Ethanoate
     Methyl Propanoate
     Methyl Butanoate
     Methyl Pentanoate
     Ethyl Methanoate
     Ethyl Ethanoate
KETONES
   Acetone
   2-Butanone
   3-Metyl-2-Butanone
   2--Pentanone
   3-?entanone
ALDEHYDES
    Propanal
    Butanal
    3-MethyIbutana1
    Pentanal
    Hexanal
                    Negative Mode
HALOCARBONS
     Methylene Chloride
     Chloroform
     Carbon Tetrachloride
     Trichloroethene
     1,1,1-Trichloroethane
     Tetrachloroethene
     1,2-Dichloroethane
     1,1,2,2-Tetrachloroethane
CHLORINATED AROMATICS
    Chlorobenzene
    o-Dichlorobenzene
    2-Chlorotoluene
                                   163

-------
                             Table 2
                    AVM Reproducibility Study


Description                   Mean*      Rel. Std. Dev.  (%)
Reactant Ions
Peak Height
Positive Mode
Negative Mode
Reduced Mobility
Positive Mode
Negative Mode
Product Ions
Peak Height
Positive Mode
Positive Mode
Negative Mode
Reduced Mobility
Positive Mode
Positive Mode
Negative Mode

6911
2109
1.87
1.60

935
679
1687
1.64
1.39
2.22

11.2
22.2
2.01
2.18

8.65
8.22
8.77
1.19
0.98
0.99
*:  Mobilities reported in cm2 V -1 s ^ and peak heights  reported
in millivolts.

**
  :  Ethylbenzene had 2 product ions.
                                    164

-------
            Ion Mobility Spectrometer
         Reaction Region          Drift Region    Vent
       Drif
       Car
              Ni63
          Gas
          ier Gas
       Repeller
                   Shutter
                                             Detector
Figure 1.  Schematic of ion mobility  spectrometer

               Acetone                2-Butanone

                            KO
        50   100   '50    200   250   50   100   <«>   200   250
                           Temper aluie
           3-Methyl-2-Butanone

 Figure 2.  Behavior  of  selected ketones over  the 9
 temperatures studied.   Legend for Figures  2 and 3: P:
 that moved over the  course of the elution.  The P marks the
 extremes of the mobility.   X: Distinct stable peak.
 Extremes of a drift  tube reaction broadened peak.
 Approximate center of  the peak associated  with a dri
 reaction.
                               165

-------
                                                                        Toluene
                              SO    100    '50     200     250    50     100     150     200
                                                          Temperature
                            200
                            Z50
                                                                                           1.75
                                      Ethylbenzene
                                                                         Sfyrene
                   Figure  3.    Behavior  of selected  aromatics  over  the  9
                   temperatures  studied.   See figure  2  for key.
                                                     DISCUSSION
COLLEEN PETULLO: Did you use the same IMS in the IMS-MS study or
were several used?

SUZANNE BELL: The IMS-MS instrument was different than the heated
instrument we used in New Mexico State. That's simply because we didn't have
an IMS-MS available, so we simply used one that PCP was gracious to rent us
for a week.

COLLEEN PETULLO: But you only used one in the study at any given time,
right?

SUZANNE BELL: Right. The nine temperatures and 43 compounds were all
run on one instrument. The IMS-MS was on another instrument, and then the GC/
IMS was yet another instrument.

COLLEEN PETULLO: How long would it have taken you to train the neural
networks if you would have programmed it for the 43 compounds?
SUZANNE BELL: I would assume it would take eight to ten hours, at the worst.
The training time gets longer as you get more and more similar data. If we gave
it. for example, 25 examples of benzene spectra over a wide concentration range,
that would let the network generalize but you pay the price in training lime. It
could take hours or weeks to train the computer.

COLLEEN PETULLO: You had mentioned that you didn't do this because of
time constraints.

SUZANNE BELL: Right.

COLLEEN PETULLO: How many did you ultimately program?

SUZANNE BELL: We ultimately trained 23 in the combined data set. This was
about half.
                                                               166

-------
              Remote and In  Situ  Sensing  of  Hazardous Materials
                     by  Infared Laser Absorption,  Ion  Mobility
                             Spectrometry  and Fluorescence
                                          Dr. Peter  Richter

                   The  Institute of  Physics, Technical University of  Budapest
                              1111. Budapest,  Budafoki u't 8.  HUNGARY
ABSTRACT
Three  instruments  will  be  described  that  were
developed  at  the  Technical  University  of  Budapest
for the  sensing of  hazardous  materials.   A remote
sensing  infrared   differential  absorption   lidar
based on  the coherent  detection  of  backscattered
CO2 laser light  has been built.   The  lidar can be
used for the detection of a  wide  range  of  molecular
pollutants  in  the atmosphere from  ranges of  a few
kilometers   along  a  path  to  a topographic  target.
Results  of field  measurements to  detect  molecular
pollutant clouds from km ranges  will  be presented.
The experiments  were   carried  out  on  NHs  and
DDVP but detection of  more  than  80 air-polluting
components  such   as  Freons,  SC>2, etc.  is  also
potentially   possible.   In  addition,  an  ion  mobility
spectrometer  will   be   discussed  which  has  been
developed   for in  situ  measurements  of impurities
in  air.  The  impurities  are  identified  with  the  help
of a dynamic dual-grid  cell.   Upon  evaluation of
the  frequency-ion   current   spectrum,  the  detection
of several  impurities  (e.g., NHs, DDVP,  HF  etc.) was
demonstrated.   The instrument can  operate either
in  a stand alone  or a  remote controlled mode and
can be  connected  to   a   central  computer.    A
fluorescence  detector for the  detection of surface
contamination  will  also be discussed.   Based on
chemical  indicator  reactions, UV  excitation  and
fluorescence  detection  via  fiber  optics,  a  mobile
instrument   for    detection    of    pesticide
contamination  and  control  of decontamination has
been built.   Reliability  detection  of  concentrations
of  0.1  mg/cm2  for DDVP  was  achieved  with   a
measurement  time of  less than 5  sec.   Applications
of  the instruments  and   methods  will   also  be
discussed.

INTRODUCTION	

Sensing hazardous  materials  is a  task  that  should
be   approached   using   techniques   that   are
appropriate   not   only   for  the  materials   to   be
detected  but  also  for  the  measurements  required.
A  variety  of  sensing  techniques  are  available  to
accomplish  this  end.    In this  presentation, three
different  methods and  instruments  that  have been
developed at  the  Technical  University  of  Budapest
will   be   described.    As  will  be  noted,  these
instruments   are  applicable  for  different   specific
purposes.    The   sensing  techniques  that   will   be
discussed are  as  follows:

       A remote  sensing  lidar to measure  pollutant
       clouds  in  the  atmosphere  from km ranges;

       An  ion  mobility  spectrometer  for  in   situ
       measurement of  air samples;  and

       An  UV  fluorescence   detector  to   measure
       surface   contamination   without   direct
       surface  contact.

REMOTE SENSING LIDAR	

Lidars are  laser  radars sensing  backscattered laser
light  from  long  ranges making  use of  the special
characteristics   of  laser  light.     Differential
absorption  lidars  measure  light intensities   at  two
wavelengths   corresponding  to  absorption  maxima
and    minima  of   the  absorbing   atmospheric
component  along the  beam  path.    Due   to their
broad  tunability   range  in   the   infrared  region
around the  10  u.m   wavelength   where    several
molecular    pollutants    have    characteristic
absorption   spectra,  systems  based  on  CO 2 laser
sources are  of major importance [1].   In the group
of more than  80 detectable  pollutants,  some of  the
more  important ones are: NH3, C2H4, 03, SC>2, SFg,
C2H3C1, as  well as pesticides such as  DDVP  (2,2
dichlorovinyl  dimethyl  phosphate).
                                                    167

-------
Two major problems  associated with  this technique
were  eliminating  the  disturbances  due to the  open
path   and   keeping   the   system  compact   and
transportable.   These  problems  were  solved  by the
development  of   the  system,  the  optical   part  of
which  is  shown  schematically  in Fig.  1.   Electronic
separation  of  the signals  at  the  two  wavelengths
allow  the  measurements  to  be  simultaneous  and
coincident,  thus  avoiding,  for  example, problems
due  to  turbulence  and  differential  backscattering.
Using   the   internal   amplification   of   the
backscattered   light  by the  lasers  and  heterodyne
detection, make  application of  small CW lasers and
a  transmit-receive  telescope   of  diameter   15 cm
only  possible.    Topographic  backscattering makes
long  path  absorption  measurement possible.    The
system used in the field tests is shown in Fig. 2 and
the  results   of  a   field   test   using  stationary
topographic  backscattering  from  500m   range  with
an  artificial  cloud of NH3  is shown in Fig.  3.  It  is
the  time dependence  of  the  differential absorption
signal
           K>.2.t)
E(t) =ln
that  is  displayed   where   ICXi^.t)   are  the
normalized  detected  light  intensities  at  the  two
wavelengths  at  time t.    The column content along
the  beam path  cL  (molecular concentration  c times
the path length L)  is given by
where   A (T     is  the  absorption  cross  section
difference  of   the   molecule   for   the   two
wavelengths.

The temporal variations of   E   in  Fig. 3 are due to
the  concentration  changes   in   the   cloud  blown
across the beam  path.  The time resolution is 1  sec.
Due to the atmospheric window around  A, = 1 0 |4.m ,
the  reference  range  of the  system  is about 3 km
(material   dependent)   and   is   not  significantly
influenced  by  the  visibility  conditions.

The  measurement   wavelengths  and  sensitivities
for some  specific molecules are as  follows:
                                   (cL)min
                            (ppb)(km)  (mg/m3)(km)
NH3       10.33    10.32

C2JU      10.53    10.59

O3         9.49     9.59

SO2        9.02     9.02

SF6       10.51    10.50

C2H3C1    10.61    10.50
8

8

22

710

1.5

34
                                      8.6 x  10 '3

                                      9  x  10 '3

                                      4.2 x  10 '2

                                      1.7 x  10 '2

                                      9  x  10 '3

                                      8.4 x  10 ~2
This system  can be used  in a  stationary mode when
with  a  scanning  attachment it can  monitor  either
large  area  (~ 30  km2)   pollution   distribution
(immission),  or  emission  from   certain   selected
sources.    When  coverage  of  a  larger   area   is
necessary,  it can  be  used from  a  flying platform as
well.

ION MOBILITY SPECTROMETRY _

A  simple  and  cost-effective  technique for  the  in
situ detection  of air  pollutants is through  the  use
of  ion  mobility  spectrometry.   Here  the  air sample
is ionized  by a radioactive source  in  a chamber  and
the  ions produced  are  moved  by  the use of  an
electric  field.   The arrival  time  and current of the
ions    characterize    the  products   and   their
concentration.     However,   as  the  predominant
charge   carriers  in  the   chamber   are  ion  clusters
consisting  of fragments  of  water,  Nitrogen  as  well
as  the  molecule  to  be  detected   (e.g.,  NH3,  HF,
CH3COH3, C2H5OC2H5,  HCN, different pesticides), the
selectivity  of  the system  requires the  application
of  sophisticated  hardware  and software  solutions.
12].

The structure of  the  chamber  is  shown  in Fig. 4.
Ambient air is  drawn  in across  a  semi-permeable
membrane   allowing   a  portion  of  its  component
gases  and  vapors  to  be  introduced to  relatively  dry
air in  the  ionizing  region.   An  alternating voltage
with   frequencies  sweeping  from  0-30   kHz   is
connected  to a dual  grid  of  transversal   Venetian
blind  type  in  front   of the  collecting  electrode.
Recombination   on  the  grid  is  dependent  on  the
mobility of  the  ions;  therefore,  evaluation of  the
ion  current  as  a   function   of grid  frequency
improves the selectivity  of  the  system.   In Fig. 5.
ion  currents  are  shown  as   a  function   of  grid
frequency   for   clean  air   and  air  with  NH3
Automatic  evaluation   of these curves  are carried
out  by  a  microprocessor taking derivatives of  the
ion    current    curve    at   five   characteristic
frequencies  that  correspond  to  f=OHz,  f(Imin).

       I(f2=OV) ,  f (jf = o) and f =  fmax = 30  kHz.

With  the  help  of  an  algorithm,  these  values  are
compared with  sets  of  stored  data  that  had been
determined   empirically.

Many  materials  can be  monitored  in  the  low  ppm
region.   The system shown in  Fig. 6  can  be used in
a  network   through   a  RS232  line  that   is  also
supported by  its  low   mass  and power consumption
(2kg,  1W).

SURFACE  CONTAMINATION DETECTOR _

Determining  the  contamination   of  surfaces   of
ground  areas  as  well  as equipment  and  personnel
and   the  verification   of  the  effectiveness   of
decontamination  from  hazardous   materials   are
important considerations  in  assessing the  extent  of
                             f l =
                                                        168

-------
residual   chemical   activity   such   as   in   the
application  of  pesticides  or  in  setting  clean  up
goals  for  site remediation.

With   the   technique    described    here,   the
monitoring  is based on   the  fluorescence  analysis
of chemical  compounds  produced  in   a  reaction
where  a  non-fluorescent  compound,  indole in  an
alkaline  peroxidase  solution  is  oxidized  by  the
agent  to  be  detected  to  give  highly  fluorescence
indoxyl [3].

To detect trace  impurities,  fluorescence  techniques
show  an  inherent  advantage  compared  to  methods
based  on  absorption.   Namely  while  the extinction
shows  a  logarithmic dependence  on  light  intensity
given  by
E = a c L = In
                   Io
                 Io -
the   fluorescent    light   intensity   F   exhibits
an  approximately  linear  relationship  given  by


F= QF Ia = QF Io (1 - e'ccL) = QF Io o c L  ,

where  IQ  is  the  incident  light  intensity,  Ia  the
absorbed   light intensity,  and  Q F is the  quantum
efficiency  of  the   fluorescence.    Therefore,  with
fluorescence  the   sensitivity  can  be  improved  by
increasing  the exciting  light   intensity  IQ.   Also
surface  contamination  often   appears   in   thin,
sometimes  discontinuous  layers  or droplets  where
the   additional   selectivity   provided   by    the
wavelength  discrimination   of   the  fluorescent
light   from   the   backscattered   light   can  be
exploited.

In  the  chemical  reaction  described  above,  the
material  to  be  detected  plays  the   role  of  the
catalyst;  therefore,  the  quantity of the  fluorescent
material  can  be  controlled  to   a certain  extent  by
the amount  of reagent added.

The  advantages   of  this  method  compared  with
those  requiring   probe  sampling  are,   that   this
method operates   without   physical  contact,  is  not
 influenced  by  the   surface  type,  and   is  highly
 selective.   The application  of   this method consists
 of  the following  steps:

             • spraying  the contaminated  area,

             • illuminating  it with  UV  light,  and

             • detection  of the  frequency  shifted
               fluorescent   light   and  evaluating
               the  detector  signal.

 This  system  (shown  in  Fig.   7)   consists  of  the
 following  three  units:
              •  a spray unit to store   and pump the
                chemical   reagents;

              •  an  optoelectronic  unit  housing  the
                Mercury  vapor  light  source,  the
                photomultiplier    detector,    the
                spectral  filters  matched   to  the
                compound   to  be  detected  and  the
                electronics  using  lock  in  detection;
                and

              •  a   sensor   head   unit   (containing
                optical  elements   and   controls)
                connected   with  3  m  long   hoses,
                cables  and  fiber  optic  bundles  to
                the other  units.

Experiments  carried  out with  DDVP  and  a reagent
containing   NaBOs  and  indol  in  water  solution
showed that  the response  time was less than  5  sec
after  spraying  and  the detection limit  was at  100
fig/cm^.  Time  duration  of the  fluorescence can  be
adjusted  by  proper  selection  of  concentration  of
the reagents.  This  system   can be  used  either in  a
stationary  mode  or  on  a  moving  vehicle  to  monitor
large   ground  surfaces.

REFERENCES	

1. Richter P., Proc. SPIE 883,  162 (1988)
2. Brokenshire  J.,  Pay  N., International
   Laboratory, Oct. 1989.
3. Diehl  W.,  Proc.  2nd Int. Symp.  Protection
   Chemical Agents,  Stockholm, Sweden,  173
   (1986)
                                                       169

-------
         Telescope
                                         m
                           Chopper

                                \
        Lasers
 b.s.
V
                                               Attenuator
rA   A  .	,
   Hm
       —'    V Detector
          Lens
                                          m
Figure  1.  Arrangement  of  the  differential  absorption  lidar.
                Figure  2.  The lidar system.
                               170

-------
              30    60     90     120    150    180    210    240   270    t
                1/DIV     START:!      SEC
                AVO5     1989.09.06.    15.23 "E"
  Figure  3.  Time  evolution  of  differential  absorption  signal
                      for an artificial  NHa  cloud.
                                        insulator rings
               metal housing
             source
            electrode
dual grid
collector
electrode
Figure  4.  Structure  of  the  ion  mobility  spectrometer  chamber.
                                      171

-------
l[pA]


 80 _


 60 -


 40 -


 20 -
                  air+NHg
                   /,
               I
              100 Hz     300 Hz     1kHz      3kHz
         I
                   I
                                                                       -»-  f
        10kHz     30kHz
     Figure 5. Dependence  of ion current  on grid  frequency for
                 clean air and  air with  0.2  ppm
 Figure  6.  The  ion  mobility
    spectrometer   sensor.
Figure  7.  The  surface contamination
        fluorescence   detector.
                                         172

-------
                              THE DEPARTMENT OF ENERGY'S
                            ROBOTICS TECHNOLOGY DEVELOPMENT
                          PROGRAM  FOR ENVIRONMENTAL RESTORATION
                                 AND WASTE MANAGEMENT
                                  A. C. Heywood
                     Science Applications International Corporation
                            Pleasanton, California  94566
             S. A. Meacham
      Oak Ridge National Laboratory
     Oak Ridge,  Tennessee  37831-6305
              P. J.Eicker
       Sandia National Laboratories
      AlbiK^uerque, Hew Mexico 87185
     In August  1989,  the new Office
of Environmental  Restoration and
Waste Management  (ER&WM) in the
Department of Energy  (DOE)
published an ER&WM Five-Year Plan
which established DOE's agenda and
commitment to correct existing
environmental problems, ensure
compliance with applicable  Federal,
State, and local  requirements, and
effectively execute DOE's waste
management programs.   The plan
includes a section covering the
applied research  and  development
needed to support the five-year
plan.  In November 1989, DOE Issued
a draft Applied Research.
Development, Demonstration.
Testing, and Evaluation (RDDT&E1
Plan for ER&WM  which  expands on the
applied research  and  development
section of the  five-year plan.  The
RDDT&E plan provides  guidance to
the new ER&WM Office  of Technology
Development (OTD)  for its mission:
"to manage and  direct programs and
activities to establish and
maintain an aggressive national
program for applied research and
development to  resolve major
technical issues  and  rapidly
advance beyond  current technologies
for environmental restoration and
waste management  operations."  The
development and application of
robotics technology for the
resolution of identified problem
areas at DOE sites is a major
element of the  RDDT&E program plan.
     The OTD has established a
Robotics Technology Development
Program (RTDP) to integrate
robotics RDDT&E activities and to
provide needs-oriented, timely, and
economical robotics technology to
support environmental and waste
operations activities at DOE sites.
DOE laboratories, private industry,
and universities have existing
robotics technology that provides  a
strong foundation for initiating an
aggressive RDDT&E program to
support ongoing and emerging ER&WM
functions.

     A major objective of the ER&WM
Program's five-year RTDP is the
application of robotic technology
in the resolution of DOE's
identified problem areas.  The
thrust of the application is to
reduce exposure of personnel to
hazardous substances and radiation
while increasing productivity.  An
additional goal is to integrate all
such activities to obtain the most
economical approach to resolving
site-related waste problems using
robotic technology and to
demonstrate robotic technologies
that can be applied to major site-
specific waste clean-up efforts.

     The Robotics Five-Year Program
Plan provides the focus and
direction for the near-term  (less
than five years) and guidance  for
the lona term  (five to twenty
                                       173

-------
years) R&D efforts  associated with
resolution of  site-specific waste
problems.  The goals  include: (1)
supporting the ER6WM  Program and
being responsive  to the ER6WM Five-
Year Plan, (2) focusing near-term
robotic R&D efforts to be
responsive to  application
requirements,  (3) ensuring that
robotic applications  are responsive
to site requirements  and scheduler
needs, (4) integrating all robotic
activities to  obtain  the most
economical approach to resolving
site problems while reducing
personnel exposure, and (5)
providing guidance  for the Office
of Energy Research  long-range (>5
year) robots research program.

Program Focus and Objectives

     The Program currently
addresses a number  of important
issues facing the ER&WM activities
at the DOE sites.   Among the areas
included are:

     •   underground storage tanks
        (material characterization
        and remedial actions),

     •   buried waste retrieval,

     •   waste minimization,

     •   contaminant analysis,
         automation,

     •   decontamination and
        decommissioning,

     •   basic and applied research
        and development required to
        support the above areas.

     The objectives of the Program
are to develop, test,  evaluate,  and
make available robotic technologies
that:
     •   allow workers in waste
        operations  and remediation
        to be removed from hazards,

     •   increase the speed and
        productivity with which
        ER&WM operations can be
        carried out when compared
        to alternative methods and
        technologies,

     •   increase the safety of
        ER&WM operations, and
     •  provide robotic and remote
        systems technologies that
        have lower life cycle costs
        than other methods and
        technologies.

     In addition to developing
robotics technology, the program
promotes the availability of the
technology and supports its
deployment and use in ER&WM
activities at DOE sites.  The
program further serves as a bridge
between the ER&WM robotics RDDT&E
and the basic robotics research
carried out by DOE's Office of
Energy Research, providing guidance
for the basic research program and
integrating its results in applied
research and advanced development
projects.

Program Organization

     In order to execute the
Program, the Program has been
structured as shown in Fig. 1.
Since the Program is an element of
the DOE ER6WM Applied RDDT&E
program, it is administered by the
ER&WM OTD through the Program
Manager (RPM).

     To ensure that the Program
responds to the needs of the DOE
complex, RPM is assisted by an
Operations Review Group (ORG).
This group is familiar with the
ER&WM issues facing the DOE
complex.  RPM also receives
assistance from a Technical Review
Group (TRG) of robotics and
automation experts from the DOE
laboratories and sites,
universities, industry, and other
federal agencies.  A Program and
Budget subcommittee of the TRG also
assists the RPM.

     The Robotics Applications
Coordinators (RAC) develop robotics
program plans focused on each of
the major ER&WM issues.
     The RAC is responsible for
coordinating the flow of technical
information relevant to the
applications area among those
groups having an interest in the
area.  RAC is also responsible for
keeping the other groups in the
relevant applications areas
apprised of the results of RTDP
                                       174

-------
 r   sir-
 ?: (v»«t)
  Coordinator
      ST~
 7  (east*
 ^Coordinator
 5?
    Burled
    waste*
  Contaminant
 /  Analyst*
  Automation
  •:   Waste
  Minimization
  Coordinator
   D«contam.
      and
  ;  Dftcomm.
  Coordinator
     Wasta
    Faculties
  t Operations
  Coordinator

       C
       i

       II
       n
Ficurc   1
\TDP Oreamzanon
               175

-------
funded activities.  The
coordinator,  with the approval of
the RPM also convenes occasional
conferences on the applications
area.

     The coordinators function as
the advocate for the technologies
applicable to their particular
problem area.  To facilitate the
application of the best technology
with a high probability of success
to the particular problem area, the
coordinator actively solicits
proposals from the entire robotics
and automation community for
routing to the RPM.  A thorough
familiarity with the ERSWM problems
and issues is required of the
coordinators.  This familiarity
will be maintained through site
visits, personal contacts, and
symposia where appropriate.

     Applied research is funded
through the applications center
that has identified the
technological need.  This helps
insure that the applied research is
responsive to the needs of the
group sponsoring the research.
Coordinators who put together a
team approach with industry, labs,
•universities, or other agencies are
most favorably reviewed.

     The R&D Coordinator  (RDC)
reports to the RPM and is
responsible  for coordinating the
flow of technical  information  other
than applied research.  The RPM  is
familiar with all  aspects  of the
RTDP and is  able  to  identify areas
of  future  need in robotics and
ancillary  systems  which are not
being  addressed  in the applied RSD
areas.   He is  responsible for
coordinating with universities,
industry,  DOE  laboratories,  and
other  federal  agencies to bring
proposals  for need advanced
technology to  the TRG and RPM.

Program Planning

      A comprehensive technical
program plan has been developed
during the first year of  funding.
This initial plan development is a
significant effort since the plan
 is based on the needs of  the
environmental restoration and waste
management operations as identified
by the eight DOE field offices and
the sites they administer.  A major
portion of the initial plan
development is assessing and
understanding those needs.  The
technical program plan covers a
five-year period with primary
emphasis on the one-year plan and
secondary emphasis on the two- and
three-year projections.  The plan
covers technical work, budget
requirements, and schedules and is
tied closely to the requirements
and schedules of individual site
environmental restoration and waste
management projects.

FY 1990 Accomplishments;  The RTDP
accomplished a number of
significant activities in FY 1990,
which facilitated a fast start for
robotics technology development and
established a sound basis for
program activities over the next
five years.

Program Planning:  Five priority
DOE sites were visited in March
1990 to identify needs for robotics
technology in environmental
restoration and waste management
operations.  This 5-Year Program
Plan for the RTDP was prepared on
the basis of the needs identified
at the DOE sites, and provides a
needs-based road map for detailed
annual plans for robotics
technology development.

Initiating Interactions with the
Robotics Technoloov Community:  In
July 1990, a forum was held
announcing the robotics program.
Over 60 organizations  (industrial,
university and federal laboratory)
made presentations on their
robotics capabilities.

Technology Demonstrations:  To
stimulate early interactions with
the ERSWM activities at DOE sites,
as well as with the robotics
community, the RTDP sponsored four
technology demonstrations related
to ERSWM needs.  These
demonstrations integrate commercial
technology with robotics technology
developed by DOE in support of
areas such as nuclear reactor
maintenance and the civilian
reactor waste program.
                                         176

-------
     Rapid, swing-free movement of
simulated waste containers was
demonstrated using control
algorithms developed at Sandia
National Laboratories (SNL) with
technology in computer control of
large gantry bridges at Oak Ridge
National Laboratory (ORNL).  This
technology decreases the time for
materials movement and increases
safety by eliminating the potential
for collisions of swinging
payloads.

     A scaled waste tank
remediation demonstration at SNL
integrated sensors and advanced
computer control into a commercial
gantry robot.  The extensive use of
models for robot system control
allowed graphical programming of
the system complete with operator-
supervised path planning to
increase speed of repetitive waste
removal tasks.

     A teleoperated vehicle with
advanced sensing technologies for
mapping of buried waste sites was
demonstrated at a small buried
waste site at ORNL. Navigation
technologies were coupled with the
sensing information (from
radiation, gas, and subsurface
large object sensors) to
automatically map subsurface
materials.

     A team consisting of LLNL,
SNL, LANL, SAIC, and IBM demon-
strated a robotic system for
loading powder into a furnace in a
Pu production line, and then
transferring the product to the
next operation in a mock up
facility.  This robotic system
eliminates the need for operator
hands-on transfer operations and
reduces the generation of operator-
associated waste materials such as
wipes, protective clothing, gloves,
and transfer bags.

SITE VISITS/NEEDS

     In March  1990  RTDP planning
teams visited  five  DOE sites.
Additional site visits will be
conducted  in the  future to expand
the planning basis.
     The purposes of these visits
were (1) to understand the needs
and requirements of the highest
priority environmental restoration
projects and waste management
operations at the sites, (2) to
obtain information for use in
planning the program, and (3) to
describe the RTDP to personnel at
the site and discuss development of
the program plan.  Emphasis was
placed on both technical and
schedular (i.e., compliance dates)
needs and requirements.

     The results of these visits
are documented in a Site Needs and
Requirements Document.  This
document summarizes the findings at
each site and highlights priority
needs.

APPROACH TO NEEDS DIRECTED
TECHNOLOGY DEVELOPMENT

     The visits to five DOE sites
led to selection of six areas of
need for robotics technology to
support ER&WM activities.  These
need areas are:

     •  Remediation of waste
        storage tanks,

     •  Retrieval of buried wastes,

     •  Automation of contaminant
        analyses,

     •  Waste minimization,

     •  Decontamination and
        decommissioning,

     •  Waste Facilities Operations

     Plans for development and
application of robotics technology
are based on the need areas listed
above.  In addition, the plans
reflect other aspects of needs at
the sites such as regulatory
compliance dates, planned remedial
actions, and established schedules.

     The fundamental approach to
developing robotics technology to
meet these needs couples available
and emerging technology with
advanced technology.  Near-term
needs can be met by integrating
                                        177

-------
 available  commercial  technologies
 with  emerging  technologies
 available  in RSD laboratories.   At
 the same time,  development  of
 advanced technology will  proceed to
 meet  intermediate and long-term
 needs.  In addition,  attention  will
 be given to development of  cross-
 cutting technology which  will be
 applicable to  multiple need areas.
 Technology development will be
 keyed to integrated  demonstrations
 at the DOE sites  to further couple
 the robotics technology development
 to the site needs and to  the
 deployment of remedial actions
 technology.

     The DOE sites  are evaluating
 alternative approaches to remedial
 actions. The robotics  technology
 developed  for each  application must
 meet the needs, and match the
 approach selected by  each site.
 The plans described for robotics
 technology development are  based on
 reference concepts, selected as
 reasonable and  likely concepts from
 the alternatives, which form the
 basis for identifying needed
 technology development, estimating
 schedules,  and  estimating budgets.

     The robotics technology
 development plans are also  keyed to
 demonstrations  of technology at the
 DOE sites.   Wherever possible,
 demonstration of  the robotics
 technology is integrated with
 larger integrated remediation
 technology demonstrations.

CROSS-CUTTING AND ADVANCED
TECHNOLOGY DEVELOPMENT

     Near-term applications of
 robotics to ER&WM activities is
 necessarily focused on existing
 technologies that can be readily
 adapted to the  specific cleanup
 tasks and environments.  As the DOE
 cleanup activities progress and
 evolve,  a larger  body of robotic
 technology will be  needed for
 application to  ER&WM  projects.  A
 technology development program
 targeted at relevant  cross-cutting
 and advanced technology development
 will make possible  a more rapid
 insertion of beneficial technology
 into these activities. This
 technology development will be
 focused on high payback projects
 that  offer safer,  faster,  or
 cheaper  approaches to cleanup
 goals.

      An  advanced technology
 development program including a
 long  term  research and development
 component  is a  means to effectively
 incorporate the expertise  of the
 universities, national laboratories
 and other  basic research
 organizations into the nation's
 cleanup  projects.   Also, this
 offers educational training
 opportunities consistent with the
 DOE emphasis on developing the next
 generation technical work  force.

      Needs  identified at DOE sites
 indicate that cross-cutting and/or
 advanced technology development  in
 the areas  listed below would be
 highly beneficial  to application of
 robotics in  ER&WM  activities.

 Mechanical Subsystems
      Manipulators
      End-Effectors
      Mobile  Systems

 Control Subsystems
      Computing,  Graphics and
      Modeling
      Man-Machine Interfaces
      Communications
      Telerobotic Operations
      Motion Planning and
      Control

 Sensor Subsystems
      Environmental Sensors
      Servo Mechanical Control
      Sensors
      Imaging & Vision Systems
     Multi-Sensor Integration

     Cross-cutting and advanced
 technology developments need to
 focus on near-term, mid-term, and
 long-term implementations.   By
 investing in a sustained long-term
development program, emphasizing a
balanced evolution in technology
development with implementations
continually encompassing technology
advances, steady progress may be
assured toward the technology
required for the more complicated
or demanding tasks of the decades
to come.   Development of advanced
robotics technology that is
commonly applicable to many
environmental restoration,  waste
                                        178

-------
management,  and waste minimization
activities can lead to higher
efficiency,  increased reliability,
and reduced life cycle costs in
these operations.

     Participants in this program
are the following whom we wish to
thank for their contribution.

SAIC -    Science Applications
          International Corporation
LANL -    Los Alamos National
          Laboratory
SNL  -    Sandia National
          Laboratories
LLNL -    Lawrence Livermore
          National Laboratory
ORNL -    Oak Ridge National
          Laboratory
T-12 -    Oak Ridge Y-12 Plant
RF   -    Rocky Flats Plant/EGiG
          Rocky Flats
SR   -    Westinghouse* Savannah
          River Company
'.VHC  -    Westinghouse Hanford
          Company
?NL  -    Pacific Northwest
          Laboratory
£GSG -    EGSG Idaho
INEL -    Idaho National
          Engineering Laboratory
WMC  -    Westinghouse Materials
          Company of Ohio
WINCO-    Westinghouse Idaho
          Nuclear Company, Inc.
Fernald   Feed Materials production
          Center
                                    179

-------
                                     Field  Robots  for
              Waste Characterization and Remediation
                     William L. Whittaker
                     Field Robotics Center
                  Carnegie Mellon University
                     Pittsburgh, PA 15213
                        (412)268-6559
            David M. Pahnos
          Field Robotics Center
       Carnegie Mellon University
          Pittsburgh, PA 15213
             (412) 268-7084
Abstract
field operations for waste characterization and remediation
offer real  opportunities and  compelling  motivations for
advanced robot work systems. The  application  of field
robotic technology can enhance the quality of data collected
al waste sites  through standardization,  verification, and
repeatability of methodology. It can increase the coherence
of data  by  enabling  dense  data collection, advanced
correlational databasing, and the collection of previously
unavailable  data,  such  as   position  tagged  data  or
inteipretable 3D subsurface images. Held robots can operate
were humans  are precluded,  in pipes,  tanks, abandoned
mines, and sea and river bottoms or where humans perform
inefficiently in protective clothing and breathing apparatus.
Thus, field robots can greatly increase the knowledge base
gained during  site  investigations; this  knowledge will
expand remediation options performed by human and open
the way for the use of field robots in remediation activities.
Moreover,  the  development  and  use  of  field  robotic
technologies in the service of national efforts to characterize
Mid remediate nuclear and hazardous waste will eventually
lave profound effects on large commercial industries and
open new world markets for robotic technologies.

Introduction
Hazard has been the historical justification for the use of
field robots; operations surrounding accidents at Chernobyl
"id Three Mile Island have world impact, preclude humans
 and call field robots to action. Less reactive than these crises
 are the innumerable nuclear, deep sea, military, and space
 operations  that are  inhospitable  to humans  and  are
 significant  both strategically and  fiscally.  The ultimate
 opportunities,  however, for field  automation  are  those
 immense and inefficient industries like construction, mining,
 timbering, hazardous waste management, subsea and outer
 space  that  dwarf  the  economics   of manufacturing.
 Characterization and  cleanup  of the  nation's  weaponry
 complex alone is now estimated at 100 billion dollars: efforts
 of this magnitude require new technologies. As a growing
 technology, the potential of field robots to apply sensing and
 analytical capabilities and to perform precise, repetitive, and
 dangerous tasks is virtually untapped in the world.

 Field robots work in environments as they are encountered,
 not idealized or altered to accommodate automation. While
 an assembly process can be structured into a limited number
 of predictable actions, a robot working in an unstructured
environment encounters new situations that it has not been
explicitly programmed to deal with.

Field robots are thus challenged to perform goal-driven tasks
that  defy pre-planning  in  unpredictable and changing
environments. In  order to explore, work, and  safeguard
themselves and the environment, field robots must sense
complex phenomena in a dynamic world. As these  robots
move towards  autonomy, they must plan and implement
their work tasks.
                                                     181

-------
 Robots are quickly becoming mobile in natural terrains,
 perceptive, self-navigating, and  competent in the field.
 Within the next few years, a number of robotic performance
 niches in waste  characterization  and remediation will be
 exploited where  humans are precluded from the scene or
 where  robots  offer  superior  capabilities.  Areas  of
 opportunity include reconnaissance, surveying,  subsurface
 imaging,  soil  gas sampling, perimeter monitoring,  fast
 analytical screening,  accident response, remote sampling
 and manipulation, remote coring, and excavation.

 Automated Characterization
 Perhaps the most frustrating aspect of waste characterization
 is the paucity of  reliable data that scientists and engineers
 have to work with following an investigation. Field sampling
 is expensive, time consuming, and labor intensive. Although
 methodologies are standardized,  human judgement  and
 sometimes intuition are broadly  applied when deciding
 where to sample  or survey and how to interpret data once
 collected. This is  particularly true  for the  selection of
 boreholes, the interpretation of geophysical data, and the
 selection of soil and soil  gas  sampling  points. Analytical
 instruments and techniques have improved greatly over the
 past several years, but the results  are  only as good as the
 choice of sampling points, which often  are too few  and
 chosen poorly.

 Field robots can  deploy screening instruments far  more
 rigorously, sampling hundreds or thousands of times  per
 acre, achieving total site coverage.  They can create a three-
 dimensional data  base  by  analyzing air, soil gas, and  the
 subsurface: they can screen organics on the fly and create 3D
 images of buried waste  from radar data,  sampling  at
 centimeter resolution. Held robots can survey a site and
 layout a precise grid; take samples, position tag, package,
 and label them; position tag instrument data, store the data in
 a single spatially correlated data base, and present multiple
 types of data to users in a straightforward visual format.

 Quality of Data
 Capable field robots can greatly increase the quality of data
 from a waste site  by obtaining verifiable data with a high
degree of repeatability, and they can advance the process of
 data collection to a higher standard than is possible using
 present methodologies. Ultimately, field robots can also help
 ensure  that  the  right  samples are  sent  to  analytical
 laboratories.

 Standardized Data
 Most waste sites have long lives; the time from preliminary
 assessment to the remedial action can stretch into years, and
 monitoring can take place for decades after. Throughout the
 life of a site, scores of scientists, engineers, technicians, and
 workman perform  tasks, and as a site transitions  from
 assessment to investigation to remediation, the cast of actors
 changes.

 Although  methodologies   are   standardized,  no   two
 investigations at a site are performed in exactly the same
 way; indeed, no two investigators can be relied upon to bring
 the  same experience, judgement, and skill to a site or to
 collect data in exactly the same way, thus making it difficult
 to achieve standardization.

 Moreover, because waste sites vary greatly in topography,
 soil types, geology, and the nature of contaminants, it is
 difficult to achieve  standardization across a range of sites,
 partly because humans perceive the sites differently.

 The use of field robots to collect and screen  data  can
 significantly improve standardization. Robots can be relied
 upon to  treat data in the same way in each  investigation.
 Robots  eliminate  human variables and collect far greater
 quantities of data. The data thus become more reliable, and
 data from  different sites  can be compared legitimately.
 Ultimately, a single, complete data base can follow a site for
 its entire life. Created during the preliminary assessment, a
 three-dimensional computer data base can be an interactive
 repository in which each new set of data is entered.

 Verifiable and Reoeatable Data
 Field robots can verify data taken previously at a site and
 repeat the collection  and screening  process precisely.
 Because  robots process  and store  data at  the  time  of
 collection, the chain of custody can be maintained more
reliably and securely. Repeatable outcomes translate  into
defensible  conclusions  and reduce  uncertainly  when
                                                        182

-------
planning remedial actions and issuing a record of decision.
Field robots can become an important tool in the process.

Relevant Data
Two ways to increase the relevance of data are to collect it in
quantities great enough to yield high statistical reliability and
collect several types of data at the same time. Reid robots
can build dense  data bases. They are also  capable of
deploying a range of sensors that humans cannot; e.g., three-
dimensional  laser rangefinders, infrared  sensors,  sonar,
radar, etc. In addition, they can deploy analytical instruments
smultaneously and determine their position accurately in
global coordinates.

UK site investigation robot (SIR) under  development at
Carnegie  Me lion's Field Robotics  Center collects ground
penetrating radar  data (GPR) at two centimeter intervals,
accumulating  in excess of a  400,000 data points per acre.
GPR  data are inherently  three-dimensional  and  can  be
processed into 3D images, if the data are dense enough. A
human cannot attain the positioning accuracy or deploy the
sensor with enough precision to collect dense data, as a robot
can. The result is not just more data but new and better data.
Further,  the robot can be configured to collect additional
types of data or samples simultaneously, e.g., organics in air
or soil.

hJeipretable. Usable Data
Investigators  are  often confronted with data that do not
asily yield  to interpretation  or,  at  worst, require the
investigator to make a guess as to what the data show. Field
robots can process data, making it easier to visualize and
understand.

GMU's Site Investigation Robot provides a visual image that
is not only quantitatively better but qualitatively better than
 standard GPR data bases. The user is provided with an image
 defined accurately in x, y, and z, making the  data more
 Weipretable, even to a novice.

 Dta bases become more usable when one is  able to see
 oxidations among data in new ways.  The availability of
 multiple  types of  data  superimposed on a  computer-
 derated site map will enable investigations to gain a whole
site profile in a single visual image. This kind of user power
will not only speed the investigation process but  give
entirely new insights to investigators.

Accessible Data
Finally, when data are accessible to many people over time,
the likelihood of good use being made of the data increases
significantly. Data collected by field robots can be stored on
central file servers, available to all who need to determine
what is known about a site or who have new data to add to
the file.

When Humans Are Precluded
Some  investigations  and  remedial  activities  preclude
physical human access, such as the interiors of pipes, tanks,
and ducts; abandoned mines;  and  river, harbor, and sea
bottoms. Field robotic technologies offer the best access to
collect data and to perform remedial activities.

Generations  of  competent  pipe  crawlers  have  been
developed and are in service in petroleum and natural gas
industries. In-tank inspection robots and remediation robots
are needed at DOE complexes. One such robot is being
developed by RedZone Robotics to inspect tanks containing
nuclear waste. At CMU's Field Robotics Center, we are
developing autonomous navigation  and vision systems for
underground mining equipment and  autonomous navigation
systems for  walking machines and wheeled vehicles to
traverse rough terrain. Others have significant experience
with competent sub-sea robots and have demonstrated their
capabilities and utility.

Another class of sites precludes humans  because of health
and safety concerns, e.g., high-level waste, mixed waste,
transuranics,  unbreathable atmospheres, unknown  waste,
and accident response. These sites present high-motivations
for robots to perform not only reconnaissance and sampling
activities but forceful manipulation and heavy work to a high
degree of precision. These  activities  include excavation,
loading,  haulage,  and  packaging  of diffuse materials;
removal of sludges and mixing of materials; removal of
debris; barrel handling; boring on gassy landfills,  and the
handling of explosive materials or operations in explosive
environments.
                                                        183

-------
 Field Robotic technologies have now progressed to the point
 where the robotics community can begin to build competent,
 rugged, and reliable systems to meet the performance needs
 of waste characterization and remediation programs.

 Integrated  Characterization  and  Remediation
 Systems
 Robotic technologies can fulfill the need to better integrate
 characterization and  remediation systems. An excellent
 example of this is the case of trenched transuranic wastes.
 Conditions preclude most invasive means of characterizing
 the volume and position of the waste, and having a human
 onboard of an excavator is precluded during the remediation.

 The work can, however, be performed by robots in a
 coordinated sequence. A site investigation robot (SIR), using
 ground  penetrating radar, can produce measurements of
 buried waste in x an y to a  reasonable accuracy (7 to 14
 inches), which would allow a robotic excavator to trench on
 both side of the waste to install steel sheeting. The excavator
 would have the SIR's position data  and subsurface map
 available to it to guide it through the digging process, along
 with active sensing of its own.

 The SIR also surveys the z axis, determining the depth of the
 waste and the distance from  the soil surface to the waste.
 Through a sequence of iterative sensing and excavation, the
 clean overburden could be removed, leaving 4 inches of soil
 covering the waste. The excavator could then remove the
 waste autonomously.

 In this scenario, robots working together can perform the
 tasks more efficiently and with greater accuracy than human
 operators. Five years ago sensing and control in both robots
 to the degree of accuracy described above would have been
 wishful thinking; two years ago it was beyond the reach of
the technology, today it is within reach, and although it is not
yet  ideal  for selectively  finding and  excavating,  deeply
buried hot spots, it is likely the safest, most cost effective
 approach to retrieving radioactive, trenched wastes that can
be expected in the next several years.
 Future Opportunities
  Commercial applications  for capable  field robots  will
 number in the hundreds. Among them are significant field
 robotic applications that are achievable in the near term with
 evolutionary extensions to our  current  technology base.
 Moreover, there are significant opportunities, some of which
 are unique  to the U.S., e.g., robotic timbering,  surface
 mining, and large-scale agriculture.

 Federal agencies should not miss opportunities to develop
 and apply robotic technologies in programs where they have
 a legitimate interest and obligation to protect human health,
 increase productivity, and decrease costs. Because robotic
 technologies are extensible  to many applications,  there
 should be a coordinated effort by Federal agencies to  1)
 focus performance-based research to move the technologies
 forward; 2) apply the technologies in Federal programs
 where they will produce high-leverage results, sufficient  to
 pay for the investment; and 3) ensure that programs will be
 sufficiently  stable   over  time  to  attract   world-class
 researchers to the field.

 There is an opportunity to reduce significantly the total
 cleanup costs of chemical and  nuclear waste sites through
 the programmatic development of robots to  perform site
 investigation, data collection, and remedial activities. The
 core technologies have reached a stage of development  to
begin the  task of putting together integrated, teleoperated
 and  semi-autonomous systems  for  this  purpose. The
opportunity is to alleviate a major national problem and,  at
the same time, to develop and apply new technologies that
will impact the world.
                                                        184

-------
                                                          DISCUSSION
BRIAN PETERS: You mentioned American leadership. What about the position
of the Japanese in this area? They're well known for corporation robotics on
tdomobile assembly lines, for example.

WILLIAM WHITTAKER: The Japanese are a significant force in this arena.
hiticularly, they have programs that have matured, driven in a strategic way, top
down,over several years, and they look very good. They look extremely good in
construction. They have lesser presence in subsurface and in space. Consider, if
you will, that we enjoy a 20- or 30- year history in space, and they're just building
ieirfirstrockets. But to bring it to terms here, I look for the United States to drive
Ibis agenda because we are the ones who pioneered some of the nuclear
tdmologies, and we are the ones that have the volumes and the programs to go
atelhis.

Resistance, if you looked at the navigation technologies, there aren't a lot of
places in Japan that have enough roads to drive something like that. And so if you
tola the agenda in the program, I think that is enough to really focus operations
lot I actually have a video tape of condensed Japanese technology that I just
pal together this week. After this session I'll be happy to show that.

GREGG DEMPSEY: On your  remote vehicles that stand  completely alone,
0% run on  telemetry or whatever)  is the technology such that  if there's  an
incident out  on a site or something, and you lose communication, can the
machine actually turn itself around and come back?
WILLIAM WHITTAKER: Yes, that technology is available. However, I think
it's important to know that it's in very select pockets of seasoned research groups,
and very select pockets of small organizations that  can move fast to put it
together. Specifically, that kind of technology source is from the DOD Strategic
Computing Initiatives and DARPA's Road Following Programs, which were
funded at the hundred million dollar level over a number of years, going back
three or four years.

GREGG DEMPSEY: I remember when the robots went into Three-Mile Island
there were problems with the camera lenses darkening up because of the radiation
exposure. Has that problem been solved to any great extent?

WILLIAM WHITTAKER: In the first deployment in November of 1984, it's
true that the cameras didn't function well. And that's because we were using a
CCD technology. It was small, and it was very new! But within a month that was
straightened up. And with the years that have gone by, particularly out of military
and space initiatives, rad hardened CCD's are a known  technology.  It's very
straightforward now.

GREGG DEMPSEY: So we have technology that can operate in the thousands
of roentgens per hour now?

WILLIAM WHITTAKER: Yes.
                                                                     185

-------
             SPACE TECHNOLOGY FOR APPLICATION TO TERRESTRIAL HAZARDOUS
                              MATERIALS ANALYSIS AND ACQUISITION
                                      Brian Muirhead   Susan Eberlein
                                      James Bradley    William Kaiser

                                      NASA/Jet Propulsion Laboratory
                                      California Institute of Technology
                                               Pasadena, CA
ABSTRACT

In-situ and remote measurements of elemental, molecular
and mineralogical composition of materials has been part
of the space science program since its beginnings.  There is
a great deal of commonality between space science missions
and terrestrial hazardous materials screening in the types
of measurements, methods and instrumentation used.
There are also strong parallels between the hostile
environments of space and those of a hazardous material
This paper discusses the measurements, methods and
nstrumentation used on past, present and future space
missions for in-situ and remote analysis of materials.
Specific instrumentation discussed includes gas
chromatographs, mass spectrometers, imaging
spectrometers, X-ray and gamma-ray spectrometers.
Work sponsored by the National Aeronautics and Space
Administration's Sample Acquisition, Analysis and
Preservation technology program is discussed, including
concepts and hardware for multi-spectral remote sensing.
Instrument data analysis and interpretation, material
acquisition and processing.  Some new concepts for micro
Sensors for  making various chemical measurements are
also discussed. Possible applications of space technology to
terrestrial hazardous materials field acquisition and
analysis are presented.

NTRODUCTION

In-situ and  remote measurements of elemental, molecular
and mineralogical  composition of materials has been part
of the space science program since its beginnings.  Two of
the best known surface science missions were the Viking
mission to the surface of Mars and the Soviet Venera series
to Venus. The Galileo spacecraft is carrying a probe to
sample Jupiter's atmosphere and the National Aeronautics
and Space  Administration (NASA) has just started a project
to make a variety  of in-situ measurements of the comet
Kopff. NASA is currently working on technology to enable
robotic and  human missions to the Moon and Mars.  Such
missions will include a wide variety of in-situ and remote
science and engineering measurements. There is a great
deal of commonality between space science missions and
terrestrial hazardous materials screening in the  types of
measurements, methods and instrumentation used as well
as in the hostile  nature of the environment in which  these
measurements are  made. NASA is very active in the design,
development and utilization of the instruments.  Table 1
contains a listing of some science data requirements and
associated  instrument(s) that are used and/or under
development within NASA for its past, present and future
missions.

NASA has established a technology program called Sample
Acquisition, Analysis and Preservation (SAAP) to address
the specific needs  of in-situ  science and engineering
measurements. SAAP is intended to develop critical and
significantly enhancing technologies for remote
identification, acquisition, processing, analysis and
preservation of materials for in-situ  science, engineering
characterization and earth return.  Although the  technology
being developed in the SAAP program is not currently
being applied to specific missions, the SAAP program will
broaden the base of technology available for future
missions. Specifically, SAAP is developing concepts and
hardware  for  multi-spectral  remote  sensing, instrument
data analysis and  interpretation,  material acquisition and
containment [1,2,3,4].  Some new concepts for
microsensors for making various chemical  measurements
are also under development. There are many possible
applications of space technology to terrestrial  hazardous
materials field acquisition and analysis.

SPACE INSTRUMENTS, MEASUREMENTS AND
APPLICATIONS

There  is very  high scientific value to direct surface
measurements, independent of whether a sample is
returned to a  laboratory. In particular, the analysis of
volatiles is probably best done in-situ due to the potential
for loss or chemical change after prolonged storage.  For
space  applications, in-situ measurements may be a
necessity because of the limitations on sample return.
                                                       187

-------
                    Table 1. SCIENCE DATA REQUIREMENTS vs INSTRUMENT TYPES
                        Required Data
                                       Example Instruments
                 Elemental Composition

                 Mineralogical Composition

                 Water Detection and Mapping
                 Atmospheric Composition
                 Subsurface Structure
                 Seismometry
                 Volatiles
                 Imaging
                 Exobiology
                 Magnetic Fields
                           Gamma-ray Spectrometer, a-p-x Spectrometer
                           XRF, a-Backscatter
                           Visible-Infrared Spectrometer
                           Mossbauer Spectrometer, DSC, XRD
                           Neutron Spectrometer, Electromagnetic Sounder
                           GCMS, Laser Spectrometer
                            Electromagnetic Sounder, Active Seismometer
                           Passive Seismometer
                           DSC-EGA, Visible-Infrared Spectrometer
                           Camera, Imaging Sectrometer
                           Viking Biology Instrument
                           Magnetometer
Although terrestrial applications do not face the same
limitations, major advantages in speed and accuracy can be
gained by  employing field analysis prior to selecting
samples for laboratory study.

Below are  listed some of the characteristics of a few
instruments that have been flown by  NASA or are being
proposed for NASA future missions.  The constraints on
mass and power, combined with the need to function in a
hostile  environment,  place severe requirements on these
instruments. The technology developed to meet these
requirements could benefit the production of similar
instruments for terrestrial applications.
         .AM,.
- ELECTRON
 MULTIPLIER
                          ION SOURCE
                          HOUSING
                   r
  ION
  PUMP
                              r
1

1

  MAGNET
 L"  = 2 6 CM
                                                ELECTRIC
                                                SECTOR
                              Re=47CM
                Rm- 38 CM


                       L'e = 1 7 CM-i
                           = 1 7 CM
   Figure 1. The mass spectrometer for the Viking Lander
   GCMS. The electric sector has a radius of 4.7 cm.
Chemical Analyzers

The prime example of a chemical analyzer is the Biology
Experiment on the Viking Landers.  The experiment
included a GC-MS system for analysis of organic
compounds in Martian soil [5]. The GC-MS part of the
system had a mass of 16 kg, measured 28 cm x 38 cm x 27
cm and consumed 25 to 125 W when active. When the
system was presented with a soil sample it could sift a soil
sample into a pyrolysis tube, seal the tube to a GC inlet,
perform a controlled  heating on the sample, and perform a
mass spectral analysis of the GC effluent with
exceptionally high sensitivity.  The mass spectrometer also
had a direct inlet for analysis of the Martian atmosphere.
Figure 1 shows a diagram of the mass spectrometer.

Currently under development for the Comet
Rendezvous/Asteroid Flyby mission is the Cometary Ice and
Dust  Experiment (CIDEX) instrument that incorporates  a
3-column GC system for evolved gas analysis over a
sample temperature  range of -90 to +1000 C.  The
instrument also includes an x-ray fluorescence
experiment in a 15 kg package that uses an average of
about 22 W.  The system will analyze comet dust for
organic materials and elemental composition.

New GC-MS systems have been proposed that combine the
analytical speed of microbore GC columns with  the
exceptionally high sensitivity of a focal-plane mass
spectrometer equipped with an integrating focal plane
detector.  Such a flight system would be comparable in size
and mass to the Viking Lander GC-MS, but with analytical
cycle times of a few  minutes and the ability to analyze GC
peaks separated by a few hundred milliseconds.  Such a
system could measure dynamic processes or determine
planetary atmospheric composition while descending on  a
probe or parachute.  The robust, portable nature of such an
instrument would make it a good candidate for deployment
in terrestrial field screening activities as well.  A gas
chromatogram from a laboratory  prototype is provided  in
Figure 2.
                                                        188

-------
 10000 -i
>; 8000-
(0
I
I 6000-
Q
   4000
   2000
      0-
                 4 5
                            8
              lOuil
 9
A
              50    100   150    200    250   300    350
                        FRAME NUMBER

  Figure 2. Chromatogram of a mixture of EPA priority
  pollutants. Each 50 mg frame contains a time-integrated
  mass spectrum from mass 25 to 500 amu. (Peak 1 is air
  and peak 9 is toluene.)
 Elemental Analyzers

 Gamma-ray spectrometers have been used in orbiting
 spacecraft to obtain elemental maps of atmosphere-free
 bodies such as the moon.  The Mars Observer spacecraft
 wil contain a gamma-ray  spectrometer for elemental
 mapping of the Martian surface through  its thin
 atmosphere. The recently built and proposed gamma-ray
 systems for elemental analysis have tended to follow
 commercial technology by use of cooled germanium
 detectors. These detectors use radiators aimed into cold
 space to achieve the required temperatures. The detected
 elements are those with naturally  radioactive  isotopes or
 which are excited by cosmic rays.  Long counting times are
 needed.  Related instruments may  be useful in the remote
 determination of radioactive isotope composition at
 terrestrial  sites.

 New. high efficiency x-ray fluorescence  analyzer systems
 have been proposed for lunar and  Martian landers that  use
 new toroidal focussing crystals to achieve many orders of
 magnitude increase in x-ray flux from microfocus x-ray
 lube sources to achieve rapid and high sensitivity analyses
 [6]. With the use of uncooled mercuric  iodide x-ray
 detectors, such an x-ray fluorescence system might have
 ""ass of 4 kg. consume 10 W, and occupy a volume of about
 35cm x 25 cm x 25 cm.  The same microfocus x-ray
 source could be used in a high-efficiency, toroidal-
 fccussing powder  x-ray diffractometer for identification of
 minerals. Both instruments can work in an atmosphere of
 w x-ray absorption density, such as  that on Mars, or in
 vacuum.
VISIBLE AND NEAR INFRARED REMOTE SENSING

Imaging spectrometers play a major role in both Earth
observation and planetary exploration.   The Airborne
Visible/Infrared Imaging Spectrometer (AVIRIS) images
with 20 m x  20 m spatial  resolution in 224 spectral
channels from 400 to 2450 nm wavelengths [7,8]. The
data, obtained from NASA ER-2 aircraft at 20 km altitude,
is  spectrally  and radiometrically calibrated to provide
information for disciplines such as ecology, geology,
oceanography, inland waters, snow hydrology and
atmospheric science. An AVIRIS type instrument might be
used for aircraft tracking of ocean oil spills, smoke
plumes, or other indicators of chemical contamination.

In addition to visible and near infrared imaging
spectrometers, NASA has developed a portable backpack
point spectrometer (Portable Instantaneous Display and
Analysis Spectrometer - PIDAS).  At a mass of about 30
kg, PIDAS obtains and records with integrating  detectors,
reflectance spectra in 830 bands from 400 to 2450  nm.
The instrument, developed  at JPL, has been used to support
geological and ecological disciplines, and can be calibrated
for identification of a wide  range of materials.  The
instrument field of view is  10 to 30 cm when hand held.
NASA is currently, working to develop an adaptive, reliable
and compact imaging spectrometer system for autonomous
site and sample selection and analysis of materials.  This
system will provide wide area as well as close-up
identification of minerals which  is enabling  for surface
science and engineering missions.

The key element of the SAAP remote sensing subsystem is a
multi-spectral imager based on  the  solid-state acousto-
optic tuneable filter (AOTF).  This device operates on the
principle  of acousto-optic  interaction in an  anisotropic
medium  and acts as a controllable narrow band filter.  The
current breadboard version can collect spectral images at
4 nm spectral resolution in the visible range (0.5 and 0.8
microns). It  has been implemented with a  1000x1000
fiber optic bundle between the foreoptics and the AOTF.
The fiber optic cable enables the mounting and articulation
of the foreoptics, remote from the main spectrometer  body.
Figure 3 shows the current breadboard  hardware.

By altering the pass  band sequentially, only the desired
spectral bands are collected.  Each pixel has a spectral
signature associated with  it and  classification is
accomplished on the  basis of elemental content and spatial
location.  Figure 4 shows a set of spectrometer images of a
rock containing the rare earth mineral  neodymium taken
in the range  of 783-710 nanometers. The absorption
characteristics of this mineral at around 750 nanometers
is  evident in the dark spot  in the right-center of the second
row of images.  Figure 5 shows the complete spectral
signature of neodymium as  taken  by the AOTF
spectrometer.
Although  the current instrument operates in  the visible
region, the AOTF technology will also allow construction of
tunable filters for the infrared and ultraviolet regions of
the spectrum, with a total  range  between 0.35 and 25
microns.  This may provide a new class of tunable spectral
analyzers for a variety of space and earth applications.
                                                        189

-------
CO
O

                                               Figure 3. AOTF Spectrometer Breadboard

-------
 - ;Cl •  lul    ;SJk   ;Ui   liB  -. lul

 ^ii ^^ ^ *& -' wv :'&  ^  • *«•
  iJl   iJl    iA   lA   xfll   Jl
   1200


   1150


LL?
3 1100
>
z
9. 1050
—

m 1000
i—
z

    950


    900
             560  nm  to 528 nm
  Figure  4.  AOTF Spectrometer Output  in
  710-783  nm
The completed imaging spectrometer will be capable of
collecting high resolution images at hundreds of discrete
wavelengths. Processing of such a large amount of
information  (>1 gigabit per scene)  will  strain
computational systems without some means of data
reduction.  Hierarchical analysis schemes, in combination
with neural nets, have been shown to produce several
orders of magnitude  reductions in total computation time
and are discussed below.

INTELLIGENT DATA ANALYSIS

Spectral data from a variety of instruments is used in
many areas of chemical analysis. The proceedings of the
First International Symposium  on Field Screening Methods
for Hazardous Waste Site Investigation [9] report on the
use of fieldable  instruments for mass spectroscopy,  x-ray
fluorescence spectroscopy, infrared spectroscopy and
Raman spectrosopy.  For any of these instruments, the
spectral data produced is complex, requires a highly
trained chemist to assist in the  interpretation  process,
and often requires extensive computer work for proper
analysis. In many cases the data analysis and interpretation
step presents a  significant  bottleneck which prevents the
most efficient utilization  of the  instruments.

Work done within the SAAP program has concentrated on
the the analysis of visible and near infrared spectra for
mineral determination [10].  The developing system
incorporates a number of data analysis methods and
algorithms which will transfer readily to use with  other
types of spectral data.  Application of these approaches to
the instrumental  analysis required for field screening of
toxic waste will improve the speed and efficiency of the
 analysis step. Table  2 shows a comparison for speed and
 accuracy of four  classification  methods.  The first matched
                                                                      0.7            0.6

                                                                            WAVELENGTH (urn)
                                                                                                         0.5
       Figure 5. Neodynium  Absorption  Spectrum
       from  AOTF  Spectrometer

     filter is a brute force approach using full dimensionality
     of all patterns, and requiring the most computation. By
     reducing  the dimensions used for matching, or performing
     the matching in several steps (e.g. a grouping step and a
     finer classification step), the  computation is reduced.  The
     hierarchy of neural network pattern classifiers combines
     these approaches.  Images consist of 32-band spectra for
     all pixels, and are classified as one of 28 known minerals
     in each case.


     Neural networks are trained  to recognize spectra or
     classes of spectra by presenting many examples of each
     spectrum, complete with noise and normal variation in
     features. Following training,  new variants of the spectra
     contained in the training set  may be identified with a high
     degree of accuracy. During the training  procedure, the
     network extracts the common features among the training
     examples representative of each type of spectrum, and
     learns  to recognize these as important  identifying factors,
     while the noise is discarded. Thus new spectra are
     classified based on the  presence of the diagnostic features
     specific to a type of compound, without significant
     interference due to normal variation, noise, and
     background contamination. The major components in
     mixture spectra  may also be identified, if the  mixing
     process does not obscure the critical features.

     The neural network spectrum classifiers  currently  used
     within  the SAAP system work  hierarchically, placing
     spectra into progressively more detailed classes. This
     approach allows either a rough estimate of mineral
     composition, or  a very  detailed analysis and identification.
     The final analysis step includes an assessment of the
     classification accuracy. This allows the system to identify
     those spectra which were poorly classified, and which may
     represent mixtures or other  unexpected spectra. Since the
                                                       191

-------
Table 2. COMPARISON OF 4 SPECTRAL CLASSIFICATION METHODS
METHOD
Single Matched Filter

Reduced Dimension Matched Filter

Two Step Matched Filter

Hierarchy

DATASET TOTAL OPERATIONS
Mars
AISA
Mars
AISA
Mars
AISA
Mars
AISA
16,226,560
5,017,600
8,113,280
2,508,800
6,374,720
1,971,712
4,858,284
1,006,099
ACCURACY
80%

80%

69%

89%

          Note: Mars dataset is a simulated multispectral image derived from a Viking Lander image.
          AISA dataset is a real multispectral image taken by the Airborne Imaging Spectrometer.
final application of this spectral analysis system requires
almost complete automation of the analysis process, the
results of the spectral analysis are integrated into an
automated decision making procedure. The decision making
is goal-driven: specific classes of minerals may be
searched for and analyzed in great detail, while other less
important compounds are discarded at an early step in the
analysis procedure.

The goals of the existing (planetary) spectral analysis and
decision making system include identifying interesting and
uninteresting  areas on the basis of spectral information,
and identifying samples which  should be acquired for more
detailed analysis.  Similar goal driven systems could be
designed with the objectives of finding specific types of
chemical compounds or determining which samples will
prove most informative regarding chemical distribution  in
an area.  The hierarchical goal driven architecture allows
the system to analyze many samples rapidly, and to provide
the user with information  regarding which samples are
most  important  for further examination.

Application to field screening for hazardous waste:

Two aspects of the work done for spectral data analysis in
planetary exploration will be of interest for the  field
screening of hazardous waste. The neural network based
spectral  analysis approach will be useful for the  analysis
of IR, XRF, Raman, and mass spectra, if networks are
trained with real spectra gathered under the anticipated
field conditions. The hierarchical  analysis architecture
that incorporates goal driven decision making may be
adapted to assist field workers in making rapid decisions
regarding the areas requiring special attention  during  a
field screening operation.
 Although special neural network pattern recognition
 systems will be required for each type of instrument  data,
 the basic algorithms  developed for the analysis of
 visible/near IR mineral spectra should transfer readily to
 the analysis of other spectra. A hierarchical,  neural
 network based spectral identification system will have
 several  applications:

 1. Unknown identification.

 A network based hierarchy can replace a library search
 procedure with favorable results for the identification of
 unknown spectra. Progress is being made in the
 implementation of hardware network pattern  matchers
 which will allow the  equivalent of very  large  library
 search procedure to occur in microseconds.

2. Searching for specific compounds.

A hierarchy of networks is  particularly well suited to the
search for specific compounds. A spectrum is presented to
the hierarchy, and  is progressively  classified until it
becomes apparent that the spectrum does not represent the
desired compound (or until the desired compound is found).
A negative result is usually determined fairly quickly,
since at each step of the hierarchy, a large group of spectra
may be eliminated since they are not potential  matches.

3. Searching for classes of compounds based on specific
features.

This is a variant of the hierarchical search for a specific
compound, with the difference that a positive result may
occur when a given branch point of the hierarchy is
reached, rather than only at the end of the search. The
                                                          192

-------
hierarchy is designed so that the groups of spectra that
represent important classes are together within a branch
of the hierarchy. The selection  of critical  spectral features
for identifying a class is ensured by using specific spectral
bands for training the networks. Extensive knowledge of
the chemistry is  required at  the training step for optimal
results.

4. Extracting major components from mixtures.

Identification of spectra of mixtures presents problems for
traditional library search and match techniques.  Since
mineral spectra generally derive from mixtures of pure
minerals, this problem is being addressed in the work
wttiin the SAAP program. The neural network approach
has the advantage of basing results on important features
which are extracted from the  anticipated data in advance,
rather than on complete spectral matching. This allows
identification of major components in many mixtures.
Situations where mixing causes masking or shifting of
critical spectral  features require special  treatment.

SYSTEM CONCEPTS

h-situ analysis  systems can range from  single
instruments placed on  the surface to multi-purpose,
mobile units looking for specific materials or unique
materials units.  An autonomous space exploration system
nil require the functions of planning, analysis,  execution
control,  reflex action, data processing and interpretation,
in order to operate in real time in a hostile environment.

for an in-situ analysis  subsystem, the spectrum of
possible architectures can be characterized by two
extremes.  At one end is a set of disjoint, self-contained
elements working more or less independently  to perform
the required functions.  At the other extreme  is a fully
integrated system with  many interdependent relations
between the elements.  The former case is probably more
comparable  to the terrestrial applications, where several
independent instruments are operated by humans. This
system design causes some problems for space systems
since it  is not efficient in terms  of mass or power and
compromises science due to uncoordinated measurements.
Multi-instrument  data fusion and corroboration is  an
important consideration in this system design.

An extreme example of the latter case is a multi-purpose,
foctory-like system,  implementing a set of processes that
nay vary significantly depending on the desired outcome  or
product.  Physical material, not just data, must move
Between the elements.  Current requirements and desires
brcoordinated measurements as well as mass, power and
volume limitations make an integrated design approach the
logical basis for technology  requirements, but this
approach clearly pushes technology.  Technology developed
fa such an integrated system could be applicable to the
automation of sample gathering and analysis in extremely
tostile earth environments, in cases where human
fraction must be remote and limited for safety reasons.

Technology will be validated in  the laboratory and then
toegrated into the series of evolving SAAP testbeds. The
 representative environment provided by the testbed will
 be used to verify technologies and demonstrate overall
 SAAP operational capability. By the end of September
 1992 an  initial laboratory testbed will be constructed to
 demonstrate sample identification  and acquisition.  By the
 end of 1995 a fully functional system testbed will be in
 operation which  will  transition into  a complete self-
 contained transportable testbed for end-to-end "field"
 operations.  A preliminary system  conceptual design of a
 SAAP platform with a full complement of subsystem
 components except for a regolith deep core drill is shown
 in Figure 6. This configuration can be considered a
 preliminary  model for the full-up  system testbed;  no final
 payload or mission configuration has been selected.

 SAMPLE ACQUISITION

 The capability to acquire physical samples robolically,
 without human intervention,  would  be  significantly
 beneficifial in  many hazardous waste screening
 applications. The principal requirement driving sample
 acquisition for planetary exploration is to obtain samples
 of weathered and unweathered materials from accessible
 rocks or outcrops.  These samples must not be  significantly
 altered either mechanically or thermally during
 acquisition.  Conceptual designs and early experimental
 work have been completed to help understand the
 mechanical, controls and automation issues for sample
 acquisition in  the hostile environment of a planetary
 surface. Effort has focused on mechanical designs to
 achieve functional capability and is now proceeding to
 include testing of control and automation methodologies.
 Laboratory validation  at the component level will be
 followed by further development and verification at the
 system level in a  series of SAAP testbeds.

 Various techniques have been studied for sample
 acquisition including sawing, coring and chipping.  Of
 these, core  drilling represents an efficient way of
 obtaining surface and subsurface samples that are easy to
 handle  by a preparation or storage subsystem.  Terrestrial
 coring processes, however, require  direct human
 supervision  and utilize high power and introduce large
 volumes of fluid to aid the cutting process by cooling the
 bit and removing cuttings of rock  and/or soil.

 SAAP has developed the means for core  drilling low
 porosity, high  compressive  strength  rocks without the use
 of coolant.  High  velocity diamond matrix core barrels are
 used under the control of robotic manipulators.  Under
 study are various control approaches and a  variety  of
 sensors modalities including, position, force, vision,
 spectral, temperature and vibration.  Progress in this area
 should improve the prospects for  remote robotic
 acquisition of solid samples  from hazardous  areas on earth
 as well.

 In addition to tools, work is  underway to identify and
develop end effector and manipulator technologies
necessary for  the sample acquisition operations.
 Preliminary studies of end effector and manipulator
dexterity versus  reliability,  mass,  power and performance
have been made for some mission  scenarios. The current
                                                       193

-------
   6 DOF ARM
      7 DOF ARM

          SAMPLE PREPARATION SYSTEM
                              MULT-
                              SPECTRAL
                              IMAGER
                                                                                         TOOL/INSTRUMENT
                                                                                         BOX
                                                                                             SAMPLE CANISTER
                                                                                        X-RAY FLUORESCENCE
                                                                                        SPECTROMETER
                      -REDUCED DEGREE OF
                       FREEDOM END EFFECTOR
                 X-RAY DIFFRACTOMETER

         DIFFERENTIAL SCANNING
         CALORIMETER

  •GAS CHROMATOGRAPH/
   MASS SPECTROMETER
                          Figure 6. SAAP  Preliminary System Conceptual Design
state-of-the-art in end effectors consists of either very
limited  capability industrial  vise-type grippers, or
extremely complex anthropomorphic designs being studied
in research laboratories.  In general, the fewer degrees of
freedom the better for simplicity.  However, to achieve
high inherent reliability, mechanical redundancy at each
degree of freedom will be required. Concepts that provide
adaptability  or flexibility and involve  trade-off's of
degrees of freedom with redundancy will be studied
further.

ADVANCED CONCEPTS

       NASA is interested in developing new sensing device
technology  for in-situ science investigations.  Currently
available instruments for in-situ science investigations
are often incompatible with mission requirements due to
their excessive mass, volume  and power consumption.
Science capabilities may be significantly extended by  the
development of sensing device systems which represent
smaller payloads. The sensing device development is
directed to  enable compact, low-mass, low-power
consumption instruments for a variety of mission
requirements. The advanced technology of silicon
micromachining for device fabrication will be employed to
implement highly capable, sensitive, and robust
instruments while retaining compact structure and low
mass attributes.
        The development of silicon micromachined gas
 sensors will be based on the compact gas chromatography
 (CaC) instruments recently  demonstrated in silicon
 micromachined structures.  The key components of the
 compact GC systems include a silicon micromachined gas
 dispersion column, integral gas metering valves and
 silicon thermistor gas detectors, fabricated entirely on a
 single silicon wafer. The successful operation of this
 prototype time-of-flight GC system indicates the range of
 opportunities for unique  instruments of this type.   In this
 task, specific gas detector applications will be identified
 and instrument requirements will be formulated.  Gas
 sensors and instruments will fabricated and tested for
 operation in the Martian atmospheric  environment
 Finally, with results of device testing, complete
 instruments will be designed for specific mission
 applications.

 CONCLUSION

 This paper has discussed some of the measurements,
 methods and instrumentation used on past, present and
 future space missions for in-situ  and remote analysis of
 materials.  Work sponsored by NASA's Sample Acquisition,
Analysis and Preservation technology program included
 concepts and hardware for multi-spectral remote sensing,
 instrument dafa analysis and interpretation, and material
                                                        194

-------
acquisition, and new concepts for micro sensors for making
various chemical measurements. Much of the technology
under development in the SAAP program has application to
terrestrial hazardous  waste materials acquistion and
analysis.
REFERENCES

1)     Moreno, C., Editor, In-situ  Science Investigation
System Catalog, Version 1.0, JPL  Document, June 5, 1990

2)     B. Muirhead and G. Varsi, (1990), Next-
Generation In-situ Science Concepts and Technology, IAF
90-444, 41st Congress of the  International Astronautical
Federation,  Oct. 6-12,  1990.

3)     Muirhead, B.,  et al, (1989), " Sample Acquisition,
Analysis and Preservation Technology Development",
Presented at the 2nd International Conference on Solar
System Exploration, Pasadena, CA.

4)     Plescia, J., Editor, Sample Acquisition,  Analysis
and Preservation Instrument Technology Workshop,
Proceedings, Johnson Space Center, November 14-16,
1988.

5)     D.R.  Rushneck, A.V. Diaz, D.W. Howarth, J.
Rampacek, K.W. Olson, W.D. Dencker, P. Smith, L
McDavid, A. Tomassian, M. Harris, K. Bulota, K. Biemann,
Al. LaFleur, J.E.  Biller, and T. Owen,  (1978), Viking gas
chromatograph-mass  spectrometer, Rev.  Sci. Instr.
49(6),  817-834.

 6)    D.M. Golijanin and D.B. Wittry, (1988).
 Microprobe  x-ray  fluorescence analysis - new
 developments in an old technique, Microbeam  Analysis-
 1988,  D.E. Newbury,  Ed., San Francisco  Press. 1988,
 397-402.
 7 )     W.M. Porter  and H.T. Enmark, (1987), A system
 overview of the Airborne Visible/Infrared  Imaging
 Spectrometer (AVIRIS), Proc. SPIE. 834.

 8 )     G. Vane, M. Chrisp, H. Enmark, S. Macenka, and J.
 Solomon,  (1984),  Airborne Visible/Infrared  Imaging
 Spectrometer:  An advance tool  for earth remote sensing,
 Proc. 1984 IEEE Int'l Geosciences and Remote Sensing
 Symposium,  SP215,  751-757.
 9 )     Field Screening Methods for Hazardous Waste Site
 Investigation,  First International Symposium Proceedings,
 October  11-13,  1988.

 10)   Eberlein, S., Yates, G. (1990). "Neural Network
 Based System for Autonomous Data Analysis and Control",
 In "Progress in  Neural Networks". Volume 1,  pp 25-55,
 Ablex Publishing Corp.
 ACKNOWLEDGEMENTS

 The Sample Acquisition, Analysis and Preservation
 Program is part of the Exploration Technology Program
 within the NASA Office of Aeronautics, Exploration and
 Technology.  This project is the joint effort of the Jet
 Propulsion Laboratory, Johnson Space Center and Ames
 Research Center, with JPL as the lead center.

 The research described in this paper was carried out by
 the Jet  Propulsion  Laboratory, California Institute of
 Technology, under a contract with the National Aeronautics
 and Space Administration.
                                                 DISCUSSION
BRIAN PIERCE: My question concerns the fiber optic bundle. You said
infrared. Do you mean the near infrared or closer to the mid i.R.?

SUSAN EBERLEIN: Right now the fiber optic bundle that we've actually
wwked with has only been in the visible range. We're looking this year in the near
infrared of 1.2 to 2.5 microns. In the long-term maybe more, but I gather that as
you go further into the infrared you get more trouble with your fibers.

BRIAN PIERCE: Yes, that's right. You also mentioned very intriguing hard ware
neural networks. What do you mean by that?
SUSAN EBERLEIN: What I mean by hardware neural networks is micro
silicon chips where the connection weights for the neural network matrices are
actually in the resistances in the chips. JPL is fabricating some of these. They are
still in the early stages, and not as precise as we need them. Some othercompanies
are working on making them commercially as well. If in fact they turn out to be
a viable technology that can be space qualified, they offer very, very rapid
processing for specific problems.
                                                          195

-------
                  DEVELOPMENT OF A REMOTE TANK INSPECTION (RTI)

                                            ROBOTIC SYSTEM
      Chris Fromme
      RedZone Robotics, Inc.
      2425 Liberty Avenue
      Pittsburgh, Pennsylvania 15222
      (412) 765-3064
Barbara P. Knape
RedZone Robotics, Inc.
2425 Liberty Avenue
Pittsburgh, Pennsylvania 15222
(412) 765-3064
Bruce Thompson
RedZone Robotics, Inc.
2425 Liberty Avenue
Pittsburgh, Pennsylvania 15222
(412) 765-3064
ABSTRACT:
RedZone Robotics, Inc. is developing a Remote Tank
Inspection (RTI) robotic system for Westinghouse Idaho
Nuclear Company to perform remote visual inspection of
corrosion inside high level liquid waste storage tanks.  The
RTI robotic system provides 5.8 m (19 ft) of linear extension
inside the tank to position a five degree-of-freedom robotic
arm with a reach of 1.8 m (6 ft) and a payload of 15.9 kg (35
Ib). The primary end effector is a high resolution video
inspection system. The RTI Intelligent Controller provides a
standardized, multi-tasking environment which supports
digital servo control, I/O, collision avoidance, sonar
mapping, and a graphics display. The RTI robotic system
features an innovative, standardized, and extensible design
with broad applicability to remote inspection,
decontamination, servicing, and decommissioning tasks
inside nuclear and chemical waste storage tanks.


I.   APPLICATION

    Westinghouse Idaho Nuclear Company (WINCO) will
use the RTI robotic system at the Idaho Chemical Processing
Plant (ICPP) to perform remote visual inspection of corrosion
inside high level liquid waste (HLLW) storage tanks.  The
ICPP tank farm consists of several HLLW storage tanks that
are 15.2 meter (50  ft) in diameter with a capacity of
1,135^00 liters (300,000 gallons).  The domed roofs of the
tanks are buried 6.1 m (20 ft) below ground level. The bottom
of the tanks are located approximately 12.5 m (41 ft) below
ground level. The  tanks will be drained of liquid prior to
inspection,-however a 30 cm (1 ft) layer of caustic sludge will
remain on the bottom of the tanks. The only access to the
tanks is through 25 cm (10 in) and 30 cm (12 in) diameter riser
pipes which extend from ground level down into the tank
roof dome. Accessible risers are typically located 0.8 m (2.5
ft), 3.6 m (12 ft), and 6 m (20 ft) away from the tank wall.
Currently, the RTI  system will only be deployed through the
30 cm (12 in) tank risers. Cooling coil arrangements line the
tank walls and the tank floor.
    The primary mission of the RTI robotic system is to
perform remote visual inspection of the interior walls of the
tanks for corrosion which may have been caused by the
                 combined effects of radiation, high temperature, and caustic
                 chemicals present.  Due to the location and limited number of
                 accessible risers inside a tank, the intent is to inspect only a
                 pie-shaped portion of the tank to qualify the typical
                 condition of corrosion inside the tank. Thus the application
                 does not require a robotic arm with a long reach.


                 n.   SYSTEM OVERVIEW

                      The RTI robotic system features a vertical deployment
                 unit, a robotic arm, and a remote control console and
                 computer. One of the major design constraints for the RTI
                 system is that the in-tank components are inserted through a
                 25.4 cm (10 in) diameter riser. This criteria lead to the
                 design of compact,  electric actuators for the robotic arm,
                 which provide high torque and absolute position feedback.
                 The RTI robotic system is initially lowered by a facility
                 crane into the top of the riser. The vertical deployment unit
                 then provides another 5.8 meters (19 ft) of servo controlled
                 extension inside the tank. The RTI robotic system transmits
                 minimal loading to the riser pipe since it is self-supporting
                 via a support structure that rests on the ground above the
                 riser.  Figure 1 provides an illustration of the RTI robotic
                 system installed inside a tank.
                     A five degree-of-freedom robotic arm provides 1.8
                 meters (6 ft) of articulated reach to accurately position a
                 high resolution video inspection camera to examine the tank
                 walls. The arm has sufficient dexterity to position the
                 camera normal to the curvature of the tank wall. The
                 controller provides  coordinated end point  motion so that the
                 operator can easily  jog the arm inside the  tank. A graphics
                 display is provided at the control console  to give the
                 operator a sense of how the arm is positioned inside the
                 tank.  The robotic arm also positions a pressurized spray
                 nozzle to wash down the tank walls prior to inspection. In
                 addition, the end of the arm has an interchange flange to
                 allow the robotic arm to carry a gripper instead of the
                 inspection camera.  Another camera system is mounted at the
                 top of the robotic arm to provide the operator with an
                 overview of the arm operating inside the waste tank.  The
                 RTI robotic system is capable of manual recovery to retrieve
                 the system in event of motor failure.
                                                     197

-------
                                                                                   LUTING CAGE
       ID'S" NOMINAL
        911- NOMLNAL
                 Z-AXIS ACTUATOR WITH BRAKE
                 AND HOMLVG LIMIT SWITCH
          S-5" NOMINAL
          411" TO 511-ADJUSTABLE
                                                                                                      TETHER MANAGEMENT SYSTEM
                                                                                                      19-OF CABLE
                                                                                                      COUNTER WIND ELIMLVATES NEED
                                                                                                      FORSUTRING
                                                                                                      SUPPORT STRUCTURE
                                                                                                      0 • ir ADJUSTABLE FOOT PADS
                                                                                                      GUIDE SLEEVE
  167- NOM. RETRACTED/ 35V NOW. EXTENDED
    HORIZONTAL
    REACH 6TT
    (WITHGRIPPER)
                                                                                             SHOULDER ROTATE (±180°)
                                                                                             .TILT MOUNTED OVERVIEW CAMERA
                                                                                             REMOTE FOCUS, ZOOM AND IRIS
                                                                                                  ATT VARIABLE INTENSITY LIGHT
                                                                                                    'NARSENSO:
                                                                                                       SHOULDER PITCH (±90°)

                                                                                                       EL BOW PITCH C±120->



                                                                                                       WRIST ROLL a ISO-)


                                                                                                       WRIST PITCH (±120-)


                                                                                                     CRTPPER(t!-4-.0-70LBS)
25T SOM. RETRACTED/ 431 r NOM. EXTENDED
                   41 tT TO TANK BOTTOM -
                                 Figure 1.  RTI Robotic System Deployed Inside HLLW Waste Tank
                                                                     198

-------
       The RTI system is radiation and environmentally
hardened to assure reliable performance in the tank
environment.  The design criteria requires that all in-tank
components be capable of withstanding a 20 psi washdown of
10% nitric acid and 10% oxalic acid, radiation field of 100
Rad/hr for a total accumulated dose of 10,000 Rad, and
operating temperatures of 4 to 49 °C (40 to 120 °F) at 100
percent humidity. The RTI system uses sealed components
such as connectors, video equipment, sensors, and actuators to
preclude the intrusion of decontamination fluids. Bearing
and wear surfaces are stainless steel and non-stainless
components are anodized or coated with epoxy paint to
prevent damage from caustic decontamination fluids.
    The RTI's control system uses RedZone's standardized
Intelligent Controller for Enhanced Telerobotics to provide a
high speed, multi-tasking environment on a VME bus.
Currently, the robot is controlled in a manual, joint jog mode
or a coordinated end point motion control mode.  Control
capability is available to develop a pre-programmed,
automated or teach/playback mode of operation. The
control system incorporates sensing and software safeguards
to prevent an operator from inadvertently colliding with the
tank wall. Collision prevention is implemented in software
and backed up with four proximity sensors. A sonar range
finding sensor is  used to establish the orientation of the RTI
robotic system  inside the tank.


HI. MECHANICAL DESIGN

    The major components of the RTI mechanical system are
the support structure, vertical deployment unit, robotic arm,
accessories, and strongback. These assemblies are described
in the sections  that follow.

    A.  Support Structure
         The support structure rigidly supports the vertical
deployment unit  at ground level.  It consists of the alignment
guide sleeve and  support stand assembly. The support stand
is a four legged structure that spans the riser pipe and
bunker. Its leg pads provide 1 foot of vertical adjustment and
allow the stand to be levelled. A facility crane is used to
position the support structure over the riser and to insert the
alignment guide  sleeve into the riser pipe. The guide sleeve
follows the inclination of the riser pipe to guide the vertical
deployment unit  during insertion. The objective is to avoid
loading the riser pipe if it is not absolutely vertical.

    B.   Vertical Deployment Unit
         The vertical deployment unit provides 5.8 m (19 ft)
of servo-controlled vertical extension, at speeds of up to 7.6
cm/sec (3 in/sec), to position the robotic arm inside the waste
tank. The vertical deployment unit consists of a telescoping
tube assembly,  cable management system, drive motor, and
junction box. The telescoping tube assembly contains a fixed
outer tube and an inner extending tube to minimize the
overall retracted  height of the system. With the inner  tube
extended, the wrist flange of the arm can reach the tank
floor. An adjustable hard stop is provided to safely reduce
the extent of vertical travel.  The outer tube is a 20 cm (8 in)
square stainless steel tube and the inner tube is a 15 cm (6 in)
square tube.  The vertical deployment tubes are designed for
deployment through 30cm (12 in) risers. However, the
robotic arm is designed to pass through a riser as small as
25 cm (10 in). The inner extending tube is supported and
guided along the upper tube by stainless steel linear bearings
and rails.  The rails are mounted along the length of the
inner tube and the bearing blocks are attached to the inside
of the outer fixed tube.
     An electric motor drives the lower tube, Z-axis, by a
dumb-waiter arrangement of a drive chain and pulley. The
motor package includes an integral gear reducer, brake and
resolver. The motor's output shaft is directly coupled to a
drive sprocket which drives a steel chain attached to the
upper section of the inner tube. The chain moves within the
gap between the upper and lower tubes. The drive sprocket
was designed so it can be driven from either side. In the
event of a motor failure, an identical backup motor package
can be quickly mounted in order to drive the telescoping tube
assembly. Due to the relatively large gear ratio and  large
travel of the chain, absolute position feedback on the
vertical deployment was avoided. Instead, a resolver is
attached directly to the motor shaft  and a limit switch is
used to home Z-axis position at start-up.
     After insertion into the riser, the top flange of the
vertical deployment unit is bolted to  the guide sleeve. On
top of the vertical deployment are located the cable
management drum and a junction box. Cabling is payed out
from a spring loaded cable drum which has a large diameter
so that only two wraps are required to pay out the 5.8 m (19
ft) of cable length.  This design obviates the need for
electrical slip rings. The vertical deployment junction box is
connected to the control console with 30.5 m (100 ft) of cable.
The junction box contains some pneumatic and valve
equipment and terminal strips but no circuitry. Its main
purpose is to serve as a termination point for cables routed
down the vertical deployment unit to the robot arm.
     At the base of the vertical deployment unit is a
mounting flange for the robotic arm.  Cables are routed
internal to the inner tube and exit the tube at its  bottom.  At
the bottom of the outer fixed tube, a spray ring is mounted to
spray decontamination fluid on the inner tube as it retracts
upward. This minimizes the spread of contamination inside
the telescoping tube assembly.

     C.   Robotic Ann
         The RTI robotic arm mounts to the bottom of the
lower extending tube.  The arm is a five-degree-of-frcedom
revolute arm consisting of shoulder rotate, shoulder pitch,
elbow pitch, wrist roll and wrist pitch axes.  The primary
function of the robotic arm is to position the WINCO
inspection camera system mounted to the wrist flange. The
arm has sufficient degrees  of freedom to position the
inspection camera normal  to the curvature of the tank wall.
Coordinated end point motion control allows the operator to
move the inspection camera in/out and along the curvature ol
the tank wall.  An overview camera is packaged between
the shoulder rotate and pitch joints to rotate with the arm,
allowing a continuous view of the end of arm.  A spray nozzle
is attached to the robot wrist so that the robot can wash
down the tank wall prior to corrosion inspection.
     The robotic arm weighs approximately 100 Kg (220 Ib)
and has an overall length of 2.5 m (8 ft). The arm has a 1.6 m
(64 in) length to the wrist mounting flange, providing the
                                                         199

-------
robot with a 1.8 m (6 ft) reach when positioning the
inspection camera. The last three joints of the arm, elbow
pitch, wrist roll and wrist pitch, are clustered in close
proximity to provide dexterous manipulation.  All axes are
electrically driven, feature absolute position feedback, and
are actively servoed to hold position. Upon loss of power,
the controller automatically shorts the motor leads to
provide dynamic braking. Gravity will backdrive the arm
into a nearly vertical position so the RTI system can be
removed from the riser in a manual recovery mode. Table 1
provides performance characteristics of the arm.


           Table 1. Performance Specifications
                                        Max
      Description             TraselVelocity
Shoulder Rotate
Shoulder Pitch
Elbow Pitch
Wrist Pitch
Wrist Rotate
±180°
±90°
±120°
±120°
±180°
1.0 rpm
1.0 rpm
2.4 rpm
55 rpm
5.5 rpm
      Reach of Arm          6 feet
      Coordinated End Point Motion
2.5  ips
Key: ips = inches/sec, rpm = rev/minute, ° = degrees


    The five joints of the robot arm are driven by three
different sized actuator packages as specified in Table 2.
The three actuators are similar in concept and design but
provide differing torque and speed characteristics.  The
capabilities of these actuators were optimized to meet the
goal of providing a 15.9 Kg (35 Ib) payload for the robot.
The actuators are designed into a compact, pancake-style
package. In the case of the shoulder pitch it was necessary
to keep the actuator small enough to fit sideways, in profile,
through the 25 cm (10 in) riser. Frameless DC high torque
brush motors were used as they offer the smallest size,
highest torque and lowest speeds available. Each motor is
coupled to a pancake type Harmonic Drive gear reducer,
providing a single step reduction of up to 200:1. These drive
components are integrated with slim line ball bearings and a
resolver to produce compact servo-actuators capable of large
torques.  The integral resolver is directly coupled to the joint
output allowing precise, absolute, servo control of the arm.


Table 2.  Mechanical Characteristics of Actuator Packages

Robot Joints

Shoulder
Rotate&Pitcl

Elbow Pitch
Wrist
Roll & Pitch

Actuator
Size

Heavy

Medium

Light

Dimensions

9.0" dia x 4.5'
351bs
6.5" dia x 3.5'
18Ibs
5.2" dia x 3.0'
81bs
Max
Torque
(in-lbs)

8400

2500

800
Max
Speed
(RPM)

1.1

2.4

55
    The actuators and links are constructed of aluminum,
which is anodized on all exterior surfaces. The actuators are
environmentally sealed to protect them from the
decontamination solution. Since the actuators are not
equipped with brakes, they experience a 100% duty cycle
when the arm is loaded, causing the motors to heat up
significantly. Analysis of the system indicates that the
actuators' capabilities are thermally limited.  That is, the
maximum payload of the arm is dictated by the motors
maximum winding temperature of 155 °C (311 °F) and not by
the maximum mechanical torque of the actuators. To
increase the actuator and arm payload capabilities an air
line is run into the actuators to provide cooling for the
motors. Cabling to each of the joints and tooling is routed
along the I-beam shaped linkages of the arm. Submersible,
molded connectors are provided on each motor.

    D.  Accessories
         Accessories for the RTI robotic arm comprise the
quick change interface, decon spray nozzle, gripper,
overview camera system, sonar sensor, and proximity sensors.
A description of each accessory is provided below:
  • A manual quick change interface is provided at the
    wrist mounting flange to change end effectors
    (inspection video system and gripper).  The interface
    consists of an electrical connector, pneumatic connectors,
    and a common mounting plate.
  • A decontamination spray nozzle is mounted directly
    above the wrist flange  to wash down the tank walls.  It
    has an adjustable flowrate of up to 15 liters (4 gallons)
    per minute.
  • A pneumatic parallel jaw gripper  is provided with a 10
    cm (4 in) stroke and adjustable gripping force of up to 482
    kPa (70 psi).
  • The overview camera system consists of an
    environmentally sealed color camera with a zoom and
    focus lens. The camera is mounted inside a cut-out
    section of the robot shoulder linkage.  A rotary actuator
    provides the ability to pitch the camera along the
    robot arm while zooming in for close views. Remote
    control of the camera, rotary actuator, and light
    intensity is provided at the control console.
  • A miniature sonar detector is used to determine the
    relative orientation of the robot inside the tank.  The
    sonar detector is mounted on the shoulder of the robot
    arm to calibrate shoulder rotation to distance of the
    tank wall.  Since the risers are not located on the center
    line of the tank, radial  extensions  from the riser to the
    tank wall vary in length. An applications software
    package automatically controls the sonar sensing and
    rotation of the shoulder axis. The software processes
    the data to identify the location of the wall closest to
    the riser.  Once distance to the tank wall is known as a
    function of shoulder rotation, distance of the robot's end
    of arm to the tank wall can be calculated based on
    forward kinematics. Distance to the wall is displayed
    on the graphics monitor and also used for software
    collision avoidance.  The accuracy of this  information is
    dependent the combined accuracy of the robot, sonar
    detector, data processing, arm dimensions, and the
    assumed location of the riser.
                                                         200

-------
  •  For impending collision detection, four photoelectric
    proximity sensors are mounted on the leading edge of
    the robot arm linkages to detect close proximity to the
    tank wall.

E.  Strongback
         The Strongback fixture rigidly supports the RTI
robotic system during shipment. It is designed to attach to
the bed of a semi-trailer truck. The Strongback consists of a
tubular framework to cradle and support the full 10.7 m (35
ft) horizontal  length of the RTI system. For additional
protection, the robotic arm is housed inside a reinforced cage
before it is placed onto the Strongback. A facility crane is
used to pivot the RTI robotic system vertical from the
Strongback during deployment at a riser.
IV.  CONTROL CONSOLE

    The RTI control console is the remote station from
which an operator can control and monitor the robotic arm to
perform visual inspection of the tank. The control console
will be located on a desk top inside a trailer located up to
30.5 m (100 ft) from the RTI mechanical system.  The console
consists of two side-by-side 48 cm (19 in) racks which
maximize the useful working and viewing area to the
operator.  The racks are encased in structural foam and
housed together in one self-contained shipping container.  A
removable front cover protects the monitors and control
panels during shipment. All cables enter the control console
through external chassis mounted connectors.
    The control  console is composed of industrial grade
components, rated for operation in indoor, industrial
environments. The inspection and overview camera each
have their own display monitor and camera control panel.
VCR's are provided to record the video output signal of the
cameras.
As shown in Figure 2, the control console displays the
following equipment to the operator:
  •  Operator Control Panel
  •  8-inch Color Monitor to display Overview Camera
  •  9-inch Black & White Monitor to display B&W
    Inspection Camera
  •  Two Super VHS Recorders
  •  Overview Camera Control Panel
    (camera, zoom, pan, & lights)
  •  Inspection Camera Control Panel
    (camera, zoom, pan & tilt, & lights)
  •  Control Panel (B&W Camera focus & iris)
  •  20-inch Color (Video & Graphics) Monitor to display
    inspection cameras or computer graphics
The control console also contains  the following components
within its cabinet:
  •  Intelligent Controller Rack
  •  Servo Amplifier Rack
  •  Power Box
  •  Fan Panels
       Figure 2.  Operator Control Panel (Front View)

     A.  Operator Control Panel
         The operator control panel provides the operator
with a complete interface to drive the RTI system. All
devices and accessories arc operated from the control panel
with the exception of the cameras which have independent
control panels.  The operator control panel is wired directly
to the digital I/O boards of the controller.  The controller
acknowledges operator commands by illuminating activated
switches. The controller performs safety checks of operator
commands before executing them.
     The operator control panel provides switches for speed
selection and jogging of each individual axis. To prevent
accidental activation, each "Axis Jog" toggle switch must bo
held down continuously by the operator to jog the axis. The
axis will move at the selected speed (slow, medium or fast).
Once released, the toggle switch returns to its neutral "off"
position.  In the event of a controller failure, the robot can be
driven in an open-loop mode by hooking up a battery power
supply directly to the motor amplifiers.  An emergency stop
pushbutton is provided on the operator panel.
                                                      201

-------
     The operator must depress a pushbutton to select
coordinated end point motion. A 4-position joystick is
provided to jog the end point of the arm towards or away
from the tank wall and clockwise or counterclockwise along
the tank wall.  Consistent orientation of the end point is
maintained. In coordinated motion control, the Z-axis, wrist
roll and wrist pitch axis jog keys are  also active. Depending
on the orientation of the robot arm, wrist roll and pitch
control the relative pan & tilt of the  inspection camera
mounted at the end point.
     Controls are also provided to open and close the gripper
and to control the decon spray ring and spray nozzle. The
operator control panel provides an up/down arrow and enter
key so the operator can make selections of menu commands
displayed on the graphics monitor.
     B.   Intelligent Controller
         The design of the RTI controller is based on
RedZone's Intelligent Controller for Enhanced Telerobotics,
a proprietary, standardized platform for computation and
communications for the control of a wide variety of multi-
axis robotic systems. The Intelligent Controller is housed
inside a 12-slot VME bus chassis inside the control console.
The Intelligent Controller performs the following functions,
in a multi-tasking environment, for the RTI robotic system:
  •  Translation and execution of all operator commands
     originating from the operator control panel.
  •  Digital servo control of all movement including
     individual axis joint control and coordinated end point
     motion of the robot arm.
  •  Execution of automatic routines; system self check,
     power-up, sonar map control, and shut-down sequences.
  •  Safety monitoring of proximity sensors, joint overtravel,
     joint and velocity tracking errors and overtorque
     conditions.
  •  Continuous monitoring of potential collision states.
  •  Logging significant events in a data file.
  •  Displaying on the graphics monitor: plan view and side
     view of robot arm inside tank, distance and orientation
     of end point to wall, absolute position of each axis, error
     message & diagnostics, and menu prompting of routines.
The computational  devices of the RTI Intelligent Controller
consist of the following boards:
  •  68020 CPU Boards (2) with floating point processors.
  •  RGB Video Board to interface the controller to the
     graphics display monitor.
  •  Resolver to Digital Boards (2) to transform resolver
     output to the digital signal used to compute current
     position and velocity of each axis.
  •  Digital to Analog Board to convert the digital control
     signal generated by the CPU to an analog control signal
     to drive the motor amplifiers.
  • Timer Interface Board to measure time-of-flight of
     sonar echoes generated by the sonar ranging module.
  •  SCSI Interface Board to interface to the removable
     cartridge disk drive.
  •  44 MByte Removable Cartridge Disk Drive to provide
     portability with hard disk performance.  All software
     resides on the disk drive.
  •  Digital Input Boards (2) to provide 64 opto-isolated
     channels that are interrupt driven to the controller.
     Digital I/O serves as primary interface between CPU
     and operator control panel.
  •  Digital Output Board to provide 32 dry reed relay
     outputs. Allows CPU to control devices and indicator
     lights on each switch.
     Control of robot motion is achieved by a control law
implemented in software on the main CPU. Motion control
boards are not required as servo control is flexibly
implemented in software. The CPU reads resolver inputs,
computes forward and inverse kinematics, and generates a
digital control signal.  This digital control signal is then
converted into an analog input to the motor amplifiers.  The
CPU performs all of the control calculations for robot motion,
interprets user commands from the operator control panel,
and maintains the graphics display. Two CPU boards allow
the computational load to be distributed by running the
motion planner on one board, and the remainder of the
software modules on the other. This results in stiffer motion
control and faster updating of the graphics display.


V.   SOFTWARE

     Under separate contract to the Department of Energy
(DOE), RedZone is developing an Intelligent Controller for
Enhanced Telerobotics to provide a standardized, multi-
tasking, VX Works ™ environment for software
development.  The RTI system uses the hardware and
software architecture defined  by the DOE Intelligent
Controller architecture.  All software is written in the C-
language and resides on the disk drive. Figure 3 is a block
diagram of the major software modules of the system. The
software is organized into five main modules: the task
executive, the motion planner, the motion controller, the
data processor, and the graphics module. Communication
between (and in some cases within) these modules is
performed using RedZone's proprietary Robotic
Communications Protocol (RCP) which is the heart of the
Intelligent Controller.  RCP provides both intra-cpu and
inter-cpu communications as well as global variables,
functions calls and semaphores between modules. Below,
each module is described in detail.

    A.   System Control
         The system control module is the "front-end" of the
RTI controller. It contains four sub-modules: digital
input/output drivers, task executive, health monitor, and
data logger.  The digital input and output drivers provide a
standardized software interface to the digital I/O boards.
The task executive's main function is to monitor the state of
the operator panel and of the robot. It directs action based
on these inputs. The data logger records events, errors, and
change of state into a file. The log is maintained on  the
hard disk to help understand and troubleshoot failure or
accident scenarios.
                                                        202

-------
    B.   Motion Planner
         The motion planner module provides a collection of
high level path generating modules, collision detection
modules, and kinematics utilities that operate with a
nominal cycle time of 10 milliseconds. The path generating
modules include joint space profile generation, cartesian
space profile generation, and control for sonar mapping.
Cartesian space points are transformed via inverse
kinematics into joint space goals to generate a smooth
trajectory for each joint in motion. The sonar map utility
automatically  controls the arm while the sonar mapping
sequence is in progress. The collision avoidance module
monitors the proximity of the arm to the tank wall.  The
kinematic module contains the mathematical model of the
arm, including link lengths and axes of rotations. Forward
kinematics are used to compute the end point position of the
arm based on axis joint positions for collision avoidance
checks. Inverse kinematics are used to compute the axis joint
positions necessary to achieve a desired end point position
for coordinated motion control.


Monito
153
A
Operator Console ^i,
f /jran
1
" ' -graphic-
driver

A
graphics

Graphics module
, f.
lOOHz


System am
module
T
1
	 —DID
rol drivi
1
I Task

'
f
-
r

r
n control & status bits
	

Executive L.
Monitor |
Data
I^RfilT



v ^-^ \ ^

collision
avoid.
T


Cart, space Joint space ^
profile profile
generation generation
t „
forward
kinematics



1KHz
inverse
kinematics
/

S S^ Motion planner
/ ^

Soft Limits
contro
^ module
1 Interpolation
^|r
j servo bw
1 •
•driver
J
Arm ^

il
-<

vsolvur
iriver -




Data pnxressor

. 	 "




:>nar Map
Control

i •
sonar
m.i . >inK
t
1

i
r






              Figures. Software Organization
        1. Jog Control. Robot motion is initiated whenever
 the operator holds down an axis jog toggle switch or the
 coordinated motion joystick. The controller responds to the
 switch transition state. An acceleration ramp is
 immediately generated to ramp up to the preselected speed
 range. The motion control module then generates new,
 incrementally small, position goals for the joint every 10
 milliseconds.
        2. Coordinated End Point Motion. The operator's
 primary objective is to position the robot's inspection video
 camera relative to the tank surface.  It is often difficult and
 tedious to position  the end-of-arm while jogging individual
 axes.  To facilitate easier positioning of the camera,
 coordinated end point motion is provided in two axes while
 maintaining a consistent orientation of the tool faceplate:
 horizontal extension of the arm to the tank wall and
 following the curvature of the wall at a constant distance.
 Coordinated motion for the RTI robotic system is constrained
 in the cylindrical world frame of the tank.  Control is
 simplified by requiring the arm to be in a preferred
 orientation.  Should the operator choose to deselect
 coordinated motion and jog in joint mode, a resume function is
 available to allow the operator to return to his former
 position and resume coordinated motion.
        3. Collision Avoidance. The collision avoidance
 software consists of a real-time background program that
 continuously checks the position of the arm to avoid a
 collision with the tank.  The computer checks for penetration
 by the arm into a safety zone that extends from the tanks
 walls and floor.  If the robot enters the safety zone, the
 computer executes an interrupt of the current motion and
 warns the operator of the condition. Once the robot arm is in
 the software collision state, the software only allows the
 operator to jog arm motion away from the tank surface.
 Proximity sensors are also provided to detect an impending
 collision and initiate an emergency stop. A manual override
 button is provided so the operator can override collision
 avoidance so that the RTI can touch the tank  wall or floor.

     C.  Motion Control
         The motion control module reads the joint absolute
 position from the resolver-to-digital driver every
 millisecond. The servo law, an enhanced PID control, uses
 commanded and actual position read from the resolvers to
 calculate a command output to send the power amplifiers.
 Robot motion is controlled in a position controlled mode, not
 a rate controlled mode, as commonly used on robotic
 manipulators. Position control provides stiffer motion
 control with more damping. It also allows an easy upgrade
 to programmed operation at a later date. Execution of the
 motion control task is triggered by a clock interrupt to ensure
 precise timing. The  motion control module also enforces soft
 stop limits and performs linear interpolation on the
commanded positions.

     D.  Sonar Data Processor
          The sonar data processor module reads and
processes the sonar data to map distance to the tank wall as
a function of shoulder rotation.  Radial extensions from the
RTI to the tank wall vary in length, since the RTI system is
inserted through a riser that is offset from the tank center.
The sonar sensor produces a digital pulse each time it is
                                                      203

-------
  fired. The length of the pulse is proportional to the time
  from transmission of the sonar signal to the return of the first
  echo. The sonar driver measures this time-of-flight which
  is converted into distance and recorded in an array with the
  corresponding shoulder rotation angle. The sonar mapping
  module performs pre-processing of the signal to remove
  erroneous data and compensate for the wide beam width of
  the sonar. Signal processing of the sonar signal is performed
  to derive a circular model of the tank from the raw data.

      E.   Graphics Module
          The graphics display on the large color monitor
  provides the operator with a physical sense of the robot
  arm's position inside the waste tank. Objects are portrayed
  as two-dimensional diagrammatic models. A plan view
  shows the orientation of the arm inside the tank and a side
  elevation view shows the robot arm configuration to the
  tank wall. The monitor displays robot joint angles, as well
  as the distance and orientation of the end of the arm to the
  tank. These views and information will greatly enhance the
 operator's efficiency in  operating the robot within the tank.
 The graphics software module continuously reads the current
 position of all axes and uses the kinematic model to compute
 and display the configuration of the arm. The graphics
 display module also provides menu commands, status
 information, and messages to the operator.


 VI.  CONCLUSION

     RedZone Robotics will deliver the RTI robotic system to
 WINCO in April 1990.  The RTI robotic system will then
 become one of the first robotic systems deployed to remotely
 inspect hazardous waste tanks. The initial mission of the
 RTI  will be remote visual inspection of corrosion inside the
 ICPP waste tanks.  WINCO is currently planning  additional
 development of the RTI robotic system including advanced
 tooling to sample the sludge and inspect the bottom of the
 tank, supervisory control to provide enhanced force control of
 the tooling, and a programmed mode of operation.
     The RTI robotic system provides a 15.9 Kg (35 Ib)
 payload, 1.8 m (6 ft) reach, five degree of freedom robotic
 arm that can be inserted through a 25 cm (10 in) diameter
 opening. The vertical deployment unit provides 5.8 m (19 ft)
 of servo controlled extension. The robotic arm can
 manipulate a variety of tools:  inspection viewing systems,
 gripper, spray nozzle, or other specialized end of arm
 tooling. The arm can be flexibly mounted on a variety of
 platforms or even a mobile base. Its compact, high torque,
 electric, servo-controlled actuators can be re-configured with
 different linkages to customize a rcJbotic arm of any
 configuration and degrees of freedom. The RTI robotic system
 is radiation and environmentally hardened to assure
 reliable operation in hazardous environments. The
 Intelligent Controller provides a multi-tasking environment
 to support digital servo control, I/O, collision avoidance,
 sonar mapping, and a graphics display. The controller,
based on the standardized DOE architecture, is extensible to
servo control almost any multiple axis application.  In
conclusion, the RTI robotic system and its components offer an
innovative, standardized, and extensible design with broad
applicability to  remote inspection, decontamination,
servicing, and decommissioning tasks.
REFERENCES

Griebenow, Bret & Martinson, Lori, "Robotic System for
Remote Inspection of Underground Storage Tanks,"
Proceeding of 1990 American Nuclear Society Winter
Meeting. Washington D.C., Nov. 1990.
                                                       204

-------
            AUTOMATED SUBSURFACE MAPPING
                                              Jim Osborn
                                         Field Robotics Center
                                     Carnegie Mellon University
                                         Pittsburgh, PA 15213
                                             412-268-6553
Abstract
Non-invasive imaging of the underground is an essential
component  of hazardous waste  site  investigations,  yet,
despite advances in sensor technology, high quality maps of
the subsurface are difficult to obtain. Subsurface mapping
depends on the  spatial  correlation of individual sensor
measurements taken at multiple locations. Current manual
data collection techniques, however,  are suboptimal  for
precisely positioning subsurface imaging sensors and, in
general, are quite inefficient. Use of the sensors also requires
considerable experience on the operator's part to acquire and
interpret sensor data.  In short,  locating and identifying
buried objects and geological features is a process that relies
heavily on human adeptness and expertise. Thus by applying
automation  and  computer  vision  technologies  to  the
problem, subsurface mapping can be improved.

In our Site Investigation  Robot (SIR) project, prototypical
robots are used to position ground penetrating radar (GPR)
equipment with  the  accuracy needed to  generate three
dimensional subsurface maps. Estimating its site location by
a combination of dead reckoning and inertial measurements,
a rough terrain mobile robot deploys a gantry mechanism to
scan the ground  with the  GPR  antenna. Radar data are
digitized and stored in three dimensional arrays for spatial
correlation and image enhancement on  a color graphics
workstation. We have also applied basic image processing
and visualization techniques to assist in the interpretation of
these subsurface maps. Control of the robots and access to
the software  are through user-friendly interfaces, which
facilitate the subsurface mapping process.

Introduction
For  years,  robotics  and  automation  have  increased
productivity   in   manufacturing   industries    through
standardization and repeatability. Core robotic technologies
have now progressed to the point that robots are moving into
the field and offering similar benefits performing  tasks in
unstructured settings. One class of these field robots is
emerging to meet one of the most important challenges now
facing the world: the clean up of hazardous waste sites.

One of the cost drivers in remediation of a site is the lack of
information about  the site  itself.  A detailed and costly
investigation is required to develop a knowledge base of site
geology, hydrology, chemistry, the extent of contamination,
etc., that can be used to select appropriate  remediation
technologies and effectively plan the cleanup effort. Much of
this expense can be attributed to inefficiencies in manual
data acquisition techniques, lack of standard data collection
procedures, and the cost of insuring and protecting  the
personnel who conduct the investigation. As an alternative,
automation offers the prospect to collect large quantities of
data in a form that supports more complete assessments and
at a significantly lower cost.
                                                     205

-------
 Most investigations include efforts to locate buried objects
 that are potential sources of contamination (such as drums),
 identify and measure the extent of contaminant plumes, and
 determine the morphology of geological formations that
 affect pollutant  migration. Commonly used methods to
 generate such information include resistivity measurements,
 acoustic techniques and ground penetrating radar.  While
 each  has  unique  advantages,  no  single method  alone
 provides complete information, and all have limited utility
 owing to  the inaccuracies and  inefficiencies of manual
 sensor deployment.  Ideally, the data resulting  from the
 application of these non-invasive techniques can be used to
 construct  an  accurate  graphical  representation  of the
 geometry of buried structures - a map of the subsurface.

 In this  paper we present the Site  Investigation Robot, a
 system  for  automated subsurface  mapping with ground
 penetrating radar (GPR), as one aspect of a program to
 automate hazardous waste  site  characterization. The Site
 Investigation Robot is a mobile robot that collects and
 spatially registers GPR data and recovers  them to its base
 station where they are correlated, enhanced and displayed so
 that  inferences about the shape and location of buried
 structures can be made.  This program's broader goal is to
 develop robotic systems to make the data acquisition process
 faster and  more  complete and  to  apply advanced data
 processing  techniques  that will make these data  more
 accessible and easier to interpret.

 System Overview
The  Site  Investigation  Robot  consists  of a robot  and
 controller, data acquisition system, and a body of subsurface
mapping software  to manage, process and  visualize data
collected  during  investigation  missions.  The  present
configurations of these  subsystems  are described below;
 future enhancements planned for each are  described in the
section that follows.

 Robot
The Site Investigation Robot prototype is pictured in Rgure
 1. We  have employed an existing mobile robot, the
Terregator  (short for terrestrial  navigator), a driveriess,
outdoor  vehicle   built  for  autonomous  driving  and
exploration research, for the data acquisition aspect of this
project.  Terregator  is  a  rugged,  six-wheel,  skid-steer
locomotor scaled and powered  to  negotiate moderately
rough terrain and steep slopes.

On both the right and left sides of the base locomotor, three
wheels are linked together with chains and driven by a low-
speed, DC  motor through a harmonic gear unit.  This
drivetrain, in conjunction with  off road floatation tires, gives
Terregator excellent tractive  characteristics to  overcome
obstacles and grades. For position feedback, each motor is
coupled to an incremental rotary encoder. Theoretically, this
arrangement gives  the  Terregator open loop  positional
accuracy in the  sub-millimeter range;  in practice, tire
deflections,  vehicle/ground surface interaction  and  other
non-linearities limit Terregator's dead-reckoning ability  to
distances on the order of centimeters.

To position subsurface imaging sensors, a single-axis gantry
mechanism is attached to Terregator's frame forward of the
generator such that the direction of motion is perpendicular
to the mobile  robot's path. The mechanism consists of a
buggy that is pulled along parallel fiberglass T-beams by a
chain belt driven by a DC motor. The GPR antennas are
suspended from  the buggy with threaded rods for height
adjustment. A rotary encoder directly coupled to the motor
allows the antenna  to be positioned accurately to one
centimeter over the entire two-meter length of the gantry.
Limit switches  at each  end  of the gantry  ensure safe
operation and  provide a convenient  way  to  identify the
antenna's limits of travel.

A 3kW, 120 VAC gasoline generator and ventilated, shock-
isolated electronics enclosure are mounted atop Terregator's
base to provide  power for the locomotion, computation,
sensing and communications. Raw generator output is tied in
to the base locomotor's 90 VDC power supply; the generator
output is also conditioned by an uninteruptible power supply
(UPS)  for  more sensitive devices,  including  telemetry
equipment, onboard computers and disk drives, safety logic,
sensors  and interface electronics. Substantial  auxiliary
power is  available for mission specific payloads, such  as
GPR equipment.
                                                        206

-------
At the heart of the Terregator is a VMEbus card cage that
bouses a 68020 CPU card with 4 Mbyte onboard memory,
SCSI and ethernet ports. The system CPU functions as a
multi-tasking   controller,   coordinating  and  sequencing
locomotion  and  gantry  motions, GPR data acquisition,
communications  with  the   base  station  and  system
monitoring functions. Other boards in the card cage include
a serial interface card, two 2-axis motion control cards, and
a sensor interface card with eight channels of analog-to-
digital (A/D) conversion, four channels of digital-to-analog
(D/A) conversion and 16 bits of digital I/O.  All connections
to these boards are made through an intermediate patch panel
that facilitates  the  addition  of new  sensors  and other
peripherals to the basic system. For development purposes, a
single board workstation  and  disk are  located  on the
equipment deck above  the  electronics  enclosure and
interfaced to Terregator's CPU via an ethernet cable. The
organization of these components is shown graphically in
Rgure 2.

Controller
The Site Investigation Robot is intended for use by persons
who are  much better  versed in the  practices of  field
screening, data collection  and analysis than they are in
operating a robot. It is thus essential to hide the complexities
of controlling the robot from its users and make interactions
with the SIR as simple and straightforward as possible. This
motivated us to develop a control architecture that allows
SIR users to command and monitor the robot at a high level
while masking  the details of implementing expressed user
intentions.

The SIR command interface presents the user with a set of 2-
Dsurface maps of the site, that show the size, spatial location
and orientation of boundaries, known man-made  structures
(e.g., buildings and roads) and natural features (e.g., trees
and surface water bodies) in a consistent, user defined site
coordinate system. These maps are  created with a simple
CAD package, developed specifically for this purpose, at the
outset of a site investigation, and can be updated and edited
as the investigation proceeds. To initiate a  data acquisition
ran, the  user first displays a map of the site on the base
station computer by recalling a file that contains a CAD
description of a particular region of interest. Site boundaries
are indicated by straight  line  segments while all known
objects and other obstacles to the mobile robot are shown as
polygons. Using the computer mouse, the user then draws a
bounding box (a rectangle that encloses part  of the map)
around the area of the site from which data is to be collected.

A set of routines to plan a path that covers all of the obstacle-
free ground  surface  within the  bounding box are  then
invoked. First, the dimensions of the bounding box and all
obstacles it  contains  are adjusted  using  dimensional
parameters of the SIR. In this algorithm, the robot's effective
turning radius is calculated by finding a circle within which
all parts  of the skid steered locomotor  will remain when  it
turns in place. All sides of the bounding box and all included
polygonal obstacles are 'grown' by an  amount equal to the
radius of that circle. Should the transformed bounding box
be found to intersect a site boundary, which is a pathological
case for the current path planner, the initial bounding box is
rejected  arid the user  instructed  to redraw  it. Once an
acceptable bounding box is found, the robot can be modelled
as a single point travelling through a more constricted space,
thus simplifying subsequent path planning.

Planning paths for the Site Investigation Robot is a departure
from traditional mobile robot path planning in the objective
is to cover as much of the ground surface as possible, rather
than finding the shortest route between two points. The SIR
path  planning  problem is constrained by  the mobility
characteristics of the Terregator mobile robot. Terregator
can  faithfully execute straight line motions  of specified
length by dead reckoning, in which the wheel encoders are
used to measure distance travelled; it can also make accurate
turns in place,  using  a gyroscope to measure  the angle of
rotation.  However, the indeterminacy of Terregator's skid
steering  makes  following  an  arc  of  specified curvature
difficult even on hard, flat surfaces. For this reason, we have
limited all driving to  straight line motions  and point turns.
This  is  acceptable  given  the data acquisition protocol
described below.

SIR's path planner examines the resulting free space in the
transformed bounding box and  finds a way to cover it such
                                                        207

-------
that the number of turns are minimized. If obstacles are
present  the user selected  area  is  divided into smaller
obstacle-free areas, and a path is planned for each. Since
there are often multiple ways to perform the subdivision,
solutions are not always unique. Furthermore, there is no
way to guarantee that the resulting path is optimal. However,
once a path is  found,  it is overlaid on the site map for
validation.  This affords the user the opportunity to draw
smaller bounding boxes and specify point-to-point moves
that connect the subregions of the map.

The final path description is translated into a sequence of
driving commands (straight lines and rotations) that are
placed in a queue and transmitted to the robot via a wireless
modem.  Using a  software joystick,  the robot is  then
leleoperated to its starting point and set on its route. While
driving, the robot transmits its location back to the base
station which is displayed as an icon on the site map. Other
status information is similarly relayed so that the user can
supervise the data acquisition mission.

Subsurface Mapping Software
The  Site  Investigation Robot  deploys and  supports  a
commercial ground  penetrating radar set  (Geophysical
Survey System, Inc. SBR.-3) to acquire subsurface data. A
data acquisition run  is comprised of  combinations  of
Terregator drive motions and gantry movements in which
the basic procedure is to move the antenna from one limit to
the other and then drive forward some incremental distance.
At regular intervals through the antenna's travel, a series of
radar pulses are transmitted into the ground and the energy
reflected  to the receiving antenna amplified, filtered  and
digitized. These signals are stored adjacently in a buffer until
the antenna has completed a full  scan. The result is a two
dimensional data array, in which the columns are individual
GPR waveforms, stored on disk as an image along with the
mobile  robot's  site coordinates. More  details on  the
principles of GPR are presented in the Appendix.

Every row of pixels in the GPR image contains data acquired
at a constant time delay relative to the transmitted pulse. That
time delay is converted into a distance from the antenna by
the speed  of  electromagnetic wave propagation  in  the
imaged subsurface media based on measured and/or inferred
electrical parameters. Since the position of the mobile robot
and the position of the antenna relative to the mobile robot
are measured  for  every  recorded  GPR waveform, it  is
possible to assign three spatial coordinates to each pixel in
the image. It is this position tagging that makes it possible to
spatially  correlate  and  visualize  GPR data  in three
dimensions.

Each  recorded waveform spans  a depth  range  that  is
governed by the wavelength of the transmitted energy and
the electrical properties of the subsurface medium. Generally
speaking, there is a trade-off in depth of penetration and the
physical dimensions that can be resolved.  The 500 MHz
antenna used in this work can image structures buried to
depths of 3 meters with 5-10 cm resolution in the best of
conditions  (e.g.  dry,  sandy  soils);  lower  frequencies
penetrate  deeper  at the  sacrifice  of  resolution. GPR
performance is poorer in materials with high conductivity
and high dielectric constant - conditions associated with high
moisture content - due to attenuation of the radar energy. In
saturated soils and clays, imaging potential may be limited to
depths of only one meter,

This data acquisition procedure is repeated until the robot
has covered its entire planned route. Once the robot returns
to its base station, all acquired images are transferred from its
onboard disk to mass storage devices connected to the base
station computer for archiving and processing. Acquired
GPR data are arranged in volumes, each containing a set of
parallel subsurface sections stored as images. Individual
sections are stored as files that also contain other parameters,
including location of the scan, date and time  of acquisition,
and radar gain and  time base settings. These files  are
organized in a Unix file system such that each subdirectory
corresponds to a unique site volume. Each subdirectory also
contains an additional site index file that is used to retrieve
and store individual images. Figure 3 shows an example of a
site map from which nine volumes of the subsurface would
be scanned.

Since the intuition of experienced field screening personnel
is still required to apply the appropriate processing steps and
                                                         208

-------
choose parameter values to make sense of the images, we
have developed a set of programs to process  GPR data
acquired by the Site Investigation Robot that are called by
the user through a common menu-driven interface. This
software package, known as gpr-shell, includes routines for
reading and writing data files, applying time domain filters
to individual records, displaying of 2D subsurface sections
as  color or gray-scale  images,  scaling and windowing
images, spatial correlation all GPR records in a subsurface
volume, and a variety of image enhancement functions. To
facilitate processing, Gpr-shell also provides command line
completion, prompting, and on-line help. It also provides the
user with an 'on-line lab notebook', in which the steps and
parameters used to process each  image are  automatically
recorded for future reference.

In  order to transform  raw  GPR data  scans  into  high
resolution  images,  several  processing  steps have  been
implemented, as illustrated hi Rgure  4. (We have yet to
identify a single methodology or set of parameters that can
be  successfully   employed  to   generate  interpretable
subsurface maps from all GPR data, however, the following
steps are generally taken.) First the signal is deconvolved
with the return signal from a pulse transmitted into air.
Deconvolution is a matched filter operation that removes the
effects of the secondary pulses from the return signal and
effectively  transforms a  return from the transmitted pulse
into the return  that would have been caused by an  ideal
impulse function. The resulting signal  is then low pass
filtered to  remove noise  components  introduced by the
deconvolution.

The waveform  recorded at each  grid point is  actually  a
composite  of all radar  reflections within the  antenna's
conical beam pattern due to the poor focusing of the GPR
antenna. However, since the spacing between surface grid
points is accurately measured, we are able to correlate all of
the measurements and synthetically focus the antenna. A
process known  as 'migration' is applied to convert the
deconvolved and filtered data into a representation of the
subsurface. Migration is  very  similar to the synthetic
aperture focusing techniques used for high resolution pipe
location, in that its underlying principle is data from adjacent
scans tend to reinforce one another.

A three dimensional array of GPR data is constructed by
sampling data from vertical sections in the scanned volume.
The value in each cell, or voxel (for volume element), is then
added to all array locations equidistant from the transmitter
and within the antenna beam. This effectively 'spreads' each
part of the return signal over surface that is a locus of points
with the same time of flight from the antenna. By applying
this algorithm  cell  in the  array,  the  recorded signals
originally associated with individual voxels constructively
interfere with one another. This reinforcement indicates the
presence of an impedance discontinuity at the corresponding
subsurface  location and  emerges in the  migrated  image.
Migration  can  thus be  used to effectively  focus  the
transmitted radar beam. (We note, however, that its success
requires a good estimate of the  soil's dielectric constant,
which determines the speed at which GPR waves  travel
through the subsurface, and  the antenna beam pattern  and
soil conductivity, both of which influence attenuation.)

Once a volume of data has undergone 3-D migration, vertical
and horizontal sections are extracted from it as  individual
images. These images  are then enhanced by  a number of
image processing operations, including 2D low- and high-
pass filters of varied bandwidths, edge detectors and region
growing operators,  depending on the  image features of
interest.

Figure 5 through 7 show the results of these processing steps.
All three are images of a small  metallic drum containing
water buried in sand. Figure 5 is a vertical section of raw data
and Figure 6 is the  same image after deconvolution  and
migration. In this case, the barrel cross section is best seen by
the thresholding of the image after it is finally processed by
the 2-D  high pass filter (Figure 7).

Future Enhancements
A number of enhancements to  our current system  are
planned to increase its ability to operate on waste sites, ease
its use,  and improve the quality of the subsurface maps it
generates.
                                                      209

-------
For sites with very rough terrain and/or numerous obstacles,
improving the mobility of the base locomotor will result in a
greater percentage of ground surface that SIR can cover.
This can be accomplished with suspension, greater ground
clearance, replacing the wheels with tracks, etc. An even
more significant performance increase can be realized by
improving SIR's position  cognizance, regardless of its
mobility characteristics. The most promising technologies to
provide a more accurate measurement of the robot's location
on the site are inertial navigation units (INS) and global
positioning (GPS) receivers, both of which can be deployed
onboard  and readily interfaced to the robot controller. By
providing a position estimate  that is independent of the
robot's dead reckoning, the robot can be navigated with a
closed loop path tracking control  scheme, a paradigm in
which the robot's actual (measured) position is used to
correct for deviations from the planned path that may result
from wheel slippage or other controller disturbances. Path
tracking  control using combined  INS  and  GPS  has
successfully guided our NavLab mobile robot at speeds
exceeding 20 km/hr, more recently, the same controller has
been ported to an off-road dump truck.

More  accurate  GPR  antenna  positioning  can  also  be
achieved by replacing the gantry mechanism with a tnulti-
degree-of-freedom robot arm. Our concept for such a sensor
deployment arm (SDA) is a long reach mechanism able to
position and orient sensor payloads weighing up to 10 kg.
over a 2 meter x 2 meter area, adjusting to any undulations
of the terrain. The principal advantage of an SDA is greater
integrity of the sensor position  measurements - complete a
3D data array can be  collected with a common frame of
reference, eliminating  the possibility of positioning errors
between adjacent scans due to motions of the mobile base,
which arc typically an order of magnitude less accurate than
manipulator movements.

There appears to be a synergy between the Site Investigation
Robot and geographical information systems (GIS), another
emerging   technology for  waste   site   investigations.
Geographical information systems are software tools for
cataloging; manipulating and displaying any form of data
that can be related to a cartographic map. GIS applications
include land use management, record keeping of legal
boundaries,  roads  and utility  networks, agriculture,  and
many others. A GIS can be also linked to a relational data
base to provide a powerful tool for site investigation. Many
available GIS packages include routines to enter previously
digitized terrain maps and survey data which would aid in
the development of site maps for the SIR user interface. The
other attractive feature of GIS is simplified storage  and
retrieval of data: entry of acquired position tagged data into
the GIS data base can be automated and its recall reduced to
the simple positioning of a cursor in the display window.

Two advances in subsurface mapping software are currently
being pursued. One is the development of more general three
dimensional migration algorithms that will account for the
non-homogeneous nature of the subsurface  nature of the
subsurface medium. This will entail assigning permittivity
and conductivity  values  to  each voxel in the scanned
subsurface volume in order to better model GPR wave
propagation.  Techniques  to  measure and/or infer these
parameters will have to be developed to make the best use of
this algorithm. In addition faster processing engines  and
techniques will be required to achieve results in useful time
frames. The second advancement will be the application of
three dimensional enhancement and rendering techniques to
subsurface maps. Such techniques exist in the domains of
medical imaging and geological exploration, but have yet to
be adopted for GPR.

Finally, our goal is to integrate these hardware and software
elements into the more complete system for waste  site
characterization, as shown in Figure 8.

Summary
Subsurface mapping is a discipline that has advantageously
adopted technologies from the domains of robotics and
computer science. In this research, we have successfully
implemented registration of sensor position and automated
acquisition of sensor data using a robot, and thereby created
opportunities to apply processing techniques  to create 2-D
and 3-D subsurface maps of higher quality than previously
attainable. This and other spatially correlated information
that the Site Investigation Robot generates can be used to
                                                        210

-------
more effectively characterize waste  sites  and ultimately
lower the expense of site cleanups.

More generally, robotics and automation can benefit waste
site characterization in a number of ways.

   •   The enormous data requirements will be satisfied
      faster and at lower cost when data arc acquired by
      robots.
   •   The quality  of those data will be enhanced
      through standardized, repeatable measurement
      techniques.
   •   By automatically indexing measurements by
      position in a geographical information system,
      opportunities for numerical modeling, graphical
      visualization and straightforward data correlation
      are created.
The Site Investigation Robot is an example of an emerging
class of robots dedicated to the solution of hazardous waste
problems. We view the SIR be the first in a family of robots
for environmental  applications.  Systems that  follow will
have additional perceptive capabilities and self-reliance to
perform detailed site assessments.

Acknowledgments
This research is sponsored through a cooperative agreement
with the U.S. Environmental Protection Agency and a grant
from  the Ben Franklin Technology Center  of Western
Pennsylvania. We  also acknowledge RedZone  Robotics,
Inc., for  its  participation in the Site Investigation  Robot
project.

Appendix: Principles of GPR Sensing
Ground  penetrating  radar  works  by  transmitting  an
electromagnetic  pulse  into the earth  which spreads as a
conical wavefront  as it travels further from the antenna.
When  the radar wave reaches a discontinuity  in electrical
impedance of the  subsurface, an echo is relumed, the
strength and the phase of which indicate the magnitude and
sign of the  change. Mathematical descriptions of these
interactions in all but the simplest of cases defy  closed form
solutions; even finite element methods are too cumbersome
for practical modeling of the GPR phenomenon. Fortunately,
modeling the physics using geometrical optics can produce
meaningful results. With this simplification, the transmitting
antenna is treated as a light source from which rays emanate
and are reflected to the receiving antenna. The distance to the
point of reflection (assuming a direct reflection) can thus be
estimated with time-of-flight measurements, i.e., the latency
of the echo relative to.tbe transmitted pulse.

A difficulty  with the optical assumption  is the  poorly
focused radar beam. Commercially available GPR antennas
are designed to limit beam spread of the transmitted wave to
an elliptical cone, however, for  a single return, the exact
location of an echo within this volume cannot be determined.
To resolve this ambiguity, the antenna is scanned in a line
over the ground surface to create  an ensemble of return
signals. Latency of echoes are lowest when the antenna is
directly over an object and increase as the antenna  moves
away. By combining recorded echoes from points along the
scan line, distinctive curves are  generated which are then
interpretedby GPR experts to identify subsurface features.

In practice, there are several factors that complicate the radar
return. Time of flight measurements on return echoes depend
on knowledge of the propagation velocity of the transmitted
pulse,  which is not a constant but instead depends on
electrical permittivity (or equivalently, dielectric constant)
of the subsurface material. This introduces uncertainty in the
measurements,  which  is  currently  resolved  either  by
calibration in the field or simply by estimation of subsurface
permittivity. Both the transmitted and reflected radar waves
are attenuated due to losses in the media that are governed
primarily by  its conductivity, another parameter requiring
estimation. Geometric dilution of the wave energy  as  the
beam  spreads with   distance  travelled  is  a  further
complication since the exact  shape of the antenna  beam
pattern within the subsurface medium cannot be determined.
Finally,  the difficulties of controlling the shape of  the
transmitted  pulse  at  GPR   operating frequencies (one
hundred  megahertz  to over one  gigahertz)  introduce
additional return signals that confuse the main return echo
and must be removed.
                                                      211

-------
             Figure 1. Site Investigation Robot prototype
                              VME Bus Computer
•:


Tirn^-z!
Js.







•n~^on
\ \ 4UBP | —
Optical Disk






1






i_


















o


_

















g
6
e
o
1




















-------
              PARKING LOT SUE MAP
    ildjZ

test pa

test pa

test pa

ch 3

:ctiZ

;ch 1
leslpa

test pa

test pa

ch4
\
ch5
I
:ch6
1

test pa

est pa

le§tpa

ch 9

ch 8

:ch 7
           Pzxking Cuib
                                                  Parking Cuib
figure 3. Site map with nine scanned subsurface volumes
Data Acquisition
   Rgure 4. Ground penetrating radar processing steps
                          213

-------
Time
(IIS)
0.0
1.0
2.0
3.0
4.0
5.0
:,.'•
7.0
BjO
JjQ
10.0
11.0
12.0
13.0
14.0
15.0
16.0
17.0
18.0
19.0
Depth
(mm)
0.0
74.9
149.9
224.8
299.8
374.7
449.7
524.6
599.6
674.5
749.5
824.4
899.4
9743'
1049.3'
1124.2
1199.2'
1274.1'
1349.1'
1424.0'
              < \iliu-M;ip: -I.W.I i	
              Volume: icsl volume I   Slice Niimher:  5  (400 mm)
              Antenna: 3102 I-rcq: 500 MH/. Pcnn: 4.00  Time Slcp: 50 ps.
          X (mm)
           0     149    299   449    599   749   899   10-19  1199   1349   I49S   I64K   I79S  194S
           I     I    I     I    I     I     I     I     I     I     I     I     I
            Figure 5. Vertical section of burled drum (raw GPR data)
Tune Ucplh
 UK!  (nun)
      0.0
      74.9
      149.9
      224.8
      299.8
      374.7
      449.7
      524.6
      599.6
      674.5
      749.5
      24.4
 12.0  899.4
 13.0  974.3
 14.0 1049.3
 15.0 1124.2
 16.0 1199.2
 17.0 1274.1
 18.0 1349.1
 19.0 1424.0
Figure 6
:
:

i
;
•
7.0
-


::
             Volume: Icsl volume 1   Slice Number  r>  (4()()mm)
             Aiuaina: 3H)2  l-'req: 500 MH/. Perm: 4.00  Time Step: SO ps.
         X (mm)
          0     149   299   449    599   749   899   1049   1199   1349  1498   1648  1798   1948
                    I     I      I     I     I     I     I      I     I    I     I     I
                                                                                 -
          Image of buried drum (Figure 5) after deconvolution and migration
                                          214

-------
                             ,		                i  7;i IS.'I
                Volume: lesl volume 1   Slice Number:  5  (400mm)
                Antenna: 3102 Freq: 500 MHx, Perm: 4.00  Time Step: 50 ps.
Figure 7. Drum image from Figure 6 following 2-D high pass filter and thresholding
                                      215

-------
                                              CITE IK'
                                              Motion planner
                                              Motion controller
                                              Site posrtion
                                              Data acquisition
                                              Self monitoring
                                                                             SIR System Architecture
                             :  .v-fsa

                          Display site map
                          Display acquired data regions
                          Select acquired data
                          Overlay scalar data arrays
                          Show robot posrtion and path
                          Select robot path
                          ooooooooooo
                          OODOOOOOOQO
                          CO OOOOOO O O O
                          O O O L      ^Q o D
Position and time tagged :
Sensor specific data
Site maps
Data annotation
Processing results
*B-*s
— -
KAal FILE SYSTEM
Variable length GPR data
Processing history
Processing results
VHSU&US&TOfi
2D Radar display
3D GPR 'stacks'
Solid visualization
Depth map displays

•:7'- 	 ._••..-: .
Deconvolution
2D migration
3D migration
Depth mapping
Parameter Estimation
Bd&QE PHQCESSIH3
Thresholding
Segmentation
Frequency analysis
Rltering
Object tagging
                                       Figure 8. Site Investigation Robot system architecture
                                                            DISCUSSION
BRIAN PIERCE: My first question has to do with using ground penetrating
radar, as just one example, or using a magnetometer as another type of sensing
device. And the second question has to do with the use of a pair of robots or a team
where you could take advantage of forward scattering using the ground penetrating
radar. Right now it seems to me you're just using back scattering in a monostatic
configuration.

JAMES OSBORN: That's certainly correct. If you recall the viewgraph that
Ann put up. they are actually going to pursue the magnetonietry type of sensing.
In fact, there is really a whole class of sensors that can be put on it. Each one has
unique requirements. In particular, some of the magnetic techniques can't be near
these very metallic robots. So you've got to come up with long deployment
booms. The idea of doing bistatic  radar soundings is an interesting one.  I can
think of a couple ways that do that. One is to have a multiple arm system on a
single mobile base. And the other is to actually go w ith two mobile bases. I would.
at this time, say the preferred way would be the former (two arms) because of the
ability to register a manipulator and/or affect your position with much higher
accuracy than you could a mobile robot.
            CHRISTOPHER FROMME: There are some excellent available technologies
            for registering line of sight over short ranges, like the distance between the two
            of us right now. So the idea of a pair of robots working in unison and precision
            may have some merit.

            DOUGLAS LEMON: Is this technology resident in the university or is it in the
            RedZone Robotics Company, and who has funded this?

            CHRISTOPHER FROMME: The project isfunded by EPA.And the technology
            is currently in the university, although we have had some collaboration from
            RedZone. in particular  to turn out that robot controller that drives the system. So
            we are getting some collaboration from RedZone. but the project is resident at
            CMU.

            DOUGLAS  LEMON: Do  you expect  this technology to  eventually be
            commercially available? Is that where you're headed?

            CHRISTOPHER FROMME: Yes. If not, then it doesn't make any sense to do
            it.
                                                                       216

-------
       A QUALITY ASSURANCE SAMPLING PLAN FOR EMERGENCY
                           RESPONSE (QASPER)
John M. Mateo, Quality Assurance Officer
and Christine M. Andreas, Assistant
Quality Assurance Officer, Roy F. Weston,
Inc.^REAC, GSA  Raritan Depot, 2890
Woodbridge Avenue, Building 209 Annex,
Edison, NJ, 08837-3679
William Coakley, Quality Assurance
Coordinator, USEPA, Environmental
Response Team, GSA Raritan Depot,
2890 Woodbridge Avenue, Building 18,
Edison, NJ, 08837
Abstract

Integration of critical elements into a com-
prehensive Quality Assurance Sampling
Plan (QASP) is crucial to implementation
of an effective plan.  How can a project
manager ensure consideration of all these
elements?  Utilizing a software  package
called QASPER, a project manager is
prompted to consider elements necessary
to generate a comprehensive Quality As-
surance Sampling Plan  for Emergency
Response.

QASPER is a PC-based software  package
which compiles generic text and  user
provided, site-specific information into a
draft QA/QC Sampling Plan  for the
Removal Program.  QASPER addresses
the nine sections of a QA/QC Sampling
Plan, as specified in OSWER Directive
9360.4-01,  Removal Program QA/QC In-
terim Guidance,  Sampling QA/QC Plan,
and Data Validation Procedures  (revised
April, 1990). Sections include: Initial data,
background information,  data use objec-
tives, QA objectives, approach and sam-
pling methodologies, project organiza-
tion  and  responsibilities,  QA
requirements, deliverables, and data
validation.

QASPER was created to facilitate the
timely assembly of a comprehensive sam-
pling plan for emergency response ac-
tions.  By thorough consideration and
attention to the necessary requirements
of QA/QC sample  planning through an
automated process, it is anticipated that
reliable, accurate and quality data can be
generated to meet the intended use.

The On-Scene Coordinators (OSC) or
the Technical Assistance Team  (TAT)
contractors are the primary users of
QASPER. These  individuals will have
access to the site specific information and
the  sampling  objectives which charac-
terize  a particular  hazardous waste site
investigation.  They are also responsible
for assembling the information into an
acceptable plan for implementation.
                                     217

-------
The system, however, is applicable to many
regulatory programs that require the com-
pletion of QASPs.

Features of QASPER are  numerous.
QASPER  is self contained, no  other
software is required  for support.  ASCII
outputs are generated so that files may be
uploaded to other word processing pack-
ages for further manipulation. Database
files on  all previous sampling plans are
retained. Consistency and comprehensive-
ness of sampling plan creation efforts are
maintained throughout office, region or
zone, therefore, sampling plans are created
more efficiently. Redundant data entry is
minimized by integrating repetitive infor-
mation throughout the plan after one entry.
The user is provided access to standardized
generic text with the capability to overwrite
and edit. QASPER allows for flexible data
entry throughout the plan. QASPER runs
on an IBM PC or 100% compatible, with a
hard drive, 640K RAM and a  printer (for
hardcopy output).

Introduction

The U.S. Environmental Protection Agen-
cy  (EPA) has divided the  Superfund
cleanup program into short-term and long-
term remedial activities. Short-term inves-
tigative and mitigative efforts,  typically
addressing imminent threat, are referred to
as "Emergency  Response Actions" under
EPA's  Removal Program. To ensure ade-
quate and comprehensive response, suffi-
cient time must be allocated for thorough
planning; however, planning  is often
regarded as a  luxury in an  emergency
response scenario.

The EPA has taken a number of steps to
establish planning criteria for emergency
response actions which are sufficiently
detailed to ensure that data generated will
be of known quality to serve its intended
purpose and are commensurate with the
emergency response timeframes.  The
first of these steps was the establishment
of data quality objectives (DQOs) for the
Removal Program.  Second, EPA also
established a minimum framework for an
acceptable Quality Assurance Sampling
Plan. Both of these guidelines are set
forth in OSWER  Directive 9360.4-01
released April  1990 (Publication No.
EPA/540G-90/004).

This paper will describe the Removal
Program DQO's, define the framework
of the QASP, and  describe a third, in-
novative step EPA has taken in creating
a software package which facilitates the
timely assembly of both into a com-
prehensive plan ready for implementa-
tion in an emergency  response.  The
majority of this paper will describe the
features of the software program.

Removal Program Data Quality Objec-
tives

The quality of data is determined by its
accuracy  and  precision  against
prescribed requirements or specifica-
tions, and by its usefulness hi assisting the
user to make a decision or answer a ques-
tion with confidence. OSWER Directive
9360.4-01 guides the user in defining data
quality within a framework that also in-
corporates the intended use of the data.
The guidance is structured around three
quality assurance objectives, each as-
sociated with a list of minimum require-
ments.  The three QA Objectives,
hereafter referred to as QA1, QA2 and
QA3 are described as follows:

QA1 is a screening objective to afford a
quick, preliminary assessment of site
contamination. This objective for  data
                                        218

-------
quality is for data collection activities that
involve rapid, non-rigorous  methods of
analysis and quality assurance.  These
methods are used to make quick, prelimi-
nary assessments of types and levels of pol-
lutants.  The primary  purpose for this
objective is to allow for the collection of the
greatest amount of data with the least ex-
penditure of time and money.  The user
should be aware that data collected for this
objective have neither definitive identifica-
tion of pollutants nor definitive  quantita-
tion of their concentration level.

QA2  is a verification objective used to
verify analytical  (field or lab) results. A
minimum of 10% verification of results is
required. This objective for data quality is
for data collection activities that require
qualitative and/or quantitative verification
of a "select portion of sample  findings"
(10% or more) that were acquired using
non-rigorous methods of analysis  and
quality assurance. This quality objective is
intended to give the decision-maker a level
of confidence for a select portion of
preliminary data. This objective allows the
user to focus on specific pollutants and
specific levels of concentration quickly, by
using field screening methods and verifying
at least 10% by more rigorous analytical
methods and quality assurance. The results
of the 10% of substantiated data gives an
associated sense  of confidence for the
remaining 90%.   However,  QA2 is not
limited to only verifying screened data. The
QA2 objective is also applicable to data that
are generated by any method which satisfies
all the QA2 requirements, and thereby in-
corporates any one or a combination of the
three verification requirements.

QA3 is a definitive objective used to assess
the accuracy of the concentration level as
well as the identity of the analyte(s) of in-
terest.  This objective for data quality is
available for data collection activities
that require a high degree of qualitative
and quantitative accuracy of all findings
using rigorous methods of analysis and
quality assurance for "critical samples"
(i.e., those samples for which the data are
considered  essential in making a
decision). Only those methods that are
analyte specific can be  used for this
quality objective.  Error determinations
are made for all analytes of the critical
sample(s) of interest.

Quality Assurance Sampling  Plan
Framework

There are nine sections to a Removal
Program QA Sampling Plan. Section 0.0
addresses basic information require-
ments such as site name, relevant work
order numbers, primary personnel
names and titles, etc. Section 1.0 solicits
information about the location of the
facility, type of facility, type and volume
of materials to be addressed, sensitive
adjacent environments, and action
levels. Section 2.0 addresses data quality
objectives (DQOs), i.e., regarding
decisions the data will support. Section
3.0 addresses the linkage of DQOs with
matrix and parameters.  The project
manager must decide which parameter
will be assessed, by matrix, for which in-
tended data use, at which QA objective
(QA1, QA2, or QA3).  Section 4.0 ad-
dressed the Sampling Approach and
Methodologies, including documenta-
tion requirements. This section will in-
clude a discussion of sampling design,
type  of equipment, fabrication and
whether equipment decontamination
will be employed,  standard operating
procedures, numbers of field samples
and control samples needed to achieve
the stated  QA Objectives.  It also in-
cludes a timetable for sampling activities.
                                       219

-------
 Section 5.0 addresses  information about
 what personnel are assigned which respon-
 sibilities, and which laboratories will be
 analyzing which samples. Section 6.0 dis-
 cusses the requirements  necessary to
 achieve the quality assurance objectives
 identified in Section 3.0. Section 7.0 ad-
 dresses the types of deliverables to be
 produced and what they will contain. Sec-
 tion 8.0 addresses the degree of data valida-
 tion necessary to achieve the identified QA
 Objective.
                                   for
Quality Assurance Sampling  Plan
Emergency Response (OASPER)
QASPER is a PC-based software package
which compiles  generic text and user
provided, site-specific information into a
draft QA/QC Sampling Plan  for the
Removal  Program.  QASPER addresses
the nine sections of a QA/QC Sampling
Plan, as specified in OSWER Directive
9360.4-01, Removal Program QA/QC In-
terim Guidance, Sampling QA/QC Plan,
and Data Validation Procedures.

The site manager (On-Scene Coordinator)
or contractors are the primary anticipated
users of QASPER.

These individuals will have access to the site
specific information and the sampling ob-
jectives which characterize the site inves-
tigation.  It is their responsibility to
assemble that information into an accept-
able sampling plan for implementation.

QASPER has  a database  of standard
generic text which is utilized  in an
electronic "cut and paste" process with user
provided site specific information to create
a draft QA Sampling Plan. This approach
enables the user to focus on critical infor-
mation while the software manages both
the presentation and correlation of that
information with other essential data.

Perhaps the best way to illustrate this
process is to "walk through" QASPER.
The user should progress in a sequential
manner, starting with section 0.0 because
the plan database will build on previously
provided  information.  This feature
avoids the need for redundant input of
data which must appear in several sec-
tions of the completed plan. It is possible
to skip sections,  or  avoid certain input
requirements (e.g., when information re-
quested is not yet known  to the user).
This allows the user to create those por-
tions of the database  at times that are
convenient to the user. However, it may
not be possible to complete certain sec-
tions (most notably  the DQO sections:
3.0, 6.0, and 8.0)  without providing cer-
tain  information in  preceding sections
(e.g. Section 2.0).

Figure 1. Main Edit Plan Menu
                                          Section 0.0 identifies certain information
                                          required to complete the title page of a
                                          Sampling Plan; some of information will
                                          also be utilized elsewhere throughout
                                          the completed plan. If the user chooses
                                          not to enter the information requested,
                                        220

-------
the completed plan (through the Output
menu) will be assimilated as if that informa-
tion was not requested. Should the user
wish to add alternate information currently
not requested by QASPER, this would be
accommodated through the Edit menu
after the plan has been compiled from the
database (through the Output menu).

Section 1.0 solicits background information
about the site. The user is first prompted to
geographically locate the site, characterize
its size and operating status, i.e., operation-
al or abandoned. The user is requested to
provide information about the  type of
facility.  (This information request is cur-
rently limited to  one response  per
category).  For sites with multiple facility
types, the user may enter this data through
the Edit Text menu after the file has been
compiled.  Next, the user is requested to
provide information about the materials
handled, the  surrounding environs and
populations.  Responses to these requests
are facilitated by pop-up menus of standard
responses. In the last three parts of this
section, the user provides the information
requested  by  typing onto free-form test
screens. Although there is room for multi-
ple page responses under each information
request, one to several paragraphs should
be sufficient.

Section 2.0 requests information regarding
the objective and purpose of the sampling
event.  How does the user expect to utilize
the resultant  data?  Several  standard
responses are  provided and may be ac-
cessed by the arrow keys and or selected by
the "Return" key.  The user may input an
alternate "objective" or "purpose" by select-
ing the "Other" category and specifying the
other use.  The return key is  utilized to
mark or unmark each item. A critical con-
sideration for any data collection event  is
whether the data will be evaluated against
an existing  database or action level.
Specification of the contaminants of con-
cern and their respective actionable
levels will help determine appropriate
analytical methods and quality assurance
needs later in the plan. Multiple selec-
tions are permissible from the  screen.
Selections under the "Purpose" group
will be carried forward to other sections
of the plan.  This section, therefore, re-
quires input in order to enable the user
to complete portions of Sections 3.0,4.0,
6.0, and 8.0.
Figure 2.
Menu
Section 3.0  QA Objectives
 In Section 3.0, the user will select among
 various parameters to identify the class
 of compounds to be investigated. This
 parameter  selection will  initiate the
 DQO logic for a parameter, in a matrix
 (next menu),  for a given purpose (sub-
 sequent menu), at a selected Quality As-
 surance Objective  (subsequent menu).
 At the end of the logic path, the user will
 be brought back to the parameter menu
 to make another selection, if ap-
 propriate. QASPER remembers the last
 logic path, therefore if the user wishes to
 select the same parameter, same matrix,
 same purpose, and a different QA Objec-
                                        221

-------
live, he/she need only move the highlight on
the last option.

Section 4.0 of the system solicits informa-
tion about the proposed sampling rationale
and how sampling will be conducted. There
are five subsections which address the fol-
lowing:

1. Sample Equipment

The user is requested to  identify sampling
equipment that will  be utilized, what
material  it is made of (fabrication), and
whether it is to be dedicated and/or decon-
taminated. The user must identify the sam-
pling tools which will  be used to collect
samples from the various matrices. This
process is initiated by first selecting a matrix
from among those previously identified in
Section 3.0. Next, the user will identify the
type(s) of equipment  to be used  in the
various matrices selected.  The emphasis
here is on the  equipment which will be
utilized to obtain the sample from the en-
vironment and transfer it to the sample con-
tainer.  Most of the equipment in the menu
has a corresponding Standard Operating
Procedure (SOP) available in Subsection
4.3.

Figure 3.  Sampling Equipment Decon-
tamination Sequence Menu
The user is also requested to identify the
equipment fabrication, or material of
construction. This is important so that
the quality of the sample is not com-
promised, inadvertently, by the materials
it comes in contact with during sample
collection. This is usually critical for low
concentration investigations, or situa-
tions of incompatibility between sample
contaminants and sampling  device
fabrication. If the equipment is not dedi-
cated, QASPER will import generic text
describing decontamination procedures
and solicit additional information about
the user's preference for the  decon-
tamination sequence and chemicals (e.g.
solvents) of choice.  The user will high-
light, or select,  the decontamination
steps from a menu in  the order he/she
wishes the sequence to be conducted in
the field.  A manifestation of  that se-
quence will be compiled in the plan out-
put.

2. Sampling Design

In this  section, the user will indicate the
sampling  design  or grid proposed  to
achieve the sampling event objective. It
is expected that the user will detail where
and how many samples will be collected.
A basis for the sampling scheme would
be described herein, and a sampling map
would  be  referenced.  QASPER will
print a blank page with the name of the
site and the title, "Sampling Location
Map", for incorporation of this map.

3. Standard Operating Procedures

There are three sections to the SOP sub-
section, addressing standard text for
Sample Documentation, Sampling, and
Sample Handling and  Shipment.
QASPER allows the user to choose exist-
ing generic text from  the database,  or
                                       222

-------
write new text to describe how sample
documentation will be achieved. If the user
selects "Write own Text", a free form edit
screen of several pages will appear to
receive the user's narrative.

Figure 4. Available SOPs Menu
 QASPER enables the user to choose from
 an inventory of standardized SOP texts to
 prepare a description of how the sampling
 event will be conducted. There are several
 approaches  for incorporating Sampling
 SOPs:

 -The user may  import only the titles of
 SOPs into the compiled plan. This reduces
 the bulk of the final plan document and may
 be appropriate where all users of the plan
 would have  access  to  a repository of the
 actual SOP texts.

 -The user may import title and text into the
 compiled plan. This allows the final plan to
 be a "stand alone" document.

 -The user may  import any portion of the
 generic titles and  text available through
 QASPER and/or modify and add SOPs to
 the QASPER database.
4. Schedule of Activities

The user is requested to provide a
timetable for the sampling activities.
This usually begins with the procurement
process for laboratory services and may
end with delivery of the final report.  A
tabular presentation will be created
when the plan is compiled.

5. Tables

QASPER presents a summary table of
each parameter, matrix, purpose, and
QA objective as compiled in Section 3.0.
The user will select by means of the high-
light bar and "return" key to initiate a
method  selection for each parameter,
identification of level of sensitivity, num-
ber of samples to be collected and QC
samples  needed to address the relevant
QA objective. This information will be
assimilated by QASPER into Field Sum-
mary and QA/QC Summary Tables.

FigureS. Field QA/QC Summary Tables
Menu
 In Section 5.0, the user is requested to
 identify what personnel will be perform-
 ing what tasks or responsibilities for the
 sampling event. Likewise, the user is re-
                                       223

-------
quested to provide the name of the lab and
a city or state descriptor for an address.
Labs will be characterized as either CLP,
commercial, EPA or field under the space
for lab type. Parameters may be identified
by class of compound.

Section 6.0 of the plan database receives
standardized text regarding QA require-
ments,  based  on the QA  Objectives
selected in Section 3.0. The user has the
opportunity to view and  edit  the  text  in
Section 6.0, since this is where the informa-
tion will appear in the final compiled plan.
There are also options for deleting generic
text or writing unique text (requirements).
The menu will indicate which QA Objective
requirements are  being imported  (e.g.
QA1, QA2, and/or QA3).

In Section 7.0, QASPER contains an inven-
tory  of standardized descriptions of the
types of deliverables which may be
prepared under a sampling event. The user
need  only  select the appropriate
deliverables, and the resultant plan will
contain the appropriate text.

Figure 6. Deliverables Menu
Section 8.0 contains the requirements for
validating the data generated  under the
plan.  The text in this section will be auto-
matically imported at the time the QA
Objective(s) is selected.

After completing review  and/or
modification of Sections 0.0-8.0, the user
may proceed to the output menu to com-
pile the plan for eventual printing or
sending to diskette.

Features of OASPER

If contained, requires no other software
for support

-Generates ASCII outputs - file and
hardcopy.  Files  may be uploaded to
other word processing packages for fur-
ther manipulation

-Creates (draft) hard copy QA/QC Sam-
pling Plan document ready for approval
signatures and implementation

-Retains  database files on all previous
sampling plans for future manipulation
(e.g. recreating documents, searching for
similar sampling plans by location,
facility type, contamination, etc.)

-Capable of  transmitting (compiled)
sampling plan or database via diskette or
modem

-Improves consistency and comprehen-
siveness of sampling plan creation efforts
throughout office, region, or zone

-Improves efficiency for creating and
reviewing QA/QC Sampling Plan docu-
ments

-Repetitive  use of information
throughout the plan without the need for
redundant data entry
                                         224

-------
-Provides the user access to standardized
generic text with overwrite capability for
editing

-Flexible data entry throughout

Requirements

QASPER runs on an IBM PC or 100%
compatible, with a hard drive, 640KRAM
and a printer (for hardcopy output).

Conclusion

QASPER is a PC-based software package
which compiles generic text and user
provided, site-specific information into a
draft QA/QC Sampling Plan for the EPA
Removal Program. It is envisioned that this
tool will primarily facilitate the timely as-
sembly of comprehensive QA Sampling
Plans in emergency response scenarios and,
indirectly, educate users on the correlation
of data quality objectives and sampling ac-
tivities.

Mention of trade names or  commercial
products does not constitute EPA endorse-
ment or  recommendation for use.

References

U.S. Environmental Protection Agency,
Quality Assurance/Quality Control
Guidance for Removal Activities, Sam-
pling QA/QC Plan and Data Validation
Procedures, Interim Final EPA/540G-
90/004, April 1990.
                                     225

-------
                   A RATIONALE FOR THE ASSESSMENT OF  ERRORS  IN  SOIL  SAMPLING
            J. Jeffrey van Ee*
       Exposure Assessment Division
     Environmental Monitoring Systems
                Laboratory
         Las Vegas, Nevada  89193
*0irect  questions  to  this author.
             Clare L. Gerlach
      Lockheed  Engineering &  Sciences
                  Company
          Las Vegas,  Nevada 89103
ABSTRACT

Considerable  guidance has been provided on
the  importance of  quality assurance  (QA),
quality control (QC), and quality assessment
procedures  for determining  and minimizing
errors   in  environmental  studies.     QA/QC
terms,   such  as  quality  assurance  project
plans and program plans  are becoming a part
of  the vocabulary  for  remedial  project
managers (RPMs).    Establishment  of  data
quality objectives  (DQOs)   early  in the
process of a site  investigation  has  been
stressed in  EPA QA/QC  guidance documents.
Quality assessment  practices,  such as the
use  of duplicates,  splits,   spikes,  and
reference   samples,  are  becoming  widely
accepted as  important  means  for assessing
errors   in  measurement  processes.   Despite
the  existence of various forms of  guidance
for  hazardous  waste  site investigations,
there  have been  no  clear,  concise,   well-
defined strategies  for precisely how  these
recommended QA/QC materials can be utilized.

The  purpose of this paper is  to familiarize
field  scientists  with an approach  to  these
questions:

     How many and  what  type  of samples are
     required to assess the quality of data
      in a  field  sampling effort?

     How  can the   information  from  these
     quality assessment samples  be used to
      identify and  control  sources  of  error
      and  uncertainties  in the measurement
     process?
The  primary audience  for  this  paper  is
assumed to be RPMs who have concerns about
the quality of the data being collected at
Superfund  sites  but  have  little  time  to
investigate   the   complexities   of   the
processes  used  to  assess  the quality  of
data from  the total  measurement  process.
The approach outlined in this document for
assessing errors  in  the  field sampling of
inorganics  in soils  may  be transferrable,
with modification, to other contaminants in
other media.

This  presentation   is  a  summary  of  "A
Rationale for the  Assessment  of  Errors in
the Sampling  of  Soils" by  J.  Jeffrey van
Ee, Louis J.  Blume,  and  Thomas H.  Starks,
1990.

An  in-depth treatment of  the statistical
approach is outlined in the Rationale (1),
and it is recommended reading.
INTRODUCTION

This document expands upon  the guidance for
quality control samples for field sampling
as contained  in  Appendix  C of  EPA's Data
Quality  Objectives  for  Remedial  Response
Activities - Development Process  (2).  That
report   outlines,    in  greater   detail,
strategies for how  errors  may be assessed
and minimized in  the sampling of soils with
emphasis on inorganic contaminants.

Basic guidance for soil sampling QA, which
includes a discussion of basic principles,
may be found in EPA's  Soil  Sampling Dualitv
Assurance  Users   Guide  developed  at  the
Environmental    Monitoring    Systems
                                             227

-------
Laboratory, Las Vegas (3).  The Users Guide
is  intended to  be revised  on a  periodic
basis.  It  is  anticipated that some  of  the
guidance  provided  in this  document will
eventually  be  incorporated into the  Users
Guide.

The  sampling  and  analysis  of  soils  for
inorganic   contaminants   is   a    complex
procedure  from experimental  design  to  the
final  evaluation  of  all  generated  data.
Sources  of error  abound but  they can  be
successfully mitigated by careful  planning
or isolated by  intelligent error assessment.
Error (or variability) can be either bias or
random.   Biased error  is indicative of  a
systematic  problem that  can  exist  in  any
sector  of  soils  analysis, from sampling  to
data analysis.   The first step in  analysis
of variability (or error) is to establish a
plan that will  identify  errors, trace them
to  the  step  in which  they occurred,  and
account  for variabilities to  allow  direct
action to correct them.   In anticipation  of
errors,   it  is   essential   to  ask   two
questions:

1.    How  many and what  type samples  are
      required to assess  the quality of data
      in a field sampling effort?

2.    How  can  the  information from  these
      samples  be  used   to   identify   and
      control     sources   of   error    and
      uncertainty in the measurement?

Error assessment should be understood by  the
field scientist  and the  analyst.   To  aid
scientists in the estimation and evaluation
of variability, the Environmental Monitoring
Systems Laboratory-Las Vegas  (EMSL-LV)  has
developed a computer program called ASSESS.
ASSESS can trace errors to their sources  and
help  scientists  plan  future  studies that
avoid the pitfalls of the past.
BACKGROUND

Superfund and RCRA site  investigations are
complicated by: the variety of media being
investigated, an assortment of methods, the
diversity of investigators, the variety of
contaminants, and the numerous risks to and
effects on human health and the environment.
Many   phases   exist   in  Superfund  site
investigations. An initial phase, generally
described as  a  preliminary investigation,
consists  of   collecting   and   reviewing
existing  data   and  data   from   limited
measurements    using   practically    any
available   method.      The  next   phase,
generally    described    as    site
characterization, uses selected methods and
prescribed  procedures to  characterize  the
magnitude and extent of the contamination.
Later  phases   include  an  examination  of
remedial   actions,   which   involve   an
assessment  of treatment technologies,  and
continued monitoring  to assess  the degree
of cleanup  at  a site.  A  final  phase  may
require    long-term    monitoring    to
substantiate  that  no  new  or  additional
threats occur  to affect human  health  and
the environment.  Throughout Superfund site
investigations  QA/QC  procedures  change as
data quality objectives vary and different
phases proceed.
RANDOM ERRORS

Random errors can result in variations from
the true value that are either positive or
negative but  do  not  follow a  pattern  of
variability.     During   the   measurement
process, random errors may  be caused  by
variations  in:

      1) sampling
      2) handling
      3) transportation
      4) preparation
      5) subsampling
      6) analytical procedures
      7) data handling

The greatest source of error is usually the
sampling  step.     In  the  Comprehensive
Environmental Response, Compensation,  and
Liability  Act   of  1980   (Superfund,   or
CERCLA) and the  Resource  Conservation  and
Recovery Act  (RCRA),  site investigations,
analytical,  and  data  handling variability
are checked  by the CLP protocol.  When more
than one laboratory is involved, handling,
transportation,    subsampling,     and
preparation  can  be checked  at Level  IV.
All analyses  are performed in  an  offsite
Contract    Laboratory    Program    (CLP)
analytical     laboratory    following    CLP
protocols.
                                              228

-------
But how can the analyst know that the sample
in  the  jar   is  representative   of   the
surrounding samples at  the site?   How  can
the field  analyst  know that  the more  (or
less)  contaminated  soil  didn't stick to  the
auger or  split-spoon?

It  is   strongly  recommended   that   the
traditional  approaches  used in  mitigating
the error in the last six  steps  be  applied
to sampling itself, i.e., use of duplicates,
splits,  spikes,  evaluation  samples,   and
calibration standards.   A  certain amount of
random  error   is    inherent   in   samples
themselves.    In  fact,  the total variance
equals the measurement  variances plus  the
population  variances,   as  defined  by  the
equations:
where at  =  total  variability
      am  =  measurement  variablity
      a  =  population  variability
and
                                    and
where as =  sampling  variablity
            (standard deviation)
      Qh =  handling, transportation
            preparation variability
      0SS= preparation variability
            (subsampling variability)
          laboratory analytical  variability
           between batch variability
      aa =
      a =
NOTE:  It  is   assumed   that   the  data   are
      normally  distributed   or   that   a
      normalizing  data  transformation  has
      been performed.

We can address the variance  in measurement;
the population variance,  however,  is  a  true
picture of the complexity of the  soil.
BIAS ERROR

Some sources of error are systematic,  that
is,  in a  given  situation conditions exist
that   consistently   give   positive   or
consistently give  negative  results.    This
skewing of data can be introduced early in
a  sampling  regime,  e.g.,  by  a sampling
device that alters  the  composition  of the
                                                   soil matrix.   It can occur in the middle of
                                                   the   sampling  regime,   e.g.,   by   the
                                                   preferential   handling of  a  sampler  who
                                                   isn't trained in the intricacies of sample
                                                   handling and  preparation.   Or  it  can  be
                                                   introduced in the later, analytical stages,
                                                   where  it  is  easier to  trace  because  of
                                                   interlaboratory comparisons  and  frequent
                                                   calibration checks.   The pervasive quality
                                                   of an early bias error is its resistance to
                                                   detection   and  the   fact   that   other
                                                   variabilities  are   added  throughout  the
                                                   process until, finally,  the reported data
                                                   may be significantly non-representative of
                                                   the true value.  Bias errors  can be traced
                                                   to:
                                                           faulty sampling design
                                                           skewed sampling procedure
                                                           systematic operator error
                                                           contamination
                                                           degradation
                                                           interaction with containers
                                                           displacement of phase (or chemical
                                                           equilibria)
                                                           inaccurate instrument calibration
PREVENTION

To avoid both random and  bias errors (or at
least   to   be  able  to   pinpoint  their
occurrence and estimate  their  extent),  it
is wise to plan a study well, anticipating
possible sources of  error.   The inclusion
of  quality  assurance  samples  used  for
quality assessment and quality control can
help isolate  variability  and  identify its
effect.

An  effective  technique is  to  concentrate
duplicate sampling early  in the study and
send   the   samples   off  for   rapid   CLP
analysis.  Dependent on the  results, it may
not be necessary  to  include  as many quality
assessment  samples  after  these  samples
demonstrate  reliability   in the  sampling
process.   Early detection  of  sources  of
error   can   help   the   field   scientist
customize the remainder  of the  study  to
meet the specific needs of the project.
                                                   QUALITY ASSESSMENT SAMPLES

                                                   A Remedial  Project Manager (RPM) must ask:
                                                   how many samples  are  needed  to adequately
                                                   characterize the  soil  at this  site?   The
                                             229

-------
key word  is  "adequately."   By  determining
the  data  quality   objectives   (DQOs)   in
advance,  the   RPM  can   assure   adequate
sampling at a site.  Too little sampling,  as
well as  too  much, is  a  waste of  time and
money.    The  extent  of  QA/QC  effort  is
dependent on the  risk  to  human  health, the
nearness  of  action  levels  to  detection
limits,  and  the  size,   variability,  and
distribution of contamination.  Ultimately,
the number of quality assessment samples  is
determined by  the DQO  for the site.  Table
1   explains    various  types  of   quality
assessment samples and their uses.
SOME STATISTICAL CONCERNS

Confidence in quality assessment sample data
can  be  expressed  as an  interval  or  as  an
upper limit.  All  confidence  levels/limits
are  based  on the  number  of  degrees  of
freedom and  the limits  get lower   (or the
intervals  get smaller)  as the  number  of
degrees of freedom  increases.  For  example,
if 15 samples are  taken at a site,  and each
split   is   extracted   twice  at   a   CLP
laboratory,   and   2  injections  of   each
extraction  are  made  into  an  Inductively
Coupled Plasma/Mass Spectrometer  (ICP/MS),
the  total  number   of  degrees  of  freedom
associated  with  this  experimental  design
would be calculated as:

 15 samples X 2  preparations splits   =  30
            X 2  CLP extractions       =  60
            X 2  injection replicates  = 120

120  degrees  of  freedom   for the  whole
process.    But,   if  only  the population
variability  in  the field  samples  (which
includes  the  sampling  error)   is  being
estimated, the number of degrees of freedom
is 15-1,  or  14.   There are 15 independent
samples but  one degree  of freedom  is  lost
with the estimation  of the mean. Therefore,
there  are 14 degrees  of  freedom   for  the
sampling  variance   estimate.    As  another
example, to estimate the variability in the
extraction  step,  one  has  30  independent
pairs of numbers, each pair associated with
one extraction.  Thus,  there are 30 degrees
of freedom  associated  with the extraction
error.

Obviously, the  confidence associated  with
any particular sampling is directly related
to the number of samples taken.  In Table 2
(also Table 3 of the Rationale Document) or
in a  statistics  manual, guidance  is given
for   the   number  of   quality  assessment
samples that must be used  with the routine
site  characterization  samples.     These
tables  assume   that   data   are  normally
distributed.  The tables will show the user
the confidence interval associated with the
degrees of freedom. Then, decisions may be
based upon the requirements  of the DQOs.  A
synopsis of  this targeted  approach  can be
seen  in  Figure  1.  The total  measurement
error is comprised of error  in the sampling
(a),   subsampling  (ass),    handling   (ah),
batch (ab), and  analysis (aa)  steps.  Each
is  addressed  in the  regime  depicted  in
Figure 1.
SAMPLE COLLECTION CONSIDERATIONS

If Level  IV  CLP analysis  is  performed on
the soil,  we can assume that  very little
error occurs  in the  analytical  stage.  This
focuses our attention on sources of error
in the sampling, handling,  and preparation
steps.   The  two major  considerations  in
collection of environmental samples are:

1. Will the collected data  give the answers
   necessary  for a  correct  assessment of
   the contamination or a  solution to the
   problem?

2. Can sufficient sampling  be done well and
   within  reasonable cost and  time limits?
ASSESS

The  EMSL-LV  has developed  an  easy-to-use
program   to   calculate   the   necessary
statistics, as  described  in the Rationale
(1),  from  the  generated   data  for  an
accurate  determination of  precision  and
bias.  ASSESS  is a public domain,  FORTRAN
program that is available from EMSL-LV and
written for personal computers.  It may be
applied
to cases where no field evaluation samples
are available  as well  as  cases where they
are.  ASSESS  is  user-friendly  and  its use
will greatly aid both field scientists and
RPMs  in  decision-making  based  on  soil
studies.
                                               230

-------
                                            TABLE  1

                           QUALITY ASSESSMENT SAMPLES AND THEIR USES


•  ALLOW STATEMENTS TO BE MADE CONCERNING THE QUALITY OF THE MEASUREMENT SYSTEM

•  ALLOW FOR CONTROL OF DATA QUALITY TO MEET ORIGINAL DQOs

•  SHOULD BE DOUBLE-BLIND:

      Types  of Samples           Description
      Field Evaluation
      (FES)
      Low Level  Field
      Evaluation (LLFES)

      External  Laboratory
      Evaluation (ELES)

      Low Level  External
      Laboratory (LLELES)
      Field  Matrix  Spikes
      (FMS)

      Field  Duplicates
      (FD)

      Preparation Splits
      (PS)
• SHOULD  BE  SINGLE-BLIND:

      Field  Rinsate
      Blanks (FRB)

      Preparation  Rinsate
      Blank  (PRB)

      Trip Blanks  (TB)
Samples of known concentration are introduced in the field
as  early  as possible  to check  for measurement  bias and  to
estimate precision

Low concentration FES samples check for contamination in
sampling,  transport, analysis, detection limit

Similar to FES but without exposure in the field, ELES can
measure laboratory bias and, if used in duplicate, precision

Similar to LLFES but without field exposure, LLELES can
determine the method detection limit, and presence of laboratory
contamination

Routine samples spiked with the analytes of interest in the
field check recovery and reproducibility over batches

Second samples taken near routine samples check for
variability at all steps except batch

Subsample splits are made after homogenization and are used
to estimate error occurring  in the  subsampling  and analytical
steps of the process
Samples obtained by rinsing the decontaminated sampling
equipment with deionized water to check for contamination

Samples obtained by rinsing the Blanks sample preparation
apparatus with deionized water to check for contamination

Used for  Volatile  Organic Compounds  (VOC),  containers  filled
with American Society for Testing and Materials  Type  II water
are kept with routine samples  through the sampling,  shipment,
and analysis phases
• MAY  BE  NON-BLIND:   AS  IN  THE  INORGANIC  CLP PROTOCOL
                                              231

-------
                       TABLE 2


Some 95 Percent Confidence Intervals for Variances

Degrees of Freedom            Confidence Interval

       2
       3
       4
       5
       6
       7
       8
       9
      10
      11
      12
      13
      14
      15
      16
      17
      18
      19
      20
      21
      22
      23
      24
      25
      30
      40
      50
     100
0.27s2
0.32s2
0.36sJ
0.39s2
0.42s2
0.44s2
0.46s2
0.47s2
0.49s2
0.50s2
0.52s2
0.53s2
0.54s2
0.54s1
0.56s2
0.56s2
0.57s2
0.58s2
0.58s2
0.59s2
0.60s2
0.60s2
0.61s2
0.62s2
0.64s2
0.67s2
0.70s2
0.77s2
< a2
< a2
< a'
< o2
< a2
< a2
< o2
< a1
< o2
< a2
< a2
< a2
< a2
< o1
< a2
< a2
< a2
< a2
< a2
< a1
< a2
< a2
< a2
< a2
< a2
< a1
< a2
< a1
< 39.21s2
< 13.89s1
< 8.26s2
< 6.02s2
< 4.84s2
< 4.14s2
< 3.67s2
< 3.33s2
< 3.08s2
< 2.88s2
< 2.73s2
< 2.59s2
< 2.49s2
< 2.40s1
< 2.32s2
< 2.25s2
< 2.19s2
< 2.13s2
< 2.08s2
< 2.04s2
< 2.00s2
< 1.97s2
< 1.94s'
< 1.91s'
< 1.78s'
< 1.64s2
< 1.61s2
< 1.35s2
                           232

-------
                                            Figure  1.

                                       QUALITY ASSESSMENT SAMPLES
                        DUPLICATES AND SPLITS
                                              EVALUATION SAMPLES
                                                                                BLANKS
SAMPLE TAKING
PREPARATION
ANALYSIS
SOURCES OF ERROR
III!
| ROUTINE | | FIELD | , 	 , , 	 , , —
1 QAMPI F 1 	 InilPtTTATFl 1 FFS 1 I FFC 1 1 F

1 1 1
	 	 	 _ 1 	 	 	 I 	 . 	 I 	 „„ 	 1 „ 	 ....__..... 	 	 	
	 1 	 1 	 j 	 | 	
	 I I II
1 1 1 II
III II
| ROUTINE | JPREP. SPLIT] 1 FD 1 1 FES j 1 FES |
| SAMPLE |-| SUBSAMPLE | | SUBSAMPLE | 1 — , — 1 ' — , — 1 NO PREPARATION
i it ii i
1 1 1 IELESI |ELES|
»]
r-1
| PRS |
1
1
1
III II
| RS | I PS | 1 FD | | FES | | FES | ' |ELES| |ELES| | FRB | | PRB |
? oss.Oa a, ah. ass. aa os.ob.oh aa oa.ok.os oa.oh
ACKNOWLEDGEMENT

This work is based on the in-depth treatise,
"A Rationale For the Assessment of Errors in
the Sampling of Soils"  by J. Jeffrey van Ee,
Louis Blume,  and Thomas  Starks.
NOTICE

Although research described in this article
has been funded wholly by the United States
Environmental   Protection   Agency   under
contract  number   68-03-3249   to  Lockheed
Engineering & Sciences Company, it has not
been   subjected  to   Agency   review  and
therefore does  not  necessarily reflect the
views  of  the  Agency,   and   no  official
endorsement should be inferred.  Mention of
trade names or commercial products does not
constitute   Agency   Endorsement  of  the
product.
REFERENCES

(1) van Ee, J.J., L.J.  Blume,  T.H.  Starks,
    A  Rationale  For   the  Assessment  of
    Errors  in  the  Sampling of  Soils,  U.S.
    EPA, 1990, 600/4-90/013.

(2) U.S.   EPA.   1987.      Data   Quality
    Objectives   for    Remedial    Response
    Activities   -    Development   Process.
    EPA/540/6-87/003.

(3) U.S. EPA. 1989.  Soil  Sampling  Quality
    Assurance  Users  Guide  (2nd.  Edition).
    Environmental    Monitoring    Systems
    Laboratory,  Las Vegas, Nevada.    EPA
    600/8-89/046.
                                               233

-------
                                                             DISCUSSION
REX RYAN: You did an admiral job of explaining the strategy of breaking down
what we call a "nugget effect" by using ANOVA techniques. I was a little bit
shocked that you didn't discuss  the amount of variance distance contributes
within a sampling program. I was also surprised that you didn't discuss variograms
or any of those kind of issues that would affect a sampling team's success in
determining what is in fact going on at a site.

JEFFREY VAN EE: The two methods go together. The method I've described
is useful in pinpointing sources of variability in the measurement process if you
want to make changes. But the  points that you're making address the larger
question of where your samples are located and whether they're going to be
representative of the site, assuming that the measurement variability is relati v'ely
low. That certainly needs to be looked at how representative are your sampling
locations to the contamination throughout the site.

REX RYAN: In your experience which do you think is larger—which in fact
could—in your professional judgment be a larger contribution to total variabil-
ity: the problem of extending samples in distance or trying to replicate samples
at  the same location?

JEFFREY VAN EE: I don't think I have enough data to answer that question.
I can pose a few questions for all of you to consider. Let's say that we're sampling
volatile organics or a contaminant that varies with depth. This approach would
be useful in determining whether  the sampling of that contaminant is being done
well. If you take a field duplicate sample and you go down, say, four inches and
your contamination is in the first two inches of the surface, then this method will
allow you to see that kind of variability from how the samples actually collected.
This method w ould also allow you to look at the loss of volatile organics. By the
time the samples get to the lab, it's more difficult with volatile organics and we
need to do some more research to see if this approach is applicable. But those are
some of the questions that can be answered by using this approach.

Both methods have been used together—at a site in Region VII.  and they both
yielded very useful information.  The CEO Statistical Approach again, looks at
the question of how many samples you need to collect to characterize a site and
then our method looks at whether those samples are being collected properly,
handled properly, those kinds of questions.

NABILYACOUB: I have aquestion about a statement you made about a second
sample collected at about an inch and a half and two inches from the original
which relates directly to this concern. Would this be a measure of the effect of
sample handling, the performance of the laboratory, containers, etc.? I beg to
differ because we are introducing  here a variable that  might bias the  results.
Would you consider this sample as a split sample? If not, would you consider a
split sample more representative of the effect of these operations rather than this
end?

JEFFREY VAN EE: You need to use a combination of samples together. We are
assuming, (although we can disprove it,) that the spatial  variability in those two
inches is insignificant. We can disprove it by the introduction of other samples
throughout the  process. Once we collect a field duplicate, we could split that
sample and then analyze it separately to get a handle on errors down the line:
handling in the subsampling of the core or analytical errors. If we do come back
with this analysis and see that we do indeed have tremendous  differences in
moving two inches away and we compare that to the GEO Statistical Approach
then we've got  some real problems in characterizing that site.

A lot really depends how  the contaminant was distributed at the site. If the
contaminant was uniformly distributed at a site, then I would expect the spatial
variability to be low. If we have  leaking drums, we might just happen to hit on
that area, and if we move two inches over we would get a dramatically different
result. But the more samples we collect, the more field duplicates we collect.
presumably we will get a more representative idea of where the variability is. If
we were to rely on just one field duplicate or a few. then we would really be prone
to some of the misjudgments that you're alluding to.

ROY  KAY: As I  understand it,  the objective of sampling and  population
comparison within samples, is to provide a cost effective means of reducing the
total sampling costs while maintaining a  high level of accuracy. Am I correct
there so far?
JEFFREY VAN EE: Yes.

ROY KAY: Has there been any cost evaluation information developed on the
relative cost of going through the process of designing and multiple batching
your samples versus simply expanding randomly the samples that you take—
particularly if you're starting  from an nonhistorical, time-zero point of view?

JEFFREY VAN EE: I think a lot depends on the objectives that you establish
for that site. You need to look at the economics of collecting more samples, what
type of samples, versus the kind of action that you're going to be taking. It you
know that you're going to be cleaning up the site in large pan then taking a lot
of samples may not be appropriate.

But if  the cost of that clean-up is significant, if the cost of disposing of the
contaminant  is significant, then you  will want to pay more attention to how
accurately you can characterize the site. And then, of course, you want to know
whether the data that you're  getting  represents the site or whether it is more
representative of variabilities in the measurement process.

I'm not sure I really answered your question well. It's a difficult question to
answer, because it varies depending upon the site.

ROY KAY: I'm looking at a situation where in a time-zero, first evaluation of
a site, there are certain theoretical  things that you had assumed, like if you have
an explosion of some kind, it  would naturally be expected to disperse contami-
nants. Whereas a leaking drum would expect to leach in a continuous fashion and
probably in all geometric dimensions. That is. of course, is a seat-of^the-punts
guess in each individual case. But lacking historical experience on that particular
site, do the sampling techniques dial in on the proper variables and reduction of
their influence faster than simply  expanding the sampling population?

JEFFREY VAN EE: In  a situation like that I would weigh more QA samples.
as well as more samples, period, early on in the process. You can hopefully back
off as  you learn more about the site. Now that's assuming you don't have
historical information on how well that particular contractor performs out in the
field, or how well  that particular sampling method performs.

Let me demonstrate very quickly  another value that comes out of this process.
Say you're out sampling the site and you're concerned about the change of the
contaminant overtime, you may have different labs involved, and you may have
different sampling crews involved. If you do not have a rigorous QA program
instituted, then when the data comes back out of the lab. it's very difficult for you
to say whether that data reflect the pollutant changing over time or whether it's
your measurement process changing over a period of time. So, at some point.
you've got to pay your dues and you've got to start developing that data. We have
a tremendous amount of data right now on how well the contract labs perform.
but we don't have  enough data on  how well those samples are transported to the
lab and how well they're prepared.  Say there's a rainfall event during your
sampling study, how do you know that the data you collect after that significant
event is comparable to the data before that event?

J ANINE ARVIZU: Could you describe some of the programmatic applications
of the  program and whether or not there were any good real world experiences
learned?

JEFFREY VAN EE: The philosophy I'm espousing today is relatively simple
and it's relatively  new. My hope is that more people will pick up on it whether
they're in RCRA or Superfund Programs. I think we really do need to demon-
strate where the variability is throughout the measurement process. Right now
I'm simply advocating that we try it. How well it's used remains to be seen. We
have applied it to a Superfund  site in the middle part of the country and we looked
at the  spatial variabilities. As a result of our efforts using GEO Statistics, we
saved about 6 million dollars in the sampling effort at this particular site. We w ere
able to demonstrate that the sampling method that they were using, while it was
crude,  was sufficient to meet  data quality objectives. We were able to tell them
that they could back off on a number of samples that they're taking in certain
areas,  because the measurement  variability was relatively low. They weren't
getting a lot of variability in the compositing of the samples. We have had a few
success stories, but not nearly enough. We can just hope with time there will be
more stories like that.
                                                                         234

-------
                      A REVIEW  OF  EXISTING SOIL QUALITY ASSURANCE MATERIALS
            Kaveh Zarrabi, Chemist
Amy Cross-Smiecinski, Quality Assurance Officer
      Thomas Starks, Senior Statistician
        Environmental  Research  Center
       University  of  Nevada,  Las Vegas
           4505 S.  Maryland Parkway
           Las Vegas, Nevada  89154
                   ABSTRACT

Assessment of the quality of environmental  data
often depends  on the  availability of  quality
assurance  (QA) materials to measure errors  at
various stages of the measurement process.   A
rigorous   approach  has   been   developed   to
evaluate the quality of  data  from  the  sampling
of  metals  in  soils.    "A Rationale  for  the
Assessment of Errors  in  the Sampling of Soils"
was written for application to  hazardous waste
site investigations.   The rationale described
is  based  primarily  upon  duplicate  and split
samples and QA materials  known as performance
evaluation materials.   The rationale  depends,
in  varying  degrees,  on performance evaluation
materials being readily  available  for  use  in a
hazardous    waste    site    investigation.
Unfortunately, early experiences in testing the
rationale  indicate  that  inadequate   numbers,
types,  and  volumes  of performance evaluation
materials and other  types of  soil  QA materials
exist to  fully implement the  rationale.

In order to begin to answer questions  as to the
necessity  of,  and  alternatives  to,  soil  QA
materials, it is  necessary to know the current
availability  and  the state  of  research  and
development of soil  QA materials.  The intent
of this paper is  to provide such  information -
what materials are available  and what  is being
done to provide more  materials.
                   INTRODUCTION
SCOPE
Millions of dollars are spent in designing and
implementing monitoring and remediation programs
for hazardous waste sites.  It is  the Agency's
responsibility to ensure that the data resulting
from these programs  are of adequate  qua!i ty to be
defensible  in a court of law as well  as  to be
considered  scientifically sound.

Quality assurance (QA) materials are an important
part of many environmental sampling and analysis
programs today.  Results from the analyses of
hazardous waste  site samples are often accepted
or rejected solely on  the basis of  data obtained
from QA samples  analyzed for Agency programs
ranging from  water  quality  monitoring  to
hazardous waste remediation.  It is  alarming that
only a 1 imited supply of  these QA materials is
available for soil  sampling and  analysis (Table
1).  What does a project  manager  do when no QA
materials exist?  It is the intent of this report
to discuss the need  for soil  QA materials in many
environmental programs'1'3 and to demonstrate the
limited availability of  these  materials.  An
alternative to  the  use  of manufactured QA
materials is briefly described as are approaches
for increasing the supply and variety of the most
commonly needed soil QA materials. This report
does  not  purport  to have the answer to the
scarcity of soil QA  materials, but simply to
point out the problem  and explore some solutions
with the hope that more attention  will be given
to the issue.

RESEARCH

Research in the area of QA materials has been
limited.  In fact,  the bulk of the information
gathered for this  report came  from catalogs,
personal communications,  and  internal reports.
The following examples were obtained through  a
literature search.  Recently, Taylor[3] published
a  comprehensive book,  Quality  Assurance of
Chemical Measurements. The book discusses the
basic concepts of quality  assurance and provides
details on evaluation  samples, traceability, and
                                                  235

-------
reference materials.  Sewardt41 of the National
Institute of  Standards  and Technology (NIST),
formerly  the  National  Bureau  of  Standards
(NBS),  published  a  book  which contains  25
papers  describing national  and  international
programs  for  the  development  of  reference
materials.    The  selection  criteria, use  of
statistics,  and  steps  for  certification  of
standard  reference  materials are  discussed.
Reports of 15 panel sessions  reviewing the use
of  and  needs  for  reference  materials  are
included.

Calit5]  of  NIST,   in  another NBS  monograph,
examines the general use of standard reference
materials  and their  role  in  the measurement
system.  Further, procedures  for certification
of standard reference materials are discussed,
and examples of several  selected  industries are
given  in which  standard  reference materials
have made a significant  contribution.  Steger
compiled   the  information   on   all   of  the
available certified reference materials through
the   Canadian  Certified  Reference  Material
Project.   Taylor    published a  handbook for
standard   reference  material   users.     The
preparation and analysis of reference materials
has  been discussed and documented  by several
programs. C8'9'10>11«12'13]    In  other studies, the
design      and   stability      of   reference
materials have  been  evaluated.

Another search of "Chemical Abstracts" from the
year  1979 to  the  present resulted in just five
more  references.    Studies  in  which the  QA
materials  were  used  range  from  proficiency
samples discerning between immunoinhibition and
electrophoretic   measurement  to   soil   and
geological  reference materials.
SOIL QA MATERIALS

DEFINITIONS

The uses  of  QA materials have  been  predefined
for  the  purposes  of  this  paper in  the  EPA
report   referenced  in   the  abstract:     "A
Rationale  for the  Assessment of  Errors  in  the
Sampling of Soil."m  Briefly summarized, there
are two  basic uses of  QA materials:   quality
assessment or  evaluation  (QAS)  and  quality
control  (QC).  QAS samples  are  intended to  aid
in evaluating data quality and can  be  used in
QC.   QC  samples are  used  specifically on  a
real-time  basis  to detect and correct problems
before  a  large  body  of  erroneous or  out-of-
control data is generated.  The  main  difference
between the  two  uses  becomes evident when  the
data generated from  them is  interpreted.   QAS
data  are  usually  analyzed at   the   end   of
studies, whereas QC data  is analyzed as it is
generated; hence, the quality is "controlled."

QAS and QC samples exist in several types such as
reference materials and performance evaluation
materi al s.  Reference materi al s are defi ned as
having  "one  or  more properties  which  are
sufficiently well established to be used for the
calibration of an apparatus, for
the assessment of a measurement method, or for
assigning values to materials.""   Reference
materials are typically used as QC samples but
can be used as QAS samples. Originally, soil QA
materials began existence as reference materials
and are slowly evolving as important components
of QA programs.

Performance evaluation materialst2'171 often are
associated with an analytical program  in which
participants  submit  results  to  a  central
authority  who "grades"  the  data  either  in
comparison to the pooled results of all of the
participants or against a "referee" laboratory in
order to judge  the overall performance  or
accuracy  of  the  laboratory.    Performance
evaluation materials are, therefore, examples of
QAS samples.

Whether the data is used  on a real-time basis
(QC) or  at the end of a study (QAS),  the overall
effect of a QA sample is to evaluate measurement
system performance.  The sample may be used to
evaluate a whole system, from sampling through
data validation,  or  a part of the system; such as
extraction efficiency.

An  important issue  for soil   sampling  and
analytical QA is how  closely soil QA  samples
represent the routine samples of interest.  A QA
sample should be similar to the routine samples
for the  analytical parameter in order for a true
correlation to exist between the  two.  Analytes
spiked onto potter's clay or sand probably do not
accurately mimic environmental samples  visually
or analytically and, therefore, test  only the
recoverability of the analytes from the clay or
sand in combination with  the competence of the
analysts.  In the chemical analysis of natural
soil samples,  it is especially important that a
QA sample be of a similar soil type as that of
the  samples  being  analyzed to eliminate the
effects  of various matrices effects on analytical
measurements  and final  results.

This paper deals with three basic types of QA
soil samples which are non-blind, single-blind,
and double-blind soil QA samples.  Non-blind QA
samples  are used for internal quality control and
for calibration.  Single- and double-blind QA
samples  are  used  in  quality assessment  and
external quality control.  All  three  types of
blind  QA materials have  been  successfully
                                                   236

-------
utilized to  control  and  evaluate laboratory
measurements.c18'191

Non-blind QA Samples

These samples  are not  blind to  the analyst.
The identity and reference values  of the  sample
are known.   Reference materials and laboratory
control   samples  are  examples   of  non-blind
samples.

Single-blind QA Samples

Single-blind QA samples are used principally as
a reference  point in analyses,  the data  from
which  serve  as  a  guide  to  acceptance  or
rejection of  routine sample  data.   A single-
blind QA sample is known to be a QA sample, but
its composition  is not  known to  the analyst.
A performance evaluation material  is an example
of a single-blind QA sample.

Double-blind QA Samples

Double-blind QA samples are used  as a basis for
acceptance or rejection  of  routine  sample  data
and  for quality  assessment.   The difference
between single- and double-blind  QA samples is
that the double-blind QA sample is  intended to
be  indistinguishable  from a routine  sample.
Visually, the QA  sample resembles the routine
sample  in container  type, number system,  soil
texture and  soil  color.   Analytically,  the QA
sample   resembles   the   routine  sample   in
interferences, coanalytes, etc.  This minimizes
bias in processing the sample batch. A double-
blind  QA sample  is  even  more   difficult  to
compose  or   develop  because,  in  addition  to
having  the  same or  similar chemical make-up,
the sample must  appear to be of  the same  soil
type.   For example,  if the soil   being sampled
for analysis is a Hagerstown  silt loam  (a  fine
textured medium brown  soil  with a neutral  pH),
an  acidic  red-colored  sand  would not  be  an
appropriate double-blind sample.   Spiked field
samples  and  field duplicates are examples of
double-blind QA samples.   Manufactured double-
blind QA materials are rare.

Use of Single-blind and Double-blind QA Samples

Quality  assurance samples  are  used to  detect
bias   and   to   estimate   precision   in   the
measurement  system.   The advantage of double-
blind  QA samples is  that  they  are  treated
exactly  like   the  routine  samples  in   the
analytical  laboratory  and  hence  should  be
exposed to the same  types and levels of  errors
in the preparation and analytical processes.

Unfortunately, it  is often  difficult to  employ
double-blind   QA  samples   for   studies   of
environmental pollution. Difficulties in using
double-blind QA samples usually arise for one of
two reasons. The first reason is that the nature
of the pollutant may make it impossible to carry
out the drying, grinding, sieving, homogenizing,
and subsampling (to obtain  a laboratory sample)
of  routine samples  outside  the  analytical
laboratory. This series of preparatory steps is
essential  for obtaining homogeneous soil QA
materials.  Such treatment produces QA samples
that look different  from the routine samples,
provided the routine samples did not go through
the same process before entering the laboratory.
The second most  probable reason  is that an
appropriate soil QA material is not available,
and there is insufficient time prior to field
sampling to characterize the soil QA material for
double-blind samples.  It should be noted that no
matter how many soil QA materials are available,
it is unlikely that  a soil QA material exists
that is appropriate for double-blind samples
unless the material actually comes from the site
under investigation.

If it is not possible to employ double-blind QA
samples  in an  investigation,  an alternative
procedure has been suggested based upon single-
blind samples and additional  field duplicate
samples.["  The  additional  field duplicate
samples in this alternative procedure allow the
estimation of total measurement error (i.e., the
precision  of the measurement  system)  and the
estimation of the variance contributions of
several  of the  possible sources  of error.
Depending on where they are  incorporated into the
sampling  and analytical scheme,  the single-blind
samples provide means for detecting  bias from
sample handling,  preparation, and  analysis.
Unfortunately, the single-blind QA samples may
miss some of the bias in the laboratory, owing to
special  handling  by the chemist, to which  a
double-blind  sample  would   not  have  been
subjected.  A  research  study  by  Rumley[20]
evaluated the effects of favorable treatment of
samples and of alteration  of results to reduce
bias  on  indices  of  performance in  external
quality assessment (EQA)  schemes.  He concluded,
in  fact,  that EQA schemes can be  affected by
giving  favorable treatment  to single-blind
samples.

Since there will always be a need  for single-
blind soil QA samples, and the need will often
involve situations requiring rapid response, it
seems imperative that an extensive inventory of
soil QA materials be prepared and maintained for
future environmental  pollution studies.  Double-
blind soil QA samples  should be employed where
practicable, and facilities should be available
to produce such samples in an expeditious manner.
                                                 237

-------
AVAILABILITY

The establishment  and expansion of monitoring
and enforcement  programs  by  federal  agencies
requires the use of many QA samples.   Federal
agencies such as the  U.S.  EPA and the Food and
Drug    Administration     (FDA)    established
repositories  of   QA  materials  out  of  the
necessity to support their own programs.  The
private sector, although originally interested
in  producing  standards   for  calibration  of
different instruments,  produces QA samples in
various media  and  for  specific environmental
programs (e.g., RCRA) in  a limited  variety.

Although  a  listing  of  many  of the  soil  QA
materials  described  that  are  available today
(Table 1) may appear sizable,  many analytes are
not represented.  At this time, the  authors are
unaware of any sources of soil  QA materials for
volatile   organic   analytes.     The  natural
variability  of soils,  however,  is  the  factor
that  makes  a  large number  of  QA materials
necessary.  The same factor limits  the ability
to manufacture sufficient materials to provide
realistic  and/or  blind QA materials  for all
hazardous   waste    sites  that   are   being
investigated.   This   deficiency  makes    it
difficult  to  plan   and  implement  many  soil
sampling and analysis QA programs.
SOIL  QA  MATERIALS NEEDED

AGENCY NEEDS

Clearly  there is  a need  for more sources  of
soil  QA  materials.   This   leads  to  certain
questions.   Which types are most often needed?
 Which materials  should  be manufactured first?
A  survey1211  of U.S.  EPA officials shows  that
all 10 Regions share an  interest in a national
QA  material   program for  Superfund  analyses,
primarily for use  by the Contract  Laboratory
Program  and  by Potentially Responsible Parties
(PRPs).

Although each Region has specific needs,  there
is some  agreement on analytes.    Most interest
is  in  materials  containing  Target  Compound
Listt213  analytes.    Special  requests  include
tetrachlorodibenzo-p-dioxin     (TCDD)     and
pentachlorodibenzo-p-dioxin     and     -furan
(PCDD/PCDF)  isomers; explosives (RDX); benzene,
toluene,   and  xylene   (BTX),   solvents;   and
polycyclic   aromatic hydrocarbons  (PAHs)  in
sediment.   The number of QA materials  needed
per year and their concentrations vary  among
the   Regions  (Table  2).c211     One   Regional
official   commented  that   site-specific   QA
materials are needed.1  The value of the soil
QA  materials distributed by the EMSL-LV  CLP
Performance  Evaluation   Program   has   been
proven,   '   but as demonstrated in Table 1, this
program offers a limited variety of  samples and
analytes.   The  EMSL-LV  program  would  need
additional resources  in order  to  be able to
provide a wider variety of materials.

INDUSTRIAL POLLUTANTS
                                      ble
                                       ]
Industrial organic chemicals presently comprise
the highest volume of hazardous waste produced,
followed  by  wastes  from  general  chemical
manufacturing, petroleum refining, and explosives
(Table 3)12  . According to the Comprehensive
Environmental Response Compensation Liabil ity
(Act) Information Systems    database, the most
abundant pollutants on the National Priority  List
(NPL) of Superfund sites are from the industrial
and  general  organic   chemicals  industries,
petroleum refining, and explosives industries
(Table 3).  The pollutants found most often on
these NPL sites (Table 4) are Pb, As, Cd, Cr, Hg,
Cu, and cyanides for inorganic pollutants, and
trichloroethylene (TCE),  other  chlorinated
solvents, and BTX for organic pollutants.  The
highest volumes of organic pollutant/waste are
volatile organic compounds (VOCs) , while heavy
metals comprise the greatest volume of inorganic
wastes.      It would seem that soil QA samples
containing the pollutants specified by the users
(e.g., Regional users) and/or those most commonly
found at the NPL sites  should be the first to be
produced.

SUGGESTED RESEARCH

Supplying Blind  Soil  QA Materials

At this time, preparing and stocking complete
(adequate analytes) and realistic (double-blind
as well as single-blind) soil QA materials is not
feasible  due  to  the  tremendous  natural
variability of soils.  On the other hand, as
stated previously, variety of QA samples  from
present sources  is limited  (Table 1).

Two general approaches,  that overlap somewhat in
their philosophy,  are presented for manufacturing
both single-blind and double-blind QA materials.
These are:  industry-specific QA materials in
which a limited number of soils are produced  that
contain   analytes  specific   to   polluting
industries; and site-specific QA materials in
which soils found at hazardous waste  sites are
prepared  to  contain  analytes  or  analyte
combinations commonly found at hazardous waste
sites. Either approach would require a rigorous
multi -laboratory characterization study.  As one
example,  soils  naturally  rich in particular
metals could be obtained and processed for either
industry-   or  site-specific  QA  materials
                                                   238

-------
representing mining  industry wastes  for sites
with similar soil  characteristics.

Industry-specific  Materials

Using historical  industry data  as  well  as NPL
data, information  such  as  geographic location,
contaminant  types, and  concentrations  can be
mapped and evaluated for any general geographic
trends.     This   information   can   then  be
correlated with  10-15 general  soil-types1 '
to  narrow  the  choices  of  industry-specific
soil/analyte combinations. The  next  step would
be  to  collect  and  homogenize  the  selected
soils.  During homogenization some  of the soils
would   be   spiked   with   contaminants   for
characterization  and  distribution.  This would
result  in samples  that  could be used  for non-,
single-,  and perhaps  double-blind,  blank, or
contaminated  soil  QA  materials.  The materials
could then be stored at distribution  centers to
fill   user   requests  for  various   industry-
generated hazardous waste sites.
 Site-specific  QA Materials

 Relying  on NPL  site  data  in  combination  with
 geographically related  soils, a  set of  site-
 specific  soil  QA  samples  could  be developed.
 In this  approach, the selected soils could  be
 collected   for  spiking   and   processing,   as
 described  in the  previous  section;  or,  using
 site-specific  soil/analyte combinations,  the
 materials   could  be  collected   from   actual
 hazardous   waste  sites,   with  blanks   being
 obtained  from nearby uncontaminated soils  of
 similar composition.  The artificially composed
 materials  and the materials obtained  from waste
 sites  could be  used  during the  investigation
 and  remediation of sites  having  similar soils
 characteristics, or  they could  be stored  and
 used  throughout  the  study of the  site  from
 which  they were obtained.

 Site-specific     QA    materials    have    been
 successfully   manufactured   and    used   for
 treatability     studies     for     similarly
 characterized   sites,c283  as   single-blind  QA
 samples   with  routine   samples,     and  for
 integration  of   QA   data    (site   comparison
 soils)1293  among  several  projects  on a large (21
 square mile) site for the duration of the site
 investigation  and remediation.

 A disadvantage in preparing site-specific soil
 QA materials is that often  they  cannot  be used
;as double-blind  samples  because their  visual
 characteristics   may   be   altered   by   the
 processing  that  is   employed  to  prepare  QA
 materials.  The site-specific approach  is very
successful, however, when the site is  fairly
dryt151 and sieving  is  not necessary.

CONCLUSION

Increased public interest in environmental issues
has led to new legislation at both the state and
federal levels.  As a result of these laws, many
contaminated  sites  have been  or  will  be
evaluated.  A large number of  these sites have
been  grossly contaminated by  a  variety  of
hazardous chemicals at different concentrations.
A parallel increase in the number of sites added
to  the  National Priority List  (NPL) and the
number  of contaminants regulated by RCRA and
Superfund Amendment Reauthorization Act (SARA)
(CERCLA) and other federal and state regulations
demands  a  comprehensive  suite  of quality
assurance samples111  or a mechanism to produce
such  on short notice.  The QA samples should
represent  the variety  of contaminants  at
appropriate concentrations and  natural  soil
characteristics to provide a true comparison to
real world samples.  The authors of this report
recommend that the rationale document113 described
previously be consulted to determine whether the
information and conclusions presented there pose
serious problems for the investigator. If the
quality  of  environmental  data   cannot  be
adequately assessed because suitable QA materials
do not exist, then more effort clearly needs to
be  made  to increase  the supply  of soil  QA
materials.

Future  research should  include a preliminary
study  comparing  approaches  for   producing
realistic soil QA materials.  It  is felt that
such  a  study may show that the site-specific
approach produces  the  most useful  soil  QA
materials.  A multi-laboratory pilot  study  would
evaluate the advantages and disadvantages of each
approach and should lead to a long-term plan for
providing a supply of soil QA materials.128'293

                        NOTICE

Although the research described in this paper has been funded wholly
or in part by the United States Environmental Protection Agency inder
Cooperative Agreement No. CR 814701 with the Environmental Research
Center of the University of Nevada, Las Vegas, it has not been
subjected to Agency review and therefore does not necessarily reflect
the views of the Agency and no official endorsement should be
inferred.

                     REFERENCES

 1.     U.S. EPA.  1989.   "A Rationale for the
       Assessment of Errors in the Sampling  of
       Soils."  EPA 600/X-89/203, Environmental
       Monitoring Systems Laboratory-Las Vegas,
       NV.
                                                   239

-------
2.    Hertz, H.S.   1988.   "Quality Assurance,
      Reference Materials,  and the  Role  of a
      Reference  Laboratory  in  Environmental
      Measurements."       Proceedings,    The
      International Symposium on Trace Analysis
      in  Environmental  Samples and Standard
      Reference Materials.   Honolulu,  HI, pp.
      5-8, January  6-8.

3.    Taylor,  J.K.     Quality  Assurance  of
      Chemical Measurements.  Lewis Publishers,
      Inc., Chelsea, MI,  1987,  pp.159-163.

4.    Seward, R.W., editor.  Standard Reference
      Materials and Meaning/Measurement.   NBS
      SP  408.   National  Bureau of  Standards,
      Gaithersburg,  MD,  1973.

5.    Cali,   J.P.      The  Role  of  Standard
      Reference   Materials    in   Measurement
      Systems.   National  Bureau  of Standards
      Monograph  148.   NBS,  Gaithersburg, MD,
      1975.

6.    Steger,   H.F.      Certified  Reference
      Materials  Report 80-6E.   Canada Centre
      for Mineral and Energy Technology, Ottawa
      Canada,  1980.

7.    Taylor,  J.K.   Handbook  for SRM Users.
      NBS SP  260-100.    National  Bureau of
      Standards,  Gaithersburg, MD, 1985.

8.    U.S.  EPA.    1984.    "Quality Assurance
      Support:  Project Plan for the Superfund
      Standards     Program."       Tr-506-112A
      (Internal Report).   Project  Officer J.G.
      Pearson.

9.    Bowman, M.S., G.H. Faye, R. Sutarno, J.S.
      McKeague,  and H.  Kodema.   5o/7 Samples
      SO-1,  SO-2, SO-3,  and 50-4:  Certified
      Reference Material.  Report 79-3. Canada
      Centre for Mineral and Energy Technology,
      Ottawa  Canada,  1979.

10.   Stoch,   H.,   and   E.J.   Ring.      The
      Preparation  and Analysis  of Reference
      Materials    and    the   Provision   of
      Recommended Values. Progress  Report No.
      5,  Report No.  M.   Council  for Mineral
      Technology, Randburg,  South Africa,  1983.

11.   Holynska,  B., J. Jasion, M. Lankosz, A.
      Markowitz,  and  W.  Baran.     "Soil   SO-1
      reference  material for  trace  analysis."
      Fresenius    Z   Analytical    Chemistry,
      322:250-254,  1988.

12.   Campana, J.E.,  D.M. Schoengold,  and L.C.
      Butler.     "An  environmental   reference
      material  program:    Dioxin  performance
      evaluation materials." Chemosphere 18(1-
      6):169-176, 1989.

13.   Inn, K.G.W., W.S.  Liggett,  and  J.M.R.
      Hutchinson.   "The National Bureau  of
      Standards  Rocky  Flats  Soil  Standard
      Reference Materi al." Nuclear Instruments
      and Methods in Physics Research 223:443-
      450, 1984.
14.
Jorhem, L.,  and S. Slorach.  "Design and
use  of quality  control  samples  in  a
collaborative study of trace metals in
daily diets."  Fresenius Z Analytical
Chemistry,  322:738-740,  1988.
15.   Thiers, R.E., G.T. Wu, H. Reed, and L.K.
      Oliver.  "Sample stability: A suggested
      definition and method of determination."
      Clin. Chem.  2212:176-183,  1976.

16.   McKenzie,  R.L.,  ed.    "NIST  Standard
      Reference Materials Catalog 1990-1991."
      NIST Special Publication 260, January,
      1990.

17.   U.S. EPA.  "Annual  Summary  Report FY89,
      Quality Assurance in Support of Superfund."
      EPA600/X-90/033, Environmental Monitoring
      Systems Laboratory-Las Vegas, NV, February,
      1990.

18.   Frank, D.J.  "Blind sample submission as a
      tool for  measurement control." Institute
      of Nuclear Materials Management, 14(3):112-
      117,  1985.

19.   Glenn, G.C., and T.K. Hataway.  "Quality
      control  by  blind  sample  analysis."
      American Journal of Clinical Pathology
      72(2):156-162, 1979.

20.   Rumley, A.G.  "External Quality Assessment
      (EQA):   The  effect  and  implications of
      favourable  treatment of EQA samples."
      Medical Laboratory Sciences, 41:295-298,
      1984.

21.   Bleyler,  R.  Viar, and Company.  "Survey of
      Quality Control for  Superfund Programs."
      April, 1989.

22.   Gaskill,  A.      "News   and   Views:
      environmental   reference  standards."
      Environmental  Lab,  Z:  12-15,  1990.

23.   Butler, L.C.  Personal communication.  U.S.
      EPA,  Environmental Monitoring Systems
      Laboratory-Las Vegas, NV,  1990.

24.   Krieger, J.  "Hazardous waste management
      database starts to take shape." Chemical &
                                                   240

-------
      Engineering News,  pp. 19-21. February 6,
      1989.

25.    McCoy,  D.E.     '"301"  Studies  provide
      insight  into  future of  CERCLA.'   The
      Hazardous  Waste Consultant, March/April
      1985,  McCoy  and  Associates,  Lakewood,
      Colorado,  Vol.  3/2:  18-24,  1985.

26.    U.S.  Department  of Agriculture.   Land
      Resource and Major Land Resource Areas of
      the   United   States.       U.S.   Soil
      Conservation    Service,    Agriculture
      Handbook  296,  1981.

27.    U.S.  Geological  Survey.    The National
      Atlas of  the United States of America.
      Department of the Interior.  Washington,
      D.C.   pp.  85-88, 1970.

28.    Esposito,  P.,  J. Hessling, B. B.  Locke,
      M.  Taylor,  M.  Szabo,  R.   Thurman,  C.
      Robers,    R.   Traver,  and   E.    Barth.
      "Results   of treatment  evaluations of  a
      contaminated synthetic  soil."  JAPCA 39:
      294-304,  1989.

29.    Barich III, J.J., G. Raab, R, Jones, J.
      Pasmore.     "The  Application  of  X-ray
      Fluorescences  Technology in the Creation
      of Site  Comparison Samples  and   in the
      Design of Hazardous Waste Treatability
      Studies."  First International Symposium
      Field Screening Methods  for Hazardous
      Haste  Site  Investigations,  Symposium
      Proceedings.   Las Vegas, NV, pp.  75-80,
      October 11-13,  1988.
                                                 241

-------
                       TABLE 1. LIST OF GOVERNMENT AND PRIVATE SOURCES FOR SOIL/SOLID QA MATERIALS*
8
SUPPLIER
Environmental
Research Associates
5540 Marshall St.
Arvada, CO 80002
USA
1-800-372-0122
QA
MATERIAL
Sludge
CLP-priority
pollutant in soil
Hydrocarbon
fuel in soil
Total petroleum
hydrocarbons
(TPH) in soil
Benzene,
toluene, ethyl
benzene and
xylene (BTEX)
in water/soil
DESCRIPTION
Certified QC standards in a
sludge matrix for volatile
(Benzene & TCE), semi-
volatiles (5 BNA), pesticides/
PCB, and metal analysis (11
metals)
Certified QC standards in soil
matrix for Superfund volatiles
(6 to 8 VOCs), semi-volatiles,
trace metals, and cyanide
analysis
Standards of gasoline, No. 2
diesel, heating oil, and crude
oil in a soil matrix
Standardized 50 g QC soil
samples, one specifically
designed for analysis of TPH
in soil in the presence of fatty
acids in screw top bottles
QC set containing two
standard concentrates and
one soil matrix
TYPE & CONCENTRATION
RANGE
Volatiles (5-500 ug/kg)
Semi-volatiles (300-30,000 ng/kg)
Pesticides/PCBs (10-10,000
"g/kg)
Trace metals (1-5,000 mg/kg)
Volatiles (5-500 ug/kg; Sealed
ampoule containing VOCs in
methanol to be spiked into 10 g
of soil)
Semi-volatiles (300-30,000 ug/kg)
Pesticides/PCBs (10-10,000
ug/kg)
Trace metals (1-5,000 mg/kg)
20 g QAS containing unleaded
gasoline (5-500 mg/kg)
No. 2 diesel fuel, heating oil or
crude oil (10-5,000 mg/kg)
Standard 1 - 50 g (100-2000
mg/kg) level
Standard 2 - the presence of fatty
acids (100-2000 mg/kg)
Ampulated 5-500 ug/kg in
CH3OH to be spiked onto 10 g
soil
APPLICATION
40 CFR 503
Evaluation of
laboratory
performance -
especially for
CLP-type
analysis
Evaluation of
specific analysis
for Underground
Storage Tanks
(UST program)
UST program
UST program

-------
TABLE 1.    LIST OF GOVERNMENT AND PRIVATE SOURCES FOR SOIL/SOLID QA MATERIALS (Continued)
      SUPPLIER
      QA
  MATERIAL
      DESCRIPTION
TYPE & CONCENTRATION
         RANGE
                                                                                                APPLICATION
 Fisher Scientific
 711 Forbes Avenue
 Pittsburgh, PA 15219
 USA
 (412)  562-8300
Solid waste
Real world samples,
homogenized for consistency
and tested for accuracy
                                            Fly ash (4 metals)

                                            Waste water treatment
                                             media (3  metals)

                                            Diatomaceous earth filter cake (4
                                            metals)

                                            Circuit board coating sludge
                                             (5 metals)
                                                                 Electroplating tank bottoms
                                                                   (5 metals)

                                                                 Raw sludge, chrome plating
                                                                   process (4 metals)

                                                                 Incinerated sludge (5 metals)
                                                                 Municipal incinerator ash (8
                                                                  TCLP metals, 4-4000 ppm)
                                                                 PAH-contaminated soil
                                                                  (14 PAH and PCPs, 20-1200
                                                                  ppm)

                                                                 Custom Orders
                             SW846

                             Water treatment
                             facilities

                             SW846
                                                          Waste from
                                                          electronic
                                                          industries

                                                          Waste from
                                                          electroplating

                                                          Waste from
                                                          electroplating

                                                          Waste from
                                                          incinerators

                                                          SW 846,
                                                          Methods 3050,
                                                          6010

                                                          SW 846,
                                                          Methods 3540,
                                                          3550

                                                          As required

-------
TABLE 1.   LIST OF GOVERNMENT AND PRIVATE SOURCES FOR SOIL/SOLID QA MATERIALS (Continued)

SUPPLIER
National Institute of
Standards and
Technology
Chemistry Bldg. B-
311
Gaithersburg, MD
20899 USA
302-975-6776










QA
MATERIAL
Ore, minerals,
and refractories








Solid organics








DESCRIPTION
QC reference materials for
critically important material
balance in mining and
metallurgical industries






QA materials for analysis of
materials for constituent of
interest





TYPE & CONCENTRATION
RANGE
Copper ores (5 metals,
0.03 ppm to 0.84%)

Fluorospar (CaF2) (97.4 to
98.8%)

Iron ores (Fe, 58 to 90.8%)

Bauxite ores (Al, 21.1 to
28.8%)
Powdered lead-based paint
(Pb, 12%)

Trace mercury in coal
(Hg,0.13ng/g)
Lead in refinery fuel
(5 varieties, 11.0 to
780.0 ug/g)

APPLICATION
Mining and
metallurgical
processing







Lead-based paint
analysis

Heavy metals in
fuel




-------
TABLE 1.   LIST OF GOVERNMENT AND PRIVATE SOURCES FOR SOIL/SOLID QA MATERIALS (Continued)
SUPPLIER
National Institute of
Standards and
Technology
Chemistry Bldg. B-
311
Gaithersburg, MD
20899 USA
302-975-6776
QA
MATERIAL
Trace elements
Urban dust
Diesel
particulate
matter
PAH in solid
matrices
Polychlorinated
biphenyls in
sediments
Organics in
marine
sediments
DESCRIPTION
Trace elements in solid
matrices (12 to 42 elements)
Urban dust QA materials for
analysis of organic
constituents
QA materials for analysis of
diesel particulate matter and
its organic constituents
QA materials with variety of
PAHs on solid matrices
QA materials of sediments
contaminated by PCBs
QA materials made of marine
sediment contaminated by
organics
TYPE & CONCENTRATION
RANGE
Urban particulate
(1.0-860 ug/g)
Coal - bituminous
(0.1-100 ug/g)
Coal - fly ash, 4 varieties
(0.2-200 ug/g)
Coal - subbituminous
(0.1-20 ug/g)
Estuarine sediment
(0.5-375 ug/g)
Buffalo River sediment
(0.1-555 ug/g)
10 g
100 mg/ampoules
6 varieties, 1.0-4000 ug/g
In preparation
In preparation
APPLICATION
Evaluation of
laboratory
performance
especially for
analysis of trace
elements in
variety of
matrices
Air pollution
Air pollution
SW 846 or
similar analytical
programs
SW 846 or
similar analytical
programs
General

-------
TABLE 1.   LIST OF GOVERNMENT AND PRIVATE SOURCES FOR SOIL/SOLID QA MATERIALS (Continued)

SUPPLIER
Canada Centre for
Mineral and Energy
Technology
555 Booth Street
Ottawa, Canada
K1A OG1





United States
Geological Survey
Geochemistry
Branch
P.O. Box 25046 MS
973
Denver Federal
Center
Denver, CO 80225


U.S. Environmental
Protection Agency
RREL, Releases
Control Branch
Edison, NJ
08837-3079 USA
201-321-4372
QA
MATERIAL
Soil Samples
SO-1, SO-2,
SO-3, SO-4








GXR-1-6










Synthetic Soil
Matrix/I


Synthetic Soil
Matrix/II


DESCRIPTION
Compositional Reference
Materials









Jasperoid soils, Cu millhead
tailings, B horizon soil









30% clay, 25% silt, 20% sand,
20% topsoil, 5% gravel
High organic, low metal

Low organic, low metal


TYPE & CONCENTRATION
RANGE
Clayey soil, sandy podzolic B
horizon with a high organic
content, a calcareous till, and a
chernozemic A horizon







Chemical and physical soil and
mineral properties









Organic: 400-8200 mg/kg
Metal: 10-450 mg/kg


Organic: 40-820 mg/kg
Metal: 10-450 mg/kg


APPLICATION
General
analytical and
earth science for
agricultural,
forestry, and
environmental
applications,
especially for
mining and
metallurgical
operations.
General
analtyical and
earth science for
agricultural,
forestry, and
environmental
applications,
especially for
mining and
metallurgical
operations.
Soil treatability
studies


Soil treatability
studies


-------
TABLE 1.   LIST OF GOVERNMENT AND PRIVATE SOURCES FOR SOIL/SOLID QA MATERIALS (Continued)
SUPPLIER
U.S. Environmental
Protection Agency
RREL, Releases
Control Branch
Edison, NJ
08837-3079 USA
201-321-4372
U.S. Environmental
Protection Agency
EMSL-LV, QAD
P.O. Box 93478
Las Vegas, NV
89193-3478
702-798-2114
FTS 545-2214
U.S. Environmental
Protection Agency
EMSL-LV, QAD
P.O. Box 93478
Las Vegas, NV
89193-3478
702-798-2114
FTS 545-2214
QA
MATERIAL
Synthetic Soil
Matrix/Ill
Synthetic Soil
Matrix/IV
Dioxin
performance
evaluation
materials
Base-neutral-
acid PEMs
Pesticide PEMs
DESCRIPTION
Low organic, high metal
High organic, high metal
Real World samples
contaminated by dioxin and/
or selected matrices fortified
by dioxin
Sand fortified with selected
BNAs
••„
Real world samples
contaminated with toxaphene
and other pesticides or
selected soil fortified by
selected pesticides & PCBs
TYPE & CONCENTRATION
RANGE
Organic: 40-820 mg/kg
Metal: 500-22,500 mg/kg
Organic: 40-8200 mg/kg
Metal: 500-22,500 mg/kg
Kiln ash, XAD Resin, filter
paper, florisil, clay, sand (20 ppt
to 6 ppb)
TCDD/PCDF soil
Times Beach soil
Times Beach & PCDD/PCDF
soil
Times Beach & Region 9 soil
Low level BNA (400 ppb)
Medium level BNA (15 ppm)
High level BNA (75 ppm)
Mixed level BNA
Toxaphene soil
Pesticide soil 1 (4-40 Hg/kg)
Pesticide soil 2 (4-40 ug/kg)
Pesticide soil 3 (30-100 ug/kg)
+ PCB 1016
Pesticide soil 4 (30-60 Ug/kg)
+ PCB 1266
APPLICATION
Soil treatability
studies
Soil treatability
studies
SW 840, 8280
SW 846, 8250,
8270
SW 846, 8080

-------
TABLE 1.    LIST OF GOVERNMENT AND PRIVATE SOURCES FOR SOIL/SOLID QA MATERIALS (Continued)
SUPPLIER
U.S. Environmental
Protection Agency
EMSL-LV, QAD
P.O. Box 93478
Las Vegas, NV
89193-3478
702-798-2114
FTS 545-2214
QA
MATERIAL
Inorganic PEMs



DESCRIPTION
Selected soil samples fortified
with metals and cyanide



TYPE & CONCENTRATION
RANGE
LCS metals (1 ppm-200,000 ppm)
LCS, cyanide (4-8 ppm)



APPLICATION
SW 846, 6010



      Information containedin this table was obtained in September 1990 and may not include some sources of QA materialsTaespite
      the authors' efforts to be accurate and complete.

DISCLAIMER: Mention of trade names or commercial products does not constitute endorsement or recommendation for use.

-------
TABLE 2.  SOIL AND WATER PE SAMPLES NEEDED BY THE 10 REGIONS OF THE U.S. EPA™.
Region
I
II
III
IV
V
VI
VII
VIII
IX
X
Analytes
VOA, BNA, PEST/PCB
soil blanks for VOA and BNA
Dioxin
Unspecified
«TCE 25 ppb
toluene
vinyl chloride
phenols
napthalene
pentachlorophenol
2 or 3 mixes for each fraction; e.g.
5 analytes
7 analytes
3 analytes (determine in workgroup)
•VOA and BNA from CLP-TCL
PEST/PCBs
Metals
VOA and BNA
case by case; not routine enough to predict
levels or analytes
PCBs
Pest/Herb
PCP
TCE and solvents
dioxin congeners, tetrachloro-specific isomcrs
Complete TCL (grouped aromatics, PAH,
etc.)
EDB
RDX explosives
TCDD only
PCDD/PCDF
chloroform, carbon tetrachloride
BTX
chlorinated hydrocarbons
VOA and BNA
Heavy metals
include most common and possibly some
more difficult compounds
PAH (sediment)
Levels
same as CLP PE
no detectable levels
isomer specific; not only 2,3,7,8
Unspecified
100 ppb
100 ppb
100 ppb
50 ppb
100 ppb
2 x (CRQL)"
5 x (CRQL)
10 x (CRQL)
-1.5 ppb
-CRQL
-CRQL
•CRQL
100-80,000 ppm (soil)
300-10,000 ppm (soil) ,
300-30,000 ppm (oily matrix)
low ppb (water)
Low (10 x CRQL)
Med (50 x CRQL)
100 ppt; 1 ppb
Ippb
1 ppb; 5 ppb; 10 ppb (soil)
10 ppt (water)
10 ppb (soil)
20 ppb
wide variety
high for
high soils;
low for drinking water
asbestos needed but don't expect it
in this effort
low and high (within DOT
regulatory limits)
# PE samples/year
100/type/year
100, or if replace MS/MSD* 1/50
samples

unknown
15-20
15-20
15-20
15-20
up to 100 if convenient and
flexible schedule
200 water, 200 soil
50
20
1500 soil
50 water
50
-30;
contractor!, would like 2
50-75/matrix/analyte set
if replace MS/MSD, 1 per data set
 * Matrix spike/Matrix spike duplicate




 *Soil samples not requested.




 "Contract required quantitation limit
                                                       249

-------
TABLE 3.    VOLUME OF WASTE GENERATED BY INDUSTRIAL ACTIVITIES PER
             YEAR'24'.
Standard Industrial
Classification
2869
2800
2911
2892
2821
4953
2879
2865
2816
2812
Category
Industrial organic chemicals
General chemical
manufacturing
Petroleum refining
Explosives
Plastic materials/resins
Refuse systems (commercial
TSDR* facility)
Agricultural chemicals
Cyclic crudes/intermediates
Inorganic pigments
Alkalis/chlo rine
Hazardous Waste Volume,
Millions of metric tons
60-80
40-50
20-30
10-15
6-10
5-8
5-8
5-8
3.5-5
2.5-4.5
* Transportation, storage, disposal, or recycling
                                          250

-------
TABLE 4.  MOST FREQUENTLY REPORTED SUBSTANCES AT 546 NPL SITES'25'.
Rank
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Substance
Trichloroethylene
Lead
Toluene
Benzene
Polychlorinated biphenyls
(PCBs)
Chloroform
Tetrachloroethylene
Phenol
Arsenic
Cadmium
Chromium
1, 1, 1-Trichloroethane
Zinc and compounds
Ethylbenzene
Xylene
Methylene chloride
Trans- 1,2-Dichloroethylene
Mercury
Copper and compounds
Cyanides (soluble salts)
Vinyl chloride
1,2-Dichloroethane
Chlorobenzene
1, 1-Dichloroethane
Carbon tetrachloride
Percent of Sites
33
30
28
26
22
20
16
15
15
15
15
14
14
13
13
12
11
10
9
8
8
8
8
8
7
                                  251

-------
                                                            DISCUSSION
JANINE ARVIZU: Have you considered as one of your options for preparation
of these materials, reconstruction of some simulated soils from stockpiles of
individual soil constituents (clays and gravels) and so forth? Based on compo-
sitional analysis of the soils, would you be able to reconstruct QA materials on
a site-specific basis?
AMY CROSS-SMIECINSKI: Yes, we have considered this possibility and
have tried to locate large stockpiles of various types of soils. Most of the sources
of soils that we have found are not extensive. They're small volumes and the
people who distribute them are apprehensive about sending out large quantities.
They are used mostly for a routine soil sample analysis.
JANINE ARVIZU: I'm curious as to how you would envision addressing the
problem of accurately dealing with active soils, (e.g., biologically active soils or
natural soils that  have  absorptive  properties) and  being able to accurately
determine  the recovery of analytes from those types of materials?
AMY CROSS-SMIECINSKI: In another study we have in the poster session,
we have looked into various types of soil, specifically volatile organic preserva-
tives, to prevent  those kinds of degradations and  activities. But then it's
something that would be a real problem  for any type of soil QA material.
LLEW WILLIAMS: I might just comment on something we've been wanting
to try to see if we can get better representative spiking into QA materials. I think
this has always been a concern that spiked materials frequently don't reflect in
recoveries for instance. The same analytes, if they were naturally in a waste
material, we may get fifteen percent (15%) recovery, we spike them and then we
get ninety percent (90%) back.

One of the things that we're looking into right now and some of you who have
the facilities might want to play around with it a little bit, too, is looking at the
concept of using super critical fluid to put analytes back into matrices, rather than
taking them out. If the concept is a good one to reach down into the pores and draw
analytes out of a matrix, it's possible to be able to release the pressure and put
analytes deeply into a matrix in a way that they may better assimilate natural
materials.

JANINE ARVIZU: Your concerns about double blind QA samples for soils, I
think are really legitimate. Have you considered the introduction of single blind
QA samples with every analytical batch as an alternative to having a double
blind? Would it serve some of the same purposes?

AMY CROSS-SMIECINSKI: We believe it does and it has. Single blind QA
samples have been used this way for some  time, particularly in the dioxin
program. But we feel that the double blind QA samples, although they're very
hard to manufacture, would be the most realistic type of soil QA samples at this
point.
                                                                       252

-------
                   EVALUATION OF  EMISSION  SOURCES  AND  HAZARDOUS
                    WASTE SITES USING  PORTABLE  CHROMATOGRAPHS
                                   R.  E.  Berkley
                         Environmental Protection  Agency
             Atmospheric Research  and Exposure Assessment  Laboratory
                            Research  Triangle Park,  NO
ABSTRACT

Portable  gas  chromatographs  (PGC)  cap-
able of direct detection of ambient  con-
centrations of  toxic organic  vapors  in
air were operated in field studies while
simultaneous data were taken for  compar-
ison by the Canister/TO-14 Method.   Sam-
ples were obtained downwind of Superfund
hazardous  waste  sites,  highways, chem-
ical  plants,   and  in  locations  where
there was  concern  about  odors or  nasal/
respiratory irritation.    In  some cases
two PGCs  equipped  identically were  used
side-by-side or upwind/downwind.   In ot-
hers, different  columns  were used side-
by-side  to analyze  a  larger  group  of
compounds.  Reasonable agreement  between
methods was found,  even  though sampling
techniques  were  not  equivalent.    Such
agreement  suggests  that  both   methods
were free  of  sampling errors,  and  that
the data were substantially accurate.

This paper  has  been reviewed  in  accor-
dance with  the U.  S.  Environmental  Pro-
tection  Agency's  peer  and  administra-
tive review policies  and  approved   for
presentation  and publication.    Mention
of  trade  names  or commercial  products
does not  constitute endorsement  or   re-
commendation for use.

INTRODUCT I ON

Toxic organic compounds are usually  pre-
sent in  ambient  air at such  low  levels
(typically about one ppB) that they  can-
not  be  analyzed without  preconcentra-
tion.   In  the  TO-14  Method,  six-liter
»ir samples are  collected in passivated
canisters  and  stored  pending analysis.
Just prior to analysis they are cryogen-
ically  preconcentrated (1).   Use  of a
portable gas  chromatograph (PGC) equip-
ped  with   a   photoionization  detector
(PID) sensitive enough to  detect organic
compounds at sub-ppB  levels without pre-
concentration offers  an alternative sam-
ple  collection  method  which  produces
data on-the-spot in near real-time.

PID  detectors  are  no  longer  novel.    In
1984-5  Verner  (2)  and Driscoll  (3)  re-
viewed more than a  decade  of PID use  in
gas  chromatography.   There have  been
several  reports since 1980  describing
analyses of airborne  organic vapors with
them.   However,  none  of  the instruments
were portable,  and  sample  preconcentra-
tion was  always required  because those
PIDs were not  significantly more sensi-
tive  than  other   kinds   of  detectors
(4-8).   Then  Leveson  and  coworkers dev-
eloped  a 10.6  electron-volt PID of sig-
nificantly  greater  sensitivity  and  in-
corporated it into a  PGC (9).  The  light
source  was  an  electrodeless  discharge
tube which  was excited by a radio-fre-
quency oscillator to  produce an intense
emission  line.    The  chromatograph  was
claimed to  detect  benzene  without  pre-
concentration at 0.1  ppB  (10-13).   How-
ever, the lamp  is restricted to low-tem-
perature  operation  because heating   it
would decrease sensitivity by broadening
the  emission  line.   For  Leveson1s  PGC
(Photovac Model  10A10), Berkley  estim-
ated a  benzene  detection  limit equival-
ent  to  0.03 ppB.   The smallest  sample
actually analyzed,  one microliter  con-
taining 1.6 picogram  of  benzene,  produ-
                                          253

-------
ced  a 2.3  volt-second peak  at maximum
gain.  A  linear  response  to benzene was
observed over a wide concentration range
(0.5  to  130  ppB),   and   injections  as
large  as  one  milliliter   could  be  made
without  significant  loss  of  chromato-
graphic resolution.  Similar sensitivity
to  other   aromatic   compounds  and  to
chloroalkenes  was  also  observed  (14).
Such  an  instrument obviously  should  be
useful for  air monitoring,  but  few re-
ports  of  it have  appeared.   Lipsky an-
alyzed  vinyl  chloride   from  landfills
(15),  and Hawthorne  analyzed indoor air
in a  "research house"  (16).  Jerpe  est-
imated a  benzene  detection  limit of  20
picograms  using   a Model  10A10 PGC  to
which  an  external capillary  column  and
constant-volume   sample   loop  had  been
connected  (17).     Users   of  the  Model
10A10  PGC  experienced difficulty  with
battery  endurance, baseline  drift,  and
on-site data interpretation.  These pro-
blems  were  mostly resolved by the later
series of Model   10S-  PQCs.   Since  PGCs
can  be  more  easily  transported  than
large  numbers  of canisters,  they  more
readily produce large  volumes of data in
the field.  Their  disadvantages are that
(a) at present they are  limited  to low
resolution  chromatography,  (b)  they id-
entify, by retention time only, the lim-
ited number of compounds  which they can
detect at  low  ppB levels,  and  (c)  they
require a skilled  operator.

It is  difficult to be  certain that  pre-
concentrated samples are  not being spoi-
led by sampling errors.   Although sample
integrity during   storage  in passivated
canisters has  been demonstrated  in  the
absence  of  highly  reactive  compounds
(18),  artifact formation  can be caused,
for example, by HCl (19).   We have eval-
uated  PGCs  in  both laboratory and field
operation  (20, 2t).    Because PGCs  are
not affected by breakthrough of analytes
from  a preconcentration  trap,  by  chem-
ical  reactions between   collected  com-
pounds, or  by  sample  degradation during
storage,   use  of   them  in  parallel  with
the Canister/TO-14 Method could identify
such  problems,  should they  ever  occur,
if  the two  methods  could  be shown  to
consistently   produce   similar  results
under  field  conditions.    That requires
much parallel  use over a  long period  of
time at a variety of sites under differ-
ent ambient conditions using many  kinds
of operating parameters.  Herein are re-
ported  an  accumulation   of  comparative
data obtained during the  past two years.
EXPERIMENTAL

Spherical  6-liter  electropolished  can-
isters  (SIS,  Incorporated)  were used to
collect air samples and store PGC calib-
ration standards.  Canisters were clean-
ed  by  heating to  90°C  while evacuating
through a liquid nitrogen trap to a fin-
al  pressure  below  10 micrometers  (mer-
cury equivalent)  for two hours.   Samp-
ling for  direct comparison  of  canister
and PGC data  was  done by  holding a can-
ister with its inlet less than 10 centi-
meters from the end of the PGC probe and
opening the valve  to  fill  it during the
time the  PGC  sample pump  was  running.
Another method of comparison was to per-
form  consecutive   PGC   analyses  while
time-integrated  canister  samples  were
being  collected.    For  time-integrated
measurements,   evacuated canisters  were
fitted  with  pre-calibrated  mechanical
flow controllers, and air was sampled at
25 mi 11i1iters/minute for two hours. Air
samples  collected   in   canisters   were
transported  to  a   laboratory,  cryogen-
ically  preconcentrated,   and  analyzed
using  a  modified  Hewlett-Packard  Model
5880A  gas  chromatograph  equipped  with
flame  ionization  and  electron  capture
detectors.      A   Hewlett-Packard  Model
5970A mass  selective detector  was  used
for some samples.  Calibration was based
on  41  organic  compounds  cited in  the
Canister/TO-14 Method (1).

Microprocessor-controlled PGCs (Photovac
Model  10S70)   were  used.     They  were
equipped with  constant-temperature  col-
umn enclosures and  0.53 millimeter  ID X
10 meter fused-silica wall-coated  open-
tubular  (WCOT)  columns,  a   1.67  meter
section of which was  backflushable  pre-
column.    Chemically-bonded  stationary
liquid  phases  were used, either CPSilSCB
or CPSil19CB  (Chrompak).   A KCl/Alumina
porous-layer open-tubular  (PLOT)  column
of the  same  size and configuration  was
used for  extremely volatile  compounds.
Ultrazero air  (less than 0.1 ppM carbon)
was the carrier gas.  An  IBM-compatible
laptop   computer,  using  vendor-provided
software via  an  RS-232  interface,  con-
trolled chromatograph operation and data
storage.  Chromatographic peaks were id-
entified and quantitated using retention
times  and  response  factors  stored  in
nonvolatile memory of the  PGC micropro-
cessor.    The  calibration  library  was
created by  analyzing mixtures of  anal-
ytes (10 ppB)  produced  by  flow-dilution
of  commercially-prepared  standards  as
described above.   Compounds  with ioniza-

-------
tion  potentials  greater than  10.6  elec-
tron-volts were  not detected by  PGCs  at
ambient  (below 10  ppB)  levels.    Before
beginning  to sample,  a stable  baseline
Mas observed,  and the  library  was recal-
ibrated  with a single-compound  standard
(approximately  10  ppB)  which  had  been
certified  by GC/FID  analysis.    Chloro-
benzene  or tetrachloroethylene were used
as calibrants with  the  WCOT  columns,  and
vinyl idene chloride with  the  PLOT  col-
umn.   During sampling, automatic  recal-
ibration was performed  every 4 or 5 runs
using  the  single  compound standard,  af-
ter  which  the  microprocessor  corrected
the  retention  time  and response  factor
for  the  calibrant,   then  corrected  pro-
portionally  the retention times  and res-
ponse  factors of other  compounds.   Samp-
les  were taken  every  15 minutes.    Air
was drawn  into  the  sample probe  (3 met-
ers  long  X  2  millimeter  ID  stainless
steel  tubing) for 45 to 60 seconds. Then
the  sample  was   injected  for  7  to  15
seconds, after which the sample  loop  was
removed  from  carrier  flow  to  minimize
peak  tailing.   The  precolumn  was  back-
Hushed  by  the  carrier  stream  except
iihile  calibrated  compounds  were  passing
through  it.   Calibration  runs  differed
from  sample  runs  only  in  that the  loop
received calibration  mixture instead  of
an air sample.  PGCs were sheltered from
drafts and direct  sunlight   inside  a  ve-
hicle  or building,  and  a stainless  steel
Sample probe was extended through  a win-
dow  or a  sampling  port.    External  re-
phargable  12-volt   batteries    (Johnson
Controls  GC12800 or  PP12120  Gel-Cell,
and Sears  Die-Hard  Marine)  were  used  to
supply power.

RESULTS AND  DISCUSSION

|n comparing Canister and PGC  data  it  is
important to  remember  that  samples  col-
lected by  the  two methods  are  not  equi-
Sfalent.  A PGC analyzes only  one  of  50
to 70 milliliters of air which enter  the
probe  during  sampling,  whereas a  repre-
sentative sample  of the entire six lit-
ers  collected  by the  canister is  anal-
yzed.  If the air is well-mixed  and dev-
oid of reactive  or  corrosive materials,
then canister and PGC data should  resem-
ble each other, and generally  do.   How-
sver,  if a heterogeneous  plume  is  samp-
Jed, or  if highly reactive materials  en-
ter the  canister, then PGC and  canister
tfata   could  differ  significantly   even
though the "same" air was sampled.

Complaints about  episodes of  stench  at
Marcus Hook, PA were  investigated at the
request  of  EPA Region  III.    A  PGC was
operated  in  a  van  at several sites, and
canister  samples were taken  for compari-
son.  The results  are shown in TABLE  1.
The PGC  twice  failed to recognize  small
benzene  peaks  which  eluted  in the tail
of the large initial  peak.   The CPSilSCB
column eluted  compounds so  close toget-
her that  resumption of  backflush always
interfered  with  some  peak,  no  matter
when  it  occurred.   In this  case toluene
was missed.   Trichloroethylene, reported
by the PGC,  was never found in the can-
isters. That peak was undoubtedly due  to
some  other  compound which had a similar
retention time.   For  other  compounds,
agreement between   the  two  methods was
reasonab1e.

TABLE 2 shows samples taken  at hazardous
waste sites near Wilmington  and New Cas-
tle,  Delaware.    Concentrations  at the
Superfund  remediation  sites  were  low,
typical of  sub-ppB  background levels  in
remote areas,  showing that  buried  waste
was not  emitting  significant  quantities
of these  compounds  into the air.    Rela-
tive  agreement  between  PGC  and canister
data  seemed  to improve  with  increasing
concentration.    PGC  data  for  tetra-
chloroethylene  at  the waste lagoon were
not reported because  of a persistent co-
eluting  peak.    Samples  taken  by  both
methods   near   the  waste   incineration
plant show toluene  and higher  homologues
at  significant  levels.    High  levels   of
benzene and  chlorobenzene were found  by
both  methods downwind  of  the Standard
Chlorine  plant.   For compounds found  by
both  methods,  agreement  was  reasonable
over  a wide range of  concentrations.

Under  Project   02.01-12  of  the  US-USSR
Environmental  Agreement,  samples  were
taken at  a  roadside site about 12  kilo-
meters  from  Vilnius,  Lithuania.    Two
PGCs were operated while time-integrated
canister  samples were collected.   A mo-
bile  laboratory stood  about  20  meters
from  the  highway  on  ground  about  2 me-
ters  below  it.   Daytime  traffic volume
was moderate-to-heavy without stop-and-
go  congestion  and  subject  to a  100   km
per hour  speed  limit.  No industrial ac-
tivity was  visible  in the immediate vi-
cinity.   Two  identically equipped  PGC's
were  compared  side-by-side  and then up-
wind/downwind.  During  side-by-side op-
eration   inside the  mobile  laboratory,
the sample  probes   extended  to about   18
meters  from the  roadway and  one  meter
above  it.   TABLE 3  compares  colocated
                                          255

-------
  and  upwind/downwind  PGC  analyses  with
  time-integrated canister  data.    During
  colocated sampling canisters  were  placed
  3 and 10 meters downwind of the  highway.
  Sampling  was   done  during nonturbulent
  movement  of  air  across  the  site   and
  while traffic  density   was  fairly  con-
  stant.   Average levels  of benzene,  tol-
  uene,  ethylbenzene, m,p-xylene  (reported
  as  one  compound)  and  o-xylene  found by
  the PGC's were in reasonable agreement
  with  data from  the  canisters.  PGC  data
  for toluene,  and  sometimes  m,p-xylene,
  exceeded average concentrations found in
  the 10  meter  canisters,  even  though the
  PGCs  were farther  from  the  highway.
  This  discrepancy may have occurred  be-
  cause  the PGCs  often  sampled  the plumes
  of  passing vehicles.  When the PGCs were
  deployed  across the  highway  from  each
  other, PGC-1 was  inside  a van parked 12
  meters   downwind  while  PGC-2  remained
  upwind in  the mobile laboratory.  Canis-
  ters  were again placed  3  and  10  meters
  downwind  of the  highway.    Scheduling
  constraints allowed only a half  hour of
 PGC sampling to be  compared to  the can-
  isters,  but downwind PGC results  agreed
 substantially with canister data.
 At  a  Superfund  remediation   site   in
 northwest  Georgia,   airborne  emissions
 produced strong  odor  but contained  low
 levels of compounds which could  be  det-
 ected by  the  PI Ds.    Two PGCs  equipped
 with  CPSil19CB  columns   were   operated
 side-by-side while canister  samples  were
 taken for comparison.   Data  are shown  in
 TABLE 4.   Toluene and  xylenes  were  con-
 sistently seen by both methods at  sim-
 ilar  levels.    Some  styrene  was   also
 seen.    These  compounds  probably   came
 from  trucks  and  earth-movers on  the
 site.    The  CPSil19CB  columns   provided
 better resolution  than  CPSilSCB columns,
 but  benzene  peaks smaller than one ppB
 were missed  because  the PGC  peak-recog-
 nition algorithm  could  not find them on
 the  tail  of  the large initial peak.

 Compounds which can  be  analyzed without
 concentration  by  a   PGC  are  those  to
 which the PID  is sensitive and  which can
 be separated from  each  other by an iso-
 thermal column  at  low  temperature (50°C
maximum).  The number of compounds which
can  be analyzed can  be  increased by op-
erating two  PGCs  side-by-side with  dif-
ferent columns.  An  example  is  shown in
TABLE 5.    The  site was about 40 meters
downwind of a dry  cleaning plant.  PGC-1
was  equipped  with  a  KCl/Alumina  PLOT
  column  and used to analyze vinyl  chlor-
  ide and vinylidene chloride.   Since  the
  PLOT column had very low bleed, the  PGC
  could  be  operated   at   maximum  gain
  (1000).   PGC-2 equipped with a CPSilSCB
  column  was calibrated for the usual list
  of  compounds.   Traces of vinyl chloride
  and vinylidene  chloride  were  found   by
  PGC-1  but  not  found in  the  canisters.
  These  concentrations were  below   detec-
  tion  limit (approximately  0.2  ppB)   for
  the Canister/TO-14 Method.    PGC   detec-
  tion  limits for vinyl  chloride and vin-
  ylidene  chloride  were  0.005  and  0.010
  ppB,  the  amounts  which would  have pro-
  duced  5  millivolt-second   peaks.     The
  integration algorithm does not  process
  smaller  peaks.    Canister  and  PGC data
  showed  tetrachloroethylene  at  elevated
  concentrations.    They  did  not  agree
  closely, probably  because the  plume   was
  poorly mixed.    To  measure the  extent  of
  agreement between PGC and canister data
  a  criterion for  evaluation is  needed'
 The absolute difference  between  results
 was  chosen because  it  does not  change
 drastically  with  concentration.     For
 each compound,   the averages of  absolute
 differences are  shown  in  TABLE 6.   For
 the  CPSilSCB  column  these  differences
 (from data  in TABLES 1, 2,  and  5)  range
 approximately  from 1  to  2  ppB.    Appar-
 ently, absolute differences do  increase
 slightly with  increasing concentration
 Supposing   they  did not, then  at  about
 100  ppB, relative  differences  would  be
 5%.    At 10 ppB they would be approx-
 imately  10%, and  at  one ppB,  100%.    A
 difference  of 100%  seems  large,  but sup-
 pose one method  reported one ppB of tol-
 uene while  the  other  reported  two ppB.
 That  difference  would  arouse   little
 concern;  the data  would  be considered
 similar   because   both   results   are
 "small".     Detection  limits   for  the
 Canister/TO-14  Method  (about  0.2  ppB)
 prevent  making  such comparisons  at sig-
 nificantly   lower  concentrations.    For
 data  taken  with  CPSil19CB   columns
 (TABLE 4),.  agreement  was  much  better,
 because  those  columns  retain  compounds
 longer and  resolve  them better, so  peaks
 are more likely to  be identified and in-
 tegrated properly.   Agreement  for  ben-
 zene  and styrene  was  poorer  than for
 other compounds  because benzene was lost
 in the tail of  the  initial  peak on every
 run, while styrene was crowded  by an ar-
 tifact peak produced  by  column  bleed.
 PGC  performance  could most  readily  be
 improved  by  using  a column  with  better
 resolution  and   less  bleed,   perhaps   a
thicker-phase CPSilSCB,  which would pro-
                                          256

-------
vide better  resolution of  early-eluting
compounds  and sufficient  space  between
later  peaks  to  accommodate the  minute-
long baseline disturbance  which  erupts
when backflush  resumes.   Improvement  of
resolution will  ultimately  be  limited  by
flow system  configuration.   Another ad-
vantage  of  using  a  column  with  less
bleed would  be  that operation at  higher
gain  could  result  in  lower  detection
limits.

CONCLUSIONS

Portable gas  chromatographs can  rapidly
produce reasonable estimates of  ambient
background  concentrations  of  many  vol-
atile  nonpolar  and  semi-polar   organic
air  pollutants  which  ionize below  10.6
electron-volts.    Because  they   process
data  immediately,   they  are  useful for
evaluation  of   hazardous   waste   sites,
chemical  spills,  and  other  sources  of
airborne organic vapors.   PGC data  gen-
erally  agree well  with  data  from the
Canister/TO-14   Method,  which  provides
further  indication that  the  latter  is
generally valid  for sampling atmospheres
not  contaminated  with  highly  reactive
compounds,  even when  analyses  are de-
layed.    Combined  Canister/PGC analyses
should be  used  at uncharacterized  sites
or  where  highly reactive  compounds are
suspected.  Positive  interferences  could
affect either PGC  or  canister data, but
negative  interferences  might  be   less
likely to influence PGCs because  they  do
not  store  or   preconcentrate  samples.
Furthermore,  when  analyses  using  dif-
ferent  sampling  methodologies   produce
similar results,  a preponderance of ev-
idence  is  created  that  sampling  errors
did  not  occur  and that  data  are  sub-
stantially correct.   Comparison of  can-
ister and  PGC  sampling should be  exten-
ded  to  include  additional  classes  of
compounds, especially  polar  compounds.

REFERENCES

1.  Compendium of Methods for the  Deter-
   mination of  Toxic  Organic Compounds
    in Ambient Air.  Environmental  Pro-
   tection Agency, Atmospheric Research
   and Exposure Assessment  Laboratory,
   Research Triangle  Park,  NC  27711.
   EPA-600/4-84-017.  June  1988.

2.  Verner, P.   J. Chromatogr. 1984,
   300, 249-264.

3.  Driscoll, J. N.    J. Chromatogr.
   Sci., 1985,  23, 488-492.
4.  Driscoll, J. N.; Atwood, E.  S.;  He-
    witt, G. F.  Ind. Res. Dev. ,  1982,
    24,  188-191.

5.  Cox, R. D.; Earp, R. F.   Anal.
    Chem., 1982, 54, 2265-2270.

6.  Rudolph, J.; Jebsen, C.    Int.  J.
    Environ. Anal. Che., 1983,  13,  129-
    139.

7.  Nutmagul, W.; Cronn, D. R.;  Hill, H.
    H., Jr.  Anal. Chem.,  1983.  55,
    2160-2164.

8.  Langhorst, M. L.   J.  Chromatogr.
    Sci. , 1981 , 19,  98 - 103.

9.  Leveson, R.  Ger. Offen. DE  3031358,
    3-19-83.  Leveson, R.  C.  US-
    4398152, 8-9-83.  Leveson,  R. C.;
    Barker, N. J. CA 1158891 A1.  12-20-
    83.

10. Barker, J. J.; Leveson, R.  C.    Am.
    Lab., 1980, 12,  76.

11. Leveson, Richard C.;   Barker, Ni-
    cholas J.  Proc. of the Annu.  ISA
    Anal.  Instrum.  Symp.. 27th,  St.
    Louis, MO, Mar.  23-26  1981.   Paqes
    7-12.

12. Collins, M.; Barker, N. J.    Am.
    Lab., 1983, 15,  72.

13. Clark, A. I.; Mclntyre, A.  E.;  Les-
    ter, J. N.; Perry R.   Intern. J. En-
    viron. Anal. Chem., 1984,  17, 315-
    326.

14. Berkley, R. E.   Evaluation  of Photo-
    vac  10S50 Portable Photoionization
    Gas Chromatograph for  Analysis  of
    Toxic Organic Pollutants in  Ambient
    Air.  EPA/600/4-86/041. PB87-132858.

15. Lipsky, D.   Proceedings of  the  APCA
    Mid-Atlantic States Section  Confer-
    ence, Wilmington, DE April  18-19,
    1983.  Paper D.

16. Hawthorne, A. R.;  Matthews,  T.  G.;
    Gammage, R. B.   Proceedings,  78th
    Annual Meeting - APCA, Detroit.  Ml
    June 16-21, 1985.  Paper  85-30B.

17. Jerpe, J.; Davis A.  J.  Chromatogr.
    Sci., 1987, 25,  154-157.

18. Oliver, K. D.; Pleil,  J. D.;  McClen-
    ny, W. A.  Atmos. Environ.,  1986,
    20,  1403-1411.
                                          257

-------
19. Gholson, A. R.; Storm, J. F.; Jayan-     21.  Berkley,  R.  E.;  Varns, J. L.; Mc-
    ty, R. K. M.; Fuerst, R. G.;  Logan,          Clenny, W. A.; Fulcher, J. Proceed-
    T. J.; Midgett, M. R.  JAPCA 1989,           ings of the  1989 EPA/AWMA Symposium
    39, 1210-1217.                               on Measurement of Toxic and Related
                                                 Air Pollutants, AWMA, Pittsburgh.
20. Berkley, R. E.  Field Evaluation of          PA, 1989, pp.  19-26.
    Photovac 10S50 Portable Photoion-
    ization Gas Chromatograph for Anal-
    ysis of Toxic Organic Pollutants in
    Ambient Air.  EPA/600/D-88/088.
       TABLE  1. MOBILE  PGC  AND  CANISTER  SAMPLING  AT MARCUS HOOK,  PENNSYLVANIA
 April  25,  1990.   PGC  in  van with  probe  one meter  above  roof  on  upwind side.
 CPSilSCB  column.  Concentrations  are  parts per  billion  by  volume.


                   Tri-          Tetra-
                chloro-          chloro-  Chloro-   Ethyl-    m,p-
        Benzene  ethylene Toluene  ethylene benzene benzene Xylene  o-Xylene Styrene

 Market  Street  at  Railroad Overpass.   77°C.
PGC
CAN
PGC
CAN +
PGC
CAN +
Rt.
PGC
CAN
Rail
PGC
CAN +

6
4
7
6
10
ND
.8
.86 2
.7
.59 2
.2
1 3 at Trai 1 er

1
road
4
4
ND
.6
Station
.83 1
.7
ND
ND
. 18
ND
.20
ND
Park
ND
ND
*
15.9
*
15.6
*
20. 1
Street,
*
3.7
SW Parking Lot
.99
ND
*
7.6
4.90
0. 1
ND
0. 1
ND
0. 1
Trai ner ,
ND
0.4
. 77°C.
ND
0.6
0.53
ND
0. 13
ND
ND
ND
PA. 77°
ND
ND

ND
ND
ND
I .4
ND
2.3
0.47
3.4
C.
0.03
0.5

ND
1 .0

6
2
7
7
13

0
1

0
3
ND
. 1
.82
.5
. 14
.0

.84
.8

. 63
.4
ND
ND
ND
1 .4
ND
ND

ND
ND

ND
ND
ND
1 .9
0.36
2.5
ND
4. 1

ND
0.8

ND
1 .7

 +    An appreciable concentration of hydrocarbons  (not calibrated) was observed
      in the canister sample.
 *    Toluene detection by PGC prevented by  incorrect placement of valve time.
 ND   Not detected.  Peak was absent or smaller than 5 millivolt-second.
                                           258

-------
TABLE  2.  PGC  AND CANISTER DATA AT HAZARDOUS  WASTE SITES  IN  NORTHERN  DELAWARE

April, 1989.
mounted i
column .


Samples taken
n a van with probe
Concentrations


Benzene
April 5,
* PGC
CAN
1989
0.59
0.93
Tri-
ch 1 oro-
ethy 1 ene
Grantham
ND
ND
are

at Superfund haza
one meter above
rdous waste sites.
roof on
parts per bi 1 1 ion
Tetra
_
chloro- Chi
To 1 uene ethy 1
Lane

1

ND
.00 <0

by vol

oro- Ethy
upwi
ume .

1-
ene benzene benzene

INT
. 15

0


.60
ND

ND
ND
nd s i de


m.p-
Xy 1 ene

ND
0.62
PGC was
CPSil



Styrene

ND
ND
5CB



o-Xy 1 ene

ND
0.38
Army Creek
* PGC
CAN
Delaware
* PGC
CAN
April 6,
PGC
CAN
PGC
CAN
0.39
0.75
Sand
ND
0.45
1989
ND
1 .00
ND
0.50
ND
ND
& Gravel
ND
ND

0


0
ND
.69 <0

ND
.27
INT
. 15

I NT
ND
0


0
<0
.54
ND - 
-------
    TABLE 3. COLOCATED AND UPWIND/DOWNWIND PGC AND CANISTER OPERATION IN USSR
Vilnius  June 1989.  Colocated:  PQCs in mobile laboratory.  Probes 2.5 cm
apart 18 m from highway.  Canisters sited on same side of road as PGCs and
filled continuously between 1631 and 1815.  Data not shown if either PGC
recalibrating.  Upwind/downwind:  PGC-1 in van 12 meters downwind of roadway
with probe extended 1.5 meters above roof.  PGC-2 in mobile  laboratory.
Canisters downwind of road and filled continuously between 1100 and 1300.
CPSil19CB columns.  Concentrations are parts per billion by volume.
Start
Time

COLOCATED

1631

1701
1716
1731

1801
          Benzene
          Toluene
                                       Ethyl-
                                      benzene
                      m, p - X y 1 e n e
          DATA
          (D
          0.8
 June
 (2 )
 0.8
                    1, 1989
                        2.5
(2)
2.8
          1.3  1.0
          0.6  0.7
          1.1  1.0

          0.7  0.8
          2.7  2.7
          2.1  2.1
          2.4  2.2

          2.3  2.6
ND

ND
ND
ND

ND
(2}
 ND

 ND
 ND
 ND

 ND
(D
0.5
(2)
 ND
                                                    0.7   ND
                                                    0.7   ND
                                                    0.9   ND
                                                    0.5
                             N'D
Average PGC values during the canister sampling oeriod
          0.9  0.8      2.4  2.5      0.0  0.0      0.7  0.0
Canister sample values (distance from roadway in meters)
                        (3) (10)      (3) (10)      (3) (10)
          (3)
          2.1
(10)
 1.1
                        3.1  1.3
                                      0.4  0.2
UPWIND/DOWNWIND DATA June 2, 1989
          (1)  (2)       (1)  (2)
1229      1.3            4.5
1244           6.3
1259      3.3  0.4       7.4
                                      (D
                                       ND

                                       ND
                             (2)

                              ND
                              ND
                       1.2  0.5
             (1)
             2.5

             3. 1
1244           6.3           2.0            ND            ND
1259      3.3  0.4      7.4  1.3       ND   ND      3.1   ND

Average portable chromatograph values during the canister sampl
          2.3  3.3      5.9  1.6      0.0  0.0      2.8  0.0
Canister sample values (distance from roadway in meters)
          (3) (10)      (3) (10)      (3) (10)      (3) (10)
          2.1  1.1      3.1  1.3      0.4  0.2      1.2  0.5
                           o-Xy1ene
                                                                  ( 1 )  ( 2 )
                                                                   ND  0.4
                                      ND
                                      ND
                                      ND

                                      ND
                   0. 1
                   0. 1
                    ND

                    ND
                                                                  0.0  0.1

                                                                  (3) (10)
                                                                  0.5  0.2
                                     ( 1 )   ( 2 )
                                      ND
                                           ND
                                      ND    ND

                                  ing period
                                     0.0   0.0
                                                                  (3)
                                                                  0.5
                                                                      (10)
                                                                       0.2
ND
     Not detected.  Peak was absent or smaller than 5 millivolt-second.
                                         260

-------
       TABLE 4.  SIDE-BY-SIDE  PGC  AND  CANISTER  DATA  AT  LAFAYETTE,  GEORGIA
June 6,  1990.   Shaver's  Farm  Superfund  Site.   PGCs  in  van  were moved  to  several
sites.   CPSil19CB  columns.  Concentrations  are parts per billion by volume.


                  Tri-           Tetra-
               chloro-          chloro-   Chloro-  Ethyl-    m,p-
     Benzene  ethylene Toluene  ethyiene benzene benzene Xylene o-Xylene  Styrene
PGC-1
PGC- 2
CAN
PGC-1
PGC- 2
CAN
PGC-1
PGC- 2
CAN
PGC-1
PGC -2
CAN
PGC-1
PGC- 2
CAN
PGC-1
PGC- 2
CAN
ND
ND
0.7
ND
ND
0.3
ND
ND
0. 1
ND
ND
3.0
ND
ND
0. I
ND
ND
0. 1
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
1 . 10
1 .75
1 .0
0.99
0. 18
0.7
0.59
ND
0.4
0.32
0. 10
0. 2
0.08
ND
0. 1
0. 19
ND
0. 3
ND
ND
0. 1
ND
ND
0.2
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
0.67
0.62
1 .0
0.46
0.69
0.8
ND
ND
0.2
ND
ND
0. 1
ND
ND
ND
ND
ND
0.2
1 .31
1 .84
1 .6
0.28
0.58
0.8
ND
ND
0.4
0.04
0.07
0.2
ND
ND
ND
ND
ND
0.4
0. 12
ND
0.8
ND
ND
0.4
ND
ND
0.2
ND
ND
0.2
ND
ND
0. 1
ND
ND
0.3
*
18.03
5.9
*
13.68
4.5
*
ND
0.2
*
ND
0.2
*
ND
ND
*
ND
0.4

*    PGC-1  was  not calibrated for styrene because of a persistent interfering
     peak  probably caused  by column deterioration.
ND   Not detected.   Peak was absent or smaller than 5 millivolt-second.
                                        261

-------
                  TABLE 5. TANDEM PGC DATA AND CANISTER DATA IN
                     RESEARCH TRIANGLE PARK, NORTH CAROLINA
March 23, 1990.  PGC-1 analyzed vinyl chloride and 1 , "i-dich loroethy 1 ene with
KCl/Alumina PLOT column.  PGC-2 analyzed other compounds with a CPSil5CB column.
PGCs in car with probes 1.5 meters above the on upwind side, 40 meters downwind
of dry-cleaning plant.  Concentrations are parts per billion by volume.


              1,1-Di-                      Tetra-
      Vinyl-  chloro-                     chloro-
    chloride  ethylene  Benzene  Toluene  ethylene m,p-Xylene  Styrene  o-Xylene
PGC
CAN
PGC
CAN
0.01
ND
ND
ND
0.02
ND
0.01
ND
ND
0.63
ND
0.76
ND
0.56
ND
0.62
1 .41
3.38
4.09
3. 15
ND
0.35
ND
0.29
ND
0.20
ND
ND
ND
0. 20
ND
< 0. 20
ND
Not detected.  Peak was absent or smaller than 5 millivolt-second.
       TABLE 6. AVERAGE ABSOLUTE DIFFERENCES BETWEEN PGC AND CANISTER DATA
Absolute values of differences between PGC and canister results for each
compound were averaged.  CPSilSCB data taken from TABLES 1, 2, and 5.
CPSil19CB data taken from TABLE 6, in which two PGC values for each analysis
were averaged.  CPSilSCB is methyl si "ii cone.  CPSil19CB is 7% cyanopropyl-
silicone, 1% pheny1si1icone, 85% methyIsi1icone. and 1% vinylpolysiloxane.
Phase thicknesses 2 micrometers.  Differences have dimensions of Darts per
billion by volume.
Compound
Benzene
Tr ich loroethy lene
Toluene
Tetrach loroethy lene
Ch lorobenzene
Ethy Ibenzene
m,p-Xy lene
o-Xy 1 ene
Styrene

	 UO 1
CPSi 15CB
1 .49
2.03
0.75
1 .27
1 . 17
0.97
1.69
0.40
1 .53

umn 	
CPSi 1 19CB
0.72
*
0. 15
0.05
*
0.20
0.30
0.31
3.69

     No data were available for these compounds.
                                          262

-------
                                                           DISCUSSION
Q)WARDFURTAUGH:WealsohaveaPhotovaclOS70.Iuseditinasmoking
lounge and at a gain of 100, there was a monster peak occurring near the retention
time of toluene. Any suggestions what it could have been? The Photovac people
Ait I've talked to haven't been able to shed much light on it.

RICHARD BERKLEY: Indoor air is a pretty tough thing to deal with. You
normally see ambient background levels of things like toluene as a result of single
photon absorptions. In indoor air you can have ppm concentrations so you can
teethings which are ionized in double photon absorptions. You can, for example,
calibrate things like carbon tetrachloride and chloroform, which you can't see at
iBatambient background levels. So, in indoor air all bets are off, and I have seen
some horrendous things in indoor air which are probably relatively high levels
ofthingsthat the instrument normally can't see. If it didn't have the retention time
of toluene, and if you were using a constant temperature column accessory the
dunces are very good that it was not toluene. It may be a much larger level  of
something else.

TOM SPITTLER: We just did an air study of our building in Boston using the
Photovac. And we found 1 to 2 ppb of benzene and toluene inevery place because
it's > very well ventilated building. But in the smoking room we found about 100
ppbof toluene and about 50 of benzene. There wasn't any question, the retention
tees matched beautifully. We took samples back and confirmed them on GC/
MS. You get benzene and toluene in all smoking rooms. I'm not sure why  your
(leak wasn't exactly there, but I bet anything that's  what it was.  A question
tough: you were using canisters  and the Photovac  with what?  Occasional
sampling or regular sampling? How often did you sample with the Photovac in
onfcr to cover the period of time you were drawing the canister sample?

RICHARD BERKLEY: In most cases what I did was take a  canister  grab
sample by holding the canister within ten centimeters of the tip of the Photovac
Make and opening the can so that it filled during the same time, during the minute
or so that the Photovac pump  was running.  These  samples are  necessarily
nonequivalent. In the canister you get six liters, and you take a representative
sample of that to analyze it. The Photovac  takes a milliliter or something that
happened to be flying through at the moment when it decided to inject. These are
not equivalent, but if the same air is being sampled they ought to resemble each
other.

TOM SPITTLER: Yes, I agree. I think it's really a nice correlation.

RICHARD BERKLEY: So, in all cases expect variances while taking those
{tab samples. Variances were seen  with two-hour integrated canister samples,
and we were taking Photovac runs during the time.

TOM SPITTLER: You just averaged them then?

RICHARD BERKLEY: Well, the canister samples were shown as dotted and
•lashed lines because they were the time-integrated samples. If you could go back
and compare those slides, you'd find that all those dotted and dashed lines  were
a the same level on all of those slides. We couldn't quite figure out how to show
4e continuity there.

TOM SPITTLER: No, I thought it was really nice data. This afternoon a couple
of guys from the Regional Lab up in Boston are going to show some Photovac
versus canister standards and calibrated by different  techniques. They are
actually directly comparable samples, and you see the same basic kind of
correlation. It may be a little tighter now because they're sampling exactly the
same way and they're sampling the same known mixture of air.

RICHARD BERKLEY: Something I forgot to mention and it'll be important
to some people, we are using canisters to hold our calibration standards, and
we're preparing the standards the same way we prepare the standards for the
method that is used to analyze the canisters. There is no independence on that
point. These two methods are locked together, and if we make a mistake on one,
we make a mistake on the other. What's independent here is sampling methodology,
and I should have said that.

JOSEPH EVANS: My question pertains to detection limits. I notice that you're
down measuring at very low levels (1,2 ppb). Your worst agreement was at those
levels. When you got to the higher levels you had much better agreement. And
I was wondering about how close you were to your detection limits for the two
different methods?

RICHARD BERKLEY: Well, there are two limits to talk about here. One of
them is detection limit, and for single photon ionizations, compounds that ionize
well below 10 eV, such as benzene, its homologs and the chloroethylenes. We
measured detection limits by extrapolation, three times the baseline noise, using
an old  10A10 with a gain turned all the way up, and it appeared that the absolute
detection limit was somewhere in the neighborhood of 18 femtograms. That
would translate out to down in the neighborhood of 1/100 ppb in a 1 mL sample.
That's just a detection limit.The instrument in fact will refuse to process any peak
that is smaller than five millivolt seconds. And of course, when we did that
detection limit we were only extrapolating — our smallest sample was 1.6
picograms, and it produced a peak of about 2.3 volt seconds. We were no where
near this extrapolated detection limit with any sample we actually delivered to
the instrument. So, we're just guessing. But, we do have a substantial basis to
guess that a 5  millivolt second peak is way above that. And all you have to do if
you want to really get tough about what the detection limit is, is to run a sample
on a blank  library, then shift to a calibrated library and  calculate how much it
would take to make a five millivolt second peak, assuming linear response.

JOSEPH EVANS: What levels were your calibration standards?

RICHARD BERKLEY: We generally try  to use between 10 and 20 ppb.  If
something is very convenient to prepare like chlorobenzene or tetrachloroethylene,
we like to use one of them on one instrument and the other one on the other
instrument, because there is some tendency, if the calibration gas valve is a little
bit weak, to have some carryover contamination, usually no more than 0.5 ppb.
You do need to look at that for your standard, whatever your standard compound
JOSEPH EVANS: You were actually measuring below your lowest calibration
standard?

RICHARD BERKLEY: We're using a single point calibrations. We did a lot of
work on this thing early on and found that we were getting pretty consistent linear
responses from as low a sample as we could inject all the way up to higher than
we could inject.
                                                                     263

-------
                    HIGH SPEED GAS CHROMATOGRAPHY FOR AIR MONITORING
                  Levine,  S.P.  (A,*),  Ke, H.Q. (A), Mouradian, R.F. (A)
                          Berkley.  R.  (B) and Marshall, J. (C)


                 A) Department of Environmental  and Industrial  Health,
                 University of  Michigan, Ann Arbor, Michigan 48109-2029
                  (B)  U.S. EPA,  AREAL/MRB (MD-44), 79 TW Alexander Dr.,
                           Research Triangle  Park,  NC   27709
                (C)  HNU Systems,  160 Charlemont, Newton Highlands, MA 02161
                 (*) Author to  whom correspondence should be addressed.
Abstract

Gas chromatography  has  the potential
to  be   a   much   faster  method   of
separation  than  is usually realized.
If  column  operating conditions  are
optimized  for  speed  and  injection
band width is  minimized, some  simple
separations can be completed in  a  few
seconds. In  the work  described here
the system  was evaluated using  common
organics      including      alkanes,
aromatics,    alcohols,    ketones   and
chlorinated              hydrocarbons.
Quantitative  trapping and reinjection
was   achieved    for    all    tested
compounds.   Limits of detection  (LOD)
for many compounds,  based  on  a  1  cm3
gas sample, were less than  1 ppb,  but
for one  carbon-chlorocarbons  the  LOD
when   using   a   flame    ionization
detector was inadequate.    By  using
the cold trap  inlet with  a low dead
volume  detector  and  a  high   speed
electrometer,      the       efficiency
available  from  commercial  capillary
columns  can  be  better  utilized  and
retention  times   for   some  routine
separations  may  be reduced to  a  few
seconds.
Introduction

Gas chromatography  (GC)  is  often  used
for  routine,  repetitive  analysis  of
simple mixtures.    For  some of these
applications,  the  use  of  2 to  5  m
capillary columns operated  at  linear
velocities of  100 to 200 cm/s  offers
the possibility  of greatly  decreased
analysis  times.    This  potential  for
high   speed    analysis   has    been
documented  in the  literature  (1-7).
Under  optimal  conditions,  a 0.25  mm
i.d.  column  should  be  capable  of
achieving  5000   to  7000   effective
plates with  retention times of  5  to
10  seconds  (4,8).    Although  this
number of plates is  low compared  to
most   capillary    systems,   it    is
comparable  to the  number  of  plates
achieved   by   many  packed   column
systems   with  retention   times   of
several  minutes   or more.  Therefore,
some routine  GC  separations that are
currently   performed   using   packed
columns or non-optimized  open tubular
columns   could   be  performed   much
faster with  a capillary system  that
is optimized for speed.

While  the  theoretical   potential  of
capillary  columns   for  high  speed
analysis  is  well  known,  limitations
in  commercially  available  equipment,
especially    inlet    systems,    have
prevented general application of  high
speed    techniques.       With    most
commercial   instruments,   the  major
                                        265

-------
factors that  limit  analysis  speed  are
the   width   of  the   initial  band
produced by  the inlet system  and  the
response  time  of  the  electrometer.
Efficient  separation  with  retention
times of 5 to 10  seconds  and a column
diameter  of   0.25  mm   requires   an
initial band  width of about 20 ms  or
less  and  an  electrometer   response
time of about 5 ms.   For  purposes  of
comparison, most  capillary GC  systems
produce  injection band  widths of  50
to  500 ms  and  feature  electrometer
response times of 150 ms or longer.

In  response  to  the  requirement  for
narrow  injection  bands,  a number  of
experimental    inlets    have   been
described  (5,  9-13).  Our  group  has
described a  prototype cold trap that
was used as a vapor collection device
and  which   may  also   serve  as   a
focusing system for rapid  analysis  of
simple mixtures  (14-15).   The  design
reported by our group, which  expanded
on the innovative work of  Hopkins  and
Pretorius (16),  featured a cold trap
that was cooled by a continuous flow
of cold nitrogen, and  was  resistively
heated  using  a  current  pulse. This
design was  a  marked improvement over
that  reported  earlier,  which  had  a
number of unrecognized serious flaws
that   prevented    reliable    and/or
quantitative operation (17-18).  More
recently,  van Es  et  al  described  a
fast  GC   system   that   utilized   a
similar inlet  (19).  In their  design,
a 50 micron capillary column was used
for the separation.

Experimental Section

The design and operation of the cold
trap  is  given  in detail  elsewhere
(14,15),  and   is  shown schematically
in Figure 1.

Operating       conditions       and
chromatographic   equipment.      All
chromatograms      were     collected
isothermally  at  column   temperatures
of 35 to 60 °C using a 5 m long, 0.25
mm  i.d.  fused  silica column  with  a
0.1  micron  bonded  methyl  silicone
stationary  phase   (Quadrex).     The
carrier  gas  was hydrogen,  which  was
supplied at  a flow rate of  2.5 to  3
ml/min  to  produce  linear  velocities
of 85 to 102 cm/s.   The injector  and

detector were heated to  225 °C.    A
flame  ionization  detector  (FID)  was
used in  all  experiments.  To minimize
the effective dead volume, the  column
was moved  close to  the base  of  the
flame.   Both a  Varian 3700 or  an  HNU
301 GC were used.

For   trap    recovery   studies,   test
mixtures      were    prepared    either
without  solvent or  in high   purity
carbon  disulfide provided  by The  Dow
Chemical  Company.     The  injection
volume  was   2.5 uL  in  all cases  and
the  split  ratio   ranged   from   about
50:1 to  500:1 depending on the  sample
concentration.  For   vapor  studies,
samples  were injected  in  humidified
or  laboratory  air   in  volumes   of
0.025-1.0 cm3.

Results and Discussion

Design  Considerations.  A  number   of
design  considerations were  found  to
be   important  in   determining   the
durability   and  performance  of   the
system.  The  choice  of  trap material
and dimensions affects durability  and
reinjection  performance.    An   ideal
material would  have  high  electrical
resistivity, low chemical activity,  a
low coefficient of thermal expansion,
would be highly malleable  and  would
not  work   harden.      A   number   of
materials, including  stainless  steel,
nickel,  platinum,  Monel 400,  and  an
alloy  of thirty  per  cent  copper  -
seventy   per   cent   nickel   were
evaluated for use as  trap tubes.   The
work reported  here was done  using  a
trap  made  of  Monel  400.   Stainless
steel,  which was  used  in  some  early
studies   (17,   18),  is   the   least
desirable   choice   because   of   its
tendency  to  work  harden  and   become
brittle.  For a  trap  made  of   hard-
tempered Monel  400  with  an internal
diameter of  0.25 mm,  a wall thickness
of   0.18    mm   provided   a   good
combination     of     strength     and
performance.
                                       266

-------
Trapping  and Reinjection  Efficiency.
Cold  traps  have been  used in  GC for
many years  (19-23).   Since the short,
open  tubular   trap  used   in  these
experiments  may   be   less  efficient
than  some   other  designs   (23),   a
careful   evaluation    of    trapping
efficiency was necessary.

In   order   to   test   trapping   and
reinjection  efficiency,  samples  were
injected without  using the  cold  trap
and   average    peak    areas    were
calculated  for  each  compound.    In
addition  to  comparing  peak  areas
obtained  with  and without  trapping,
the FID response was monitored during
the  entire  process   to  allow   any
breakthrough  of   the  sample   to  be
detected.   At temperatures of  -100  °c
or   colder,   each  of   the   tested
compounds was  guantitatively  trapped
and   reinjected.         Peak    area
reproducibility for all compounds was
very  good   with   coefficients    of
variation ranging   from  1  to  5  per
cent, or  less  in  all  cases in which
trapping was used.

Compounds  tested were  (given  in order
of    increasing    boiling    point):
isoprene,   pentane,  dichloromethane,
acrolein,chloroform,methanol,  hexane,
carbon  tetrachloride,  acrylonitrile,
2-butanone,benzene,propanol,  heptane,
i-octane,      toluene,     n-butanol,
tetrachloroethylene, octane,  m-  &  o-
xylene,  nonane,  4-ethyltoluene,  and
1,3-dichlorobenzene. Detailed  results
are  given   elsewhere   (15).  Trapping
efficiency was  also measured  for  1%
solutions  of aromatics  prepared  in
carbon  disulfide.     The  trapping
efficiencies   obtained    in   those
experiments   were  not  significantly
different  than  those measured  without
solvent.   These   materials   can    be
effectively  trapped and reinjected  at

temperatures of     -100 °C.  However,
trapping  behavior  is  not  easily
predicted  on the  basis  of   boiling
point or freezing  point,  and in  most
cases an  effective temperature   must
be experimentally  determined for  each
type   of  sample.     Highly  volatile
 materials,  which may be gases at room
 temperature,    and   low   volatility
 materials,  which may be  difficult to
 revaporize,  have not yet been tested
 and  may  be  difficult   to  trap  and
 reinject  with this  system.
 Limit   of    Detection    (LOD).   For
 monitoring    volatile   organics   in
 ambient or  workplace air, the LOD of
 the  method must be  very low.  As  of
 early  November,  1990,  the  LCD's  for
 pentane,   hexane,   heptane,   octane,
 benzene,        toluene,        xylene,
 ethylbenzene,  4-ethyltoluene,  1,3,5-
 /1,2,4-trimethylbenzenes,         and
 chlorobenzene have been  measured  and
 been shown  to be in  the  range of  0.2
 -  5 ppb,  with the most  recent results
 all being <1.0  ppb.  (The drop in  LOD
 has occurred  as a  result of improved
 methodology   as   work  has  proceeded
 over  the  past  few months.  There  has
 not  been  time  to  re-do  some  of  the
 earlier work.)

 All  of these  values were  determined
 based  on  an injection of  a maximum of
 1  cm3  of  air, and the use of  an FID.
 The  LOD  was  calculated  based on  a
 definition    of    three    times   the
 standard  deviation  of the noise.

 One of the  major factors  contributing
 to   the    reduced   LOD   was   the
 optimization  of the  custom-designed,
 high speed  electrometer  supplied  for
 this   project  by  HNU  Co.   A  filter
 setting  of  12   Hz  was   found  to  be
 optimal for GC peaks  in the  retention
 time range of 5-10  seconds.

 Note   that   these   LOD's   are   not
 achievable     for     one     carbon-
 halocarbons.  LOD's  in the sub-20  ppb
 range  for  certain  halocarbons will
 only be achievable  with the  use of  an
 electron   capture   detector   (BCD).
 Unfortunately,   an   BCD   has,    of
 necessity, a  certain internal  volume
that may  significantly  spread  peaks,
 and reduce the advantage  of  the Fast-
GC method. This may require  assays  to
be performed on a 30-60 second basis,
rather than on a 5-10 second basis.
                                        267

-------
In  addition,   it  is  important  to
remember  that the  Fast-GC technique
trades chromatographic resolution for
speed.  Although  the  cost  of  this
trade is reduced by tuning the column
for  high  speed,  low  retention  time
use   (8,14),   the   separation   of
components  of  complex mixtures  may
not always be possible.

Further,  the limitations  imposed by
the  use of  an isothermal  GC method
(necessitated  by  the  short analysis
times)  limit the ability  to monitor
compounds of widely differing boiling
point   simultaneously.   While   this
might be overcome by  flow-programming
methods,  the  extent  to  which  such
strategies   will   allow   effective
ambient air  monitoring  is  unknown at
this time.

Acknowledgements

The    authors    acknowledge    Lauri
Mendenhall   and  George   Capps   of
Prototype Design Inc. for engineering
and   technical   assistance   in   the
development    of    the    capacitor
discharge     power      supply     and
temperature    measurement    devices.
This  research was supported  by U.S.
EPA (AREAL/MRB) cooperative agreement
CR-817123-01-0. Earlier  work leading
to this  stage had been  supported by
the   Centers   For  Disease  Control,
National  Institute   for  Occupational
Safety and Health Grant R-01-OH02303,
the  U.S.   EPA (OER)   R814389-01,  and
The  Dow Chemical Company  Health and
Environmental Studies Laboratory.

References

1.   D.  H.  Desty.  capillary columns:
Trials  tribulations   and  triumphs.
Advances  in  chromatographv.  Vol  1.,
J. C. Giddings and R. A. Keller eds.,
Marcel Dekker, NY, 1965, pp.  199-228.
2.   D.  H. Desty, A. Goldup and W. T.
Swanton.   Performance    of   coated
capillary         columns.         Gas
Chromatography.  N.   Brenner,  J.  E.
Callen and M. D. Weiss eds., Academic
Press, New York, 1962, pp.  105-135.
 3.   J.  C.  Sternberg.   Extra  column
 contributions to  chromatographic band
 broadening.        Advances	   jn
 Chromatoqraphy Vol 2., J. c.  Giddings
 and   R.   A.  Keller,   eds.,   Marcel
 Dekker: N.Y. 1966, pp. 203-270.
 4.   G. Caspar, R.  Annino,  c.  Vidal-
 Madjar and  G.  Guiochon.  Influence  of
 instrumental  contributions   on   the
 apparent  column  efficiency   in  high
 speed gas Chromatography. Anal.  Chem.
 50: 1512-1518 (1978).
 5.   G.  Caspar,  P.  Arpino  and   G.
 Guiochon.   Study  in  high  speed  gas
 Chromatography.  i	Chromatogr.   sci .
 15: 256-261 (1977)..
 6.   A. van Es,  J. Janssen,  R.  Bally,
 C.  Cramers  and  J.  Rijks.   Sample
 introduction in high speed  capillary
 GC ;  Input  band  width  and  detection
 limits. HRC&CC.  10:  273-279  (1987).
 7.   C.  P.   M.  Schutjes,    E.    A.
 Vermeer,   J.  A.  Rijks  and  C.   A.
 Cramers.  Increased speed of  analysis
 in   isothermal   and   temperature-
 programmed  capillary GC  by  reduction
 of  the  column   inner  diameter.   J.
 Chromatoqr.  253: 1-16 (1982).
 8.   R.  Villalobos and R. Annino. The
 computer    aided    optimization    of
 capillary  columns for  minimum  time
 analysis  and  minimum  detectability.
 HRC&CC. 12:  149-160  (1989).
 9.   R.  L.   Wade  and  S.   P.  Cram.
 Fluidic logic sampling and  injection
 system  for  gas  Chromatography.  Anal.
 Chem. 44: 131-139  (1972).
 10.   R. Annino  and J. Leone.  The use
 of coanda   wall   attachment   fluidic
 switches as  GC valves. J. Chromatogr.
 Sci.  20: 19-26 (1982).
 11.   C.   P.   M.   Schutjes,   C.   A.
 Cramers,  c.   Vidal-Madjar  and  G.  J.
 Guiochon.     fast     fluidic     logic
 injection at pressures up to 25  bar
 in   high-speed    capillary    GC   J.
 Chromatogr.  279:  269-277 (1983).
 12.  R. J. Jonker, H. Poppe and  J.  F.
 K.  Huber.   Improvement  of  speed  of
 separation in packed  column GC.  Anal.
 Chem. 54: 2447-2456 (1982).
 13.  R. Tijssen, N.  van  den  Hoed  and
M.   E.   van  Kreveld.   Theoretical
 aspects  and  practical  potentials  of
rapid gas  analysis in  capillary  GC.
Anal.  Chem.  59:  1007-1015 (1987).
                                       268

-------
14.  Mouradian,  R.F.,  Levine,   S.P.,
Sacks,    R.D.   and    Spence,    M.W.
Measurement  of  Organic Vapors at Sub-
TLV  Concentrations  Using  Fast  Gas
Chromatography.    Amer Ind  Hya  Assoc
J.   51:90-95 (1990).
15.  Mouradian,  R.M.,  S.P.  Levine  and
R.D.  Sacks.   Limits of  Detection  and
Recoveries    for    Fast-GC.        i..
Chromatoar.. 28:  643-648 (1990)
16.   B.   J.   Hopkins   and   V.   J.
Pretorius.    Rapid   evaporation   of
condensed    GC     fractions.      «L=_
Chromatoar.  158:  465-469 (1978).
17.  B.  A.  Ewels  and  R.   D.  Sacks.
Electrically heated  cold  trap  Inlet
system  for high-speed GC. Anal.  Chem.
57:  2774-2779 (1985).
18.  L.  A.  Lanning,  R.  D.  Sacks,  R.
F.  Mouradian, S.  P. Levine, and J. A.
Foulke.  Electrically heated cold trap
inlet  system  for  computer-controlled
high-speed  gas  Chromatography.  Anal.
Chem.  60: 1994-1996 (1988).
19.  A.   van   Es,  J.   Janssen,   C.
Cramers   and    J.   Rijks.    Sample
enrichment  in  high speed narrow bore
capillary gas Chromatography. HRC&CC.
11:  852-857  (1988).
20.  G.  Schomburg, H.  Husmann  and F.
J. Weeke. Aspects  of  double-column GC
with   glass   capillaries   involving
intermediate trapping. J.  Chromatogr.
112: 205-217 (1975).
21. J.   A.   Rijks, J.   Drozd and J.
Novak.     J.   Versatile    all-glass
splitless sample-introduction  system
for trace  analysis by  capillary GC.
J. Chromatoar. 186: 167-181 (1979).
22.  D.  Kalman,  R.  Dills,  C.  Perera
and F.  DeWalle.  On-column  cryogenic
trapping  of  sorbed   organics  for
determination  by  capillary GC.  Anal.
Chem.  52: 1993-1994 (1980).
23.  J.  W.  Graydon and  K. Grob. How
efficient are capillary cold  traps?
J. Chromatoar. 254: 265-267  (1983).
24.  Mouradian,   R.F.,   S.P.   Levine,
H.Q. Ke  and H.H.  Alvord.  Measurement
of  Volatile  Organics   at  Parts  Per
Billion  concentrations  Using  a  Cold
Trap   Inlet   and  High  Speed  Gas
Chromatography.  submitted  to  J_.	Air
Haste Manag. Assoc.
                                         269

-------
Figure 1
Fast-GC system;
A: syringe or gas sampling
loop injection port;
B: silica transfer line;
C: low dead volume unions;
   electrical contacts;
   trap tube;
   upper chamber cold trap;
   lower chamber cold trap;
   baffle;
   capillary column
   flame ion. detector;
                    K:  capacitor power supply
     270

-------
                                                       DISCUSSION
BANK WOHLTJEN: How much energy did your capacitive discharge heater
we?

STEVEN LEVINE: It's running about 30 to 70 volts discharge with a few tens
rfamps.

HANK WOHLTJEN: How big are the capacitors? Are they a tenth of a farad
ursomething like that?

SIEVEN LEVINE: All the details of the design is in that paper in Analytical
Chemistry.

HANK WOHLTJEN: You mentioned electric cooling of the trap. What do you
Jhiai you'd use for that, a refrigerator or a thermal electric?
STEVEN LEVINE: It would have to be a thermoelectric cooler. We are
investigating that at this moment.

JOHN SNYDER: I was curious as to the diameter of the columns you're using.

STEVEN LEVINE: They're just 0.25 mm columns. They're very traditional
columns. They're not megabore. They're not ultra small.

JOHN SNYDER: You also spoke about the dead volume in the detectors. Are
you modifying traditional detectors or are you making your own detectors?

STEVEN LEVINE: We have a 90 pi dead volume BCD from HNU Systems at
this point that we're working with. We feel that size is probably too big.
                                                                 271

-------
              SCREENING VOLATILE ORGANICS BY DIRECT SAMPLING ION TRAP AND
                               GLOW DISCHARGE MASS SPECTROMETRY*
            Marcus B. Wise, G.B. Hurst, C.V. Thompson, Michelle V. Buchanan, and Michael R. Guerin

                                         Analytical Chemistry Division
                                        Oak Ridge National Laboratory
                                       Oak Ridge, Tennessee  37831-6120
ABSTRACT

   Two  different  types   of  direct  sampling  mass
spectrometers  are currently  being  evaluated  in  our
laboratory for use as rapid screening tools for volatile
organics in a  wide range  of environmental matrices.
These include a commercially available ITMS ion  trap
mass spectrometer  and a  specially  designed  tandem
source  glow discharge  quadrupole mass  spectrometer.
Both of these  instruments  are equipped  with versatile
sampling  interfaces which  enable direct monitoring of
volatile organics at part-per-billion (ppb)  levels in air,
water,  and  soil  samples.     Direct  sampling  mass
spectrometry does not utilize  chromatographic or other
separation steps prior to admission of samples into the
analyzer.   Instead, individual  compounds  are measured
using one or more  of  the  following methods:  spectral
subtraction, selective chemical  ionization, and  tandem
mass  spectrometry  (MS/MS).   For  air  monitoring
applications, an active "sniffer" probe is used to achieve
instantaneous response.  Water and soil samples are
analyzed by means of high  speed direct purge into the
mass spectrometer. Both instruments provide a range of
ionization options for added  selectivity and the ITMS
can  also  provide  high  efficiency  collision   induced
dissociation  MS/MS for   target  compound  analysis.
Detection  limits  and  response  factors  have  been
determined for a  large  number volatile organics in air,
water, and a number of different soil types.
 INTRODUCTION

   Direct  sampling   mass   spectrometry  for  the
 measurement  of  trace  levels of volatile  organics  in
 environmental matrices has a wide range of important
 field  screening   applications.    These  include  the
measurement of volatiles in waters, soils, oily wastes, stack
emissions, and ambient air, among others. In addition, real-
time "sniffing" capability provides a convenient means of
detecting soil gas emissions, leaking waste containers, and
probing the atmosphere in enclosed storage facilities.

    Because  of their  small  size,  relative simplicity,
ruggedness, and low power consumption, conventional
quadrupole mass spectrometers and quadrupole ion trap
mass  spectrometers   are  especially  attractive  for
transportable field screening applications. In fact, several
commercial quadrupole based instruments are currently
available for field monitoring applications and recently,
several different research groups have been developing and
demonstrating transportable ion trap mass spectrometers for
on-site GC/MS applications (1-3).

    This paper describes the  use of an ion trap mass
spectrometer and a tandem source glow discharge mass
spectrometer for the direct measurement of ppb levels of
volatile organics in air, water, and soil.   Because these
instruments do not use chromatographic separation prior to
admitting a sample into the mass spectrometer, the response
time is virtually instantaneous and accurate quantification of
target analytes can be accomplished in less than 2 minutes.
Although the tandem source quadrupole mass spectrometer
is somewhat limited in its ability to handle complex samples,
the  ion  trap mass  spectrometer has  the capability of
selective ion storage and multiple stages of collision induced
dissociation for much greater specificity.

    Laboratory-based instruments arecurrently being used
to develop and validate methods for direct air monitoring
and the screening of water, soil and waste  samples. A
transportable ion trap mass spectrometer for field use is
under construction in our laboratory and will be initially
tested in 6-9 months.
                                                       273

-------
 EXPERIMENTAL

 Instrumentation

 Ion Trap Mass Spectrometer
    All ion trap experiments were performed with a
 Finnigan  MAT  Corporation  ITMS  ion  trap  mass
 spectrometer.   Our instrument is equipped  with a
 specially   designed   vacuum   chamber   which   is
 electropolished  on  the  inside  and pumped to high
 vacuum with two air cooled 330 L/sec turbomolecular
 pumps.   The vacuum chamber and analyzer cell  are
 maintained  at  a constant  temperature of  120°  C by
 means of infrared heating lamps which help to minimize
 the adsorption of contaminants on the analyzer surfaces.
 This  instrument is also  equipped  with  the necessary
 hardware and software to perform electron impact (El)
 and chemical ionization  (CI), as well  as selective  ion
 ejection, and collision induced dissociation multiple-step
 (tandem)  mass spectrometry  experiments  (MS/MS).
 Control  of the instrument  and data acquisition  are
 performed with an IBM AT compatible computer using
 software provided by the manufacturer.

    The standard chromatographic  interface provided
 with  the  ITMS instrument  has  been  replaced with a
 custom designed interface developed in our laboratory.
 This interface consists of a short length (14 inches) of
 110  micron ID  uncoated fused  silica  capillary tubing
 which is maintained at atmospheric pressure at one end
 and high vacuum at the other end.  The high vacuum
 end of the capillary is inserted directly into the ITMS
 analyzer  cell  and  the atmospheric  pressure  end is
 connected to a quick-coupling device which  allows rapid
 switching of sampling modules for different monitoring
 applications.  The gas flow rate through the capillary
 restrictor is approximately 0.5-1.0 mL/min. Because  the
samples are introduced directly  into the ion trap cell,
 the manifold pressure is maintained  at a lower pressure.
This  is believed to  help  reduce deterioration of  the
electron  filament  and the  electron  multiplier.   For
example, even  when sampling  water-saturated  air  for
extended periods of time, the electron  filament lifetime
has been  approximately  6 months  and  the multiplier
lifetime has been in excess of 12 months.

ITMS Air Sampling Probe
   For direct  air  monitoring  experiments,  a  special
sampling system has been developed as  shown by  the
diagram in Figure 1.  This system consists of an 1/4 inch
OD teflon transfer line which is connected at one end
to the air sample generation system and at  the  other
end  to a sampling  "cross"  arrangement which  allows
helium to be mixed with the air sample prior to entering
the ITMS. The helium is necessary as a buffer gas in
the ITMS to collisionally cool ions, thus reducing loss of
ions   from  the  trap  and  improving  the  overall
performance.  A pulsed valve is used to meter helium
into the air stream providing approximately an order of
magnitude increase in sensitivity relative to a fixed-ratio,
continuous mixing of helium with the air. A vent port also
located on the inlet "cross" of the sampling system allows
the gas stream to be continuously sampled at a high flow
rate, thus decreasing the response time for the  mass
spectrometer.   The  other port of the inlet "cross" is
connected to  a short section of uncoated fused silica
megabore capillary which is used as an "open/split" interface
with the ITMS by inserting 1 inch of the microbore capillary
restrictor  into the other end of the megabore tubing.
Approximately 2  L/min  of air is  drawn through the
megabore tubing  by means of a small sampling pump;
however, a metering valve located between the pump and
the splitter can be used to reduce the pumping speed if
desired. This combination of active pumping and the use of
the open/split capillary interface minimizes the dead volume
in the inlet system leading to a response time of only a few
seconds.

Purge Device for Water and Soil samples
   For the measurement of volatile organics in water and
soil samples  (slurries), the air sampling probe is simply
replaced with a high speed needle sparge purge device as
shown  in Figure 2. This device accepts standard 40 mL
VOA vials which mount directly on the needle sparger. A
pressure regulator and a precision needle valve control the
flow of helium purge gas through  the sample  and the
purged components  exit through  a 10 inch length  of
megabore capillary tubing. Normal helium flow rates vary
from 100 to 200 mL/min which efficiently purges the volatile
components from a room temperature sample in less than 5
minutes.   The purge device connects directly with the
capillary restrictor interface in an open-split configuration
with a split ratio of approximately 100:1. The bulk of the
sample is diverted to the vent port.  As an added feature
for screening applications, the vent port is  capable of
accepting resin cartridges for trapping of components that
would normally be vented. This enables the collection of an
archived  sample which may be  sent back to a central
laboratory for confirmatory analysis by GC/MS.

Tandem Source Quadrupole Mass Spectrometer
   The tandem-source quadrupole mass spectrometer
(TSMS) is a prototype instrument constructed using an
EXTREL C-50 quadrupole mass spectrometer as the basic
system. This instrument was configured with 3/4" diameter
rods for high transmission efficiency and a 300 watt RF
power  supply  for  a maximum mass range of 500 amu.
Control of the instrument is provided by a Dell 325
computer using software written in our laboratory.  An axial
El source was purchased with  this instrument for testing
purposes and for generating conventional 70 eV electron
impact spectra.

   In  order  to  produce a  versatile  instrument  for
environmental monitoring applications, the configuration of
the standard C-50 mass  spectrometer was extensively
                                                        274

-------
modified.  In addition to the axial El source which was
purchased with  the  spectrometer,  a  glow  discharge
ionization source was designed and constructed for this
instrument.  This source is  housed in a differentially
pumped vacuum chamber which is separated  from the
rest of the mass  spectrometer by a  1.5 mm  diameter
vacuum  conductance limit as shown in Figure 3.   The
glow discharge source  is  typically  maintained  at  a
pressure of 0.25 torr while the analyzer is maintained at
2 x  10"5  torr.   Ions  generated  by  glow  discharge
ionization pass through a lens assembly into the high
vacuum portion of the instrument where they  enter the
lens  assembly of   the  axial  El   source  and  are
subsequently focussed into the mass analyzer.

   Air  samples can  be introduced  into  the tandem
source quadrupole mass spectrometer by  two different
methods, either through the differentially pumped glow
discharge source chamber, or directly into the electron
impact source  by means of a simple capillary restrictor.
Both inlet systems have been designed so that they are
directly compatible with the same sampling devices used
with the ion trap mass spectrometer.  Thus, essentially
the  same  apparatus and experimental conditions are
used for  direct purging of water and  soil  samples
regardless of the mass  spectrometer used.   The only
difference is the ability of the glow discharge ionizer to
sample air directly without the need for the air sampling
pump and open/split interface used with the  ITMS.

Dynamic Sample Generator
   A dynamic sample generation  apparatus is used to
produce  known  concentrations   of  volatile  organic
analytes  in an  air stream. This apparatus was used for
the  determination of instrumental detection limits for
real-time  air  monitoring experiments.   It  basically
consists of a variable speed syringe pump and a dilution
air manifold.   The  syringe  pump continuously  meters
small amounts of organic compounds into a controlled
stream of air.   Concentrations of the analytes can be
easily varied by adjusting the speed (metering rate) of
the  syringe pump and/or by changing  the flow rate of
dilution air through  the manifold.  Turbulent mixing of
the organic compounds and the dilution air occurs in the
manifold  line  which   provides  a  homogeneous
concentration at the sampling ports.

."  Components   of  the dynamic  sample  generator
include a Razel Instruments  model A-99 syringe pump
equipped with a 5 mL syringe, a 100 psi air supply line
equipped with an on/off toggle valve and  a  precision
metering valve, a 1.5 m x 6 mm  Teflon line (dilution
manifold), and two  1/4 inch Swagelock sampling ports.
The  apparatus   produces   continuous  and   stable
generation of organic concentrations  in  air  and  also
; allows rapid changes in concentration without having to
wait excessively to reach a steady-state concentration.
    Air containing the desired concentration of individual
organic compounds is typically generated by metering a
(1:1) water/methanol solution containing approximately 400
ug/mL of the organic compound into the dilution air stream
using the syringe pump. The flow rate of the syringe pump
can be continuously varied from 8.47 x 10"4 mL/min to
0.0503 mL/min. The dilution air flow is typically adjusted
for a rate of 25 L/min through the manifold.  As this air
flows rapidly past the syringe pump  needle, it quickly
vaporizes the volatile organics and the solvent. Liquid flow
from the syringe, however, must be maintained low enough
to prevent condensation in the system. By knowing the
concentration of the organic in the liquid solution, the flow
rate out of the syringe, and the flow rate of the dilution air,
the concentration of the organic compounds in the air can
be readily calculated. This assumes that there is minimal
adsorption of analytes  on the walls of the manifold and
complete vaporization  of the liquid into the dilution air.

Operating Conditions

Ion Trap Mass Spectrometer

    Most of the ion trap data presented in this paper was
generated using electron impact ionization conditions. Scan
functions for the acquisition of mass spectra were written
using the scan function editor program supplied with the
commercial software.  Typically, for optimum sensitivity the
electron ionization time was 50 msec.  Low mass cut-off was
60 amu, preventing the storage of ions due to water and air.
The mass scan range was approximately 50 to 200 amu
which enabled the detection of major ions for each of the
volatile organic compounds.  In order to improve the signal-
to-noise ratio, 16-25 microscans were averaged per displayed
scan. Axial modulation was used for  all experiments in
order to achieve optimum instrument performance. Helium
buffer gas was admitted into the system exclusively through
the sample transfer line.

Tandem Source Quadrupole Mass Spectrometer
    The glow discharge ionization source  is specifically
designed  for  high  sensitivity  direct air  monitoring
applications.  Air is  admitted into the ionization region
through a metering valve at  a flow rate of 0.5-1.0 standard
mL/min while a 160 L/min roughing pump maintains the
pressure in the ionizer at a constant 0.25 torr.  Coaxial
ionization electrodes are used for the discharge and consist
of a 1 cm diameter x 2  cm long hollow cathode with a 20
gauge wire anode. A potential difference of approximately
600 volts is sufficient to strike and maintain a discharge in
the source. Ionization of organic compounds in this source
is the result of ion molecule reactions which produce proton
transfer and charge exchange reaction products.  Conditions
within  the glow discharge source  can be adjusted to
optimize  either proton transfer  or  charge exchange
reactions.  The proton transfer reactions provide high
sensitivity for compounds which  have proton affinities
greater than that of water (which is the primary proton
                                                        275

-------
transfer reagent).  Charge exchange on the other hand,
is  a  much  more universal  ionization  method and
produces  fragmentation  spectra  which are similar  to
electron impact ionization spectra.   By operating the
glow discharge source at low pressures, the formation of
water  cluster  ions  which  often  hamper  API  mass
spectrometers is nearly eliminated, improving sensitivity
and decreasing  the complexity of the spectra.

   Direct sampling using the electron impact ionization
source  of  the   quadrupole  mass   spectrometer  is
accomplished by means  of a  1  meter length  of 110
micron  ID  uncoated fused silica capillary tubing.  A
simple on/off valve between the capillary and the source
allows the restrictor to be isolated when not in use. The
conditions in the ionizer include an electron current of
0.5 to 1.0 milliamps and  an electron energy of 17 to 20
eV.   The  use of lower  electron energies  helps  to
minimize fragmentation,  thus concentrating ion  current
in fewer ions.

Samples and Chemicals

   Individual samples of 31 different volatile  organic
compounds  from  the  USEPA Target Compound List
were   obtained from Ultra   Scientific  Company  as
solutions of the neat compound dissolved in methanol at
a concentration of 10,000 ppm.  Solutions for use in the
dynamic sample generation system  were prepared from
the methanol stock solutions using ultra-pure water and
spectroscopic grade methanoL   In order to  verify the
proper  calibration and  performance of the dynamic
sample generation system, certified standards  of  volatile
organics  hi nitrogen  were  purchased  from  Scott
Specialty Gases.

   Water samples were prepared using distilled water
containing 0.15 g/L of sodium chloride and 0.17 g/L of
sodium sulfate. A series of concentrations of individual
volatile organics from approximately 1 ppb to 200 ppb in
water was prepared by injecting a known  concentration
of a methanol solution  into water and then carefully
pipetting the water standard  into a 40 mL pre-cleaned
VOA vial.  The vials were  capped  with Teflon  lined
septa until  used.   Most  samples were prepared at
approximately  pH 7; however,  samples of  benzene,
trichloroethylene,  and  tetrachloroethylene  were also
prepared at pH 2 and pH 10.

   A total of 5 different soil samples were examined as
part  of this  study including 2 soils provided by the U.S.
Army   Toxic   and   Hazardous  Materials   Agency
(USATHAMA), 2 local soils, and a potting soil.  These
represent a range of soil types including clay, sand, and
high humic content.  The soil samples were prepared by
injecting a pre-weighed  5 gram sample of soil  in a 40
mL VOA vial with a known  quantity of the  volatile
organic in methanol  and allowing it  to sit for  a short
period of time. Slurries of the soil samples for direct purge
experiments were prepared by adding 25 mL of water to the
sample and allowing them to sit for at least 1 hour prior to
analysis.

RESULTS AND DISCUSSION

Volatile Organics in Air

   The primary objective of the air monitoring study was to
optimize the experimental conditions and determine the
real-time detection limits for a representative sample of
volatile organic pollutants. This sensitivity assessment was
performed using standard electron impact ionization on
both the tandem source quadrupole mass spectrometer and
the ITMS. This enables comparison of our results with
other mass spectrometer systems which are commercially
available and use electron impact ionization. For all ITMS
experiments, the electron ionization time was  50 msec.
Mass scan ranges were selected as appropriate for each
compound although the lower mass cut-off was normally at
least 40 amu or higher. This prevented water, nitrogen, and
oxygen ions from being stored in the ion trap simultaneously
with the analyte ions, thus minimizing the effects of space
charge and unwanted ion-molecule reactions. Future studies
will  involve a comparison  of sensitivities  for  chemical
ionization and electron impact ionization.

     Using the ITMS instrument, sensitivities for the 31
volatile organics were determined. However,  pumping
problems  with  the  tandem  source  quadrupole  mass
spectrometer restricted experiments to the determination of
detection   limits  for  only  3 compounds:  benzene,
trichloroethylene,  and  tetrachloroethylene.  For  both
instruments,  response  curves (instrument response vs.
concentration in air) were prepared for each of the
compounds studied. The range of concentrations examined
was generally between 4 and 200 ppb. A typical experiment
involved  the acquisition of a background level signal,
followed by the acquisition of spectra for  a  series of
decreasing concentrations in air generated with the dynamic
sample generator. Instrument response vs. time produced a
"stair-step" curve as the concentration of organic was
reduced to successively lower levels. Each concentration
level was maintained for several minutes to ensure that a
steady state concentration  was reached before further
reducing the level.

Ion Trap Mass Spectrometer
    An electron impact mass spectrum of a  mixture of
volatile organics in air is shown in Figure 4.  This mixture
contained carbon disulfide, benzene, chloroform, toluene,
and ethyl benzene at concentrations of approximately 1 to
10 ppm.   As shown in  this  figure, space-charge-induced
peak broadening and mass shifting are not significant.

    A typical "stair-step" air monitoring response curve
acquired with the ITMS is  shown in Figure 5.  This is a
                                                         276

-------
reconstructed plot of the ion current for m/z 83 as"seen"
by the  ITMS  instrument vs.  time  for  a sample of
chloroform  in  air.    As  the  concentration of  the
chloroform was decreased to lower values over a period
of  time,  the   response  of  the   ITMS  decreased
proportionally.  This same type of plot can be generated
in real-time continuous monitoring applications, allowing
changes in the concentration to be  readily visualized.
As shown in Figure 5, the response time of the ITMS to
changes in concentration was very fast  (less than 15
seconds) and the time required for the sample generator
to reach  steady state at a new  concentration  was
typically less than 3 minutes.

   In addition  to the continuous plotting of the ITMS
total ion response, it  is also possible  to monitor the
actual mass  spectrum  in real time in order  to  detect
changes in specific ion  intensities.   This is  especially
useful whenever multiple components are present  in a
sample.  All of the  information which is generated in
real-time may be stored on a hard disk  as a temporal
series of mass spectra, allowing response curves for any
ion in the mass range to be reconstructed, plotted, and
integrated.    An example of  a post-processed  mass
spectrum of chloroform in air is shown in  Figure 6.

   An  important  feature  of  the  response   curves
generated  with the  ITMS  is  the  pseudo-sinusoidal
waveform  superimposed on  the curve.   This  is not
actually noise, but is actually an effect due to the pulsed
valve addition of helium into the air stream.  Maxima
correspond to the optimum helium/air ratio and minima
correspond to the least effective helium/air ratio.   By
synchronizing the pulsing of  the helium valve with the
acquisition of the  spectral scans, this effect should be
nearly eliminated.

   The experimentally determined detection limits for
the 31 volatile organic compounds in air  are  presented
in Table 1.  As shown in this table, the detection limits
are generally in the low ppb range which  is comparable
to the sensitivity of some commercially  available  API
mass  spectrometers.    Exceptions  to   this  include
bromoform,  chloroethane,  and   chloromethane.
However, because  chloromethane and chloroethane are
extremely volatile (boilding points of 24°C and +12.3°C,
respectively), it  is likely that these compounds were lost
during preparation of the standard.  Bromoform, on the
other hand,  is less  volatile than most  of the compounds
examined, with a boiling point of +150.5°C.  Bromoform
probably condenses on the walls of the vapor generating
system  at  room  temperature and never reaches the
ITMS inlet.  With proper sample preparation techniques
and a shorter, heated sampling line, detection limits for
chloromethane,  chloroethane,  and  bromoform  would
probably be more  comparable to the other compounds
studied.  This is a reasonable assumption since these
 compounds  are   chemically  very   similar   to  other
halogenated hydrocarbons  that have  been successfully
measured and would be expected to have similar ionization
efficiencies under electron impact ionization conditions.

   The detection limits which are reported  for volatile
organics in air, were calculated using the RMS (root mean
square) variation in the signal measured with no sample
present (a blank). This is an accurate determination of the
analytical detection  limit  and represents  the lowest
concentration of a compound in air that can reliably be
observed with the current sampling interface and ITMS
operating parameters.  For these calculations, the lowest
reliably measured signal is defined as the average of the
blank signal plus three times the  RMS variation in this
signal.  From the lowest reliably  measured signal, the
detection limit can be calculated from a calibration curve
relating signal to concentration.   Linear  least squares
calibration  curves were constructed for the 31 volatile
organics  studied.    Due  to  space  charging  effects
encountered with a few compounds, a quadratic model was
necessary to describe a better fit for the data.

Tandem  Source  Quadrupole Mass Spectrometer
    Detection limits for benzene,  trichloroethylene, and
tetrachloroethylene in air were also determined using the
tandem source quadrupole mass spectrometer.  Various
concentrations of the individual compounds were generated
using the .dynamic sample generator as previously described.
One signal averaged mass spectrum (n=36) was acquired
and stored for  each  concentration.   Signal  averaged
background samples were also acquired and subtracted from
the mass spectra of  the  actual samples.  Experimental
difficulties arising from a high hydrocarbon background in
the instrument complicated these low-level  analyses. The
background problem was due to backstreaming of diffusion
pump oil and condensation on the ionization source.

    Linear regressions of the data were calculated and both
data and regression were plotted for each compound. Due
to the nature of the signal averaging experiments, an
accurate detection limit could not be determined for the
three compounds using the same RMS noise calculation
method as  the ITMS. Rather, the detection limit was
determined by calculating the standard deviation of the
linear  regression plot  and  then  determining the
concentration at which the signal is equal to the standard
deviation4 as shown in Figure 7.  The regression curve for
benzene in air is shown  in Figure 8 and the calculated
detection limit was determined to be approximately 11 ppb.
Based on the linear regression curves for trichloroethylene
and tetrachloroethylene,  detection  limits  for  these
compounds were determined to be approximately 42 and 29
ppb respectively.

    Although the electron impact ionization was used
predominantly for this study, earlier experiments with the
glow discharge ionization source indicate that the detection
limits are very similar to or slightly  better than those
                                                        277

-------
achievable with the electron impact ionization source.
In  fact,  the  tandem  source  configuration  of  the
quadrupole  mass spectrometer is  unique and provides
extra versatility in terms of  sample introduction  and
ionization options relative to a  conventional electron
impact ionization quadrupole.  For example, air may be
sampled  and ionized directly  with the glow discharge
source  or  it may  be sampled  through  a capillary
restrictor and ionized with the  axial electron  impact
ionization source.  Since both  ionization sources are
simultaneously installed on the spectrometer, switching
between ionization modes or  sample inlet systems  is a
simple matter  of opening the  appropriate  valve  and
turning on the electronics for the selected source.

     The  advantages of  the  glow  discharge  source
relative to the electron impact ionization source are  that
it is more rugged for long term operation, the response
time is virtually instantaneous, and the  source  is very
tolerant   of   high   oxygen   and  water  saturated
atmospheres.  Primary advantages of the axial electron
impact ionization source are ease of operation and the
ability to produce library searchable mass spectra.   A
major problem with the electron  impact  source is  that
the  filament assembly is very susceptible to oxidation
and burn-out if exposed to large amounts of oxygen or
water.    For example,  when  performing  direct  air
monitoring experiments with the electron impact source,
the filament must be replaced  every 3 to 4 weeks.

Volatile Organics in Water and Soil

    The sample handling apparatus and methods for the
determination of  volatile  organics in water and  soil
slurries are  identical for both  the  ITMS  and the TSMS
experiments. Volatile organics are purged from a water
or soil slurry directly into the mass spectrometer without
any preconcentration  such  as  trapping on  a  resin
cartridge.  In the simplest case, conventional electron
impact ionization spectra are continuously acquired over
a mass range of approximately 40-200 amu in order  to
observe  the response for ions  corresponding  to the
purged  volatile organics.   As shown  in  Figure 9, the
purge profiles for a particular ion can be reconstructed
as a plot of response versus purge time.  At a helium
purge flow of 200 mL/min, purging is  normally 90% or
more complete after 3 minutes.  The area beneath a
purge profile correlates well with the concentration of
the  analytes in the sample  as  shown  in Figure  10.
Quantification is accomplished simply by  integrating the
area of a  reconstructed purge  profile  for the  ions
corresponding  to  the target  analytes.    A  typical
calibration curve for benzene in water from 1  to  100
ppb is shown in Figure 11.   Using carefully prepared
standards, correlation coefficients of better than 0.998
are possible.  Quantitative reproducibiltiy of less than
10% at the 95% confidence level can also be achieved
for water samples without the  use of internal standards.
   A series of experiments were conducted in which the
detection limits, relative response factors, and standard
spectra were generated for a series of volatile organics in
water. In addition, studies with benzene, trichloroethylene,
and tetrachloroethylene were also conducted in order to
examine the effects of pH and soil type on  the purge
efficiency of water samples and soil slurries relative to
solutions of volatile organics in pH-7 water. Data for these
samples were acquired simultaneously using both the ITMS
and the TSMS instruments in order to compare detection
limits and quantification accuracy.

   The detection limits for 21 different volatile organics in
pH-7 water using the ITMS and electron impact ionization
are shown in Table 2. These range from approximately 3
ppb  for benzene to approximately 60 ppb for dichloro-
ethane and appear to be routinely achievable using the
direct purge method. For comparison, the detection limits
for compounds purged into the TSMS are also typically less
than 200 ppb, although they are generally not quite as good
as can be achieved with the ITMS.  Accurate detection
limits for acetone, 2-butanone, and 4-methyl-2-pentanone
have not yet been established due to much lower purge
efficiencies.

   The matrix effect experiments which were conducted for
benzene,   trichloroethylene,  and  tetrachloroethylene
appeared to show essentially the same purge efficiency at
pH-2,  pH-7, and  pH-10.    Similar results for  these
compounds were also obtained for a potting soil leachatc
with a high humic content.  These results suggest that
accurate quantification maybe achieved without the need
for extensive sample preparation or the use of internal
standards for many water samples.  An exception to this
may be water samples which contain a high surfactant
concentration, although comparative data have not yet been
generated.

    As opposed to the water samples, differences in the
purge efficiencies for volatile organics in soil slurries are
more pronounced. As shown in Table 3, the relative purge
efficiency   for  benzene,   trichloroethylene,   and
tetrachloroethylene ranges from approximately 25% to 90%
relative to pH-7 water.  The least efficient purging was from
the soils which had a high clay content and the most
efficient purging was from soils having the highest sand
content.  Although the general trend exhibited by these
results is probably reasonable, the actual purge efficiencies
are probably better than the data indicate. For example,
comparative purge profiles for benzene, trichloroethylene,
and tetrachloroethylene in pH-7 water and a potting soil
slurry are very similar as shown in Figure 12.
    Apparent differences in purge efficiency most likely
reflect inefficient stirring and sample purging using a single
needle sparger. Further studies have also shown that there
was probably significant loss of volatiles from the soil
                                                         278

-------
samples  during the  preparation  step  using our  soil
spiking procedure. Improvements in the purging of soils
samples  could  probably be achieved by  simultaneously
stirring samples to ensure more homogeneous sparging.
Further, the use of an internal standard would be useful
to help minimize quantitative errors due to differences
in purge efficiency.
CONCLUSIONS

   The results of these studies have demonstrated the
feasibility of using direct sampling mass spectrometry for
the real-time detection of trace organic compounds in
air, water, and  soils.    Detection limits for  both the
tandem source quadrupole mass spectrometer and the
ion trap mass spectrometer are generally in the range of
5 to 200 ppb for water and  soil samples without any
sample preparation or preconcentration. The  detection
limits for volatile organics in air using the ITMS range
from approximately 1 to 45  ppb for the 31 volatiles
studied which is approximately 1,000 times lower than
the threshold limit values (TLV's) for these compounds.
These detection limits are comparable to those that can
be achieved with API  mass  spectrometers.  Detection
limits for the compounds studied using the TSMS are
slightly worse  than  those  obtained with  the  ITMS;
however, they also  are well below the published TLV's.
This suggests that the ITMS or TSMS could indeed be
useful for field  monitoring of stack emissions and soil
gas emissions at hazardous waste sites.

   Although    it  is   not    likely   that   significant
improvements  can be  made  in  the detection limits
achieved with the TSMS, modification and optimization
of the sampling interface for the ITMS will probably
result in even better detection limits than reported in
this document.  In addition, the ITMS instrument also
has the capability of chemical ionization which can be
used  to  selectively  enhance certain  target  analytes
relative to other compounds in a sample stream.

    Both the TSMS and ITMS have excellent detection
limits for volatile organic compounds in air, water, and
soil; however, experience with the  two different mass
spectrometer systems  suggests that the ion trap mass
spectrometer overall is a more  useful  instrument for
continuous air  monitoring.    Specifically, the ITMS is
highly reliable, easier to operate,  and more stable than
the  tandem  source  quadrupole  mass  spectrometer.
Further,  the  ion  trap  mass  spectrometer  has  the
capabilities of controlled chemical ionization, selective
fan storage, and collision induced  dissociation  (CID)
tandem mass  spectrometry  (MS/MS).  These features
are especially important in helping to identify  individual
components in  a complex sample, especially since  no
chromatographic separations  are performed -on  the
sample  prior   to  entering   the  mass  spectrometer.
 Without  these  features,  the TSMS is restricted  to
 monitoring samples that typically have fewer than 10-15
 components. Finally, due to the simplicity of the ion trap
 analyzer assembly, this type of instrumentation lends itself
 to downsizing, portability, and remote operation better than
 the TSMS.

    While  the  results of this study have been quite
 successful  and   demonstrate the  potential  of  the
 instrumentation for screening of environmental samples,
 much  work remains.   Especially  important  is  the
 development of methods  for the identification and
 quantification of compounds in complex mixtures.  This
 work will involve a thorough examination of chemical
 ionization reactions, the generation of MS/MS spectra of
 commonly encountered organic pollutants and potential
 interferences, and the development of computer programs
 to process this information in real time.
 ACKNOWLEDGEMENT

 'Research sponsored  by the  U.S.  Army Toxic  and
 Hazardous Materials Agency under Interagency Agreement
 1769-A073-A1 under U.S. Department of Energy Contract
 DE-AC05-84OR21400  with  Martin Marietta  Energy
 Systems, Inc.
REFERENCES

1.      McClennen, W.H.,  Arnold,  N.S.,  Sheya,  S.A.,
       Lighty, J.S., Meuzelaar, H.L.C., "Direct Transfer
       Line   GC/MS   Analyses   of   Incomplete
       Combustion Products from the  Inceneration of
       Medical Wastes and the Thermal Treatment of
       Contaminated Soils", Proc. 38th ASMS Conf. on
       Mass Spec. All. Topics, Tucson, AR, 1990, 611-
       612.

2.      Hemberger, P.H.,  Alarid, I.E.,  Cameron, D.,
       Leibman,  C.P.,  Cannon, T.M.,  Wolf,  M.A.,
       Kaiser,   R.E.   "A   Transportable   Gas
       Chromatograph/Ion   Trap Detector for  Field
       Analysis of Environmental Samples", Int. J. Mass
       Spectrom. Ion Proc.. In press.

3.      Wise,  M.B.,  Buchanan,  M.V., Guerin,  M.R.,
       "Rapid  Environmental  Organic   Analysis  by
       Direct   Sampling   Glow  Discharge    Mass
       Spectrometry and Ion Trap Mass Spectrometry",
       Oak Ridge National Laboratory TM-11538, Oak
       Ridge Tennessee, 1990.

4.      Hubaux, A.; Vos, G.,  Anal. Chem.. 235,  1967,
       849-855.
                                                       279

-------
                                Table 1

Detection Limits for Volatile Organics in Air using Direct Sampling FTMS


      Compound                              Detection Limit (ppb)

      1,1,1-Trichloroethane                              2
      1,1,2,2-Tetrachloroethane                          3
      1,1,2-Trichloroethane                             20
      1,1-Dichloroethane                               16
      1,1-Dichloroethene                                6
      1,2-Dichloroethene                                3
      1,2-Dichloropropane                             45
      2-Butanone                                     48
      4-Methyl-2-Pentanone                            17
      Acetone                                        22
      Benzene                                         5
      Bromodichloromethane                            4
      Bromoform                                    >  80
      Bromomethane                                 >280
      Carbon Disulfide                                25
      Carbon Tetrachloride                             16
      Chlorobenzene                                   2
      Chloroethane                                  >209
      Chloroform                                      3
      Chloromethane                                 >268
      Cis-l,3-Dichloropropene                           6
      Dibromochloromethane                           12
      Ethylbenzene                                    2
      Methylene Chloride                              12
      Tetrachloroethylene                               8
      Toluene                                         3
      Trans-l,3-Dichloropropene                         7
      Vinyl Acetate              •                    44
      Vinyl Chloride                                   5
      O-Xylene                                        4
                                    280

-------
                                         Table 2

         Detection Limits for Volatile Organics in pH-7 Water using Direct Purge ITMS
              Compound

              1,1,1-Trichloroethane
              1,1,2,2-Tetrachloroethane
              1,1,2-Trichloroethane
              1,1 -Dichloroethene
              1,2-Dichloroethane
              1,2-Dichloroethene
              Benzene
              Bromoform
              Carbon Disulfide
              Carbon Tetrachloride
              Chlorobenzene
              Chloroform
              Cis-l,3-Dichloropropene
              Ethylbenzene
              Methylene Chloride
              Styrene
              Tetrachloroethylene
              Toluene
              Trans-l,3-Dichloropropene
              Vinyl Chloride
              Xylenes (total)
           Detection Limit (ppb)

                   12
                   28
                   18
                   33
                   27
                   21
                   3
                   15
                   18
                   16
                   5
                   20
                   6
                   4
                   60
                   5
                   5
                   4
                   15
                   5
                   4
                                        Table 3

      Purge Efficiency of Volatile Organics in Soil Slurries Relative to pH-7 Water
Soil Sample   Soil Type

THAMA 1    Clay                 29

THAMA 2    Sand/Clay             51

Local 1        Sand/Clay             61

Local 2        Sand/Clay/Humic      46

Potting        Sand/Humic           91
Relative Purge Efficiency (%)

    Trichloroethvlene    Tetrachloroethvlene
           20

           48

           45

           42

           77
19

46

61

42

53
                                           281

-------
                     ITMS DIRECT AIR  INLET
            VENT/AUX SAMPLING PUMP
                   t
                                                 MICROBORE
                           MEGABORE CAPILLARY        RESTRICTOR
    AIR INLET.
                   PULSED SOLENOID VALVE
                                                            TO ION TRAP CELL
                                                   METERING VALVE
          HEUUM INLET
                                       TO SAMPLING PUMP
                     Figure 1   Air sampling interface for ITMS.
                       TEFLON
HELIUM
PURGE
  GAS -
AN
Lll
•
t
9
•
A
•t
SF
ME


;


ER
\
"f*~* * — ^" *



SPLIT (
WAI \/t=N
i
•
i


X
3APILLARY
X3B33




ION
MC
Mo


i —





' — i
                                                              FPUMP |
         Figure 2  Device used for direct purge of volatiles from water and soil samples.
                                   282

-------
  1.5 mm Vacuum Conductance Limit
Aa*Iyz»r Vacuum Chamber
                                                                    Clew Discharge
                                                                    Senrcv Vacuum
                                                                    Chamber
                                                      To Reo(h Pump
      Figure 3   Diagram of the tandem source quadrupole mass spectrometer.
         500  -
                                                         (Cone. Low PPM)
             0
                                                       100
                                            Mass (AMU)
110
120
             Figure 4   ITMS electron impact mass spectrum of ppm levels of VOCs in air.
                                        283

-------
                                  ITMS Response to Chloroform in Air
                170
                160
                •53
                140
                130
                120
                no
                •oo
                 90
                 80
                 70
                 60
                 50
                 4-0
                 30
                 20
                 •0
                  0
170ppb
           56ppb
                28ppb
                      14 ppb
                                               blan
                              0.2     0.4
                    0.6      0.8
                    (Thousands)
             Elapsed Time (seconds)
   Figure 5   ITMS response for m/z 83 at various concentrations of chloroform in air.
INT
               Chloroform in Air
                                                83
                    56   61  65  69  73       81
                                                   85
                                                     8?
                                         100
                                               105
                                                                  100                 120

          Figure 6    ITMS post processed mass spectrum of chloroform in air.
                     80
               Mass (AMU)
                                                284

-------
   t>
   <
  130



  120 -



  110 -



  100



   90



   80



   70



   60



   50



   40



   30



   20



   10



    0
                    Estimate of Tetrachloroethylene Detection Limit

                        Tandem-Source Quadrupole Detection
                                          *95% Confidence Limits


                                 ^Calibration Curve
                        I   11	I   I	I	!    I	I
                                                        I    I    I   I    I   I
                        20      40     60     80     100


                                    Concentration (ppb)
                                                           120     UO
                                                                         160
  Figure 7   Graphical determination of detection limits for the TSMS instrument.
                                     BENZENE

                              Tandem Source Ouadrupola Air Monitor
i
S.
'c
800




700 h



600



500




400



300




200




 100



   0



-100
                                         I    I     I	I	L
                        20      40       60       80


                                   Concentration in Air (ppb)
                                                         100
                                                                 120
                                                                         140
     Figure 8   Linear regression curve for benzene in air using the TSMS instrument.
                                        285

-------
                            RESPONSE FOR m/z 78
           0:38
         1:18       2:05

             TIME (min)
                                           2:45
3:25
      Figure 9  Reconstructed purge profile for 100 ppb of benzene i
                                     in water.
         SOLUTION PURGE PROFILES OF AQUEOUS
        VINYL CHLORIDE STANDARDS (ppb = ng/ml)
 z
 LLJ
 H
 Z

 LLJ
 LU
 DC
                                                40 ppb
                    20 ppb
                  10 ppb
     2 ppb
  7:09
           T-T-      |  i  i  i   i	r-r-r
9:32       11:54      14:17       16:39      19:01
                         TIME (min)

Figure 10  Direct purge profiles for 4 different concentration of vinyl chloride in water.
                            286

-------
                   RESPONSE FOR m/z 78
     0
20          40          60          80
         CONCENTRATION (ppb)
                                                                  100
Figure 11  Response curve from 1 to 100 ppb for direct purge of benzene from water.
                             287

-------
     VOLATILE ORGANICS PURGED FROM PH-2 WATER
       770





       385 -\










       840-
                     50 ppb each compound


                    BENZENE



                                m/z78
   z

   O
                      TRICHLOROETHYLENE



                                   m/z 130
    1000-




     500-
                      TETRACHLOROETHYLENE



                                   m/z 166
                          3     4


                           TIME (MEM)
VOLATILE ORGANICS PURGED FROM POTTING SOIL SLURRY


                     50 ppb each compound
                      BENZENE
                                    m/z 78
a    630



I    315H
UJ
                      TRICHLOROETHYLENE


                                    m/z 130
       420





       210 H
                    TETRACHLOROETHYLENE


                                 m/z 166
     Figure 12 Comparison of VOC purge profiles for pH-7 water and potting soil.
                                  288

-------
                       DEVELOPMENT AND TESTING OF A MAN-PORTABLE
                    GAS CHROMATOGRAPHY/MASS SPECTROMETRY SYSTEM
                                       FOR AIR MONITORING
                         Henk L.C. Meuzelaar, Dale T. Urban and Neil S. Arnold
                    Center for Micro Analysis & Reaction Chemistry, University of Utah
                                  214 EMRL, Salt Lake City, UT 84112
ABSTRACT

A fully man-portable, GC/MS system based on the
combination of an automated vapor sample inlet, a
"transfer-line" gas chromatography module and a
modified Hewlett Packard model 5971A quadrupole MS
system is described.  The current prototype weighs
approx. 70-75 Ibs and uses 150-200 W of battery power.
Th&mass spectrometer and computer are carried in front
of the operator by means of a shoulder harness whereas
battery pack, carrier gas supply and roughing vacuum
system are carried as a backpack.  Air samples can be
malyzed using a special automated air sampling inlet.
TTie man-portable GC/MS system  is designed to be
Supported by a vehicle transportable  "docking station".

BACKGROUND

In situations involving severely contaminated hazardous
sraste sites, industrial accidents or natural disasters, as
srell as special military or law enforcement operations,
mobile laboratories may be of little use  because of
Emited site access,  restrictions due to contamination or
terrain constraints.  Under such conditions, man-portable
analytical instruments may offer the only acceptable
means of carrying out on-sitc analyses.

Obviously, man-portability puts severe constraints on
Wight, size  and power requirements as  well as on
fflggedness and user-friendliness.  Consequently, the man-
fortability requirement may also function as a convenient
benchmark for the development of analytical equipment
for a variety of special operational environments ranging
fiom remotely operated devices (e.g., robotic vehicles,
femes or probes) space stations  and operating rooms.
All of the above environments require a high degree of
miniaturization, reliability and ease of operation.

The past decade has witnessed impressive progress in
miniaturization of mass spectrometric systems.  Besides a
broad range of commercially  available benchtop
instruments, including the Hewlett Packard MSD (Mass
Selective Detector) and Finnigan MAT ITD (Ion Trap
Detector), several specialized  MS instruments have been
developed for applications where transportability  is a
prime requirement.  Well known examples include the
Bruker Franzen MM1 system, originally developed for
military applications involving chemical agent detection,
and the Viking Spectratrak system primarily designed for
environmental applications.

As shown in Figure 1 most commercially available
miniaturized systems are characterized by a combination
of relatively low weight (typically 100-300 Ibs, excluding
power source) and modest power requirements (600-1800
W range).  In spite of these marked advances in system
miniaturization, however, man-portability  and some of the
other  abovedescribed applications require  even more
stringent size, weight and power limitations.

This prompted us to undertake a study aimed at obtaining
maximum power and weight reduction using  the Hewlett
Packard MSD as a starting point.  Although the project is
still under continuing development, some  preliminary
results and conclusions are starting to take shape, as  will
be discussed in the following  paragraphs.

            SYSTEM DESIGN CONCEPTS

      An overview of the selection criteria for the main
system modules and components is given  in Table I.
                                                      289

-------
Automated Vapor Sampling Inlet Module
Transfer Line Gas Chromatography Module
Transfer line gas chromatography (TLGC) is defined here
as a form of GC in which the column connects two
environments, viz. an atmospheric environment at
ambient pressure and the vacuum environment of the MS
ion source region.  In other words, column inlet and
outlet pressures are more or less fixed and, consequently,
optimization of column flow requires suitable adaptation
of column length and/or diameter. This sets TLGC apart
from the more widely used short column gas chromato-
graphy (SCGC) technique in which column inlet
pressures can usually be adjusted while column length is
kept below 5 meters or so.

Although most TLGC applications reported thus far do
use short to very short column lengths, optimum GC
conditions for a 500 p.m i.d., ambient inlet transfer line
columns connected to a vacuum detector (e.g., MS) may
dictate column lengths in the 50-100 m range (see Figure
2). In view of the abovedescribed distinctive differences
between TLGC and SCGC we feel justified in adding yet
another term to the already baffling jargon of the
chromatographer.

When sampling condensable and potentially labile vapors
from air, the main challenge is to avoid compound losses
through irreversible adsorption and/or decomposition  in
the transfer line section.  To this end, a novel, automated
vapor sampling method was recently designed at the
University of Utah Center for Micro Analysis & Reaction
Chemistry (1,2). The most characteristic property of this
sampling method, illustrated in Figure 3, is the absence
of any valves or other mechanical obstructions in the path
of the molecules between the ambient environment and
the ion source. Only quartz walls and/or surfaces coated
with inert stationary phases (e.g., poly-dimethylsilicones)
are seen by sample molecules on their way to the ion
source.

A second advantage of the new sampling technique is the
potentially very short switching time. Sampling times as
short as 60 msec have been used already (2) and 20 msec
or less may be achievable in the near future.  This
enables "injection" of a narrow sample plug into the
TLGC column, thereby minimizing peak broadening  due
to sample injection and allowing repeat GC analyses  at 6-
60 sec intervals (3). All air flows in the inlet are
sustained by means of a Graseby Ionics miniaturized dual
air pump (max. capacity 2 x 500 ml/min, max power
consumption 1 W) whereas rapid switching of air flows is
performed with a Skinner micro valve (5 msec response
time).
The GC oven module consists of a simple heated
aluminum cylinder which houses the capillary GC
column, e.g., a 29 cm long, 50 \im i.d. fused silica
capillary coated with a 0.2 pirn thick layer of poly-
dimethylsilicone (DBS) and providing a continuous  He
flow of approx. .02 ml/min.

At present the oven is used in isothermal mode only.  A
temperature programming option as described by Arnold
et al. (4), which would allow a broader range of
compounds to be analyzed in a single GC run and also
help protect the column from oxidative degradation, has
not yet been implemented in the present prototype.  A
direct consequence of the rapid  GC run time is  the  need
for very high temperature programming rates, e.g, 10-20
C/sec.  This requires significantly larger power  supplies
than necessary for isothermal operation.

A small (2 ft3) compressed gas cylinder with flow
controller provides more than 36 hours of He or N2
carrier gas flow. The theoretical relationship between
inner column diameter, max. resolving power, column
length and retention time is depicted in Figure 2.
Obviously, the use of a 50 urn i.d. column (primarily
selected to keep gas flows as low as possible) has the
advantage of allowing very rapid separations, although
limiting maximum achievable resolving power.

Quadrupole MS Module

A Hewlett Packard Model 5971 MSD  (Mass Selective
Detector) was modified extensively in  order to reduce
system weight and power requirements and increase
overall manoeuverability.  The original housing was
completely discarded and the relative positions of the
electronic boards were changed to enable convenient
operation of the air sampling inlet.  The new
configuration is shown in Figures 4 and 5. Most
importantly, the original AC and DC power supplies were
removed and replaced by a battery powered 12  V DC
supply with DC/DC converters for the various DC
voltages required for mass spectrometer, computer and
sampling inlet operation.  Total power consumption of
the modified MS system was determined to be  43 W  (see
Table I).

Vacuum System

The vacuum system of the HP model 5971 MSD was
completely reconfigured to provide operating pressures in
the lO^-lO"5 torr range while minimizing roughing
vacuum requirements.  The original  60 I/sec diffusion
                                                        290

-------
pump was exchanged for an Alcatel Model 5010 MDP
JMolecular Drag Pump) with a max. pumping speed of 8
 sec'1 for N2 and a roughing vacuum requirement of < 30
millibar.  This enabled us to replace the original rotary
pimp (power requirement approx. 160 watts; weight 14
Jbs) with a simple vacuum buffer capable of maintaining
iroughing vacuum of better than 10 millibar for up to 12
iours at the specified GC column flows.  The vacuum
assembly configuration can be seen also in Figs. 4 and 5.
Micro Computer Module

A Toshiba model 5200, 20 Mhz, 80386 lap top is used to
control all GC/MS functions by means of a standard PC
interface and software available from Hewlett Packard.
In addition the PC system controls the operation of the
air sampling inlet. The only modification of the Toshiba
SlOO consisted of removing the built-in, relatively heavy
DC and AC power supplies and connecting the unit
directly to the specially constructed DC power supply
down in Figures 4 and 5.

SYSTEM INTEGRATION

Mechanically, the various components described thus far
were integrated  by means of a specially designed
Shoulder harness and backpack frame, as shown in Figure
5. The aluminum backpack frame carries the two
batteries as well as the vacuum reservoir whereas the
entire mass spectrometry assembly with MDP and PC is
Suspended from the shoulder straps and stabilized by two
ilp straps.  Due to the difficulty of typing in detailed
(omputer commands  during field use, especially when
tearing gloves, a beach ball type  mouse was installed to
fflable direct communication with a single (gloved) hand.
Alternatively, one could envisage  the use of a built-in  PC
Computer card (without display screen or keyboard)
icmotely controlled by a second, more completely
outfitted PC using standard PC software such  as Carbon
Copy® or PC Anywhere®.

The most simple remote control option would be to use
in umbilical cord carrying a twisted pair cable in addition
to AC power. The latter option would eliminate the
heavy (28 Ib) battery pack, thus resulting in greatly
reduced overall  size and weight.  Finally, as also shown
kFigure 4, a special transportable "docking station" (still
under construction) enables vacuum system regeneration,
fettery recharging and  carrier gas  refills at 6-10 hour
intervals.
PRELIMINARY TEST DATA

TLGC/MS curves generated with a 100 cm long,  100 jim
i.d. capillary column, coated with 0.25 jim
polydimethylsilicone (DB5, Supelco) while sampling a
mixture of 10 ppm vapor components in air for 1 sec at
30 sec intervals  are shown in Figure 6.  Obviously, a
highly useful level of chromatographic separation is
achieved with the very short transfer line.  Also the
narrow peak shapes (half height width < 1 sec) illustrate
the efficiency of the rapid sampling air inlet. Overall
peak height reproducibility (approx. + 10%) is influenced
by the limited resolution of the sampling time due to
manual operation.

From the selected ion profile (tropylium fragment ion at
m/z 91) in Figure 6 the minimum detectable
concentration in direct air sampling mode  appears to be
approx. 1 ppm.  Although this is 1-2 orders less than the
minimum concentrations detected by means of ion trap
type MS systems when using the automated vapor
sampling inlet (2), the MSD system has not yet been
fully optimized for operation under the present vacuum
and flow conditions.  However, since it may be
anticipated that some of the most promising applications
will require detection limits in the lower ppb range, a
suitable adsorption/desorption module is currently under
development in our laboratory.

Figure 7 illustrates the performance of the automated air
sampling TLGC/MSD system with polar compounds
under similar experimental conditions as in Figure 6.
Note the rapid separation of a mixture of ketones into its
components and the relatively minor degree of peak
tailing due to  the heated, all quartz vapor sampling inlet.
Finally, Figure 8 shows selected ion chromatograms for
several chemical agent simulants, demonstrating the fast,
repetitive (17 sec interval) analysis capability of the short
(29 cm) narrow bore (50 jxm i.d.) capillary column used
while maintaining adequate chromatographic resolution.

Although it is tempting to envisage the use of man-
portable GC/MS instruments for military reconnaissance
purposes, e.g., when venturing into contaminated regions
with high levels of background interferents, it should be
pointed out here that the current sensitivity of the MSD
based TLGC/MS system is insufficient for such appli-
cations.  Partially, this is due to the relatively low sample
mass flow  through the narrow bore capillary columns
used. In principle, this could be corrected by closing up
the MSD ion source thereby increasing the residence time
                                                       291

-------
of the vapor molecules in the source which would result
in increased ionization efficiencies.

Additionally, the use of rapid absorption/desorption
methods for sample preconcentration should be
considered.  Assuming a 10 second absorption interval at
10 times normal flow, followed by a 1 second desorption
interval at normal flow, it should be possible to obtain a
100 times enrichment factor without sacrificing analysis
speed.  Basically,  the 10-15 seconds necessary for
chromatographic separation is then being used to  collect
and preconcentrate the next sample.

Finally, we are investigating the use of rapid (10-20
C/sec)  temperature programmed heating in order to
broaden the range of compounds that can be analyzed in
a single chromatographic run.  The feasibility of this
approach has been demonstrated by Arnold et al.  (4).  A
second, important advantage of rapid temperature
programming is that the initial "air peak" passes through
the column at low temperature, thereby considerably
reducing the likelihood of oxidative degradation of the
column.  This then allows  programmed heating of the
column to high temperatures (e.g., 300 C) thus enabling
separation and detection of large polar molecules such as
underivatized trichothecenes, as demonstrated by
McClennen  et al.  (5). Many commercially available,  air
sampling mass spectrometry and ion mobility
spectrometry systems use silicone membrane interfaces,
thereby the detection of large, polar compounds.

CONCLUSIONS

The feasibility of constructing a fully man-portable
"transfer line" GC/MS system  with automated vapor
sampling capability has been demonstrated.  In its present
form, the  system weighs 72 pounds, consumes  160 W of
electrical power and can operate continuously for 6-10
hours.  Application of novel battery technologies, further
integration of the microcomputer module and use of
alternative vacuum pumping strategies is expected to
reduce overall system weight to less than 50 Ibs.
Without vapor preconcentration, practical detection limits
appear to  be in the low ppm range. Development of
rapid temperature programming capabilities is being
considered in  order to facilitate detection of relatively
nonvolatile species and to  increase the range of
compounds that can be analyzed in a single run.  The
ultralow power and weight requirements of the technique
would seem to offer promise for a broad spectrum of
field applications ranging from hazardous waste sites  and
industrial  or natural disaster areas to reconnaissance
drones, space stations, interplantary probes and
autonomous vehicular robots.
REFERENCES

1.      McClennen, W.H., Arnold, N.S., Meuzelaar,
       H.L.C., Apparatus and Method for Sampling. U.S.
       Patent 4,970,905.

2.      Arnold, N.S., McClennen, W.H., Meuzelaar,
       H.L.C., "A Vapor Sampling Device for Rapid,
       Direct Short Column Gas Chromatography/Mass
       Spectrometry Analyses of Atmospheric Vapors",
       Anal. Chem.,  in press.

3.      McClennen, W.H., Arnold, N.S., Sheya, S.A.,
       Lighty, J.S., Meuzelaar, H.L.C., "Direct Transfer
       Line GC/MS  Analyses of Incomplete Combustion
       Products from the Incineration of Medical Wastes
       and the Thermal Treatment of Contaminated
       Soils", Proc. 38th  ASMS Conf. on Mass Spec.
       All. Topics, Tucson, AR,  1990, 611-612.

4.      Arnold, N.S., Kalousek, P., McClennen, W.H.,
       Gibbons, J.R., Maswadeh, W., Meuzelaar, H.L.C.,
       "Application of Temperature  Programming to
       Direct Vapor Sampling Transfer Line GC/MS",
       Proc. 38th ASMS Conf. on Mass Spec. All.
       Topics, 1990, 1401-1402.

5.      McClennen, W.H., Meuzelaar, H.L.C., Snyder,
       A.P., "Biomarker Detection by Curie-point
       Pyrolysis in Combination with an Ion Trap Mass
       Spectrometer", Proc.  1987 CRDEC Conf., 271-
       277.
              ACKNOWLEDGEMENTS

       The authors acknowledge Jean-Luc Truche and
John Fjeldsted (Hewlett Packard Corp.) for their valuable
ideas and continued technical support and thank William
H. McClennen and Pavel Kalousek (University of Utah,
Ctr. Micro Analysis  & Reaction Chemistry) for their
expert  technical advice and assistance. This work was
financially supported by Hewlett Packard  Corporation
(University of Utah Instrumentation  Grant) and by the
Advanced Combustion Engineering Research Center.
Funds  for this Center are received from the National
Science Foundation, the State of Utah, 23  industrial
participants and the U.S.  Department of Energy.
                                                         292

-------
                  TABLE I:  PRIMARY SYSTEM COMPONENT SELECTION CRITERIA
                        Automated Vapor Sampling Inlet Module
                              fully automated
                              only inert quartz and fused silica materials
                              ultrashort sample "injection" pulse
                        Transfer Line GC Module
                              interferent rejection
                              rapid analysis capability
                        Hewlett Packard 5971A Mass Selective Detector
                              low power requirements (43 W)
                              lightweight (7 kg)
                        Alcatel 5010 Molecular Drag Pump
                              low power consumption (17 W)
                              high backing pressure up to 40 mbar (no backing pump
                              needed)
                              light weight (2.35 kg)
                        Toshiba 5200, 20 mhz, 386 Computer
                              low power consumption (40 W)
                              high speed, capable of running existing MSD software
                                                                   "o
                                                                    O
                                                                       1E5
                                                                       1E4
                                                                      1000
                                                                       100
N>
CO
CO
                    200
                    150
                Sf

                B   100
                O
                     50
                                          FINNIGAN ITD
BRUKER MM-1

      O
                         INCOS 500

                              D
            HP MSD 5971A

                 V
                                          SPECTRATRAK 600
                          MAN-PORTABLE
                                                                                     10         100        1000
                                                                                        Transfer Line Length (cm)
                                                                                                                                                 1E4
                                                                                                  100
                                                                                                    1E-2    1E-1
                                                                                                                      1       10      100
                                                                                                                      Retention Time (s)
                                                                                                                 1000
                                                                                               1E4-
                              500
           1000
                                              1500
2000
                                  2500
                                      POWER (Watts)

                Figure 1.  Power requirements and weights of typical
                miniaturized GC/MS systems (note that man-portable
                system includes power and carrier gas sources).
                                                                        Figure 2.  Theoretical relationships between internal
                                                                        column diameter (in \am), maximum achievable resolving
                                                                        power, column length and retention time for a compound
                                                                        with capacity factor k=5.0.  (Triangles indicate points of
                                                                        minimum plate height operation.)

-------
                                                              To Transfer Line
                                                                and Detector
                                                                 Sampling
                                                                 Mode
                                                       Inert
                                             Vacuum   Carrier
                                                       Gas
                        Sample  He    He +
                                     Sample
Control
System
J Vacuum



                                                      Inert
                                            Vacuum   Carrier
                                                      Gas
                                                              To Transfer Line
                                                                and Detector
Separation
Mode
Figure 3.  Operating principle of automated vapor sampling inlet developed at University of Utah (US patent no.
4,970,905).
            HP MSD ANALYZER
             RF GENERATOR


QUADRUPOLE
DETECTOR

HP HARDWARE
INTERFACE

VACUUM SYS


CEN
PORTA
MSDD
INLET (
TEM
MOLECULAR DRAG PUMP
VACUUM RESERVOIR
                                               VAPOR INLET
                                             SAMPLE PUMP
                                             GC OVEN HEATERS
                                             CARRIER GAS
                            CENTRAL DATA SYSTEM
                           PORTABLE 386 COMPUTER
 SAMPLE IN
                                                                         "DOCKING STATION"
                                                POWEF  SYSTEM
                                             24 VDC BATTERY
                                             DC/DC CONVERTER
       REFILL CARRIER GAS

       RECHARGE BATTERY

       EVACUATE RESERVOIR
        Figure 4. Block diagram of man-portable GC/MS system and docking station interface.
                                                294

-------
                                       Figure 5.  Schematic outline of GC/MS man-portable
                                       system with operator.  A) vapor inlet/transfer line GC
                                       column; B) MSD analyzer;  C) control electronics; D)
                                       portable 386 computer; E) molecular drag pump; F)
                                       vacuum hose;  G) vacuum reservoir; H) carrier gas, and I)
                                       24VDC battery.
                                     12000-1
                                     10000
                                   3  8000
                                   —
                                   B
                                   a
                                      6000
                                   3  4000
                                   o


                                      2000
                                            1.20   1.40   1.60   1.80   2.00   2.20
                                                                 Time (minutes)---
                                                                               2.40   2.60   2.BO   3.00   3.20
Figure 6.  Selected ion chromatogram profile of an alkylbenzene mixture at m/z 91 obtained by TLGC/MS using
the automated vapor sampling inlet in combination with a  100 cm long, 100 [xm i.d., DBS coated fused silica
capillary column.  (1) toluene; (2) ethyl benzene; (3) m-xylene; (4) o-xylene.  Approximate vapor concentrations:
lOppm.  Arrows  indicate air sampling events at 30 second intervals.  Note that o-xylene (peak 4) elutes after the
next sampling event.
         3944
                                   i.2o '' i.« " i.to   i.bo
                                      Time (minutes)—	>
 Figure 7.  Total  ion chromatogram (TIC) for a mixture of 4 ketones. 1) acetone; 2) methyl ethyl ketone; 3) ethyl
 acetate; 4) 3-pentanone; 5) methyl iso-butyl ketone.
                                                        295

-------
    «      100 -
                                                                  m/z 79, DiVEVEP
                                                                   z 111, DEEP
                     6 sec
                                              Time (s)
Figure 8. Selected ion chromatograms of 4 chemical agent simulants (DMMP=dimethyl methyl phosphonate,
DEEP=diethyl ethyl phosphonate, DlMP=diisopropyl methyl phosphonate, DEM=diethyl malonate). Arrows
indicate  air sampling points (17 sec interval).  Note separation of all 4 simulants within 6 sec. Star symbol (*)
indicates "pseudo" peak due to effect of eluting air on MS system.
                                                    296

-------
                                                         DISCUSSION
IALPH SULLIVAN: With these high flow rate systems, how did you go about
calibrating it and how do you introduce the gas to it to know what you have in
the system?

HENK MEUZELAAR: You make diluted air — the flow rate doesn't have to
fe above 100 mL per minute, or even 50 per minute. So, if you have a dilution
tjrstem that can give you that kind of output, you can just calibrate it with a
calibrated dilution system.

AUDIENCE PARTICIPANT: Could you repeat that?

HENK MEUZELAAR: All right. What I said is the high flow of the outer tube,
tie first sampling tube, can be as little as 50 or 100 mL per minute. So, if you have
tvapor dilution system that can give you a couple hundred mL output you can
4) a loose coupling for such a system and get very good results. If you have a
tapor dilution system that just puts out a few mL per minute it would be more
difficult to do that. You could do it from a bag if you could fill a bag and keep it
at atmospheric pressure for several minutes, you could obtain a sample without
changing the pressure or the concentration in the bag.

BILL McCLENNY: I was wondering what the prospects would be for using
some type of preconcentration that involved a cold trap, using thermo electric
cooling  or something of that sort, and what that would add to the  power
requirements for this unit?

HENK MEUZELAAR: I think almost any type of absorption, desorption, or
preconcentration by any method I know that would keep the high response
characteristic intact, would certainly require power because you would have to
desorb for a relatively short period of time. And the only way to make gain is to
absorb for let's say 60 seconds and flush desorb in one or two seconds. That's
going to require power. We are currently looking at a number of  different
methods. The power requirement is just needed, for a second, or maybe even less
than that. I think it's a doable thing, but it certainly will  add to the  power
requirement.
                                                                   297

-------
                                 ON-SITE MULTIMEDIA ANALYZERS:
                       ADVANCED SAMPLE PROCESSING WITH ON-LINE ANALYSIS
      S. Liebman
      GEO-CENTERS, INC.
      c/o U.S. Army Cml
        Rsch, Dev & Engr Ctr
      Attn:  SMCCR-RSL
      Aberdeen Proving Ground,
        MD  21010-5423
M. B. Uasserman
U.S. Army Cml Rsch,
   Dev & Engr Ctr
Attn:  SMCCR-RSL
Aberdeen Proving Ground,
   MD  21010-5423
E. J. Levy and S. Lurcott
Computer Chemical Systems, Inc.
Rt. 41 and Newark Rd.,  Box 683
Avondale, PA  19311
ABSTRACT

The need for on-site chemical analysis
of air,  water,  and soils has led to
development of two highly automated
prototype instruments in the field of
trace organic analysis:   EPvA, the
Environmental £yroprobe  Analyzer and
CHAMP,  the Chemical Hazards Automated
Multiprocessor.   In the  EPyA unit, a
purge and trap module permits routine
determination of target  chemicals in
water and hazardous wastes.  A thermal
desorption module permits controlled
thermal desorption of air sampling
cartridges,  as well as dynamic
headspace/pyrolysis analyses of
solids.   CHAMP is based on supercriti-
cal fluid extraction (SFE) with liquid
C0£ mobile fluid for solid samples in
amounts from milligrams  to over
several grams in six individually
heated extractors.  Specialty
interfaces,  such as TRANSCAP. provide
on-line analysis by chromatographic
and/or spectral detectors.

Both benchtop,  microprocessor-based
systems are newly designed for in-
field operation, as well as laboratory
or plant sites.   Highly automated
instruments such as EPyA and CHAMP
operating with external expertise
provided by artificial intelligence
(AI) software,  illustrate the Inte-
grated Intelligent Instrument (I3)
approach which is focused on multi-
media analyses for hazardous
materials.
                     INTRODUCTION

                     Advantages  of precision,  accuracy,  and
                     reproducibility are realized with the
                     use of automated instruments to per-
                     form thermal  and nonthermal sample
                     processing with on-line chromato-
                     graphic and/or spectral analyzers.
                     New.engineering designs are required
                     to bring this analytical power on-site
                     to the field, mobile lab, or plant to
                     provide rapid, validated information
                     to analysts.   Two prototype analytical
                     systems are described to meet these
                     needs; the Environmental Pyroprobe
                     Analyzer,  EPyA (1) and CHAMP, the
                     Chemical Hazards Automated Multi-
                     Processor (2).  The prototypes are
                     designed for  compactness with inte-
                     grated specialty separation and/or
                     detector units that are important to
                     the hazardous waste field for on-site
                     use.   Figure  l(a,b) shows the bench-
                     top units,  each about 2'x2'x3' and
                     weighing ca.  eighty pounds.   The pur-
                     pose of this  report is to describe the
                     ongoing development of specialty in-
                     strumentation that is based on proven
                     analytical methodologies in trace
                     organic analysis.

                     I.   Thermal Sample Processing - EPyA

                     The thermal analyzer system,  EPyA.  is
                     the result of over fifteen years of
                     engineering design and manufacture  of
                     microprocessor-based instrumentation
                     used throughout the world for trace
                     organic analysis of vapors,  liquids
                     and solids.   Studies in the  70's and
                                                299

-------
80's  developed purge  and trap modules
for water analyses  and  thermal desorp-
tion  methods for  rapid  analyses of air
sampling cartridges that contained
treated charcoals,  porous polymers,
Ambersorb, Tenax, etc.  (3a).  Figure 2
shows a test air  mixture with 40 ppb
levels of typical solvents  (benzene,
toluene, chloro-benzene,  heptane, o-
dichlorobenzene,  and  dodecane) sampled
for 90 sec at 0.5 mL/min on a Tenax
sorbent bed (100  mg)  which  was then
thermally desorbed  for  GC/FID analysis
(3b-e).  Figure 3 shows an  analysis
conducted for gasoline/fuels using a
cryofocusing concentrator module and
on-line capillary GC/FID detection
(4).  Figure 4a,b,c give results from
other studies (4) using remote air
sampling cartridges for analyses of
outside air, laboratory air, and paint
shop  air (all 500 ml  samples) with
GC/FID analysis.

Recently, the thermal desorption/cyro-
trapping module was used in trace
particulate analysis  of a mlcroencap-
sulated pesticide,  Diazinon, in an air
sampling cartridge  with on-line analy-
sis by GC-MS (5)  (Figure  5).  A
corresponding dynamic headspace/py-
rolysis method using  the  Pyroprobe Pt
coil  pyrolyzer on a few micrograms of
a microencapsulated sample  also
provided trace detection and
identification of the Diazinon core,
which gives a parent  ion  at m/z 304
and a base peak at  m/z  179.  Clearly,
thermal desorption, rather  than C$2
solvent stripping,  proved to be the
optimum analytical  method which is now
used  throughout the world in the
industrial R&D,  forensic, and
environmental fields.   However, some
thermally sensitive samples required
additional effort for reliable
analyses.   A more effective method
than  solvent extraction was needed,
both  for analyzing  thermally labile
materials,  as well  as to  eliminate
solvent wastes.   The  traditional
Soxhlet solvent extraction  method has
further disadvantages of hour or day-
long  extraction times and off-line,
more  labor intensive, multistep
analyses for complex  environmental
samples.
 II.   The Nonthermal Sample Processing
      Analyzer - CHAMP

 The  nonthermal multiple sample proces-
 sing system,  CHAMP, using supercriti-
 cal  fluid (SF) technology (6)  permits
 the  conduct of trace organic analysis
 on diverse samples, including cart-
 ridge sorbent beds (7), soils, coals,
 or hazardous waste solids.  Six
 individually heated sample extractors
 may  contain up to five grams or more
 of material to be treated near or at
 supercritical fluid conditions in the
 2, 4, or 6 mL extractor vessels.

 Automated SF extraction (SFE)-capil-
 lary GC analysis of gasoline from
 charcoal filters may be routinely
 analyzed with either single or
 multiple SFE units.  Analytes re-
 quiring well-established capillary GC
 methods use the automated SFE system
 configured for GC separation.
 Alternatively, in Figure fc, the SFE-
 SFC  analysis is shown of a phosphonate
 chemical in soil (ca. 500 mg)  with
 detection by FID at estimated ppb
 levels.  The SFE was conducted at 3000
 psi, 100°C with C02, which is a
 nontoxic, safe and inexpensive mobile
 fluid.  The SFC was conducted with a
 Nucleosil CN microbore column at 120°C
 and  pressure programming from 2000 to
 6000 psi at 300 psi/min with a FID
 unit.

 As with the thermal processing
 analyzer, EPyA. it is necessary to
 have a variety of detection systems
 for  adequate analytical sensitivity
 and  specificity.  The SFE process has
been used with FID, ultra-violet  and
mid-infrared  (ir) spectrometers using
fiber optic monitors  (FOM)  (6,8).
Figure  7  represents the recent on-line
SFE-SFC analysis of a polyolefin/
naphthalene mixture.  An ion trap MS
detector  (ITD) was used to detect the
molecular ion  from naphthalene (m/z -
128)   (9).  Other detectors show
similar potential for trace on-line
analyses with highly specific and
sensitive responses to hazardous/toxic
substances, e.g., fluorescence/uv with
fiber optic technology and advanced
data  analysis with applied AI  (10).
Both  EPvA and CHAMP incorporate new
design engineering features that
emphasize compact, transportable
systems.  Sample processing,
integrated with separation and
                                                300

-------
detection units are controlled by
microprocessors with programmable,
interactive software.  External AI
software  will  provide guidance in the
use  of  the total system.


 III.  Applied Artificial  Intelligence
      - Expert  System Networks

 The I3 approach combines  data
 generation using highly automated
 modular/interfaced systems with
 external intelligence for development,
 data analysis, interpretation and
 validation.  Development  of a
 proprietary expert system network for
 SF technology, MicroEXMAT, has been
 reported using CCS SF hardware and
 methods (11).   Currently, a multi-
 variate experimental design based on a
 Box-Behnken central composite is
 linked explicitly in the network  via
 an expert system, EXBOXB.  Further
 integration of MicroEXMAT into a  full
 laboratory information management
 system (LIMS)  was also outlined
 previously (12).  Applications to EPyA
 and CHAMP are  being developed.  The
 recent  ACS Symposium on expert systems
 applied to the environmental field
  (13) indicates the growing importance
  of AI  in analytical chemistry.

  IV.  Summary

  Newly designed instrumentation for
  multimedia (air, water,  solids)
  environmental trace organic analysis
  is described  for on-site applications.
  The automated prototype units feature
  advanced sample processing with
  interfaces for on-line analyses with
  chromatographic and/or spectral
  detectors.  Thermal sample processing
  is provided by EPyA. including modules
  for purge and trap/thermal desorption,
  dynamic  headspace,  and pyrolysis.
  Nonthermal multi-sample  processing is
  conducted with CHAMP based on super-
  critical fluid extraction and
  specialty interface units.   Analyses
  of low ppb levels  of vapors,  aerosols/
  particulates,  gasoline,  and soils
  illustrate the proven capabilities of
  the integrated modular systems.   A
  developing expert  system network,
  MicroEXMAT, encodes expertise to guide
  analysts in analytical strategy,
  instrumental  configurations,  and
  aethod development  for the  proposed
  on-site  analyzers.
REFERENCES

1.  Manufactured by CDS Instruments,
Division of Autoclave Engineers,
Oxford, PA.
2.  Manufactured by Computer Chemical
Systems, Inc., Avondale, PA.
3.  (a)  Michael, L.C., Pellizzari,
E.D., Norwood, D.L., Environ. Sci.
Technol.. 25, 150-155 (1991), "Appli-
cations of the Master Analytical
Scheme to Determination of Volatile
Organics in Wastewater Influents and
Effluents."
    (b)  Applications Laboratory,
Chemical Data Systems, Inc., Oxford,
PA.
    (c) Liebman, S.A., Ahlstrom,

 D.H.,  Sanders,  C.I.,  First FACSS
 Mtg.,  Atlantic  City,  NJ,  Nov 1974.
 "Automatic Concentrator/GC System for
 Trace  Analysis."
     (d)  Ahlstrom,  D.,  Kilgour,  R.,
 Liebman,  S.,  Anal  Chem..  47.  1411
 (1975),  "Trace  Determination of Vinyl
 Chloride Monomer by a Concentrator/GC
 System."
     (e)   Liebman,  S.A.,  Wampler,
 T.F. ,  Levy,  E.J.,  EPAInternat.
 Sympos.  on Recent  Advances in
 Pollutant Monitoring of Air,  Raleigh,
 NC, May  1982, "Advanced
 Concentrator/GC Methods  for Trace
 Organic  Analysis."
 4.  Applications Lab.,  CDS
 Instruments/ Division of Autoclave
 Engineers,  Oxford,  PA.
 5.  Liebman,  S.A.,  Smardzewski, R.R.,
 Sarver,  E.W., Reutter,  D.J.,  Snyder,
 A.P.,  Harper, A.M.,  Levy,  E.J.,
 Lurcott,  S.,  O'Neill,  S.,  Proc. Poly-
 meric  Materials  Science  and Engineer-
 ing. 5.2,  621-625, Amer.   Chem.
 Soc.,  Los  Angeles,  CA,  September  1988.
 6.   (a)   Liebman,  S.A.,  Levy, E.J.,
 Lurcott,  S.,  O'Neill,  S.,  Guthrie,  J.,
 Yocklovich,  S.,  J.  Chromatogr.  Sci..
 27, 118-126  (1989),  "Integrated
 Intelligent  Instruments:  Supercritical
 Fluid  Extraction, Desorption, Reaction
 and Chromatography."
 7.  Raymer, J.H., Pellizzari, E.D.,
Anal.  Chera..  59, 1043, 2069  (1987),
 "Toxic Organic Compound Recoveries
Using  SF C02  and Thermal  Desorption
Methods."
 8.  Liebman,  S.A.,  Fifer,  R.,
Griffiths, P.R., Lurcott,  S., Bergman,
B., Levy,  E.J.,  Pittsburgh Conf..March
 1989,  Atlanta, GA,  Paper  No.  1545,
 "Detection Systems  for Supercritical
Fluid/GC Instrumentation:   Flame
                                                 301

-------
lonization Detector (FID) and Fiber
Optic Monitor (FOM) Units."
9.  Liebman, S.A., et al., Pittsburgh
Conf., March 1990, NY, Paper No. 546,
"New Applications of I3  in Trace
Organic Analysis."

 10.  (a)  Siddiqui,  K.J.,  Eastwood,  D.,
 Lidberg,  R.L.,  SPIE.  1054.  77-90
 (1989),  Fluorescence Detection III:
 Soc.  Photo-Optical Instrument.  Eng. ,
 Bellingham, WA, "Expert System for
 Characterization of Fluorescence
 Spectra for Environmental
 Applications."
     (b)  Eastwood, D., Lidberg, R.L.,
 Simon, S.J., VO-Dinh, "An Overview
 Advanced Spectroscopic Field Screening
 with In-Situ Monitoring Instrumenta-
 tion and Methods, private communica-
 tion.
 11.  Liebman,  S.A., Fifer, R.,
 Morris, J., Lurcott, S., Levy, E.J.,
 Intelligent Instruments and
 Computers. May/ June 1990, pp  109-120,
 "An Expert System Network for
 Supercritical  Fluid Technologies."
 12.  Liebman,  S.A., Snyder, A.P.,
 Wasserman, M., Brooks,  M.E.,
 Watkins, J., Lurcott,  S., O'Neill, S.,
 Levy, E.J., Internat.   Conf. on Anal.
 Chem., University  of  Cambridge, UK,
 July/Aug,  1989,  "Integrated
  Intelligent Instruments in Materials
  and  Environmental  Sciences."
  13.  Hushon, J.M.,  Ed.,  ACS  Sympos.
  Series  431, Amer.  Chem.  Soc.,
  Washington DC, 1990,  "Expert Systems
  for  Environmental Applications."
                       (a)
                              EPyA
                                                                CHAMP
                                                                        (b)
                                       SPECIALTY INTERFACES TO
                                       FTIR AND MS, MS/MS SYSTEMS
                            ANALYTICAL
                            PYROLYSIS MODULE
                            —PYROPROBE*
  TRANSCAP INTERFACES
TO GC/SFC, FTIR, MS SYSTEMS
           Figure 1.   (a)  EPyA, the Environmental Pyroprobe Analyzer
                       (b)  CHAMP, the Chemical Hazards Automated Multiprocessor
                       (c)  TRANSCAP Interface to Finnigan TSQ MS/MS
                                                  302

-------
                   CARTRIOQE SAMPLING FOB LOW PPB LEVELS
                 OF HALOCAHBOMS, ALIPHATIC!. AND AROMATIC*

                           TEST AIR MIXTURE ANALYSIS
                              TENAX CARTRIDGE
                   CHLOHOMItZtNl      «» frt

•INZCNE



J











--»_>^.






8AHPLINQ

• MIN. 40 •! N«
oc.
OURAIONO 01 « 3 OH K O >I m 1 U FILH
• */MI«. TO 17**C
ATT'M 10"11 * 11
Figure  2.   Cartridge Sampling  for Low PPB  Levels of Halocarbons,
            Aliphatics, and Aromatics in Air with Thermal Desorber  Module
                      TUT SAMPLES WITH WIDE-RANQINQ VOLATILES
                FOR CBYOTHAPPIMO. DESORPTIOM AND CAPUAMY OC ANALYSIS
                        SAMPLE CONCENTRATOR CDS S3O/GC
                                        iri.ITl.Ilt C»HU.»«» OC
                                         WITH CHYOFOCUIM*
   Figure 3.  Gasoline and  Diesel Fuel  Test Mixture  Analyzed with
               Cryotrapping,  Thermal Desorption, and  Capillary GC/FID System
                                     303

-------
                                                      (a)
                                     OIRECT COtUMH CUTOFOCUtlMQ

                                        SflO -1 OUTSIDE Alfl
                               • ETCNTION TlMf 10
                                          ISO ml LABOtUTOftT AMI ( b )
                                                     (Cj


                                             SOOml PAINT SHOr A.IN
                                MIEHTKM TIME 10
Figure 4.  Air Monitoring on Tenax Cartridge with Direct Colu»n
           Cryofocusing, 500 ml Sample
           (a) Outside Air, (b) Lab Air,  (c) Paint Shop Kir
                           SAMPLE CONCENTRATOR - QC/M*1S SPECTROMETER
                            CARTRIDGE AEROSOL/PARTICULATE
                                                      (a)
                                                 RECOWTHUCTTD Kttt
                                                  CKftOHATOOAAM
                                                       (b)
                Figure 5.  (a) Reconstructed Ion Chromatogram of Cartridge
                               Aerosol/Particulate.  Thermal Degradation
                               (260°C/5 min) GC-MS Analysis of Microcap fl
                           (b) Electron Impact MS of Scanset 1529
                                              304

-------
                SUPERCRITICAL FLUID EXTRACTION-CHROMATOGRAPHY
                       EXTRACTION
                                SIMULANT IN SOIC
                               1880,Hgm
                        CHROMATOGRAPHY
                              BIS-CETHYLHEXYL)
                                PHOSPHONATE
                                               >
Figure 6.   Supercritical Fluid Extraction-Chromatography (SFE-SFC/FID)
            of Bis-(Ethylhexyl)phosphonate,  3000 psi CO2 Mobile Fluid
          SUPERCRITICAL FLUID EXTRACTION-CHROMATOCRAPHY
                                               (a)
                                         SFE-SFC/ITD
                                              (b)
                                         SPECTRUM
                     11'I. IP A*"'.*?'!1 """."?'" w. .,*J
                           m    in    MI    s
 Figure  7.   SFE-SFC Interfaced to Ion Trap  Detector (ITD) of Polyolefin
             Mixture with Naphthalene
             (a)  Reconstructed Chromatogram
             (b)  Mass Spectrum,  m/z 128
                                  305

-------
               USING A FID-BASED ORGANIC VAPOR ANALYZER IN CONJUNCTION WITH
              GC/MS SUMMA CANISTER ANALYSES TO ASSESS THE IMPACT OF LANDFILL
                GASSES FROM A SUPERFUND SITE ON THE INDOOR AIR QUALITY OF AN
                                  ADJACENT COMMERCIAL PROPERTY
                                              Thomas H. Pritchett
                                     U.S. Environmental Protection Agency
                                                  Edison, NJ
                                      David Mickunas and Steven Schuetz
                                         IT Corporation, REAC Contact
                                                  Edison, NJ
The ERT was tasked to access the degree that VOCs, which
may have been co-migrating with methane from a Superfund
site, were affecting the indoor air quality of a shopping mall.
Of particular concern to the Region was the fact that the mall
had actually been built on top of the site prior to its being
added to the NPL, The actual assessment used a combination
of both field screening methods and fixed laboratory meth-
ods to gather two separate sets of data: one set on the landfill
gases and the other set on the air inside the mall. OVA,
Explosivity, and HNU readings from all of the landfill vent
were used to select the vents from which the Summa canis-
ters would be taken for GC/MS and permanent gas analyses.
Concurrent with Summa sampling, the inside of the mall was
screened using an OVA - particularly at all of the likely
entry points for subsurface gases.
The analytical results were interpreted as follows: The
Summa results were used to determine the "worst case" ratio
of target compound to methane observed in the vent gases.
These values were then multiplied by the worst OVA read-
ings observed in the vicinity of a likely soil gas entry point
in order to predict the highest possible concentration of
VOCs that could have been present due to co-migration with
the methane from the landfill. These "worst case" predic-
tions clearly indicated that there was not an apparent long-
term health risk due to VOC migration from the landfill.
                                                    307

-------
               FIELD ANALYTICAL  SUPPORT PROJECT (FASP) USE TO PROVIDE DATA FOR
                 CHARACTERIZATION OF HAZARDOUS WASTE SITES FOR NOMINATION TO
                              THE NATIONAL PRIORITIES LIST (NPL):
                      ANALYSIS OF POLYCYCLIC AROMATIC HYDROCARBONS (PAHS)
                                  AND PENTACHLOROPHENOL (PCP)
                   Lila Accra Transue,  Andrew Hafferty, and Dr. Tracy Yerian
                                    Ecology and Environment
                                   101  Yesler Way,  Suite 600
                                  Seattle,  Washington  98104
ABSTRACT

The path  from  initial discovery of a site
as potentially contaminated to its inclu-
sion on the National Priorities List (NPL)
requires  numerous activities,  most impor-
tantly the identification and quantitation
of hazardous wastes or contaminants asso-
ciated with  the site and the surrounding
area. New guidance for NPL nomination
places greater emphasis on accurate deter-
•ination  of  the areal and volumetric extent
of contamination during the site assessment
phase of  work.  Under this guidance, exten-
sive sampling  is a prerequisite for charac-
terization of  a site.  This places a heavy
burden on the  United States Environmental
Protection Agency (EPA) regions' ability to
provide quality assurance oversight for
data generated by Contract Laboratory Pro-
gram (CLP) analysis of these samples, and
adds considerable costs and time to the
nomination  process.  If the contaminants of
concern have been identified previously, it
•ay be appropriate to characterize  the site
using field  analytical support.  In Region
 10,  the  Field Analytical Support Project
 (FASP) program has been integrated  into  the
 Screening Site Inspection  (SSI) and Listing
 Site Inspection  (LSI) process  to provide
 cost savings and near real-time analytical
 information about  the site.  FASP methods
 are designed  to  meet  the data  quality ob-
 jectives (DQOs)  established for each site.
 All FASP data used for site characteriza-
 tion are confirmed by analyzing 10  percent
 of the samples collected for  full Target
 Compound List  (TCL)  analysis  through  the
 CLP. Gas chromatographic  methodologies  for
 field analysis of  selected PAHs and PCP
have been developed for FASP in response
to a regional need for site characteriza-
tion at wood treating facilities.  FASP
methods are developed for small volumes,
rapid extraction and analysis, and minimum
labor intensity.  Methods developed for
FASP will be presented, as well as the
results from two LSIs, including a compari-
son of .FASP data to CLP confirmation  re-
sults at each site.
INTRODUCTION

The United States Environmental  Protection
Agency  (EPA), under  the  Superfund  Amend-
ments and Reauthorization  Act  of 1986
(SARA), uses  the National  Hazardous  Waste
Site Investigation program to  identify
hazardous waste sites  for  inclusion  on the
National Priorities  List (NPL).  Ecology
and Environment, Inc.  (E & E)  holds  the
Zone 2  Field  Investigation Team  (FIT)
contract, under which  potential  hazardous
waste sites are investigated,  and  relative
risks and threats  to human health  and  the
environment are evaluated.  FIT  assists the
EPA  in  its goal of  identifying sites for
the NPL in three stages:  1) Preliminary
Assessments  (PAs),  2)  Screening  Site
Inspections  (SSIs),  and  3) Listing Site
Inspections  (LSIs).   A potential hazardous
waste site would go through all  three
phases  before it could be listed on the
NPL.

In 1988,  EPA released the proposed re-
visions to  the Hazard Ranking System (HRS),
which  is used to  score potential hazardous
waste  sites  based  on an assessment of rela-
                                                 309

-------
tive risks.  Prior to the revised HRS
(rHRS), the extent of contamination at a
hazardous waste site was determined only
after the site actually was placed on the
NPL (during the remedial investigation
phase of site cleanup).  The rHRS includes
nev guidance for nomination to the NPL, and
places greater emphasis on accurate deter-
mination of the areal and volumetric extent
of contamination during the site assessment
phase of work.  Coupled with congressional
mandates aimed at streamlining the listing
process, the new guidance places a heavy
burden on the limited analytical resources
available in terms of the number of samples
required for accurate site characteriza-
tion, and rapid turnaround of analytical
data after sample collection.

The site assessment program obtains most of
its required data through the EPA CLP,
since the CLP provides cost-effective
analyses for a large number of
contaminants.  Sometimes, however, it may
be impractical to utilize the CLP to
characterize a site if preliminary data are
already available that identify the target
analytes of concern.  The costs and time
involved with a large-scale sampling plan
can be minimized by tailoring the type of
sample analyses performed to the specific
project needs.  Also, information obtained
from the laboratory during the sampling
event may allow the field team to optimize
sample locations for proper identification
of site boundaries, while minimizing the
total number of sample analyses required.
These types of laboratory interaction and
sample location tailoring are currently
difficult to obtain through CLP Routine
Analytical Services (RAS).

In addition, RAS contract required quanti-
tation limits may not be adequate to deter-
mine the extent of on-site contamination at
sites where the NPL listing criteria estab-
lishes a need for the lowest obtainable
quantitation limits.  It also is possible
that CLP methodology may be inappropriate
under specific matrix conditions present at
a site, potentially resulting in further
elevation of the quantitation limit above
required action levels.  Determination of a
matrix interference in advance, through
real-time analysis, may allow for modifica-
tion of the CLP method as requested through
the Special Analytical Services (SAS) pro-
cess, to minimize the necessity of resamp-
ling.
This paper describes an alternative to the
exclusive use of full organics and in-
organics CLP RAS analysis of samples col-
lected during the SSI and LSI processes.
When compared to CLP RAS, this alternative
often results in cost and time savings
while providing analytical information that
satisfies the data quality objectives
(DQOs) for each site.
DQOs

DQOs are statements regarding the level of
uncertainty that a data user or decision-
maker is willing to accept in results de-
rived from environmental measurements.  The
DQO process is designed to help the data
user match quality needs with the appro-
priate analytical laboratory and methods so
that the right type, quality, and amount of
data are collected (1).

When applied to hazardous waste site inves-
tigations, the DQO process provides a quan-
titative basis for designing rigorous, de-
fensible, and cost-effective investiga-
tions.  The DQO planning process recognizes
that decision making is driven by regula-
tory requirements and by risks to public
health and that the uncertainty in deci-
sions will be affected by the type and
quality of data collected.  DQOs provide a
qualitative and quantitative framework
around which data collection programs are
designed, and can serve as performance
criteria for assessing projects (2).

DQOs determine the level of analytical sup-
port necessary to provide decision-makers
with sufficient confidence upon which to
select options with known levels of uncer-
tainty.  Choice of specific analytical op-
tions may be determined by:

o  Health-based concerns,
o  Sample analysis cost,
o  Analytes of concern or target/indicator
   analytes,
o  Regulatory action levels that dictate
   method quantitation limits,
o  Sample matrices,
o  Sample collection, handling, and storage
   requirements, and
o  Statistical uncertainty in the qualita-
   tive identification of analytes and
   errors associated with the quantitation.
                                                  310

-------
All of the above considerations must be
weighed to determine the appropriate analy-
tical needs for the project data.  Rarely,
if ever will a single analytical program
provide the best technical information and
the most cost effective solution to address
all concerns at the site.

The "art" of field analytical support is  to
natch analytical capability to the DQOs re-
quired for a specific site in a cost-effi-
cient manner.  Once the acceptable level  of
error in the result is determined, the
acceptable level of inherent error in the
Measurement system can be addressed.
FIELD ANALYTICAL SUPPORT PROJECT (FASP)
PROGRAM

Broadly defined, field analytical support
is the use of chemists in an analytical
laboratory at or near the site of a hazar-
dous waste investigation, removal, or  re-
Bed ial action.  Field analytical support is
•ore than a facility or vehicle stocked
with instrumentation, glassware, and
expendables; it is the interactive
management process by which decision-makers
and the personnel who provide the
analytical results integrate planning,
execution, and assessment of analytical
data collection into environmental studies.
These procedures form the basis of the FASP
program.

In the late 1970s and early 1980s, field
analytical support for determinations  of
 contaminants at hazardous waste sites  was
 almost exclusively restricted to health and
 safety monitoring of on-site personnel.
 Early site screening was limited primarily
 to air monitoring for volatile organic com-
 pounds with hand-held instruments such as
 the HNu PI101 (photoionization detection)
 and the Foxboro OVA (flame ionization
 detection).  Within the last decade, more
 sophisticated analytical instrumentation,
 such as portable (hand-carried) and  trans-
 portable  (mobile laboratory supported) gas
 chromatographs and light-weight, compact
 X-Ray fluorescence and atomic absorption
 analyzers, have begun to be employed rou-
 ;tinely  in hazardous waste site investiga-
 tions.  These new instruments, coupled with
 field-experienced chemists, have provided
 near real-time organic and inorganic
 analyses  for contaminants in air, soil,
 Vater,  and other matrices (3).
Under E & E's Zone 2 FIT contract, a FASP
program was initiated in 1984.  The main
purpose of FASP is to support the PA, SSI,
and LSI process by utilizing field analyti-
cal methods to provide useful information
about site contaminants on a real- or near
real-time basis.  FASP can be a cost- and
time-effective alternative or supplement to
conventional laboratory sample analysis in
many situations.  Turnaround time for
conventional laboratory analyses, such as
CLP RAS is 40 days after receipt of the
samples.  CLP data for site assessment
activities must undergo data validation by
a FIT chemist which takes approximately two
weeks.  By contrast, FASP data are
generally provided verbally within 24 hours
of sample receipt, and a final deliverable
is often available approximately 14 days
after the project is completed.  FASP data
are evaluated during laboratory projects.
Additional data validation time is not
required.

The EPA recognizes that field analytical
methods such as FASP provides, are appro-
priate for many decisions made in Superfund
(American Environmental Laboratory, October
1990).  The EPA encourages the use of  these
field analytical methods for screening,
monitoring and other assessments requiring
rapid turnaround of data, and for decisions
where unconfirmed analyte identity and
estimated concentrations are appropriate.
FASP methods are currently included in
EPA's revised Field Analytical Methods
Catalogue.  FASP data have been used  to:

o  Optimize sampling grids,
o  Select groundwater well screen depths,
o  Guide remedial disposal requirements,
o  Provide guidance to  cleanup contractors,
o  Assist in spill response,
o  Select well locations based on soil gas
   monitoring,
o  Provide enhanced site characterization,
o  Identify the most appropriate samples
   for CLP analysis,
o  Estimate waste quantities,
o  Determine extent of  contamination  migra-
   tion, and
o  Find  "hot-spots".

FASP  is  not a replacement  for or an equiva-
lent  of  the EPA CLP.   FASP does  provide
real-time data of known (legally admis-
sible) quality, which  may  be  used  in  situa-
tions where data generated by a  certified
laboratory and  standard methodology  is not
a  requirement for decision making.  All
FASP  analytes are,  by  definition,  tenta-
                                                 311

-------
lively identified, and all FASP quantita-
tive data are estimated concentrations be-
cause methods and quality control (QC) are
a subset or variants of standard CLP QC.
Although both qualitative and quantitative
accuracy and precision may nearly equal
CLP, no attempt is made to alter these
limitations.  Therfore, to properly iden-
tify FASP data as tentatively identified
with estimated concentrations, all FASP
data in Region 10 are annotated with the
qualifier "F".  This qualifier also indi-
cates that field methodologies were
employed to generate the data.

FASP often is used at sites where previous
sampling has been performed and target
analytes have been identified.  When
analytes have been identified previously,
unambiguous identification (i.e., mass
spectral detection) may not be required.
FASP is used most efficiently in the
analysis of samples for a limited group of
analytes requiring only one or two analyti-
cal methodologies.  FASP is not used rou-
tinely for analysis of samples for unknown
contaminants.
FASP STANDARD OPERATING GUIDELINES (SOGs)

The FASP program functions under SOGs that
provide guidance on general QC and analyte-
matrix-specific methodologies which have
been developed within the FASP program.
Methodologies are developed on an as-needed
basis, to accommodate the FIT program, or
any other program in which FASP is uti-
lized.  FASP methods are designed to pro-
vide near real-time data to field person-
nel.  To accomplish this goal, the methods
utilize simplified sample preparation tech-
niques (disposable glassware, smaller scale
extractions) based on more exhaustive con-
ventional laboratory methods, such as CLP
methods.  As field analytical methodologies
and the associated QC are generated, they
are standardized, reviewed by FASP
chemists, and submitted to EPA for review
by the Analytical Operations Branch (AOB)
Field Methods Workgroup for final approval.
By the use of standardized and approved
SOGs, consistent data of known quality are
generated.

Like EPA or other standard methods, SOGs
prepared for field analytical support pro-
vide information on the approximate pre-
cision and accuracy that the methods may
provide for sample analysis.  However, FASP
methods often are tailored to meet site-
specific requirements.  This increases the
probability of obtaining useful data by
overcoming matrix problems, establishing
appropriate quantitation limits for the
project DQOs, or focusing on specific
target analytes.
QC
FASP QC is based on the needs of the FIT
program and may vary according to the
analytical method and/or specific project
needs.  There are, however, some general
guidelines provided by SOGs which are
consistently employed.

Instrument Calibration

Gas chromatographic response to target
analytes for the external standard method
of quantitation is measured by determining
calibration factors (CFs), which are the
ratio of the response (peak area or height)
to the mass injected.  An initial calibra-
tion designed to demonstrate the instru-
ment's linear response is generated for
each target analyte by analyzing a minimum
of three standard concentrations which
cover the working range of the instrument.
Using the calibration factors calculated
from the initial calibration, the percent
relative standard deviation (ZRSD) is cal-
culated for each analyte at each concentra-
tion level.  The percent relative standard
deviation generally is required to be less
than or equal to 25 percent.

The mean initial calibration factor for
each analyte is verified by the continuing
calibration during each operational period
(daily) to ensure detector stability.  Mid-
range standards are analyzed, and calibra-
tion factors are compared to the mean
initial calibration factor for each
analyte.  The relative percent difference
generally is required to be less than or
equal to 25 percent.  If the continuing
calibration criteria are not met for each
target analyte, a new initial calibration
is performed.

Final calibrations are performed at the end
of a project, or sampling effort to ensure
analytical instrument stability.  The cali-
bration factor from the final calibration
is compared to the mean initial calibration
factor for each analyte.  The relative per-
cent difference is required to be less than
or equal to 50 percent.  If the relative
percent difference meets continuing cali-
                                                  312

-------
bration criteria, the final calibration
also may be used as a continuing calibra-
tion.

Analyte Identification and Quantitation

Qualitative identification of target
analytes is based on both detector selec-
tivity  and  relative retention time as com-
pared to known standards, using the
external standard method.  Generally,
individual  peak retention time windows
should  be less than ±5 percent for packed
columns.

The  concentration of an analyte in the
sample  is calculated using the calibration
factor  for  that  analyte calculated from the
continuing  calibration.   Reported results
are  in  micrograms per kilogram (ug/kg)
vithout correction for blank results, spike
recovery, or percent moisture.

Sample  chromatograms may not match identi-
cally with  those of analytical standards.
Vhen positive  identification is question-
able, the chemist may calculate and report
a maximum possible concentration (flagged
as < the  numerical value)  which allows the
data user to determine if additional (e.g.,
CLP  RAS or  SAS)  analysis is  required or if
the  reported concentration is  below action
levels  and  project  objectives  and DQOs have
been met.

Similarly, when  sample concentration ex-
ceeds the linear  range,  the  analyst may
report  a  probable  minimum  level  (flagged  as
> the numerical value) which allows the
data user to determine if  additional (e.g.,
CLP  RAS or  SAS) analysis  is  required or if
the  reported concentration is above action
levels  and project objectives and DQOs have
been met.

Blank Analysis

A method blank is performed  with  every set
of samples extracted;  a minimum of  one
•ethod  blank per 20 samples  is performed.
The  method blank must  contain less  than the
project  quantitation limit,  the minimum
reportable value, for  each target analyte.

Matrix  Spike Analysis

Accuracy is  defined as the closeness  to

-------
analyte concentrations may be used as a
comparison of the two data sets.
FASP POLYCYCLIC AROMATIC HYDROCARBONS
(PAHs) ANALYTICAL METHODOLOGY

FASP PAH methodology provides identifica-
tion of a subset of the base/neutral acid
(BNA) compounds included on the CLP Target
Compound List (TCL).  The method provides
tentative identification of the PAH com-
pounds listed below, at estimated concen-
trations:

     Naphthalene
     Acenaphthylene
     Acenaphthene
     Fluorene
     Phenanthrene
     Anthracene
     Fluoranthene
     Pyrene
     Chrysene
     Benzo(a)anthracene
     Benzo(b)fluoran thene
     Benzo(k)fluoranthene
     Benzo(a)pyrene
     Indeno(1,2,3-cd)pyrene
     Dibenzo(a,h)anthracene
     Benzo(g,h,i)perylene

For the soil matrix, a veil homogenized 2
or 3g sample is weighed into a disposable
culture tube with a Teflon-lined cap. The
sample is extracted with 6 mLs of methylene
chloride twice by vortexing for 2 minutes,
combining the extracts.  The final extract
is dried with a small amount of sodium
sulfate and then solvent exchanged into
isooctane.

Isolation of the target analytes is accomp-
lished by a small-scale silica gel column
cleanup.  A disposable glass 4 mL giant
pipette is filled with a plug of glass
wool, silica gel, and sodium sulfate.  The
column is eluted first with methylene
chloride, then petroleum ether (10 mLs of
each).  The sample, in isooctane, is then
introduced onto the column.  After the
sample is introduced to the column, the
column is first eluted with petroleum ether
(6 mLs) in order to allow interfering
contaminants, such as hydrocarbons, to be
removed.  The PAHs are then eluted with
methylene chloride (10 mLs), and the final
volume of the extract is reduced to 1.0
mL under a stream of nitrogen.
The sample is analyzed by gas chromato-
graphy, using a J&U 0.53 mm x 15 m DB-5
fused silica megabore column and employing
flame ionization detection.  A temperature
program is utilized to optimize separation
of the analytes.  The gas chromatographic
analysis time is approximately 30 minutes.

Samples are quantitated using the external
standard method.  Standard mixes are pur-
chased from a commercial manufacturer and
diluted to appropriate concentrations for
instrument calibration.  Calibration
factors are calculated for each analyte in
the initial and continuing calibrations.
The concentration of the analyte(s) in a
sample is calculated based on the analyte
calibration factors calculated from con-
tinuing calibrations.

The quantitation limits for the FASP PAH
methodology are 1,000 ug/kg, while CLP RAS
required quantitation limits are 330 ug/kg.
As the CLP samples do not undergo silica
gel cleanup, the final matrix potentially
contains a higher degree of interference
from petroleum hydrocarbons, which are
often present along with the PAHs.  When
petroleum hydrocarbon interferences are
present, the sample often requires dilution
before an accurate analysis can occur.
This results in an elevation of the actual
contractual quantitation limits.  Samples
analyzed by FASP methodology are relatively
free of these interferences, and generally
do not require dilution.

The total time for preparation and analysis
of 10 soil samples is 490 minutes.  In a
10-hour day, the maximum capacity for a
field analytical laboratory equipped with
one gas chromatographic system is approxi-
mately 11 samples during the first day of
operation, and 20 samples each day there-
after.  This projected capacity does not
take into account any dilutions which may
be required when high target analyte levels
are present.

This method employs only disposable glass-
ware, eliminating time required for clean-
ing glassware, and minimizing  the potential
for cross contamination.   Solvent volumes
are minimal, requiring a total of only 40
mLs per sample, compared to  the CLP method
for BNAs which requires 300 mLs of solvent
per extraction.
                                                  314

-------
FASP PENTACHLOROPHENOL (PCP) ANALYTICAL
METHODOLOGY

For soil,  a well homogenized 2 or 3g sample
is weighed into a disposable culture tube
with a Teflon-lined cap.   The soil is dried
by adding  a small amount  of sodium sulfate.
The sample is then extracted with methanol
(10 mLs) by vortexing for 2 minutes.  Five
mLs of the extract is transferred into a
clean culture tube.

The extract is derivitized with a solution
of pentafluorobenzyl bromide and hexacyclo-
octadecane (18-crown-6 ether) in 2-pro-
panol.  One mL of the derivitization solu-
tion is added to the sample extract, along
vith 3 mg  of potassium carbonate.  The
culture tube is then capped,  gently shaken,
and left in a hot water bath at 80°C for 4
hours.  The culture tube  is allowed to
cool,  then the sample is  extracted with 5
•Ls of hexane by vortexing for 1 minute.
Five mLs of carbon-free water are added to
the culture tube,  and vortexed for an addi-
tional minute.   The hexane layer, which
contains the derivitized  PCP, is trans-
ferred to  a clean culture tube and dried
Vith a small amount of sodium sulfate.   The
extract is then ready for analysis.

The extract  is  analyzed by gas chromato-
graphy using a  1.0 m,  glass column packed
with 1.5%  SP-2250/1.95% SP-2401 and  employ-
ing electron capture detection.   The iso-
thermal column  oven temperature is 275°C,
and gas chromatographic analysis time is
approximately 20 minutes.

Sanples are  quantitated using the external
standard method.   Standards,  blanks,  and
appropriate  quality control samples  are
prepared with each batch  of samples  de-
rivitized.

The quantitation limit  for PCP using this
•ethodology  is  50  ug/kg.   The quantitation
Unit  for  PCP by CLP BNA  methodology is
significantly higher (1,600 ug/kg).   FASP
•ethodology  allows  for  the lower quantita-
tion limit by isolating the PCP  present in
the sample and  removing matrix interfer-
ences,  and  then  using a more  sensitive  in-
strumental  technique (GC/ECD).

The total  time  for  preparation and analysis
of 10 soil samples  for  PCP is  530 minutes.
In a 10-hour day,  the maximum  capacity  for
afield analytical  laboratory  equipped  with
one gas chromatographic system is approxi-
mately 10 samples during  the  first day  of
 operation, and 20 samples each day there-
 after.   This projected capacity does not
 take into account any dilutions which may
 be required due to high target analyte con-
 centration in the sample.

 This method, like the PAH method, employs
 only disposable glassware, and consumes
 only minimal solvent volumes (21 mLs total)
 compared to CLP solvent volumes of 300 mLs
 per sample extracted.
 CASE STUDY 1

 E  &  E was tasked to perform an LSI at an
 active wood treating facility occupying 19
 acres in Oregon.  The facility operations
 involve pressure treating wood products
 using creosote (containing PAH compounds)
 and  PCP in a petroleum oil carrier.   The
 determination of the extent of on-site sur-
 face contamination was defined as one of
 the  objectives of the LSI, requiring
 analysis of 56 on-site grid surface  soil
 samples.   Since the target analytes  were
 known,  it was determined that site-specific
 DQOs could be met by using FASP at a sub-
 stantial cost and time savings compared to
 a  full  CLP sample analysis scheme.

 Sixty-two surface soil samples were  col-
 lected  at the site for FASP analysis,  in-
 cluding six duplicate,  or colocated
 samples.   The samples were shipped to the
 FASP Seattle Base Laboratory for analysis,
 as the  project was not large enough  to
 justify mobilization.   The sample analyses
 were completed within 24 hours of receipt
 of the  last  sample shipment.

 A cost  comparison was calculated for FASP
 versus  CLP RAS analysis of the samples.
 The  total FASP costs  included the purchase
 of required  expendables,  which totaled
 approximately $4,546.00 and labor, which
 totaled approximately $13,300 for 350  hours
 of effort.   If CLP had  been utilized for
 these analyses,  the total cost would have
 been $27,308,  which accounts  for laboratory
 charges and  data validation.   This amounts
 to a savings  of  $9,461  by utilization  of
 the  FASP  program.   This comparison indi-
 cates that  full  organics  CLP  RAS would not
 be appropriate for these  samples.  Rather,
 a focused  analysis, such  as CLP SAS  or FASP
would be  more  appropriate.  For near real-
 time availability  of  sample data,  FASP
would be  the  preferred  alternative.

The confirmatory samples  were analyzed for
                                               315

-------
 BNA  compounds  by  a  CLP  laboratory  at  a  fre-
 quency  of  approximately 10  percent  (8
 samples).   Sample quantltation  limits vere
 consistently higher for the CLP data  set
 due  to  the matrix interferences from  the
 oil  present  in the  samples.   For most
 samples, quantitation limits were  elevated
 2  to 300 times above the contract-required
 quantitation levels.

 Correlation  between the FASP and CLP  data
 sets was excellent.   FASP identification of
 PAHs and PCP vas  confirmed,  and relative
 trends  in  concentrations generally  agreed.
 A  statistical  analysis  of the data  sets was
 performed  using correlation coefficients.
 FASP and CLP data sets  were compared  for
 analytes where four or  more pairs  of  data
 points  were  available (i.e.,  four  or  more
 samples sent for  confirmatory analysis  had
 results above  method quantitation  limits
 for  the analyte).   The  calculated  correla-
 tion coefficients are summarized in Table
 1.
Table  1.  CORRELATION COEFFICIENTS  FOR  FASP
AND CLP DATA:  CASE  STUDY  1
Analyte
Data
Pairs
Used
Correlat ion
Coefficient
    (r)
Phenanthrene/
  Anthracene              6      0.999
Fluoranthene              6      0.999
Pyrene                    6      0.999
Chrysene/
  Ben7o(a)anthracene      8      0.9997
Benzo(b)fluoranthene/
  Benzo(k)fluoranthene    8      0.9775
Benzo(a)pyrene            A      0.9703
Pentachlorophenol         6      0.9696
As a result of  the FASP analysis and CLP
confirmation,  the data generated by FASP
were determined  to be acceptable for use  in
determining the  on-site hazardous waste
quantity.  This  allowed data  users  to
accurately measure the relative risks
resulting from on-site contamination.
CASE STUDY 2

An LSI was performed at an  inactive  pipe-
coating facility, which had generated coal
tar, coal tar epoxies, asphalt, and  cement
mortar wastes over  the 51 acres for
approximately 30 years.  Several target
analyte groups had been identified pre-
viously, including volatile organic com-
pounds, PAHs, and polychlorinated biphenyls
(PCBs).  The project objectives required
on-site surface soil contamination to be
characterized.  An on-site grid sampling
pattern was used, resulting in collection
of 54 samples.

Previous site sampling events had identi-
fied the target analytes, allowing for FASP
analysis of the on-site surface soil
samples while maintaining the project DQOs.
The soil samples were analyzed for volatile
organic compounds, PAHs, and PCBs at the
FASP Seattle Base facility.  It was more
cost-effective to analyze the samples at
the base facility due to the variety of
analyses required and the relatively small
size of the project.

The cost of FASP analysis of the 54 samples
and four field duplicate samples vas
$20,900 ($1,900 for supplies, $19,000 for
labor) compared to CLP analysis costs which
would have totaled $57,408.  This amounted
to a total savings of $36,508 by utilizing
FASP.  All sample analyses were completed
within 7 days of the last sample shipment
date.

Six samples (approximately 10 percent of
the total number of samples) were split and
sent to a CLP laboratory for confirmatory
volatile, BNA, and pesticide/PCB analysis.
Again, matrix interferences prevented CLP
BNA analysis without elevated quantitation
limits due to the presence of oil.  FASP
methodology, involving sample cleanup for
specific analyses, removed much of the oil
interference.

Correlation between the two data sets was
excellent.  FASP identification of volatile
compounds, PAHs, and PCBs was confirmed by
CLP data, and relative trends in analyte
concentrations agreed.  Calculated correla-
tion coefficients were generated where four
or more data pairs were available.  One
split sample contained extremely high
levels of PAHs.  CLP results were signifi-
cantly and consistently higher than the
FASP results for all PAHs detected in this
sample.  It is most likely that this
phenomenon was due to the non-homogeneous
nature of the soil matrix.  Therefore, this
data pair was not included in the correla-
tion coefficient calculation.  The correla-
tion coefficients are presented in Table 2.
                                                  316

-------
 Table 2.   CORRELATION  COEFFICIENTS FOR FASP
 AND CLP DATA:  CASE  STUDY 2

                           DataCorrelation
                           Pairs  Coefficient
 Analyte	Used	(r)

 Fluoranthene              4      1.000
 Pyrene                    4      1.000
 Chrysene/
   Benzo(a)anthracene     4      1.000
 Benzo(b)fluoranthene/
   Benzo(k)fluoranthene   4      1.000
 Benzo(a)pyrene            4      1.000
 Indeno(l,2,3-cd)
   pyrene/Dibenzo(a,h)
   anthracene              4      0.999
 Benzo(g,h,i)perylene     4      0.999
 Aroclor 1254              5      0.945
 A statistical analysis  of matrix spike re-
 covery data for eight samples collected at
 both of the sites described above is pre-
 sented in Table 3.

 CONCLUSION

 Recently,  EPA has placed  a greater emphasis
 on the determination of extent of contami-
 nation during site assessments.   FASP was
 initiated under E & E's Zone 2 FIT contract
 in 1984,  and is a viable  alternative or
 supplement available to address the
 analytical demands for  determining relative
 risks at  hazardous waste  sites.   FASP
 provides  data of known  quality,  using
 standard  methodologies  and QC modified to
 meet the  project DQOs.  FASP data can be
 obtained  at a substantial cost and time
 savings when compared to  conventional CLP
 analysis,  and has been  used successfully
 for characterization of sites with known
 target analytes.
 Table 3.  AVERAGE MATRIX SPIKE RECOVERIES
 FOR SOIL SAMPLES AT HAZARDOUS WASTE SITES
 Analyte
Average
Percent
Recovery
Standard
Deviation
 Naphthalene            70.0        36.8
 Acenaphthylene         103          47.3
 Acenaphthene           94.3        36.6
 Fluorene               90.3        24.9
 Phenanthrene/
   Anthracene           93.4        38.9
 Fluoranthene           118          53.5
 Pyrene                 123          53.8
 Chrysene/Benzo(a)
   anthracene           121          34.5
 Benzo(b)
   fluoranthene/
   Benzo(k)
   fluoranthene         107          26.6
 Benzo(a)pyrene         112          24.0
 Indeno(l,2,3-cd)
   pyrene/
   Dibenzo(a,h,)
   anthracene           98.7        25.8
 Benzo(g,h,i)
   perylene             88.0        28.8
 Pentachlorophenol      122          51.4
REFERENCES

1.  Cram, S.P.,  American Environmental
Laboratory,  September 1989, pp.  19.

2.  Neptune,  D., E.P. Brantly, M.J.
Messner, D.I.  Michael, May-June  1990,
Hazardous Materials Control, Volume  3,
Number 3, pp.  19.

3.  Hafferty,  Andrew, September  1989,  "A
Cost Summary  of  Field Screening  Implementa-
tion in Region 10", Division of  Environ-
mental Chemistry, Proceedings, American
Chemical Society National Meeting, Miami
Beach, Florida.
                                             DISCUSSION
DOUG PEERV: You were talking about doing 20 samples in a ten-hour day with
30-minute run time. Does thai include your QA/QC or did you have another four
hours of work time to cover that?

LILA ACCRA-TRANSUE: We did ten samples or 20 sample analyses. So that
includes the QC samples that we need to run.

DOUG PEERY: So you're talking about your standards and your QC's within
that 20 number.

LILA ACCRA-TRANSUE: Right.
VICKI TAYLOR: How many split sample pairs did you take?

LILA ACCRA-TRANSUE: We take approximately 10%. For the first project
we'd taken eight and for the second project, six.

VICKI TAYLOR: So you were basically presenting a correlation coefficient for
all the split samples that you took?

LILA ACCRA-TRANSUE: Right. All of the comparable data pairs are reported
where they were hits in both samples.
                                                     317

-------
                           Thermal Desorption Gas Chromatography-Mass Spectrometry
                             Field Methods for the Detection of Organic Compounds
    A. Robbat, Jr., T-Y Liu, B. Abraham, and C-J Liu,
    Tufts University, Chemistry Department, Trace
    Analytical Measurement Laboratory,
    Medford, MA 02155
INTRODUCTION

The overwhelming amount of information required to characterize
purported hazardous waste sites, as well as to support Superfund
site cleanup and closure activities, have catalyzed the development
of field instrumentation capable of  providing site managers with
immediate access to chemical and physical data. The demand for
field "practical" methods and instrumentation has been recognized
by the U.S. Environmental Protection Agency (1, 2).

Faster data turnaround times and  ease of operation have been the
primary motivation for selecting field gas  chromatographic (GC)
methods  of analysis.  Despite recent advancements in field  GC
instrumentation, typical applications focus on the detection of EPA
listed volatile organic compounds (VOCs) in water, air, or  soil
gas. The primary limitation of commonly employed field GC's is
the non-definitive signal response  of  the detectors (including
photoionization, flame  ionization,  thermal  conductivity,  and
electron capture) which are incapable of providing unambiguous
identification of the wide variety  of organic compounds that may
be present in a highly contaminated sample.  Generally, ten to
twenty percent of the  samples analyzed  on-site are "split"  for
confirmation by GC with  mass spectrometric  (MS) detection.
Since  most commercially  available mass spectrometers have
traditionally been housed and operated in a clean air, temperature
controlled room and the notion that economies  of scale require
highly trained MS operators to be based in multi-MS laboratories,
misapprehensions have arisen as to whether MS's can be operated
successfully (and profitably) in the field.

The limited availability of field GC-MS's is not a function of MS
operating requirements, but more, the perception that significant
sample cleanup and QA/QC procedures will be required to obtain
useful data as well as  the  apparent reluctance of instrument
manufacturers to enter the field marketplace. Until recently, these
misconceptions have perpetuated the myth that GC-MS's belong
solely in the laboratory.
Over  the last  several years,  we have  discussed field GC-MS
applications   utilizing   Bruker  Instruments'   mobile   mass
spectrometer (2-6).  The MS, initially designed for NATO as a
chemical warfare detector, was manufactured from the outset as a
field instrument.  In our studies, the MS was transported from
site-to-site in a mid-sized truck and was battery operated for - 8
to 10-hr at ambient conditions. For example, samples have been
analyzed with outdoor conditions, where; temperatures have been
between 10 "F and 90 "F, rain, snow, and high  humidity.  Gas
cylinders were not  necessary for GC operation since charcoal
filtered ambient air served as the carrier gas.

Simple  field  methods have been developed based on analyte
introduction by thermal desorption  CTD) followed by fast GC
separation  and  MS  detection.  Screening level and  more
quantitative TDGC-MS methods  have been submitted to EPA's
EMSL-Las Vegas for VOCs in water, soil/sediment, soil gas, air
and polychlorinated biphenyls (PCBs)  and polycyclic aromatic
hydrocarbons  (PAHs)  in  soil/sediment for  inclusion into the
compendium of field methods that  will be published  by EPA's
Analytical Operations Branch.  The methods include a menu of
QA/QC procedures  whose implementation depends upon a given
study's objectives.  The goal is to provide a practical GC-MS tool
that can deliver the quality of data  required for the study with
minimal sample cleanup.   Presented in this paper are  typical
examples of data quality and a comparison of field and laboratory
results  one  can  expect  from  both the  screening  and  more
quantitative field TDGC-MS methods  for PCBs, PAHs, and
pesticides.
EXPERIMENTAL SECTION

A mobile mass spectrometer (Bruker Instruments, Billerica, MA)
was  used  in these  studies.  The TDGC-MS was  powered by
battery or electrical supply from the site. The MS was transported
to Superfund sites in Westborough (Hocomonco Pond; PAHs) and
                                                            319

-------
North Dartmouth, MA (Resolve; PCBs) in a Chevrolet Blazer, In
addition to the instrument's internal data collection and monitoring
system, the MS was equipped with an external data system and
thermal desorption  sampling probe.   Sample introduction  was
made by thermally  desorbing (TD) the analyte  directly  from
soil/sediment or from an organic extract through the TD sampling
probe's (SP) short 3.5 m fused silica capillary column. For direct
TD soil/sediment experiments, 0.5 g of soil was placed on an
aluminum foil covered petri dish.   An internal  standard  was
injected into the soil before the measurement  was made.  In
contrast, the more  quantitative  measurements required  several
additional steps:  1) 0.5 g of soil  was  weighed and  extracted  with
2  ml of solvent; 2) prior to extraction, a known quantity of
surrogate (or target) compound(s) was added to the soil (or  field
blank) to  determine extraction efficiencies  (note:  this step was
required since a  single 2 ml extraction yielded analyte recoveries
of less than  100%;  3)  co-inject known aliquots of extract and
internal standard onto aluminum foil  covered  petri dish; 4)
thermally desorb analyte.    Shown  below are the TDGC-MS
operating and PCB, PAH, and pesticide  experimental conditions:

Operating Conditions

Mass Spectrometer       Bruker Instruments (Billerica, MA)
electron energy          70 volts (nominal)
mass range              45 to 400 amu
scan time                2 sec
MS tune                autocalibrate (HjO,,,; FC-77); 18,  69,
                             119, 169,331 amu
mass resolution           set to  unity; ca. 10% valley
                             definition
ion detection             17 stage Cu-Be dynode  electron
                             multiplier with self-scaling
                             integration amplifier (108
                             linearity)

Sampling Probe  Head     260 °C
 GC Column
 dimensions
DBS (J & W Scientific, Folsom, CA)
3.5m x 0.32mm i.d.; 0.25/x film
     thickness
ambient air purified through carbon
     filters
3 to 4 ml/min
 earner gas

 flow rate

              PCBs         PAHs             Pesticides

 initial temp   140°C, 30 sec  70°C, 40 sec      120 °C

 temp prog    120°C/min    35°C/min         17°C/min

 final temp    200°C, 90 sec  233°C, 80 sec     233°C

 Internal      d,«-pyrene     d,-naphthalene     d,0-
 Standards                  or d,0-pyrene      phenanthrene
                                       Data were acquired by using the internal monitor's selected ion
                                       monitoring program.  The data system reported the total ion
                                       current as a logarithmic  value.  The antilog value is used in
                                       conjunction with MS response factors and  analyte recoveries to
                                       calculate concentrations in the sample.  Standards were purchased
                                       commercially from  the  following  companies:  PCBs (Ultra
                                       Scientific, Hope,  RI); PAHs (Supelco,  Inc., Bellefonte,  PA);
                                       Pesticides (Chem Service, West Chester, PA); internal standards
                                       (Cambridge Isotope Laboratories, Woburn, MA).  All standards
                                       and soil recovery experiments were  prepared with high purity
                                       solvents (> 96 %) as received.
                  RESULTS and DISCUSSION

                  The objective of  this study  was to develop fast TDGC-MS
                  methods  (<  20 min/sample including sample cleanup).  Two
                  methods  were developed.   Analyte introduction for quantitative
                  measurements were  made  by co-injecting  organic  extracts  (or
                  standard  solutions) of PCBs,  PAHs, or pesticides  and  internal
                  standard(s) onto an  aluminum covered  petri dish followed by
                  TDGC-MS and  for  screening measurements by direct  thermal
                  desorption from soil/sediment.

                  The surface monitor program mode was employed in this study.
                  Target compounds (maximum number twelve) were detected by
                  selected ion monitoring (SIM) MS. The (logarithm) ion current
                  was recorded and displayed visually on the system's monitor.
                  Found in Figure 1 are typical PCB and pesticide outputs. Three
                  fragment  ions representative  of each  compound(s)  and  an
                  impossible ion  (see below for rationale) were  selected  for
                  detection. For example, in cell A the target ions and their relative
                  intensities for the three monochlorinated PCBs were  188 (100%),
                  190 (33.5%), 152  (31.1%), and 189 (0%).  Similarly, cells B-H
                  in Figure la illustrate the SIM four ion current responses for
                  chlorination levels 2 - 8, respectively; cell I, d,0-pyrene (internal
                  standard); cells J - K, PAH surrogates; and cell L, hydrocarbon
                  signals indicative of matrix complexity.  Detection was made, and
                  printed on screen,  when the signals from the four ions relative to
                  each other agreed  to within preset criteria over a predetermined
                  retention time window.  In this mode, SIM response  may be
                  considered analogous to selective GC detection. Note above, that
                  the last fragment ion for the monochlorinated PCBs had a relative
                  intensity of 0%.  Inclusion of an impossible ion served to provide
                  selective detection.  For example,  an increase in fragment 1 ion
                  current relative to fragments  2-4 within the target compound's
                  retention window precluded compound identification.  Thus, the
                  mathematical  algorithm assisted in screening out interferants
                  present in the sample.
 solvent      C»H,4
 extraction
    CH,C12
C6HM
                                                              320

-------
              EURFACE MONITOR
              M PC8/DIOXINS/CL
               8

               7-

               6-

               5-

               4-

               3-

               2-
  Die-PHEJWTHR
  D-12-CHRYSENE
  HYDROCHRBONS
  CL3-BIPHET1YL
  CL4-BIPHEWL
  CL5-BIPHEWL
  CLS-BIPHEJ1YL
B CL2-BIPHO1YL
G CL7-BIPHENYL
F 4.8
F 2.9
F 5.3
F 4.9
F 5.0
F 3.3
F 3.4
                  fiBCDEF6HI3KL
                  Figure  la
5SFKZ KiiTO?. 28.18.315:44
V FESTICIuE
ft fl,3/E/0-BHD F 4.S
5 Die-Ft£?MHF£ F £.7
S-j CHEPTACllOR F 4.3 153

7-

6-

3-
4-
3-
2-
>H
^'-DOD F3.3 2
D ftLDRI/1 F 4.5
F DIEffi!?! F 3.8
EffiT-EFCKIDE F 4.4 [p
B 4,4 ' -DDE




1 VTT
A6!
.4 ' -rjDT

I
1
U F5.2
F5.4
)

7
DEFSHim
184
 H
The fast GC  linear temperature programs and  MS  detection
provided sufficient separation to identify compound(s) as shown in
Table 1.  Figure 2 is a typical instrument print out for the amount
(4-ion total current count, in log values, left vertical axis) vs. time
response curves  (horizontal  axis) for four of the chlorinated
pesticides shown in  Figure Ib. In addition to the compound and
amount detected, other information visible on the display included
"real-time" monitoring of: logarithm of ion current, left vertical
axis;  MS  vacuum  pressure,  right vertical axis;  and column
temperature, above  right vertical  axis.
ajon toinot 38.u.a IBS
V PESTICIDE fflOUHT VS Tfft
I ft^E/HHC F 4.8
8-
7-
S"
4-
3-
2-
1
IS?
2
5
'V^'U !
# 1
tSSs
SUJFflK tttHITOR 3fl.1B.58 18:45
V PETICIIE5 fraW VS THE
t i£FT.EPIKir£ F4.4
1-
t
3-
4'
3-
1
172
2
i
H
i
WWHWMH ^V-V ^WM) 7
183s
aim nonna 38.ia.aa u:c
D MEW F 4.C
7-
£•
t
4-
3-
2-
171
2
!
f
,V*, .v"' A :
•nren KwrV n •?
imii
tffil
ssFsumnai 2a.ta.3B u;«
v PEsriciio nmin vs ire
S4/4'-0£ F5.2
J-
7-
£•
5-
3-
2-
1
173
2
S
1
i
':•.;•, It
Ssi'V ?
188 s
                 Figure  Ib
igure 1. Typical Field TDGC-MS SIM response of a standard
 ution containing PCBs (la) and chlorinated pesticides (Ib).
                                           Figure 2.  Amount  versus time  curvfr for several  chlorinated
                                           pesticides shown in Figure 1.
                                                           321

-------
 TDGC-MS experiments were performed between the concentration
 range of 40 and 4000 ng/compound.  Repetitive measurements at
 each concentration yielded differences in the log value of + 0.13
 producing ion current differences of less than 30%. Table 1 lists
 typical  response factors  (RF)  and  percent  relative  standard
 deviations (%RSD) calculated for PCBs, PAHs,  and pesticides
 thermally desorbed from an organic extract.  Plots of signal versus
 concentration were linear (r=  0.999) with the %RSD for the
 average  RF  less than 30%,  meeting initial and  continuing
 calibration criteria in the Contract Laboratory Program.  Table 2
 lists representative RF and RSDs for PCBs  and PAHs thermally
 desorbed  directly from soil.   Despite somewhat  larger percent
 RSDs for some PAHs, measurement precision at  this level will
 only be  critical at  site cleanup "action" levels.  It should be
 pointed out that thermal desorption extraction efficiencies differ
 greatly  for some PAHs  (see Table 3 for  minimum detectable
 quantity.   Note:  RF in Table 2 calculated  over linear  range as
 shown  in Table  3).    Minimum  detection  levels for  most
 compounds were  ~ 1 ppm for soil/solvent extraction and slightly
 higher for direct soil thermal desorption.   Because TDGC-MS
 experiments can be performed in  5  to 20 min depending on the
 method employed (with known data quality), many more  analyses
 can be performed than currently practiced for site characterization,
 stockpiling, and  worker/community protection activities.  The
 frequency for performing  continuing calibration checks  may be
 determined (on-site) by following surrogate compound RF values
 (see below).

 Research has shown that compound recoveries vary with soil-type.
 For example, PCB/hexane (0.5 g/2 ml hexane, 2 min) extraction
 recoveries were 69 + 5% for 50 ppm backyard (organic) soil, 80
 ± 2 %, for 25 ppm sandy material from the Resolve Superfund
 site in North Dartmouth,  MA, and 73  ±  5 %  for an ERA, 35
 ppm, soil.  Therefore, appropriate surrogate  compound(s) and/or
 target standards must be added to samples as the soil-type varies.
 Such  experiments  can  be  used  to  determine  instrument
 performance as well.
Tables 4-7  illustrate typical examples of data quality  one can
expect from the field  TDGC-MS methods.  Split samples were
collected by EPA's Region 1 oversight contractor and analyzed in
the field (Tufts) and lab (Lockheed ESC, Las Vegas, NV). Table
4 compares field  and lab  GC-MS measurements for total PCB
present in several samples obtained from the Resolve site while
Table 5 delineates chlorination level comparisons for two of the
samples.  The field and lab results are in excellent agreement.

 Shown in Tables 6 and 7 are  field and lab comparisons  for four
 PAH samples from the Hocomonco Pond (Creosote contaminated
 Superfund) site. Note that the samples in Table  6 and the sample
 labeled HP-SB5  in Table  7 were performed by SIM using the
 system's  internal  monitor  as described above.  In contrast, the
 sample labeled pond (Table 7) was analyzed  by  total ion  current,
 selected  ion  monitoring  extraction.   The advantage  of  this
detection method  was that full mass spectral fragmentation data
 and compound library matching was applied.  On the other hand,
 the disadvantage was that ion current from matrix components may
 add to the SIM signal resulting in higher concentrations than  what
might actually be present.  This, however, is no different than
what can occur using traditional CLP, MS methods. Field and lab
comparisons  for  PAH  samples  also  appear  to  be  in good
agreement.

Additional data will be presented describing further  application of
the field TDGC-MS methods.    Illustrations  will  be  given
documenting cost effectiveness.  Results will show  that GC-MSs
can be operated  in the field,  provide rapid access of data, and
allow project managers to make decisions on-site.
ACKNOWLEDGEMENTS

Partial financial support for this project was provided by the U.S.
Environmental  Protection  Agency,  EMSL-LV;  New  Jersey
Institute of Technology's Northeast Hazardous Substance Research
Center;  and  Tufts  University's  Center  for  Environmental
Management.   The  authors  wish  to thank EPA's  Region  1
Hazardous Waste Division for providing access to Superfund sites
and samples and to the oversight contractors for their cooperation.

REFERENCES

1) Williams, L.R.,  Editorial Article, American Environmental
Laboratory, October, 1990 (see additional articles by EPA).

2)  U.S.  Environmental Protection Agency, Sixth Annual Waste
Testing and Quality  Assurance Symposium, July  16-20,  1990,
Washington, DC; Field Analytical Methods Workgroup sponsored
by Analytical Operations Branch. See Proceedings, Robbat,  A.,
Xyrafas,  G.,  Abraham,  A., "A Fast Field  Method for The
Identification of Organics in Soil", 1-350..

3)  "Method Evaluation for Field Analysis of PCBs and VOCs
Using a Field Deployable GC-MS", Xyrafas, G., Ph.D. Thesis,
Tufts University, Chemistry Dept., Medford, MA.  O2155.

3)  "A fieldable GC-MS for  the Detection and  Quantitation of
Hazardous Compounds:   Analytical  Chemistry  in the Field?"
Robbat, A., Jr., Xyrafas, G., 198th American Chemical Society
National Meeting, 411, 29(2), 1989, Miami Beach, Florida.

4)  "Evaluation of a  Field-Based, Mobile, Gas Chromatograph-
Mass Spectrometer for the Identification and Quantification of
Volatile Organic  Compounds on EPA's Hazardous  Substance
List", Robbat, A., Jr., Xyrafas, G., In Proceedings of the First
International  Symposium on Field  Screening  Methods  for
Hazardous Waste Site Investigations,  October 11-13,  1988,  Las
Vegas, Nevada, Pg. 343.

5)  "On-Site Soil Gas Analysis of Gasoline Components Using a
Field-Designed Gas Chromatograph-Mass Spectrometer", Robbat,
A., Jr., Xyrafas, G., In Proceedings of the First International
Symposium on Field Screening Methods for Hazardous Waste Site
Investigations, October 11-13, 1988, Las Vegas, Nevada, Pg. 481.
                                                             322

-------
Table 1. Thermal Desorption Field GC-MS Response Factors and
Percent Relative Standard Deviations - from Extract (Quantitative
Method)

                 Polvchlorinated Biphenvls
Table 2. Thermal Desorption Field GC-MS Response Factors and
Percent Relative Standard Deviations - Direct from Soil
                 Polvchlorinated Biphenvls

Chlorination Level
CM
Cl-2
Cl-3
Cl-4
Cl-5
Cl-6
Cl-7
Cl-8


Ave RF(n=5)
0.47
0.26
0.27
0.16
0.15
0.10
0.06
0.03


%RSD
20
17
17
15
12
15
17
10

Chlorination Level
Cl-1
Cl-2
Cl-3
Cl-4
Cl-5
Cl-6
Cl-7
Cl-8

Polvcvclic Aromatic
Ave RF(n=5)
13.44
3.75
3.91
2.55
2.02
1.61
1.04
0.36

Hydrocarbons
Polvcvclic Aromatic Hydrocarbons

naphthalene
acenaphthylene
acenaphthene
fluorene
phenanthrene/anthracene
fluoranthene/pyrene
chrysene/benz(a)anthracene
Chlorinated
BHCs
Heptachlor
Aldrin
Heptachlorepoxide
Dieldrin
4,4'-DDE
4,4'-DDD
4,4'-DDT

1.37
8.63
0.82
0.58
4.59
9.52
0.90
Pesticides
0.10
0.02
0.07
0.03
0.02
0.32
0.16
0.12

123
123
24J8
123
123
95
133

96
235
163
109
25.4
163
186
19.7
naphthalene
acenaphthylene
acenaphthene
fluorene
phenanthrene/anthracene
fluoranthene/pyrene
chrysene/benz(a)anthracene










2.39
1.21
0.33
0.16
0.25
0.06
0.003










                                                                                                                RSD
                                                                                                                 19
                                                                                                                 25
                                                                                                                 23
                                                                                                                 16
                                                                                                                 16
                                                                                                                 16
                                                                                                                 23
                                                                                                                 19
                                                                                                                  75
                                                                                                                 356
                                                                                                                 52.1
                                                                                                                 165
                                                                                                                 213
                                                                                                                 310
                                                                                                                 229
                                                         323

-------
 Table 3. PAH Dynamic Range Directly Desorbed from (0.5 g)
 Soil Matrix.
               Concentration         Signal        Linearity
 Compound(s')       Cne)             (11=5)           (rt
Table 4.  Comparison of Field and Lab GC-MS Results for Total
PCBs in Samples  from the  Resolve  Superfund  Site, North
Dartmouth,  MA
Naphthalene







Acenaphthylene





Acenaphthene





Fluorene







Phenanthrene &
Anthracene






Fluoranthene &
Pyrene


4000
2000
1600
800
120
80
40

4000
2000
1600
800
80
40
4000
2000
1600
800
80
40
4000
2000
1600
800
120
80
40

8000
4000
3200
1600
240
160
80

8000
4000
3200
1600
240
160
510084 + 22.9% 0.999
255648 ± 22.9%
210541 ± 31.2%
110357 ± 12.9%
28371 ± 13.2%
5312 ± 23.4%
1995'± 22.4%

255648 ± 22.9% 0.999
129245 ± 26.1%
94858 ± 10.8%
51454 ± 26.1%
3575 ± 13.2%
794'+ 17.2%
94858 ± 10.8%' 0.999
23397 ± 12.7%
20307 ± 22.9%
9936 ± 34.5%
740 ± 12.7%
251"+ 16.8%
52714 ± 11.0% 0.999
24197 ± 32.8%
18585 ± 12.7%
8971 ± 13.2%
371 ± 12.8%
794a± 13.4%
251"+ 15.4%

42578 ± 35.4% 0.999
21245 + 12.2%
16271 ± 26.1%
8084 ± 22.9%
877 + 24.3%
3981'+ 18.6%
316" ± 20.2%

17498 ± 24.3% 0.999
8629 ± 13.8%
6854 + 13.8%
3221 ±40.1%
371 ± 12.8%
195'+ 15.3%
Quantitative Screening Level
TDGC-MS TDGC-MS Lab GC-MS
EPA ID# (ppm) {ppm) (ppm)

TUF-RS-SO-A26-2-4 368.3 309.4 298.6
TUF-RS-SO-AI-5-2 274.6 213.6 260.0
TUF-RS-SO-A42-6-8 23.1 7.2 15.9
TUF-RS-SO-A37-0-2 9.1 3.2 1.3
TUF-RS-SO-AI4-0-2 7.6 1.6 5.0
TUF-RS-SO-A5A-2-4 1.7 1.7 0.4
TUF-RS-SO-NH24-2-4 1.7
TUF-RS-SO-A14-6-8 1.3 - 3.0
TUF-RS-SO-A7-4-6 ND ND ND

ND, compound not detected
Sample comparison on an as collected basis (i.e., soils were not
dried)
Lab GC-MS performed by Lockheed ESC, Las Vegas, NV
Field GC-MS performed by Tufts University
Sample collected by EPA's Region 1 oversight contractor
-, Samples were not analyzed


Table 5. Comparison of Field and Lab GC-MS by Chlorination
Level (ppm), Resolve Superfund site, North Dartmouth, MA.


ID Sample » TUF-RS-SO-A 1 5-2TUF-RS-SO-A42-6-8
Cl-level Field Lab Field Lab
TDGC-MS GC-MS TDGC-MS GC-MS

Cl-1 12.5 ND 0.5 ND
Cl-2 7.6 10.8 1.5 1.0
Cl-3 60.3 56.5 4.5 4 1
Cl-4 121.4 122.8 5.1 5.3
Cl-5 59.5 53.6 6.3 4.3
Cl-6 20.9 15.9 3.0 1 2
Cl-7 1.7 0.4 0.3 ND
Cl-8 0.7 ND 1.9 ND
total PCS 274.6 260.0 23.1 15.9
These values were not included in the dynamic range.
                                                            ND, compound not detected
                                                            Sample comparison on an as collected basis (i.e., soils were not
                                                            dried)
                                                            Lab GC-MS performed by Lockheed ESC, Las Vegas, NV
                                                            Field  GC-MS  (Quantitative  Method)  performed   by  Tufts
                                                            University
                                                            Sample collected by EPA's Region 1 oversight contractor
                                                         324

-------
 Table 6.  Comparison of Field and Lab GC-MS Results for PAH's
 From the Hocomonco Pond Superfund Site in Westborough, MA,
 in ppm.
 Table 7. Comparison of Field and Lab GC-MS Results for PAH's
 From the Hocomonco Pond Superfund Site in Westborough, MA,
 in ppm.
DSTB22(0'-2') DSTB22(2'-4')
Lab Field1 Lab Field1
Naphthalene 0.1 ND 2.2 ND
Acenaphthylene 0.1 0.1 ND 0.7
Acenaphthene 1.4 0.1 6.0 0.2
Fluorene 2.9 1.5 16.3 3.0
Anthracene & 8.3 40.3 81.8 72,7
Phenanthrene
Pyrene& 11.8 10.6 112.2 60.5
Fluoranthene
Chrysene& 6.0 6.2 37.2 37.2
Benz(a)anthracene
Benz(b)fluoranthene, 3.2 23.8 17.7 22.3
Benz(k)fluoranthene, &
Benz(a)pyrene
ND, compound not detected
Sample comparison on an as collected basis (i.e., soils were not
dried)
Lab GC-MS performed by Lockheed ESC, Las Vegas, NV
Field GC-MS performed by Tufts University (Thermal Desorption
of Methylene Chloride Extract)
'Data collected by Selected Ion Monitoring (Internal Data System)
POND HP-SB5
Lab Field2 Lab Field1
Naphthalene 1.3 1.9 54.8 32.0
Acenaphthylene 1.4 ND ND ND
Acenaphthene 0.7 ND ND 1.2
Fluorene 2.5 ND ND 0.8
Anthracene & 16.7 10.4 ND ND
Phenanthrene
Pyrene& 30.7 43.6 ND ND
Fluoranthene
Chrysene& 37.2 55.2 ND ND
Benz(a)anthracene
ND, compound not detected
Sample comparison on an as collected basis (i.e., soils were not
dried)
Lab GC-MS performed by Lockheed ESC Las Vegas NV
Field GC-MS performed by Tufts University (Thermal Desorption
of Methylene Chloride Extract)
'Data collected by Selected Ion Monitoring (Internal Data System)
'Data collected as Total Ion Current Chromatogram and quantified
by Selected Ion Monitoring Extraction (External Data System)
                                                       DISCUSSION
ALAN CROCKETT: I found your presentation and the results extremely
informative and the accuracy or the precision you were getting was fantastic. Did
you say that you were using two tenths of a gram sample or a two-milligram
sample?

AL ROBBAT: A half a gram.

ALAN CROCKETT: That's impressive just being able to sub-sample a jar of
soil as repetitively as you've been able to. What's your preparation procedure for
homogenization of soil that comes into your facility?

ALROBBAT: These samples were all homogenized by EPA Region I. We didn't
do anything more after we got them, except stir them up a little bit.

ALAN CROCKETT: How did they  homogenize when you get them so
homogeneous?
ALROBBAT: Basically they screen them and then they collected them in a large
jar and simply just rotated them. We did not do any of the real homogenization
of the sample.

ALAN CROCKETT: What's the cost of the instrumentation by the way?

AL ROBBAT: I think it's about $ 180,000 but your best bet is to ask Bruckner
Instruments.

JON GABRY: What are your power requirements for the unit?

ALROBBAT: We use six 24-volt batteries. Six 24 volt batteries out in the site.
We also can power-up at the site if there's electrical supply. So again, if you're
interested in those types of details. 1  would suggest you visit the Bruckner
Instruments booth.
                                                                325

-------
        RAPID DETERMINATION OF SEMIVOLATILE POLLUTANTS BY
           THERMAL EXTRACTION/GAS  CHROMATOGRAPHY/MASS
                                SPECTROMETRY


             T. Junk, V. Shirley,  C. B. Henry, T. R.  Irvin,  E. B. Overton
                     LSU  Institute for  Environmental  Studies
                     42  Atkinson  Hall,  Baton  Rouge,  LA   70803

                      J. E. Zumberge, C.  Sutton, R.  D.  Worden
           Ruska Laboratories,  Inc.,  3601 Dunvale,  Houston,  TX   77063
 Abstract

 There  is  considerable interest  in rapid,
 field   deployable  analytical  systems!
 Conventional  gas  chromatography/mass
 spectrometry   analytical   techniques
 provide  sensitivity and  specificity  but
 require     cumbersome    solvent
 extractions.   Thermal  extraction  offers  a
 fast  and  safe  alternative  to  classical
 extraction procedures  for a  wide range
 of  semivolatile  pollutants.     In   this
 technique  samples   are  loaded  into
 porous   quartz   crucibles   with  no
 preparation   other   than   weighing
 required prior to  analysis.  Analytes are
 volatized  into  the helium  carrier  gas
 flow  at  controlled  preprogrammable
 temperature   profiles  and  subsequently
 cyrocondensed onto a conventional  gas
 chromatographic  column.   The  method
 was demonstrated by  analyzing  for  a
 representative   group   of   organic
 pollutants  covering  a  wide  range of
 polarity/volatility  contained  in  natural
 soil  matrices at concentrations as low as
 0.5  ppm  using  a  Pyran  Thermal
 Chromatograph.      Analyses    were
 independently  performed   by   three
 different   laboratories   (Institute  for
Environmental  Studies, Louisiana  State
University;   Engineering   Toxicology,
Texas   A   &   M   University,  Ruska
Laboratories,  Inc.)   using  an   on-line
Finnigan    Ion   Trap   Detector   for
identification and  quantification.

Average  correlation  coefficients  for
calibration  curves  ranged  from 0.938 to
0.997 for compounds less volatile than
naphthalene.    Naphthalene   and  more
volatile    compounds    experienced
variable  losses  during  open-air  sample
loading.    Dialkylphthalates  underwent
partial   decomposition   during   the
thermal  extraction process.    Recoveries
varied  depending on  soil types  as well
as on the  physical and  chemical  nature
of analytes,  with  generally the  highest
thermal  extraction yields for river silt
and  the lowest yields for clay.   Typical
recoveries   were   10   to   30%   for
polynuclear   aromatic hydrocarbons,  60
to  70%  for  hexachlorobenzene,  and
nearly   100%  for  chloronaphthalenes.
However,  the  pesticide   aldrin   showed
recoveries of at most 19%.   A  majority
of the  analytical  results  are  within  an
accepted range  for quantitative  analysis.
The  Pyran  system can be adapted  to be
                                           327

-------
deployable.    With  sample  turn-around
times  of  typically  30-60  minutes  this
instrument   should   greatly  facilitate
remediation  and   hazardous   waste
cleanup efforts.

Introduction

Transportation,    field    deployable
analytical    systems   that   provide
unambiguous  data  on  the amount  of
semivolatile  organic pollutants can  aid
in  the rapid  assessment and cleanup  of
hazardous    waste    sites.       By
complementing   the   Environmental
Protection  Agency's  Control Laboratory
Program   through   interactive   field
management,  the efficient  remediation
of   hazardous  wastes   sites  can   be
accomplished  (R.  J.   Bath,  personal
communication).

Mass   spectometry   provides    the
specificity  and  sensitivity  necessary for
the  identification and  quantification  of
most    environmental    pollutants.
However,  to  introduce  analytes into the
mass  spectrometer,  the  pollutants  must
first   be   extracted  from  the   soils.
Normally, organic solvents are used for
this purpose, a  cumbersome  and  labor
intensive  approach.   Thermal  extraction,
in  contrast,  desorbs  analytes  from  their
matrices  (soils)   by controlled  heating
under  conditions  which avoids analyte
decomposition (as opposed  to  pyrolysis).
In  this report, we describe results  from
a   study   aimed  at   verifying   the
suitability  of  thermal  extraction  as
alternative  to  conventional  extraction
for  a  representative cross  section  of
semivolatile  organic pollutants.    We
establish  the  factors controlling analyte
recoveries   from   different  types  of
matrices.        Three    laboratories
participated   in   this   study,   using
identical  instrumentation   (Institute for
Environmental Studies,  Louisiana  State
University,  Texas  A  &  M  University,
College Station, and Ruska Laboratories,
Inc., Houston).

Instrumentation

A   Level  2   Thermal  Chromatograph
(Ruska  Laboratories,  Houston,  Texas)
was  interfaced  with a  Finnigan Ion Trap
Detector.    Samples  were  heated  in  a
quartz   chamber   using   a   linear
temperature  program  and  semivolatile
analytes purged  with helium gas.  These
analytes   were  cyrocondensed   onto  a
fused  silica  chromatographic   column
(Hewlett Packard HP-5, 12 m  x  0.2 mm)
cooled  with   liquid  carbon  dioxide,
separated,  and  identified  by  mass
spectroscopy.     Thermal   extraction
efficiencies  for specific  toxicants  were
also   monitored  by thermal  extraction
under   identical   conditions   in   an
identical  quartz  chamber  coupled  to  a
flame  ionization  detector  (Level   1
Thermal Extractor).  Schematic diagrams
of these instruments are shown in Fig. 1.

Other  experimental   parameters  were
chosen as  follows:  30 ml/min He carrier
flow  during thermal  extraction  phase,
30:1   split  ration   between   thermal
extraction  chamber and  GC  column,  1
ml/min carrier flow through GC column.

Standards   Preparation

Test  soils  were  prepared  by  adding
stock  solutions  of  20  semivolatile
organic pollutants covering a  wide range
of polarity/volatility   to  three  different
organic-lean   natural   soil   matrices:
kaolin  clay,  sandy  river  silt,   and
subsurface   terrestrial   soil    from
Livingston Parish,  Louisiana  containing
30% clay, 66%  silt, and  4% sand with  a
total  organic content  of  0.11%.   stock
solutions  of the 20 standards  (see Table
1)  were  prepared  by  weighing  pure
                                            328

-------
compound   standards   (primarily  from
Aldrich  Chemical Co.) and diluting 4000
Hg/ml  stock  standard  (PP-HC8,  Chem
Service,  Inc.;   lot   #25-121B)  with
dichloromethane   to  20   ng/|il   per
component.

The  three  soils  were  crushed  using  a
mortar and  pestle and  sieved through  a
850  |im sieve.   The  sieved  soils were
slurried  for 1 hour with  the  appropriate
amount  of  stock   standards  (pure
dichloromethane   for   controls),   the
solvent   then   removed    at  room
temperature  by  evaporation   under  a
fume  hood  to produce  two sets of  test
soils with concentrations of  50  ppm  and
0.5 ppm, respectively, per analyte.   The
soil  standards  were  then  sent to  the
three  participating   laboratories   for
independent  analyses  in    well-filled
teflon lined  screw  cap vials  and  stored
at 6°  to avoid analyte losses.

Methods

Soil  samples were  weighed   into  the
porous   fused  silica   crucibles,  while
standard stock solutions (20  ng/^il) were
injected onto the  porous fused  silica  lids
of the  sample crucibles using  a  10 u.1
syringe  just prior  to  loading  into  the
thermal   extraction   chamber.     All
samples  were heated from  30° to 260°
at 30°/min  and  held  isothermally  at
260° for 10 min before cooling to 30°.
The  "trap"  and "splitter" regions  (see  Fig.
1)  were held isothermally  at  300°  and
310°,   respectively;  interface   and
transfer  line temperatures  to  the  MS
were  held  between  280°  and  290°.   The
column  was  held at 5° until the thermal
extraction  process   was  complete,  the
temperature   programmed  to  285° at
10%nin and kept  isothermal  for 5 min.
Total cycle time  was  59  min.  The  ion
trap  detector  was  scanned from  47  to
440  amu  at  1  scan/sec,  peak threshold
was  set at 2,  and a mass defect of  100
mmu/100  amu  was  used.   Full  scan
mass  spectra   of   eluting  compounds
standards  were  verified  using the  NBS
mass  spectra  library.    Areas   and
retention  times  of  characteristic   ion
masses were recorded after each  run for
each of the 20 compounds and  internal
standards.    Calibration  curves for  each
of  the 20  compounds   in  the  stock
solution  (20  ng/jil)  were obtained  by
injecting 2, 5,  10,  15  and  20 ill onto the
crucible lids  (corresponding to 40,  100,
200,   300,   and   400   ng/component,
respectively).       Ten   u 1    (200
ng/component)   of   the  deuterated
internal standards  (Table  2)  were  also
added  to  the  lid  prior  to each  of the
above  five runs.   This experiment  was
done in triplicate  at  Ruska Laboratories,
using  a Finnigan  Ion  Trap Detector for
two  runs  as  described   above   and  a
Hewlett Packard Mass Selective Detector
(MSD) once for  comparison. Just prior to
each run  of  the standard  soils (10.0 to
13.8 mg  for the 50  ppm standards  and
approx.   100   mg  for  the   0.5   ppm
standards),  10  ul  (200   ng/component)
of   the  deuterated  internal   standards
were  injected  into  the   soil/sediment.
Response  factors   (RF)  and  percent
relative   standard  deviations  (%RSD)
were  calculated  for  each  compound
based  on EPA's  "Test  Methods  for
Evaluating   Solid   Waste,   Physical,
Chemical  Methods",  SW-846,  Third
Edition,   Method  8270   (GC-MS   for
semivolatile  organics,  capillary  column
technique).   RF values  are  based  upon
the results of  the on-lid injections of the
stock  solutions.

Soil/sediment  samples   were   also
analyzed   using  the  Level   1-  FID
instrument  (see  Fig.   1)  to  further
                                            329

-------
eludicate  the  thermal  extraction process
in an  independent  study  at  Louisiana
State   University.     This    set   of
experiments  seeks  to  identify  factors
influencing   analyte   recoveries   by
systematically   varying    operator-
controllable variables including gas flow
rates,  additives  to  facilitate  extraction,
extraction  temperature  and  duration; as
well  as  to  define  limiting  factors  for
target  analytes  and  matrices.    Three
analyte  solutions  were  prepared:   n-
triacontane   ("C-30"),   pyrene,   and
hexachlorobenzene  ("HCB").      These
compounds   were  chosen   for   their
thermal   stabilities    and   chemical
inertness.   Two  are structurally  similar,
all  three  are  neutral  and   devoid  of
reactive  functionalites.   Ten  u.1  of stock
solutions  in dichloromethane  (10  mg/ml
for  pyrene,  HCB;  2  mg/ml  for  C-30)
were spiked onto  the  soils immediately
prior to  analysis.   The  resulting  FID
signals  were  integrated  to  calculate
analyte  recoveries  (Table  3), with  the
FID  signal  of  the pure  analytes  (no
matrix)  as  reference.

Conclusions

Level   1  Thermal Extraction/FID

Thermal   extraction  efficiencies  vary
considerably with the nature  of analytes
as  well  as  matrices  (Table  3).   While
conventional    solvent    extraction
procedures   would  be   expected   to
produce  similarly high  recoveries  for n-
triacontane,      pyrene,      and
hexachlorobenzene,  thermal  extraction
produced  markedly different  results  for
clay as matrix (Fig. 2a).  HCB recoveries
were  quantitative,  while   C-30   and
pyrene  recoveries  ranged  at   approx.
30%.  Variation  of  the  matrix had  a  less
pronounced  effect  on  the  recovery  of
pyrene.      These  results   cannot  be
 explained solely  in  terms of polarity  or
 volatility.    Not  surprisingly,  percent
 deviations    of   recovery   decrease
 dramatically  in  the  presence  of a  soil
 matrix (Fig. 2b).   The increase of helium
 flow   during  the  thermal   extraction
 process from  40 to 100  ml/min  did not
 increase  the  extraction  yields of  C-30
 significantly  (see  Table  3); however,
 addition  of polar  additives  to the  soil
 samples   immediately   before  thermal
 extraction,  such  as  water or phosphoric
 acid,  improved  the recovery  of  pyrene
 from clay markedly (Fig. 3c).   Figure 2d
 illustrates blockage  of  reactive sites of
 the  soil  matrices by  repeated  spiking of
 the  same river  silt  sample.   Thermal
 extraction efficiences increased from  25
 to  65%.   Simple physical obstruction of
 the  carrer gas  flow is  certainly  one of
 the   factors   contributing  to  reduced
 recoveries.   The  soil  samples  "cake"  and
 block  the desorption of analytes into the
 carrer gas flow.   Thus, recoveries  sank
 to 69% for pyrene and  to  82% for C-30
 when  standards  were spiked  onto  the
 lids  of crucibles filled with 100 mg  clay
 without  direct  contact  between  analyte
 and  matrix  (Table 2).   Repeated  thermal
 extraction,  the  increase   of  extraction
 temperatures    above     450<>,  or an
 extension of  extraction  times  were  not
 promising,  as  illustrated  by  Fig.  4a-c.
 These  figures  compare  the  thermal
 desorption   of  identical  amounts  of
 pyrene  (100 ng)  from  a   porous  quartz
 crucible  (Fig.  4a)  and  spiked  into  a
 kaolinite  clay  sample  (Fig.  Figure  4b)
 using  the  temperature profile shown in
 Fig.   4c   under   otherwise   identical
 conditions.    Not  only  is  the thermal
 desorption  of  the  standard  from   the
 spiked  clay  considerable   below   100%,
 but  it   is  also   shifted  to  higher
 temperatures.    At  450°,  no  further
analyte was  released  upon  prolonged
heating.   The  fate  of  the unextracted
                                             330

-------
 analytes  is  currently  unknown   and
 subject  to future  investigations.

 Level  2  Thermal  Extraction/GC/MS

 The  results  of  analyses  from  all  three
 laboratories  are  summarized  in  Table 1.
 The   20   organic   compounds   and
 corresponding characteristic  ion  masses
 are listed  along with  linear correlation
 coefficients   (r)  derived  from  the   five
 point  calibration  curves  of  the on-lid
 stock  solution injections.   Fig.  3 shows
 examples   of four  calibration  curves
 from  one laboratory;  the  more  volatile
 components    (e.g.    naphthalene)
 experience variable rates  of evaporation
 after  injection  of the  standard  stock
 solution  onto  the  porous  quartz  crucible
 lids prior to  sample   insertion  into  the
 pyrocell  (approx.  2  min  from  injection
 onto   the   lid   until   sample  loading).
 Dioctyl phthalate signals were  relatively
 low except  at high concentration levels
 (300-400  ng);  after  it   appears  that
 much  of  this   compound  degraded  to
 phthalic  anhydride  (which  was  always
 detected)   during the   on-lid calibration
 runs.   Diethyl phthalate, in  comparison,
 showed    good   linearity   and   less
 degradation.       Pentachlorophenol
 linearity   was  not  as   good  as  that of
 other compounds in  the same  volatility
 range.    All  other compounds  showed
 good  linearity.

 Also  listed in Table  1 are  the  percent
 relative standard deviations  (%RSD) of
 calculated  response factors based  on  the
 on-lid  injections  of 20  ng/u,l mix of 20
 compounds  plus  the  deuterated  internal
 standards listed in Table 2.  Since %RSD
 values  are  also  a  measure   of  the
precision  for  each  compound, it is  not
surprising   that most volatile  compounds
also  show   the  highest  deviations.
Although   there   is   some   variation
between  the  participating  laboratories,
specific  compounds  tend  to yield  high
%RSD   values   while  others  showed
consistently good precision.   The  same
holds for  deuterated standards.   Again,
the  more  volatile  naphthalene-d8  and
dichlorobenzene-d4  showed  the   most
variation,    phenanthrene-dlO    and
chrysene-d!2  the least.

From the  obtained  data set,  recoveries
could   be  calculated   either   by  the
external  standard  method  using  the
least square  fits  of   the  five  point
calibration  curves for all compounds or,
alternatively,   by   internal   standard
quantitation   based  on  the  response
factors  calculated  for  each compounds.
Table  1  lists  results for  both  methods,
which  do  not   reflect  the  expected
improved  accuracy  for   the   internal
standard   method.        Due   to   the
considerably  different  chemical   and
physical  environments  the  standards
experience   while   being   partially
adsorbed   by  the   soil  samples   and
partially  by  the porous  crucibles,  no
high degree of accuracy can be expected
by  the  internal  standard  method.    The
implicit    assumption    made    in
conventional  chromatography,  namely
that   standards   and   analytes   are
subjected  to  identical  environments,
cannot  easily  be realized  in  thermal
extraction.

Percent   recovery   appears   to   be
dependent  on   a number  of factors
including  polarity,  molecular  weight,
and  interactions   with constituents of the
soil  matrix, both organic  and inorganic.
Not   surprisingly,    recovery   was
significantly    greater    for    many
compounds from  the  river  silt than  from
the  clay  or  subsurface   soils  (e.g.,
phenanthrene:   11% from  clay  and  31%
from  silt)  while  chloronaphthalene  was
close to 100%  for both clay  and  silt.   The
                                           331

-------
recovery of  diphenylamine  was  equally
low (approx. 5%) for clay  and  silt.  Since
the subsurface  soil  contains  about  30%
clay,  percent recovery is  generally  in
between  those  for  clay and  silt.  It is
interesting  to  compare  recoveries  for
the   structurally   similar   tricyclic
compounds  dibenzothiophene, fluorene,
and  carbazole.    In  all three soil  types
the  order  of  recovery  efficiency  was
dibenzothiophene>fluorene>carbazole,
which  likely reflects  increasing  binding
to  the  soil  matrix.    At the 0.5  ppm
concentration    levels,   naphthalene,
chloronaph thalene,     fluorene,
hexachlorobenzene,   dibenzothiophene,
phenanthrene,  aldrin,  and  pyrene  were
all detected  in  the  soil  standards  in at
least two of the three laboratories.

It is apparent  from  these  results  that
small aliquots  of soils can be  analyzed
by  thermal  extraction/GC/MS   without
any  prior  sample  preparation.    While
the  method  is  generally   suited  for
situations  requiring  high  precision  or
low  detection limits, it performs well  a
analyte   concentrations>50   ppm,  is
amenable  to  full  automation and  will
serve   for   rapid   screening of   soils
contaminated   with  thermally  stable
organic  semivolatiles,   a  class   of
compounds  that includes PNA's PCB's,
most petroleum  products  and pesticides
and  is   commonly   encountered  in
hazardous  waste cleanup  efforts.
References

Bath,  R.F.,  personal  communication
(1989).

Environmental Protection  Agency,  "Test
Methods   for  Evaluating   Solid  Waste,
Physical,  Chemical  Methods",  SW-846,
Third Edition, Method 8270 (GC/MS  for
semi-volatile  organics:  capillary column
technique).

Henry, C.B., Overton, E.B., and Sutton, C.,
"Applications  of  the  Pyran  Thermal
Extraction-GC/MS   for   the   Rapid
Characterization  and  Monitoring   of
Hazardous Waste Sites";  Proceedings  of
the  First International Symposium  for
Hazardous  Waste  Site  Investigations,
399-405  (1988).

Overton,  E.B.,  Henry, C.B., and Martin,
S.J., "A  Field Deployable Instrument  for
the    Analysis   of    Semi-volatile
Compounds   in  Hazardous   Waste";
Pittsburgh Conference and  Exposition  on
Analytical   Chemistry   and   Applied
Spectroscopy,   New  Orleans,   LA.,
Abstract  (1988).

Zumberge, J.E.,  Sutton, C., Martin,  S.J.,
and  Worden,  R.D.,   "Determining  Oil
General  Kinetic  Parameters by Using  a
Fused  Quartz  Pyrolysis  System";  Energy
and  Fuels, 2,  264-266 (1988).

Junk, T., Irvin, T.R.,  Donnelly, K.C.,  and
Marek, D.,  "Quantification  of Pesticides
on  Soils  by  Thermal  Extraction-GC/MS",
in  preparation.
                                                 Acknowledgements

                                                 We thank Drs.  R.J. Bath and D. Flory for
                                                 helpful comments and  suggestions.
                                            332

-------
u
U
                                              AIR
                                                Sample Crucible




                                                  	LC02
                       LEVEL I-FID ANALYZER


                       Figure  1
                                                                                               sScale


                                                                                               10cm
                                                                              MS
COI.UUN LXII 	

LLTT OR RIGHT

   SIDE
                                                                                   PYROCEU

-------
                  MATRIX vs RECOVERY
                         Flow 40 ml/min
       % Racovery
N°n«         Clay


 •I Pyrene    E3 C-30
                                 Subsoil         Sill



                                  Hexachlorobenzene
Fig 2a
             ADDITIVES DURING EXTRACTION
                          Pyrene on Clay
     so/
             Nona
                                        Phosphoric Acid
                                                                     DEVIATION OF RECOVERY
                                                                       Comparison of Different Matreces
                                                                    % Sid Dev.
                                                              Fig 20
Nons         Clay         Subsoil         Silt



 ^ Pyrene   IZ2 C-30   [^ Hexachlorobenzene
                                                                  REPEATED SPIKING OF SOILS
                                                                            Pyrene on Sand
                                                                                      Extraction #
                                                              FJg 2d

-------
          NAPHTHALENE
HEXACHLOROBENZENE
   Area(m/e 128)(Thousands)
                                       Aroa(nVe 284)(Thousands)
Fig 31
                                    Fig. 3D
         PHENANTHRENE
   Area(nVe 178)fT"housands)
        100     200    300    400     500
   BENZO(A)PYRENE
                                            100    200
                                                             400    500
Fig. 3c
                                    Fig 3D

-------
'ABLE i. Precision ol semi -volatile standard calibration curves and percent recovory


COfvPCUvD
2-chlorophenol
4-methylphenol
2.4-dichlorophenol
naphthalene
4-chloro-3-melhylphenol
l-chloronaphlhalene
2.s-dinitrotoluene
lluorono
cJlo ihylphlhalu lo
diphcnylamine
hexachlorobenzene
dibenzothiophc-ne
peniachlorophenol
phenanlhrene
carbazole
nldrin
pyrcno
bis[2-ethylhexyl)ph!halati
benzo(a)pyrene
Based on 25 on-lid injeclio
2 Response (actors (RF) and
(area cua/i mass of dcul. st
3 Percent theoretical recovery
50 ppm per component)
•• Rased on deutcratcd intcrna
5 Average linear correlation
These compounds were also

QUAN
MASS (m/z)
128
107
152
128
107
162
165
166
1 45
169
284
184
266
178
167
66
202
149
252
252
TS; 4 different
sercent relative
. (ng compoun
based upon I
3 different in
standards sp
II
coefficient lor
II
detected from
II
I
(vCAN
r5
0.9152
0.9481
0.8962
0 7484
0.9708
0.9385
0.9848
0.0007
0.07 1 "/
0.9801
0.9901
0.9946
0.9537
0.9943
0.9932
0.9924
0 9971
0.9922
0.9931 |
inslrumen
standard c
d)
nearily of
Irumenls a
kcd diroc[l
the on-lid
	 !—
the 0.5 pp
MEAN1
%RSO
110
85
67
66
29
23
26
1 -1
34
20
10
44
8
10
1 1
7
38
37
s: 3 differ
eviation ("/
%RSD . 1
id injeclio
t 3 diflere
Y on soil c
live poin
	 H-
m slandar

%RSC
RL
147 73
96 76
78 50 78
88 35 59
27 26 27
29 6 22
18 41 16
10 9 24
14 GG 14
21 32 14
674
734
51 56 28
835
11 115
845
730
26 19 39
38 10 36
>nt laboratories.
»RSD) were calc
00 (SD/RF)
ns (slope and y
nl laboratories
nd corrcspondin
calibration fror
ds in at least 2

OF F
TAW
;>4
R4
68
43
1 9
??
?4
0
i a
8
22
1 5
28
1 9
1 3
29
1 4
53
38
38
40
ulaied
interc
n RF
~T
n lhr£
nf the
I

IF 2
LSU
64
103
46
35
31
22
GO
25
9
5
56
7
1 1
8
5
70
6'
o 400
I
based
cpl) ant
values-
I
e diffo
i
3 labor
n

Cl
RL TAM LSI
15 46
8 10 V
9 26 33
60 100 72
7119
69 77 141
10 6 4
4159
835
71 76 57
12 28 19
~8 13 12
434
234
4 12 5
10 1 1
9 1
:g per componenl
on EPA method 8
not on deuterale
only PNAr. nro cc
ent laboratories.
atones.


.AY
\ AVE
30
1 0
23
' 771
9
96'
7
9'
5
68'
20'
11 '
4
3
7
4
4
per in
>70 ol
d inter
mparo

% RECOVERY 2
50 PPM STANDARDS
SILT
RL TAM LSU AVE
19 25 - 22
8 6 6 77
23 23 15 20
93 128 138 120'
14 12 9 12
83 118 109 103'
11 12 7 10
20 31 17 25'
04 5 '6
71 67 54 64'
40 39 26 • 35'
36 33 23 31'
10 7 6 8
23 20 15 19'
20 24 13 19'
4779
14 5 6 8
eclion. 	
SW-846; i.e.: RF: (area qu<
nal standards spiked inlo
3 duo to tho chemical sim


SUBSOIL
RL TAM LSU AVE
2 o
9847
25 44 9 26
27 41 121 63'
9 15 4 9
>1 78 - 70'
10 16 4 10
12 27 10 19'
87 46
58 78 103 SO'
•2 53 66 50'
8 47 46 40
8 12 7 9
1 21 13 15
1 32 17 20
018 4 11
0 20 2 11
in mass compound) (nrj
soil. Average oJ 9 re
aritics with the slnndn


% RECOVERY 4
CLAY SILT SOIL
33 93 67
3.1 55 45
50 78 85
35 65 52
	 	 	
3 15 10
2 149
deut. internal standard)/
ilicates (about lOmg soi
ds.

-------
TABLE 2

Variations of Internal Standard Areas
Deuterated
Internal Standards
Quan. Mass
                                                 Mean %S     Ruska     Ruska    Ruska    TAMU
                                                                         LSU
.1 ,4-dichlorobenzene-d4
naphthalene-d8
acenaphthene-dlO
phenanthrene-dlO
chryscnc-d!2
perylene-dl2
152
136
164
188
240
264
106
57
17
8
7
34
75
41
7
6
6
43
155
74
22
9
11
38
139
70
13
11
11
26
73
43
10
6
3
19
89
57
32
8
4
45
* Values are based on 25 on-lid injections recorded at 3 different laboratories; 200 ng/component,
 injection.

-------
          TABLE 3

          Evaluation of Thermal Extraction on Pyran Level 1

            Exp#     Analyte     Amt(ug)       Matrix    He ml/minNo.of Obs.    Avg      %Dev
1
2
3
4
5
6
7
8
9
10
pyrene
pyrene
pyrene
pyrene
C-30
C-30
C-30
C-30
HCB
HCB
100
100
100
100
20
20
20
20
100
100
none
clay
subsoil
r.silt
none
clay
none
clay
none
clay
                                                                     40
                                                                     40
                                                                     40
                                                                     40
                                                                     40
                                                                     40
                                                                    100
                                                                    100
                                                                     40
                                                                     40
                                                       3
                                                       3
                                                       3
                                                       3
                                                       3
                                                       3
                                                       3
                                                       3
                                                       3
                                                       3
                                              2837.3
                                               717.0
                                               716.3
                                              1063.0
                                              1101.0
                                               341.0
                                              1241.3
                                               396.7
                                               613.7
                                               667.7
                                                   3.7
                                                   1.8
                                                  18.3
                                                  29.7
                                                   1.4
                                                  18.5
                                                   2.0
                                                   2.4
                                                   8.4
                                                   2.1
                                           %Rec

                                             100.0
                                              25.3
                                              25.2
                                              37.5
                                             100.0
                                              31.0
                                             100.0
                                              32.0
                                             100.0
                                             108.8
          Addition of phosphoric acid, 85%, 0.1 ml
              11       pyrene

          Addition of water, 0.2 ml
                  100
          r.silt
                 40
                    1246.5
              18.1
                                                                                          43.9
              12       pyrene            100       clay

          Standards on lid of crucible filled with soil
                                             40
                                               868.0
                                                             30.6
              13
              14
pyrene
 C-30
100
 20
clay
clay
 40
100
1984.0
1018.7
                                                                                                       4.7
69.9
82.1
         Repeated spiking of previously extracted soil
              15       pyrene
              16       pyrene
              17       pyrene
         C-30: n-triacontane
         HCB:  hexachlorobenzene
                  100
                  100
                  100
          r.silt
          r.silt
          r.silt
                 40
                 40
                 40
        Extr.#
          1
          1
          1
                                                       DISCUSSION
 768.0
1330.0
1860.0
27.1
46.9
65.6
AL ROBBAT: Have you tried looking at organic extracts? Can you place an
organic extract in the soil and look at the thermal desorption properties? In other
words, lake the soil, extract it with melhylene chloride, taken out of part of the
extraction, and run your experiment?
THOMAS JUNK: In other words, you introduce an organic extract that has been
extracted in a conventional procedure  to see how that behaves in the instrument
itself? I'm not quite sure that I understand your question.

AL ROBBAT: We found the same thing. For example, for PAHs, if you take the
thermal desorption sample probe and place it  directly over the soil, you get
between 7% and 15% extraction recoveries. What I'm suggesting is that if you
use the simple 2 mL extraction procedure that I described, take a half a gram of
soil, add 2 mLs of methylene chloride,  extract it, add that extract to your cell, can
you perform that experiment? Can you use say, 1Ou]s or 20 uJs of extract in your
cell. Have you tried that experiment?
                                             THOMAS JUNK: Yes, you can. You can use a conventional extract and absorb
                                             it onto the porous quartz crucible. In order words you could go through the
                                             addition of a small amount of solvent into the soils that would then by and large
                                             produce a similar effect as the one you just mentioned.
                                             AL ROBBAT: Right.

                                             THOMAS JUNK: Yes, we've tried that. And in some cases it  produces
                                             satisfactory results. However we have not consistently found an improvement
                                             over classic extraction procedures. I think I can cover that together with the
                                             addition of various cold solvents that I just mentioned, such as with phosphoric
                                             acid or water. And yes, we do see an increase, bu the increase for phosphoric acid
                                             for example, was much more significant than that for other cold solvents.

                                             STEPHEN BILLETS: I want to  say from the  standpoint of testing this
                                             technology that Ruska Thermal Extraction System  is in the EMSL-Las Vegas
                                             Laboratory currently  undergoing evaluation  for possible  use in a field
                                             demonstration study. So we are conducting, in our laboratory, a complementary
                                             effort to what Thomas described.
                                                                  338

-------
           THE APPLICATION OF A MOBILE ION TRAP MASS SPECTROMETER SYSTEM
                       TO ENVIRONMENTAL SCREENING AND MONITORING
                          William H. McClennen, Neil S. Arnold, Henk L.C. Meuzelaar,
                                     Erich Ludwig* and JoAnn S. Lighty**
    Center for Micro Analysis &
    Reaction Chemistry, University of
    Utah,  Salt Lake City, UT 84112
*GSF Munchen Institut fur
Okologische Chemie, Ingolstadter
Landstrasse 1, 8042 Neuherberg,
Germany
**Chemical Engineering
Department, University of Utah,
Salt Lake City, UT 84112
ABSTRACT

This paper presents examples of the use of a mobile Ion
Trap Mass Spectrometer (ITMS, Finnigan MAT) for on-
site environmental screening and monitoring of vapors by
gas chromatography/mass spectrometry (GC/MS). The
instrument is built around a miniaturized ITMS system,
with a novel direct vapor sampling inlet and coupled to a
high speed transfer line GC column (short capillary
column with fixed pressure drop).  The column is
temperature controlled inside the standard ion trap
transfer line housing.  This provides for high speed
analyses at 10-60 s intervals using an  automated sampling
system constructed with only inert materials  in the
sample path.

Specific laboratory and field applications exemplify  key
characteristics of the system including sensitivity,
specificity for a broad range of compounds, ruggedness
for field testing in harsh environments, and general speed
and versatility of the analytical technique. The system
has been calibrated for alkylbenzenes  at concentrations as
low as 4 ppb in air and used to monitor these compounds
in an office space.  Both the MINITMASS and a simpler
Ion Trap Detector (ITD) based system have been used to
monitor organic vapors from acetone through 5 ring
polycyclic aromatic  hydrocarbons produced in  laboratory
scale reactors for studying the thermal desorption and
incineration of hazardous wastes.  The ruggedness of the
MINITMASS system has been demonstrated by vapor
sampling in the Utah summer  desert and  at a 600 MW
coal fired  power plant. Finally, the analysis speed and
versatility  are described for vapor monitoring of volatile
organic compounds at  an  EPA national priority list waste
site.
                   INTRODUCTION

                   Preliminary data obtained with a miniaturized Ion Trap
                   Mass Spectrometer (MINITMASS) system developed in
                   close collaboration with the manufacturer (Finnigan MAT
                   Corp.) were presented at the first International
                   Symposium on Field Screening Methods for Hazardous
                   Waste Site Investigations (1).  The MINITMASS system
                   was shown to  be capable of performing tandem MS
                   (MS") analyses in electron ionization (El) as well as
                   chemical ionization (CI) mode and featured a special air
                   sampling inlet in combination with so-called "transfer line
                   gas chromatography" capability (2,3).  Due to its
                   relatively low  weight (approx.  280 Ibs.), the
                   MINITMASS  system was readily  transported inside a
                   small mobile laboratory mounted on a regular 3/4 ton
                   pick-up truck (1).  Some of the main shortcomings of the
                   MINITMASS  system included: insufficient sensitivity
                   (high ppb/low  ppm range), limited mobility for many
                   field screening applications (due to a shock sensitive
                   turbomolecular pump),  untested performance with low
                   volatile (e.g., PAH type) compounds and lack of field test
                   data at actual hazardous waste  sites.

                   Since October  1988 the MINITMASS system has been
                   tested under a  variety of conditions at several outdoor as
                   well as indoor locations.  Moreover, several hardware and
                   software improvements have markedly increased its
                   sensitivity (currently ~1 ppb for alkylbenzenes (3)) and
                   applicability to low volatile compounds (e.g., 3-5 ring
                   PAH's (4)) while enabling true mobility through the
                   installation of  a more rugged vacuum pump.  In addition,
                   a simplified mobile ITD (Ion Trap Detector) system was
                   constructed and tested for dedicated hazardous waste
                   combustion applications which do  not require MS"
                                                      339

-------
capabilities (5,6). Hazardous waste related monitoring
applications of both systems will be described in the
following paragraphs. Some of these results have been
presented elsewhere (2-6) but are included with various
new data for completeness in this overview  of the
instrument's current performance.

EXPERIMENTAL

General Parameters

The direct atmospheric vapor sampling inlet described in
detail elsewhere (2,5,6), consists of three concentric tubes
with appropriate flow control plumbing and electronics.
The inlet system is made from deactivated fused silica,
quartz and  glass, or glass-lined metal tubing. The sample
path contains no moving parts.  When sampling, the gas
is exposed  to the column inlet for a controlled period of
time (0.3 to 2 s) while 30 to 200 ^L  of sample is
admitted to the column. Helium carrier gas flow is then
restored  for the rest of the sampling cycle and GC
separation of the sample takes place.

This inlet is coupled to a 1 m long fused silica capillary
column which enables nominal GC separation of
components and provides a pressure drop between the
near-ambient sampling environment and the high vacuum
of the mass spectrometer ion source.  With  the fixed
pressure drop, the chromatographic conditions are
controlled primarily by column  length, radius and
temperature (2). The fused silica capillary column used
in this work was either a 0.15 mm  ID x  1.2 urn methyl
silicone film (CP-Sil-5 CB, Chrompak) with 1.5-2 ml/min
He How or a 0.18 mm ID x 0.4 urn film DB-5 (J&W
Scientific) with ca. 4 ml/min He flow.

This transfer line GC inlet system was used with both a
regular Finnigan MAT ITD and the MINITMASS system
with axial modulation and tandem MS capabilities (1),
although this paper describes results from use in only the
single MS mode. In addition to these special capabilities,
the MINITMASS system permitted higher flow rates by
virtue of the axial modulation feature and was thus the
only system which could use the 0.18 mm ID column.
The combination of increased MS resolution and
increased flow rates resulted in higher sensitivity.

Field  Testing

In field tests sponsored by the EPA, vapor standards were
diluted into a 5 m long x 2.5 cm ID glass and Teflon®
manifold with a 2.2 1 s'1 total flow.  These standards
included both 50 ppm compressed gas mixtures and
equilibrium  headspace vapors of pure compounds injected
 into the manifold with a motor driven syringe pump.  The
 50 ppm gas standards were diluted to calibration mixtures
 of 20 to 350 parts per billion (ppb) in air, while the
 syringe pump produced mixtures ranging from 16 ppb to
 over 10 ppm, depending on the compound vapor pressure,
 syringe diameter and motor speed.

 Vapor standard calibration data were obtained scanning
 from m/z 45 or 50 to 200 at 4 scans s"1. The inlet and
 transfer line were maintained  at 25°C,  while the ion trap
 was maintained at 85°C.  The temperature of the mixing
 chamber was ambient and not controlled. The vapor  inlet
 drew up to 120 ml min"1 from the EPA vapor manifold.
 A .5 s vapor sample pulse was frequently used, although
 the EPA experiments involved varying the sample time
 from 330 ms to 2.5 s with a routine pulse width of 715
 ms.

 Combustion Monitoring

 Exhaust from a rotary kiln simulator was monitored for
 gas phase hydrocarbons during the combustion of
 polymeric medical supplies (6). The 11 g batch samples
 were loaded into the kiln and incinerated at 600°C.
 Rapid on-line analyses were obtained using the
 unmodified ITD system.  A sample flow of 25 to 50 ml
 min"1 was drawn from the kiln exhaust gases  in a
 transition area preceding the afterburner. Vapor samples
 were taken at 10 s intervals to monitor concentration
 transients during sample combustion.  The .15 mm ID
 column was used in this system at a constant temperature
 of 30°C or 82°C with the vapor inlet at 60 or 100°C, and
 the mass spectrometer was scanned from m/z 35 to  120
 or 50 to  148 at 4 scans s"1.  Exhaust from a thermal bed
 reactor for hazardous waste studies (7) was monitored
 using the MINITMASS during thermal dcsorption of
 polycyclic aromatic hydrocarbons  (PAHs) from contam-
 inated soils obtained at former coal gas plant sites (4).
 The soils were heated to 400°C under  a radiant heater
 with a preheated nitrogen flow of .5 1  min'1 above the bed
 of soil.  The  exhaust gas was sampled  at 60 s intervals
 with 2 s  vapor sampling pulses. The inlet was operated
 between  150 and  175°C and the transfer line was
 maintained at 125°C.  The separation was performed in
 the .18 mm ID column specified above.  The ion trap
 manifold was 200°C and the MS was scanned from  m/z
 100-300  at 4 scans s-1.

 RESULTS AND DISCUSSION

Sensitivity and Dynamic Range

The basic objective of vapor sampling short column gas
chromatography is to provide sufficient separation of the
                                                        340

-------
organic compounds of interest from the major
atmospheric constituents to allow optimum use of the
sensitivity, specificity and speed of the detector.  The
sensitivity of the MINITMASS system utilizing this
principle is demonstrated by the analysis of toluene vapor
in air in Figure 1.  This figure shows ion chromatograms
for six repetitive samples of toluene in air at a
concentration of 16 parts per billion (ppb, volume or
molar ratio).  The sampling points are indicated in the
total ion chromatogram (TIC) of Figure la by the
baseline depressions at 25 second intervals.  The short
pulses of air, ca. 70 jil in 0.7 s, suppress the baseline by
overloading the ion trap with air ions so that even back-
ground  ions in the MS are  not detected. However, the
well resolved toluene  peaks elute from the short column
20 s later with excellent sensitivity and signal to noise
(s/n) as indicated by the selected ion chromatogram of
summed ion peaks at  m/z 91  and 92  shown in Figure Ib.

In addition to this 16  ppb data,  Table 1 presents a set of
calibration points for toluene showing the degree of
sample repeatability expressed as  relative standard
deviation.  For statistical reasons, a minimum of 5
consecutive vapor samples were taken at each
concentration.  These  concentrations were prepared via
the syringe pump method described above, and response
was measured via the  peak area of the m/z 91 trace from
a .715 s vapor sample. A linear fit to the full set of data
points was obtained with the  correlation coefficient R =
.998, indicating a linear dynamic  range of 3 or more
orders of magnitude.

A practical application of alkylbcnzcne analysis is shown
in Figure 2 using the  thin film column (0.4  urn  rather
than 1.2 ^im used in Figure 1).  Figure 2 presents two  ion
chromatogram traces from indoor atmospheric sampling
in one of our office work areas. The m/z 91  trace clearly
shows peaks for toluene, ethylbenzene, m- and p-xylene
and o-xylenc.  The estimated toluene concentration is 70
ppb, presumably derived from the glue used on  the
recently installed ceiling tiles. The MINITMASS system,
with its axial modulation capability and the higher flow
of the .18 mm ID column has shown alkylbenzene
detection limits near 1 ppb with a s/n greater  than 2.  The
normal  ITD-based system, limited to the 1.5 ml/min of
the .15 mm ID column, has shown detection limits of
approximately 20 ppb.

Speed and Selectivity

An example of the GC/MS vapor analysis of a 7
component standard gas mixture is shown in the partial
chromatograms  of Figure 3.  Figure 3a shows the total
ion chromatogram (TIC) while the concurrent selected ion
 chromatogram profiles show the major ions from four of
 the test compounds. Arrows in the TIC profile indicate
 the beginning points for 3 subsequent 715 ms samples
 with a 30 second sampling interval.  Although the 1,1,1-
 trichlorocthane (111TCA) and benzene peaks are not
 completely resolved in the TIC, they are readily
 quantitatcd based on the selected ions from their unique
 mass spectra. Note that the small vinyl chloride peaks at
 scan numbers 5, 125 and 245 are partially cut off by the
 large air pulse baseline disturbances from which they are
 incompletely separated.  However, even this early eluting
 compound had a reproducible, linear response curve over
 the range of 20 to 350 ppb which we tested.  In other
 words, the limited resolution of the short GC column is
 sufficient to greatly enhance the specificity and selectivity
 of the mass spectrometer.

 A major asset of the speed of short column GC/MS is the
 ability to do  on-line monitoring in a nearly real-time
 mode. Figure 4 shows a set of chromatograms
 monitoring the evolution of volatile organics from the
 combustion of polypropylene materials in a rotary kiln
 simulator (6). For  these experiments the ITD system
 with the .15 mm ID, thick film (1.2 ^im) column sampled
 gases just prior to the afterburner. A 0.5 s  sampling
 every 10s with the column  at 82 C was sufficient to
 follow the transient concentrations of aromatics during
 the 2 min experiment.  Selected ion traces at m/z 78 and
 91 show the specific benzene and toluene peaks in the
 repetitive analyses.  Figure 4b explicitly plots the
 quantitatcd evolution curves obtained from peak areas of
 selected ions for benzene, toluene, phenol and styrcne.
 With Ihc column at 30°C, compounds as small as  acetone
 were separated from air with benzene eluting at 7s.  The
 peak concentrations of these hydrocarbons occurred
 before and after the point at which the melting plastic
 was totally engulfed in flame in the 600 C rotary  kiln
 simulator.

 Figure 5 illustrates the high  boiling range of compounds
 which can benefit from  the speed and selectivity of short
 column  GC/MS in an analysis of polycyclic aromatic
 hydrocarbon (PAH) vapors during thermal treatment of a
 contaminated soil (4).  By elevating the temperature of
 the thin film column to  125  C, compounds with boiling
 points ranging from 218 to 340 C were readily analyzed
 with a 60 s sampling internal. The numbered peaks in
 the TIC represent 1) naphthalene, 2) methylnaphthalcnes,
 3) C2 naphthalenes,  4) fluorene, 5) phenanlhrene and 6)
anthracene.  Although the GC resolution was insufficient
 to separate some of these isomers, the selected ion
chromatograms show the obvious benefits.  In Figure 4b
the m/z  154 trace shows a prominent peak for biphenyl
with additional unresolved humps for fragments from
                                                        341

-------
larger compounds. The m/z 168 trace indicates that the
short column is able to resolve the methylbiphenyls (at
12 s) from the dominant dibenzofuran peak (at 14 s).
The m/z 184 trace shows separate peaks for the C4
naphthalenes at 22 s, dibenzothiophene at 37 s, and ions
associated with the intense phenanthrene (m/z 178) peak.

Instrument Ruggedness

Although the commercial versions of both the ITD and
ITMS  instruments were designed  principally for
stationary laboratory operation,  the modified ITMS,
which  was built around a normal  ITD chassis, has held
up  well under the rigors of harsh  transportation and
environmental operating conditions.  Two examples of
applications in particularly harsh environments include
the Utah summer desert and flue gas sampling from a
coal fired 600 MW power plant.

The desert testing was the maiden use of the system to
discover potential field problems. It involved operation
of the  instrument in the mobile  lab on the  back of a 3/4
ton pickup truck  lo sample various chemical vapors
released from a permanent dissemination line (1). The
instrument was severely bumped and jostled as we
maneuvered the truck to sample in the shifting winds on
the brush covered terrain.  Operation was complicated at
the time by the necessity of venting and then restarting
the instrument before and after  each move to avoid a
turbomolccular pump crash. Typical down time  was
approximately one hour between data acquisitions,
including cool down and warm  up.  Despite the rigorous
handling, the instrument's only  failure during the 2 weeks
of testing was an overrated power transistor which had
previously failed in normal lab  operation on a different
ITMS.  The turbomolecular pump has recently been
replaced with a more rugged model to allow true
"mobile" use in addition to "transportable" operation but
has not yet been  rigorously tested in the "mobile" mode.

The flue gas sampling involved rolling the MINITMASS
instrument to a seventh floor site at a Utah Power and
Light Company 600 MW coal fired power plant.  The
objective was to  do on-line analysis for aromatic
hydrocarbons in the 350 C flue gas  as it exited the main
boiler  sections.  No organics were observed  in the 1.5%
excess oxygen combustion  products although our on-site
detection limits were 4 and  10 ppb for alkyl benzenes and
alkylnaphthalenes respectively.  However,  simply
operating the instrument was a  major accomplishment in
this harsh environment. Fly ash was continuously raining
in the  ambient atmosphere  from the overhead structures;
the whole work site was constantly  rocking and rumbling;
and the ambient temperature ranged from ca. 4 to 35 C
(40 to 95 F). A makeshift plastic tent with a crude
window fan and filter was assembled over the instrument
to supply some measure of environmental control (in
addition to the chassis mounted fans and filters).

A recent example of the instrument's more "routine"
transportability  was demonstrated in EPA testing in New
Jersey.  The MINITMASS system was successfully
driven across the country in its mobile lab and brought
into operation with verified performance capability within
6 hours after arrival at the national priority list (NPL)
landfill site.  This start up time included transferring the
instrument from the mobile lab to the EPA site trailer and
diagnosis and repair of a broken thermocouple.
Operation in the mobile lab itself had previously been
verified in less  than 1 hour.

Versatility of Inlet Sampling

One of the unique features of our vapor sampling inlet is
the ability  to readily vary the sample volume injected
onto the column by a simple change  in the sample pulse
width.  Figure 6 shows calibration data for toluene from a
gas standard sampled with a 495 ms  pulse compared to
the typical 715 ms sampling.  The linear regression lines
along with 95% confidence intervals are shown for each
of the two data sets.  The shorter sampling pulse has very
similar repeatability with the smaller response slope
corresponding directly to the smaller sample and
reduction  in  pulse time within 4%.  Figure 7 also
demonstrates the linearity of mean peak area response
versus sample pulse width for two different scries of gas
mixtures.   The lower concentration data are from the
analysis of an "unknown" gas mixture (ca. 30 ppb of the
gas standard shown in Figure 3) which was run with 2.5
s sampling to maximize sensitivity for the later eluting
compounds.  The high concentration data was a test of
increasing GC resolution for the overlapping peaks
dichloromethane (DCM), 1,2-dichloroethane (12DCE) and
1,1-dichloroelhane (11DCA) in a gas mixture standard by
decreasing the pulse width to 330 ms.  The combined
data sets demonstrate excellent linearity over an eightfold
change  in sample size.

Figure 8 illustrates  the effect of sampling duration on
resolution and sensitivity. The same mixture of DCM,
12DCE, 11DCA and  tetrachloromethane at 350 ppb each
is used  in  8a with a sampling time of 330 ms and in 8b
with a time of 715  ms.  Clearly the 330 ms time
improves  the resolution of the early eluting compounds,
but comparison of the m/z 82 selected ion trace indicates
that the tetrachloromethane which elutes at 7 s has only
70% of the peak height obtained at 715 ms.  For
detection  of compounds at the lowest levels, peak height
                                                          342

-------
becomes the limiting factor as the ion counts must exceed
the noise threshold.

Another aspect of the vapor sampling performance of our
inlet is the ability to  readily sample from atmospheres at
different ambient pressure and limited small total volume.
Most of the EPA testing involved sampling standard gas
mixtures from a  high flow manifold operating at reduced
pressure conditions similar to those which might exist in
a system drawing samples from many separate remote
points on a fence line.  However, at the end of this series
of experiments, several Tedlar bag samples were also
analyzed.  In  order to maximize the number of separate
samples from a single 1 liter Tedlar bag, the total sample
drawn into our inlet was reduced from 120 ml/mm to 25
ml/min.  One of  the gas mixture standards was then
diluted to 2.5 ppm  in a Tedlar bag and our duplicate
analysis of it  showed responses within 4% of perfect
linearity for the manifold calibration lines.

Figure 9 shows the total ion chromatograms for the
analysis of landfill  wellhead vapors from the NPL site
which were sampled  into Tedlar bags and diluted by 1/5
with clean air. The identified compounds from these two
analyses are listed in Table 2 with detected concentrations
for those previously calibrated.

CONCLUSIONS

The results presented here have demonstrated
MINITMASS performance in several key areas of
capability which  might be expected from an on-site,
vapor sampling,  short column  GC/MS system based on
an ion trap mass spectrometer.  These include
specifically: detection limits of less than 10 ppb for a
variety of volatile organic compounds; selective analysis
of 21 compounds or  more in a single one minute
chromatogram with boiling point windows depending on
column type and temperature;  repetitive sampling as
frequent as each  10 s for monitoring transient vapor
concentrations; and direct variation of sample size with
sample pulse  time to readily optimize GC resolution
versus ultimate sensitivity. The examples of operation in
harsh environments and at remote sites further suggest
that the instrument is rugged enough for most  field
screening and hazardous waste site investigations. These
specific capabilities also apply to a similarly equipped
standard ITD  system except for the order of magnitude
sensitivity difference with the ITMS enhancement.
Although the  MINITMASS has the additional  advantages
of capabilities for MS"  and truly mobile operation as
compared to the  ITD described here, many  applications
do not require tandem MS and the more rugged turbo
could be user installed. However, the main advantage of
 the standard ITD for vapor sampling GC/MS is its high
 sensitivity in a commercially available instrument.
 Coupling the vapor inlet to bench top MS such as the
 ITD or a Hewlett Packard MSD is also a much more
 economical way to get transportable GC/MS into a field
 screening, stack monitoring or even process control
 application than the prototype MINITMASS.

 This paper also suggests the need  for further development
 in associated areas of field testing instrumentation.
 Foremost are the advantages in compound range and
 analysis speed which could be gained by broader
 temperature range operation and temperature
 programming for the transfer line  column.  Second is the
 capability for rapid on-line enrichment in case of more
 dilute target compounds or less sensitive detectors. And
 finally, there is the need for continuing development  in
 all aspects of miniaturization (size, weight,  power
 requirements) and ruggedization of fieldable GC/MS
systems.

ACKNOWLEDGEMENTS

This work was funded by the Advanced Combustion
Engineering Research Center (ACERC, which is
supported by the National Science Foundation, the state
of Utah,  23 industrial participants  and the U.S.
Department of Energy), U.S. Environmental Protection
Agency (EPA), Finnigan MAT Corporation, the U.S.
Army Chemical Research Development and Engineering
Center (CRDEC) and Utah Power  and Light Company
(UP&L).

The authors also wish to acknowledge the invaluable
assistance of Dale Urban for general field preparations,
David Mickunas (USEPA), A.  Peter Snyder (CRDEC)
and Kenneth R. Thompson (UP&L) for specific project
arrangements and ACERC participants Sue  Anne Sheya,
David Wagner, Eric Lindgren,  and David Pershing for
support of rotary kiln and desorption studies.

REFERENCES

1.     Mcuzelaar, Henk L.C.; McClennen,  William H.;
       Arnold, Neil S.; Reynolds,  Tim K.; Maswadeh,
       Wallace; Jones, Patrick  R.;  and Urban, Dale T.,
       "Development of the MINITMASS,  A Mobile
      Tandem Mass Spectrometer for Monitoring
      Vapors and Particulate  Matter in Air," First
      International Symposium on Field Screening
      Methods for Hazardous  Waste Site Investigations,
      Las Vegas, Nevada, October 1988.
                                                       343

-------
2.     Arnold, Neil S.; McClennen, William H.; and
      Meuzelaar, Henk L.C.; "Vapor Sampling Device
      for Direct Short Column GC/MS Analyses of
      Atmospheric Vapors," Analytical Chemistry, 1991,
      in press,

3.     Arnold, Neil S.; McClennen, William H.; and
      Meuzelaar, Henk L.C.; "On-site Transfer Line
      GC/MS" Analysis of Environmental Vapors Using
      a Modified  Ion Trap Mass Spectrometer," ACS
      Preprints, Div. of Environmental Chemistry, Vol.
      29, No. 2, Miami Beach, Florida, September 1989.

4.     McClennen, William H.; Arnold, Neil S.; Roberts,
       Kenneth A.; and Meuzelaar, Henk L.C.; Lighty,
       JoAnn S.; and Lindgren, Eric R.; "Fast, Repetitive
       GC/MS Analysis of Thermally Desorbed
       Polycyclic Aromatic Hydrocarbons (PAHs) from
       Contaminated Soils,"  Combust. Sci. and Tech.
       74(1-6), 1990, 297.

 5.     McClennen, William  H.; Arnold, Neil S.; Shcya,
       Sue Anne N.; Lighty, JoAnn S.; and  Meuzelaar,
       Henk L.C.; "Development of Novel, Mass
       Spectrometric Combustion Monitoring
       Techniques,"  ACS Preprints, Div. of Fuel Chcm.,
       Washington, D.C., August 1990.

 6.     Lighty, JoAnn S.; Wagner, David; Deng, Xiao
       Xuei Pershing, David W.; McClennen, William
       H.; Sheya,  Sue Anne N.; Arnold, Neil S.; and
       Meuzelaar, Henk L.C.; "On-line GC/MS Sampling
       of Exhaust Gas from a Rotary Kiln Simulator,"
        Air and Waste Management Assoc. International
        Specialty Conf. on Waste Combustion in Boilers
        and  Industrial Furnaces, Kansas City, Missouri,
        April 1990.

 7.     Lighty, JoAnn S.; Pershing, David W.; Cundy,
        V.A.; Linz, D.G.; "Characterization of Thermal
        Dcsorption Phenomena for  the Cleanup of
        Contaminated Soil,"  Nuclear Chem.  Waste
        Management, 8, 1988, 225.
                     Table 1
       Concentration   Peak Area     RSD
                                                     Table 2
                                 Tedlar Bag Samples of NPL Site Wellhead Vapors
Scan
No.
8
9
21
26
36
46
48
55
60
68
72
91
102
165
179
210
217
396
430
460
490
Compound2
11DCE
DCM
12DCA
Benzene
Trichloroethene
Cyclooctane
Chlorocyclopentene
Cyclooctane
Cyclooctane
Toluene
Cyclooctane
Cyclooctane
Perchloroethene
Elhylbenzene
m,p-Xylene
Styrene
o-Xylene
C3-benzene
Cj-benzene
C3-benzene
C3-benzene
Concentration
1-4 (1/5) 1-6
85
480
94
340





30,000 :


430
1300
3900

800




(Ppb)
(1/5)3
~
42
210
15
--
-

t
--
1,500
t
t
—
82
270

75
—
—
«
—
                                      1   Scan number corresponding to chromatograms
                                         in Figure 9.  This number divided by four
                                         equals retention time in seconds.

                                      2   Positive identification for standard compounds
                                         which were used in EPA testing while others
                                         are tentative, e.g. cyclooctanes could be
                                         octenes.

                                      1   Compounds  in 1-4 but not seen in 1-6 indicated
                                         by dashes while t indicates traces detected.
             16
            161
            245
            9450
 227
 1430
 2540
86530
14
3.4
1.0
1.7
                                                          344

-------
                                                                100%
                                                                                                         a. TIC (Total Ion Chromatogram)
                                   12C
                                           150
26 -
13 -
Q
C






b) m,

ll
30

'z 91 +

l
60
Tu
92
I
j
90 1
->e (s)


I !
JO

Figure 1.  (a) Total ion chromatogram for 6 repetitive
samplings of a 16 ppb toluene vapor standard. The
points of injection  can be identified by the suppression of
the baseline, (b)  A selected ion trace of m/z 91 and 92
for the same 6 samples. The threshold setting of the MS
prohibits exact calculation of the signal  to noise ratio, but
it appears to be 8 to 1 based on height,  and  >20  to 1
based on area. Adapted from reference 2.
   100
   =
  0)
  _o
   o
          m/z 91
                           2l 4
                         T i • i • r • i' i' i' i' i''' I''' i' i
       0      20      40      60     80      100
                       Time (s)
Figure 2.  Repetitive sampling of background levels of
alkylbenzenes in  a room with a  recently tiled ceiling.
Labelled compounds in (b)  the selected ion trace for m/z
91 are:  1) toluene; 2) ethylbenzene, 3) m,p-xylcnes and 4)
o-xylene.
                                                                        inyl chloride
                                                                            ' I ' I '           I            I
                                                                       ^ 12DCE
                                                                  33%
35<
Sea
Tin

7o
1 i ' i
n #
>e(s)
(1 benzene
i • i ' 1 ' i
i ' i ' | ' i
50
12

i ' i ' i • | • i •
Ji toluene
l ' i • i • | • i ' i •
100
25
« d. m/z 78 *
1 11
1 l ' ' ' ' ' ' ' ' ' ' ' '
\ e. m/z 91
150 200 250
37 50 62
                                                                            Figure 3.  Example analysis of a 7 component gas
                                                                            mixture at 350 ppb showing 3 repetitive samplings  at 30
                                                                            s intervals in a) total ion chromatogram and selected ion
                                                                            chromatograms for quanlitation of b) 12DCE, c)  111TCA,
                                                                            d) benzene and e)  toluene.  Note presence of ions m/z 61
                                                                            from vinyl chloride, 111TCA, and trichloroelhene and
                                                                            m/z 97 from trichloroethene that do not  interfere with
                                                                            quanlitation because of chromatographic separation.

-------
Co
».
O)
                      100 -
                       50 -
                   a
                   0)
                   a
                   a
u
«
    7.4 -

    3.7 ~

    0.0
                      3.4 q

                      1 .7 -

                      0.0
                    1300  -,
XI
p,
a

C
O
—>
C
U
«J
C
EJ
U
a
a
o
                     975  -
                     650  -
                     325  -
                                                Total  Ion Current
                                                                                   100%
                                                    m/z  78
                                  m/z 91

                                                               >%
                                                               05
                                                               o
                                                               t->-p-, , ^|-,T-,-^|

                          0   10   20   30   40  50   60   70   80  90   100
                                              Time (s)
                                                                 2.5%

                                                              1
                                                              «
                                                                      b)
                                                                                — 4.4%
                              10   20  30   40   50   60  70   80   90  100
                                             Time  (s)

                  Figure 4. (a) Ion profiles for  a sequence of vapor
                  samples  taken during the combustion of 11 g of poly-
                  propylene materials in a laboratory scale rotary kiln
                  simulator.  The ion m/z 78 is due to benzene,  while  m/z
                  91 indicates first  toluene, and  then partially resolved
                  ethyleenzcne and xylene isomers as in Fig. 2.  (b) Concen-
                  tration profiles in parts per billion  (ppb) for 4  compounds
                  obtained  from the integrated peak areas of selected ions
                  78, 91, 104 and 94 for benzene,  toluene, styrcnc and
                  phenol respectively. Adapted from  reference 2.
                                                                  0.2%
                                                                                                                      SOIL "A" UftPORS
                                                                                                                                             a) TIC
                                                                                                                      5,6
                                                                              •  I i                |       I    I    I   |       I    i       |   I

                                                                                                                           b) m/z 154

                                                                                                                                             c)  m/z 168
                                                                                                                                             d) m/z 184
                                                                               I  I                                I | T             T  |  I |  1  p

                                                                        °      8                22   Time(s)     37                «
                                                                           Figure 5. On-line vapor sampling GC/MS analysis
                                                                           durinr a 400 C thermal desorption of a contaminated soil.
                                                                           Chromatograms of the total and selected ion signals are
                                                                           shown for the vapor sample taken 12 min into run.
                                                                           PAHs labeled in the TIC trace: 1) naphthalene, 2)
                                                                           methylnaphthalenes,  3)  C2 naphthalenes, 4) fluorene, and
                                                                           5,6) unresolved phenanthrene and anthracene.  Additional
                                                                          compounds indicated as prominent peaks in the selected
                                                                          ion traces are: biphenyl, m/z 154; dibenzofuran, m/z  168;
                                                                          and dibenzothiophene, m/z 184.  Adapted from ref. 4.

-------
  3000 -
o 2000 -
  1000
Toluene (m/z 91)
 O  22May, 715 ms
 V  23May, 715 ms
    24May, 715 ms
    22May, 495 ms
    23May. 495 ms
    24May, 495 ms
                                                                      2000
                                                                      1500
                                                                   o
                                                                   4)
                    100            200

                            Concentration (ppb)
                                                 JOO
                                                               400
                                                                      1000
  Figure 6. Comparison of toluene calibration data from
  715 ms and 495 ms pulse widlh sampling.  These
  standard concentrations were diluted from a 50 ppm in air
  compressed gas mixture.  Note that the spread of
  individual data points comprises three separate days of
  system operation.
                                                                   O
                                                                   o
                                                           c
                                                           o
                                                                        500
Compound, Cone.
 O  12DCE, 30ppb
 V  111TCA, 30ppb
 D  Benzene, 30ppb
 A  Toluene, 30ppb
 •  DCM,  350ppb
 T  12DCE, 350ppb
 •  11 OCA, 350ppb
 A  CCI   350ppb
                                                                             0         500       1000       1500       2000

                                                                                             Sample  Pulse Width (ms)


                                                                              Figure 7. Mean peak area response versus sample pulse
                                                                              width data for one gas mixture at ca.  30 ppb and  another
                                                                              at 350 ppb.
                                                                                                                            2500

-------
    50 H
 c
 I)
 -   0
                        Total Ion Current
 B   o
 c
 o
                     Retention Time (s)
                       Total Ion Current
                                                a)
0
"""*
u
>
^J
o

Bi

26 -



13 -


0 -

	 m/z 61
JYi 	 m/z 65 x 3.3
M. 	 m/z 82 x 2.5 ,
1 M ' \
/' i '' •
/' V ''

                                                   10
                                                b)
/& -
13 :
0 -
	 m/z 61
fl\i ~ ~ m/z 65 x 3.3 >'\
// 1, 	 m/z 82 x 2.5/ \
0 5
1 i
'0
                     Retention Time (s)

Figure 8.  Ion profiles showing the effect of sampling
time on chromatographic resolution  and sensitivity, (a)
was obtained from a 330 ms vapor sample and shows the
separation of dichlorornethane (the first peak)  from the
complex of 1,2 dichloroethene (m/z 61) and 1,1
dichloroethane (m/z 65). Separating these latter two
compounds may be possible with even shorter sample
pulses.  The ion profiles in (b) were obtained at 715 ms
and illustrate the relationship between sample duration
and sensitivity.  The m/z 82 (tetrachloromethane) trace
has a 50% increase in  peak height which correlates to
absolute sensitivity of  the instrument.  Adapted from
reference 2.
 100%
     xlO
                                                                                                    TEDLAR BAG 1-4, 1/5 dilution
                                                                                    vA
Scan No.
Time (s)
100%
               50
               12
 r i' i -1 -1'  i -1
100        150
25         37
                                                     A
                                    ' | '  1 ' 1 ' 1  ' 1 '  ' 1
                                   200        250
                                    50          62
                                          TEDLAR BAG 1-6, 1/5 dilution
 Scan No.
 Time(s)
               i  ' i ' i '  i i | i i  ' i
50         100        150
12          25          37
                                                   I • i • i  • i M"' | i I i I i  i i i i
                                                 200        250
                                                  50          62
                Figure 9. Total ion chromatograms from analysis of 1/5
                dilution Tedlar bag samples from NPL landfill site
                injection wells 1-4 and 1-6.

-------
                                                           DISCUSSION
MAHADEVA SINHA: In all your slides when you lalk  about sensitivity, 1
couldn't get a good feeling. How did you find your detection limit of sensitivity
of 1 ppb or  10 ppb?

WILLI AM McCLENNEN: Basically it is a matter of looking at peak areas that
we're doing quantitation on. Routinely we would compare the peak area of any
peak we could find at the retention time we were looking at to any background
peak in that vicinity or another area of the chromatogram. So the one picture I put
into the paper that will accompany the proceedings of this  will show a 16 ppb
toluene peak with virtually no baseline. There is, I think, three little blips on the
baseline. We estimated that there are signal to noises,  at least the factor of 20 or
more. So basically it's a matter of looking at any discernable peak, any disruption
of the baseline noise in the area of the retention time that we're identifying the
compound.

MAHADEVA SINHA: Toluene works very good because you have almost the
base peak for the entire mass spectrum there. Other peaks are pretty small. But
when you get the mass spectrum itself, do you just identify the peak itself, one
parent peak, or the complete mass spectrum of that?

WILLIAM McCLENNEN: The mass spectrum. Typically we take data with a
full scan, so we can look at all the ions that we would typically expect. And we
can compare them to library spectra or spectra that we know.

MAHADEVA SINHA: What  if I change that compound and go to, let's say.
carbon tet or a dichloroethylene, what becomes your sensitivity at that point?

WILLIAM McCLENNEN: Again, that's where I've hedged a little bit. For
compounds that do not show one nice peak or one nice fragment ion or mo lecu lar
ion we have a slightly lower detection limit. All of our quantitation for the results
I've shown you has been on single ions, not trying to combine several  ions. But
we still have detection limits for the chlorinated compounds that I showed. We
were still looking at standards that were less than 20 ppb and getting good spectra
for them.
                                                                     349

-------
            FIELD MEASUREMENT OF VOLATILE ORGANIC COMPOUNDS
                            BY ION TRAP MASS SPECTROMETRY
              M.E. Cisper, J.E. Alarid, and P.H. Hemberger, Chemical and Laser Sciences Division
                      E.P. Vanderveer, Mechanical and Electronic Engineering Division
                                     Los Alamos National Laboratory
                                        Los Alamos, NM 87545
ABSTRACT
We have developed a second generation transportable
gas chromatograph/ion trap detector (GC/ITD) for the
in-situ characterization of chemical waste sites. This
instrument is extensively based on commercial
instrumentation and can be used for field analysis of
volatile organic compounds (VOCs) in soil and water.
A purge and trap GC is used for sampling and
separation of VOCs from the environmental matrix
before their introduction to the ion trap detector for
mass spectral analysis. A secondary microprocessor
controls the sampling and GC hardware in parallel
with the ion trap detector, which in turn is controlled
by the host PC. The analysis of water samples is
demonstrated by using surrogate samples spiked with
the 24 VOCs contained in the Supelco A & B
Purgeable Standards, acetone, methyl ethyl ketone, and
methyl isobutyl ketone.  Our first transportable
GC/ITD was demonstrated at a chemical waste site for
the analysis of volatile organic compounds in soil [1].
The second generation instrument incorporates
significant improvements in several areas and is nearly
ready for  field deployment. This instrument  has been
extensively characterized in the laboratory. In these
tests, we have found anomalies in quantitation that
might arise during field use. Once these problems -
which may occur with any ion trap based field
instrument - are resolved, the second generation
GC/ITD will be tested and demonstrated in the field.
The second generation transportable GC/ITD will be
described in this manuscript. Some comparisons will
be made to the first generation instrument where
appropriate.

INTRODUCTION
We describe and compare two modular field-
deployable gas chromatograph/ion trap detector
systems for characterization of hazardous waste sites .
Extraction of the analyte from the matrix occurs in the
sampling module. Although the sampling module
could be readily adapted for extraction of volatile
organic compounds from air [2], the focus of this work
has been soil and water analysis. A separations
module, i.e., a gas chromatograph, provides separation
of the extracted VOCs. A Finnigan MAT ion trap
detector (ITD) provides a simple and reliable mass
analyzer on which to base field instrumentation [3].
The ITD serves as a universal detection module. A
turn-key operating system has been developed for this
instrument. This operating  system, which incorporates
additional hardware and software, allows the
instrument to be operated by personnel with minimal
technical background. Because the instrument
operates under nearly full computer control, very little
operator interaction is required. The first generation
instrument (GC/TTD-1) was developed using a purge
and trap gas chromatograph built in our laboratory;
commercially available equipment was used wherever
possible in the second generation transportable
instrument (GC/ITD-2).

EXPERIMENTAL
Ion Trap Detector
The ion trap detector was used  in conventional fashion
and without modification in GC/ITD-2.  An SRI
Instruments 8610 gas chromatograph was connected
directly to the ITD transfer  line. Electron impact
ionization was used with the Finnigan Version 4.10
software with automatic gain control (AGC) [4]. The
Finnigan Programmer's Option Package  was used to
generate FORTH subroutines and keystroke sequences.
The ion trap was tuned using perfluorotributylamine
(FC-43) as a tuning and mass calibration standard with
the automated tuning procedures contained  in the ITD
software.

A schematic diagram of GC/ITD-2 is shown in Figure
1. Automation software consists of the host interface
                                                  351

-------
software, the extended FORTH Finnigan ITD
software, and keystroke sequences (i.e., macros) added
to the ITD software. The SRI Model 8690 purge and
trap device and Model 8610 gas chromatograph are
controlled via the Microstar Laboratories DAP 1200/4
data acquisition processor.  In this fashion, all
components required for the second-generation
transportable GC/ITD  are readily and inexpensively
available.

Although no special considerations were given to
reducing the size, weight, and power requirements of
GC/ITD-2, it easily fits into the back of a 4X4 vehicle
and can be generator powered. GC/ITD-2  is self-
contained in a housing approximately 60 cm on each
side.  This housing contains the gas chromatograph,
ion trap detector, sampling system, and a 6.8-1 cylinder
of ultrapure helium. The mechanical backing pump
for the ion trap vacuum system was mounted next to
the housing.  The instrument weighs ca. 80 kg and
requires less than 1.5 KVA of power.

Modifications of the Finnigan ion trap were required
for use in GC/ITD-1. The conductance-limiting
interelectrode spacers and the open-split interface were
eliminated. A wide-bore capillary column was directly
coupled to the ion trap and the standard 50 1 s'1
turbomolecular pump was replaced with a 240 1 s'1
pump to maintain the ion trap at the proper pressure
with the higher gas load imposed by the direct coupled
column.  A system based on the Hitachi
HD637BO5ZOF microprocessor was designed and
built to control the sampling and chromatographic
hardware.

Surrogates and Samples
Water standards and surrogates were prepared by
adding the Purgeable Mixture A (200 |ig ml'1 of each
of 13 VOCs) and Purgeable Mixture B (200 (Jg ml'1 of
each of 11 VOCs) (Supelco, catalog nos. 4-8851 and 4-
8852, respectively) to HPLC grade water (J.T. Baker).
Aqueous solutions of acetone, 2-butanone  (methyl
ethyl ketone), and 4-methyl-2-pentanone (methyl
isobutyl ketone) were also prepared and added at the
appropriate concentration.  Standards at 1 part-per-
million (ppm), 100 part-per-billion (ppb), 10 ppb, and
1 ppb concentrations were prepared by dilution of a
stock solution containing the A & B Purgeables and
three ketones at the 2 ppm level.  Standard solutions
were refrigerated to prevent evaporation of the VOCs.
Fresh stock solutions and standards were prepared at
approximately 3 day intervals. Fluorobenzene was
used as an internal standard at the 32 ppb level in the
aqueous standards and surrogates. A 10 ppm standard
of fluorobenzene (Aldrich) in reagent grade methanol
was prepared and used as the spike solution. Water,
methanol, and the fluorobenzene/methanol solution
were periodically analyzed with GC/ITD-2 as a check
for impurities.
Purge and Trap Chromatography
For water analysis, a 5 ml aliquot of the standard or
surrogate was loaded into a Pyrex glass tube (12.5 cm
length by 16 mm i.d.). The sample was heated to 80°C
during purging with helium at a flow rate of 60 ml
min'l. Sample recovery under these purge conditions
is shown in Figure 2.  Trap effluent during the purge
cycle is diverted away from the analytical column and
vented.

After the sample purge, the adsorbent trap (50/50
Graphpak GB and Chromosorb W) was ballistically
heated to 375°C from ambient temperature at a rate of
1800°C min'1 and flushed with helium at the carrier
flow rate of 3 ml min'1 (carrier gas velocity 21 cm s'1).
The analytical column is a "VOCOL" fused silica
capillary column (30 m length by 0.53 mm i.d.) with a
3.0 nm film (Supelco, cat no. 2-5320M). The oven
temperature for the chromatography column was held
at 30°C during the purge cycle. Following heating of
the adsorbent trap, the oven temperature was ramped  at
20°C mhr1 to 40°C, held for 1 minute, then ramped at
4°C min to a final temperature of 150°C. The
adsorbent trap is continually flushed with helium
during the chromatographic run; the chromatographic
oven is heated to 250°C for system cleanout after data
acquisition.

RESULTS
A list of the volatile organic compounds used in these
studies (the A & B purgeables and ketones) and the
reconstructed ion chromatogram (RIC) from the purge
and trap analysis of 5 ml of water containing  100 ppb
of these 27 compounds is shown in Figure 3.  Working
calibration curves from 1 ppb to 100 ppb of 1,1,2-
trichloroethane in 5 ml of water are shown in Figure 4.
The value of the average slope of these three lines is
0.82 +/- 0.02 and the correlation coefficient for these
data is 1.00. This calibration curve was obtained using
the Finnigan Automatic Gain Control (AGC) software.
The use of the automatic gain control does not in itself
ensure that the ion trap will not operate under space-
charge conditions. At higher sample concentrations
(microgram levels) the AGC should be manually tuned
to  prevent mass spectral degradation due to space-
charging.  Although the calibration curve for 1,1,2-
trichloroethane is quite good, calibration curves for
other compounds can exhibit non-linearity, especially
at lower concentration levels.  For these compounds,
relative sensitivity factors will vary as a function of
concentration as shown in Figure 5. This non-linearity
can be partly explained by the occurrence of
ion/molecule reactions of analyte molecules with ions
from water (H2O+, H3O+) and methanol (CH3O+,
CH3OH+, and others). In the AGC mode, the Finnigan
ion trap detector is pre-programmed (in the
instrument's firmware) to store all ions above m/z 20
(ionization DAC value of 125). However, water ions
below m/z 20 can remain in the ion trap during and
after ionization and are able to react with analyte ions
                                                     352

-------
during the analysis sequence.  A plot of relative
intensity of HaO* ions (m/z 19) as a function of the
mass storage level during ionization is shown in Figure
6. These data were obtained with a Finnigan Ion Trap
Mass Spectrometer (ITMS), which does allow the
operator to select the storage level (DAC value) during
ionization. With the Finnigan ITD, the storage level
during ionization is fixed in the AGC mode (i.e., it is
not user adjustable) and it is likely that the non-
linearities displayed in Figure 5 will be observed with
any purge and trap GC/ITD system.  It is also possible
for analyte ions to react with neutral water and
methanol molecules. The creation of protonated water
molecules, HjO"1", via reactions of analyte ions (shown
in the upper cnromatogram) with water background in
the ion trap detector, is demonstrated in Figure 7. We
estimate water loss during our purge cycle to be about
1-2 mg mur1. Moisture may either adsorb or simply
condense on interior surfaces of the sampling system
and subsequently be introduced to the ion trap during
trap desorption.  It is important to remember here that
nanogram levels of analyte are being measured and
that seemingly insignificant amounts of moisture in the
ion trap (for example, water introduced by venting the
instrument to transport it from one area to another) can
affect the reliability of quantitation.  An effective
solution to this problem will require modification of
the ITD. We are presently developing methodologies
to reduce the ion/molecule chemistry that occurs in
GC/ITD-2.  These new techniques should improve
response linearity for most of the VOCs in this study.

The chromatographic retention times obtained with
GC/ITD-2 are very reproducible. Table 1 shows the
precision of retention times for several compounds at
concentration levels from 1 pbb to 100 ppb. We have
found that changes in retention time are often
accompanied by changes in mass spectral peak
distribution and intensity. However, because of the
excellent retention time reproducibility shown here,
retention time measurement of the internal standard
(fluorobenzene) can be a readily observed metric of
performance during field analysis and provides a real-
time check on instrument performance. A retention
time control chart is shown in Figure 8.  The three out-
of-control points occurred following  scheduled power
outages in our laboratory.

Compounds are identified on the basis of their
chromatographic retention time and their
experimentally obtained mass spectrum via an
automated identification  routine.  Mass spectral library
matches are  highly accurate on a "first-hit" basis for
nearly all compounds in the A & B Purgeable and
ketone mixture with GC/ITD-2. Exceptions are
toluene and those compounds that coelute from the
chromatograph (1,1-dichloroethylene and methylene
chloride; benzene and 1,2-dichloroethane). Coeluting
compounds can often be identified with some mass
spectral interpretation. At this time, this capability is
not programmed into our automated library search
routines and, in fact, such interpretation might be
better left to those reviewing the data after die field
work is completed. Table 2 shows the accuracy of
library matches with GC/ITD-2. The "ID Hit" is the
rank of the correct compound identification in a list of
the 10 most probable compounds selected by the
library search routine.  The purity value provides an
indication of how closely related the sample mass
spectrum is to the library spectrum on a scale of 0 -
1000. A purity of 800 or higher implies a very good
match between the two.  Even though some of die
identifications shown in Table 2 do not have high
purity values, the purity values obtained so far have
been quite reproducible.  This reproducibility provides
a means to screen library search data for consistency.
For example, c/s-l,3-dichloropropene is identified on a
"second-hit" with a purity value of about 400. If an
analysis of an unknown mixture were to identify a
compound as cw-l.S-dichloropropene with a
significantly higher purity value, say 800, that
identification might be suspect even though the purity
value suggests otherwise. In general, we have found
that the number of "first-hit" identified compounds is
higher with GC/ITD-2 than  with GC/ITD-1. This is
most likely due to better regulation of the helium
buffer gas partial pressure in GC/ITD-2 where the
conductance limiting spacers are used and the ion trap
is coupled to the chromatography column via the open-
split interface.

DISCUSSION
The design of this transportable  instrument, coupled
with the flexible control  system  provided by the
Microstar Laboratories DAP1200 data acquisition
processor, is well suited  to problems in field analysis.
The ability to address and control ancillary
instruments, such as sampling devices,  via the host
computer provides great flexibility for different
analytical problems. Keystroke  sequences can provide
customized data reduction procedures for different
applications.

The combination of the data acquisition processor,
Forth interface software, and Forth keystroke
sequences added to the ITD software allow the
instrument to be  operated in a turnkey fashion.  That is,
in the intended mode of operation for the transportable
GC/ITD, the operator only needs to select the
appropriate operational procedures from the menu.
Figure 9 shows the menu by which the instrument is
controlled. Once selected, each  item proceeds
automatically. For example, one can choose to
calibrate the instrument ("Trap Setup")  or to generate
an analytical report ("Quantitation"). If "Acquisition"
is selected from the menu, a sub-menu is displayed to
provide optional analysis procedures, e.g., to analyze
for a suite of VOCs in  soil or to monitor a single
compound in groundwater.

At this time, an analysis with GC/ITD-2 requires about
60 minutes. We have recently purchased the SRI
                                                   353

-------
Model 8680 Purge and Trap Device; that device should
allow us to significantly reduce the analysis time
required by GC/TTD-2.  Detection limits with GC/TTD-
2 have not been determined. These limits cannot be
reliably estimated until the quantitation anomalies
described above are resolved. The response of
GC/ITD-2 to many compounds at the 1 ppb level is
quite good and we expect that detection limits will be
below the part-per-billion level for laboratory
standards.  However, it is quite likely that the ultimate
detection limits for the instrument will be limited by
the nature of the sample in the field.

The gas chromatograph/ion trap detector (GC/ITD)
configuration described here has other advantages over
commercially available transportable gas
chromatograph/quadrupole mass spectrometers [5].
Ion trap detectors are inherently simple. The ion trap
does not suffer from the complexity added by external
ion sources and ion lenses. The ion trap electrode
assembly is small and rugged although the radio-
frequency and power transformers, gas supplies, data
system, and other equipment can offset the size
advantage.  When operated as an ion trap detector (that
is, as a simple single-stage mass analyzer), no direct
current (dc) potentials are applied to the trap
electrodes.  Without dc potentials applied to the trap
electrodes, charging phenomena are minimized as the
electrodes become dirty. Another advantage derives
from the inherently high sensitivity of the ion trap
detector. The ITD is roughly 10 to 100 times more
sensitive than a conventional transmission quadrupole
[6]. This sensitivity advantage can be pushed  even
further with the addition of axial modulation to the ion
trap [7]. Finally, the ion trap can provide mass
spectrometry/mass spectrometry capabilities (MS/MS)
[8] far more simply than tandem quadrupole
instruments, which require at least two independent
quadrupole mass filters [9].  The advantages added by
axial modulation and MS/MS capabilities do however
require more expensive versions of the Finnigan ion
trap than the Ion Trap Detector used in this work; these
are the Finnigan ITS-40 Ion Trap System and  the Ion
Trap Mass Spectrometer (ITMS).

The configuration of GC/ITD-2 has advantages in
several areas over the configuration of GC/TTD-1.  The
first advantage is a cost savings. Savings are realized
through the use of commercially available equipment
The time and costs associated with the development,
construction, and trouble-shooting of chromatographic
hardware and the custom interface were eliminated.
GC/ITD-1 also required an expensive turbomolecular
pump and mechanical modification of the Finnigan
vacuum manifold.  The second advantage is reliability.
The gas load imposed by the removal of the
conductance limiting spacers and the high flow rates in
GC/ITD-1 resulted in reduced lifetimes for the
ionizing filaments and the electron multiplier.  The
DAP 1200 processor also appears to be more
compatible with the Finnigan operating system than
the custom interface and computer "hang-ups" have
been eliminated.  The third advantage is the ease with
which future instruments may be built. The
construction of GC/ITD-2 by other laboratories should
be quite straightforward due to the extensive use of
commercially available equipment. In general, copies
of instruments that use custom designed and built
hardware are more difficult to build with the same
performance of the original instrument vis-a-vis
instruments that are based on proven "off-the-shelf
hardware. Finally, the Hitachi processor used in
GC/TTD-1 is no longer available and therefore not
suitable for new designs.

CONCLUSION
Work is currently in progress to fully characterize the
second generation instrument in the area of
quantitation accuracy.  Once quantitation methods are
developed and tested in the laboratory, detection limits
for a wide number of compounds of environmental
significance can be determined. Analysis time will be
reduced. Tests with soil samples will be performed.
We do  not anticipate significant problems in the purge
and trap analysis of soils since the SRI Purge and Trap
devices used in this instrument have been developed
for both soil and water samples. New containment for
GC/ITD-2 is being designed; the present containment
is designed to minimize instrument footprint but it
does make minor maintenance operations cumbersome
to carry out.

The ion trap detector provides many advantages as a
mass analyzer in this application.  It is simple to
maintain and operate. The high sensitivity of the ion
trap and the inherent universality of the modular mass
spectrometer system are perhaps the most important
features for a field analytical instrument. The
instrument provides high specificity for compound
identification due to the two-dimensional information
provided by chromatographic retention time and mass
spectral library identification. Mobile ion trap mass
spectrometers operating in the MS/MS mode have
been successfully applied for the direct, continuous or
near-continuous analysis of permanent gases and
condensable vapors  [10].  Ion trap mass spectrometer
systems have also been developed for rapid screening
of volatile organics in environmental matrices by
MS/MS techniques [11].  Such instruments can
provide highly complementary information to that
obtained from a transportable GC/ITD system. Both
types of instruments will find widespread use in
environmental restoration activities. Other instruments
used in field applications, such as a gas chromatograph
[12] or Fourier-transform infrared spectrometer [13],
do not provide the combination of sensitivity,
specificity, and universality demonstrated by the
transportable ion trap instrument.
                                                     354

-------
REFERENCES
1.    Hemberger, P.H., Aland, J.E., Cameron, D.,
      Leibman, C.P., Cannon, T.M., Wolf, M.A., and
      Kaiser, R.E., Int J. Mass Spectrom. Ion
      Processes., in press.

2.    Cameron, D., Alarid, I.E., Hemberger, P.H., and
      Leibman, C.P., "Real-Time Analysis of
      Organic Pollutants by Ion Trap Mass
      Spectrometry," Ion Trapping in Mass
      Spectrometry Conference, American Society for
      Mass Spectrometry. Sanibel Island, FL,
      January, 1990.

3.    Stafford, G.C., Kelley, P.E., Syka, J.E.P.,
      Reynolds, W.E.,and Tood, J.F.J., Int. J. Mass
      Spectrom. Ion Processes, 60,1984,85.

4.    Yost, R.A., McClennan, W., and Snyder, A.P.,
      "Picogram to Microgram Analysis by Ion Trap
      Mass Spectrometry," Proc. 35th Ann. Conf.
      Mass Spectrom. Allied Topics, Denver, CO,
      1987, p. 789.

5.    Cannon, D.R., Ind. Chem. News, 7,1986,1.
      Specific instruments include:
      Mobile Environmental Monitor (MEM) Broker
      Instruments, Manning Park, Billerica, MA.
      PETRA, VG Gas Analysis Systems, Aston
      Way, Chesire, England.
      TAGA, Sciex, Inc., Thornhill, Ontario, Canada.
      SpectraTrak, Viking Instrument Corp., Reston,
      VA.

6.    Nourse, B.D., and Cooks, R.G.,  Anal. Chim.
      Acta,228, 1990,1.

7.    Weber-Grabau, M., Kelley, P.E., Bradshaw,
      S.C., and Hoekman, D.J., "Advances in MS/MS
      Analysis with the Ion Trap Mass Spectrometer,"
      Proc. 36th Ann. Conf. Mass Spectrom. Allied
      Topics, San Francisco, CA, 1988, p. 1106

8.    Louris, J.N., Cooks, R.G., Syka, J.E.P., Kelley,
      P.E., Stafford, G.C., and Todd, J.F.J.,  "New
      Advances in the Operation of the Ion Trap Mass
      Spectrometer," Proc. 33rd Ann. Conf. Mass
      Spectrom. Allied Topics, San Diego, CA, 1985,
      p. 707.

9.    Yost, R.A. and Enke, C.G., Anal. Chem., 51,
      1979,69.

10.    McClennan, W.H., Arnold, N.S., Sheya, S.A.,
      Lighty, J.S., Meuzelaar, H.L.C., "Direct
      Transfer Line GC/MS Analyses of Incomplete
      Combustion Products from the Incineration of
      Medical Wastes and the Thermal Treatment of
      Contaminated Soils," Proc. 38th Ann. Conf.
      Mass Spectrom. Allied Topics, Tucson, AZ,
      1990, p. 611.

11.   Wise, M.B., Buchanan, M.V., Guerin, M.R.,
      "Rapid Environmental Analysis by Direct
      Sampling Glow Discharge Mass Spectrometry
      and Ion Trap Mass Spectrometry," Oak Ridge
      National Laboratory TM-11538, Oak Ridge,
      TN, 1990.

12.   Blanchard, R.D. and Hardy, J.K.,  Anal. Chem.,
      58,1986,1529.

13.   Spartz,  M.L., Witkowski, M.R., Fately, J.H.,
      Jarvis, J.M., White, J.S., Paukstelis, J.V.,
      Hammker, R.M., Fately, W.G., Carter, R.E.,
      Thoma, M., Lane, D.D.,  Marotz, G.A., Fairless,
      B.J., Holloway, T,. Hudson, J.L., and Gurka,
      D.F., Amer. Environ. Lab., 1, 189, 15.

ACKNOWLEDGEMENTS

P.H.H. wishes to acknowledge the efforts and
contributions of C.P. Leibman in the development,
testing, and demonstration of the first generation
transportable GC/ITD. P.H.H. also thanks Dale Spall
for his invaluable technical contributions, support, and
encouragement throughout this  project.

This work has been financially supported through the
Department of Energy - Office of Technology
Development.
                                                 355

-------
Compound

Methyl Ethyl Ketone
1,1,1 -Trichloroethane
Carbon Tetrachloride
Benzene
Fluorobenzene
Trichloroethylene
1,2-Dichloropropene
Bromodichlorome thane
Toluene
1,1,2-Trichloroethane
Tetrachloroethylene
Dibromochloromethane
Chlorobenzene
                                           Table 1.

                             Reproducibility of Retention Times
  tR*
(1 ppb)

  23.3
  25.0
  25.7
  26.4
  27.1
  28.2
  28.9
  29.6
  32.4
  33.6
  34.4
  35.2
  37.2
(lo'ppb)

  23.5
  25.0
  25.8
  26.4
  27.1
  28.3
  28.9
  29.6
  32.4
  33.6
  34.3
  35.2
  37.2
(lOlfppb)

  23.4
  25.0
  25.7
  26.4
  27.1
  28.2
  28.9
  29.6
  32.3
  33.5
  34.4
  35.2
  37.2
                                                               Average        %rsd
                                    23.4
                                    25.0
                                    25.7
                                    26.4
                                    27.1
                                    28.2
                                    28.9
                                    29.6
                                    32.3
                                    33.5
                                    34.3
                                    35.2
                                    37.3
0.3%
0.07%
0.1%
0.1%
0.1%
0.2%
0.1%
0.1%
0.1%
0.1%
0.2%
0.2%
0.2%
* Retention times in minutes. The start of the purge cycle is time t = 0.
** Average of 9 runs from 1 ppb to 100 ppb.
             Compound
              Table 2.

Reproducibility of Library Matching

         ID Hit        Purity*        %rsd
             Methyl Ethyl Ketone         1            931
             1,1,1-Trichloroethane        1            599
             Benzene**                  1            759
             Trichloroethylene           1            801
             Bromodichloromethane      1            861
             c-l,3-Dichloropropene       2            413
             Toluene
             Tetrachloroethylene         1            657
             Chlorobenzene              1            826
             Bromoform                1            856

             * Average of 3 analyses at 100 ppb.
             ** Coelutes with 1,2-dichloroethane.
                                     5%
                                     7%
                                     2%
                                     2%
                                     5%
                                          356

-------
                   Figure 1. A Schematic Diagram of the Second Generation
                   Transportable Gas Chromatograph/Ion Trap Detector.
  SRI 8690  or 8680
Purge & Trap Device
  SRI 8610 Gas
 Chromatograph
     Finnigan
Ion  Trap  Detector
Helium Supply |
    Microstar
Data Acquisition
    Processor

-------
8!
CD
                o
               <
                CO
               6
                   100 A
                                               Figure 2.
                                 Sample Recovery for 100 ppb of
                                       VOC / Ketone Mixture
                                                  O Methyl Ethyl Ketone
                                                  • Methyl Isobutyl Ketone
                                                  A A & B Purgeables
                                          Analysis Number
                                                                     Los Alamos
                                                                      CLS-91-630

-------
                           Figure 3. Reconstructed Ion Chromatogram from the Purge and Trap
                           Analysis of 5 ml of Water Containing 100 ppb each of 28 Volatile
                           Organic Compounds.
                                       11,12
T01-
             2   3-4
             -   A.
               I
       fl
  13:21
1280
   I
 1609
26:41           33:21
    Scan Number
Retention Time (min)
 2400
40:01
                          1. Acetone (not observed)
                          2. Trichlorofluoromethane
                          3.1,1-Dichloroethylene
                          4. Methylene Chloride
                          5. t-12-Dichloroethylene
                          6. 1,1-Dichloroethane
                          7. Methyl Ethyl Ketone
                          8. Chloroform
                          9.1,1,1-Trichloroethane
                          10. Carbon Tetrachloride
                          11. Benzene
                          12.1,2-Dichloroethane
                          13. Fluorobenzene
                          14. Trichloroethylene
                          15.1,2-Dichloropropene
                          16. Bromodichloromethane
                          17.2-Chloroethyl Vinyl Ether
                          18. Methyl Isobutyl Ketone
                          19. c-13-Dichloropropene
                          20. Toluene
                          21. r-1,2-Dichloropropene
                          22.1,1,2-Trichloroethane
                          23. Tetrachloroethylene
                          24. Dibromochloromethane
                          25. Chlorobenzene
                          26. Ethyl Benzene
                          27. Bromoform
                          28.1,1,2,2-Tetrachloroethane

-------
                         Figure 4.
    Calibration Curve for 1,1, 2-Trichloroethane
         O Dec 90
         D Dec 90
         A Dec 90
1E-2
    1                        10                       100
     Concentration (parts-per-billion) in a 5 ml Water Sample
                                                  Los Alamos
                                                     CLS-91-578

-------
                                   Figure 5.


            Sensitivity Factor Relative to Fluorpbenzene

                   as a Function of Concentration

                32 ppb Fluorobenzene in 5 ml VOC / Water Standard
    10.0
o
W
c
o
(0
o


I

-------
                               Figure 6.
V)

4)
DC
              Relative Intensity of M/Z 19 as Function
                      of lonization DAC Value
100
90
80
70
60
50
40
30
20
10
0
12
2(


I
ft










Puncorr =
A






^





r

•





>
•





>
•





I
1





^





_..
8x10'8torr ~





d





«
—
—
—
—
—
~








'5 135 145 155 165 175 185 195 205 215 225
3 23 26 29 33 36
                             DAC Value
                               Mass
LOS ALAMOS
  CLS-91-632

-------
                                Figure 7.


              RECONSTRUCTED ION CHROMATOGRAMS FOR
                    A & B PURGEABLES AND KETONES
100%-
 IT)
 ID
 OJ
 i
        Methyl Ethyl
        Ketone
2-Chloroethyl
Vinyl Ether
Methyl Isobulyl Ketone


  1.1,2-Trichloroethane
15% -,
 O)

 N

 E

                                        fromXH+%H2O-> X'+H3Of
                I
   1400       1600
     I
   1800
   2000
2200
2400
                             SCAN NUMBER
                                                             LOS ALAMOS
                                                                CLS-91-628

-------
             Figure 8.
Retention Time Control Chart for
     Dibromochloromethane
VSL.
+99%
+95%


^f
-95%
-99%
I I I I I I I I I I I


	 0 	
o o o o
0
J 0^ -0 f—
— — — — — — — — — — — — —
— — — — — — — — — — — — —
ocan
2116
2114


2110
2106
2104
1234 56789 10 11
Run
                                   Los Alamos
                                      CLS-91-580

-------
                                                         Figure 9.

                                                Instrument Control Menu
                                                SELECTION MENU
                                 1) Trap Setup
                                 2) Bakeout
                                 3) Acquisition
                                 4) Trap cool down
                                 5) DAP reset
                                 6) Quantitation
     (SETUP)
     (BAKEOUT)
     (ACQMENU)
     (TRAPCOOL)
     (DAPCLR)
     (QUANT)
                                 9) Quit Menu

                                 ENTER SELECTION NUMBER:


                                                    DISCUSSION
RUSSELL SLOBODA: Once you finished prototype work, what was the
average length of time when you turned off the motor and the vehicle driving the
trailer off the site and when your calibration could actually be done?

CHRISTOPHER LEIBMAN: Start again you mean? Actually we have a fairly
good size  turbo pump,  the 240 liter/second  turbo pump. That  repre-
sents one of the modifications. It pumps out fairly quickly and we're back in
business within a day. We drive up to the site at night, and let the unit pump down;
that's been our practice. Clearly that's something we're going to have to look at
very hard.

RUSSELL SLOBODA: Do you need power throughout the night for one day's
events?

CHRISTOPHER LEIBMAN: The way it is currently configured, yes.

RUSSELL SLOBODA: Can you peel off the back of the system and then shut
things off and then turn it on the next day?

CHRISTOPHER LEIBMAN: We've considered that. But the power require-
ments when the instrument is in the standby mode just aren't considerable. The
GC's off and the turbo's up to speed, so there's very little load. The one thing I
would like to also add is, I have this demonstration on a diskette. I suspect it will
run a little more smoothly on your instrument than in this form. But if you would
like to see any data after this presentation, write and let me know.

BRIAN ECKENRODE: I was wondering about your success on the library
searching. For example, I noticed you had one spectrum, tetrachloroethylene,
that had peaks beyond the molecular ion. How is that affecting your ability to get
a hit in the library?
CHRISTOPHER LEIBMAN: We've done quite well in terms of the library
searching, the library matching. We're looking both based on retention time and
the mass spectra. So while it may see those other ions in the mass spectrum, it's
only looking for the presence of certain target ions. Operated in that mode, the
coelution does not pose a problem.

PHIL HEMBERGER: You had mentioned going to a hydrogen generator. Is
that also to supply buffer gas to the ion trap?

CHRISTOPHER LEIBMAN: Yes. And in fact some interesting work has been
done in that area. Scott McClucky and Gary Clish and O'Krish have looked at
the presence of hydrogen in the ion trap, and what they've observed is an
enhancement in sensitivity by a factor of two.

PHIL HEMBERGER: With hydrogen you would expect much different
collision of cooling of the ions then would  be provided by helium. The
fragmentation patterns should also change. Do you anticipate that you're going
to have to build your own library?

CHRISTOPHER LEIBMAN: I think you need to look at the search algorithm
that we're using. The library we're using to conduct the search is one we create.
So when we go through the chromatogram and identify each of the peaks, we
essentially then identify that fragmentation pattern. So if it changes with different
collision gases so be it. It's taken into account in the way we set upour calibration
files. But, a very good point.
                                                             365

-------
                  TRANSPORTABLE GC/ION TRAP MASS SPECTROMETRY FOR TRACE FIELD
                                      ANALYSIS OF ORGANIC COMPOUNDS
                 Chris P. Leibman. David Dogruel, Health and Environmental Chemistry Group, HSE-9,
                 Eric P. Vanderveer, Instrumentation Group, MEE-3 Los Alamos National Laboratory,
                 M/S K-484, Los Alamos, New Mexico, 87545
Abstract

A transportable purge and trap/GC/MS based on the Finnigan
Ion Trap Detector (ITD) has been developed at Los Alamos
National Laboratory for the identification and quantification
of volatile organic compounds present at chemical waste sites.
This instrumentation is being evaluated for use to  support
environmental surveillance and the characterization/clean-up
of hazardous  waste sites.  A custom  purge  and trap/GC
sampling system  was  integrated  with  a modified  ITD  to
achieve instrument operation consistent with field activities.
The   sampling  system  is   controlled  by   an ancillary
microprocessor designed at Los Alamos  National Laboratory.
The instrument is extensively automated and can be operated
with minimal  training.  Instrument operation transparent to
the field user has been achieved by integrating sampling
system control software with the operating software of the
ITD.

The instrumentation and associated methods parallel those
outlined  in   method   8260,   SW-846.    Qualitative  and
quantitative analysis for the  68  target compounds and the
associated internal standards and surrogates is completed in an
automated sequence that is executed every 25 minutes. Sample
purging, analysis,  data reduction, and preliminary report
generation  proceeds automatically.  The instrument can be
operated in a continuous mode, pausing only  for sample
loading and data file specification. All data are archived on
floppy disk for subsequent review by a skilled analyst.

Part-per-trillion detection limits can be attained for many
compounds from either 5  gram  soil or 5 milliliter water
samples.
Introduction

The  development and  use  of  field transportable analytical
instrumentation can significantly reduce the cost associated
with environmental surveillance and restoration activities.
Field analytical  support  minimizes  the  analytical  data
turnaround time, which can expedite site characterization and
provide analytical data  to  field personnel  for guidance  of
ongoing  work.   Clean-up personnel can  be  used  more
efficiently since these teams will not have to be released and
reassembled weeks later after receipt of analytical results from
a remote laboratory.
Field analytical support can directly impact the  expense of
environmental clean-up by reducing  the cost-per-analysis.
Cost  for  sample  packaging,  shipment,   receiving  and
management are eliminated if analyses are performed on site.
Field analytical support improves the chances that schedules
and monetary constraints associated with remedial activities
are met.

Performing  analyses  on-site  can  enhance  the  quality of
analytical data generated. Field analyses reduce the possibility
that  samples  will  be  compromised  from  transport  and
handling.  Reduced sample  handling and  the  analysis of
samples within  minutes of collection minimizes the potential
loss of volatile components.  Near real-time data  can also be
used to direct  subsequent sampling  efforts.  Additionally,
initial site characterization can help  delineate the sampling
grid  used for collection of samples to be sent to a remote
laboratory.

A transportable purge and trap/GC/MS has been developed at
Los Alamos National Laboratory to provide field analytical
support  for  environmental  restoration activities.    The
instrument is based on a Finnigan Ion Trap  Detector (ITD)1,
a rugged and simple mass spectrometer.  This transportable
GC/ITD has been designed  specifically to support field
operations and to provide analytical data of sufficient quality
to meet higher  level data quality objectives.  Our focus has
been to attempt to meet the quality control criteria outlined in
chapter 1 of SW-846 and to use procedures which  parallel
method 8260, SW-846.
Experimental

Purge and Trap/Gas Chromatograph

A custom purge and  trap/GC was fabricated for sampling
volatile organics in water or soil samples.  The purge and
trap/GC has two sampling loops, each loop consisting of a
needle sparger and an adsorbent resin trap.  A schematic of
this sampling system is shown in Figure 1.   In position  A,
simultaneous with  the  purging  and  concentration of one
sample onto trap B the contents of trap A are desorbed onto
the capillary column. Subsequent to the analysis of trap A, the
ten port valve (Valco Instruments Co.) is rotated to position B
                                                         367

-------
and the contents of trap B are desorbed onto the capillary
while purging/concentration occurs on  the other sampling
loop.  Backpressure regulation via a split maintains  column
carrier gas flow.  Splitless  injection  is  performed for 20
seconds during adsorbent trap desorption.  This is sufficient
for quantitative  transfer of trapped  target  analytes while
serving to minimize water transfer to  the analytical  system.
Additionally, the capillary column is maintained at 10 °C
during desorption. This serves to cryofocus target analytes
onto the head of the column while allowing any water to pass
unretained.

The adsorbent traps are packed with equal amounts  of  2,6-
diphenylene polymer and silica gel. Traps are heated to 200°C
at a rate of 500°C/min. Heater jackets are provided for the 5
milliliter sparger tubes and maintain a purge temperature of 35
°C.   The  temperature  programmable  GC  oven   can be
programmed with up  to 35 multiple ramps and is capable of
sub-ambient operation.   All sample  transfer  lines  are
deactivated  fused silica and are heated to 85 °C.  All valves,
heaters and the GC oven associated with this sampling system
are controlled by a  dedicated microprocessor.  A 30m x
0.32mm i.d. DB-624  (J&W Scientific) fused silica capillary
column with 1  um film thickness was used.  The capillary
column was directly coupled to the ITD via a heated  transfer
line.

In addition to soil or water sampling, soil gas analysis can be
accomplished  by replacing the  needle  spargers  with an
adsorbent trap thermal desorption unit. Soil gas or air samples
are collected on an adsorbent trap using an air sampling pump.
The air sampling tubes are then transported to  the instrument
and  placed  in a  heater assembly  whereby trap contents are
thermally desorbed onto the primary adsorbent traps shown in
Figure 1.  Conversion  from soil/water  analysis to air  analysis
can be accomplished within 5 minutes. Instrument operation
is modified via the computer to accommodate the air sampling
trap desorption/analysis.

Mass Spectrometer

A Finnigan MAT Ion Trap Detector1  was used with the
following modification. The supplied  transfer line and open
split interface were eliminated. The 50 L/sec turbomolecular
pump  was replaced with a 240 L/sec  turbomolecular pump.
The larger pump was required to handle the increased  gas load
realized from the direct coupling  of the capillary column to
the ITD.  The larger turbomolecular pump also reduces pump
down time following system venting.
Electron impact ionization was used; the ionization period was
regulated  using  Finnigan supplied automatic gain  control
software.

Data System/Automation

A Zenith Supersport 286 laptop computer was used for data
acquisition  and  instrument control.   All aspects  of  mass
spectrometer and sampling system operation were controlled
through the dedicated laptop computer.  Finnigan supplied
ITD control software  (version 4.10) with the programmer's
option served as the platform for system automation. FORTH
subroutines  and keystroke sequences were incorporated  with
the  Finnigan supplied  software to automate  ITD  data
acquisition,   quantitation,   and   report  generation.
Communication to the sampling system  microprocessor was
through the serial port of the  laptop computer.  Sampling
system  control  was   achieved  using  assembly  language
programs. Parameters for sampling system event sequencing
and heater or GC oven temperatures  were written into the
Finnigan ITD software using the programming option.

Physical Requirements

The total instrument dimensions are 17.5" x 23.5" x 26" ( H x
W x D ) exclusive of the laptop computer. The instrument can
be deployed in a vehicle equipped with compressed gas supply
and a small liquid nitrogen dewar if cryogenic operation is
required.  A  portable  generator or  line power is required.
Power consumption is less than 1.5 kW. For field test to date,
the instrument has been deployed in a  12 foot trailer.
Methods/Operational Sequence

The  methods  used  with  the  transportable  purge  and
trap/GC/ITD parallel those outlined in method 8260, SW-8462.
Following instrument pump down and warm up of all heated
zones, filament continuity and water concentration in the ITD
are  checked.     Mass   calibration  is   verified   using
perfluorotributylamine (PFTBA).  Depending on the end use
of field generated data, different  levels of quality control can
be used.  An ITD tuning  check can be performed  using 4-
bromofluorobenzene (BFB) to ensure  that abundance criteria
recently specified in method 524.2 is met. Figure 2 shows ion
chromatogram  derived  from the   molecular  ion of 4-
bromofluorbenzene obtained by purging a solution containing
50 ng 4-bromofluorobenzene. The mass spectrum obtained at
the scan indicated by the cursor shown in Figure 2,  is shown
in Figure 3. This mass spectrum meets the abundance criteria
specified  in   method  8260,  SW-846.   Following   tune
verification,  a calibration  curve  can be  established or
continued adherence to the calibration curve  can be checked
using a midpoint standard. The midpoint standard check can
also  be  used to  update  retention  times on a daily  basis.
Currently our target list comprises the  68 target compounds
shown  (with their  corresponding internal  standards and
surrogates) in Table  1. Following analysis of a blank, sample
analysis is performed in a continuous  mode, pausing only for
sample loading and data file specification. Data acquisition is
followed  automatically by data reduction. Target compounds
are identified by 1) elution of sample component at the
appropriate elution window and 2) comparison of the sample
mass spectrum with the standard reference  mass spectrum.
Standard  reference mass spectra are obtained from the analysis
of calibration mixtures.   If any targeted compounds are
detected,  a  hardcopy  preliminary  report  is  generated
immediately.   All data  are  archived on machine  readable
media for subsequent review by a skilled analyst.

Results/Discussion

We have  successfully deployed the transportable GC/ITD at
waste sites at Los Alamos National Laboratory. To date field
trials have been  completed without  significant  instrument
failures.  The qualitative and quantitative analysis for the 68
target compounds and the associated internal standards and
                                                           368

-------
surrogates is accomplished  in  the  field in an  automated
sequence executed every 25 minutes. A portion of the total
ion chromatogram obtained in the field from 5 mis of a 50 ppb
water standard is shown in Figure 4. Retention times for the
target compounds reflected in Figure 4 are given in Table  1.
Chromatographic development is completed within 16 minutes.
The need to obtain the Chromatographic resolution displayed
in Figure  4 is dependent on the site specific data quality
objectives established.  If only screening is required or the
target list is more limited, Chromatographic development can
be reduced by using a steeper GC oven temperature ramp for
faster elution. An example of instrument operation in the fast
screening  mode is shown  in   Figure  5,  representing  a
chromatogram obtained from the internal standards/surrogates
spiking mixture used in this work.
During  field  trials, co-located  samples  were  taken for
comparisons  between the  transportable   GC/ITD  and  a
laboratory based GC/quadrupole mass spectrometer. Table 2.
shows the comparison between an analysis performed at a
waste site with the GC/ITD and  results obtained from the
remote laboratory.  The results shown in Table 2. reflect a 1 to
100  dilution  (high level-methanol   extraction   method)2.
Difference between the field results to those obtained at the
remote laboratory may reflect the loss of volatile components
during sample transport.
Low part-per-trillion detection limits have been achieved for
some compounds from 5g soil or 5 mis water samples with this
instrumentation. An extracted ion current profile of m/z 98,
the quantitation ion of toluene d8 (a surrogate), obtained from
a 20 ppt solution of the compounds listed in Table  1. (100
picograms/component in 5mls) is shown in Figure 6.  The
signal  to noise ratio  for m/z 98 in this chromatogram  is
approximately 10:1.   The complete background subtracted
mass spectrum for toluene d8 is shown in figure 7. No toluene
d8  was  detected  in  the  blank  which  preceded the ion
chromatogram shown in figure  6.   Low part-per-trillion
detection limits  cannot  be  routinely achieved in the field.
However, the high sensitivity  of the  instrument increases the
degree  of   confidence  in the  automated   mass spectral
identifications performed in the  field for a higher working
concentration  range.  Our targeted  working concentration
range for field studies is from 100 ppt to 100 ppb for most
compounds.
ensure data quality.   A high  degree  of specificity  for
compound identification is achieved with retention time and
mass spectral information. The instrument is fast, analysis for
the 68 target compounds outlined in method 8260, SW-8462
can  be  achieved  in  25 minutes.   For screening,  the
Chromatographic performance can  be reduced  to  reduce
analysis time. Additionally, part-per-trillion detection limits
have  been demonstrated for  many compounds with this
instrument.
References

1) Stafford, G.C., Kelley, P.E., Syka, J.E.P., Reynolds, W.E.,
and  Todd, J.F.J. "Recent Improvements in  and Analytical
Applications of Advanced Ion Trap Technology " Int. J. Mass
Spectrom. Ion Processes, 60, Special Issue, Sept., 1984, 85.

2) "Test Method for Evaluating Solid Waste Physical/Chemical
Methods, SW-846, Third Ed., Update I, Method 8260. Office
of Solid Waste and Emergency Response, U.S. Environmental
Protection Agency, Washington, D.C.
Acknowledgements

C.P.L. wishes  to  acknowledge  the  contributions of  P.H.
Hemberger, T.M. Cannon, and M.A. Wolf during the initial
development of this instrumentation.

This work has been financially supported by the Department
of Energy Office of Environmental Restoration  and Waste
Management, Los Alamos National Laboratory Environmental
Restoration Program, Applied Instrumentation Technology.
Conclusion

A transportable GC/MS for the qualitative and quantitative
analysis of volatile organic compounds present at chemical
waste  sites  has  been  developed  at Los  Alamos National
Laboratory.   System  components  have  been  integrated  to
produce an instrument which is extensively automated and can
be  operated  with  minimal training.  Protocols for field
technicians with subsequent data review by a skilled analyst
                                                         369

-------
                                                     TABLE 1.

                            VOLATILE INTERNAL STANDARDS WITH CORRESPONDING
                                    ANALYTES ASSIGNED FOR QUANTITATION
 Pentafluorobenzene
 Dichlorodifluoromethane
 Chloromethane
 Vinyl Chloride
 Trichlorofluoromethane
 l,l-Dich!oroethene
 Trichlorotrifluoroethane
 lodomethane
 Carbon Disulfide
 Acetone
 Methylene Chloride
 Acrylonitrile
 trans- 1,2-Dichlorethene
 1,1-Dichloroethane
 Vinyl Acetate
 2,2-Dichloropropane
 cis-1,2-Dichloroethene
 2-Butanone
 Bromochloromethane
 Chloroform
 1,1,1 -Trichloroethane
 Carbon Tetrachloride
 1,1 -Dichloropropene
 Benzene
 1,2-Dichloroethane-d4 (surrogate)
 1,2-Dichloroethane
1.4-Difluorobenzene
Trichloroethene
1,2-Dichloropropane
Dibromomethane
Bromodichloromethane
2-Chlorovinylether
trans-1,3-Dichloropropene
Toluene d8 (surrogate)
Ret. Time
(min: sec)

  5:45
   :04
   :18
   :28
  1:18
  2:00
  2:03
  2:10
  2:09
  2:08
  2:43
  3:13
  3:09
  3:59
  4:18
  4:59
  5:02
  5:09
  5:28
  5:28
  5:34
  5:45
  5:46
  5:58
 .5:56
  6:01
  6:24
  6:36
  6:48
  6:54
  7:04
  7:20
  7:26
  7:38
 Chlorobenzene d5
 4-Methyl-2-pentanone
 Toluene
 cis-1,3-Dichloropropene
 1,1,2-Trichloroethane
 Tetrachloroethene
 1,3-Dichloropropane
 Chlorodibromomethane
 2-Hexanone
 1,2-Dibromomethane
 Chlorobenzene
 1,1,1,2-Trichloroethane
 Ethylbenzene
 m,p-Xylene
 o-Xylene
 Styrene
 Broinoform
 Isopropyl Benzene
 1.4-Dichloroethane-d4
4-Bromofluorobenzene (surrogate)
Bromobenzene
1,2,3-Trichloropropane
1,1,2,2-Tetrachloroethane
n-Propylbenzene
2-ChIorotoluene
4-Chlorotoluene
1,3,5-Trimethylbenzene
t-Butylbenzene
1,2,4-Trimethylbenzene
S-Butylbenzene
1,3-DichIorobenzene
1,4-Dichlorobenzene
p-Isopropyltoluene
1,2-Dichlorobenzene
n-Butylbenzene
1,2-Dibromo-3-Chloropropane
1,2,4-Trichlorobenzene
Napthalene
Hexachlorobutadiene
1,2,3-Trichlorobenzene
Ret. Timg
 (min: sec)

     8:56
     7:36
     7:42
     7:55
     8:05
     8:10
     8:13
     8:25
     8:19
     8:31
     8:58
     9:04
     9:07
     9:15
     9:39
     9:41
     9:52
    10:04
    11:36
    10:15
    10:24
    10:29
    10:26
    10:33
    10:39
    10:46
    10:46
    11:09
    11:12
    11:24
    11:31
    11:38
    11:34
    12:03
    12:04
    13:03
    14:11
    14:21
    14:27
    14:54
                                                         370

-------
                                       TABLE 2.




       COMPARISON OF FIELD ANALYSIS TO LABORARORY BASED ANALYSIS*











       Compound                  Field                             Laboratory




1,1,1-Trichloroethane               1900 ppb                         1340 ppb




Tetrachloroethene                  1800 ppb                         1500 ppb




2-Butanone                        140 ppb                           40 ppb




Trichlorotrifluoroethane            2950 ppb                          810 ppb











*100x Dilution Required Prior to Anaylsis
                                           371

-------
                           POSITION A
                      ANALYSIS • TRAP A (-
                 SAMPLE COLLECTION - TRAP B (
H« PURGE
  GAS
                          POSITION B
                    ANALYSIS • TRAP B (
               SAMPLE COLLECTION - TRAP A (-
    Figure 1. Schematic of Transportable GC/ITD Sampling System.
                              372

-------
 o
 (X
U
 c
 o

TJ
    174-
             569
             9:21
 9:41         18:61
 620
10:21
10:41
                                        Scan Number
                                    Retention Time (min:sec)
     Figure 2.   Extracted  Ion Current Profile Derived from Molecular  Ion of 4-Bromofluorobenzene
     (50ng).
                                  95
I   INT
75
                    6168
                            81
                                                                 174
                                     104113   128  141
                                             i        i        i
                                           120     140     160
                                         m/z
      Figure 3. Mass Spectrum of 4-Bromofluorobenzene Which Meets Tune Criteria.
                                               373

-------
r;

U
*
2  to
s275
c
c
o
u

a:
                                  <-- xlO
            50    180   150    206   250    300   358   408    450   500    550   608    650   706    750   800    850   906

           0:51  1:41   2:31   3:21   4:11  5:01   5:51   6:41   7:31  8:21   9:11  18:01  10:51 11:41  12:31 13:21  14:11 15:01
                                                          Scan Number
                                                       Retention Time (minisec)
                                 Figure 4. Total Ion Chromatogram Obtained in the Field from 50 ppb Standard.

-------
 3
 u
 I   58
 "2   tO
 §275
 c
 o
                                               fyrr
              50    100    150    200   250   300   350
             0:51   1:41  2:31   3:21   4:11   5:01   5:51
       Figure 5. Total Ion Chromatogram of Internal Standards Obtained in Fast Screening Operational
       Mode.
    100%
 o
 lH
ft,

 a
U
-o
01
X
W
             7:41
8:01
 500
8:21
 520
8:41
 540
9:01
      Figure 6. Extracted Ion Current Profile of Toluene d8 Molecular Ion from 20 ppt Standard Solution

      (100 picograms/5 mis water).
                                                375

-------
    1062
3
ea
•i—'


<

CO
>
                     55
                  52
                49
                       57
                                       73
                                         76
82
                                                    87
                                                             96
,   IB   113    120
i    ,i  i
                                                                           i
                                                                108
                                                                                   120
                                            m/z
          Figure 7.  Background Subtracted Mass Spectrum Obtained from 20 ppt Standard Solution.
                                                 376

-------
                 The Use  of Field Gas Chromatography to Protect
                              Groundwater Supplies
                               Thomas M. Spittler
                          Director, USEPA Regional Lab
                      60 Westview St., Lexington, MA 02173
Abstract

The  Use  of  field  instrumentation  to
detect the  presence  of  volatile chem-
icals  in  the environment has undergone
rapid and dramatic  change  in  the past
fifteen years.   This  paper will give a
brief overview of  this  development and
indicate some  of the  promising uses to
which this equipment can  be put  in the
service of groundwater protection.

The use of non-specific detectors to de-
termine the presence of  volatile organ-
ics in  the environment is really a pre-
liminary phase  of  field  gas chromato-
graphy.   Such instruments can determine
low ppm levels of most volatiles and are
now equipped  with on-board data-logging
capability.

Portable  gas  chromatographs  are obvi-
ously more useful for identification and
guantitation of mixtures of volatile or-
ganic  contaminants.    While  there are
many different instruments  on  the mar-
ket,  only  a  limited  number  meet the
qualifications of true portability, rug-
gedness  and  high  sensitivity that are
frequently  required  in  field studies.
The  capabilities   and  limitations  of
several field  instruments  will  be de-
scribed in some detail.

The remarkable sensitivity of some field
gas  chromatographs   enables   a  field
chemist  to  detect  very  low levels in
ambient or vadose zone  soil gas surveys
(low  ppb  levels  on  a  wt/wt  basis).
Also, the  use of  headspace analysis in
the field provides ppt sensitivity for
volatile organics in water.   These tech-
niques provide  a field  analyst with an
ability   to   detect  contamination  in
potable water even when  levels are well
below any need for health concerns.

The above  techniques and equipment pro-
vide the basis  for  a  truly preventive
strategy  to  protect  groundwater  sup-
plies.   Some discussion  of the various
stages  useful  in developing and imple-
menting a  groundwater  protection stra-
tegy will be discussed.

Introduction

Contamination  of   groundwater  in  the
1970's was primarily  a  matter  of con-
cern  for  bacterial, odorous or visible
constituents  deemed   undesireable  for
potability.   With the  advent of sensi-
tive   detection   equipment,  attention
began  to  focus  on the presence of or-
ganic contaminants as well.  It  did not
take  long  to realize that in most areas
of the country groundwater contamination
was principally caused by volatile, low-
solubility  solvents  and  hydrocarbons.
Now it  is widely  acknowledged that the
greatest threat to  groundwater  is from
fuel  leaks  and other  solvent losses to
the ground and water table.

Total Organic Analysers

Field measurement  of  volatile organics
at  waste  sites began historically with
the use of portable  total organic vapor
detection  equipment.(1)   Typical exam-
ples  of this equipment  were the  HNu PM
101,  Century Systems OVA and the AID PID
                                          377

-------
analyser.   Using either photoionization
or   flame  ionization  as  the detection
principle,  these  instruments were able
to   detect  most  volatile  organics  at
about  the  1-5 ppm range.  Despite their
obvious  lack of specificity, such equip-
ment is  still widely used to detect the
presence of volatile organics  from fuel
leaks,   spills  and improper disposal of
solvents in pits, ponds  and lagoons.(2-
5)

Depending on the nature of the detector,
some instruments were  much  more sensi-
tive   to  certain  classes  of  volatile
organics.  For example, the PID detector
can  detect  about  1 ppm of benzene and
chlorinated ethylenes, but only about 40
ppm  of  the  chlorinated  alkanes.   In
fact,  some PID detectors of an early de-
sign (HNu  and AID) had almost no sensi-
tivity to chloroalkane molecules.   This
was  based  on a paper on photoionization
theory as first proposed  by Driscoll of
HNu.(6)   When Photovac introduced their
total  PID detector it  was apparent that
this   detector,  while  still  operating
with lamps at  10.2  or  10.6  ev, could
readily detect compounds with ionization
potentials above the rated energy of the
lamp.  While no theoretical  explanation
of  this  vastly  heightened sensitivity
has been published, users have known for
years  that 111 TCA can be  detected with
the Photovac gas chromatograph.

About  three  years ago a significant ad-
vance  on the total analysers appeared on
the market  in the  form of data-logging
capability.   First came the "Smart Port-
able"  from  Thermo Environmental Instru-
ments  (formerly AID)  which  was followed
closely by the Photovac "Microtip".   Now
field data  could be  stored in computer
memory and dumped via an interface cable
to  a  computer  for  later  display  in
either tabular or graphic format.

?ield Gas Chromatographs

The limitations  of non-specificity soon
led investigators to taking small field-
designed  gas  chromatographs  with them
for field  studies.(1,7)   In the 1970's
this  usually  meant   either the Century
OVA equipped with an ambient temperature
gas   chromatographic   column   or  the
isothermal   unit    by   AID.(8)     Both
instruments  had the advantage of careful
attention  to  field   needs:   they  were
ruggedly  designed,   had  a  good  track
record    for    field   usefulness   and
contained unique   features  that made for
versatility  in addressing a wide range
of  field  problems.     For Example,  the
AID  unit  could  accept  five different
detectors in  the same  gas handling  and
electronics package.    The  Century unit
contained features  that allowed a field
chemist to rapidly screen  samples using
a  total  analysis  mode and at any time
switch into the gas chromatographic mode
when the detector showed the presence of
volatile organics in the sample stream.
(1)

Headspace Analysis

To demonstrate the  usefulness  of field
chromatography, consider screening water
samples  by  analysing  headspace  above
collected drinking water samples. These
samples can  be rapidly screened by sim-
ply injecting 200  ul  of  headspace  gas
into  the  GC  septum  and observing  the
total response of  the  detector  in  the
backflush configuration.  When the back-
flush peak  exceeds some  low limit,  the
presence of ppb levels of dissolved vol-
atiles is indicated.(1)  At this point a
sample can  be injected  into the septum
using  the  GC  configuration  and rapid
analysis with rather good resolution  can
be achieved for the typical list of vol-
atiles  found   in  contaminated  ground
water.  Identification  is  performed by
comparing  peak  retention time to known
standard mixtures  in the  field.  Quan-
titation  is  achieved  by comparing  the
unknown peaks  to  known  standards with
identical retention times.(9)  Where re-
tention times  are  ambiguous,  it  is a
simple matter  to change  the column and
repeat the sample  and  standard  to  de-
termine  retention   times  again  on  a
different packed column.

Soil Gas Analysis

In addition to field headspace analysis,
a second  powerful tool  to aid in field
investigations was  the  technique known
as   "Soil   Gas   Analysis".   It   was
principally  the  remarkable sensitivity
of the  Photovac PID  detector that made
this  method  of   vapor   detection  so
widespread.    Now  it  was  possible to
detect typical  aromatics or chlorinated
alkenes  at  the  ppb  (wt/wt)  level in
air.(10,11)      With   this   increased
sensitivity, many investigators began to
determine the presence of volatiles from
spills or  underground tank leaks by the
simple expedient of measuring their con-
centration in the vadose zone. (12-14)

Soil   gas    analysis,   coupled   with
headspace analysis provides the field
                                           378

-------
investigator  with  tools  to locate and
track under-ground plumes or  tank leaks
and  determine  rapidly  their impact on
local ground-water.    There  are certain
features of the Photovac GC which enable
an experienced field analyst  to do this
work  more   effectively.    First,  the
instruments are  typically supplied with
two  columns  so  that the field analyst
can not  only  perform  reliable identi-
fications using the two-column technique
but he  can also  use a  short column in
one  position  to  speed up screening of
samples.(15)     Second,   the   use  of
inexpensive  zero-grade  compressed  air
eliminates   the   typical   problem  of
sample  matrix   interference  with  the
early part of the chromatogram since the
carrier  gas  is  now  identical  to the
usual  sample  matrix,  ambient  air  or
headspace above soil or aqueous samples.
(12)    Third,  the use of an isothermal
oven  containing  a  wide-bore capillary
column  has  greatly  enhanced the reso-
lution of field gas chromatography.

When discussing the  necessity  for high
resolution in field work, this observa-
tion has  been made frequently by exper-
ienced field investigators.   Most field
contamination incidents do not involve a
large number of  volatile  compounds. In
fact, it  is more  common to find two to
five  volatiles  in  the  typical  field
study.  Even in this situation, only one
or two  compounds predominate  at a site
and  are  the  principal  reason for the
investigation    and    remediation.(16)
Where a  larger number  of gas chromato-
graphic peaks are found  the experienced
field  chemist  will immediately suspect
the presence of some type of hydrocarbon
fuel.   In these cases it is much easier
to continue the investigation  by merely
observing the pattern and relating it to
a type  of fuel  (e.g. gasoline, diesel,
jet  fuel  etc.).   Even number two fuel
oil has enough of a volatile fraction to
be  rather  readily  detected  in under-
ground tank leaks and spills.(17,18)

Groundwater Protection Strategy

Using some of the above observations and
equipment, it  is possible now to devise
a  practical  strategy  for  groundwater
protection  which  has  much more of the
"prevention" aspect  than the "reactive"
component  so  often  found  in environ-
mental contamination incidents.(19)

The first  consideration in  a real pro-
tective  strategy  would  be the unusual
sensitivity of today's field gas chro-
matographs.    For example, the Photovac
has a sensitivity to volatiles in air in
the range  of parts per billion (pg/cc).
(11)   Consider a  volatile like benzene
dissolved  in  groundwater.  The benzene
will partition  into the  headspace of a
closed  vial  with  a distribution coef-
ficient of about 1  at room temperature.
(20)    This  means  that if benzene was
present in the aqueous phase at  the ppb
level, it  is present in the vapor phase
at the ppm level.  But we have said that
the Photovac PID is capable of detecting
one ppb of benzene  in air.   It follows
that  when  water  concentrations are in
the ppt range, it  is still  possible to
detect   them    using   the   headspace
technique.

Consider now  the fact that  many if not
most  public  water  supplies drawing on
ground water are surrounded by test bor-
ings  or  other experimental wells often
drilled  when  the  original  supply was
under  consideration  for  exploitation.
Where this is  not  the  case,  it  is a
relatively  simple  matter to place such
test wells  in  strategic  locations up-
gradient  of  the  supply  wells so that
they can be  used  as  an  early warning
monitoring  field.    Instead of regular
testing of production wells,  these test
wells can and should be regularly tested
for possible signs of early incursion of
contaminant  plumes  into the production
well field.  When  subsequent tests show
the presence  of increasing levels (even
at  the  ppt  level)  of  volatiles like
aromatics from  gasoline leaks or chlor-
inated hydrocarbons  from other sources,
it  is  time  to  investigate  potential
sources.  The  mere  fact  of increasing
levels,  low  though  they may be at the
present time, is clear evidence that the
water supply  is under  threat of future
higher contamination  which  might even-
tually render it unfit for consumption.

In the early investigation stage of this
process, soil gas analysis often can and
will play a key role.   Establishing soil
gas sampling  profiles  will  often show
clearly whence the contaminant plume or-
iginates.  Because of  the low  levels of
contamination  present  in the perimeter
of the  production well   field,  there is
still  adequate time to lay out a plan to
find and  stop the  source of contamina-
tion.    In this exercise great care must
be taken that  not  only  the  source be
removed, but  also the contaminated soil
beneath the source be  remediated as soon
and as  efficiently as  possible to stop
any further discharge  of contamination
                                         379

-------
to the  aquifer.  For this purpose there
are now many  alternatives.    Among the
most practical is vacuum extraction.(21,
22)

Vacuum Extraction Technique

Vacuum extractions is a method of choice
for removal  of volatiles  from soil for
several  reasons.    First, it addresses
the problem at the point where volaltile
organic  contamination  directly  enters
the aquifer.  Second,  vacuum extraction
can rapidly  and effectively remove pre-
cisely  that  fraction  of  the  organic
contamination which is most volatile and
water soluble.  In  fact, volatility and
water solubility of hydrocarbons go hand
in hand.  Hexane  is about  13 ppm water
soluble, octane  is 0.6  ppm soluble and
n-dodecane is only 4  ppb water soluble.
Thus, spills  involving fuels less vola-
tile than number 2  fuel  oil  have very
little  tendency  to  dissolve  in  rain
water and reach  the  aquifer.    Even a
product  layer  of  these heavier hydro-
carbon fuels will  only  contribute very
low  levels  of organic contamination to
the  underlying  groundwater.    On  the
other  hand,  the  very soluble aromatic
fraction of  gasoline,  diesel  and fuel
oil has sufficient volatility that it is
rapidly removed from the  soil by vacuum
extraction.

It has  been the  experience of many who
use  this  removal  technique  of vacuum
extraction  that  once  the  bulk of the
more volatile  constituents  are removed
from  the  spill  site, normal bacterial
action is enhanced and  rapidly consumes
the  higher  boiling  constituents which
cannot be removed  from  soil  by vacuum
extraction.(23)     The  combination  of
enhanced aeration  and  reduced volatile
content in the soil is precisely the set
of conditions which  most  favor natural
degradation of the residual hydrocarbons
from any  fuel  contamination situation.

Interceptor Well Installation

The last  step in  a prevention oriented
clean-up strategy should  be  to install
interceptor  wells  in  the  path of the
groundwater  contamination  plume.   For
this  purpose,  it  is  imperative  that
careful depth profiling of contamination
in  the  aquifer  precede any attempt to
install  interceptor  or  barrier wells.
The  literature  is replete with studies
which indicate  that  groundwater plumes
are  usually  confined in their vertical
and horizontal dimensions by the natural
geological  features  and the wide range
of permeabilities  of aquifer materials.
(24)  By careful placement of screens in
the  interceptor  wells  a contamination
plume  can  be  cost-effectively removed
from the aquifer while  permitting clean
water from other parts of the aquifer to
continue to supply the production well.

Field GC and Groundwater Protection

Routine  field  monitoring  of  the test
wells and withdrawn and aerated water by
a field chromatograph  will  insure that
the  water  supply  remains in a potable
condition.  At the same time  vacuum ex-
tracted  volatiles  can  be monitored to
prevent air pollution.  The use of real-
time monitoring will prevent inadvertant
environmental damage during the cleanup.

Spray Aeration Technique

Regarding techniques to aid in restoring
the  quality  of  water removed from the
plume zone, I wish  to discuss  a method
developed several   years ago in Miami by
Paul  Wood.(25)    Wood's  technique was
spray aeration  as  an  alternative to the
more conventional stripping towers.   In
his system, water is sprayed upward in  a
box the dimensions  of  which are not cri-
tical as  long  as   about 14 ft in height
is  used.  At  the base  of the  box is  a
tank  to  receive   the sprayed water.   A
second pump and spray  head is  placed in
series in   a  second  box to  form a two-
stage system  of  spray  aeration.   Wood
measured  an  average  90%  efficiency per
aeration  stage.   It   is   then  a simple
matter to   place enough stages in series
to  remove  dissolved  volatile   organic
contamination down   to whatever level is
satisfactory  for recycling back into the
aquifer.

Contrasting   Wood's  spray aeration with
stripping   towers,   the   following   five
points  should   be made.      1)  Spray
aeration  is   considerably  less  expensive
to  build  and  maintain.   2) Biofouling of
the aeration  system is almost non exist-
ent.    Compare  this  to  the  continual
buildup of  large  bacterial  colonies in
stripping towers.    3)  Buildup of oxide
films on  the  stripping   tower  coils is
absent  in  the  spray aeration box.  4)
Concentration  of   removed  volatiles is
much higher in  the  natural draft  exhaust
than in the large   volume-high  flow  rate
coming  from  a  stripping   tower.  5) The
highly  concentrated  exhaust   stream of
the aeration  box  is more  easily captured
and can be  more cost-effectively
                                           380

-------
prevented  from  becoming  an air pollu-
tion problem.

Field GC and Spray Aeration

Once more, the  use  of  a  portable gas
chromatograph to monitor the performance
of the vapor handling system is obvious.
Data  can  be  obtained  in real time to
monitor the efficiency of spray aeration
using  headspace  analysis  of inlet and
finished water.  Vapor  concentration of
the  volatiles  from  the aeration tower
discharge can also be  monitored in real
time  to  assist  in designing and oper-
ating a vapor  recovery  system.   Where
vapor  recovery  is  necessary,  systems
such as the CRS technique (26) will also
benefit   from  ongoing  field monitoring
for optimization of performance.

Conclusions

In conclusion,  we  have  seen  that the
availability  of  inexpensive, sensitive
and rugged field gas  chromatographs can
make a  substantial contribution towards
a practical groundwater protecticr. stra-
tegy.    There  are now many examples in
the  northeast  where  these  ideas have
been applied  at the local municipal and
county 1evel.(17,19)   With  the limited
resources  of  federal  programs and the
dwindling  funding  of  state  programs,
this :r.ove towards local self-help is not
only welcome but long  overdue.   Such a
strategy,  under  the guidance and tech-
nical  overview  of  state  and  federal
programs promises to be a cost-effective
way  to  insure  the  future  purity  of
groundwater resources.   There is also a
growing awareness  that such inexpensive
techniques  as  field gas chromatography
are   sorely   needed   in   many  other
countries  where  emerging environmental
awareness cannot keep pace  with limited
budgets,  but  where  a  high  level  of
dependence on groundwater is a practical
necessity.

Bibliography

1.  Spittler,  T.M.,  "Use  of  Portable
Organic  Vapor  Detectors  for Hazardous
Waste  Site  Investigations", HMCRI Con-
ference,  Washington,  DC,  Oct.  15-17,
1980.
2. Fitzgerald, J . J . , "On-Site Analytical
Screening of Gasoline Contaminated Media
Using   a    Jar   Headspace  Procedure",
Petroleum  Contaminated  Soils,  V.  II,
Lewis  Publishers,  Inc.,  Chelsea,  MI,
1989, pp. 119-136.
3. Robbins, G.A., Bristol, R.D. & Roe,
V.D.,  "A  Field  Screening  Method  for
Gasoline  Contamination  Using  a  Poly-
ethylene  Bag  Sampling  System",  GWMR,
Fall, 1989, pp.  87-97.
4.  Robbins,  G.A.,  Deyo,  E.G.  et al,
"Soil   Gas   Surveying  for  Subsurface
Gasoline   Contamination   Using   Total
Organic   Vapor   Detectors   -  Ft.  I,
Theory", GWMR, Summer,  1990, pp. 122-31.
5.  Robbins,  G.A.,  Deyo,  E.G.  et al,
"Soil   Gas   Surveying  for  Subsurface
Gasoline   Contamination   Using   Total
Organic Vapor Detectors - Pt. II, Field
Experimentation", GWMR in press.
6.  Driscoll,  J.N.  &  Spaziani,  F-F;,'
"Trace Gas Analysis by Photoionization",
Analytical  Instrumentation  Division of
ISA, King of Prussia, PA, May,  1975.
7. Clay, P.P. & Spittler, T.M.,  "The Use
of  Portable  Instruments  in   Hazardous
Waste Site  Investigations",  HMCRI Con-
ference, Washington, DC, Nov. 29-Dec. 1,
1982, pp. 40-44.
8. Mehran, M. , Michael, J. S. Sirota, E. ,
"Delineation  of Underground Hydrocarbon
Leaks by Organic Vapor  Detection", Pro-
ceedings of HMCRI Conference, Washington
DC, Oct. 31, 1983, pp 94-97.
9.   Clark,   A.E.,   LatailU,  M.M.   &
Taylor, E.L., "The Use of a  Portable GC
for  Rapid   Screening  of   Samples  for
Purgeable Organic Compounds  in  the Field
and  in  the Laboratory",  USEPA Region  I
Lab  SOP, Lexington, MA, 1982.
10.  Barker,  N.J.  &  Leveson,   R.C., "A
Portable  Photoionization  GC for Direct
Air  Analysis", American Laboratory, Dec.
1980, pp. 76-83.
11.  Leveson,  R.C.  &  Barker,  N.J., "A
Portable   Multi-Component   Air  Purity
Analyser   Having   Sub-ppb  Capabililty
without  Sample  Preconcentration", 27th
ISA  Conference  Proceedings, St. Louis,
MO,  Spring, 1981, pp. 7-12.
12.  Spittler,  T.M.,  Clifford,  W.S.  &
Fitch,  L.G.,  "A New Method  for  Detection
of Organic  Vapors in  the Vadose Zone",
AWWA Conference Proceedings, Denver, CO,
Nov. 19-21, 1985, pp. 236-246.
13.  Marrin,  D.L.  &   Thompson,  G.M.,
"Remote  Detection  of  Volatile Organic
Contaminants  in  Groundwater via Shallow
Soil  Gas   Sampling",  NWWA Conference
Proceedings, Petroleum  Hydrocarbons and
Organic Chemicals in Groundwater, 1984.
14.  Nadeau, R.J., Stone, T.S. & Klinger,
G.S., "Sampling  Soil  Vapors   to Detect
Subsurface  Contamination:   A   Technique
and   Case    Study",   NWWA   Conference
Proceedings,  Denver,  CO,   Nov.  19-21,
1985, pp.  215-226.
15.  Spittler,  T.M.,  Lataille,  M.M.  &
Parks,  P.A.,  "Correlation  between  Field
GC Measurement  of Volatile   Organics and
                                          381

-------
 Laboratory   Confirmation  of  Collected
 Field  Samples  Using  a   GC/MS",   APCA
 Specialty Conference,  Chicago, IL, Mar.
 1983.
 16.   Doty,  C.B.   &  Travis,   C.C., "The
 Superfund   Remedial   Action   Decision
 Process:   A Review  of Fifty   Records of
 Decision", JAPCA,  v. 39, # 12,  1989,  pp.
 1535-43.
 17.   Stiefel,   C.L.  &  Heufelder,  G.R.,
 "Regulation  and   Testing of  Residential
 Underground Fuel   Storage Tanks" Journal
 NEWWA,  Dec.,  1988, pp. 256-266.
 18.   Wilhelm,   R.W.  &  Bouchard,   R.J.,
 "Assessment and  Remediation  of Residen-
 tial   Properties   Contaminated  with Home
 Heating  Oil",    Petroleum Contaminated
 Soils,   v. II,  ch.  27, Lewis  Publishers,
 Inc,  Chelsea,  MI,  1989,  pp. 329-346.
 19.   Anon.,   "A  Groundwater  Protection
 Strategy    for    Local  Municipalities",
 Water Connection,  NEIWPCC Newsletter,  v
 3, #  2,  1986.
 20. Mackay, D.  &  Shiu, W.Y.,  "A Critical
 Review   of Henry's  Law  Constants  for
 Chemicals  of Environmental  Interest",  J
 Phys. Chem., v. 10,  #  4,  1981,  pp.  1175-

 21. Malot,  J.J.,  "Cleanup  of a Gasoline
 Contaminated Site   Using Vacuum Extrac-
 tion  Technology",  Petroleum Contaminated
 Soils, v.   II,  Lewis  Publishers,  Inc.,
 Chelsea, MI, 1988,  pp. 283-301.
 22.Bowman,  R.S.,   "Manipulation  of  the
 Vadose  Zone  to  Enhance  Toxic Organic
 Chemical Removal",  2nd.  Int'l.  Workshop:
 Behaviour.of Pollutants  in Porous Media,
 Bet Dagan,  Israel,  June,  1987.
 23. Connor,  J.R.,  "A  Case Study of  In-
 Situ  Soil  Venting   in   Conjunction with
 Field  Recharged  Carbon  Adsorption  for
 Gasoline Removal",   Unpublished, Private
 Communication.
 24.  Cherry,  Prof.  J.A., University of
Waterloo, Waterloo,  Ont. Canada, Private
 Communication.
 25.  Malot,  J.  J.  & Wood, P.  R.,  "Low
 Cost,  Site Specific  Total  Apoproach to
 Decontamination",   Soils Contaminated by
 Petroleum, Wiley and Sons, NY,  1988, pp
 331-354.
 26.  Patterson,   J.H.,  "Case  History:
 Soil  Venting as  a  Construction Safety-
Remediation  Method  for  Development of
Contaminated  Property",   Proceedings of
Seminar   for  Ct.  State Civil  Engineers,
Session  IV, Berlin, CT, Nov. 2-3, 1989.
                                          382

-------
                   FIELD SCREENING PROCEDURES FOR DETERMINING THE,PRESENCE
                            OF VOLATILE ORGANIC COMPOUNDS IN SOIL
                              Alan B.  Crockett and Mark S.  DeHaan
                    Idaho National Engineering Laboratory,  EG&G Idaho,  Inc.
                                   Idaho Falls, ID  83415
ABSTRACT

Many  field  screening  procedures  have  been used
to detect the  presence  of volatile organic
compounds (VOC)  in  soils but almost none have
been  documented  and verified.  Users  of these
procedures  have  not really known whether their
objectives  in  screening were met.  A  reliable
VOC screening  procedure could significantly
reduce the  number of  samples currently being
submitted to laboratories, thereby reducing
costs and improving site characterization.  The
Environmental  Protection Agency's Environmental
Monitoring  Systems  Laboratory in Las  Vegas
(EMSL-LV) has  therefore sponsored a research
effort to evaluate  and  improve headspace methods
for screening  soils for VOC in the field.  The
research involved comparing several extraction
procedures  using soils from actual waste sites,
and determining the agitation and mixing
necessary to achieve equilibrium.  Headspace was
analyzed using a relatively simple portable gas
chromatograph with  a short column.  The results
were variable  and show that several procedures
should be attempted and the results evaluated
before selecting a  screening procedure.

INTRODUCTION

A recent study by the Office of Technology
Assessment and the National  Academy of
Sciences has indicated that the U.S.
Environmental Protection Agency (EPA)  should be
collecting 10 times as many samples as is the
  Work performed for the U.S.  Environmental
Protection Agency under Agreement No. DE-SA07-
90ID012989, through the U.S. Department of
Energy, Contract No. DE-AC07-76ID01570.
current practice under the Resource Conservation
and Recovery Act (RCRA) and Superfund.
Considering that the approximately 80
laboratories in EPA's Contract Laboratory
Program (CLP) are already operating at full
capacity, that there will be no extension of
clean-up deadlines and no increase in funding,
there will have to be major changes in the
programs to increase efficiency.  One way of
increasing efficiency is to reduce the number of
samples being analyzed under CLP protocols that
show no or only very low contamination levels.
At present, 80% and 90% of the samples submitted
to CLP laboratories for analysis of volatile and
semivolatile organics, respectively, fall in
this category (personal communication, Dave
Bottrell, EMSL-LV).  One means of reducing the
numbers of such samples is to screen samples
prior to submission for CLP analyses.  In
theory, 80% of the volatile samples being
submitted to CLP laboratories could be
eliminated or the CLP productive capacity could
be expanded by a factor of 5 if adequate
screening methods for VOC were available.

Specific examples would be the Department of
Energy's (DOE) Hanford Site and Savannah River
Plant, where about 126 soil or sediment samples
were collected and analyzed for volatile
organics as part of the DOE Environmental Survey
(1, 2).  Of the samples collected, approximately
59% equaled or exceeded the Contract Required
Detection Limit (CRDL) for one or more volatile
compounds.  Many of these samples were flagged
with a B, indicating blank contamination.

Only 29% of the samples had positive detections
above the CRDL and no B flag.   Therefore, in
theory, somewhere between 41% and 71% of the
samples could have been rejected if screened in
                                                  383

-------
the field to CRDL limits.  If it costs DOE
$250-5500 for each VOC analysis, the savings
could have been $13,000-$44,000 for analytical
services, which does not include costs for data
management and report preparation.

This research was designed to evaluate several
headspace methods for screening soil samples in
the field for the presence of VOC.  The
objective was to determine whether or not to
send a sample to a CLP laboratory for gas
chromatography/mass spectrometry (GC/MS)
analysis.

BACKGROUND

The term "volatile organic compounds" refers to
a group of chemicals that readily pass from a
solid or liquid form to the gaseous phase
(volatile) and are composed of carbon-based
molecules (organic).  Many VOC, toxins,
carcinogens, or mutagens, are hazardous to the
health of human and nonhuman organisms and are
common environmental contaminants.   Volatile
organics (Table 1 for EPA list of VOC) are
particularly significant because they constitute
15 of the 25 most frequently identified
substances at 546 superfund sites (3).

A standard operating procedure for field
screening of VOC could decrease the demand for
CLP analyses, and at the same time result in
improved characterization of hazardous waste
sites.  More samples could be collected and
screened, thus increasing the size of the area
characterized at a site, the intensity of that
characterization, and maximizing the usefulness
of those samples sent to the laboratory for
analysis.  Near real time data would permit the
field sampler to redesign the sampling effort
while still in the field to characterize "hot
spots" and plumes.  Other potential uses include
preliminary input for risk-assessment studies,
monitoring for efficacy of clean-up actions, and
research on the transport of VOC in soils (4).

Laboratory Analysis of Soils.  EPA Method 8240
(5) for VOC in soils uses an inert gas to purge
(11 min.) VOC from a mixture of 5 g of soil and
5 mL of water into a Tenax  trap.   The VOC is
thermally desorbed and swept into the GC/MS for
analysis.  The "high-level method" involves
extraction of 1 g of soil with 10 mL of methanol
(including spiking solution) by hand shaking for
     2  Mention  of  specific  products  and/or
manufacturers in this document implies neither
endorsement or  preference,  nor  disapproval  by
the U.S.  Government,  any of  its  agencies,  or
EG&G  Idaho,  Inc., of  the  use  of a  specific
product for any purpose.
2 min., transfer of an aliquot of the extract to
a purge and trap device, and analysis by GC/MS.

Laboratory Screening of Soils by Headspace
Analysis.  EPA Method 3810 (5) allows rapid
screening of large numbers of samples.  Ten g of
soil are placed in two 125-mL septum-seal glass
vials and one is spiked with calibration
standards.  The two vials plus a third
containing only the standards are allowed to
equilibrate in a 90°C water bath for 1 hour.
Then, 2 mL of headspace is withdrawn and
injected into a GC.  Detection limits for this
method may vary widely among samples because of
the large variability and complicated matrices
of waste samples.  The sensitivity of the method
depends upon the equilibria of VOC between the
vapor and dissolved phases.

Field Screening.  The most commonly used
procedures for field screening soil samples
involves analyzing headspace with an organic
vapor analyzer such as a Photovac TIP, HNU
PI-101 or Century Systems OVA-108.  These
instruments respond to flame or photo ionizable
materials in air.  They are very portable, easy
to use but do require relatively large sample
flow (0.25-1 L/min.) and have detection limits
in the lower ppm range.  Very little data have
been published on their effectiveness as field
screening devices for determining the presence
of VOC in soils.

To improve detection limits and reduce sample
size, some field personnel have used portable
gas chromatographs which are up to 3 orders of
magnitude more sensitive and only require
headspace samples of 1-2 mL or even low jzL
quantities.  Much more has been published on the
use of such devices for screening and analysis
of soil headspace in the field.  Most screening
methods for soil samples are based on headspace
methods used for water samples.

Cheatham (6) effectively used a close support
laboratory to provide rapid assessment of
presence or absence of organic contamination.
The study used two portable HNU 301 GCs equipped
with a photo ionization detector  (PID) connected
in series with a flame ionization detector, or
an electron capture detector.  The method gave
unacceptable resolution of the indicator
compounds when no column packings were used.
So, direct headspace and purge and trap
techniques were studied using packed columns.
Samples were prepared by sealing  10 g of soil  in
tared 100 ml serum bottles, which were placed  in
a 90° water bath and allowed to equilibrate
approximately 1 hour, after which time the
headspace gas was analyzed.  Samples were
submitted to a CLP laboratory for confirmation.
The results were of sufficient quality to
increase the accuracy of the site
                                                   384

-------
          TABLE  1.  CLP ANALYTICAL  DATA  FOR TEST  SOILS
TARGET COMPOUNDS
Acetone
Benzene
Bromodichloromethane
Bromoform
Bromomethane
2-Butanone
Carbon dtsulfide
Carbon tetrachlorlde
Chlorobenzene
Chloroethane
Chloroform
Chloromethane
c 1 s-1 . 3-d1ch loropropene
Dlbromochloromethane
1 , 1-Oichloroethane
l,2-D1chloroethane
l,l-D1chloroethene
1,2-Dichloroethene
(total)
1 , 2-Dichloropropane
Ethyl benzene
2-Hexanone
Hethylene chloride
4-Nethyl-2-pentanone
Styrene
1,1,2,2-
Tetrachloroethane
Tetrachloroethene
Toluene
trans-l,2-d1chloroethene
trans-1,3-
dlch loropropene
1,1, 1-Tr Ichloroethane
1,1,2-Trlchloroethane
Trlchloroethene
Vinyl acetate
Vinyl chloride
Xylenes (total)
m-xylene
o-xylene
SAMPLE
KC804203
(Ug/kg)
0
66
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
24
0
170
0
0
0
0
0
0
84
0
0
0
0
0
0
31
870
0
0
KC804214
OigAg)
0
45
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
26
0
160
0
0
0
0
0
0
62
0
0
0
0
0
0
0
860
0
0
LA61201
(|ig/kg)
95
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
62
0
0
0
0
0
0
0
LA82301
(ligAg)
180
0
0
0
0
120
0
0
0
450
59
0
0
0
510
0
140
6200
0
3
0
190
0
0
0
0
47
0
0
160
9
33
0
24
0
0
0
LA82302
(Ug/kg)
260
0
0
0
0
440
0
0
0
300
320
0
0
0
1100
0
210
5800
0
0
0
160
0
0
0
0
72
0
0
1200
32
55
0
26
0
0
0
SR52702
(Ug/kg)
1200
220
0
0
0
220
17
0
0
0
0
0
0
0
0
0
0
10
0
200
0
65
0
0
0
40
1300
0
0
0
0
16
0
1100
380
0
0
Batch 1
(Ug/kg)
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
5
0
0
0
37
0
0
0
0
0
0
0
6-4
(Ug/kg)
0
3
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
47
2
1
0
0
2
0
4
0
0
0
0
0
PARK LOT
(|lg/kg)
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
2
0
0
0
0
138
36
0
0
0
3
0
30
0
0
0
0
680
characterization and improve plume mapping.  The
quick turnaround time (-2 hours) allowed the
field staff to better understand the site and
select sampling locations using known data
rather than "best guess", thus optimizing the
limited project budget.

Clark, et al. (7) used a method developed by
Spittler in which 10 mL of soil was added to a
tared 40-mL VOA vial containing 20 mL of water
and 20 /*L of 2% mercuric chloride.  In the lab,
the vials were warmed to room temperature,
shaken for 1 minute, and the headspace analyzed
using a Photovac model 10A10 GC.  Headspace
sample volumes varied from 10 /iL to 1 mL.  The
method was used to screen samples prior to GC/MS
analysis to avoid overloading the GC/MS and to
provide an indication of the presence or absence
                                                  385

-------
of organics.  Use of the method has reduced
expensive GC/MS time and greatly reduced lost
analysis time in the laboratory.  When screening
showed no organics, Clark thought it safe to
assume that no priority pollutants were present
at the ppb level, but that there might be
exceptions.

Griffith (8) and Spittler et al. (9) spiked soil
with three known VOC and placed aliquots of that
soil in water.  The soil/water mixture,
contained in a glass vial, was placed in a water
bath and allowed to equilibrate.  The air above
the water was then analyzed for VOC using a
Century Systems Model 128 GC portable organic
vapor analyzer.  Results showed good recovery of
the compounds introduced to the soil specimens.
The method was tested under field conditions
with duplicate samples sent to an independent
laboratory.  Comparison of field and laboratory
results showed good correlation for the aromatic
compounds under study.

OBJECTIVES

The objective of this research was to evaluate
eight headspace procedures for determining the
presence or absence of volatile organic
compounds (CLP list) in soil at less than ppm
detection limits.  The intent was to be able to
screen soils for the presence of VOC and to
decide whether the samples need further analysis
by CLP methods or could be considered clean,
(i.e., contain insignificant levels of VOC.)

To achieve this objective, a variety of
information was required.  First, an extraction
technique was needed that would maximize the
concentration of volatile compounds in the
headspace (more important for instruments
without low detection limits), in a reasonable
period of time, using practical field
procedures.  This extraction procedure should be
fast (minutes), efficient (90% of equilibrium),
and easily accomplished under field conditions.
Secondly, a suitable analytical device was
required for detection of VOC in headspace.  A
Photovac 10A10 field portable GC with a PID was
selected although other instrumentation could be
substituted.

APPROACH

This study was designed to build upon previous
studies.  While a quantitative approach is
reasonable for water, many uncontrollable
factors make soil headspace analysis much more
complicated.  This study was not designed to
investigate theory but to develop empirical
evidence on the utility and limitations of soil
headspace analysis using "naturally"
contaminated soils.
The first step was to compare a variety of soil
extraction methods to maximize headspace
concentration of volatiles when used on
"naturally" (as opposed to spiked) contaminated
soils.  The second step was to determine the
"best" method of achieving 90% of headspace
equilibrium using the extraction method
identified in Step 1.  The third step (not
conducted) would have been to use the screening
procedure in the field on samples that would
also be analyzed by GC/MS.  The methods and
procedures are compared to each other, as well
as to a modified Method 3810 (5) in which the
10A10 was substituted for the prescribed
analytical instrumentation.
TABLE 2.  EXPERIMENTAL TREATMENTS
1
2
3
4-6
7
8
1 g of soil in 29.5 mL of water
in a 40 mL VOA vial
5 g of soil in 27.5 mL of water
in a 40 mL VOA vial
20 g of soil in 20.0 mL of water
in a 40 mL VOA vial
same as above with a saturated
NaCl solution
10 g of soil in a 125 mL septum
cap bottle heated to 90*C in a
water bath for 1 h.
5 g of soil in 5 mL of methanol
followed of 0.6 mL of methanol to
29.5 mL of water in a VOA vial
STEP 1:  COMPARISON OF SIX EXTRACTION METHODS TO
ACHIEVE MAXIMUM HEADSPACE CONCENTRATION.  The
first tests were to determine maximum headspace
concentration after vigorous agitation using the
treatments listed  in Table 2.  The first
treatment was suggested by Spittler and has been
investigated by Griffith (8) and Spittler et al.
(9).  A salt solution was selected for testing
since it is well known that adding salts to
water samples can  increase headspace
concentration (salt also can be used as a
nonhazardous preservative).  Treatments 1-6 and
8 were planned around 10 mL of headspace in
volatile organic analysis (VOA) vials.
Treatments 7 and 8 were included for comparison
to  a standard laboratory screening procedure
(Method 3810) (5)  and a solvent extraction
method (Method 8240, "high level" method which
uses methanol as the extracting solvent) (5).
                                                   386

-------
The importance of achieving maximum
concentrations in the headspace was to improve
the detection limit.  While a GC/MS may not be
as sensitive as the Photovac 10A10 to some
compounds, a much larger sample can be actually
analyzed by using the Tenax sorbent in Method
8240.  Essentially all the VOC contained in the
5 g sample are  placed in the instrument using
Method 8240 while with headspace techniques,
only a small fraction of the VOC are injected.
Thus, to approach CLP detection limits for as
many compounds as possible, the objective was to
maximize static headspace concentration.

Obtaining fresh soil samples with good
analytical data was a problem, so archived
samples, known to have contained volatile
contaminants, were used.  Although the samples
were quite old they still had detectable
concentrations of VOC, so their age did not
matter.  The interest was in how efficient the
extraction procedures were and how long it took
to reach equilibrium (Step 2, below).

The headspace was sampled with standard syringes
(2-40 nt injections) and analyzed using the
Photovac 10A10 which had a 10.6 eV PID.  Ultra
zero air was used as the carrier gas.  While it
was not possible to specifically identify or
quantify the VOC present, it was the relative
concentration among the treatments that was
important.

The sources of the soil samples are given below
and the contaminants originally reported present
are listed in Table 1.  Some of the VOC have
high ionization potentials and thus are
difficult to detect with the 10.6 eV PID.

Batch 1     provided by EPA and collected from
            the Times Oil Superfund site in
            Tacoma,  Washington.

G-4         laboratory column sample provided by
            EPA.

KC804203    subsurface soil  sample collected
            from under a pond at the DOE's
            Kansas City Plant.

KC804214    very similar to the above.

LA61201     soil  collected from a canyon wall  at
            DOE's Los Alamos
2LA61202    National  Laboratory.

LA82301     sludge samples from an inactive
            septic tank at DOE's Los
LA82302     Alamos National  Laboratory.

Park Lot    EPA provided sample  from Tacoma,
            Washington.
 SR52702     subsurface soil  sample collected
             from an oil basin at DOE's Savannah
             River Plant.

 These bottles of soil  were to be homogenized and
 cored to obtain the required aliquot which would
 be added to a tared vial  containing the
 extractant.  This proved  impractical and a
 laboratory scoop was used to accomplish the
 transfer as quickly as possible.  Vials were
 weighed again to obtain the  actual  amount of
 soil.  Tap water was used for the extractions
 since headspace blanks showed no major
 interference.  Extractant volume was measured
 using a graduated cylinder (most practical  for
 field work).

 If it is assumed that  the VOC are totally
 desorbed from the soil  in the extraction tests,
 the 20 mL soil  test should result in a 29.5
 times higher  concentration in water than 1  mL
 soil  (20 times  more organic  in about 1/3 less
 water).   While  salts in an aqueous  medium will
 reduce the solubility  of  VOC in water thus
 increasing headspace concentration  in a two
 phase system, their effect on headspace
 concentration in a soil-water-air equilibrium is
 not readily predictable.

 To rapidly achieve static equilibrium,  a Spex
 Mill  was used to violently agitate  the  soil
 water mixture for 5 min.   Headspace was analyzed
 using a  6 in. SE-30 column at ambient
 temperature.   Because  the treatment's
 effectiveness was expected to vary  greatly  for
 different soils  and VOCs,  it was deemed
 preferable to test several soils once rather
 than  perform  repeated  tests  on the  same sample.
 The limited volume of  each sample was also  a
 factor.   It was  expected  that  violent shaking
 for 5 min. would  approximate equilibrium (or  at
 least be  as rigorous as any  practical field
 extraction technique).

 STEP  2:   TIME TO  REACH  EQUILIBRIUM.   The  second
 step  was  to minimize the  time  it takes  for  soil,
 water, and headspace to reach  near  equilibrium
 (goal was  to  obtain at  least 90% of equilibrium
 conditions in headspace)  using  practical  field
 procedures.   "Naturally"  contaminated soils were
 again  used instead  of spiked soils  because  a
 common criticism  of spiking  is  "easy  on,  easy
 off".

The four extraction procedures  listed in  Table 3
were  compared with 5 min.   of agitation  on the
Spex Mill.  The null hypothesis was  that  there
 is no difference  in extraction  rate  among the
treatments.
                                                 387

-------
TABLE 3.  EXTRACTION PROCEDURES
1
2
3
4
Violent hand shaking for 1 min.
Agitation in a sonic bath for 5
min.
Agitation using a vibrator for 1
min.
Combinations of the above and
repeated analysis over time
STEP 3:  COMPARISON TO CLP PROCEDURES.  The
final evaluation of the selected extraction
procedure identified in Steps 1 and 2 was to be
a field trial of the recommended procedures.
This would involve making a decision as to
whether volatiles were present or not and then
evaluating those decisions based on laboratory
analytical data (ideally using an improved
Method 8240 where 5 g samples were sealed in the
field and never opened in the laboratory).  This
step still needs to be conducted.

Water and NaCl blanks were also prepared and
treated as samples.  A benzene in water standard
was run as a retention time and instrument
response check.  No significant attempts were
made to identify contaminants or quantify
concentrations.  The number of samples analyzed
varied, but ideally, duplicate samples with
duplicate injections were used.  Peak height
data were measured and recorded by hand.  The
raw data were normalized before statistical
analysis to eliminate differences in actual
versus planned soil weight, attenuation setting,
and volume of sample injected into the GC.  An
example of a normalized data set is presented in
Table 4 which shows:  sample number, treatment
(D- duplicate sample), nominal amounts of soil
and extractant used, and normalized peak heights
with replicate injection data.
          TABLE 4.  NORMALIZED PEAK HEIGHT DATA FOR ANALYZED SAMPLES
TREATMENT.
NOMINAL AMOUNT
OF SOIL (g).
and
EXTRACT (mLl
Treatment
Ambient
5.0 0
5.0 0 0
Treatment H20
5.0 27.5
5.0 27.50
1.0 29.5
1.0 29.50
Treatment
Heated Bot
10.0 0
10.0 0 0
Treatment
Heated Wet
10.0 10
10.0 10 D
Treatment
Met ha no 1
5.0 5.0
5.0 5,0 D
Treatment NaCl
5.0 27.5
5.0 27.50
1.0 29.5
1.0 29. 50
NORMALIZED HEIGHT. PARK LOT SAMPLE
(cm)

PEAK 1



0.78
0.85

0.69
0.77
0.17
0.21


20.16
20.73


14.59
9.70


1.23
0.53

0.91
0.79
0.21
0.29

PEAK 1R






0.91
0.70
0.00
0.25


21.09
26.97


18.16
14.71


1.07
0.00

0.61
l.OZ
0.00
0.00

PEAK 2



0.35
0.39

0.25
0.25
0.35
0.18


7.44
8.23


11.02
3.44


0.61
0.42

0.40
0.31
0.18
0.41

PEAK 2R






0.69
0.25
0.00
0.69


12.40
12.21


13.62
7.51


0.61
0.00

0.30
0.40
0.00
0.00

PEAK 3



1.02
0.89

1.23
1.99
0.42
0.46


21.71
23.57


20.75
12.52


4.70
5.56

1.28
1.67
0.48
0.53

PEAK 3R






2.26
1.75
0.00
0.71


23.26
29.81


27.24
18.78


4.09
0.00

0.99
2.01
0.00
0.00

PEAK 4



0.00
0.00

0.00
0.00
0.00
0.00


0.00
0.00


0.00
0.00


0.61
0.53

0.00
0.00
0.00
0.00

PEAK 4R






0.00
0.00
0.00
0.00


0.00
0.00


0.00
0.00


0.51
0.00

0.00
0.00
0.00
0.00

PEAK 5



2.04
2.09

1.79
1.65
1.83
1.49


10.85
12.49


3.89
3.76


1.84
2.12

1.78
0.94
1.66
1.57

PEAK 5R






1.86
1.44
0.00
1.59


10.85
15.05


4.86
5.01


1.58
0.00

0.99
1.11
0.00
0.00
                                                   388

-------
RESULTS

Initial work showed major problems with
conducting the standard laboratory procedures
(treatments 7 and 8, heated headspace and
methanol extraction).  After some early eluting
peaks, a massive tailing peak was produced which
masked all other data.  It was thought that the
level of methanol in the headspace was
interfering with the analysis and that condensed
water may have produced a similar effect.
Therefore, early experiments eliminated these
treatments.  Later, the procedures were tried
again with better results in some cases.

Statistical Methods

The normalized data were analyzed with analysis
of variance (ANOVA) techniques to test for
statistically significant treatment effects (a
treatment is a specific extracting method and
soil amount combination).  In many cases the
effect of the amount of soil extracted was
acceptably linear, but in some cases the
relationship did not appear linear.  Therefore
the results were also analyzed with the soil
amount log-transformed to better linearize the
response.  It was found that the
log-transformation did not change the results.
Because the treatments might have different
effects when used on the different soil samples,
the analyses were run separately for each
sample.

For each experimental run with a given
treatment, from four to six peaks were measured
and used.  There were high correlations among
the peak heights for a given treatment, i.e. if
a treatment affected peak 1 height, it similarly
affected all the other peaks.  To take advantage
of this correlation, a Multivariate Analysis of
Variance  (MANOVA) was used to analyze the
treatment effects.  The soil amount was treated
as a covariate in the analysis.  The analyses
were done using the GLM procedure of the SAS*
(10) statistical software package.

For the experiments using soil types KC804203,
KC804214, and LA61201, there were no true
replicates analyzed, therefore it was impossible
to directly estimate the experimental error or
natural variation.   Instead, subsampling or
measurement error was used as a lower bound
estimate of experimental error, and all
analytical results  for these samples should be
viewed conservatively.

Statistical Results of Extraction Tests

Sample SR52702

Sample SR52702 was  a subsurface soil sample
collected from an oil basin at a depth of 6-12
in.  Table 1 shows that the sample contained 12
VOC when analyzed by GC/MS for the DOE
Environmental Survey.  In appearance, the soil
was heterogeneous with black specks and rocks.
Because of concern for sampling error, the
contents of the sample jar were dumped into a
beaker, the rocks removed and the aggregate
broken up as much as possible to improve
homogeneity prior to subsampling.  Some VOC were
certainly lost in this process but there were
enough left to use for testing.

The MANOVA involved comparison of treatments 1-6
(treatments 3 and 6 used 10 g instead of the
planned 20 g of soil due to limited amount of
soil available) and showed that the NaCl and
water treatments did not result in significantly
different peak heights (Wilks' Lambda F=2.2,
p=0.23) .   However,  the amount of soil (1,  5,
and 10 g) had a statistically significant effect
on the peak heights, with larger soil amounts
resulting in higher peaks (Wilks' Lambda F=12.8,
p=0.01).  The height of the 10 g peaks were from
1.6 to 3.4 times higher than the 1-g water
extract peaks.

Sample Batch 1

The Batch 1 sample was collected by EMSL-LV
personnel from the Times Oil superfund site in
mid March, 1989.  The sample was very dense,
black, and contained small pebbles.  The
chromatograms showed 2 large peaks of very early
eluting compounds (possibly including vinyl
chloride).  Because they were so large and
early, they could not be separated and measured
accurately with the 6-in. column and only the
later 4 peaks were analyzed.

The analysis of treatments 1-6 showed that the
NaCl and water extractions did not result in
significantly different peak heights  (Wilks'
Lambda F«3.2, p-0.14).  The amount of soil used
had a statistically significant effect on peak
heights, with larger soil amounts resulting in
higher measured peaks (Wilks' Lambda F=80.6,
p=0.0004).  The 20-g samples produced peaks 15
to 48 times larger than the 1-g water extract
samples.  It is interesting to note that the
effect the amount of soil had depended on the
extractant used.
     3  The  Wilks'   Lambda   statistic  tests
whether   the   treatments   have  significantly
different effects on the measured peak heights.
The p-values given are the  probability that the
observed  results are  simply due to chance, and
not due to treatment  effect.
                                                  389

-------
 Sample KC804203

 The experiment included all  eight treatments
 (data for the 20-g NaCl treatment was lost)  but
 the methanol  extract was diluted 100:1 instead
 of 50:1.   The analysis (5 peaks) showed that the
 different extraction methods resulted in
 significantly different peak heights (Milks'
 Lambda F=5.6, p=0.003).  The water extract gave
 consistently higher peaks,  with NaCl second
 highest.   The methanol extracting and the  heated
 dry treatment gave generally lower peaks than
 the other two treatments.

 The amount of soil  used also had a statistically
 significant effect on the peak heights (Wilks'
 Lambda F=157, p^O.OOOl).   For all  the peaks, the
 greater the amount of soil  used, the higher  the
 measured  peak, i.e.  there was a positive
 correlation between  amount  of soil  used and  the
 peak height.   For the water  extraction,  the  20-g
 extract increased peak heights from 2 to 6 times
 compared  with the 1-g extracts.   For the five
 peaks measured,  methanol  produced  from 0.2 to
 0.5 times the response compared to the 1-g water
 extract.   Methanol  appears to be relatively  more
 efficient in  recovering the  late eluting peaks
 (higher boiling  compounds).   For the heated  dry
 bottle, response varied from 0.5 to 0.8 times
 the comparable peaks height  for the 1-g water
 sample.

 Sample  KC8Q4214

 This  experiment  included  all  8 treatments  but
 there were no replicate injections  or duplicate
 samples.   Therefore  the statistical  tests  were
 not very  powerful  or accurate.   Additionally,
 the results for  the  5-  and 20-g  water extraction
 appear  to be  outliers  (peak  heights  for  both
 treatments were  very similar,  contrary  to  the
 NaCl  data).   The analysis  (6  peaks)  using  all
 the data  showed  that the extraction  methods  did
 not result in significantly  different  peak
 heights (p >  0.3).   Also, the  amount  of  soil
 used  did  not  seem to affect  peak heights (p  >
 0.1).   However,  visual  interpretation  of the
 data, excluding  the  outliers,  showed  results
 very  similar  to  sample  KC804203.

 Sample  LA6120I

 The experiment included all 8  treatments but
 about 20 g  of  soil was  used for  the  heated
 bottle  (Treatment 7)  instead of the  planned  10 g
 (the  normalizing program therefore cut peak
 heights in  half  for  data analysis).  The
 analysis of all 6 peaks showed a fairly
 significant extraction  treatment effect  (Wilks'
 Lambda F=11.9, p=0.03).  The results were mixed
 though, with water giving very much higher peak
2 readings, and NaCl  giving somewhat higher peak
6 measurements.  Overall, water and NaCl
 extracts produced comparable peak heights for
 the same soil amount, methanol  produced smaller
 early eluting peaks than the 1-g water or NaCl
 extractions,  but the last peak was 2.9 times
 larger.   The  greatest peak heights were achieved
 using the heated dry bottle (11-66 times than
 the 1-g  water extract).

 The amount of soil  used  also had a very
 significant effect  on the peak heights (Milks'
 Lambda F=2678,  p=0.0001).  For all the peaks,
 the greater the amount of soil  used,  the higher
 the measured  peak.   Peak heights were from 4-12
 times higher  using  20 g  of soil  than  1 g.

 Sample LA82301

 The experiment  involved  only treatments 1,2,4,
 and 5 (1- and 5-g extractions using water and
 NaCl) because of the small  soil  volume
 available.  The analysis of 7 peaks showed that
 the different extraction methods resulted in
 significantly different  peak heights  (Wilks'
 Lambda F=240.5,  p=0.05).   The NaCl extraction
 gave consistently higher peaks  than the water
 (up to about  3  times greater peak height).

 The amount  of soil  used  also had a very
 significant effect  on the peak  heights (Milks'
 Lambda F=3304,  p=0.01).   The 5-g treatments
 generally produced  3-5 times higher peaks  than
 the 1-g  water treatment.

 Sample G-4

 The analysis  of  data (4  peaks)  from treatments
 1-6 showed  no significant difference  in  peak
 heights  for the  water and NaCl  extractions
 (Wilks'  Lambda  F=2.1,  p=0.2).   The amount  of
 soil  used had a  statistically significant  effect
 on  peak  heights  for  the water treatments, with
 larger soil amounts  resulting in  higher  measured
 peaks  (Wilks'  Lambda  F=31.7,  p=0.001).   The  20-g
 water  extraction  peaks were  2-6  times  larger
 than  the  1-g  peaks.   It was  interesting  that
 this  soil effect  was  not  apparent  with  the NaCl
 extractant.   These data,   however,  are  suspect
 since  they  had to be  reanalyzed  several  days
 after  initial  extraction  due to  analytical
 problems.

 Sample PARK LOT

 The experiment included treatments  1,   2, 4,  5
 7,  and 8  (the  20-g water  and NaCl  treatments
were omitted due  to limited  soil volume
 available) plus a modification of  the  heated
 bottle (Treatment 7) technique (10 ml of water
 added  to bottle).  Additionally, a 5-g sample
was placed in  a VOA vial   at ambient room
temperature.  The analysis of data  (Table 4)
from treatments 1-6 showed no statistically
significant differences between the water and
                                                  390

-------
NaCl extractions (Wilks' Lambda F=0.95, p=0.6).
However, the heated bottle (treatment 7, listed
as "Heated Bot" in Table 4) produced the
greatest peak heights followed by the same
treatment with 10 ml of added water ("Heated
Wet" in Table 4), and methanol.  Interestingly,
the VGA vial with 5 g of dry soil at ambient
temperature ("Ambient B" in Table 4) provided
greater peak heights than the 1-g water and NaCl
extractions.

The amount of soil (1 versus 5 g) did somewhat
affect the measured peak heights, with larger
soil amounts resulting in higher peaks (Milks'
Lambda F-90, p=0.002).

Statistical Results of Equilibrum Tests

Sample 2LA61201

Step 2 experiments for this soil included the
treatments in Table 3.  Additionally, since Step
1 data on treatments 7 and 8 were limited
(heated bottle and methanol) those treatments
were included to supplement Step 1 data.

Sample 2LA61201 was extracted using hand
shaking, the Spex Mill, vibration and sonication
as well as the same treatments over a period of
2+ hours.  Statistical analysis of the data
showed there were some differences in the
treatment effects (Wilks' Lambda F=4.2,
p<0.001), but only for 2 of the 5 peaks
analyzed.  To help compare treatment effects,
both Duncan's multiple range test and Scheffe's
multiple comparison were calculated at the 0.05
significance level.  While there were some
differences, no treatment clearly emerged as
superior.  Although not tested statistically, it
appears that initial extraction using the Spex
Mill is more efficient since peak heights
increased less over time that with the other
treatments.  The Spex Mill initially provided
the highest average peak height for all 5 peaks;
up to 2-3 times higher in some cases.  After a
couple of hours, all treatments show very
similar results.

The heated dry bottle extraction produced much
greater peak heights than the reference method
of 1 g of soil extracted by the Spex Mill.  The
response was about 8-34 times greater for the 5
peaks analyzed.  A methanol extraction treatment
was also conducted but the results were unusable
due to interferences from the methanol.

Sample LA82302

The Step 2 treatments for this soil included
hand shaking for 1 min.,  Spex Mill shaking for
5 min., sonication for 5 min. after hand shaking
for 1 min., and vibration treatment for 1 min.
after hand shaking for 1 min.  The  samples from
the hand shaking and sonication treatments were
shaken again by hand after 3 hours and
reanalysed.  The MANOVA results indicated that
there were some significant differences (Wilks'
Lambda F«2.9, p=0.04).

The individual ANOVA analysis for 3 of the 4
peaks analyzed showed no statistically
significant differences in treatment effects in
spite of some peaks being twice the size of
those in other treatments.  For one peak, the
initial Spex Mill treatment was superior all
other treatments (p=0.006).  For all peaks, the
vibration treatment produced the lowest peaks.
No other obvious patterns appeared that would
suggest one treatment was better than the
others.

Both a heated bottle and methanol treatment were
run at the same time to supplement the Step 1
data.  The methanol results showed gross
interferences from the methanol even though a
fresh bottle of HPLC grade methanol was used.
The heated bottle treatment produced
significantly greater peak heights for all peaks
than any of the treatments discussed above
(p<.01).  Compared to the initial hand and Spex
Mill extractions, the heated bottle method
produced 5-13 times higher peaks than the
initial 1-g soil sample extracted with water
using the Spex Mill.

CONCLUSIONS

The comparison of six treatment combinations of
soil amount and extracting solution (water or
NaCl), showed variable results.  While NaCl
extractions produced significantly larger peaks
for one test sample, that was the only data
demonstrating clear superiority.  Even then, the
differences were only a factor of about 3.  The
conclusion is that water is generally a superior
extractant to the saturated NaCl solution for
soil headspace analysis.

The tests of the effect of extracted soil
amounts, clearly showed that larger quantities
of  soil extracted into the same volume of
headspace produces higher headspace
concentrations.  For the 5- and 20-g soil
samples, one would expect a 5 and 29 fold
increase in headspace concentration over  a  1-g
sample  if all soil contaminants were transferred
to  the water  (ideal but not possible).  Table  5
shows the effect of sample size on VOC headspace
concentration, by sample.  The 20 g sample
produced from 3.6 to 24 times greater response
than the 1 g  sample extracted with water  using
the Spex Mill.  Overall, the increase was about
a factor of 6.5  improvement calculated using  a
geometric mean.  The  5 g soil  samples provided
between 1.8 and  7.8 times greater  response  with
a geometric mean of 3.3.  While  increased soil
                                                  391

-------
amounts do increase headspace concentration, the
increase is usually not as great as
theoretically possible.

The methanol extraction data shown in Table 6
shows that methanol is sometimes superior
(factor of 4.2) and sometimes inferior (factor
of 0.3) to the water extraction using 1 g of
soil and the Spex Mill.  Overall, there was
little reason to select methanol over water
(geometric mean = 1.1).  The disadvantages of
methanol extraction for field screening are that
it involves the transport, use, and disposal of
a hazardous chemical, and requires the
additional field steps of quantitatively
transferring an aliquot to a VOA vial after
settling.  Methanol may also interfere with
analysis on some instruments and it may be
difficult to obtain clear supernatant from some
samples.  The advantages are that the extraction
step is quick, the sample in methanol should be
relatively stable, composite samples can be
collected, the same sample used for field
screening can be sent to a laboratory for
analysis, and the extraction method is based on
a standard EPA analytical procedure.

Dry heated head space analysis method using 10 g
of soil was sometimes far superior to the 1-g
soil/Spex Mill treatment for maximizing peak
heights.  As shown in Table 6 the relative
headspace concentration for the heated bottle
treatment compared to the 1 g water extraction
varied from 0.7 to 48 times greater, with a
geometric mean of 6.8.  The disadvantages of the
heated bottle approach are the time requirement
of heating for 1 hour, the need for a water bath
(requiring electrical power) and possible
analytical problems related to condensed water
in the analytical device.  Some GC columns do
not work well with saturated vapor samples.

For extracting samples in the field, violent
shaking such as provided by the Spex Mill is
efficient but not possible without a power
supply.  Hand shaking for 1 min. seems slightly
inferior initially, but with time, headspace
concentration becomes the same as with Spex Mill
extraction.

The overall recommendation  is that several
procedures should be evaluated and compared
using site specific contaminated soils.  A
standard operating procedure that helps a user
select the best  screening procedure for the
intended use should be developed.  Also,
documentation on the effectiveness of screening
procedures versus standard  quantitative methods
is needed so that screening effectiveness can be
evaluated including a rough estimation of
detection limits.
            TABLE  5.   RELATIVE  HEADSPACE  CONCENTRATION  OF  VOC
                       for  20, 5,  and  1  g  soil  samples extracted  with  water
Sample
Size
20 g
5 g
i g
Batch 1
24
7.8
1
G-4
3.6
3.5
1
KC804203
3.6
1.8
1
KC804214
5.1
5.6
1
LA61201
7.2
2.9
1
LA82301
	
3.8
1
Park Lot
__
2.3
1
SR52702
2.4*
1.8
1
Arithmetic
Mean
8.7
3.7
1
Geometric
Mean
6.5
3.3
1
             *Sample size was 10 g., not included in means.
           TABLE 6.   RELATIVE HEADSPACE CONCENTRATION OF VOC for Heated Bottle and
                     Methanol  Treatment Compared to 16 Soil/Sample Extracted Water
SAMPLE

Heated
Bottle
Methanol
1 9
soil/H20
KC804203

0.7

0.3
1

KC804214

2.0

1.6
1

LA61201

28

0.8
1

2LA61201

26*

—
1

LA82302

7.6

--
1

PARK LOT

48

4.2
1

Arithmetic
Mean
17

1.7
1

Geometric
Mean
6.8

1.1
1

           *Not included in means.
                                                    392

-------
 REFERENCES                                                  7.

 1.     EG&G  Idaho.   Hanford Site  Sampling  and
        Analysis Data Document, Vol.  1-4,
        EGG-ES-7953,  Rough  Draft,  1988.

 2.     EG&G  Idaho.   Savannah River Sampling and
        Analysis Data Document, Vol.  1-4,                 8.
        EG&G-ES-8042, Draft, 1989.

 3.     McCoy & Assoc., Hazardous  Waste
        Consultant,  Vol. 3,  No. 2,  1985.

 4.     Barth, Del.   Literature Review on                 9.
        Preparation  of Soil  Samples for  OC
        Analysis,  Environmental Research Center,
        University of Nevada, Las  Vegas, 1988.

 5.     U.S.  EPA.  Test Methods for Evaluating
        Solid Wastes, SW-846, Vol  IB, Office of
        Solid Waste  and Emergency  Response,
        Washington,  D. C.,  1986.                           10.

 6.     Cheatham,  Richard,  Jeffrey Benson,  Jeralyn
        Guthrie, William Berning,  and Roger Olsen,
        "Rapid, Cost-Effective GC  Screening for
        Chlorinated  Pesticides and Volatile
        Organics at  CERCLA  Sites"   Monograph
        Series  Screening Techniques,  HMCRI 9300
        Columbia Boulevard,  Silver Spring,  MD, p.
        86, Undated.
        Clark,  Arthur, Moira Lataille, and Edward
        Taylor, "The use of a Portable PID Gas
        Chromatograph for  Rapid Screening  of
        Samples for  Purgeable Organic Compounds  in
        the  Field and in the Lab."   US EPA Region
        I Laboratory, 1983,  unpublished.

        Griffith, J.  Tyler.   A New Method  for
        Field Analysis of  Soils Contaminated with
        Aromatic Hydrocarbon Compounds, M.S.
        Thesis, University  of Connecticut,  Stores,
        1988.

        Spittler,  Thomas, Mary Cuzzupe, and  J.
        Tyler Griffith,   "A  Field Method for
        Determination of Volatile Organics  in  Soil
        Samples",   First International  Symposium,
        Field Screening  Methods for  Hazardous
        Waste Site Investigations, Oct  11-13,
        1988.

        SAS Institute, SAS/STAT User's  Guide,
        Release 6.03  Edition,  SAS Institute, Cary,
        North Carolina,  1988.
                                                 DISCUSSION
RANDY GOLDING: What is the correlation between what might be perceived
as standard accepted analytical practices and the results you get from these
screening methods. What's their predictive value and which one is the best?

ALAN CROCKETT: I don't real ly have a good answer for you, I'm afraid. How
predictive are they? That was to be step three which was never completed. I don't
know right now how well the CLP data would correlate with the field screening
data. That's why I'm asking people right now, who have the interest in following
these procedures, to analyze their samples two different ways and publish the
results.

THOMAS SPITTLER: Justone brief response to that, because we had used this
•water extraction for volatiles in soil for quite a few years in our own region, and
I've been in touch with other people who have been doing it, particularly some
of the people doing research at the University of Connecticut. They have found
extremely good reproducibility and very high sensitivity in extracting volatiles
out of soil. The only thing is most of this was done with synthetic soil samples.
There's a group at the Cold Regions Research Lab in Hanover, New Hampshire,
who have also done spiking of soil samples, and had done some very nice work
on extracting volatiles from soil samples. Tom Jenkins is one of the two. I think
he's got some very interesting insights on this problem of volatiles in soil
analysis. It's a major issue because so many people are out there digging up tanks,
trying to comply with state regulations on how much is too much contamination
in soil. An incredible amount of lousy data is coming  out of samples collected,
shipped off to laboratories, and reported back months later only to find that what
was obviously there when the sample was collected is no longer there when the
sample is analyzed. That's a lesson in  biodegradation and vapor loss, among
other things.
                                                         393

-------
               COMPARISON OF FIELD HEADSPACE VERSUS FIELD SOIL GAS ANALYSIS VERSUS STANDARD
                  METHOD ANALYSIS OF VOLATILE PETROLEUM HYDROCARBONS IN WATER AND SOIL
                            RANDY D. GOLDING    MARTY FAVERO     GLEP>
                                          TRACER RESEARCH CORPORATION
               THOMPSON
ABSTRACT

Twelve sites in the state of Iowa were evaluated for hydrocarbon
contamination associated with the use of existing underground
storage  facilities.   Using driven  probe technology, soil gas,
groundwater and/or soil samples were taken from each sampling
location.  Each soil  gas, groundwater and soil sample was
analyzed on location using a field mobilized gas chromatograph.
Each groundwater and soil  sample was also  analyzed by  a
contracted  laboratory using the appropriate standard method.
Correlations between the various analytical approaches were
examined.

The correlation between field analytical results and the laboratory
analytical results was 0.87 over four orders of magnitude for
twenty-five  samples.  The correlation for toluene in  soil gas
samples versus toluene in soil samples was 0.81 over five orders
of magnitude.

INTRODUCTION

In order to examine the utility of soil gas investigations and field
analyses in evaluating the contamination at underground storage
tank (UST) sites, temporary approval was granted to Tracer
Research Corporation (TRC) to apply soil gas methods at several
UST sites in Iowa. The results of the soil gas investigations were
to be compared to analyses of soil and water samples collected at
the same time. The results of this comparative study were to be
used  to  establish  appropriate  action criteria  for soil  gas
investigations used at UST site audits for insurance purposes.

BACKGROUND

The background section consists  of two  parts.   First a brief
description or definition of soil gas methodology and second a
discussion of how the instrument that makes the total petroleum
measurement (TPHC) works. This background is very important
in understanding the first part of the results section.

Shallow Soil Gas Investigation - Methodology

Shallow soil gas investigation  refers to a method developed by
TRC for investigating underground contamination from volatile
organic chemicals (VOCs) such as industrial solvents, cleaning
fluids and petroleum products by looking for their vapors in the
shallow soil gas.  The method involves pumping a small amount
of soil gas out of the ground through a hollow probe driven into
the  ground and  analyzing the gas for the presence of volatile
contaminants. The presence of VOCs in shallow soil gas indicates
the  observed compounds may either be in the ground near the
probe or in groundwater below the probe. The soil gas technology
is most effective in mapping low molecular weight halogenated
solvent chemicals and petroleum hydrocarbons possessing high
vapor pressures and low aqueous solubilities. These compounds
readily partition out of the groundwater and into die soil gas as a
result of their high gas/liquid partitioning coefficients. Once in
the soil gas, VOCs diffuse vertically and horizontally through die
soil  to die  ground surface  where  they dissipate into die
atmosphere.  The contamination acts as a source and the above
ground atmosphere acts as a sink,  and typically a concentration
gradient develops between the two. The concentration gradient
in soil gas between the source and ground surface may be locally
distorted by hydrologic and geologic anomalies; however, soil gas
mapping generally remains effective because distribution of the
contamination is usually broader in area! extent than die local
geologic barriers and is defined using a large data base. The
presence of geologic obstructions on a small scale tends to create
anomalies in the soil gas-groundwater correlation, but generally
does not obscure the broader area! picture of the contaminant
distribution.

Soil gas contaminant mapping helps to reduce the time and cost
required  to  delineate underground contamination  by volatile
contaminants. The soil gas investigation does this by outlining the
general area! extent of contamination.

How the Hydrocarbon Measurement is Made

To illustrate some of the advantages of soil gas hydrocarbon
measurements, representative chromatograms produced during
the comparative study at selected Iowa UST sites are presented
with a brief explanation.

A chromatogram is a graph of the analytical signal output by the
chromatograph.  When a sample is  analyzed  using a gas
chromatograph, it is injected into a tube through which a gas is
flowing towards a detector. The sample is carried toward the
detector by the flowing gas stream. Between the injection point
and the detector is a long tube called a column that contains a
powder or fluid that  absorbs substances  like gasoline vapors.
Gasoline is a mixture of many different substances that are very
similar. The column is most absorptive to substances with high
boiling points such as xylenes (280 degrees F) and less absorptive
to substances with low boiling  points such as methane (-260
(below zero) degrees F).  Thus  low boiling or very volatile
substances like methane flow rapidly through die column and high
boiling or not so volatile substances like xylenes are retarded by
the column and flow through die column more slowly.

When a substance exits from the column and is carried by the gas
stream into the detector, it is burned in the flame (flame ionization
detector) and an electrode senses an increase in combustible
                                                             395

-------
 substances in the (lame and the result is an increase in voltage at
 the signal output.

 The chromatogram is a plot of this voltage versus time. When
 nothing combustible is entering the detector the recorder draws
 a straight line (baseline) along the left side of the page.  When a
 combustible substance such as methane enters the detector, the
 increase in voltage from the chromatograph causes the recorder
 pen to move until the substance is completely burned. The
 recorder pen then returns and stays at the baseline until the next
 combustible substance enters the detector. This triangular shaped
 deflection is called a peak. The point in time that the peak occurs
 indicates what kind of substance it is and the time is printed on
 the chromatogram next to the peak. The distance that the pen
 moves or how tall the peak is indicates how much of the substance
 entered the detector.
    INJECT  86/98/9813:86:13 STORED TO BIN »  85
     /       ^
               •Wane
Benzene1
     "^    XyleneA
 Figure 1. Standard mixture for calibration.
 Figure 1 is a chromatogram of a mixture of methane, benzene,
 toluene, and xylene, all substances found in gasoline and most
 other petroleum fuels. Note the exit  lime for toluene is 1.01
 minutes.
 CtWKl A    INJECT 86/88/98 H:3B:89  REPUWtJ mi BIN « 9<
       n i
   .     1.56
Figure 2. Gasoline vapors from a gasoline tank. NOTE; Most of the
components of gasoline exit before toluene.
Figure 2 is  a  chromatogram of gasoline  vapors taken from a
gasoline tank.
  INJECT  86/86V»» H:I4:33  REPUWED FICM BIN I 93
    —   5.32
                     Figure 3 is a chromatogram of kerosene vapors taken from a
                     kerosene tank.
                      •JECT  06X08/90  13:35:47   REPLAYED FROM
                               AZ 1
                                                                       1 .8
                                                                                                                         .54
                                                15 .89
                                                                                          19 .01
Figure J. Kerosene vapors from a kerosene tank. NOTE:  Most of the
components exil after toluene. Kerosene is not as volatile as gasoline.
                     Figure •«.  Diesel vapors from a dlesel tank. NOTE: Almost all
                     components exil after toluene. However, there are some volatile
                     substances in dieseL
                     Figure 4 is a chromatogram of diesel vapors taken from a diesel
                     tank. Only a small portion of the constituents of the diesel vapors
                     exit before toluene.

                     Kerosene is more volatile than diese 1 and gasoline is more volatile
                     than kerosene.  These figures illustrate that the more volatile
                     substances exit the column first and the less volatile substances
                     exit the  column later. They also serve to illustrate that diesel
                     contains a substantial amount of relatively volatile substances.

                     PROCEDURES

                     The following describes in a general way the procedures used to
                     acquire the data in this study.

                     Equipment

                     Tracer Research Corporation utilized a one-ton analytical field
                     van that is equipped with  one  gas chromatograph  and two
                     computing  integrators. In addition, the van  has  two built-in
                     gasoline powered generators that provide the  electrical power
                     (110 volts  AC) to operate  all  of the gas chromatographic
                     instruments and field  equipment.  A specialized  hydraulic
                     mechanism consisting of two cylinders and a set of jaws was used
                     to drive and withdraw the sampling probes. A hydraulic hammer
                     was used to assist in driving  probes past cobbles and  through
                     unusually hard soil.
                                                                   396

-------
Soil Gas Sampling Procedures

Sampling probes consist of 7-14 foot lengths of 3/4 inch diameter
hollow steel pipe that are fitted with detachable drive tips. Once
inserted to the desired  depth, the above-ground  end of the
sampling probes were  fitted with an aluminum  reducer and a
length of polyethylene tubing leading to a vacuum pump. Gas flow
is monitored by a vacuum gauge to insure that an adequate flow
is obtained.

To adequately purge the volume of air widiin the probe, 2 to 5
liters of gas is evacuated with a vacuum pump. During the soil gas
evacuation, samples are collected in a glass syringe by inserting a
syringe  needle through  a silicone rubber  segment in  toe
evacuation line and down into the steel probe. Ten milliliters of
gas are collected for immediate analysis in the TRC analytical
Geld van. Soil gas is sub-sampled in volumes ranging from 1 uL
to 2 mL, depending on  the VOC concentration at any particular
location.

Soil Sampling Procedures

Soil samples were collected by pushing the soil gas  probes into
the ground without the  detachable drive point, thus allowing soil
to accumulate in the probe. The soil was  removed by inserting a
one-half inch diameter pipe through the  probe to push the soil
out.  Approximately 10 grams of soil and 10 mL of water were
placed in a 40 mL teflon sealed VO A bottle leaving approximately
20 mL of headspace. Each VO A was then shaken vigorously for
30 seconds before the headspace was analyzed.  This allows for
the desorption of volatile compounds into the headspace of the
vial. Headspace vapor is sub-sampled in volumes ranging from 1
uLto2mL.

Groundwater Sampling Procedures

Groundwater samples  were  collected by  driving  the hollow
probes with detachable drive points below the water table. Once
at the desired depth the probe was withdrawn several inches to
permit water inflow into the resulting hole.  Once inserted into
the ground, the above-ground end of the sampling probes were
fitted with a vacuum adaptor (metal reducer) and a length of
polyethylene tubing leading to a vacuum pump. A vacuum of up
to 24 inches of mercury was applied to the interior of the probe
and open hole for 1 to 15 minutes or until the water was drawn up
the probe. The water  thus accumulated was then removed by
drawing a vacuum on a 1/4 inch polyethylene tube inserted down
the probe to the bottom of  the  open  hole.  Loss of volatile
compounds by evaporation is minimized  when water is induced
to flow into the very narrow hole, because it can be sampled with
little exposure to air. The polyethylene tubing was used once and
then discarded to avoid cross-contamination.

Groundwater samples were collected in 40 mL VOC vials that are
filled to exclude any air and then capped with Teflon-lined septum
caps.    Groundwater  samples  were  analyzed by  injecting
headspace in the  sample container created by decanting off
approximately half of the liquid in die bottle.  Headspace analysis
is the preferred technique when a large number of water samples
are to be performed daily. The method is more time efficient for
the measurement  of  volatile organics  than direct injection.
Depending  upon  the  partitioning coefficient of  a  given
compound, the headspace analysis  technique  can also yield
greater sensitivity than the direct injection technique.

Field Analytical Procedures

A Varian 3300 gas chromatograph, equipped with a  flame
ionization detector (FID), was used for the soil gas, soil, and
groundwater analyses. Compounds were separated by a 6" by 178"
OD packed column widi O v-101 as die stationary phase at 100°F
in a temperature controlled oven.  Nitrogen was used as the
carrier gas.
Hydrocarbon  compounds  detected  in soil  gas,  soil,  and
groundwater were identified by chromatographic retention time.
Quantification of compounds was achieved by comparison of the
detector response  of the sample with the response measured for
calibration standards (external standardization). Instrument
calibration checks were run periodically throughout the day and
system blanks were run at the beginning of the day to check for
contamination in the soil gas sampling equipment. Air samples
were also routinely analyzed to check for background levels in the
atmosphere.

The  GC was calibrated for soil  and groundwater headspace
analysis by decanting 10 to 20  mL off of the known aqueous
standard so as to leave  approximately die same amount of
headspace dial is in die field samples. The bottle is then resealed
and  shaken vigorously  for  30  seconds.  An  analysis  of the
headspace in the vial determines the Response Factor (RF) which
is then used to estimate soil or groundwater concentrations.

Detection  limits for the compounds of interest are a function of
the injection volume as  well  as the detector sensitivity for
individual  compounds. Thus, the detection limit varies with die
sample size. Generally, the larger the injection size the greater
the sensitivity.  However, peaks  for compounds of interest must
be kept within die linear range of the analytical equipment. If any
compound has a high concentration, it is necessary to use small
injections,  and in some cases to dilute the sample to keep it within
linear range. This may cause decreased detection limits for other
compounds in die  analyses.

The detection limits for the selected compounds vary depending
on the conditions of the measurement, in particular, the sample
size.   If any component  being analyzed is not detected, the
detection limit for that compound in that analysis is given as a less
than" value (e.g.  ug/L).  Detection  limits obtained from GC
analyses are calculated from the current response factor, the
sample size, and the estimated  minimum peak size (area) that
would have been visible under the conditions of the measurement.

Laboratory Analytical Procedures

Groundwater samples were analyzed using analytical protocols
outlined hi EPA methods 5030 and 8015. A purge and trap step
is used to strip the hydrocarbons out of die water.

Soil samples were analyzed by a method stipulated by the Iowa
Department of Natural Resources referred to as OA-1.  The
method  is substantially  derived from  EPA  methods 5030 and
8015. Methanol is used to extract hydrocarbons  from the soil.
The Methanol extract is then diluted at least 25 to 1 in reagent
water. The water is then analyzed in essentially the same manner
as die groundwater samples.

Quality Assurance/Quality Control Procedures

Tracer  Research Corporation's normal  quality  assurance
procedures   were  followed   in  order  to   prevent  any
cross-contamination of soil gas, soil, and groundwater samples.
  • Steel  probes were used only once during the day and then
    washed with  high pressure soap and hot water spray or
    steam-cleaned   to   eliminate    the    possibility   of
    cross-contamination.  Enough probes were carried on each
    van to avoid the need to reuse any during the day.
  . Probe adaptors (TRC's special design) were used to connect
    the sample probe to the vacuum pump. The adaptor was
    designed to eliminate the possibility of exposing the soil gas
    stream to any part  of the adaptor.  Associated tubing
    connecting the adaptor to the vacuum pump was replaced
    periodical as needed during the job to insure cleanliness
    and good fit. At the end of each day die adaptor was cleaned
    with soap and water.
  • Silicone tubing (which acts as a septum  for the syringe
    needle) was replaced as needed  to insure proper  sealing
                                                              397

-------
    around the syringe needle.  The tubing does not directly
    contact soil gas samples.
  »  Glass syringes were used for one sample only per day and
    were washed and baked out at night.
  .  Injector port septa through which samples were injected into
    the chromatograph were replaced on a daily basis to prevent
    possible gas leaks from the chromatographic column.
  .  Analytical instruments were calibrated each day by analytical
    standards from Chem Service, Inc. Calibration checks were
    also run after approximately every five sampling locations.
  .  Sub-sampling syringes were checked for contamination prior
    to sampling each day  by injecting nitrogen into the gas
    chromatograph.
  .  Prior to sampling each day, system blanks were run to check
    the sampling apparatus (probe, adaptor, and 10 cc syringe)
    for contamination by drawing ambient air from above ground
    through  the  system  and comparing the  analysis to a
    concurrently sampled ambient air analysis.
  •  All   sampling   and    sub-sampling   syringes   were
    decontaminated each day and no such equipment was reused
    before being decontaminated. Microliter size sub-sampling
    syringes were  reused only after a nitrogen blank was run to
    insure it was not contaminated by the previous sample.
  •  Soil gas pumping was monitored by a vacuum gauge to insure
    that  an adequate  gas  Dow from  the vadose zone was
    maintained. A reliable gas sample can be obtained if the
    negative pressure reading on the vacuum gauge was at least
    2 inches Hg less than the maximum pressure of the pump.

RESULTS

Twelve UST sites were  evaluated by comparative  mediods
previous to May 29, 1990. In all cases the condition of the site
determined by soil gas or Geld analytical measurements agreed
with the results obtained from soil and water samples, if the soil
gas action levels recommended by TRC were used to  interpret
the soil gas data and the current Iowa UST Board action levels
were used to interpret soil or water data.  When the data from
soils or water indicated that the site was contaminated, the data
from  soil  gas  samples  also  indicated  that  the  site was
contaminated. In Iowa, a site  is considered contaminated if the
level of TPHC in soil is greater than 100,000 ug/Kg or the level of
benzene in the groundwater is greater than 5 ug/L.

There was no existing standard or action level for soil gas or for
field analytical methods. The action levels proposed for TPHC
levels were 1000 ug/L, 10,000 ug/Kg, and 500 ug/L for soil gas,
soils, and water, respectively.

In addition, the results of soil and water analyses using TRC field
analytical methods were compared to results obtained using
standard laboratory methods. The correlation between field
measurements and laboratory measurements was good  for water
samples.   The  correlation between  Geld measurements and
laboratory measurements was not as good for soil samples. This
result is  not surprising, however, since water samples can  be
homogeneous and soil samples are not. In many cases the soil
samples being compared were taken from the same bore hole but
were taken from different cores. It is very likely that discrepancies
between field and lab results for soils  represent real differences
between samples as much as disagreements between analytical
methods.

Duplicate samples were also sent to two laboratories to check the
inter-laboratory reproducibility.  The  agreement between TRC
field methods and the samples sent to the two laboratories is as
good as the agreement between the two laboratories.
Soil Gas Sampling Versus Soil Sampling: Sample Integrity
        INJECT  B5/22/W 16:54; 12  STORED TO BIN I  12
                 ri.25
Figure 5. Soil gas sample #1 taken at a depth of 6 feet. NOTE: '
peaks are much larger than the later peaks, like fresh gasoline.
The early
     A      INJECT  05/22/9017:39:31  STORED TO BIN «  22

               733
                   =5

                    1.25
Figure 6. Soil sample #1 taken at a depth of 6 feet NOTE: The early
peaks are smaller tnan the later peaks. Most of the volatile substances
were lost while handling the soU
Figure 5 is a chromatogram of a soil gas sample taken at a farmer's
coop in north-central Iowa. Notice that the early peaks are much
larger than the later peaks.  Also notice that the chromatogram
looks very similar to the chromatogram for fresh gasoline vapors
(see Figure 2). Compare the chromatogram in Figure 5 to Figure
6 which is a chromatogram from the analysis of a soil sample taken
from the same location within a few inches.  Notice that the early
peaks in Figure 6 are smaller than the later peaks. This is because
most of the volatile compounds have been lost during the handling
of the soil sample. Also die addition of water and agitation to the
soil sample prior to headspace analysis increases the signal from
lower volatility compounds.
                                        T(l BIN * 59
Figure 7. SoU gas sample #8 taken at a depth of 3 feet NOTE: The early
peaks are larger than the later peaks, like fresh gasoline.
        IN.IFC.T  M/MM U:3ft:47  STORED TO BIN »  48
                                                                     Figures. Soil sample #8 taken at a depth of3 feeL NOTE: The early
                                                                     peaks are smaller than the later peaks. Most of the more volatile
                                                                     substances were lost while handling the soiL
                                                                     A comparison of Figure 7 and Figure 8 illustrates the same
                                                                     principles. The chromatogram in Figure 7 is the result of the
                                                                     analysis of a soil gas sample from Sioux City, LA, and the analysis
                                                                 398

-------
in Figure 8 is the result of the analysis of a soil sample taken from
the same location.  Once again the analysis of TPHC in soil gas
sample is more similar to that of  fresh gasoline than that of
petroleum product left in the soil  sample after Handling and
exposure of the soil to ambient air.

Site By Site Comparison

The results from four  of the twelve skes are presented here.
These  examples are intended  to  be representative of what
occurred in the study.

• SIOUX CITY, IOWA

  This  is a relatively new site where the UST's were recently
  installed. The site is fairly clean, except for isolated
  contamination near the elbows in  the piping where the turn
  is made  toward the pump islands  (see Table 1 and Figure 9
  at sampling location 8). A leak in the pipe is indicated. There
  is excellent agreement between the  soil gas measurement
  and the analysis of soil samples at  this site. Both would have
  located the problem.  However, having the analytical
  laboratory on site allowed additional samples to be taken at
  locations 12,  13  and  14, which verified that location 8 is
  within a few feet of the release point of the hydrocarbon
  contamination.
TABLE 1
Sioux City, Iowa
05-23-90
CONDENSED DATA
SAMPLE
\-V
2-8'
W
4-8'
5-8-
t-vr
7-3-
•w
w
10-3'
11-3'
Shaded valu
TRC
TPHC
SG
ug/L
<4
<4
<4
a?
<4
<4
<4
5800
<4
2
SO

TRC
TPHC
SOIL
ug/l
nil
6
4
<2
<2

-------
• OLWEIN, IOWA
 See Table 2 and Figure 10 for a summary of the results of
 this  site investigation.   The results from the soil  gas
 investigation, the field analyses of soil samples, and the
 laboratory analyses  of soil samples all indicate that
 contamination is not  general throughout the site but is
 localized around the eastern end of the tank pit, the piping
 trench between  the building and the pump islands and
 around the southern pump island. On the map in Figure 10,
 the largest indications of contamination  are at sampling
 locations 4, 6, 7, and 9. Analysis of soil samples indicated
 problems at two locations where soil gas measurements did
 not, but analysis of soil gas samples indicated contamination
 problems in  two locations where analysis of soil samples
 indicated no problem. At this site, the same conclusions are
 reached by using either investigative method  as long  as
 multiple sample locations are examined in reaching those
 conclusions.
TABLE Z
Olwein. Iowa
05-25.90
CONDENSED DATA
SAMPLE
1-9'
2.9-
3-W
4-0'
5-9'
6-91
7-3'
i-y
9-31
10- J1
ll-J'
TRC, TPHC. SG. ug/1
<1
<4
2700
58
as
1500
24000
5
5300
<4
3300
TRC, TPHC, SOIL.
ut/l
<8
<6
2200
5400
<13
16000
%000
too
21000
^4
310
KEYSTONE TPHC
SOIL uj/1
< 5000

-------
• GAS STA TION, DBS MOINES, IOWA

  Because the surface of this tank pit was covered with grass,
  which makes the tank pit a groundwater recharge area, and
  the backfill material is native clayey soil, the soil gas survey
  was not relied upon to survey this site.  Soil samples were
  collected  at depths equal  to or greater  than  the tank
  bottoms.  The samples were  analyzed by the TRC field
  method for soils and by  two  different independent
  Laboratories. The agreement between the TRC results and
  the  Laboratory results was as good as  the  agreement
  between the two laboratories (see Table 4 and Figure 12).
  This illustrates the variability of soil samples.

  The result of the investigation is that the site is contaminated
  throughout the tank pit at the depth of the tank bottoms.
                          TABLE*
                    Gas Sution/D«s Moinc*. Iowa
                          06-11-90
                      CONDENSED DATA
SAMPLE



1
TRC,
TPHC
GW-HS
ug/L
600
NET
TPHC
GW
ug/L
140
PACE
TPHC
GW
ug/L
190





SAMPLE



2
3
4
5
6
7
TRC
TPHC
SOIL
"g/Kg
<0.5
1,200,000
62.000
26J
290,350
200
NET
TPHC
SOIL
Ug/Kg
< 15.000
860.000
43.000
< 15.000
26,000
< 15.000
PACE
TPHC
SOIL
ugrtCg
clO
40.000
HOW
<10
984.000
<10
 Shaded value* represent concentraiions above action leveti.
 TRC umple* analyzed by Tracer Research Corporation in mobile lab.
 PACE tamplet analyzed by Pace Laboratories, Coralville, LA.
 NET samples analyzed by NET Midwest Laboratories, Cedar Falls, LA.
 TPHC signifies Toul Petroleum HydroCarbons.
 SOIL signifies a «>iJ sample.
 GW signifies a groundwater sample,
 GW-HS signifies a groundwater mole analyzed using a bead space method
 EXPLANATION

  • 1      Prob* Location

	 .   Pipeline location


  o,      Fill
                                                                                                             Pump Island
                                                                                                 10
                                                                                                            '11
                                                                                                                               '12
                                                                                       3A
                                                                                    3B
                                                                                      N
                                                                                                        FIGURE 12. GAS STATION,
                                                                                                        DES MOINES, IA.
                                                                           Figure 12. Facilities and sampling locations al a gas station In DCS
                                                                           Moines, Iowa. Set Table 4 for analytical results.
                                                                   401

-------
I CONVENIENCE STORE 2, DES MOINES, IOWA

 High levels of contamination were discovered at all sample
 locations at this site (see Table 3 and Figure 11). When
 contamination was discovered early in the investigation, the
 objective  of the investigation changed to an effort  to
 determine  the extent of  the contamination.   It  was
 discovered that the contamination extends mainly eastward
 from the north end of the tank pit and  does not diminish at
 least to the border of the  property.  The extent of this
 contamination indicates that petroleum  is being released
 underground currently and has been for  some time.   The
 very high levels of hydrocarbons in the soil gas (100,000 ug/L
 or greater) over a widespread area is typical of a significant
 ongoing contamination problem.  This  is also the  ideal
 condition for the best correlation  between different
 investigation approaches. The results at this site were unlike
 the results at almost all the other sites where isolated pockets
 of minor contamination were indicated.
                          TABLES
                 Convenience Store ?2/Des Moines, Iowa
                           06-10.00
                      CONDENSED DATA
  SAMPLE
           TRC.TPHC,
            SG.ug/l
             110,000
             10,000
             25.000
             23,000
             53.000
             S2.000
             100,000
TRC.TPHC,
GW-HS, aifl
TRC.TPHC,
SOIL, ug/kt
                                 60.000
  PACE,
  TPHC,
GW-HS. ug/l
  PACE,
TPHC, SOU,
                                                    LSO.OOO
 gK«^«4 values represent concentration* above action level
 TRC samples analyzed by Tracer Research Corporation in mobile lab.
 PACE samples analyzed by Pace Laboratories, CoraMIe, IA.
 SG signifies toil gas sample.
 SOIL signifies a soil sample.
 TPHC signifies Total Petroleum HydroCarbons.
 GW signifies a groundwaler sample.
 GW-HS signifies a groundwater sample analyzed using a bead space method.
                                                             EXPLANATION

                                                              .pj     Prob* Location

                                                            	    pipallna Location


                                                              o.     HII
                                                                           PI 3
                                                                                            •P14
                                                                                                           PIS
                                                                                              P12
                                                                                        P17
                                                                                                    •P7
                                                                                                             P11
                                                                                                                     -a
                                                                                                                               P16
                                                                                       N
                                                                                               20
                                                                                                         FIGURE!I. CONVENIENCE
                                                                                                         STORE, DES MOINES, IA.
                                                                            Figure 11. Faculties and sampling locations at a convenience store In
                                                                            DCS Monies, Iowa. See Table 3 for analytical results.
                                                                      402

-------
SUMMARY RESULTS

The following  discussion  is  concerned  with the  general
conclusions that can be drawn from the data discussed above.

Comparison of Field and Laboratory Analytical Results.

Figure 13  is  a plot  of the field analytical results versus the
Laboratory analytical results for total petroleum hydrocarbons in
water samples collected at the same sampling locations. The data
are presented in log-log scaled plots because of the wide ranges
of data values.  Table 5 is a summary of a regression analysis of
the data The regression results are reported as  logarithms. As
can be seen in Figure 13, the agreement between the two methods
is good. The correlation coefficient yielded by the data is 0.87.

                         TRACER VCTSJS LABORATORr
Figure L3.Comparlson of field analytical results using a head space
method with laboratory results using a standard method for hydrocarbons
In water.  See Table 5 for a list of linear coefficients and statistical results.
TABLES
LINEAR COEFFICIENTS AND STATISTICAL RESULTS OF A
COMPARISON OF FIELD HEADSPACE ANALYSIS AND STANDARD
METHOD ANALYZING OP HYDROCARBONS IN WATER

Slope (theoretically - 1)
Intercept (theoretical • 0)
Correlation Coefficient (r)
Thmbold value for
Mauttkal ligniTiciiKe of r
(at 99% confidence levd)
REGRESSION
RESULTS
a«s
ai4
.87
0.2B
STD.
ERROR
an
aw
.10

Also the theory line, which is generated by the assumption that
the field results should be equal to the laboratory result (slope =
1, intercept = 0), is just as good a representation of the data as is
the line generated by a linear regression analysis.

Some differences between field and laboratory water analyses and
between field and laboratory soil  analyses are parallel. In both
cases the Geld analysis uses what is called a head space method,
and also in both cases the laboratory analysis uses what is called
a purge and trap  method.  For  these reasons  the TRC field
analysis for  soils  should  compare favorably  with  standard
laboratory methods for soils if identical samples are analyzed.

The correlation between field and laboratory analyses of soil
samples should not be expected  to correlate as well as water
samples.  Soil samples  are  not typically homogeneous and,
therefore, should not be rigorously considered as split samples.
Very different samples can be collected form nearly the same
location. After examining the data in Table 4, it can be seen that
duplicate soil samples collected  in this study are not reliably
similar.  Also, the  laboratory method uses methanol to extract
hydrocarbons from the soil The field headspace analysis relies
on  water to wet the  soil particles  and displace the absorbed
hydrocarbons.   A methanol  extraction of the hydrocarbons
should be more efficient than a water displacement because the
hydrocarbons are more soluble in methanol while the methanol
is able to strongly wet the soil particle.

Figure 14 compares the field and laboratory analytical results.
While there  is  more  scatter  than in  the comparison of
groundwater analyses the correlation is still strong (0.69  See
Table 6). However, the line generated by a regression analysis of
the data does not agree well with the theory une which assumes
both results should be equal.  The log mean value of the ratio of
the laboratory analytical result divided by the field analytical
result is 13, which indicates a ratio of 20. This could be related
to the better extraction efficiency of the methanol.

If, as is the case for low level analyses, the soil sample is not
extracted with methanol but is simply mixed with water and placed
in the purge vessel, the field headspace method would yield results
that are roughly equivalent to the standard laboratory method.
                                                                                                TKAOEJt VERSUS LABCMATCAY
                                                                      Figure 14. Comparison of Held analytical results using a headspace
                                                                      method with laboratory results, using a standard melbanol for
                                                                      hydrocarbons in soils. See Table 6 for a list of linear coefficients and
                                                                      statistical results.
TABLE 6
LINEAR COEFFICIENTS AND STATISTICAL RESULTS OP A
COMPARISON OF HELD HEADSPACE ANALYSIS WITH STANDARD
LABORATORY ANALYSIS OF SIMILAR SOIL SAMPLES FOR
HYDROCARBONS

Slope (theoretically - 1)
Intercept (theoretically » 0)
, , field null
(Aver.,, of to, ^^
Correlation Coeflkienl
Tbreahold value for lUtiatiol lipiifionce
of r (at the W* confidence lever)
REGRESSION
RESULT
12
-IT
U"
O.M
0.24
STD. ERROR
0.16
U

aw

• See Equation 3
" See Equation 4
Comparison of Soil Gas Measurements with Soils Analyses

If the correlation of hydrocarbons in sol gas is represented by Cg
and the level of hydrocarbons absorbed to the soil and dissolved
in the water, which is adsorbed to the soil is represented by Cg,
the ratio of tie two values could be represented by K.
Equation 1:
                  Cs
                       K  or,
K is not  an equilibrium  constant, but if the system were at
equilibrium, K would be proportional to the equilibrium constant.
Note also that K contains many variable factors such as the surface
area of the soil, the water content of the soil, the soil porosity and
the soil temperature. Since the data is plotted in the log-log scale,
the logarithm of Equation 1 becomes Equation 2:
                                                                 403

-------
           log -£ = log K or log Cg =
               Cs
Which is equivilent to Equation 3:
                  logCg=logCs

A set of hydrocarbon levels governed by a single ratio, K, plotted
on a log-log scale would fallon a line with a slope of 1.0 and an
intercept equal to log K.  In this  way the intercept  from a
regression analysis might be related to log K. See Tables 7-9.

An alternative method of obtaining log K would be to average the
logarithms of the ratios of the soil gas and the soil hydrocarbon
levels.  Equation 4:
                Average log K = \  log
                                      CS(ri)
Figure  15 is a  plot  of soil  gas levels  of total  petroleum
hydrocarbons  with   measurements   of   total   petroleum
hydrocarbons in soils taken from nearly the same locations.  The
correlation coefficient  for the data set  is 0.73, which is highly
significant. Also, the slope of 0.81 is more than the std. error
different from 1.0, but  the theoretical slope of 1.0 is within the
95% confidence interval of the regression calculated slope.

There are many reasons why the correlation is not perfect.  The
ratio of the amount of petroleum hydrocarbons in the soil gas
versus the amount adsorbed to the soil changes depending upon
the soil type and water content. The ratio also depends on the type
of fuel, length of time in the ground, or the distance between the
sampling  point and  the  original source  of contamination.
Hydrocarbons can be detected in soil gas at greater distances
from the source than in soil samples. Also, the condition, species,
and concentration of microbes in the soil have an effect. Finally,
the amount of volatiles lost from soil samples during handling
varies a great  deal  with soil types, water  content, sampling
operator, and analyst.
Figure 15. Comparison of soil gas levels of total petroleum hydrocarbons
(TPH) with TPH levels In soil samples taken from nearly the same
location. Both samples were analyzed In the field. See Table 7 for a list of
linear coefficients and statistical results.
TABLE 7
LINEAR COEFFICIENTS AND STATISTICAL RESULTS OF A COMPARISON
OF FIELD SOIL GAS ANALYSIS AND SOIL ANALYSIS BY A FIELD METHOD

Slope (theoretically » 1)
Intercept (log K)'
Average of log K**
Correlation Coefficient (r)
Threshold value for statistical significance
of r (at the 99% confidence levd)
REGRESSION
RESULT
(177
-.117
..74
.73
.27
STD.
ERROR
0,10
\-2

.10

•See Equation 3
••See Equation 4
To illustrate the effect of some of these factors, consider Figure
16. This is a plot of the level of toluene in soil gas versus the level
of toluene in soil samples.  (Note that in this analysis toluene is
not completely separated from the other hydrocarbons  in the
sample.) Toluene is a common component of gasoline that is less
volatile than most of the components  in gasoline (see Figure 2).
It is therefore less susceptible to loss during sampling. Therefore,
as expected, the correlation coefficient (0.81) for toluene  in soil
gas versus toluene in soils is better than for total hydrocarbons in
these two kinds of samples (0.73).

Also, the slope of 1.0 calculated by the regression analysis is in
excellent agreement  with theory.   The  exact  agreement  is
probably only circumstantial. The scatter in the data sets  is best
represented by the standard error of the intercept. The standard
error of the intercept for the plot comparing toluene levels (0.82)
is also reduced from 1.2, which is the standard error of the
intercept for the comparison of total hydrocarbon measurements.

Figure 17 is a plot of total petroleum hydrocarbons in soil gas
versus  total petroleum hydrocarbons  in soils as determined by
keystone laboratories. The correlation coefficient (0.63) for the
data set is highly significant and the slope of the regression line is
in excellent agreement with expectation.

Although the scatter in each of these plots that compare soil gas
levels to levek in soils is large, as great as 12 (remember that this
is the error in log K) it is easily accounted for by allowing for the
possible variation  in soil surface area alone. In other words,
clayey soils would tend to give rise to data skewed towards the X
axis and sandy soils would tend to give rise to data skewed towards
the Y axis.
                                                                       Figure 16. Comparison of soil gas levels of Toluene with Toluene levels in
                                                                       sou samples taken from nearly the same location. Both samples were
                                                                       analyzed In the Held. See Table 7 for a list of linear coefficients and
                                                                       statistical results.
TABLES
UNEAR COEFFICIENTS AND STATISTICAL RESULTS OF A
COMPARISON OF SOIL GAS LEVELS OP TOLUENE WITH TOLUENE
LEVELS IN SOIL SAMPLES

Slope (theoretically - LO)
Intercept (log K)*
Anri|e log K"
Correlation Coefficient
Threshold value for italistical tignificance
of r (it 99% confidence level)
REGRESSION
RESULTS
LO
-.079
•0,79
0,81
033
STD.
ERROR
O.U
O.E

0.12

•See equation J
" See equation 4
                                                                   404

-------
Appropriate Action Levels for Soil Gas

The cleanup action level for TPHC in soils in Iowa, and some
other states, is 100,000 ug/Kg or 100 mg/Kg.

If a vertical line is drawn through the graph (Figure 17) at the value
of 100,000 ug/Kg, it divides the data into two groups, those above
the action level, called positives and those below the action level
called negatives.  From a total of 51 samples, 15  are positives.
These soil samples are classified as contaminated above the action
level.

If a horizontal line is drawn through the intersection of the vertical
line and either the regression line or the theory line, it will intersect
the Y axis of the graph in Figure 17 at a value of approximately
200 ug/L. This horozontal fine divides the data into two sets.
Those levels above the line are called soil gas positives and those
below the line are called soil gas negatives.

When the analytical result for a sampling location falls above the
action  level for soils and for soil gas, both  methods  are in
agreement.  When the measured  levels of TPHC falls below the
action level for soils and for soil gas, once again both methods are
in agreement. The frequency of agreement between field soil gas
and laboratory soil measurements by this approach is 0.8.

If, however, the intent is to use the soil gas survey as a screening
method, and the occurence of one or more contaminated samples
causes a site to recieve a closer look, the discrepancies that cause
the greatest concern  are those in which soil gas analysis gives a
negative results when soils analysis would have yielded a positive
one. This mighr be called a false negative. An estimate of the
frequency or probability of false negatives from the data in Figure
17 is 0.08. Finally, it should be noted that soil gas samples are less
costly to  collect and analyze than are  soil samples. If therefore,
multiple soil gas samples are analyzed, the chances of a continued
false negative becomes (0.08)", in which n is the number of soil
gas samples.  Very quickly, the chances of obtaining a repeated
false negative becomes vanishingly small ((0.08)2 = 0.006, [0.008]3
= 0.0005).

After this evaluation the action level for TPHC in soil gas was set
at 1000 ug/L as a compromise between false negatives and false
positives and to compensate for the  fact that  soil gas samples
would be collected closer to the contaminated sources. The use
of 1000 ug/L as the soil gas action  level raises  the frequency of
false negatives for soil gas to 10% in  the data set in Figure 17. The
frequency of agreement of 0.8 is not affected.
TABLES
LINEAR COEFFICIENTS AND STATISTICAL RESULTS OF A
COMPARISON OF FIELD SOIL GAS LEVELS OF TOTAL PETROLEUM
H YDROCARBONS (TPHC) WITH TPHC LEVELS IN SOIL SAMPLES BY A
STANDARD METHOD"'

Slox (theoretically " 1)
Intercept flog K)*
Average log K"
Correlation coefficient (r)
TtimtioM value for statistical significance
of rlalW. confidence level)
REGRESSION
RESULTS
o.«
-13
•2,6
0.63
0.32
STD.
ERROR
0.17
L:

.012

*See equauon 3
" See equation 4
•"Metliod a OA-1 which is an lowi modification of EPA 8023
CONCLUSIONS

After a review of this data set it can be concluded that soil gas
investigations are a useful complement to soil and water sampling
approaches to site evaluations. It is important, though, to use
appropriate action levels for soil gas measurements.

Reasonable and practical guidelines can be written to ensure that
soil  gas  investigations be  used at locations for which  it is
appropriate. For those locations where soil gas measurements are
inappropriate, soil or water samples can be collected. It has been
shown that the Geld headspace analysis  of volatile petroleum
hydrocarbons yields results that correlate very well with results
from standard purge and trap methods. The correlation between
the measurement of hydrocarbon levels' in soils by the  Geld
headspace method and by OA-1 was not as good. Whether the
differences arose from heterogeneous samples or differences in
extraction efficiency was not determined.

It should be remembered that these results were obtained using
laboratory grade analytical equipment which was mobilized for
field use. Some soil gas investigations are conducted using hand
held instruments or portable gas chromatographs with little or no
temperature control of the sample stream.  Since these devices
are not as reliable, caution should be used in applying these results
to those approaches.

This study has shown that total hydrocarbons in soil and water
samples can be reliably assessed using Geld analytical methods.
Figure 17. Comparison of soil gas levels of total petroleum hydrocarbons
(TPH) with TPII levels in soil samples taken from nearly the same
location. The soil samples were analyzed by a contract laboratory using
OA-1 which is an Iowa modification of EPA method 8015. See Table 9 Tor
a list of linear coefficients and statistical results.
                                                                 405

-------
                                                           DISCUSSION
STEVE KNOLLMEYER: You seem to assume that the soil gas emanates from
the same place you collect the soil sample, rather than from the water table or soil
contamination deeper in the ground. Is that something you always see, or did you
study that at all? And secondly, did you correct any of the soil gas readings for
methane that may be there naturally?

RANDY GOLDING: The chromatograph was able  to separate the natural
methane and it wasn't included in any of these numbers. We didn't assume that
the source for the soil gas vapors were in the same region of space as the sample
collected. But, since we were evaluating this method  as a  screening tool, we
simply collected the soil gas sample from the same region of space that any other
contractor would have collected the soil sample. We were comparing whether or
not the answers would agree.

DAVID  CLIFT: What did you use for the standards?

RANDY GOLDING: This will surprise you, but we simply used a mixture of
benzene, toluene, and xylene, and we averaged the response factor for those to
calculate the total hydrocarbon number. That should over-estimate the hydrocarbon
number. We didn't use gasoline samples as our standard.

DAVID  CLIFT: That was just gasoline that you're looking at, right?

RANDYGOLDING: Well, wedidn'tknow what we were looking at necessarily.
since the tank pits all contain multiple products.

DAVID  CLIFT: You couldn't determine if they were aliphatic or aromatic then?

RANDY GOLDING: Not from the FID detector. Only from retention time. We
were limited by resolution problems because we were using this as a screening
method, therefore speed was also a factor. If we had  a water sample in later
studies, the action level for water was defined by benzene levels. The action level
for soil was defined by total hydrocarbons. And so at that point we would take
the gas chromatograph, cool the temperature down, and try to separate benzene
out from the other products. And we easily attained the 5 ppb action level for
benzene. It was easy toobtain using a headspace method. We achieved a 0.02 ppb
or 20 ppt detection limit for benzene, if we did a good separation. Then we tried
to look to see whether or not benzene was the problem if we had groundwater in
the sample.
DOUG PEERY: What was the spectral difference between your soil gas and
your soil samples? How much apart were they?

RANDY GOLDING: It varied because being able to collect the soil sample in
that probe wasn't as reliable as one would like. But typically I would say that it
was within two feet. Sometimes we had to go down the same hole repeatedly
because we didn't get enough sample. Remember we would try to collect split
samples, one for the lab, one for us. And so sometimes, not always, the sample
was collected by two different excursions into the same hole, and so it's possible
that shavings from different depths were included in different samples. That's a
disadvantage and a weakness of the study.

DOUG PEERY: If I understand correctly, you did your soil gas and sometimes
you went back into the same hole and collected your soil samples?

RANDY GOLDING: Well, we would collect soil gas after the probe arrived.
And then before removing the probe we would usually collect the  soil sample.

DOUG PEERY: Would that not bias  your results in your soil samples because
you did your soil gas first, because the volatiles would be removed?

RANDY GOLDING: The partitioning ratio that you would get from the data
was always at least as large as you would predict just by doing a batch study in
the laboratory in a very controlled environment. There isn't any evidence in the
results that this occurred. And I agree that it's a concern that should have been
raised. We didn't evacuate large volumes of ground. In order to flush the probe
adequately, you only had to evacuate soil gas from a sphere that had a radius of
a few inches, perhaps four inches. But then we would often push well beyond that
point 10 or 12 inches to collect the soil sample.

DOUG PEERY: Okay. 1 know in some procedures for soil gas there is a purging
of the probe, and then we have found, (Dr. Spittler said of his previous work) that
over a period of time there is an equilibrium that is reached.

RANDY GOLDING: Then it would recover. There is a steady state that you
reach if you keep pumping, and then  if you stop pumping it will recover.

DOUG PEERY: Right. But I take it that you did not go to the  steady state
position?

RANDY GOLDING: No, we only Hushed the probe. We only tried to get a grab
sample of the soil gas.
                                                                       406

-------
               FIELD SCREENING OF BTEX IN GASOLINE-CONTAMINATED GROUNDWATER
                 AND SOIL SAMPLES BY A MANUAL, STATIC HEADSPACE GC METHOD
James _D_1_St_uart:,  Suya Wang, Dept. of Chemistry, U-60, Gary A. Robblns, Dept. of Geology and Geophysics,
U-45, The University of Connecticut, Storrs,  CT 06269-3060 and
Clayton Wood, HNU Systems, Inc., 160 Charlemont St., Newton Highlands, MA 02161-9987.
                   ABSTRACT

     A manual,  static  headspace  GC  method  has
 been developed  and used  in  the field  for the
 screening  of  gasoline-contaminated  groundwater
 and soil samples.   This  developed,  static
 headspace  method has focused  primarily on  the
 analyses of benzene (B),  toluene (T),
 ethylbenzene  (E),  and  the three  xylene isomers
 (X) (often collectively  abbreviated as BTEX).
 However, this method also allows for  the
 determination of methyl-t-butyl  ether (MTBE),
 trichloroethylene  (TCE)  and tetrachloroethylene
 (PCE)  in the  headspace above  the aqueous layer
 as detected by  a photoionization detector  (PID)
 of a field-portable gas  chromatograph.  The
 headspace  method is performed in the  same  40-mL
 VOA vial in which  the  sample  is  collected, hence
 reducing the  possibility of sample  loss due  to
 volatilization.  Good  agreement  was found
 between the field, static headspace method,  a
 laboratory-based manual,  static  method and a
 laboratory-based,  purge-and-trap method.   The
 results of field screening  for BTEX,  MTBE, and
 PCE at several  sites in  the New  England area
 will be presented.
                   INTRODUCTION

      Simple  field  methods  associated  with the
 use  of  portable  instruments  have  been reported
 to give dependable data  while  saving  time and
 money (1,2).   These methods  can  provide  for  the
 rapid screening  of large numbers  of samples  in
 the  field,  thus  providing  for  more  effective and
 timely  site  assessment and evaluation of
 on-going remediation efforts.   In addition,
 sample  loss  due  to volatilization and/or
 bacterial alteration of  the  targeted  compounds
 compounds can  be effectively avoided.
     Reports by Spittler, et_al^ (3-5) and
Grob, et_al^ (6-7) have shown that the static
headspace method can be used as a rapid and
effective method for the analysis of various
volatile organic pollutants in groundwater and
soil samples.  Wylie found that using optimized
conditions and the same analytical instrumention
that the static headspace method can be as
sensitive and as reproducible as the dynamic,
purge-and-trap method.  He noted that the static
method is obviously more portable and better
able to be used on a variety of environmental
samples, such as soils and sludges (8).
Recently, we published a brief technical note
that described using a manual, static headspace
method for the analyses of BTEX in gasoline-
contaminated groundwater and soil samples (9).
Since the time of that publication, we have
employed the manual, static headspace method in
the field at four sites in Connecticut that have
experienced contamination due to leaking
underground storage tanks (LUST). A report of
our findings will be described in this paper.
                   EXPERIMENTAL

Instrumentation.  The field separations were
performed on a portable gas chromatograph (HNU
Systems, Model 311).  A splitless injection was
employed onto a narrow bore, 0.32 mm i.d., 25 m
in length, capillary column having a 1.0 micron
film thickness of dimethyl polysiloxane
(Nordibond NB-30, HNU Systems).  A column flow
rate of 5.0 mL/min was used.  The column's
eluent was passed to a photoionization detector
(PID) equipped with a 10.2-eV lamp whose output
was to a built-in integrator on the Model 311.
The column was set isothermally at 60°C, while
the injector's and detector's temperature were
set at 90°C.  Manual injections of the head-
                                                407

-------
space vapors were accomplished using 50- or
100-microliter, gas-right, fixed needle
microsyringes (Scientific Glass Engineering).

For comparison work, a laboratory-based manual
static headspace method was performed using the
splitless injection mode onto a capillary
column, gas chromatograph (Hewlett-Packard Model
5890A).  A megabore capillary column, 0.55 mm
i.d., 30 m in length with a 3.0 micron film of
dimethyl polysiloxane (DB-1,  J&W Scientific) was
used.  The column's eluent, at a flow rate of
8.0 mL/min, was passed through a PID (HNU
Systems, Model 52-02A) equipped with a 10.2-eV
lamp, followed by a flame ionization detector
(FID).  The output  of each detector was
displayed on an integrator (Hewlett-Packard,
Model 3396A).  The following column oven
temperature program was used: initial
temperature, 40° C, initial time 1 min,
temperature program rate, 8 °C/min, final
temperature, 190° C.  For the laboratory-based,
megabore columns, a 200-nncroliter portion of
the headspace was injected using a
250-microliter, gas-night microsyringe.

For the laboratory based purge-and-trap
comparisons,  an equivalent procedure to the one
described for Method 52A.2, "Measurement  of
Purgeable Organic Compounds in Water by
Capillary Column GC/MS", was employed with the
following equipment: a dynamic headspace
concentrator (Tekmar, Model 2000) equipped with
sixteen port, 5-mL glass sparge chambers on an
automat*.]c liquid sampler (Teckmar, Model 2016).
A cryogenic focusing interface (Tekmar Capillary
Column Interface) was used to attach the purge-
and-trap system to the splitless injection
systems of a Hewlett-Packard 5890A gas
chromatograph.  The detector system was a mass
selective detector (Hewlett-Packard 5970) with
an associated Hewlett-Packard 5895 Chem Station.
A narrow bore, 0.32 mm i.d. capillary column, 30
m in length with a 1.8 micron film thickness of
DB-624 (J&W Scientific) was used.  The column's
flow rate was adjusted to 2.0 mL/min.

In subsequent  studies, a second purge-and-trap
unit: was used which consisted of a dynamic
headspace concentrator (Teckmar, Model 4000)
with ten port, 5-mL glass sparge chambers
(Teckmar Model ALS) that: was connected to a
packed column, gas chromatograph (Perkin-Elmer,
Model 3920B) equipped with two detectors, a PID
(HNU Model 52) with a 10.2 ev lamp and a FID.
The output of both detectors were sent to a
2-pen recorder (Perkin-Elmer Model 023) and to
two integrators (Hewlett-Packard Model 3390A).
A packed column, 8 ft. long,  0.125 in. o.d.,
0.085 in. i.d. packed with 1% SP-1000 on
Carbopak B 60/80 Mesh (Supelco, Inc.) was
employed; a column flow rate of 40 mL/min was
 used.    The following column oven temperature
 was used:  a 4 min.  hold at  an initial column
 temperature of 15°C, followed by a 8&C/min.
 temperature program to 220° C, with a variable
 final  temperature hold.   With these chromato-
 graphic conditions,  the  peak for MTBE eluted at
 about  16 min and excellent:  resolution for MTBE
 and the BTEX compounds were  obtained.

 Vials  for the Static Headspace Method.   The
 gasoline-polluted groundwater samples were
 direcly collected in 40-mL glass vials (Supelco,
 Part No.  2-3278), with hole  caps (Supelco, Part
 No.  2-3283) and Teflon®-faced septa (Supelco,
 Part No.  2-3281). Prior to  the field sampling,
 100 microliters of a 24,000  mg/L aqueous
 solution  of mercuric chloride were  added  to each
 vial.    A final concentration of 60 mg/L  of
 mercuric  chloride in groundwater samples  was
 proven  to be an effective method of BTEX
 preservation against microbial degradation (9).
 Immediately after sampling,  the capped  vials
 were inverted to reduce  the  loss of volatile
 organics  and placed  in a 25.0°C water  bath if
 analysis  were to be  performed in the  field or
 packed  on  ice and returned to the laboratory
 where they were kept refrigerated at  4dC.

 Analyses  of Groundwater  Samples.  The VGA vials
 containing the  40 mL sample  of the  groundwater
 were placed in  a  25.0°C water bath in  order to
 reach thermal  equilibrium.   Then  a  1.5  In.  long,
 22-gauge needle  was  inserted  through  the  septum
 to allow air  to  enter.   Next,  a  similar needle
 attached to a  10-mL  Luer-Lock  syringe was  used
 to remove  10.0 mL of  the aqueous  phase.   The
 vial was kept  in  an  inverted  position and  shaken
 thoroughly  for  2  min.  The vial,  with the  10.0
 mL of headspace,  was again placed In  the  25.0°C
 water bath  and allowed to reach  thermal and
 phase equilibrium.   At the time  of  analysis,
 normally 50 microliters  for the  portable GC  and
 200 microliters  for  the  laboratory  GC were
 withdrawn with a  gas-tight syringe  and Injected
 into the gas chromatograph.
Analyses of Soil Samples.   First, an
identification label needs to be placed on each
clean, empty vial equipped with its individual
holed-cap and septa.  Then the weight of the
empty vial is measured to within ± 0.010 g.
Thereupon 25.0 mL of distilled water is
carefully pipetted into each vial and 100
microliters of the 24,000 rag/L mercuric chloride
added as a preservative.  The vial with its cap
and septa is then reweighed.  During the field
sampling, the soil sample with a range from 5 to
10 g. is carefully added to the vial which is
then quickly capped. The vial and its contents
are then thoroughly shaken for 2.0 min and the
entire contents reweighed.  The weight gain
corresponded to the weight of the soil sample
                                                  408

-------
 taken  for  analysis.   Depending upon  whether the
 sample is  to  be  analyzed  in  the field  or  in the
 laboratory, the  vial  is either placed  in  the
 25.0°C water bath or on  ice  for transportation
 back  to the laboratory.

           RESULTS AND DISCUSSION

 Figure la  shows  the separation obtained on  the
 HNU-311 portable gas  chromatograph for an eight
 component  aqueous standard.   The concentrations
 for the BTEX  components in the aqueous phase
 were at: the 880  ppb levels, while MTBE was  1820
 ppb, TCE 3008 ppb and PCE 1747 ppb.  Referring
 to Fig.  la, it may be seen that almost complete
 return to  baseline occured between the peak due
 to ethylbenzene  (peak 6)  and  the peak due to the
 co-elution of m- and  o-xylene  (peak  7). However,
 it should  be  noted that only  a very  small peak
 (peak  1) is obtained  by the static headspace
 method for MTBE,  even at  a significant
 concentration of 1820 ppb in  the aqueous phase.
 This is  because  the Henry's Law constant for
 MTBE is  very  small.   A preliminary estimate of
 <0.01  (in  unitless terms) has  been obtained in
 our work.  This  means that MTBE tends to remain
 in the aqueous phase  and  does  not significantly
 partition  into the headspace.  Table  1 summarizes
 average  Henry's  Law constants  for the compounds
 used in  this  paper.

 On the laboratory-based,  HP-5890 gas chromato-
 graph,  a series  of monthly calibrations had
 established that  for  MTBE and  the BTEX compounds
 that there were  linear increases  in  peak areas
 with increases in  concentration  over 3- to
 4-orders of magnitude  for both  the PID and  FID
 detectors  .   Table 2  gives values for the method
 detection limits  obtained for  the static
 headspace method  using the portable  gas
 chromatograph (HNU Systems, Model 311) with its
 associated detector and integrators  settings
 commonly employed  in  the  BTEX  analyses of
 groundwater samples.

 Table  3  presents  data  that compares  the results
 of analyses of the same gasoline  contaminated
 groundwater sample performed in  the laboratory
 by the manual, static headspace  method to an
 automated, purge-and-trap GC/MS method,
 equivalent to  EPA  Method  524.2.   It may be  seen
 that for most  of  the  comparison  of the results  .
 by the two very different analysis methods  that
 there  is in general very  good agreement.   As
 expected the  purge-and-trap method reported
 concentrations in  the lower ppb  range that were
 not detected using the headspace method.   For
 the more contaminated samples, the headspace
method tended  to give higher concentrations.
 Upon examining the chromatograms, it appeared
 that more peaks coeluted with the peaks due to
MTBE,   benzene and/or toluene in the short, about
 20 min. analysis time, in  comparison to the
longer, about 40 min. elution time employed by
the purge-and-trap.  Also, significant column
overload of the narrow-bore, capillary column of
the portable GC was observed.   As may be seen
in Fig. Ib, the integrator's plot of the
portable GC remained above scale for a
significant portion of the chromatogram, and
reported values of 0.00 ppb for both MTBE and
benzene, whereas the laboratory-based GC
reported values of 54,600 and 1260 ppm levels,
respectively for the same groundwater sample
(not listed in Table 3).

In July of 1990, the opportunity arose to
perform field analyses on groundwater samples
from nineteen monitoring wells at a State Lust
site in Westbrook, CT.  The site is located in
the middle of the small town at a busy
intersection of coastal Route //I (Boston Post
Rd.).  This is a complex site where it was
believed that gasoline-contamination may have
been caused from two leaking underground storage
tank (LUST) locations.  In January of 1989, in
response to a report of gasoline fumes in an
nearby commercial building, the State Dept. of
Environmental Protection authorized a private
engineering firm to perform investigatory and
remedial action.  After conducting geographical
studies and historical record searching, soil
gas probings and volatile organic analyses of
groundwater and soil samples were performed.  It
was surmised that a plume of underground
gasoline-contamination eminated from at least
one of the underground storage tanks (UST) and
travelled in a general NNWest direction towards
and around the northside of the commercial
building, in a line generally delineated by
monitoring wells (MW1, MW2 and MW7, Table 4).  A
total of six, underground storage tanks  were
removed from the site during the spring of 1989.
Holes were found in several tanks, and a film of
free product noted on the water in the
excavation pits.  At approximately the location
where the underground tanks were removed, a
recovery shed housing groundwater pumps and a
stripping tower were installed.   Table 4
summarizes the analyses for MTBE and BTEX
performed in the field on the HNU-311 at the
Westbrook, CT site.

In the two days of intensive sampling at the
Westbrook Site, the HNU-311 portable GC was able
to rapidly and effectively screen for MTBE and
the BTEX compounds in the many groundwater
samples.  It should be noted that analyses of
MW 4,5,9 and 17 located at a distance and to the
southwest of the expected plume  were found to
have only significant levels of  MTBE.  Fig. Ic
shows the chromatogram of the headspace for MW5,
showing only a single peak for MTBE.  In an
interesting article, Garrett, et al., have
suggested that MTBE provides an  excellent
indicator for the outer limits of a gasoline
                                                409

-------
   plume  because It spreads further and faster  than
   gasoline (10).   Also, It is expected that  MTBE
   is  not readily degraded by the subsurface
   bacteria.

   In  August  of 1990, the opportunity presented
   itself to  use the HNU-311 portable GC at a site
   known  to be contaminated with tetrachloro-
   ethylene (PCE).  In a small shopping center  in
   Weston, CT, a dry cleaning shop had been  in
   operation for a number of year.  In recent
   months, levels of PCE in the  10-100 ppb range
   had been found in drinking water wells  of  homes
   downgradient from the shopping center.   In
   one day of field work, in conjunction with State
   regulatory officers, samples  from  about five
   home drinking water wells, groundwater
   monitoring wells and soil samples  were  all
   analyzed on-site.  A shallow  hole  was manually
   dug almost at the back-door of the dry  cleaning
   shop.   Soil samples were taken at  various
   depths.  The results of the field  analyses on
   these soil samples by the manual,  static
   headspace method using both the HNU-311 and  a
   Phot ovac 10S50 (operated by a State regulatory
   personnel) and later in the laboratory  by
   purge-and-trap, packed column GC with PID and
   FID detectors are presented in Table 5.  These
   results indicated that, the manual, static
   headspace method was an excellent, field
   screening method for the determination  of  PCE
   and other such unsaturated, branched chlorinated
   solvents in soil and water samples.  It was
   found  that  less loss of TCE occured if  the
   samples were analyzed in the  field.

                      CONCLUSION

   A manual static headspace method has been  used
   in the field on groundwater and soil samples at
   several sites found  to have a wide variation in
   organic contamination.  The method works  well
   for BTEX, TCE and PCE as they are  readily
   detected in the low  ppb levels by  the PID
   detector.  The method has been found to be
   especially valuable  in that it is  portable and
   may be relatively easily performed.  But,  above
   all, the analytical  results are available  almost
   immediately to aid in evaluating any on-going
   site characterization and/or  remediation
   efforts.
                  References

1. "Field Measurements.  Dependable Data When
   You Need It", EPA/530/UST-90/003,  Office of
   Underground Storage Tanks,  Washington, DC
   20460, July 1990,  pp.  1-92.
2. Kerfoot, H.B., Amick,  E.N.,  Pierett, S.L. ,
   Lewis, T.E., Bottrell,  D.W.,  "Field Eval-
   uation of Portable Gas Chromatographs",
   EPA/600/X-89/030,  USEPA Environmental
   Monitoring Systems Lab,  Las Vegas, NV, Feb.
   1989, pp. 1-76, Appen.  A-G.
3. Spittler, T.M., "Field Measurement of PCBs in
   Soil and Sediment  Using a  Portable Gas
   Chromatograph", Proc.  of the Nat.  Conf.  on
   Management of Uncontrolled Hazardous Waste
   Sites, Hazardous Materials Control Rsch.
   Inst., Silver Springs,  MD,  1983, pp. 105-107.
4. Clay, P.F. and Spittler, T.M.,  "The Use of
   Portable Instruments  in Hazardous  Waste
   Site Characterization",  Proc.  of the Nat.
   Conf. on Management of Uncontrolled
   Hazardous Waste Sites,  Hazardous Materials
   Control Rsch. Inst.,  Silver Springs, MD,
   1983, pp. 40-44.
5. Spittler, T.M., Siscanaw,  R.J.  and Lataille,
   M.M., "Correlation Between  Field GC
   Measurement of Volatile Organics and
   Laboratory Confirmation of  Collected Field
   Samples Using the  GC/MS",  Proc. of the Nat.
   Conf. on Management of Uncontrolled
   Hazardous Waste Sites,  Hazardous Materials
   Control Rsch. Inst.,  Silver Springs, MD,
   1982, p. 57.
6. Umbreit, G.R. and  Grob,  R.L.,  "Experimental
   Application of Gas Chromatographic
   Headspace Analysis to  Priority  Pollutants",
   •L._!!iy.ll°!-l:.J>ci- _Health, A15 (5),  1980,
   429-466.   ~   •-••••-     -
7. Kiang, P.H. and Grob,  R.L.,  "A  Headspace
   Technique for the  Determination of Volatile
   Compounds in Soil",   J^ JEn^rijron L_Sci^Health,
   A21 U), 1986, 71-100.         -    -
8. Wylie, P.L., "Comparing Headspace  with Purge
   and Trap for Analysis  of Volatile
   Priority Pollutants",  Ji_Am^_Wa_ter_Works
   Assoc^, 80, 1988,  65-72.
9. Roe, V.D.7 Lacy, M.J.,  Stuart,  J.D. and
   Robbins, G.A., "Manual  Headspace Method  to
   Analyze for the Volatile Aromatics of
   Gasoline in Groundwater  and Soil Samples",
   A-nali_c_heJ5i> II> 1989>  2584-2585.
lO.Garrett, P., Moreau, M.  and Lowry, J.D.,
   "MTBE as a Ground  Water  Contaminant",
   Proc. of Conf. of  Petroleum Hydrocarbons and
   Organic Chemicals  in Ground Water, 1986.
                                            DISCUSSION
MATT STINCHFIELD: Have you looked at using other disinfectants such as
an organomercury compound, which is much more toxic and might allow you to
use much lower concentrations.
JAMES STUART: No
THOMAS SPITTLER: We have, with things like acids. And I believe in our
Cincinnati lab they did quite a bit in biodegradation studies. Herb Brass in
particular. He'll tell you a lot of the different things that he tried. Mercuric
chloride just seemed to be the simplest. It didn't change the pH, was easy to use
and it worked.
                                                    410

-------
Table 1.  Certain aqueous properties of the compounds studied, at 25 °C

Compound                  Henry's Law Constant1'2              Solubility1
                             (dimensionless)                   (mg/1)

benzene                          0.22                          1800
toluene                          0.27                           510
ethylbenzene                     0.35                           160
m-xylene                         0.29                           160
p_-xylene                         0.29                           180
o-xylene                         0.20                           190
1,1,1-trichloroethylene (TCE)    0.42                          1000
tetrachloroethylene (PCE)        0.70                           400
MTBE3'4                         <0.01                         43000

1 Calculated from the data of Mackay, D.,  Shiu, W.Y.,  "Critical review of Henry's Law
  constants for chemicals of environmental interest",  J.  Phys. Chem.  Ref. Data  10
  No. 4, 1981,  1175-1199.                                                       "
2 Calculated from the data of Ashworth,  R.A.,  Howe,  G.B., Mullins,  M.E.  and Rogers,  T.N.,
  "Air-water partitioning coefficients of  organics in  dilute aqueous  solutions",
   ^_M2!rdous_Materj.als, !§» W&&, 25-36.
3 Approximate Henry's Law constant for
                                       MTBE from our studies.
  Garrett,  P.,  et al.   (10).
  Table 2.   Method detection limitsl in ppb for the manual, static headspace method,
            obtained on the portable gas chromatograph (HNU Systems, Model 311).

  On the headspace of a 8.8-30 ppb aqueous standard using the photoionization detector.

      Benzene     Toluene     Ethylbenzene     m-and p_-xylene     o-xylene

        3.2         3.7            7.3              7.9            15.3


       MTBE         TCE            PCE

        5.0         6.9            7.4


  1  Method Detection Limits measured according to: Appendix B, Part 136, Federal Register,
     40 CFR Ch.l (7-1-88 Edition), pp. 510-512.
                                         411

-------
Table 3.  Comparison of results, In ppb, between the static headspace method
          automated purge-and-trap GC/MS method.
to an
Sample   Method     MTBE   Benzene   Toluene   Ethylbenz.   m-and  p_-xylene   o-xylene
TMW-1
TMW-1
TMW-6
TMW-6
MMW-1
MMW-1
MMW-4
MMW-4
MMW-6
MMW-6
MMW-7
MMW-7
MMW-9
MMW-9
MMW-10
MMW-10
MMW-1 1
MMW-1 1
heads pace
purge&trap
heads pace
purge&trap
headspace
purge&trap
headspace
purge&trap
headspace
purge&trap
headspace
purge&trap
headspace
purge&trap
headspace
purge&trap
headspace
purge&trap
ND
<2
ND
8.7
72
130
483
390
ND
11
co-elut
165
ND
<50
ND
1.3
ND
2.2
ND
ND
ND
ND
ND
ND
17.3
12.7
1144
709
496
330
84.5
ND
10.5
8.1
21.2
18.0
ND
ND
ND
ND
ND
ND
61.8
40.6
4320
2170
3180
2800
ND
ND
ND
ND
ND
ND
ND
ND
13
ND
861
411
28.5
15.3
1500
1420
489
379
365
361
23.3
13.5
ND
ND
ND
ND
ND
ND
1830
1220
690
315
6940
4170
4260
4430
664
944
74.8
70.2
ND
1.4
ND
ND
ND
ND
426
276
399
322
2030
1980
2080
2120
33.5
ND
ND
ND
ND
ND
                                           412

-------
 Table 4.  Results of field analyses of MTBE and BTEX at the Westbrook Site, July 1990,
           In ppb.

 MW Location   MTBE    Benzene    Toluene    Ethylbenx.   m-and g.-xylene    o-xylene
 MW1
 MW2
 MW3
 MW4
 MW5
 MW6
 MW7
 MW8
 MW9
 MW10
 MW11
 MW12
 MW15
 MW16
 MW17
 MW18
 MW19
 MW20
 B-l
 B-2
 B-3
 B-4
 B-5
 B-6
 ND =
     -MW14
Not accessible used for groundwater air stripping
ID       5950     27100             2180            6220     4690

                                     ND              ND       ND
                                     ND              ND       ND
                                     ND              ND       ND
                                   1490            4530     4400
                                     ND              ND       ND
                                     ND              ND       ND
                                     ND              ND       ND
                                     ND              ND       ND

                                    791             778      832

                                      1.1             1.6      1.9
                                      2.6             3.0      3.8
                                     ND              ND       ND
                                      9.5            10.2      8.8
                                     ND              ND       ND
ND
5950
27100
Abandoned
117
453
ND
52100
ND
1570
5.8
ND
Used for
1730
Abandoned
469
ND
0.4
ND
ND
ND
ND
ND
NI
ND
16.
0.
ND
soil
818

ND
ND
ND
1.
ND
0.3
ND
ND
12400
ND
2 ND
6 ND
ND
vapor extractions
790

3.8
1.6
ND
1 4.1
ND
              nv         nu        nu
              not sampled, had septic leachate
            4210        262       512
              ND         ND        ND
              ND         ND        ND
          813000*     33500*   123000*
                                     94             375      146
                                     ND              ND       ND
                                     ND              ND       ND
                                   5610           17500*   10800*
      Not Detected, NI • Not Integrated, * = Integration count exceed limit
Table 5.  Comparison of the analyses for PCE in soil samples, Weston, Ct.  during
          August of 1990, expressed as mg of PCE per kg of soil.
Soil
Depth
        Method & Instrument
                                  PCE Cone.
1 ft.   headspace, HNU-311        0.395
        headspace, Photovac 10S50 0.19
        purge&trap, Perkin-Elmer^ 0.135
                               Soil
                               Depth

                               2 ft.
                                                     Method & Instrument
                                                                               PCE Cone.
                                                     headspace,  HNU-311         36.8
                                                     headspace,  Photovac 10S50 12.82
                                                     purge&trap, Parkin-Elmer^  1.52
3 ft.   headspace, HNU-311        3.71
        headspace, Photovac 10S50 3.70
        purge&trap, Perkln-Elmer^-- 0.17
                                              3.5ft. headspace, HNU-311         1.08
                                                     headspace, Photovac 10S50  0.61
                                                     purge&trap, Perkin-Elmer-^ ND
    Lab analyses performed 7 and 8 days after sampling.
    in polyethylene bags with a significant headspace.
    Integrator reported over-ranged.
                                                         Soil samples improperly stored
                                           413

-------
             la.
                                                                            03 OQ QQ (D GQ OQ OQ
                                                                            a. CL. a. a. o- a. o.
                                                                            0. Q- 0. tt- 1- Q- fl-
             Ib.
               1C.
                         U
                                                                             z a. a. o m OQ m
                                                                            ue a. o, a. a. a. a.
                                                                            •-»•    a. a. ft* A.
                                                                                 C  C J3 * *
                                                                                 ;sxix
                                                                                 -  c r X X
                                                                                 a  * — i  i
                                                                                 >- m LU i o
                                                                             £"'
Fig. la. Separation  of  an  eight component mixture by  injecting the headspace of an aqueous standard
         solution onto  the HNU-311 portable gas chromatograph.
     Ib. Chromatogram showing significant column overload of  a highly gasoline-contaminated ground-
         water sample onto the narrow-bore, capillary column  of the HNU-311 portable gas chromatograph.
     Ic. Headspace sample  of a groundwater sample found to  contain only methyl-t-butyl ether  (MTBE).
                                                   414

-------
                                          Comparison of
                 Aqueous Headspace Air Standard vs STJMMA Canister Air Standard
                           for Volatile Organic Compound Field Screening

                                             H.  Wang
                                 Roy F.  Weston, Inc.,ESAT Project
                              Landmark One, One Van de Graff Drive
                                       Burlington, MA 01803
                                          W. S. Clifford
                           United States  Environmental Protection Agency
                              Region I, Environmental Service  Division
                                         60 Westview Street
                                       Lexington, MA 02173
Abstract
Introduction
   This  paper   describes  the  application  of
SUMMA  canister  and  aqueous  headspace  air
standards   for   ambient   air   volatile   organic
compound(VOC)  field screening to perform quick
on-site   analysis    using   a   portable    gas
chromatograph(GC).  Studies  were  conducted
comparing   aqueous   headspace  standards  to
SUMMA canister standards using a  portable  gas
chromatograph. A comparison of SUMMA canister
analytical  results  from  the portable GC versus
GC/MS  (gas  chromatograph/mass  spectrometer)
was provided. Research on time dependent stability
and temperature dependency of SUMMA canister
standards was also  conducted. A  Photovac 10A10
portable gas chromatograph(GC), an HP 5890/5970
gas chromatograph/mass select detector(GC/MSD)
and a Tekmar 5000 thermal desorber modified for
canister analysis were employed.
    Toxic volatile organic air pollution is a growing
concern because of its widespread presence in the
atmosphere, adversely affecting public health. There
has been  much interest in monitoring ambient  air
for these  toxic compounds.  The  United  States
Environmental  Protection Agency(U.S.  EPA) has
developed  several  methods  for measuring  toxic
organic compounds in ambient  air. These  include
collection  on solid adsorbents,  such as  Tenax GC
and spherocarb traps, as well as the  collection of
whole  air  in  suitable  canisters[l,2].  With  the
increasing interest in air analysis, field  screening for
ambient air  is becoming more important.  When
performing on-site ambient air  analysis for  volatile
organic compounds  (VOCs)  using  portable  gas
chromatography, it is important to have a suitable
standard  to  be able  to identify and  quantitate
compounds of interest.  The headspace  above a
                                                415

-------
          Headspace
10 //g/L  aqueous  standard kept  at a  constant
temperature can be used as a VOC field screening
standard to perform quick on-site air analysis using
a  portable  gas  chromatograph.  The  headspace
standard is a very simple and inexpensive technique
for  standard  preparation.  VOCs,   with  their
relatively  high vapor  pressures,  have a  natural
tendency to  migrate from water  into  air.   In a
closed  VOA(volatile  organic analysis) vial  filled
three-quarters full with  an aqueous VOC standard
                            (Figure  1), VOCs
                            will move from the
                            water into  the air
                            above   the   water
                            (headspace) until an
                            equilibrium   is
                            reached.   This air
                            above the water is
                            a perfect medium for
                            an air VOC standard
                            since it consists of
                            air and the migrating
                            VOCs   from   the
                            water.
                                                     the  mole  fraction  or  the  volume-concentration
                                                     (ppb/v) of a component i in is:
   Aqueous Solution
   Figure 1. Aqueous Headspace
   By the  ideal gas equation of state:

                   PV=nRT

   P: pressure           V: volume
   R: gas constant       n: moles
   T: absolute temperature (K)

for a single component i in a gas  mixture:

                   PiV=niRT

   Pi: Partial pressure of component i
   n;.- Moles of component  i
                                           (1)
                                                               Mi=ni/2nj=piRT/2pjRT=pi/PT
                                              (2)
     MJ: Mole fraction or volume concentration of
      component i
     PT: The total pressure of the gas mixture(=2pj)

Because the  aqueous solution  is  very diluted(10
fig/L),  it  can  be  treated  as  a  ideal  solution.
According  to Henry's law:
                                                                        pi=ki(T)Xi
                                              (3)
    kj(T): Henry's law constant;
    Xj:   Mole  fraction of solute i  in the water
          solution.

Therefore, the concentration  in the  headspace  is
shown as:

               Mi=Pi/P=ki(T)Xi/PT

Henry's law constant k; varies  with temperature(T),
therefore, the concentration in the headspace  is a
function of temperature(T), the total pressure above
the solution(P-i) and the mole  fraction in the water
solution(Xj).

                  Mi=/(Xi,T,P)

Whether an  aqueous  headspace standard can be
used for an air analysis standard depends upon the
ability to control the concentration of VOCs  in the
aqueous solution, the pressure  of the headspace and
the temperature of the standard. The first variable
Xj which reflects the concentration of a VOC in the
                                                  416

-------
water   solution  can  be  simply  controlled  by
preparing a solution with a  known concentration of
the VOC. The temperature(T) of the standard can
be  controlled  and kept constant by placing the
VGA  vial  in  an  ice-water  bath,  keeping the
aqueous  standard   solution   in  the   vial  at
approximately  CF  -  l°C. The  total pressure(Pr)
above the solution is equal to atmospheric pressure.
Relative  changes  in  atmosphere  pressure  are
negligible. Therefore, for this screening application,
the  total  pressure   (Pj) can  be  considered a
constant. With the ability to  control the variables
above,  the  aqueous  headspace  can be  used for
ambient air field screening  analysis  as an external
standard.
    SUMMA canister based sampling systems have
gained  wide  acceptance   for  the  collection  of
integrated  whole  ambient  air samples  containing
volatile organic  compounds.  Utilization  of  this
sample    collection    method    has   increased
significantly.   Some   recent  research   has   used
SUMMA  canisters as VOA  standards[4]. As  an
application, SUMMA canisters are able to be used
as  field screening  standards  as well.   Canister
standards present  the true concentration of  VOCs
within  the  can and the VOCs stored in  a canister
exhibit relatively long term stability. In addition, the
transportation  of a  canister   standard  is   easy;
therefore,  the development  of this method can be
a very  effective approach  of  SUMMA  canister
methodology and VOC field screening for  ambient
air.
   The following work is on the method  studies of
SUMMA   canister   application   and    aqueous
headspace  as  ambient  air  standards  for  field
screening of VOCs. The comparison of  aqueous
headspace standards to SUMMA canister standards
using   a  portable  gas  chromatograph  and  a
comparison of SUMMA canister analytical results
from  the  portable   GC  versus  GC/MS  were
performed.  Research  on time  dependent  stability
and temperature dependency of SUMMA  canister
standards was also conducted.

Experimental

(1)  Evaluation of aqueous  headspace standards:
     Experimentation  was performed to  determine
the  actual concentration of selected VOCs in  the
headspace above  a   10 ^g/L  aqueous  standard
contained in a closed 40ml VOA  vial filled with
30ml of aqueous  standard kept at a temperature of
0°-lt.   Analysis was performed on  a Photovac
10A10 portable gas chromatograph equipped with a
4'   1/8"   SE-30  column  and  a  photoionization
detector,  calibrated   with   a Research  Triangle
Institute(RTI) certified mixture of VOCs traceable
to NBS primary gas standards. (Table 1.)  The
Table  1. RTI Certified  Concentration  for IS Component  Mixture
    Containing  Volatile Toxic Organic  Compound
     Traceable  to NBS Primary  Gas  Standards
    Cylinder  No. ALL 21378
Compound
Vinyl Chloride
Bromomethane
Trichlorofluoromethane
(Freon 11)
Methylene chloride
Chloroform
1,2-Dichloroethane
1,1,1-Trichloroethanc
Benzene
Carbon tetrachloridc
1,2-Dichloropropanc
Trichlorocthylene
Toluene
1,2-Dibromocthane
Tetrachloroet hylene
Chlorobenzenc
Ethyl benzene
o-Xylene
Concentration (ppb/v)
5.19
5.68

5.15
4.48
4.86
5.02
5.22
5.15
5.02
0.5
0.6

0.7
OS
OS
OS
0.4
0.3
OS
5.15 ± 0.3
5.11 ± 0.3
5.19 ± 0.3
4.83 ± 0.5
5.24 + 03
5.27 ± 03
4.85 ± 0.3
5.12 t 0.5
                                                 417

-------
Photovac portable GC was calibrated by running a
syringe  blank  and  a  single  point of  the RTT
certified cylinder standard. Concentrations of VOCs
in  the  cylinder  were  approximately 5ppb. An
aqueous standard (working standard) was prepared
using 5.0 fA of  commercial  (Supelco) and EPA
repository standard mix solution (200 ^g/ml for
                 each component) diluted to 100 ml with VOC-free
                 water giving a  final concentration of 10 ng/L for
                 each component. Table 2 shows the component and
                 concentration of the  Supelco and EPA repository
                 stock standards. The working standards were stored
                 in 40ml VOA vials with zero headspace at 0° - It.
                 Before the analysis, 10 ml of water solution was
   Table 2. Components  and the
                     and of the Canister
of the Stock  Standard  Solution  (Supelco  Purgeable  A and B, and EPA  Repository)
                  Compounds
                Concentration
        Stock Solution       Canister  Standards
            (^g/ml)        0/g/m ^     (ppb)
Trichlorofluoromethane
1,1-Dichlorocthylcne
Methylene Chloride
t-l,2-Dichloroethylene
1,1-Dichloroethane
Chloroform
Bromochloromethane
1,1,1-Trichloroethane
Carbon Tctrachloride
Benzene
1,2-Dichloroelhane
Trichloroethylene
1,2-Dichloropropanc
Bromodichlorome thane
1,3-Dichloropropene
Toluene
1,1,2-Trichloroethanc
Tetrachloroethylene
Chlorobenzene
Ethyl Benzene
Bromoform
m-Xylene
o-Xylene
1,1,2,2-Tetrachloroethane
1,3-Dichlorobenzene
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
55.6
55.6
55.6
55.6
55.6
55.6
55.6
55.6
55.6
55.6
55.6
55.6
55.6
55.6
55.6
55.6
55.6
55.6
55.6
55.6
55.6
55.6
55.6
55.6
55.6
9.9
14.0
16.1
14.0
13.8
11.5
10.6
10.2
12.0
17.5
13.8
10.4
12.1
8.3
12.3
14.8
10.2
8.2
12.1
12.9
5.4
12.9
12.9
8.1
9.3
withdrawn from  the vial,  leaving  30 ml  of the
standard water solution and 10 ml headspace in the
vial. The vial  was then placed  into  an ice-water
bath to  equilibrate  for about 30 minutes  at  0" -
fC.  After   equilibration,   the   headspace   was
analyzed on a portable GC. Four trials  of analysis
were conducted separately.  The  data in Table 3
presents the analytical results from the portable GC
of selected VOC concentrations  in the headspace.
                 (2). Preparation   and   GC/MS   Calibration   of
                 SUMMA canister standards:
                     Duplicate  standards were prepared in 6L pre-
                 vacuumed  Anderson  made   SUMMA  passivated
                 canisters. The canisters  were  cleaned by vacuum
                 and heating. Canisters were vacuumed to < 5 mmHg
                 at  about lOOt  for 4 hours. 5.0 n\  of  the  stock
                 standard solution (Supelco purgeable A and B, and
                 EPA VOC repository) was injected into the each
                 canister. The canisters were then pressurized  to 30
                                                  418

-------
       Tible 3. Coaccntntiaa   of Selected  VOCs in Aqucouc HeadspaccC)
Compound
Benzene
Trichloroethylene
Toluene
Tetrachloroethylene
Ethyl benzene
o-Xylene
Concentration
Trial 1
146
163
166
212
128
98
(ppb/v)
Trial 2
161
129
153
229
154
122
Trial 3
161
141
191
177
134
95
Trial 4
135
134
127
187
115
47
Average
1S1
142
159
201
133
91
RSD%
7.3
9.2
14.5
10.2
10.6
30.1
   (•): Above a 10 /ig/L water solution  at 0°-1°C

psi with 25% relative humidity. The concentrations
of each VOC in the canister was 55.6 ng/1. Table
2 shows the VOC concentration  of the canister
standards. Following a 24 hours equilibration period
at room temperature,  the canisters were  analyzed
on a  Hewlett-Packard  5890 gas chromatograph
equipped with a 60 m megabore  capillary  column
and 5970 Mass Selective Detector. A Tekmar 5000
thermal desorber modified for canister analysis was
used  for desorbing.  The  calibration  results are
shown  in Table  4.  The  canister   cleaning  and
analysis were  performed according to EPA Method
          Table  4.   GC/MS  Certified Concentration   of Standard  Canisters
Compound
Trichlorofluoromethane
(Freon 11)
Methylene chloride
Chloroform
1,2-Dichloroelhane
1,1,1-Trichloroethane
Benzene
Carbon tetrachloride
1 ,2-Dichloropropane
Trichloroethylene
Toluene
Tetrachloroethylene
Chlorobenzene
Ethyl benzene
o-Xylene
Calc.O

9.9
16.0
11.4
13.7
10.2
17.4
11.9
12.0
10.3
14.8
8.2
12.1
12.8
118
Concentration

10.7
12.2
9.7
12.0
9.3
15.7
10.0
11.3
9.1
13.6
7.3
10.7
12.5
11.4
(ppb/v)
Can A

± 1.3
± 2.0
± 0.6
± 0.9
± 1.1
± 2.6
± 1.0
± 1.2
± 0.9
0.7
0.6
0.7
1.6
0.9
GC/MS
Can B

11.4
13.6
10.9
13.0
10.7
17.4
11.2
12.2
10.4
14.8
8.1
11.5
13.2
11.7

±
*
±
±
±
+
±
±
t
+
+
±
*
+

2.6
2.9
1.3
1.7
1.7
2.6
19
1.9
1.3
1.6
1.0
1.6
13
15
          (•): Calculated  according  to dilution.
TO-14 and EPA Region I  draft SOP for ambient
air  VOC analysis.

(3). Portable  GC analytical  results  of  canister
standards:
    The GC/MS certified canister standards were
analyzed on a Photovac 10A10 portable GC which
was calibrated with the aqueous headspace working
standard of VOCs. Table  5 shows the  analytical
results  of  benzene,  trichloroethylene,   toluene,
tetrachloroethylene, ethyl benzene and o-xylene.
                                                  419

-------
        Table S. Comparison  of GC/MS results  versus portable  GC on Canister  Standards
Concentration (ppb/v)
Compound

Benzene
Trichloroethylene
Toluene
Tetrachloroethylene
Chlorobenzene
Ethyl benzene
o-Xylenc

CALC.
17.4
10.3
14.8
8.2
12.1
12.8
12.8
Can A
GC/MS
15.7
9.1
13.6
73
10.7
12.5
11.4

PGC
12.0
8.7
14.0
9.9
10.5
12.0
10.0
Can B
GC/MS
17.4
10.4
14.8
8.1
11J
13.2
11.7

PGC
11.6
8.7
14.1
10.1
9.0
12.0
13.0
(4).Comparison of  aqueous  headspace  standards
versus canister standards:
    The certified  canister standards and  aqueous
headspace  standards were used for calibrating the
portable GC to analyze prepared air samples. Four
                              air  samples   of  different  concentrations  were
                              prepared in  SUMMA  canisters  and were certified
                              on GC/MS. The samples were then analyzed on the
                              portable GC. Table 6  shows the analytical results.
                              Based on different calibration standards, two groups
         Table  6. Comparison  of Aqueous  Headspace   Standards  vc Canister  Standards  on Portable  GC Analysis
         Compound
                   Concentration   (ppb/v)
Sample:          #1             #2             #3
         (MS)'  AH  Can  (MS) AH   CAN (MS)  AH
          #4
CAN  (MS)  AH   CAN
Benzene
Trichloroethylene
Toluene
Tetrachloroethylene
Ethyl benzene
o-Xylcne
(27)
(16)
(23)
(13)
(20)
(20)
22
24
32
17
18
22
32
29
34
14
16
20
(14)
(8.1)
(12)
(6.4)
(10)
(10)
12
8.1
13
7.6
9.0
13
18
9.7
14
6.0
8.0
12
(6.8)
(4.1)
(5.8)
(3.2)
(5.0)
(5.0)
6.1
35
6.5
5.0
63
4.5
8.8
4.2
63
4.2
55
4.2
(2.7)
(1.6)
(2.3)
(1.3)
(2.0)
(2.0)
3.3
1.6
2.4
1.3
ND"
ND
5.0
2.1
2.4
1.6
ND
ND
     (•):  Results  of GC/MS  analysis
     (*•): Non-detected
of data were obtained. The  data in the columns
under  "AH"(Aqueous Headspace) are  the  results
from the  portable  GC calibrated by the aqueous
headspace standard and those in the columns under
"CAN"(CANister) are the results from the portable
GC  using the  canister calibration. Comparisons
between the  two data groups on benzene, toluene,
trichloroethylene,    tetra-chloroethylene,    ethyl
benzene and o-xylene were performed.

(5). Stability of canister standards  and temperature
                              dependency:
                                   The  study  of  time  dependent  stability  of
                              different manufactures' SUMMA canisters has been
                              reported[3]. The study in this paper focused on the
                              time  dependent   stability  of  benzene,  toluene,
                              trichloroethylene, tetrachloroethylene, ethyl benzene
                              and xylenes. Two canister standards were analyzed
                              on a periodic basis using HP 5890/5970 GC/MSD.
                              With seven months of analytical results, there were
                              no significant variation of VOC concentrations in
                              the  canisters. Figure  2 shows the time  dependent
                                                   420

-------
   Cone. (ng/L)
            so
                    100       150
                      Time (day)
                                      20O
                                                260
      Chloroform   ~*~ Bvnzon*
      Chlorob«nz«n«—*- o-Xyl«rM
TCE      -B- Tolu
T*trachloro*thyl«na
    Figure 2. lime dependent Stability of Canister Standards

stability  for several selected compounds.
    As   a   standard   for   field   screening,   the
temperature variation is an important factor for the
canisters. The temperature dependency study  was
performed at  various temperatures ranging from
 -20C to 45°C. A canister standard  and a duplicate
were  stored at  each temperature environment for
at least three hours, then analyzed on GC/MSD
    Cone,  (ppb/v)
  14
  12
  10
  8
  6
  4 -
  2 -
  -30   -20    -10    0     10    20    30     40    60
                   Temperature (Dg. C)
                   -•- TCE          -•- Tolu.n.
   -°- o-Xyl«n»        -«- T«tr«chloro«thyl«n«
   (a). Portable CC Analysis

   Cone. (ng/L)
  60
  20
   -30   -20   -10    0     10    SO    30
                   Temperature (Dg. C)
              40     60
       Chloroform  -*- Blnan*
       Chlorobinunt~»- o-Xylint
TCE      -o- Tolinn.
Tttraohloro«lhyl«nt
   (b). GC/NS Analysis
   Figure 3. Temperature Dependency of Canister Standards
 and  on  the  portable GC.  Figure  3  shows  the
 temperature  dependency  result of  canisters. The
 data shows  those standards were  very stable over
 the  temperature range of  -20°C to 4Q°C.

 Conclusions

     According  to   both   the   theoretical  and
 experimental  results,   the   aqueous   headspace
 standard  is  a suitable  VOC  standard for ambient
 air  field  screening  analysis.  The  field  screening
 headspace standard is easy to prepare with materials
 that  are  readily  available  in any  environmental
 laboratory.  It takes very little  time to prepare and
 the  cost to prepare this type standard is minimal.
     Canister standards possess  high accuracy  for
 most of the VOCs and reflect real concentrations of
 the  VOCs inside. Canisters are easy to store and
 transport,  and  are   reusable.  The  temperature
 dependency study on canister standards showed that
 VOC concentration in canisters remain stable over
 the  normal  field  condition ambient  temperature
 range of -20^ to 4Q°C. The time dependent canister
 stability tests showed that canister VOC standards
 have long term stability. Over a seven month time
period,  VOC  concentrations  in  canister standards
remained  stable.
     Compared with aqueous headspace standards,
 canister standards  are  relatively expensive. Because
of all the  necessary accessary equipment needed for
standard   preparation,   this   method   is   only
recommended to those laboratories  which have a
canister analysis  set up.
                                                   421

-------
Acknowledgment


    The  authors  wish  to  thank  the   Regional
Laboratory of the  U. S. EPA Region I, in which all
of the  experiments were conducted.


Reference


[1].W.  A.  McClenny  and  K.  D.   Oliver.'Toxic
    Monitoring  with   Canister-Based   Systems",
    APCA 80th Annual Meeting, 1987, New York,
    New York, June 21-26, 1987.
[2]. A.  R. Gholson, etc., "Evaluation of Canister for

    Measuring Emissions of  Volatile Organic  Air
    Pollution  from Hazardous Waste Incineration",

    JAPCA 39  1210-1217,  (1989)
[3]. R.   E.  Berkley   and  K.   Kronmiller,   etc.,

    "Performance Optimization  of Photovac 10S70
    Portable  Gas   Chromatograph"     p849-854,
    Proceedings   of   the   1990   EPA/AWMA
    International   Symposium,    Raleigh,   North
    Carolina,  May 1-4, 1990.

[4]. R. W. Harrell, W. J. Mitchell, etc., "Humidified

    Canister Stability of Selected VOCs"     p726-
    730,  Proceedings  of  the  1990 EPA/AWMA
    International   Symposium,   Raleigh,   North

    Carolina,  May 1-4, 1990.
                                                     DISCUSSION
 THOMAS SPITTLER: I have to say that I saw this data just about two or three
 days before I left for this meeting, and I was literally astounded at how well that
 the simple, inexpensive, standard preparation correlated with the incredibly
 expensive, complicated GC/MS canister technology in our lab. We didn't show
 you half the slides he had of all the equipment required to prepare and analyze
 the canister standard.

 RALPH SULLIVAN: When you remove about ten cc's of liquid, did you back
 fill with air, or make any provision to clean up that air as it went in, if you did that?

 HUI WANG: No, that's just the same procedure as we prepared the headspace
 standard for soil and water screening. We just pull the air out and just use it. We
 have a mobile  lab that's relatively  clean. Is your question about the cross-
 contamination from the air getting into the headspace?

 RALPH SULLIVAN: Yes.

 HUI WANG: The mobile lab is relatively clean. It has probably very, very low
 levels of those target compounds. It's never going to affect our standard.

 THOMAS SPITTLER: We've actually looked at the lab air in our building and
 it'sascIeanastheoutsideair.Andthatmeansabout 1 ppb of benzene and toluene
 and nothing else. We've never seen much need to put a big charcoal scrubber in
 for that make-up air.
 RALPH SULLIVAN: But you do put another hypodermic syringe in there to
 dissolve the air?

 THOMAS SPITTLER: Yes, the needle is just stuck in there. Room air just goes
 in to replace the drawn water.

 RALPH SULLIVAN: The next question has to do  with the canisters. I saw
 nothing in the diagram that indicated that you put any water into the canister. Did
 you put water into the canister with a syringe?

 HUI WANG: Yes, in many canisters there's about  25% relative humidity. I
 calculated the amount of water we need and injected more water to the canister
 directly.

 RALPH SULLIVAN: Do you have mixtures of standards or single component
 standards in the vials and in the canisters?

 HUI WANG: Yes, composed of multi-components not only single components.

 RALPH SULLIVAN: So, you added various quantities of say, neat compounds
 into the vials and also into the canisters?

 HUI WANG: Yes. Actually, I used the same stock solution. We use the same
 stock solution for canister and  headspace standards.
                                                               422

-------
          Quantitative Soil Gas Sampler Implant for Monitoring
                  Dump Site Subsurface Hazardous Fluids
      Kenneth T. Lang
   Douglas T. Scarborough
U.S. Army Toxic and Hazardous
       Materials Agency
 Technical Support Division
  Aberdeen Proving Ground,
     Maryland 21010-5401
          Mark Glover
          D.P. Lucero
    IIT Research Institute
  Maryland Technology Center
       4600 Forbes Blvd.
    Lanham, Maryland  20706
ABSTRACT

     In conjunction with a
triservice  (Army, Navy, and
Air Force)  program to develop
a cone penetrometer with
associated  sensors and
detectors,  a prototype soil
gas sampling system has been
fabricated  and  functionally
tested.  The system, referred
to as TerraTrog,
quantitatively  samples
hazardous soil  gases and
vapors.  TerraTrog can be
deployed by a cone
penetrometer to depths of  100
ft far less expensively than
drilling monitoring wells.
The device  may  be permanently
implanted or may be retrieved
and deployed at multiple
locations using the cone
penetrometer.
     TerraTrog  comprises two
modules: an implant of small
dimensions  containing a
gas-permeable membrane of  high
diffusion impedance  (located
at subsurface levels) and  a
sampling and calibration
interface with  a pneumatic
manifold  (located at ground
level).  Unlike conventional
non-quantitative soil gas
sampling techniques  requiring
vacuum to operate, TerraTrog
relies only on  soil gas
diffusion for subsurface soil
gas collection  and a carrier
gas stream  flowing at a slight
positive pressure for lifting
the sample to the surface.
Because the sampling is
diffusion-limited by a
membrane of known impedance,
the sampling rate and sample
size are independent of soil
permeability.  Sampling does
not deplete the local soil gas
or vapor, guaranteeing the
accuracy of measurements made
with the device even after
long periods of continuous
sampling.  The system has a 15
rain. maximum
time-rate-of-response.
     Functional and
performance testing has been
performed with
trichloroethylene in soil,
water, and air, using a
Photovac 10S70 portable gas
chromatograph.  The implant
has been demonstrated to
operate as designed, i.e., is
diffusion-limited with implant
response directly proportional
to external soil gas partial
pressure.

INTRODUCTION
     A major problem in the
cleanup process or assessment
of sites contaminated by
hazardous waste and toxic
chemicals stems from the
paucity of information
regarding site subsurface
characteristics, composition,
and aerial and volumetric
extent.  Performing a general
prospecting or screening
                                     423

-------
survey of the site hazardous
fluids and their mobility or
stability is of significant
value in developing
preliminary overall
containment and treatment
plans (1).  A network of
relatively low-cost implanted
soil gas samplers deployed
throughout the site vadose and
peripheral zones as well as
adjacent aquifers and high
permeability strata can be
utilized effectively for site
prospecting and
characterization.  The notion
of an implanted sampler
network is a viable concept
only if waste characterization
data can be provided quickly
and inexpensively and if the
sampler can provide samples of
all hazardous soil fluids and
contaminants and can interface
at the dump site with a
variety of analyzers or
monitors and secondary
samplers.
     This soil gas sampler
system, called TerraTrog for
easy reference, is described
below and addresses the above
requirements satisfactorily,
offering features that promote
simple, low cost sampler
deployment; minimal soil
disturbance from deployment;
minimal sample extraction
during each sampling episode,
providing a correspondingly
more representative sample of
soil gases; minimal hardware;

and  small dimensions.  The
TerraTrog implant has  a  1-in.
lateral dimension and  can be
deployed by cone penetrometers
available commercially (2).
In addition, sampler operation
is independent of the  soil
permeability over a range of
0.1  to 1000 mD, and therefore,
quantitative data are  obtained
for  sandy as well as clay soil
types.  These operational
features also render the
sample obtained independent  of
sampling chamber volume, line
length, sampling pump  head,
and  corresponding pressure
losses.
     An important consequence
of using implants in the
initial prospecting process
and then progressing to the
characterization and
monitoring phases is that the
network of permanent implants
deployed initially can be used
for the life of the dump site
cleanup and monitoring tasks.
Thus, an implanted sampler is
a very attractive concept for
long-term site monitoring
requirements.

THE SAMPLER IMPLANT SYSTEM
     TerraTrog comprises two
modules:  the subsurface
implant and the surface
control interface.  Figure 1
illustrates the system.  Soil
gases enter the implant at
flow rates proportional to the
individual gas partial
pressures and the partial and
vapor pressures of dissolved
and pure liquids,
respectively, regardless of
the soil permeability.  The
soil gases are lifted to the
surface by the carrier gas
stream, which enters the
surface module and flows at a
controlled and measured flow
rate to and through the
implant and returning to the
surface as shown.  Soil gas
analysis and monitoring is
accomplished by the analyzer
or monitor attached to the
carrier gas stream return line
at the interface.  The
                          SUBSURFACE
                          lUPUNT
    FIGURE 1 - TEHRATROG SOL GAS SAMPLER SYSTEM
                                        424

-------
 analyzer/monitor  and  carrier
 gas used  are  compatible with
 all aspects of  TerraTrog  and
 the data  quality  requirements
 of the application.   In situ
 calibration of  the implant  is
 performed with  an innocuous or
 surrogate gas,  which  is
 carried to the  implant by the
 calibration gas line.  During
 one calibration episode,  it is
 anticipated that  less than
 lOOO^il of calibration gas  is
 injected  directly into the
 soil  surrounding  the  implant.
      In addition, a secondary
 sampling  device (grab bag,
 bubbler,  etc.)  may be attached
 to the interface, and soil  gas
 may be collected  in batches
 for subsequent  laboratory
 analysis.  With a sufficiently
 large carrier gas stream  flow
 rate, one or  more
 analyzers/monitors and/or one
 or more secondary samplers  can
 be attached to  the carrier  gas
 outlet of the interface and
 can be operated concurrently.

 Implant
     The  cross-sectional
 illustration  of the implant
 (Figure 2) depicts a
 cylindrical array of eight
 metal rods approximately  6  in.
 long, contained within a  1-in.
 diameter  envelope.  These are
 surrounded externally by  a
 1-in. diameter, 0.002-in.
 thick Teflon  tubular membrane.
 The rods  provide mechanical
 support for the tubular
membrane.
     Both ends of the tube  are
 sealed from the surrounding
 soil by 0-rings and a top and
bottom header.  A sealing cap
 compresses the  tube and 0-ring
 into a groove on  the bottom
 header.  A nut and the cone
 tip maintain the  sealing  cap
 in place as shown.  Carrier
 gas is introduced to the tube
 interior and returned to the
 surface through the top
header, which also serves as a
gas manifold.   There is no
pneumatic communication
between the calibration gas
and the implant interior or
the carrier gas.  Calibration
gas enters the header  and
 flows directly to the
 periphery of  the calibration
 gas diffuser  cap where it is
 injected into the region
 external and  adjacent to the
 tubular membrane.   The
 calibration gas diffuser cap
 also serves as the sealing cap
 in an identical fashion as the
 bottom sealing cap.   It is
 maintained in place by a metal
 gasket and nut.  The thread
 sealing gasket ensures that
 calibration gas does not leak
 through the threads shown.

 Surface Control Interface
      The surface control
 interface module comprises a
 panel attached to a metal
 stake embedded in the soil.   A
 pneumatic gas control network
 is mounted to the panel back
 side.   Gas connections are
 made through  fittings that
 lead to the carrier and
 calibration gas supplies and
 the respective pneumatic lines
 to the implant.   The carrier
 gas return line connects to a
 manifold  for an
 analyzer/monitor and/or for
 secondary sampling devices.
      All gas  connections are
 made at the panel  face.   All
 pneumatic lines contain inline
 filters.   The carrier gas flow
 rate is controlled by a
 precision pressure regulator
 and flow adjustment  valve.   A
 0  to  60  psig  pressure gauge
 measures  the  regulated
 pressure,  and a  rotameter
 measures  the  carrier gas  flow
 rate.   Calibration gas  flow
 rate is  controlled by the gas
 supply  pressure  regulator and
 a  flow-limiting  orifice  in the
 surface module network.   Each
 gas line  contains  inline
 pressure  relief  and  shutoff
 valves downstream  to  prevent
 overpressurization of  the
 implant and to assist  in the
 startup and checkout process.
     When not in use, the
 sample ports  are capped or
plugged.  Note that no
electrical power is required
to either maintain or operate
the interface or the implant
as described.   All power
requirements are associated
                                      425

-------
with the analyzer or monitor.

PRINCIPLE  OF OPERATION
     Implant operation is
based on a flow of soil gases
by diffusion through the
semi-permeable  tubular
membrane of  Figure 2 (3-7).
In addition,  the soil gas flow
rate is diffusion limited by
the membrane and consequently
independent  of  the soil
permeability.   As carrier gas
flows through the implant,  the
concentration of the soil gas
species at the  surface is a
ratio of the two gas flow
rates:
= (QS/QC)
                  10-
(1)
where
     [G] = soil gas  species
     concentration in  the
     carrier gas stream  at the
     interface module, parts
     per billion (ppb, v/v);

     QS  - soil gas  species
 Gamer Gas Out
                   Carrier Gas In
                     Calibration Gas n
                         Top Haider and
                         Gas Manifold
                           Cafibratjon Gas
                           DiffuSK- Can
                          CaBbration Gas
                          Cut
                        Membrane Tubs
                     	Gamer Gas TuCe


                     /	Bottom Header
                                             flow rate into the
                                             implant, std ml/min; and

                                             QC  =     carrier gas
                                             flow rate,  std ml/min.

                                             The carrier gas flow rate
                                        is  measured at the surface
                                        interface module.   The soil
                                        gas species flow rate is the
                                        product of the soil gas
                                        species membrane conductance
                                        and partial pressure in the
                                        surrounding soil.   By lumping
                                        the membrane and carrier gas
                                        parameters into  the term,  y,
                                        the soil  gas partial pressure
                                        is  related to [G]  as follows
                                        (6,7):
                                       (2)
                                       where
                                                      sg
                                                                  7 [G]
        Rgire 2 - Sampler Implant
                                            PSO = soil gas  species
                                       partialypressure in  the  soil,
                                       torr.

                                            The system response time
                                       is the sum of the time to
                                       saturate the tubular membrane
                                       with soil gas to an
                                       equilibrium concentration, the
                                       carrier gas lag time in  the
                                       pneumatic lines, and the time
                                       required from startup to
                                       establish carrier flow through
                                       the interior gas volume  of the
                                       implant to the condition where
                                       equilibrium concentration is
                                       established.
                                            For an implant with a
                                       1.000-in.  Teflon tubular
                                       membrane,  0.002-in.  thick, and
                                       a  soil  gas diffusion
                                       coefficient of 10    cm /s (8),
                                       the time to saturate the
                                       tubular membrane is
                                       approximately 128  s.   The lag
                                       time will  depend on  the inside
                                       diameter of the  carrier gas
                                       pneumatic  lines,  the depth of
                                       the implant,  and the carrier
                                       gas flow rate.   For  TerraTrog
                                       operating  with a 50-std ml/min
                                       carrier gas flow rate stream,
                                       0.0625-in.  inside  diameter
                                       pneumatic  lines, and an
                                       implant 50  ft. below the
                                       surface, the  lag time of  the
                                       system  is  73  s.  Approximately
                                       7.7 min are required  to
                                      exchange five  implant gas
                                       426

-------
volumes at a 50-ml/min carrier
gas flow rate.  It is certain
that the tubular membrane will
saturate shortly after
deployment of the implant and
long before carrier gas is
flowing and thus contributes
insignificantly to the system
response.  Thus, the time
required to obtain an
equilibrium reading at the
ground surface from startup is
approximately 10 to 11 min for
the conditions listed above.
     Equations 1 and 2
describe the soil gas species
concentration at the surface
interface for TerraTrog
operating in the dynamic
sampling mode, i.e., the
operating mode in which the
carrier gas flows continuously
through the implant.  The
implant can also be used in
the static sampling mode,
i.e., the operating mode in
which the carrier gas does not
flow (Q  = 0) for a prescribed
period of time preceding
dynamic sampling but flows
only after the equilibrium
condition described below is
attained.  Note that soil gas
flow into the implant will
continue, regardless, until
the soil gas partial pressure
difference across the tubular
membrane is 0.  At this point,
the net flow of soil gas into
the implant is 0, and an
equilibrium soil gas
concentration internally and
externally of the tubular
membrane is obtained.  After
this equilibrium is attained,
the carrier is used to lift
the soil gas accumulated in
the implant.
     For the initial
condition, where the soil gas
partial pressure, P  ,
measured at the surface
interface is 0 (P-_» the
sampler implant intergral soil
gas partial pressure is 0),
the time required to obtain
the static equilibrium
condition with the soil gas
pressure, i.e., P   = P    is
7.22 days (6,7) f§? theig
implant dimensions listed
above and a soil gas
 permeability  coefficient  of  2
 x  10~   std ml/min-cm  -torr/cm
 (9).
     P.  will be  less than P
 at equilibrium because of   g
 mixing  and subsequent dilution
 of the  soil gas accumulated  in
 the implant by the  carrier gas
 stream.  It is estimated  that
 five implant  chamber  volume
 exchanges with carrier gas
 will be required  to remove the
 soil gas accumulated  in the
 implant and transport it  to
 the surface.  Assuming
 homogeneous mixing, the
 average soil  gas  concentration
 or partial pressure measured
 at the  surface will be
 one-fifth the soil  gas partial
 pressure during the five
 volume  gas exchanges; at  a
 carrier gas flow  rate of  50
 std ml/min/ the time  period  is
 approximately 7.7 min.
     In situ  calibration  of
 the implant is performed  with
 calibration gas supplied  to
 the external  surface  of the
 tubular membrane.   A
 calibration gas stream enters
 the calibration gas inlet,
 flows through the top header
 and gas manifold  to the
 calibration gas diffuser  cap
 and through the holes in  the
 cap to  the external surface  of
 the tubular membrane.  The
 implant operates  on the
 calibration gas as  it does on
 soil gas.

 DESIGN  AND OPERATIONAL
 CONSIDERATIONS
     Aside from fundamental
 system  analytical and
 monitoring performance
 requirements, the system
 design  constraints  are
 established by reliability and
 service life  requirements and
 deployment flexibility.
 TerraTrog reliability
 corresponds generally and most
 importantly with the
 exigencies of maintaining the
 relationship  of soil  gas
 species  partial pressure, P  ,
 and the  measured soil  gas  sg
 species  concentration,  [G],
described by equation  2.
Adherence to this relationship
 is predicated on the  design
                                       427

-------
and operational integrity of
the tubular membrane and the
pneumatic lines leading to the
surface.  It is essential that
the soil gas flow into the
implant by a diffusion process
only, and therefore, the
tubular membrane must be free
of tears, punctures, and pin
holes and other pneumatic
leaks.  Thus, pre-  and
post-assembly inspection of
the tubular membrane as well
as an implant leak  check is
required.  The tubular
membrane must not be damaged
during the deployment and
operational processes.
     Relatively inert implant
fabrication materials, e.g.,
stainless steel and Teflon,
are used, because after
deployment, every external and
perhaps some internal
component or surface will be
exposed to chemical and
physical attack.  It is of
prime importance that the
tubular membrane material be
chemically and physically
inactive with the soil and
with benign as well as
hazardous soil fluids to
ensure that the tubular
membrane material diffusion
conductance (3-5) remains
unchanged during the life of
the implant.  Teflon of any
form is regarded as the most
suitable material for the
tubular membrane.   The
membranes currently in use are
fabricated of 2 mil (0.002
in.) thick Teflon film.
     The maximum typical soil
gas sample flow rate into the
implant is approximately 0.01
std uL/min for arbitrary but
realistic conditions.  In a
relative sense, it  is a very
small sample,  yet large enough
to produce a [G] for many soil
gas species within  the
response range of many gas
phase analyzers/monitors that
may be attached to the
interface.
     There are three important
aspects to the relatively
small sample size or flow
rate:  First,  the disturbance
to the soil is minimized;
consequently,  a more
 representative  sample  is
 obtained independent of soil
 fluid conditions.   Second,  for
 soil  strata  in  and  around dump
 sites,  the soil gas flow rate
 into  the implant  is diffusion
 limited by the  tubular
 membrane and is independent of
 the gas permeability of the
 surrounding  soil.   Thirdly  and
 most  importantly, these
 conditions lead to  a
 quantitative measurement of
 the soil gas partial pressure.
      The soil gas flow rate is
 proportional to the soil gas
 species pressure  only, without
 regard to the form  of  the
 sample, i.e., gas phase,
 liquid  phase, or dissolved
 gas/liquid phase.   For
 example,  the implant can
 obtain  information  regarding
 dissolved trichloroethylene
 (TCE) in  water  or TCE
 saturated in water, and
 insoluble gases contained in
 the water.   Furthermore, the
 implant also functions as
 described immersed  completely
 in an aquifer or other body of
 water or  liquid.

 SOIL GAS  ANALYSIS AND
 MONITORING
     In a relative  sense, the
 actual  soil  gas monitoring and
 analysis  of  the transport gas
 output  stream from  the
 interface panel is  the most
 simple  and direct procedure of
 the entire system.  A variety
 of analyzers, monitors, and
 secondary sampling  devices can
 be used singularly  or
 simultaneously.  The user,
 however, must establish
 preliminary  requirements for
 the target species  and the
 lower detection limits of the
 analytical devices
 contemplated, i.e.,  it is
 essential to  consider the
 analyzer/monitor performance
 specifications to specify and
 adjust the operating
 conditions of TerraTrog.

TERRATROG PERFORMANCE
     TerraTrog response to
trichloroethylene (TCE) in
soil,  water and air has been
characterized in the
                                       428

-------
laboratory.  The test
equipment was arranged in a
configuration identical to
Figure 1, with the implant
suspended in a specially built
test vessel.  In separate
tests, soil, water and air
with measured concentrations
of TCE were contained in the
vessel to simulate implant
field deployment conditions.
The vessel was pneumatically
sealed to prevent loss of TCE
vapors except by diffusion
through the implant gas
permeable membrane and
subsequent removal on the
implant carrier gas stream.
     The response
characteristics of the implant
were determined by monitoring
the concentration of TCE
within the test vessel
external to the implant and by
monitoring the carrier gas
entering and exiting the
implant in dynamic sampling
mode, and by direct analysis
of the implant contents in
static sampling mode.  In all
tests, it was verified that
the concentration and
therefore partial pressure  of
TCE in the soil, water or air
external to the implant
remained stable and constant
during the period of the test.
All concentration measurements
were made  on a Photovac  10S70
portable gas chromatograph
with  a
10-m  CPSIL5CB capillary  column
and a photoionization
detector.  The gas
chromatograph was calibrated
using air  standards prepared
from  aqueous solutions  of  TCE
of known concentration  (9).
      Representative data  for
the TerraTrog
time-rate-of-response  in
static  sampling mode  to
dissolved  phase TCE  in water
is  shown in  Figure  3.   The
internal concentration  reaches
equilibrium with  the  external
TCE concentration in  7  days
 (168  hr.).  This  is  in
excellent  agreement with the
calculated equilibrium
response time,  7.22  days
 (6,7).
     Representative dynamic
sampling mode data for the
  I©
SMpLi
An«lyt»r;
                  004-2021
                  TCE 20 ppb- (20 u»/l)
                  IB W«t«r. 22 C
                  Static
                  Fbot«vm« 10370
     Figure 3  Implant Tim*-R4t*-of-Reiponse
           to TCE In W.tar
implant response,  [G],  to  TCE
as a function of inverse
carrier flow rate,  1/QC/ is
shown in Figure 4.  A
multipoint calibration  curve,
Figure 5, shows the
relationship between implant
response, [05] / and  external
dissolved phase TCE
concentration in water.
Figure 4 show the implant
response is linear  regardless
of whether the implant  is
deployed in soil, water, or
air. Additionally,  the  flow
rate of TCE into the implant,
Q , is constant when the
implant is sampled  in dynamic
mode in an environment  of
constant external TCE
concentration.  The sample
flow rate, Q , is relatively_5
small, ranging from 9.2 x  10~
std./Jtl/min for dissolved
phase TCE at 100 ppbm external
concentration in water, to 3.7
xlO   std. ul/min for gas
phase TCE at 227 ppmv external
concentration in air.   Figure
5 shows that the implant
response, and hence Q , varies
linearly with the extlrnal TCE
concentration and therefore
with the external TCE partial
                                      429

-------
pressure.   The data
demonstrate that Q  is
dependent  only on the
permeability, P , of the
implant  gas permeable
membrane,  and the external TCE
partial  pressure, or
concentration (6,7).
Therefore, the implant
operation  is diffusion-limited
by the implant gas permeable
membrane and the implant
response is directly
proportional to the external
TCE partial pressure in soil,
water, or  air, exactly  as
      T.rrtTrcu StH: 003-2001
      EnviranMnt:  TCZ 21 ppH
             Sail rhea.
      S«*plln| Hod.: DrnMuc
      Carrier CM:  Air
      Analyser:   Photov.i
      Figure 4  In-plane Response Co TCE In Soil
     TunTiai S/tt:     003-2001
     Envlrotwrnt:     TCE In H«t«t. 22 C
               Con««ntr*tion •• Net
     SMpUn« Hod*:     Dynmic
     Carrlvr 0«f:     Air
     Cfttift OM Flow UK*: 40 iU/.un
     An«lyt«r:       PhetevH 10S70
          ' Inplwit C«libration Fttctor K:

        O           W.t*E TCE Co«c4>ntr«tlon

      . /             lopltint R»ipon.», |C]

                    *2.92 i
                300      030      600

             VlattiT TCE CtMfttMtrctlM., ppb*>
 described by equations 1 and
 2.
      The multipoint
 calibration curve,  Figure 5,
 can be used with  the implant
 to  directly measure the TCE
 concentration  in  contaminated
 groundwater in the  field.  For
 example, a user would deploy
 the implant to the  desired
 depth in a monitoring well or
 other body of  water and
 establish a carrier flow rate
 of  40 ml/min.   The  implant
 response would be measured
 using a conventional gas-phase
 TCE analyzer,  such  as the
 Photovac 10S70.   The implant
 response would then be located
 on  the vertical axis of the
 calibration curve,  and the
 corresponding  TCE
 concentration  in  the
 groundwater read  off the
 horizontal axis of  the curve.
      Since the implant
 response is shown to be linear
 in  the multipoint calibration
 curve of Figure 5,  it may be
 replaced with  a single point
 calibration which yields a
 linear calibration  factor, K:
                                                               [Ce]/[G]
                                                 where
   Figure 5  calibration Curve for Inplanc Response
         Co TCE In Uaccr versus TCC Concencration
      K    =     implant linear
 calibration factor,  ppbm/ppbv
      [C 1 =     external TCE
 concentration,  ppbm
      [G]  =     implant
 response, ppbv

      The linear calibration
 factor may be used exactly as
 the multipoint  calibration
 curve to make direct field
 measurements.

TEST PLANS

Laboratory and  Controlled
Field Testing
      Further laboratory and
field testing is in  progress
at  the National Institute for
Petroleum and Energy Research
(NIPER).   Laboratory testing
is  planned to demonstrate the
quantitative TerraTrog
sampling characteristics over
a range of controlled soil
                                          430

-------
permeability from 0.1 to  1500
mD.
     TerraTrog will be
deployed in soil by a cone
penetrometer to depths
approaching 50 ft to develop
deployment procedures and
techniques, optimize the
pneumatic line dimensions and
configuration, and determine
the  effects of the surrounding
subsoil mass on the implant
operational integrity.
Optimum deployment procedures
will be developed regarding
the  mechanical aspects, the
pneumatic tubing, and grouting
and  sealing the bore hole.   In
addition, the in situ
calibration scheme described
above  will be implemented to
develop optimum calibration
procedures.  This work will
determine the utility and
validity of the in situ
calibration.  The TerraTrog
performance after calibration
will be assessed with
calibration gas injections.
CONCLUSION
      The development  of  the
TerraTrog is viewed as having
real  potential for future use
in the evaluation of  hazardous
waste sites.  The potential
utility of the device includes
not only initial site
assessment, but possibly of
more  importance, its  use in
the routine monitoring that is
essential to the long term
assessment of a site before,
during and after remedial
activities are accomplished.
Although initially designed to
be used in a cone
penetrometer, the utility  of
the device for routine
groundwater monitoring is  also
recognized due to its small
diameter and ability to
descend down standard well
casings.

ACKNOWLEDGEMENTS
     The authors thank Susan
Hendrickson of IIT Research
Institute for valuable
technical critique and editing
of this manuscript.
      Funding for the initial
phase of this work was
provided by the Program
Manager for Rocky Mountain
Arsenal, Gregory Mohrman,
Project Monitor.

REFERENCES

(1) S.C.  Golian,  E.E.  Dodge,  and
    B.     Bixler,     " Conducting
    Remedial    Investigations    on
    Feasibility    Studies      Under
    CERCLA,"    Superfund    '88
    Proceedings of the  9th  National
    Conference.          Hazardous
    Materials    Control    Research
    Institute,  Silver Spring,  MD,
    p. 1,  1988.

(2) L.R.  Taylor  and  N.   Berzins,
    "Subsurface       Contamination
    Screening  by  Combined   Soil
    Gas/Ground water        Survey
    Procedure,"    Superfund    '88
    Proceedings of the  9th  National
    Conference.          Hazardous
    Materials    Control    Research
    Institute,  Silver Spring,  MD,
    p. 158, 1988.

(3) R.M.  Barrer,  Diffusion In  and
    Through   Solids.    Cambridge

    University    Press,    London,
    Chapter 1, 1981.

(4)  D.P.    Lucero,    "Performance
    Characteristics   of   Permeation
    Tubes,"   ApfllyHnal   Chemistry
    43. p. 1744, 1971.

(5)  D.P.  Lucero,   "Ultra  Low-level
    Calibration Gas  Generation  by
    MuM-Stage            Dilution
    Techniques,"  Ca1i^raH"n in AIT
    Monitoring,  ASTM  STP  598,
    American  Society  for  Testing
    Materials,    Philadelphia,   PA,
    pp. 301-319, 1976.

(6)  D.P.  Lucero,  "Soil Gas  Sampler
    Implant  for   Monitoring   Dump
    Sltp   Subsurface    Hazardous
    Fluids," Proceedings of the 6th
    National     Conference      on
   Hazardous
 Wastes
 and
    Hazardous
Materials.
    Orleans,    Louisiana,
    12-14, 1989.
 New
April
                                       431

-------
        (7)  D.P.     Lucero,    "A    Soil   Gas
             Sampler  Implant  for  Monitoring
             Dump        gjt-g        Subsurface
             Hazardous   Fluids,"   Hazardous
             Mafr-pTi'als  Control _3,  5,  p.  36,
             1990.

        (8)  D.P.    Lucero,   "Performance  of
             Membrane-Covered   Polar! graphic
             Detectors,"                Analy
             Chemistry 40.  p.  707, 1968.
        (9)  T.   Spittler,   unpublished  notes
             of      December      6,       U.S.
             Environmental            Protection
             Agency  Regional  Laboratory,   60
             Westview     Street,     Lexington,
             MA, 1989.
                                                        DISCUSSION
ROBERT LUGAR: With the TerraTrog probe, have you considered how to
avoid cross-contamination between holes orcarrying it toa lower depth if you're
doing a depth profile?

KEN LANG: We've considered that a potential problem, but since we don't
have any experience with actually pushing it into the ground yet, we're not sure
whether or not it is going to be a problem, and as such we're not going to try to
engineer that problem out until we see whether or not it does happen. I really don't
think it's going to be a problem because, as was presented in the Navy's work,
on some sites that have fairly high contamination, the device seemed to be rather
self-cleansing as it was pushed down in  through the soil. They made that
determination based on the sapphire window, and the fact that their readings of
contamination dropped off rather rapidly. And we're hopeful that that's going to
happen with this device. The membrane does not come in direct contact with the
soil, so I'm hoping that that will not turn out to be a problem.

SKIP WEISBERG: When you inject the carrier gas in to drive the soi 1 gas out,
do you then  feed that directly into the GC, or  do  you add an  additional
supplemental  carrier?

KEN LANG: The testing that we've done so far involves feeding that carrier gas
directly into the gas chromatograph. We may find out later on, especially if our
aim is toward reducing detection limits, that there may be other things that we
have to do. Calculations suggest that in order to cleanse the chamber, the interior
of the membrane, you would want to get at least five volume changes. I think we
calculated that the internal volume of that is somewhere around 80 cc's. So, we
already believe that for some instrumentation we're going to have to preconcentrate
that material before it's actually introduced into the analytical instrument.

SKIP WEISBERG: I was wondering if you would suffer a dilution in loss of
sensitivity due to the injection of a carrier gas down into the soil?
STEVE KNOLLMEYER: I was wondering if the tenon membrane acts to
cause certain molecules to diffuse more through the membrane and retain some
of them outside just because of their molecular size?

DANIEL LUCERO: Yes, the conductance of the membrane varies with the
analyte. For example, benzene has a higher permeability coefficient than does
TCE. And of course what that means is that the system has to be calibrated if you
want accurate results. The variations are not great for organic molecules. But if
you want more accurate results, then you do have to calibrate. And if you implant
the device and start seeing a gas that you haven't calibrated against, then that was
the reason for the inside calibration ports that you have in there. There are no
holes in the membrane; it's not a porous membrane. It works by solution/
dissolution. And as such there is a difference in  impedance from analyte to
analyte.

THOMAS SPITTLER: Have you ever tried to push the cone penetrometer
through glacial till very far?

KEN LANG: No, we haven't pushed the penetrometer at all. That's the next step.

THOMAS SPITTLER: I don't like to be rough about it, but I think you're in
for a shock.

KEN LANG: We have not pushed the penetrometer with the TerraTrog on it. We
have made lots of pushes of the penetrometer with other devices on it, but not in
glacial till. There are obviously some limitations. One that would come to mind
right off would be once you hit bedrock  that's as far as you go. Chert is also a
problem, and we have tried to push through chert materials and it doesn't work
there, either. So, there are some limitations of the cone penetrometer that we're
already aware of.
                                                                   432

-------
               TUNABLE CO2 LASER-BASED PHOTO-OPTICAL SYSTEMS FOR
                  SURVEILLANCE  OF  INDOOR WORKPLACE  POLLUTANTS
         Harley V. Piltingsrud
  National  Institute  for Occupational
Safety and Health, 4676 Columbia Parkway
         Cincinnati, Ohio   45226
Disclaimer;  Mention of company names or
products does not constitute endorsement
by    the    National    Institute    for
Occupational Safety and Health (NIOSH).
Abstract

This   paper   describes   research   work
recently   conducted  at   the   National
Institute  for  Occupational  Safety  and
Health  (NIOSH)  on the use of  CO2  laser
lidar   for   surveillance   of   indoor
workplace pollutants. Long-term goals of
this  program  included  an  ability  to
real-time map the spatial distributions
of a variety  of  pollutants in a workplace
atmosphere.  In pursuit  of this  goal,  a
rapidly-tunable, Q-switched, low pressure
laser  was developed  for time-of-flight
aerosol    backscatter    differential
absorption measurements.   This  enabled
effective atmospheric freezing as well as
the  evaluation  of possible methods  for
suppressing  the  severe  multi-reflection
scatter   problems    common    in   work
environments. Experiments were performed
to   determine   the   practicality   of
utilizing aerosol backscatter methods in
the desired lidar system.   The results of
the  experiments  indicated that  present
practical technology did  not support such
a methodology.   Other special  hardware
requirements of a field-deployable lidar
system    were    explored,    including
high-speed,  high-sensitivity   detector
systems,    and    miniature    detector
cryocoolers.
Introduction

During discussions held  at the National
Institute  for  Occupational Safety  and
Health (NIOSH) on research needs for gas
and    vapor    monitoring,    industrial
hygienists    and    instrumentation
specialists  alike  voiced  a  need  for
portable  surveillance  equipment to  map
concentration distributions  of selected
vapor  phase chemical pollutants in  the
workplace  atmosphere,  on  a  real  time
basis.  It was suggested that this should
be  some  type  of  optical  instrument,
requiring only  line  of sight  from  it to
the areas to be  monitored,  investigating
possible  approaches  to  this  task,  a
number of  recommendations regarding  the
desirable  characteristics of a  gas  and
vapor surveillance system were gathered.
These  recommendations  indicated  that an
ideal  system should:  (a)  be  portable
(movable into various workplace settings
for surveys etc.),  (b)  be single-ended
(no  use  of  retro-reflector  arrays  or
transmitters  and  receivers separated by
large  distances),   (c)  give  real-time
mapping  of  the  spatial  distribution,
concentration and identity of  a variety
of airborne pollutants commonly found in
the  workplace,    (d),  not  present  any
hazard to the  people  in  the  workplace
(from laser radiation,  chemicals, liquid
nitrogen,  etc.),  (e)   require  minimal
maintenance (some routine  service perhaps
every  several   months),    (f)   have  a
sensitivity on the order  of 10-100  parts
per billion  (over sampling distances of
five  meters)  for  some pollutants,  (g)
have a range up to a few hundred meters,
(h) be automated  and  programmable  for
unattended  data  gathering,  (i)  have  a
spatial  resolution  of  at  least  five
meters full-width-half-
maximum  (using  a  point source),  and  (j)
                                         433

-------
survey the workplace at least every five
minutes.  In reviewing previous research
in    this    field,    two    principle
methodologies  were  identified:  Fourier
transform infrared (FTIR)  spectroscopy,
and  laser-based   light  detection  and
ranging (lidar) systems.

      Fourier-Transformed Infra-red
              (FTIR)  Systems

Open-beam, long-path FTIR systems appear
to be gaining much attention, and showing
at   least   a   limited   potential  for
qualitative  assessments  of  atmospheric
gas  and vapor contents  for  toxic waste
site monitoring,  as well  as fence-line
monitoring.  Such  a system has been used
by Herget for various outdoor studies'1'2'
as  well  as  an  workroom _study   in  an
aluminum refinery  potroom.
              (3)
Other   investigators  have   used  FTIR
equipment   similar  to  that   used  in
analytical  laboratories,  for  a  "real
time" analysis  of  gaseous pollutants in
the  workplace  by  use  of a multi-port
sampling  apparatus  with  sample  lines
running from ,work  locations  to a central
FTIR system.
(4.5)
FTIR   has   several   advantages   over
alternate  methodologies  including:   (a)
high  sensitivity,  (b)  high specificity,
(c)   a  large   variety   of   chemicals
detectable    and    identifiable,     (d)
technology  is  well  tested,  based  on
similar laboratory use,  (e) proven field
ruggedness  and  dependability,  (f)  no
support    requirements    other    than
electricity  (if  detector cooling is not
used),  (g)  commercial  availability,  (h)
an  inherent  self-calibration  of certain
parameters  of  the equipment,  (i)  "real
time"   analysis   capability,   and   (j)
portability of the apparatus, making its
use for temporary applications feasible.
Possible problems  associated  with this
approach  include:  (a)  a  lack  of  an
adequate   gas   phase   high   resolution
spectral library when peak identification
methods  are used  to  identify  unknown
compounds,    (b)  likely  problems   with
chemical interferences  in the analysis,
(c) a lack  of adequate spatial resolution
 for some applications,  such as measuring
 breathing zone chemical  concentrations,
 (d)  a  requirement for either a two-ended
 system or one using retro-reflectors,  (e)
 a  lack of adequate scan speeds to achieve
 "atmospheric   freezing"   and,   (f)    a
 reduction in the  systems  sensitivity  due
 to atmospheric water  vapor.

 Currently,  a  Nicolet  Instrument  Co.,
 Madison,  Wisconsin, custom  portable open
 beam path FTIR system is  being evaluated
 for  industrial hygiene  applications by
 the  University  of Michigan  School  of
 Public   Health   (S.  Levine)   and   the
 University  of  California,  Berkeley  (R.
 Spear).    This work  is being  supported
 by a NIOSH grant #1-\R01-OH02666-01.   The
 Bomem  Inc.,  Quebec,  Canada,  Model   DA2
 open beam path FTIR system is  currently
being used for a  variety of atmospheric
pollution studies by M.^Spartz  at   the
Kansas  State University.'"
                                                                    (7)
Laser Remote Detection Systems

Laser  systems  have been quite useful  in
making  remote measurements  of chemical
contents of the atmosphere for a variety
of  purposes and  using several methods.
Most   remote  photo-optical  pollution
detection  work   has   been   accomplished
using light detection and ranging (lidar)
methods  where laser  produced radiation
was transmitted into the atmosphere to  be
measured and the return radiation (either
aerosol    scattered    or   terrain    or
retroreflector reflected) is analyzed  to
obtain the desired information about the
chemical  contents  of  the  atmosphere.
This has been  accomplished  by a variety
of methods'8' including:
                                a.
     The  analysis  of  Raman  scattered
     radiation:   In this  method,  Raman
     scattered  short  pulses  of  laser
     radiation  (usually at ultraviolet
     (UV)  and   blue   wavelengths)  are
     analyzed for their photon intensity
     vs.  wavelength distributions  as  a
     function  of   the   time  from  the
     emission of the radiation pulse from
     the  laser.    This  results in  the
     identification and quantification of
     the  pollutant  of  interest  as  a
                                          434

-------
b.
function  of   distance   from   the
transmitter.  The  return  signal is
relatively  weak due  to  the  small
Raman scattering coefficient.   This
limits the  range of  this  technique
to a few hundred meters, even  using
very   powerful    lasers.       Its
time-of-flight  (the  time  between a
transmitted  laser  pulse  and  the
return signal pulse,  indicating by
the  time   differential  what   the
distance was between the transmitter
and  a   particular  scattering  or
reflecting    location)    position
resolution  (ranqinq)  can  be on the
order of 10 meters.   This approach
has    been    used   in    many
investigations, usually to measure
pollutants  in exhaust  plumes  from
factories or  power  plants.   Raman
lidar  has  the  advantage  of  being
applicable  to  a  large variety of
pollutants.   However, the spectral
resolution  and  bandwidth  of  the
Raman  signal  is   relatively  broad
leading to significant interferences
in complex  mixtures  of pollutants.
Interference  often   results   from
sunlight or other  bright lighting.
Principle limitations are a severe
lack in sensitivity of the technique
(a very  weak Raman  return signal)
requiring the  use of very powerful
laser  systems which  are  typically
not  "eye safe" and  would be  very
difficult to  make  "eye safe"  while
preserving  the  system performance.
Also,  spatial resolution  has  been
poor due to the  poor data statistics
of the return signal, and the near
field  interferences   in   time  of
flight measurements.

Resonance Absorption Methods:  These
methods  depend on  transmitting  a
laser beam having a wavelength very
close  to  that   of   a  resonance
absorption   line   of   a   chemical
species    being   measured,    and
measuring a return signal reflected
off of an object or from aerosols in
the  laser  beam's  path.    It  is
generally  a  much  more  sensitive
method then that of Raman scattering
due  to  the much higher scattering
                                                 coefficient   for   Mie  scattering.
                                                 This    increased    scattering
                                                 coefficient   could  result  in  an
                                                 extended  range,  the  use  of  lower
                                                 powered  lasers,   the use  of  less
                                                 sensitive  detectors,  better  range
                                                 resolution, and faster measurements
                                                 than  Raman   lidar.    It  has  the
                                                 disadvantage   of   being   able  to
                                                 measure only one pollutant  at a time
                                                 when  only one laser  wavelength is
                                                 used.  Unresolved  interferences can
                                                 result  when  only one  transmitted
                                                 wavelength is used.  One version of
                                                 such  a  system  uses  differential
                                                 absorption lidar (DIAL)  whereby the
                                                 laser   radiation   wavelength   is
                                                 shifted   between   two   (or  more)
                                                 wavelengths,  one   of which is very
                                                 near  to  the  resonance  absorption
                                                 line of the pollutant being measured
                                                 and another  line which is  not.  The
                                                 ratio of these measurements is used
                                                 to normalize the measurement system
                                                 calibration   with  regard  to  the
                                                 scattering  characteristics  of the
                                                 atmosphere   being  studied.     By
                                                 measuring     the    return    signal
                                                 intensity  for successive  locations
                                                 along  the  beam  path,  subtraction
                                                 yields  the  signal loss  due to the
                                                 incremental  decrease in the observed
                                                 signal   associated  with  a   range
                                                 increment  AR, and arising from the
                                                 attenuation    of    the   specific
                                                 molecular constituent for  which the
                                                 laser  is tuned.    This is  described
                                                 by the  following  equation:
(8)
                                            4E0. - [BU0.R) -
                                                               AR) ] - tE(Xu,R) -
                                                                                 AR) ]
                                            where:
                                                 AE0«  = the  incremental  decrease in
                                                 the  differential signal  E(Ao,R)  =
                                                 the return  signal energy at range R
                                                 and  wavelength  X0,  representing  a
                                                 maximum  absorption  by the chemical
                                                 of interest.

                                                 E(XH,R) = the  return signal energy
                                                 at   range   R  and   wavelength  A.u,
                                                 representing an  absorption off the
                                                 resonance  line of  the  chemical of
                                                 interest.
                                         435

-------
     Using conventional timing  methods,
     DIAL  systems  can  have  ranges  of
     several km, and spatial resolutions
     of as small as 10 m.


hanging Methods

Principal  methods  used   for  position
determination  (ranging),   include:   (a)
time-of-flight, (b) triangulation ranging
(such as methods used by the U.S. Bureau
of Mines for measurement  of  methane gas
concentrations at coal seams in mines<9)),
and  (c)  the   use   of  variable  focal
distance optical systems.  Variable focal
distance optical  systems  have been  used
successfully   in  laser   doppler   wind
velocity measuring  apparatus using CO2
lasers and heterodyned detection methods,
and have produced spatial resolutions of
approximately   10   m  at  a  range  of
approximately 100 m, with maximum useful
ranges  of  up  to 1  km.    However,  this
method  requires  a  coherent  detection
system.<10>   Practical problems have  been
encountered  with these  systems due  to
their   instabilities  and  engineering
difficulties,     including    frequency
instability, harmonic generation,  phase
shifted  echoes  and  loss of  wavefront
parallelism.    In  addition,  there  are
significant problems associated with the
application of coherent detection systems
with rapid wavelength shifting,  when the
parallel    retuning    of    both    the
transmitting laser and a local oscillator
laser  must  occur or when optoacoustic
wavelength   shifting  of   part   of   the
transmitter signals is used as the local
oscillator.      In   addition,  coherent
detection is  probably not practical for
short    distance    time-of-flight
measurements  (believed to be necessary
for  our application)  due the heterodyne
frequencies  required to  achieve a 30 ns
timing  resolution.    This is especially
significant when considering optoacoustic
wavelength shifting.

Laser Types

Many  laser   systems  exist  for   lidar
applications,    from    extremely
short-pulse-length,    high-energy/pulse
systems to continuous-emission,  low-power
devices covering wavelengths from the far
ultraviolet  to the  far infrared.   If a
very rugged,  low  maintenance,  compact
system  is needed, tunable over a range of
wavelengths,   the  number of available
lasers  becomes  quite small.    Possible
candidates   include  dye   lasers,   solid
state  diode  lasers and C02 lasers.    The
dye  lasers  suffer  from  rather  narrow
tuning  ranges   for  a   specific   dye,
moderate to high maintenance requirements
and  moderate  to  large  sizes.     They
operate mostly in the near ultraviolet to
near infrared region.  Operation at those
wavelengths  may  make design  provisions
for   personnel   eye   protection   more
difficult due to  lower  allowable limits
of radiation exposure.<11>

Tunable diode lasers  are available  at
less than  1 W outputs covering wavelength
ranges  from   approximately  2  -  30  /zm.
Significant   problems   associated   with
these devices are: (a)  a very
narrow tuning range, requiring an  array
of such devices in order to cover a wide
wavelength range, (b)  high cost,  (c) low
power  output, and  (d)   large size  and
weight  when  a   cooling  apparatus  is
included.
Many  gases and  vapors  of  interest  for
monitoring purposes have rich absorption
spectra  in the  near  to far  infrared,
associated    with    their
rotational-vibrational     molecular
transitions.     The  CO2   laser   emits
radiation in any of about 80 lines in the
region 9-11 j«n.  The number of available
wavelengths can  be expanded through the
use of  isotopes of carbon  or  oxygen in
the CO2.   One  of  these lines can often be
closely matched  to a rotation-vibration
absorption  line  of  a  pollutant  of
interest  so  that  it  is  feasible  to
monitor  a  large  variety  of  compounds
(though not necessarily simultaneously).
A  wide  variety  of chemicals  have  been
identified as being detectable  using  a
C02 laser  based lidar system (see Table  I
for a partial list  of gases which can be
detected using CO2  lines(12>) .  Using rapid
tuning,  a  single laser  could presumably
be  used   for  differential  absorption
monitoring   of  more   than   a   single
substance, and under many circumstances,
without interference by  other vapors or
aerosols  in  the atmosphere.    Humidity
would  be  expected to  produce  little
interference  in  monitoring pollutants in
the 9-11 /jm region from direct absorption
alone.  However, in cases  where the two
differential    absorption    lines   are
separated  by  more  than 2  x  10*Z  jra,
moisture  effects on the scattering and
absorption properties of  certain aerosols
may become a concern.03'  Other advantages
of the CO2 laser include the potential for
a  durable, compact and modest  costing
laser,  lacking  extensive  utility  and
                                           436

-------
 maintenance requirements, a more  liberal
 allowable  irradiance  level   for  laser
 radiation at long  wavelengths vs the  UV
 and visible01'  (this may  prove to be  an
 advantage,  depending  on  other  system
 performance    factors),     and    that
 wavelengths produced by the CO2 laser are
 transmitted    well    through    normal
 atmosphere.
     Tabl* I. Partial liat of C03 la««r d«t«ctabl* ga«««))   (the   parameter
characterizing     the    backscatter
efficiency)  pertaining  specifically  to
infrared    radiation    and    workplace
aerosols.  Since the  aerosol environment
varies considerably from one workplace to
another, the backscatter coefficient also
will vary.(U-15)
        Past Uses of Lidar Systems
         for Workplace Monitoring

There  have  been a  few attempts  in the
past  to  use  laser based  photo-optical
technology for workplace monitoring,  in
1981,    Britain's    Imperial   Chemical
Industries  and G.P.  Elliot  Electronic
Systems,  Ltd. made  a brief  report  on a
system  they were working  on using a  CO2
laser-based lidar system  for scanning a
workplace.06>    TO  our  knowledge  this
device   never  was   implemented.     MDA
Scientific,  Norcross,  Georgia, (formerly
Tecan   Remote  and  Environmental  Laser
Systems),  claims  to  have  obtained  an
exclusive  license  to use that technology
for their  applications.

In 1983, Egan of Bethlehem Steel's Homer
Research  Laboratories  reported  on  the
trial   of    an   Er:YAG   laser   based
differential    absorption    aerosol
backscatter  lidar  for methane detection
in mines.<17)  In 1985,  Litton of the U.S.
Bureau of Mines reported on their work in
developing  a  methane  monitor  using  a
laser diode  operating  at 3.3 jum,, and a
triangulation system  for  ranging.  This
resulted from  problems with  sensitivity
and explosion proofing of the equipment,
as  well   as  a  desire  for  tunability.
Results are not yet available/9'

In 1985, Persson, at Chalmers University,
Sweden   reported  on  a dual  CO2  laser
differential absorption detection system
for  use  in a  workplace.<18)   It  used a
continuous   wave    (CW)    laser   with
retro-reflectors placed at the end of the
laser  beam  path.    The device  measured
total  column  content  of   the pollutant
chemical,   thus   yielding   an   average
concentration along the path.

Tecan   Remote   produced   a  commercial
differential absorption CO2  laser based
system     for    workplace    pollutant
monitoring.09'  They used  a total column
content  method  and   retro-reflectors.
They have installed systems  at some major
chemical   manufacturers  facilities  for
monitoring around  process  equipment  of
special concern.   Wavelength changing to
monitor different chemicals monitored was
possible  by  a manual  retuning of  both
lasers.

Other    photo-optical    methods    were
considered  including  non-coherent pulsed
ultraviolet, visible and infrared sources
with    sensitive    spectroradiometric
                                          437

-------
detection  systems.   The method seems to
be  limited by  the  need for  both very
intense and very short  light pulses when
applied  to aerosol  scattering methods.
Generally,   adequate   high   intensity
broadband  sources  having  pulse  lengths
under  10  /is  are  not  yet  commercially
available.  This tends  to force  the use
of  long-path  measurement methods making
use   of   retro-reflectors,   and   the
acceptance   of  much   poorer  spatial
resolution.     The   system's   spectral
resolution  could  cause  problems  in the
presence of chemical interferences.

Problems  in Applying CO2 Laser
DIAL Lidar Technology to
Workplace  Monitoring

Based  on  the  above   information,  it
appeared that a C02laser DIAL  system had
significant  potential  for  leading to a
working system that would satisfy most of
the    system    characteristics    stated
earlier.      FTIR   methods  were   also
considered  promising;  however, at least
two    other   research   groups   were
investigating  that  methodology.    Many
uncertainties remained in the application
of  current  technology  to  producing  a
lidar system  satisfying the recommended
objectives.  These included:

(a)  problems in producing a system that
     can  scan  a  workplace  atmosphere
     while   keeping   transmitter   and
     receiver   beam   path   alignments
     precisely co-ordinated,
(b)  problems in achieving overall system
     sensitivities using components that
     did not  require frequent  servicing
     or  supplies  of  materials such  as
     liquid  nitrogen,   cooling   water,
     etc. ,

(c)  developing methods  for unfolding the
     identities and quantities of unknown
     contaminants  in complex  mixtures,
     using    multiple    wavelength
     measurements,

(d)   developing techniques  for  dealing
     with data errors due to the  effects
     of    rapidly    changing    aerosol
     concentrations   with   time    and
     position.  These could contribute to
     differences     in    sequential
     measurements  on  and off  resonance
     lines,

(e)   designing a system that was  field
     portable  (that was moveable into a
     workplace as one or several modules
     that can be handled by two people).
     This could place severe restrictions
     on the equipment selected for use,

 (f)  designing   a   system   that  could
     operate    continuously    in    an
     industrial    environment   without
     contamination of optical components
     or without  typical temperature and
     humidity extremes posing a problem,

 (g)  a   significant  probability   that
     non-aerosol  scattering   (scattering
     from  objects   and  walls  in  the
     workplace) could produce a sizeable
     interfering signal in the detection
     system, leading to  erroneous results
     or   a   greatly   reduced   system
     sensitivity.

There    were    significant    problems
associated    with    large    magnitude
non-aerosol scattered return signals from
objects   in   the   workplace   causing
erroneous  responses  in  the  receiver.
Scatter  rejection  could  be  the  most
severe technical difficulty to overcome.
This effect could be reduced by  using a
coincidence of time-of-flight ranging and
triangulation ranging, as well a possible
use  of both  transmitter  and  receiver
polarization   for   the  rejection   of
multiple  scattered  return signals  (see
Figure  1) .    The triangulation  ranging
uses a stationary alignment  between the
transmitted beam  and a linear array  of
receiver detector elements.   This  fixed
transmitter-receiver  relationship  could
also  help   avoid   the   difficulty  in
achieving  adequate   tracking between  a
scanning  receiver  and  a  stationary  or
scanning laser beam.  In  addition,  this
approach could help to eliminate another
potential problem associated with short-
range lidar signals,  which is a  lack of
adequate dynamic  range  in the receiver
for  return  signals  arriving  from  both
near and  distant scatter  sources.    By
having given  detectors look only  at  a
narrow distance range of return signals,
acceptable dynamic ranges should result.
It  was  anticipated  that  combining  the
narrow field-of-view with time-of-flight
ranging would  allow  a  5m FWHM  spatial
resolution.    It  was felt  that  HgCdTe
detectors  cooled  to  liquid  nitrogen
temperatures might be required here due
to the low allowable laser power levels
for eye safe conditions, as well  as many
other demands of the system design.

Calculations for a specific  system  were
performed,  using values  for  variables
                                          438

-------
suited to the desired system performance.
The choice of values for these variables
resulted   from    a   review   of   the
specifications  of  available  components
and    subassemblies    (including    a
consideration  of   their   costs).     An
example can  be  seen in Appendix I.  The
calculations  show that we would  need a
single pulse  energy of 2.75  x 10'3 J for
the   system  to   function   under  the
conditions defined.  The stability  of the
system's    electronics    and   optical
equipment may limit reliable differential
measurements  to   1  to  2  percent ;<1Z'ZO)
consequently, signal to noise ratio (SNR)
values greater then 500 to 1000 may not
be useful.

The   results   of   these   calculations
indicated that  if  the  assumptions used
were  correct,  it  would  be  possible  to
produce  a   lidar   system   having  the
required  sensitivity,  beam  irradience,
pulse  energy,  etc.     However,   these
calculations: (a)  assumed a hypothetical
value   for   the   volume   backscatter
coefficient   (/3(A0,R<>))  which  may  not
represent workplace conditions well,  (b)
did not address large scattering signals
from  surrounding   objects,   (c)  assumed
perfect  performance  of  optical  and
electronic   components,   (d)   did  not
address interfering chemical species, (e)
did not address pulsed electrical noise
from  the   laser    and  Q-switch   being
introduced  into  the  detector  signal,
degrading  the  systems   SNR,   and  (f)
assumed that  the  limited  information on
detector D*  values  were  valid for  the
fast   signals    necessary    for    the
time-of-flight measurements required.

Equipment and Methods

Laboratory Test System

Based on  the review conducted,  and  the
potential     benefit  of   a  workplace
pollutant monitoring system based on a
DIAL system, a laboratory evaluation was
conducted of certain  aspects  of  a  CO2
laser workplace DIAL system using time-
of -flight and triangulation ranging.   In
prder to make measurements of certain un-
evaluated  parameters  relative  to  the
performance  of  the  lidar  system,   to
verify  the  practicability  of  certain
design  concepts/   and   to  provide  for
experimental  development  of  a working
system,  a  laboratory test  system  was
assembled.   The goals  in designing this
system  were  to  construct a laboratory
apparatus having a maximum flexibility to
evaluate various methods of assembling a
workable lidar system.  It also needed to
be  capable  of  measuring  many   of  the
system  parameters  necessary  for  tne
development of future designs  of  a field
useable system.

The laboratory  test system consisted of
the following subassemblies:

                  Laser

System performance goals necessitated the
use of a recently  developed low pressure
pulsed  CO2  laser  in  combination  with a
high performance intercavity Q-switch for
producing near-Gaussian (no tailing as is
associated  with TEA CO2 lasers),  short,
high-intensity  pulses.   In differential
absorption lidar systems,  it is important
that differential absorption measurements
in an atmosphere take place on and off of
the  resonance   line  of   the  chemical
compound of  interest,  with a  very short
time    interval    between    the   two
measurements.  In the past,  this has been
accomplished  by  using multiple  lasers
tuned  to  different  wavelengths,  fired
sequentially with  a short time interval
between  pulses<20).    It was   considered
desirable that the two pulses  of the pair
be produced  by  the same laser, lowering
equipment   size,   weight,   .costs   and
alignment problems.

Rapid  wavelength  changing  of a  single
laser could  allow  for a rapid change of
chemicals monitored, thus  allowing a more
frequent  monitoring  of  pollutants  of
interest.   This  has  not  proved  to be
practical  in the  TEA  lasers typically
used,   due   to    both   power   supply
constraints  as well  as  the   mechanical
behavior of  the lasing medium.   It was
also desirable that the laser  be compact,
require  a  minimum of cooling, have an
extremely long lifetime (greater that 5 x
107 shots), and  use low radio frequency
interference (RFI)  components.

A laser was constructed using  a modified
Pulse Systems,  Los Alamos,  NM,  model LP30
low pressure CO2  amplifier section,  due
to the  long  upper lasinq  level lifetime
(greater than 60 ms)  of low-pressure CO2
lasers.    This   lifetime  permitted  the
Q-switching of two output  pulses  from a
single   laser    amplifier    electrical
transverse   discharge    pulse,    while
allowing   several   microseconds   for
wavelength changing between pulses.   An
intracavity beam telescope was  employed
to use the  amplifier  discharge  cavity
                                         439

-------
cross-section efficiently with the small
CdTe Q-switch crystals available.  A 1200
Hz    oscillating    grating   with    a
high-resolution  grating  position sensor
was   used   to   change   and   reprogram
wavelengths  rapidly.    Programming  of
wavelengths was accomplished by selecting
appropriate delay times from the grating
position reference signal for triggering
the  laser  amplifier  and Q-switch  (see
Figure  1) .     The resulting  laser  was
relatively compact  and had a  low mass,
and low power consumption.  It had output
pulses  of  approximately  Gaussian shape
with    full-width-half-maximum_   values
adjustable  between  50 and  100  ns;  an
ability to  produce  "pulse pairs" having
interpulse  spacings  of  5-50  /is,  each
pulse  of  the  pair  being  independently
wavelength selectable over most CO2 laser
emission lines; a repetition rate  for the
"pulse pair" up to 10 Hz;, a pulse energy
of approximately 5-10 mJ; and pulse pair
wavelengths reprogrammable between pulse
pairs.   The  output  of  the  laser  was
emitted   through   a   beam   expanding
telescope such  that  the  irradience  was
well  below  maximum  permissible  occular
exposure limits<1X>  for the  pulse widths
and repetition  rates  used.  Most of the
basic  performance goals  of the  device
were   achieved   in   the    laboratory
prototype.    A full  description  of  this
laser is presented in a paper soon to be
published.<21>

        FIGURE 1.  Q-SWITCHED DOUBLE-PULSE CO2 LASER

              •v
        Beam Power/Energy Monitor

Provision was made for extracting a fixed
percentage  of  the output beam to  a  beam
power  and  energy  monitor.    The  beam
power/energy   monitor   was   used   for
continuously monitoring the beam  pulse-
to-pulse energy differences to  normalize
the  results of  differential absorption
measurements   to   constant  beam   pulse
energy   conditions.      This   monitor
 consisted  of  a  fast  (<1  ns  rise  and fall
 time)  room-temperature  HgCdTe  detector
 (Boston  Electronics  Corp.,  Boston,  MA,
 model  R004-0)  and  a  Comlinear  Corp.,
 Loveland,  CO,  model  CLC100  low  noise
 amplifier.  The monitor  had a  short term
 (1  h)  reproducibility  better  then  +1.0
 percent   and   an   absolute  long   term
 accuracy  better  then +5  percent  (2a) .
 This detector monitored  the laser output
 radiation  intensity and   waveform  via
 radiation  reflected  from  the   tuning
 grating.

           Receiving Telescope

 The receiving telescope was a Newtonian
 type having a mirror diameter of  8 inches
 and    an     effective    aperture    of
 approximately f/4.5.   It  was  used  for
 gathering   the   return   signal   laser
 radiation  and projecting a  line  image of
 the  aerosol  scattered   laser  beam  onto
 HgCdTe detectors.

                Detectors

 Several types of detectors were available
 for  sensing  radiation   in  the  9-11  fn&
 wavelength band; however, only the HgCdTe
 detectors     had     sufficiently    high
 detectivities (as indicated  by   D*)  for
 use   in   this   application.      HgCdTe
 detectors  are manufactured in a  variety
 of  forms,   varying  in  surface  area,
 wavelength    sensitivity,    frequency
 response,   etc.,   depending   on   their
 applications.       Unfortunately,    the
 manufacturers'  data  on  their  products
 were  often  sketchy,  and  their   testing
 methods  often  did  not  include  actual
 measurements of fast pulses  of radiation.
 Thus, one  was often  left to speculate on
 their actual  performance in a particular
 application.      Competing   performance
 parameters,   frequency   response   and
detectivity,   were  both  critical  to  the
function  of   the  lidar  concepts  to  be
evaluated.   The detectors  evaluated  in
the receiver  system were  cooled HgCdTe
detectors  of  photovoltaic   (77   K)   and
photoconductive (77  K and  200  K) types,
and were  selected as representative  of
devices  commercially available   at  the
time.  The  photoconductive detectors were
specified to  have approximately 10 degree
fields of  view, 1.3  x   1.3  mm size,  D*
values of  approximately  1  x 1011  cm  Hz1/2
W"1  at  10 kHz,  and a  high  frequency
roll-off at  approximately 10 MHz.    The
elements were useable as single elements
or as  a linear  array with  5  elements.
Element six was  a photovoltaic detector l
                                          440

-------
mm diameter, had a 10° field-of-view, a D*
rating of 2 x 1010 cm Hz1/2 W1 at 100 kHz,
and   a  high   frequency   roll-off  at
approximately 50 MHz.   The detector array
was    manufactured    by    InfraRed
Associates,Inc.,  Cranbury,  NJ,  as their
model  #89-251R.    For very  high-speed
measurements (<10 ns rise  and fall times)
a   Judsen  model  J15TE4:10-MC31G-S01M
thermoelectric   cooled   (200K)   HgCdTe
photoconductive  detector  with a  model
TC-4  controller  (both manufactured  by
EG&G  Judson,  Montgomery,  PA)  was used.
It had a 1 mm diameter detector element,
a D*  rating  of 3 x 10* cm Hzvz W1 at 10
kHz,  and a  high  frequency  roll-off  at
approximately  100 MHz.   Generally, the
HgCdTe  detectors  designed  for  higher
operating  temperatures  exhibited  much
better frequency  responses,  but this is
accompanied by a significantly lower  D*
value.  Due to the photovoltaic detectors
large capacitances and consequential long
time constants, a  Comlinear model AJP401
transimpedance  preamplifier  was used to
help reduce the effective time constant.
Comlinear  Model   CLC100  video  voltage
preamplifiers   were   used   with   the
photoconductive detectors.

            Detector Cooling

The detectors  used for detecting return
signals  were    cooled  by   two methods.
The linear array detector was kept at its
operating  temperature  (45-77  K)  by  a
closed  cycle   miniature   refrigeration
system (a Philips/Magnavox Model MX 7043,
Magnavox Electro-optical Systems, Mahwah,
NJ).   This consisted of a split Stirling
cycle,  linear  motor  device  using  no
bearings  or   lubricants,  and   having
clearance seals.  The overall device was
hermetically sealed.  Mean times between
failures for this device  are guaranteed
to exceed 2500  operating hours, with test
data implying lifetimes >10,000 h.  Heat
power capacity at 77 K was approximately
1 W, allowing reasonably large assemblies
of detectors to be cooled.   Total power
consumption at those operating conditions
was approximately 50  W.    The  vibration
produced by the mechanical refrigerators
can   cause   unwanted   motion   in   the
detector,  which   may  degrade   system
performance by modulating the detector's
position relative  to  the  focused photon
beam to  be  detected.  The satisfactory
performance of such a closed-cycle cooler
could provide a considerable advantage to
the  performance  of  a  field-deployable
workplace lidar, allowing the utilization
of  high  sensitivity  detector  systems
without  the   difficulty   of  supplying
liquid  nitrogen  for  it.    The  single
element  photoconductive  detector  (see
identified  in  "Detector"  section)  was
cooled   to  200   K  by   a   four-stage
thermoelectric cooler.  Such coolers had
extremely   long   expected   lifetimes,
produced no vibration and required little
power (8 W for the one used).  The lower
temperature   limit   of   such   coolers
resulted in a  less  than optimum D* value
for   the   detector,   when   detection
sensitivity at very high frequencies was
not critical.

      Transient Waveform  Analyzer

A high speed waveform analyzer was needed
both  for general  system  diagnostics as
well as for analyzing lidar return signal
time  of  flight   information.     This
consisted  of  two LeCroy  Corp.,  Spring
Valley,   NY,   Model   TR8828  Transient
Waveform Recorders, capable of capturing
two  fast  waveforms  simultaneously  and
recording  them   in  temporary  memory.
These recorders  digitized the  waveform
information in 5  ns increments,  with an
8-bit    accuracy,    and   stored    the
information in file lengths up  to  32k.
These recorders were  interfaced via  a
CAMAC  IEEE-488   interface  to  a   data
processing  system for  display and  data
reduction.     ASYST  waveform  analysis
software  (Asyst  Software  Technologies,
inc., Rochester, NY) was used to display
and treat  lidar return signal data.   A
specialized  • fast-Fourier-transform
filtration  method utilizing a  Blackman
attenuation function*22' was used to remove
unwanted high frequency components of the
data, allowing a  better   extraction  of
return signal information.  After using a
wide range of cutoff frequencies with the
data, a 30 ns cutoff was selected which,
considering the slope of Blackman filter
function,  appeared  to have an effective
frequency  cutoff  of  approximately  100
MHz.   Some data  were  recorded  using a
model 54021A oscilloscope manufactured by
Hewlett Packard Co., Palo Alto, CA.

               Data Logger

A data logger consisting  of  at least 12
parallel channels of fast  sample and hold
amplifiers  coupled  to 10  bit ADC's, was
interfaced to the data  processing system
via  a CAMAC   IEEE-488  interface.    The
sample and hold amplifiers had gate times
as  small as  10 ns.    This  enabled the
setting  of  individual time  of  flight
range windows for individual  detectors in
                                         441

-------
a  linear  array,  and  a  rapid shift  of
signals from one set of input channels to
another    between     short-interval
pulse-pairs.

With the assembled laboratory test system
(see  Figure 2), a series  of experiments
was  performed to  help  to  understand
better  the potential  for using  aerosol
backscatter as  a signal source  for  a

       RGURE 2. LABORATORY TEST SYSTEM.
workplace DIAL system.  Measurements were
carried   out   in  a   corridor   having
dimensions of  1.85 x 2.75 x 79 m.  Major
components of the laboratory test system
were   mounted   on   a   movable   bench
positioned  at one end  of the  corridor.
The  transmitted   laser  beam  was  then
projected  along  the  major  axes of  the
corridor  under several conditions  (see
Figure  3)  including:  (a)  co-axial  with
the corridor  and  intercepted by  the end
of the corridor;, (b)  as in  (a),  except
with an object (a light source mounted on
a  tripod),  and separately,  a  generated
aerosol  (wood  dust,  and bleached flour)
in the  beam path located  28m from  the
source; (c)  with the laser beam projected
to reflect off the walls of  the corridor,
as well as  the  end  of the  corridor.
Return  signal  measurements   were  made
using both the thermoelectric-cooled and

      Figure 3. Corridor Measurements Using
            Laboratory Test System.
 Stirling-cycle-cooled  detector systems,
 located  at  the   focal   point  of  the
 receiver   telescope    for   particular
 distance intervals along the transmitted
 laser beams  path.   The profile  of the
 transmitted laser  beam was mapped  at a
 position  approximately  28 m  from the
 transmitter,     using     the
 thermoelectric-cooled  HgCdTe   detector
 operated  above   its   normal   operating
 temperature  of 200  K,  to  reduce its
 efficiency.   The detector was scanned in
 2.5  cm intervals from -15 to +15 cm  along
 both the  x  and   y  axes  on  a   plane
 orthoganal    to    the    beam     axes.
 Measurements  were  made  at  10.6  nm
 wavelength  (10P20  line),  with a   laser
 output   of  approximately   2   mJ/pulse.
 Measurement  of   the   overlap  of   the
 receiver acceptance angle and solid  angle
 of the transmitted beam at a given  focal
 distance  were  made   by   scanning   the
 detector element across the focal plane
 for  a point  reflector  backscatter source
 located  at the extremes of the beam cross
 section.     This  was   accomplished  at
 distances from  the  transmitter/receiver
 of 15 and 28 m.

Data and Discussion

Table II shows the  transmitted beam cross
section   relative   irradiance  profile.
  Tabl* II. Happing of tha R«lat
  at 30 n.
              Relative Irradlanca (H CM'3)
          -12.5 -10 -7.5
  y-ax«s
  (cm)
  -12. 5
  -10
  -7.5
  -5.0
  -2.5
  0
  2.5
  5.0
  7.5
  10
  112.5
0   15
15  60
17S  270
100  200
50 25
100 70
From the slight bipolar shape, it appears
that   there  are  probably   two  cavity
resonance modes  present.   Figure 4 shows
the output  from  the  laser as measured by
the 200 K detector.   The  approximately 6
/.s interval between two sequential pulses
on the 10P16 and 10P20 lines respectively
can  be seen.    Figures  5a-5c  show  the
return pulse (diffusely reflected from an
object  at  10   m)   as  monitored  by  a
photoconductive  detector  element (77  K)
a photovoltaic  detector element  (77  K)
and    a    photoconductive    (200    K)
thermoelectrically    cooled    detector
respectively.   It should be noted that
the  polarity  of  the  pulses  shown   in
Figure  5  varies  with  the detector  and
amplifier used.   The output of the 77 K
                                           442

-------
     Figure 4. Dual-pulse Laser Output.
,,,, ,
•;•••:
Pulse I (IOPI6)

/
Pulse 2 (IOP20)

                Time (1000 ns/dlv)
      Figure 5.  Frequency Response
      Characteristics of Three Types of
      HgCdTe Detectors.
                      Photoconductlve • 77 K
                 Time (200 ns/ulv)
                 Distance (30 m/dlv)
                /   \
                        Photovoltaic • 77 K
                      Photoconductlve • 200 K

                              *—**-A A ^-~-^
                 Time (100 ns/dlv)
                 Distance (15 m/dlv)
photoconductor and photovoltaic detectors
is   positive-going,    and   the   200  K
photoconductive  detector  is  negative-
going.   It is clear  that  the frequency
response of  the  photoresistive detector
is quite poor, with  a 1/e time constant
of  approximately  500  ns.     The  77  K
photovoltaic detector  frequency response
was much better,  with a 1/e time constant
of approximately 80  ns.   The frequency
response of  both of  these detectors was
somewhat   poorer    than   anticipated.
Manufacturer D*  specifications for  their
detectors   are  often  based   on  low
frequency measurements (1-10 kHz) and
extrapolated to  higher frequencies, with
possibilities  for  substantial   errors.
Pulse shape distortions by the detectors
for fast rise  and  fall time pulses were
often not specified by the manufacturer.
The   time    constant   of   the   200   K
photoconductive detector was small enough
 (specified as <10 ns by the manufacturer)
that the laser pulse shape detected by it
was not easily  distinguishable from that
displayed    from     the    high-speed
room-temperature    HgCdTe     detector.
Examining  the  time resolution  required
for a 10 m time-of-flight (round-trip for
the  desired  5  m   spatial  resolution,
approximately  33  ns) ,  it  seems apparent
that, based on  frequency response alone,
the 77  K photovoltaic  detector  would be
quite  marginal,  since  tailing  in  its
output  pulse excessively extended  over
neighboring spaces, and only the 200K and
room     temperature     photoconductive
detectors   would   be  fully   adequate.
Considering that  60 ns FWHM is probably a
lower  limit  for  the  transmitted  laser
pulse width,  any further  degradation of
that pulse  width would  be  unacceptable.
It  should  be  noted  that  it  would  be
difficult  to   achieve  a  5   m  spatial
resolution  from  a  continuous  aerosol
backscatter  source  using  time-of-flight
methods  and  a  60  ns  FWHM  transmitted
pulse.      We   anticipated   that   the
combination of triangulation and time-of-
flight   could   make   this   possible).
Examining   the   D*   ratings   of   these
detectors,  it  is  clear that  substantial
system   design   compromises   would   be
necessary to utilize either one, with the
room-temperature    device    being
particularly poor.  Considering the above
findings, it appears that the earlier use
of a D*  value of 1 x 1011 cm HzV2W"1 in the
example    calculations    was    overly
optimistic, and that  in practice a value
of 108 cm Hz1/2 W"1  may be more  realistic.
This is a result  of both  the pulse shape
distortion resulting from some detectors,
as well as an inaccurate extrapolation of
D* values to higher frequencies.   It  is
possible that this  could be improved  by
the use of lower capacitance photovoltaic
detectors, detectors with a much smaller
surface  area,   or  the  use  of  coherent
detection  methods.     The  use  of  much
smaller  surface  area  detectors  would
demand a much more  sophisticated optical
assembly to achieve  a  sufficiently stable
focus  of  the  return  photons   on   the
appropriate  detector   element.     Some
practical improvements could probably  be
achieved  with   this;   however,   their
magnitude would be difficult to estimate.
Additional    system    sensitivity
improvements   could   be   achieved   by
increasing   the   receiving   telescope
                                         443

-------
diameter (a 30 cm diameter would increase
AO by a factor of 2),  and by increasing
laser power (narrowing safety margins to
400%   should   allow   an   increase  by
approximately   a   factor  of   10)   to
approximately  25  mJ/pulse.   Both  the
closed-cycle  Stirling  cooler  and  the
thermoelectric cooler worked well during
the  several hundred hours  of operation
they were used.

The curves in Figure 6  show  the effect of
filtration  on  a  single  return  pulse.
Figure  7   shows   the  overlap   of  an
end-of-corridor   return   signal  with  a
space  separated  from  it  by  20  meters,
using   the  high-speed   200   K  HgCdTe
detector, and an approximately 90 ns FWHM
transmitted  pulse.   The ratio of the
end-of-corridor  signal to  the adjacent
space along the  beam path (assumed to be
an  insignificant signal  for the aerosol
backscatter component)  is  approximately
300.  Sources of this unwanted signal may
include stray  light,  aberrations in the
receiver optics, inadequate collimation
of the detector,  and the  shape and width
of the transmitted  pulse.

       Figure 6. Example of Filtration
       of Single-pulse Waveform.
        Figure 7. Effect of Large Signal
        Source on Detection of
        Neighboring Small Signal Source.
                Time <80ns/dlv)
                Distance (I 2 m/dlv)
If  the ratio  of  the  aerosol  scattered
signal   to   that   from   a   diffusely
reflective surface is approximately  10" ,
it appears that  the return  signal  from a
space  adjacent to  such  a strong  signal
would  have  to  be  isolated  from  that
signal source by a  ratio of at least 10 ,
in order for aerosol backscatter from the
space  to   be  detectable   and  useful.
Depending  on the  aerosol  concentration
                20 m
                 Time (300ns/dlv)
                 Distance (45 m/dlv)
and its reflective  properties the aerosol
scattered signal could vary over several
orders of magnitude.  It appears from the
data in Figure 7 that the "tails" of the
very    strong    specular    or   diffuse
reflections  from  walls  or other  solid
objects  cause  overlaps  of  signal  in
neighboring   spaces   such   that   the
resulting SNR  ratings for  the adjacent
space  along  the  beam  path  woulxJ  be
inappropriate  for  detection  of  aerosol
backscatter signals.

Figure  8  shows  an  example of  an aerosol
backscatter  signal produced  by a  fine
mahogany  wood  dust  aerosol  along  an
approximately 3 m pathlength of the laser
beam, at a distance of approximately 28 m
from   the   transmitter/receiver.     The
waveform   was   the   result   of   the
subtraction  of  a  return  signal  without
added  aerosol  from  one  with  the  added
aerosol.  The  limiting  background noise
in the  signal  appeared  to be  the result
of  electrical  noise  from  the  Q-switch
induced  into the  detector  signal  path,
rather than amplifier and  detector noise.
This  aerosol was  poorly  characterized;
however, it very likely represented wood
dust  levels  in  excess of  OSHA allowable
limits   (a  visible  cloud).      This
experiment was used to produce an example
of an upper  practical limit of the amount
of  backscatter  signal  that  could  be
obtained.  Normal  aerosol concentrations
in the air-conditioned  laboratory space
were   insufficient  to  observe  aerosol
backscatter with the receiver efficiency
and   transmitted   laser   power  of  the
present system.
                                           444

-------
An  examination of  the  difficulties  in
using  aerosol backscatter  as  a signal
source  for   a  workplace  DIAL  system
prompted an examination of alternatives.
The above data suggest that the scatter
from workplace objects  and  walls could
provide  a strong  signal  source  for a
"column content" system.  Time-of-flight
return  signals,  such as  in  Figure  9,
could   be  cross-correlated   with   the
transmitted   signal   to  enhance    the
separation   of  return   pulses  having
differing  time-of-flight  values,   thus
enabling  a  determination  if  only   one
significant scattering of the  transmitted


     Figure 8.  Backscatter From Wood
     Dust Aerosol.
              Time (100 ns/dlv)
              Dlntince (IS m/dlv)
      Figure 9. Multiple-scattered
      Return Signal.
                Time (200 ns/dlv)
signal  had taken  place, as  well as  to
determine  the round-trip path  length  of
the  transmitted  pulse  to  the  receiver
 (see  Figure 10) .   The continuous radial
scanning  of  a  workplace utilizing  this
methodology may provide a useful means of
generating an angular  mapping of average
beam-path     concentrations    in    the
workplace.    The  use  of more  than  one
transmitter/receiver    system,    having
overlapping  monitoring  fields,  could
allow   construction   of   a   workplace
pollutant concentration map.  It is also
possible  that  an  aerosol  backscatter
signal  could  be used  if the reflective
surfaces  the  beam  could intercept were
treated  to  reduce their  refectivities
(perhaps  by a  few orders of magnitude),
or   in   outdoor   workplaces   where   no
significant    non-aerosol   backscatter
sources   were  in  the  field-of-view.
Outdoor  settings   should  also  provide  a
substantially     larger    backscatter
coefficient.

      Figure 10 Radial-scanning Workplace
             Monitor.
 Conclusions

 This study  identified several substantial
 limits     to    the    use     of    an
 aerosol-backscatter  DIAL   system   for
 workplace monitoring.  These were  (a) the
 limitations   of   currently   available
 detectors,  when  applied to a high spatial
 resolution  time-of-flight lidar,  (b) the
 difficulty  in providing very high optical
 isolation  ratios  for   adjacent  spaces
 along  the  laser beam,  (c) the probable
 low  aerosol   concentrations   for   air
 conditioned  workplaces   (compared   to
 outdoor  concentrations),
-------
 non-aerosol backscatter  sources  in the
 field-of-view could be a suitable setting
 for using  aerosol  backscatter  methods.
 As an  alternative,  it appears  that the
 use of  backscatter from workplace objects
 may provide a  useful means of generating
 an Angular  mapping of average beam-path
 concentrations in workplace.    Further
 work is needed to  indicate the viability
 of this approach.

 Acknowledgment

 The author wishes  to  acknowledge  the
 efforts of  Gregory  J.  Deye,  physicist,
 for his work  in  the  difficult task  of
 adapting  the  Asyst  software   to   the
 requirements of the  data output  of the
 laboratory   test   system,   and   David
 Hartley,  research  physicist,   for  his
 initiation   of  the  concept  for  this
 project.    Both  are  employees  of  the
 Division   of   Physical   Sciences   and
 Engineering,  the National Institute for
 Occupational    Safety    and    Health,
 Cincinnati.  Ohio.
 References
 1.    Herget,   W.F.   and  J.D.   Brasher:
 "Remote  Fourier Transform  Infrared Air
 Pollution  Studies," Optical  Engineering
 19(4):508  (July/August)  1980.

 2.    Herget,  W.F.:   "Analysis of gaseous
 air   pollutants  using  a  mobile  FTIR
 system," American  Laboratory  (December)
 1982.

 3.    Herget,  W.F.:   "Long-Path  Infrared
 Measurements  of Gaseous  Emissions from
 Aluminum  Refinery  Potrooms,"  U.S.  EPA
 Internal Document (May)  1980.

 4.    Sawyer,   R.R.    and   J.    Coppola:
 "Automatic    Monitoring   Systems   for
 Determining    Time    Weighted   Average
 Workplace Levels,"  Symposium Proceedings
 "  Control  Technology in the Plastics and
 Resins Industry,  DHHS  (NIOSH)  Pub.  No.
 81-107.  U.S.   Govt.  Printing  Office,
 Washington, D.C. (January)  1981.

 5.   Watson,    W.M.:         "Continuous
 Environmental   Monitoring   of   Nickel
 Carbonyl by Fourier  Transform  Infrared
 spectrometry and Plasma Chromatography,"
 Environmenta1  Science  and  Technology,
Vol.  13,  (October)  1979.

 6.   Ying,  L.,  S.P.  Levine, C.R. Strang
and  W.F.   Herget:    "Fourier  Transform
Infrared    (FTIR)    Spectroscopy    for'
 Monitoring Airborne  Gases and Vapors of
 Industrial  Hygiene  Concern," Am.  Ind.
 Hyg. Assoc. J. 50(7):354-359  (1989).

 7.   Spartz,  M.L.,  M.R.  Wilkowski,  J.M.
 Fateley, J.s.  Jarvin, et aJL.: "Evaluation
 of a Mobile  FT-IR System  for Rapid VOC
 Determinations,"  American Environmental
 Laboratory (December) 1989.

 8.   Measures,   R.M.:      Laser   Remote
 Sensing, John  Wiley  and Sons, New  York
 (1984).

 9.   Litton, C.D.:   "Remote Measurements
 of   Methane   in  Underground   Mines",
 Proceedings of Topical Meeting on Optical
 Remote  Sensing of the Atmosphere,  Lake
 Tahoe,  Nevada  (January)  1985.

 10.   Schwiesow,  R.L., Cupp,  R.E.,  Post,
 M.J.   and   Calfee,   R.:      "Coherent
 Differential   Doppler  Measurements  of
 Transverse  Velocity  at a  Remote  Point "
 Applied  Optics. Vol.16,  #5, (May)  1977.

 11.   ANSI   Z136.1,    1986,     American
 National Standard  for the  Safe  Use  of
 Lasers,   American   National   Standards
 Institute,  NY,  NY (1986).

 12.   Grant, W.B.  and R.T. Menzies:   "A
 Survey   of  Laser and  Selected  Optical
 Systems   for   Remote   Measurement   of
 Pollutant Gas  Concentrations," j.  Of the
 Air Poll. Control Assoc. i3:188  (1983).

 13.   Petheram,    J.C.:      "Differential
 backscatter from the atmospheric aerosol-
 the   implications for IR  differential
 absorption lidar," Applied Optics.
 (November)  1981.               —

 14.   Post, M.J.:  "Aerosol  Backscatterina
 Profiles  at CO2  Wavelengths:  the  NOAA
 Data Base," APPl. Opt. 21:2507  (1984).

 15.   Schwiesow,R.L.,  Cupp,  R.E.,  Derr
V.E., Barrett,  E.W.,  and Pueschel,  R.F'
 "Aerosol Backscatter Coefficient Profiles
Measured  at   10.6   mm",     j^    App
Met eroro logy,  Vol.  20,  #2,   (February)"
 1981»

16.  Mannon, J.H.:    "Infrared Laser is
Key to  British Gas  Detector," Chemical
Engineering 47  (December)  1981.

 17.   DeFreez,    R.:        "Remote    DIAL
 Measurements of Methane in Coal Mines,"
 Proceedings of Topical Meeting on Optical
 Remote  Sensing of the Atmosphere,  Lake
 Tahoe, Nevada  (January) 1985.
                                          446

-------
 18.   Persson,   U.   S.,   Lundquist,   B.
 Marthinsson    and     S.T.     Eng:
 "Computer-Automated COj-Laser  Long-Path
 Absorption  System  for   Air   Quality
 Monitoring in the  Working  Environment,"
 Appl.  Opt.  2^:998 (1984).

 19.   Simpson,    0.:   "Remote    Sensing
 Technologies    for    Hazardous    Gas
 Detection," Sensors  (July)  1987.

 20.   Staehr,  H.,   Lahmann,   W.,   and
.Weitkamp,    C.:     "Range-resolved
 differential    absorption    lidar:
 optimization of range and  sensitivity",
 Applied  Optics,  Vol.  24,  #13,  (July)
 1985.
 21.   Piltingsrud, H.V. :   "A CO2 Laser for
 Lidar   Applications    Producing    Two
 Narrowly-spaced    Independently
 Wavelength-selectable Q-switched Output
 Pulses",    Submitted to  Applied Optics
 (1990) .

 22.   Blackman,  R.B.,  and  Tukey,  J.W.:
 "The  Measurement  of  Power  Spectra",
 Dover, 1958.
                APPENDIX  I
 From Measures*8'
                                 R
                                2/0 k(A0,R)dR
               A0 {(X0) AR D'
 where :
Emin = minimum laser energy pulse required
       to observe a return signal at
       range R  (J)

R = range to AR being sensed  (cm)

AR = range interval being sensed (cm)

SNRmin  =  signal  to noise ratio

/9(X0 R)  = volume backscatter
          coefficient cm"1 sr"1

£(R) = overlap factor of  laser and
       receiver beam (geometric form
       factor)

A0 = wavelength of resonance line at
     which laser is operating (cm)

A0 = area of objective lens (cm2)

5 (X0)  = receiver spectral transmission
        factor
                                             rd = detection time interval (s)

                                             D* = specific detectivity (cm Hz1/2  W"1)

                                             k(A.0,R)  = normalized attenuation
                                                       coefficient for pollutant in
                                                       atmosphere  (STP)  (ppm cm)
                                                                                -i
                                             C = concentration of pollutant (STP)
                                                 (ppm)

                                             B = detection bandwidth (sec"1)
                                                 (=l/2Td)

                                             Ad = detector area (cm2) .

                                             Using the  following values for variables
                                             in the previous equation:

                                             R = 10,000 cm

                                             AR = 500 cm

                                             SNR =1.5

                                             0(A0,R)  =  10~8 cm"1 sr"1 (this value may
                                                       be much higher in an industrial
                                                       atmosphere or lower an air
                                                       conditioned atmosphere)
                                              X0 = 9.639 /ira
                                             A0 = 314  cm2

                                             £(A.)  = 1.0
                                              rd = 5 x 10~8  s

                                              D* = 1012  cm Hz1/2 W"1 (10° acceptance
                                                   angle)   (practical values for this
                                                   variable using non-coherent
                                                   detection are probably on the
                                                   order of 1011)

                                              k(X0)  = 2.4 x 10~6 ppm"1 cm"1  (Benzene
                                                      at Ippm cnT1)  (other compounds
                                                      may have values up to 102 times
                                                      that for benzene)

                                              C = 1 ppm

                                              Ad = 1.7 x 10~2   cm2    (0.05" on  side).

                                              Then,

                                              E™1" = 4.13 X  10~6 J
                                               L

                                              If 10 sequential shots are accumulated,
                                              then:
                                          447

-------
        5 x  10
               ~7
  and
  Emin = 1<3  x 1Q-6 j per
  N
           2  "3 .T
                       of  2.75  x 10"
                                            DISCUSSION
CHUCK FLVNN: I'm curious about the decision, or finding, that it was not
useful because of the near scattering effects. If you could do away with this
scatter, would it be desirable?

HARLEY PILTINGSRUD: One of the things I mentioned in the paper, and
didn't have time to mention here, was that there possibly are some options in
some workplace situations where you could attenuate the backscatter from
objects in the workplace by some treatment. But it would be a little difficult
because you'd have to reduce it by a couple orders of magnitude and that's not
real easy to do.
                       CHUCK FLYNN: And if you took away the desire todo your ranging, could you
                       then do it easier?

                       HARLEY PILTINGSRUD: As I mentioned the second approach is one where
                       you lose some ranging. You know the direction the beam is pointed but the
                       measurement is a total column content one, so you don't know how the
                       concentration varies along the beam path. By using two such systems at different
                       angles of view, you could achieve some two-dimcntionsal spatial resolution.
                                                    448

-------
                              Immuno-based Personal Exposure Monitors
                                  Arbor Drinkwine and Stan Spurlin
                                      Midwest Research Institute
                                        425 Volker Boulevard
                                     Kansas City, Missouri 64110
              Jeanette Van Emon
      U.S. Environmental Protection Agency
  Environmental Monitoring Systems Laboratory
            Las Vegas, Nevada 89109
              Viorica Lopez-Avila
    Mid-Pacific Environmental Laboratory, Inc.
              625-B Clyde Avenue
        Mountain View, California 94043
Abstract

The feasibility of direct air monitoring using
immuno-based collection systems is being
investigated with the goal of developing personal
exposure monitors that take advantage of the high
specificity and sensitivity of immunochemical
systems.  A system is under development which
will lead to compact, diffusion-based personal
exposure monitors for specific target analytes.
The interface problem of the aqueous based
antibody system and the air medium has been
overcome using semipermeable membrane tubes
with a very high surface to volume ratio. An
immuno-based collection and analysis system using
a monoclonal antibody developed specifically for
pentachlorophenol has been investigated.  Similar
systems are being developed for aldicarb and
various nitroaromatics.

Introduction

The goal of this research is to develop personal
exposure monitors (PEMs) that use either a
polyclonal or a monoclonal antibody immobilized
onto a  silica support for collection and detection
of a specific target analyte or compound class.
Selectivity is based on the inherent characteristics
of the antibody system used in the PEM device.
The device should be applicable to short- and
long-term  monitoring and should allow the analysis
to be performed immediately after sampling and
in the field.
PEMs, both dynamic and passive, have long been
used for assessing occupational exposure to
hazardous materials. Recently, passive (diffusive)
sample collection systems have become popular as
PEMs, especially those that use sorbents. The
well characterized diffusion rates, high sample
capacity, and compact size of many of these
sorbent-based diffusion samplers make them ideal
for non-intrusive monitoring.  Sorbents such as
charcoal or Tenax which allow the PEM to collect
an array of organic compounds are most
commonly used. These compounds are then
extracted and analyzed by gas chromatography or
gas chromatography/mass spectrometry in a
laboratory setting.  This is an excellent approach
when several components are of interest and the
vapor is not characterized.  However, this method
is not particularly cost effective when only one or
two compounds are to be monitored during a
specific exposure scenario, such  as a pesticide
application or bag and drum operation.  Only a
limited number of compound-specific PEMs are
currently available for this type of monitoring (i.e.,
formaldehyde and NOX).

Immuno-based collection and detection systems
have many attributes which enhance their appeal
as PEMs.  The antibodies are generally selective
for a single compound or closely related  class of
compounds. This selectivity is an advantage for
the isolation of target compounds. Also,
                                                449

-------
antibodies generally exhibit high binding constants
for the target analytes, which means that their
collection efficiencies are high. Finally, the
antibodies can be easily regenerated to the
appropriate labeled or active form and reused.

The antibody-based detection systems available
exhibit good sensitivity, with detection limits often
in the 10 pg to 100 pg range. Recent applications
of amplification methods such as enzyme linked
immunosorbent assays (ELISA) have led to these
increases in sensitivity. The radioimmunoassay
(RIA) procedures, previously used in many
immunoassays, have been replaced with
colorimetric methods which allow for rapid
analysis of multiple samples using inexpensive
colorimetric readers (or even visual comparison to
standards in some cases).  Most immunoassay
formats are also relatively simple to use and are
readily adaptable to field laboratories.

The limitation in  using immunochemical
techniques  for air sampling is that antibodies are
designed to work in aqueous systems. The few
attempts to use "dry" antibodies have not been
very successful. Studies have been carried out
using antibodies immobilized to substrates such  as
polyethylene, Tenax, and indium [1,2].  The
immobilized antibodies were then exposed directly
to vapors without a wetting solution.  Success was
limited to those systems which developed a
response in an aqueous solution after exposure.
This limited success was most likely the result of
the target analyte binding non-specifically to the
protein covered surface and then binding
specifically with the antibody after wetting. We
have overcome this limitation by using a vapor
permeable  membrane as an interface.

The system we are investigating consists of a
vapor-permeable membrane and a cavity that
encapsulates the  aqueous medium and the
immobilized antibody.  The membrane acts as the
air-to-liquid interface, allowing vaporous analytes
to diffuse into the aqueous medium.  The  ideal
membrane  should allow the molecules to pass
freely and should not  interact with the analyte and
cause losses by nonselectively retaining the
analyte. It should also provide a high surface-
area-to-interstitial-volume ratio to keep the PEM
small and the  mass transport process rapid.
Figure 1 shows a diagram of the PEM device.
The analyte in vapor form diffuses through the
porous membrane of a microdialysis tube.  Once
inside the tube, the analyte is captured by the
antibody immobilized to the packing inside the
tube.  The capture results in the release of a
labeled compound or an enzyme product
(depending on the type of detection system) to the
aqueous  medium. At the end of the sampling
period, the tube fittings are attached to a syringe
or pump, the packing rinsed with solvent to
remove the labeled compound or the enzyme
product,  and the rinsate analyzed to determine the
concentration of the target analyte that diffused
into the microdialysis tubing.

Information on the diffusion rates of selected
compounds across the membranes  is presented in
this paper. Preliminary data on an antibody-based
collection system for pentachlorophenol (PCP) are
also presented.

Experimental

The membranes under evaluation are  regenerated
cellulose microdialysis tubing (Spectrum Medical).
Typical tube dimensions are 50 /im in external
diameter, 35 nm internal diameter, and 20 cm in
length.  Tubes are collected in bundles of 22 or 88
with pore sizes rated at 6,000 or 9,000 molecular
weight cutoff (MWCO).

The vapor chamber (Figure 2) consists of a 10 L
stainless  steel vessel with air sampling ports and a
stirrer to ensure even vapor distribution within the
chamber. Previous evaluations have indicated that
the distribution is even throughout the chamber
using  this propeller mixer.  The saturated vapor is
created by placing the material (e.g., radiolabeled
PCP) in  the bottom of the chamber and allowing
it to equilibrate until saturation is reached as
determined by collecting periodic air samples on
solid sorbents and analyzing them by gas
chromatography (GC) or liquid scintillation
counting (LSC).

PCP,  2,4-dinitrotoluene (DNT), and 2,4,6-
trinitrotoluene (TNT) were obtained from Aldrich
Scientific at 98% purity or better.  The C14
radiolabeled materials were obtained from New
England Nuclear. The radiochemical  purity of
each material was greater than 95%.  The
monoclonal PCP antibody was developed by
WBAS, Inc. of Rockville,  Maryland, and obtained
through  the U.S. Environmental Protection
                                                   450

-------
Agency, Environmental Monitoring Systems
Laboratory - Las Vegas. All other reagents were
obtained from Sigma Chemical.

GC analysis of DNT and TNT was performed on
a Varian 3700 equipped with an electron capture
detector and a DB-5 30m, 0.25mm ID fused-silica
open-tubular column.  LSC was performed using a
Packard 4170 Scintillation Counter.

The procedure used for the PCP vapor/liquid
diffusion study is as follows.  A 9-inch, 22-fiber
bundle was masked off with Teflon tape to allow
only 1.5 in. of fiber to be exposed  to the
atmosphere.  The bundle was rinsed with ethanol
and phosphate-buffered saline (PBS) to wash out
the plasticizer according to the manufacturer's
instructions. The bundle was filled with PBS,
fitted into the cell holder, and placed into the 14C-
PCP vapor chamber. At 5-min intervals for 1 hr,
2 mL of fresh PBS were drawn through the
bundle. The  2 mL of fluid, which  contained the
PCP samples during the interval, were transferred
to a LSC vial. At the end of the  1-hr exposure
time,  the bundle was removed from the chamber
and hooked up to a peristaltic pump and fraction
collector.  PBS was pumped  through the bundle at
a rate of 0.25 mL/min for 5 hr and the effluent
collected in LSC vials.  This post exposure study
was performed to help identify and quantitate any
hysteresis effect from PCP adsorbed to the tubing
but not yet migrated into the filling solution.
Quadruplicate vapor samples were taken from the
exposure chamber while the  bundles were being
exposed and analyzed for PCP.

Preliminary experiments for nitroaromatics used
water-filled dialysis tubing as the analyte collector
in the chamber.  These evaluations were
performed  in deionized water because it was
unknown at this time what buffer system would be
used. This allowed the determination of the
analyte diffusion rates into the internal filling
solution of the tubing.  The bundles were  not
masked with Teflon tape as they were for PCP.
This results in a larger sampling surface area for
these evaluations than for the subsequent  PCP
evaluations. The microtubes were filled with
deionized water after the manufacturer's extensive
solvent rinsing procedure was performed.  The
filled  tubes were placed into the chamber
containing 14C labeled DNT  or TNT for 0, 5, 10,
20 and 60 minute periods, then removed,  and the
labeled analyte amounts contained in both the
filling solution and the tubing material was
determined by LSC.

The procedure for the diffusion studies using an
antibody filling solution is as follows. The PCP
antibody was suspended in a PBS-Tween 20
solution and injected into the microtubes. The
tubes were then suspended for 15 minutes in the
chamber containing the radiolabeled PCP vapor at
3 ng/L.  Control tubes were also  placed into the
chamber, including one filled with PBS-Tween and
another with Bovine Serum Albumin (BSA)
protein suspended in PBS-Tween. The antibody-
filled  tubes were then dialyzed against deionized
water for 4 hours to remove any  unbound PCP
and then analyzed.

Results

Table 1 lists the approximate equilibrium vapor
concentration determined for the three analytes
using solid sorbents to sample the vapor  chamber.
Repetitive samples were collected until the
chamber concentration reached equilibrium.
These values are not corrected for recovery
efficiencies from the collection sorbents because
previous work at MRI has indicated recoveries of
better than 90% for these particular analytes.
      Table 1. Chamber Vapor Concentrations

              Estimated Vapor
             Concentration (Mg/L)
             from Solid Sorbent    Confirmation
Analyte           Collection         Method
DNT

TNT

PCP
  3

100

  3
GC

GC

LSC
Figure 3 contains the results from an evaluation of
the diffusion of DNT and TNT through the 6000
MWCO and 9000 MWCO dialysis microtubes.

The results for TNT and DNT are similar. The
lower apparent diffusion rates for TNT are
expected because TNT has a much lower
equilibrium vapor concentration than DNT. The
                                                451

-------
plotted values are the average of two
determinations at each time interval and have
been corrected for surface area differences.
Future work will couple this collection system with
antibody-based detection systems.

The larger pore tubing (9000 MWCO) does
exhibit faster diffusion than the smaller pore
tubing (6000 MWCO).  However, the magnitude
of the difference is not large enough to justify the
use of the larger pore tubing, which is more likely
to lose water by evaporation over extended
sampling periods.

The results from the diffusion rate studies of PCP
are shown in Tables 2, 3,  and Figure 4.  Duplicate
assays of each bundle type were performed.  The
results of assaying the PBS effluent while the
bundles were in the exposure chamber are
presented in Table 2, and representative plots of
each bundle type are presented in Figure 4.
Linear regression analysis was performed on the
cumulative PCP passed through the membrane
and into the PBS buffer as a function of time.
The data indicate an average mass sampling rate
of 14.3 ng/min for the 6000 MWCO bundle and
19.2 ng/min for the 9000 MWCO bundle.  The
effective sampling rate, calculated by division of
the mass sampling rate by the PCP vapor
concentration, averages 0.27 L/min  for the 6000
MWCO bundle and 0.30 L/min for  the 9000
MWCO bundle.

The correlation coefficients (Table 2) indicate
good linearity for both bundle types with the
MWCO 6000 bundles being greater than 0.970
and the 9000 being greater than 0.995.  The x-
intercept indicates the approximate delay time in
which the vapor permeates into the bundles and
into the PBS. These times are 6.7 min for the
MWCO 6000 bundle and 2.9 min for the MWCO
9000 bundle.

The total amount of PCP that permeated  the
membrane during and after exposure is  presented
in Table 3.  In the 5 hr after the bundles were
removed from the chamber, 1049 and 709 ng of
PCP continued to migrate through the MWCO
6000 bundle. Likewise, these values are 507 ng
and 438 ng for the 9000 MWCO bundle.  These
results indicate a large fraction (46 percent for the
6000 bundle and 30 percent for the 9000 bundle)
of the PCP takes a considerable amount of time to
permeate through the membrane and into the
buffer. However, this should not affect the usage
since a waiting time of 1 hour can be inserted into
the analysis scheme.

Representative plots of the postexposure
permeation results for each bundle type are
presented in Figure 5. These results indicate a
very rapid PCP passthrough during the first 30
min, followed by a very slow  accumulation during
the next 4.5 hr.  The slow permeation phase may
be reaching an asymptotic limit, but it is
impossible to estimate the limit from these data.
A reasonable explanation for this phenomenon is
that the fast permeation phase is the PCP
diffusing through the aqueous portion of the open
pores of the membranes, while the slow phase is
the PCP diffusing through the regenerated
cellulose portion of the membrane.  The
membrane may also become  saturated with PCP
at this vapor concentration.
To  provide a mass balance, another type of
exposure experiment was conducted  in which the
bundle was exposed while being completely static.
A 9000 MWCO bundle was filled with PBS buffer
and placed in the PCP exposure chamber.  After 1
hr,  the bundle was removed from the chamber and
10 mL of PBS buffer immediately washed through
the bundle and assayed. Next, the buffer was
pumped through the bundle and collected for 5 hr.
At  the end of the experiment, the bundle was
sacrificed and assayed for PCP.  This experiment
detected 1400 ng PCP in the first of 10 mL of
effluent, 376 ng in the postexposure effluent, and
61.5 ng in the bundles.  The  fraction of PCP
permeating through the  bundle was 20 percent of
the total PCP sampled by the membrane.  Only 3
percent of the PCP remained in the  membrane
indicating a low degree of permanent nonspecific
adsorption.  Therefore, while diffusion may be
slow, very little of the analyte becomes
permanently affixed to the tubing.

Table 4 contains the results from the preliminary
evaluations of PCP-antibody  loaded tubes.  The
tubes were exposed to the radiolabeled  PCP vapor
at 3 Mg/L for 15  minutes in the static exposure
chamber previously described.  Tubes filled with
the PBS-Tween solution and PBS-Tween/BSA
were suspended  in the chamber as controls. The
purpose of this experiment was to demonstrate
that the antibody-based collection system would
irreversibly bind the target analyte for later
                                                 452

-------
analysis.  Such binding is important because
analyte exposure may occur in an episodic manner
and the retention of the analyte is key to
accurately determining the exposure.

In assessing this limited data set, it is clear that
the PCP-antibody is binding the PCP diffusing into
the tube.  This can be inferred, in a non-
quantitative way, by comparing the amount of
PCP retained in the PCP-antibody loaded tubes
versus that retained in the control tubes after
dialysis as determined by LSC. More definitive
experiments are underway to quantify the relative
retentions.

Conclusions

The above data indicate a high probability of
success for the application of antibody-based
PEMs. Monitoring limits are of course
constrained by the detection capability of the
antibody-based  system. However, based on the
reported limit for the  PCP assay  (1 ng) [3] and the
diffusion measurements reported in this paper, the
limit of detection for the  PEM device should be
from  1-5 ng of PCP.  Based on a vapor
concentration of 5  ppb (arbitrarily chosen as a
representative air concentration for PCP), this
would convert to a minimum exposure time
around 20 minutes for the analyte to reach a
detectable quantity within the PEM device.

The data, even  though preliminary, demonstrate
the viability of using such PEMs.  Studies carried
out as part of this program have  also indicated
that the 6000 MWCO dialysis microtubing exhibits
                            sufficient collection efficiency for other
                            targetanalytes.  Several antibody systems are under
                            evaluation for use in PEM devices, and new
                            systems will be evaluated as they become
                            available.  These early studies indicate that it will
                            indeed be possible to apply antibodies to direct air
                            monitoring systems through the use of
                            microdialysis  tubing as a semipermeable barrier
                            which allows  vapor diffusion without significant
                            moisture loss.

                            References

                            1.  Lukens, H.R., "Solid Substrate Immunological
                            Assay for Monitoring Organic Environmental
                            Contaminants," EPA Report No. 600/1-77-018,
                            Environmental Health Effects Research Series,
                            1977.

                            2.  Giaever, I., "The Antibody-Antigen Reaction -
                            A Visual Observation," J. Immunology, 110, No 5
                            (1973) 144.

                            3.  EPA Internal Report, "Evaluation of
                            Westinghouse Bioanalytic Systems PCP
                            Immunoassay," EPA/600/X-90/146, July 1990.

                            NOTICE:   Although the research described in this
                            paper has  been supported by the United States
                            Environmental Protection Agency,  it has not been
                            subjected to Agency review and therefore does not
                            necessarily reflect the views of the  Agency, and no
                            official endorsement  should be inferred. Mention
                            of trade names or commercial products does not
                            constitute  endorsement or recommendation for
                            use.
               TABLE 2.    RESULTS OF UTILIZING MICRODIALYSIS TUBING FOR
                            SAMPLING PCP VAPOR

                                Linear Regression Analysis

Bundle
type
Correlation
coefficient
(r2)

x-Intercept
(min)

Slope
(ng/min)
PCP vapor
concentration
(ng/L)
Effective
sampling rate
(L/min)
         MWCO 6000
         MWCO 9000
0.970
0.972

0.996
0.995
6.81
6.69

3.08
2.86
19.5
14.8

20.5
17.9
71.8
55.4

58.3
52.8
0.272

0.267

0.351

0.340
                                                 453

-------
   TABLE 3.    DIFFUSION OF PCP THROUGH THE HOLLOW FIBER
               BUNDLES

                       Amount of PCP diffused through membrane
Bundle Type
MWCO6000

MWCO9000

With bundle in
(ng)
1135
857
1162
1049
chamber
(%)
52.0
54.7
69.6
70.5
With bundle
(ng)
1049
709
507
438
out of chamber
(%)
48.0
45.3
30.4
29.4
   TABLE 4.    COLLECTION OF PCP BY ANTIBODY SUSPENDED IN
               MICRODIALYSIS TUBES (15 min exposure time)
         Filling Solution

         PBS-Tween/PCP Antibody


         PBS-Tween


         PBS-Tween/BSA
PCP Amount (ng)'

      184
      162

       18
       29

       39
       57
Duplicate determinations by LSC; Exposure time of 15 minutes
                                   454

-------
                         FIGURE 1

                   Personal Exposure Monitor
                         Detach Tubing
                          from Holder
Badge Assembly
                                 Attach Leur-loc
                                 fittings and syringe-
                                 empty into analysis vial,
                          455

-------
                                      FIGURE 2

                        Exposure Chamber Used for Exposure Studies
                           Internal
                           Drive Shaft
       Plugs
       and Cover
       Friction Fit
 Removable
 Aluminum Lid
 Seamless
 Aluminum Can
Guide Housing
(Threaded to Lid)1
       10 Blade
       Fan
                                                            Shaft to Stirring
                                                            Motor,  Variable Speed
                                                            Control
                                       Thermocouple
                                       Probe
   Sample coupons (4 x 1.25cm) suspended
   from wire hooks
       Aluminum
       Plugs
                                                                                 Nickel
                                                                                 Support
                                                                                 Wire
      Quartz Wool
      Sample Tube
         (Two at
      Right Angles)
                                                                               Swagelock
                                                                               Connectors
To Digital
Readout and
Recorder
                                              456

-------
                                          FIGURE 3
                                Uptake of DNT and TNT by
                                Water Filled Dialysis Tubes
       270
CM



.CJ

0)




T3
0)
"o
O
Q
**—
O
4-*
c


o

E
                                                                                     52.5
                                                                                    -45.0
                                                                                   -37.5
                                                                                   -30.0
                 n Tubing-DNT


                 O Tubing-TNT
                                                                                    -15.0
CM







I
                                                                                            CD

-------
                                  FIGURE 4
                            PCP Vapor Diffusion,

                     MWCO 6000, and MWCO 9000 Bundle
1200
                                                                                      00
                                                                                      in
                                                                                       -
                               Exposure Time (min)

-------
                                  FIGURE 5
                              PCP Vapor Diffusion,
                          MWCO 6000, and MWCO 9000
                       Bundle Removed From PCP Chamber
   1050-n
O)
     0
                              120
160
200
240
280
320
                                 Exposure Time (min)

-------
          A REMOTE SENSING INFRARED AIR MONITORING SYSTEM FOR GASES AND
                                     VAPORS
           Levine, S.P. (A,*), Xiao, H.K. (A),  Herget,  W.  (B),  Spear,  R.
                             (C) and Pritchett, T.(D)

         (A) University of Michigan, School of  Public Health, Ann Arbor,
                              Michigan   48109-2029
          (B) Nicolet Analytical, 5225-1  Verona, Madison, WI 53711-0508.
         (C)University of California, School of Public Health,  Berkeley,
                                    CA 94720.
            D) U.S. EPA-ERT,  2890 Woodbridge,  Edison, NJ 08837-3679.
Abstract

A prototype transportable remote
sensing instrument has been built
that is capable of performing real
time quantitative analysis of gas and
vapor contaminants in air.

Introduction

In the early 1970's, papers appeared
in the literature on the
investigation of "remote sensing of
emissions" (ROSE) for air monitoring
(1-5). In all of those papers, the
instruments were large, and the data
analysis was off-line and depended on
an experienced spectroscopist
interpretation.

In the mid-1980's the use of FTIR
without remote sensing, but instead
with the use of a closed gas cell
into which the sample was pumped, and
using the classical least squares fit
(LSF) data analysis methods of
Haaland (8,9), was explored for air
monitoring applications (6-14).

Recently, several groups have begun
experimentation with ROSE-FTIR (15)
and differential absorption laser
systems for remote sensing of
pollutants at hazardous waste sites,
and potentially for fenceline
emergency chemical-release
monitoring. The first such study at a
hazardous waste site, a joint effort
in 1987 between the EPA-ERT and The
University of Michigan, resulted in
the definition of questions of
instrument stability under field
conditions, and aiming problems over
long (km) distances (16). Other
studies have been summarized (17,18).
All reported systems require manual
interpretation of data.

In this paper, we report on some of
the design and operational bases
behind this small, transportable
ROSE-FTIR system.

Experimental

The light source/optical bench weighs
16 kg, and measures approximately 20
cm (h) X 48 cm (w) X 36 cm (1). The
infra-red (IR) light source is an air
cooled Globar operating at 1300 K.
The optical bench contains a "porch-
swing" interferometer capable of up
to 2 cm""1" resolution at scan speeds
as high as 2 scans/second.

The receiver/detector module consists
of an 8 inch (20 cm) diameter
Cassagrain telescope equipped with
first-surface aluminum coated
mirrors, and a 3 inch  (7.6 cm)
diameter convex spherical secondary
mirror. The IR light detector is
liquid nitrogen cooled HgCdTe  (MCT),
                          2
with an image area of  1 mm
                                        461

-------
The receiver/detector is mounted on
an aluminum plate, and also weighs 16
kg. Dimensions of this module are 28
cm (h) X 38 cm (w) X 36 cm (1), plus
a 30 cm (1) X 25 cm (d) extension on
the telescope cover. The electrical
requirement for the complete
instrument is <10 amps of 115 V.
electrical service.

Mirrors may be used to direct the IR
beam around the monitoring site.
These mirrors are 1 foot square (929
cm2) first surface aluminum coated
mirrors. All modules, including
mirrors, but not including the
computer, may be mounted on tripods.

The computer is a Dell 310 20 Mhz
80386 system with an 80387 co-
processor, 150 Mb hard drive, 4 Mb of
RAM, a Dell VGA-Plus color card and
NEC Multisync II color monitor, a
Nicolet Fourier transform co-
processor board, and a Nicolet A/D
controller board. Software is Nicolet
PC/IR, equipped with special systems
to aid in ROSE operation.

Discussion

The objective of this work was to
design, build and evaluate a small,
transportable remote sensing (ROSE)
Fourier transform infrared (FTIR)
spectrophotometer system designed
specifically for use as a gas and
vapor air monitor for the workplace
or in emergency response situations.
The system has a maximum viewing
distance of 40 meters, and can yield
one or more analyses every minute.

The IR beam can be placed linearly
or, using mirrors, around monitoring
stations that are not linear. In
order to aid in the accurate aiming
of the beam,  the He-Ne laser beam
that emerges from the instrument is
co-axial with the IR light beam. The
laser beam has an intensity of 260

microwatts/cm2 at 1 meter distance
from the instrument,  so protective
glasses are not needed.
In theory, the beam could be moved
around the workplace using a digital
stepper motor-controlled aiming
mirror. Thus, the most important
paths within an entire workplace or
emergency response site could be
traversed in a few minutes. The
optimal choice of the beam path is a
question being explored using a large
exposure chamber at the University of
California-Berkeley.

Evaluation of the system has shown
that beam path length and detector
response, under conditions of
constant and uniform concentration,
are directly related for paths tested
up to 12 meters. Path lengths beyond
that have not yet been tested.

Most important appears to be the
presence of non-analyte contaminants
in the "clean" background air
spectrum. These non-analytes cause
baseline non-linearities in the
spectral regions in which analytes
must be determined.

The key advance that has made the use
of ROSE-FTIR and closed cell-FTIR
methods possible for air monitoring
applications has been the use of LSF
analysis of the data. Since LSF
techniques make assumptions with
regard to the linear behavior of the
baseline, "poorly behaved" baselines
(with non-linear regions) degrade the
performance of the LSF software.
This, in turn, results in higher
limits of detection (LOD), poorer
linearity, and degraded accuracy and
precision.


The effects of temperature (10-35 °C)
and relative humidity (20-85%) have
been evaluated and found to be
minimal. However, when the instrument
is moved between monitoring sites and
the telescope optics are realigned,
minor variations in the baseline may
be significant with respect to the
performance of the LSF method.

The solution of this problem is
therefore central to the use of ROSE-
FTIR methods. In initial tests in the
                                        462

-------
laboratory, this problem has been
solved through the use of negative
and positive least squares fitting.
With an analyte vapor mixture of five
components and a seven component non-
analyte mixture spectrum, both at 1
ppm concentration in ambient air per
component, the mean recovery of
analyte was 103% with a standard
deviation of 10%. Without this
method, the mean recovery degraded
significantly, as did the precision
of the results.

The use of an iterative classical
least squares fit (ILSF) approach has
also been evaluated for the
identification of unknown compounds
in the spectra of mixtures of vapors
in air (19). This method appears, in
preliminary tests, to be capable of
identifying unknown substances in
mixtures. However, field testing of
this method has not yet been
performed.

Conclusions

1. A transportable remote sensing
ROSE-FTIR instrument has been
designed, built and tested under
controlled conditions.
2. Positive and negative LSF methods
can be used, under controlled
conditions, to compensate for the
presence of non-analytes in the
background spectrum.
3. Iterative LSF methods can be used,
under controlled conditions, to
identify unknown components of the
spectra of mixtures.
4. Further testing is needed.

Acknowledgements

We thank the Environmental Response
Team (EPA-ERT) (contract 68-03-3255,
and related contracts) and the
Centers for Disease Control (CDC-
NIOSH)  (research grants 1-R01-02404
and 1-R01-02666)  for their generous
support.  In addition, we would like
to acknowledge the support of the
Office of the Vice President for
Research at The University of
Michigan.
References

1. Analytical Methods Applied  to Air
Pollution Measurements, R.K. Stevens
and W.F. Herget,  Eds., Ann Arbor
Science, Ann Arbor, MI, 1974.
2. McClenney, W.A., W.F. Herget and
R.K.  Stevens, A Comparative Review of
Open-Path Spectroscopic Absorption
Methods for Ambient Air Pollutants,
Chap. 6 in Ibid.
3. Barnes, H.M.,  Jr., Herget,  W.F.
and Rollins, R.,  Remote Sensing of
SO2 in Power Plant Plumes Using
Ultraviolet Absorption and Infrared
Emission Spectroscopy, Chap. 12 in
Ibid.
4. Herget, W.F. and J.D. Brasher:
Remote Optical Sensing of Emission.
Appl. Opt. 18: 3404 (1979).
5. Herget, W.F.:  Analysis of Gaseous
Air Pollutants Using a Mobile  FTIR
System. Am. Labs. 72 (1982).
6. Herget, W.F.,  J. Staab, H.
Klingenberg, and  W.J. Riedel:
Progress in the Prototype Development
of the New Multicomponent Exhaust Gas
Sampling and Analyzing System. Soc.
Automot. Engin. Conference, Feb.,
1984, paper no. 840470.
7. Herget, W.F. and S.P. Levine:
Fourier Transform Infrared (FTIR)
Spectroscopy for  Monitoring
Semiconductor Process Gas Emissions.
Appl. Indus. Hyg. 1: 110 (1986).
8. Haaland, D.M.  and R.G.  Easterling:
Improved Sensitivity of Infrared
Spectroscopy by the Applications of
Least Squares Methods. Appl.
Spectrosc. 34: 59 (1980).
9. Haaland, D.M., R.G. Easterling and
D.A.  Vopicka: Multivariate Least
Squares Methods Applied to the
Quantitative Spectral Analysis of
Multicomponent Samples. Appl.
Spectrosc. 39: 73-84 (1985).
10. Strang, C.R., S.P. Levine and
W.F.  Herget: Evaluation of the
Fourier Transform Infrared (FTIR)
Spectrometer as a Quantitative Air
Monitor for Semiconductor
Manufacturing Process Emissions.
Amer. Ind. Hyg.  Assoc. J.  50: 70-78
(1989).
                                         463

-------
11.  Strang,  C.R.  and  S.P.  Levine:  The
Limits  of  Detection for  the
Monitoring of  semiconductor
Manufacturing  Gas and Vapor  Emissions
by  Fourier Transform  Infrared
Spectroscopy.  Amer. Ind.  Hyg. Assoc.
J.  50:  79-84  (1989).
12.  Ying,  L.S., S.P.  Levine,  C.R.
Strang  and W.F. Herget:  Fourier
Transform  Infrared Spectroscopy  for
Monitoring Airborne Gases and Vapors
of  Industrial  Hygiene Concern. Amer.
Ind. Hyg.  Assoc.  J.,  50:  354-359
(1989).
13.  Ying,  L.  S. and S. P.  Levine,  The
Use of  Fourier Transform Infrared
Least-Squares  Methods for the
Quantitative  Analysis of Multi-
Component  Mixtures of Airborne Vapors
of  Industrial  Hygiene Concern. Anal.
Chem.,  61:  677-683 (1989).
14.  Small,  G.W.,  R.T. Kroutil, J.T.
Ditillo, and  W.R.  Loerop.  Detection
of  Atmospheric Pollutants by Direct
Analysis of Passive Fourier  Transform
Infrared Interferograms.  Anal. Chem.
60:264-269 (1988).
15.  Spartz,  M.L.,  et  al,   Evaluation
of  a Mobile FT-IR  System  for  Rapid
VOC Determination,  Amer.  Environ.
Lab.,  15-30 (Nov.,  1989).
16.  Levine,  S.P.,  L.S.  Ying,  C.R.
Strang,  and H.K. Xiao,  Advantages  and
disadvantages  in the  use  of FTIR
spectrometers  for  monitoring  airborne
gases  and  vapors of industrial
hygiene concern. Appl.  Indus.  Hyg.,
Appl.  Ind.  Hyg., 4: 180-187  (1989).
17.  Proceedings of the Symposium  on
Measurement of Toxic  and  Related
Pollutants,  EPA/AWMA,  Research
Triangle Park, NC,  1989,  1990.
18.  Minnich, T.R.,  R.L. Scotto, R.H.
Kagann,  O.A. Simpson,  Optical Remote
Sensors Ready  to Tackle superfund,
RCRA Emissions Monitoring Tasks,
Hazmat World,  42-59 (May,  1990).
19.  Xiao,  H.K., Levine, S.P.,  D'Arcy,
J.B.,  Iterative Least Squares Fit
Procedures for the Identification of
Organic Vapor  Mixtures by FTIR
Spectrophotometry.  Anal.  Chem., 61:
2708-2714  (1989).
                                              DISCUSSION
MAURY FEE: In your expert system, did you have to reduce the interference
by successive subtraction, like a water vapor or CO2, in identifying your species?

STEVE LEVINE: The iterative least squares, which is the algorithmic heart of
the expert system, does that essentially by doing a least squares fit and getting an
optimal fit through several successive iterations. Instead of doing a classical
library search, we'll take the spectra we think are there one at a lime, and look
at the residual. What it does is attempts, through the windows that have been
chosen in the spectra, to do that with least squares fitting. So the answer is yes,
but not in the way that people used to do it.

JUDD POSNER: Have you ever considered the use of neural networks to make
it an artificial spectroscopist?

STEVE LEVINE: I did not pay him to ask that question! In fact, we had just
published a paper on neural nets in Analytical Chemistry. We did indeed try it.
The difficulty with neural nets is that you have to train them. To train them you
have to be able to see the peaks. When you're looking at the FTIR spectrum of
environmental or industrial gases or vapor mixtures you can't see a lot of the
peaks. And so the neural net can't be easily trained. The net failed at any
concentration below about 50 ppm for anything above about three component
mixtures that we tried. It was an interesting idea. We were hoping to use it as a
pre-screening tool for the iterative least squares to speed up the process and
reduce the algorithmic load on the system.  But aside from producing an
interesting publication, it did not work.

BRIAN PIERCE: My question concerns the localization of leaks throughout
the plant. Could  you distribute an array or retroreflectors and then direct your
source at these — to enable the ability to localize such a leak?
STEVE LEVINE: That of course is what we're ultimately hoping to do. At this
point we're hoping to make everything work with a single beam, manually
operated. Again you have to look at the things that MDA has done with their
DIAL laser system and their digitally controlled mirror for moving the beam
around, and what the Army has done with their seven position moving mirror that
surveys the battlefield in their XM21 FTIR. So this is something that others have
done, and we want to be able to do it in the workplace. We haven't done it yet.

DON LAVERY: On the qualitative part of your study, you're doing something
rather similar, I think, to what folk sat the Atomic Energy Commission of Canada
have done in the UV. In their library search they look first for the candidate
spectrum that will explain the largest amount of variation. And then proceed
through the secondary and tertiary candidates so that they pay the most attention
to the most important first. Have you done anything like that?

STEVE LEVINE: We've tried three strategies. One is to take the entire spectral
library in one array and see which combination of fits matches the known peaks
that we have best. That turned out to be the least satisfactory and the slowest. We
then tried the set building method where we go through the library one compound
at a time and look for the best matches that way. And the third is the set reduction
method where we start with 16 compounds in the library at a time, and match the
best for each 16, and attempt to squeeze the positives down to the fewest number.
The set building method which, in a way, parallels what you say the Canadians
have done on the U V system, is the method that is the fastest and has given us the
best results. The difficulty with me giving you a definitive answer is, again, it's
not robust. We have one publication on it. It needs more work.
                                                     464

-------
               ADRIAMYCIN EXPOSURE STUDY AMONG HOSPITAL PERSONNEL
           R  L  Stechenson   C  I  H.,  Thomson  Consumer Electronics Inc., Marion. IN.
           CHRice   ?hD.   C.I.H.,  andJ.Dimos, M.S.. University of Clnc.no.t!,
               Department  of  Environmental  Health, Kettering Labs, Cmti., OH.
                                   ABSTRACT
Using antineoplastic drugs is one of the routine treatment regimes employed in
combatting cancer.  Nearly 250,000 cancer patients are treated annually with
chemotherapeutic agents usually in either hospital outpatient or inpatient
settings, however, some patients receive these drugs in physicians' offices or
outpatient clinics.1  The number and types of health care professionals who
are potentially exposed to antineoplastic drugs includes but is not limited to
4,000 medical oncologists, 10,000 oncology nurses, 30,000 pharmacists, and
even greater numbers of staff nurses and support personnel.1*2

A project was undertaken to utilize a visible light fluorescent method
previously developed at the University of Cincinnati, to document areas of
adriamycin contamination (skin and work surfaces).  Specific aims of the study
were to compare the degree of contamination among pharmacists or pharmacy
technicians, intravenous (I.V.) therapy administration personnel, oncology
nurses, patient care nurses, maintenance workers, and hospital laundry
workers, and to describe the population at risk of dermal exposure to this
antineoplastic agent and recommend measures to prevent exposure.

Sixty-four observations/monitoring sessions for adriamycin exposures in a
hospital setting were conducted from June - August 1988, for dermal contact
with adriamycin.  No dermal exposures to adriamycin among the hospital staff
monitored were found.  In addition, no adriamycin contamination was documented
on any work surfaces.  There were however, several noteworthy findings.  The
ability to detect fluorescence of adriamycin (2 mg/ml to 0.002 mg/ml) applied
to some of the disposable latex gloves, disposable gowns/smocks, toweling,
front covers of the infusion pumps, and several other work surfaces typically
found in the hospitals surveyed, varied according to the material/work surface
and the concentration applied.  Understanding of the sensitivity of the
methods and potential interferences will greatly aid in the interpretation of
positive results.
                                        465

-------
                                 INTRODUCTION
Background

Antineoplastic drugs such as adriamycin, cyclophosphamide, methotrexate,
mitomycin, dacarbazine, and cisplatin, are used in cancer treatment.   Several
of these drugs produce mutagenic, carcinogenic, and teratogenic effects in
some cancer patients.3>4>5  Clinical research has been concerned with
patients and their health status following drug administration.  Relatively
little attention has been given to persons exposed to antineoplastic  drugs
occupationally, during preparation and administration of the drugs, or
following treatment of the patient.  However, there has been a growing number
of studies in this area, some of which have shown mutagens or thioethers  in
the urine and sister chromatid exchanges among personnel regularly handling
cytotoxic drugs.6«7>8

Some 25 antineoplastic drugs are commonly used in cancer therapy.5
Thirty-two agents are commercially available for treatment, and another 80 are
in clinical development.1  Many of these modern antineoplastic drugs  are
highly toxic.  Health care personnel who formulate, administer, and clean
surfaces that have contacted these drugs may be at risk of developing a number
of adverse effects, including cancer and fetal loss.  The population  presently
estimated to be occupationally exposed to antineoplastic agents includes, but
is not limited to, thousands of employees in the pharmaceutical manufacturing
plants, 30,000 hospital pharmacists, 4000 medical oncologists, 10,000 oncology
nurses, and even greater numbers of general staff nurses and support  personnel
in hospital laundry maintenance and housekeeping.  Effects of chronic exposure
to these drugs at very low dosages, as would be expected occupationally,
cannot be predicted with presently available data.  Only two studies  have been
published reporting any environmental sampling for antineoplastics in the
hospital setting.  Both of these papers reported airborne levels of
antineoplastics, though neither study developed acceptable methods for
generalized use in airborne exposure assessment.9t10  |n addition, no
studies are available to confirm the effectiveness of engineering controls,
protective apparel, or work practices, along with proper handling and disposal
techniques for controlling the risk of contact with these drugs.  Few if any
reports have been published which examined drug contamination of hospital
surfaces, staff clothing, or soiled bedclothes or methods to remove residual
drugs.

Antineoplastic drugs include alkylatlng agents, antimetabolites, antimitotic
agents, antibiotics and other drugs.  The main therapeutic purpose is to
destroy cancer cells by blocking various biochemical pathways.  The specific
site of action varies, depending on the particular class of agent.  The
general mechanism of action is either through direct interaction with DNA or
inhibition of nucleic acid synthesis.11  Many of these drugs have been shown
to be carcinogenic, mutagenic and teratogenic in experimental systems, and
therapeutic doses of antineoplastic drugs have been associated with the
                                    466

-------
 development  of  secondary  tumors  in patients receiving chemotherapy.  Aside
 from  their actions  on  tumor  cells, antineoplastic agents can interfere with
 normal  body  cells resulting  in damage and, in some cases, cell death.11

 Surgery,  radiation, and chemotherapy are three types of medical treatment
 commonly  employed to combat  cancer.  Chemotherapeutic agents such as
 adriamycin,  are used because of  their cytotoxicity.  Adriamycin is one of the
 most  widely  used of all the  anticancer agents.  It is frequently used to treat
 tumors  characteristic  of  leukemias, lymphomas, Hodgkins disease, and
 carcinomas of the breast, ovaries, bladder, stomach, lung, thyroid, and
 bronchus.  Most of  the antineoplastic agents currently in use today are
 supplied  as  powders in vials or  as liquid solutions requiring reconstitution
 or dilution  prior to administration by intravenous or parenteral injection.
 Often pharmacists wearing gloves and protective smocks handle these drugs in
 biological safety cabinets.  The concentration is usually 2 mg/ral adriamycin
 hydrochloride in sterile saline  solution and the volume is adjusted for each
 patient using additional sterile saline for dilution.  The recommended dosage
 schedule  (adriamycin) for adult  patients is 60-75 milligrams per square meter
 (mg/m2) of body surface as a single I.V. infusion.  The drugs are
 administered in various schedules, such as once every three weeks,  or on three
 successive days every four weeks, until a total dose of 550 mg/n^
 (adriamycin) has been given.12  The exact regimen depends on the drugs used,
 type  of cancer and the health status and responsiveness of the patient.
 Patients  may receive drug therapy in a variety of settings; hospital
 inpatient, hospital outpatient,  in the physician's office or in the home.

 A variety of personnel are potentially exposed to the antineoplastic drugs
 including nurses, doctors, and pharmacists who prepare and administer the
 drugs,  and maintenance and housekeeping staff, who repair, clean, and/or
 dispose of equipment following administration of the drugs or work in the
 rooms or  offices where the drugs were administered.   Adrianycin, like most
 other drugs, is often not fully  utilized by the patient (the dose administered
 is not  fully absorbed: some of the drug is excreted as is and some is excreted
 in metabolized forms).  Therefore, vomitus and excreta may contain the drug
 and/or  its metabolites.  Housekeeping and custodial staff nay be exposed
 during  routine operations.  Patient care personnel must handle bed linens
 contaminated with vomitus and excreta which may contain drugs.   Unprotected
 laundry workers may unknowingly  transfer drugs froa the linens to their
 hands.  In general,  contaminated waste,  bedlinens, vomitus and excreta may be
 handled by a nunsber of persons involved in either treatment,  patient care or
 facility maintenance and the extent to which the personnel contaminate their
 skin as a result of contact with drugs,  waste or soiled linens has  not been
 documented.   Other ways that the antineoplastic drugs can be released into the
work environment include contaminated packaging (broken vials damaged during
 shipping), powders and liquid sprays (aerosols) released during preparation,
administration and routine cleanup operations,  spills or leakage froa
syringes,  I.V.  bags, residual contamination on used  syringes, gloves,  linens,
vials, I.V.  bags,  and tubing.  Although volatilization is not a property of
the currently availablp cytotoxic agents,  aerosolization of the drugs can
occur during preparation and administration.   Routes  of entry into  the body
                                  467

-------
are through skin absorption (dermal), inhalation of aerosolized drug,
accidental self-innoculation and iagestion.  Ingestion can occur during
mouth-breathing, smoking, eating, drinking, or other hand-to-mouth
contamination.  Direct skin contact and inhalation of aerosolized drug are
often the greatest sources of exposure.

Adriamvcin

Adriamycin, also known as doxorubicin, is a red crystalline solid that is
soluble in water, aqueovs alcohols and methanol.  This cytotoxic antibiotic is
isolated from cultures of Streptomyces peucntius.  It is produced by three
companies; one in Japan, one in Italy and by one domestic manufacturer.
Spectrofluorometric methods have been used for identification and estimation
of the drug in biological fluids and tissues.12

By knowing the excitation and emission wavelengths characteristic of a
compound, one can use the fluorescence phenomenon to identify and quantitate
such compounds.  One of the physical characteristics of adriamycin is that it
fluoresces when activated by certain wavelengths of visible and short wave
ultraviolet light.  In prior research studies conducted by Rice, Van Raalte,
and Dimes et. al.13 at the University of Cincinnati, a spectrophotometer was
used to characterize the abs'orption spectrum of adriamycin hydrochloride in
saline solution with lactose, as it is constituted for patient
administration.  They found that absorption in the visible range took place,
with a peak at 470nm.  Using a spectrofluorometer they examined the
excitation/emission spectrum for adriamycin hydrochloride in saline solution
and found a maximum intensity occurring at 580 nm.  The examination of
fluorescence excitation/emission was confined to the visible region since
ultraviolet illumination was not considered as an option for the project.
                                      468

-------
                              EVALUATION METHODS
To insure the easy availability of the equipment used for this project, only
readily accessible materials were considered for the various components shown
in Figure I.  A Kodak model AF-1 Ektagraphic slide projector was used as a
light source to stimulate fluorescence; the optical system of the projector
was equipped with a condensing lens and an infrared filter.  The projector was
equipped with a 300 watt tungsten-halogen projection lamp and a gliss filter
(BG-12 4084 Filter) which selectively passed short wave (blue) visible light
was placed into the slide projection compartment.  A 35mm single lens reflex
camera with a Vivitar 55mm 1:2.8 macrolens and a Kodak Wratten number 21
gelatin filter (75mm x 75mm) was used to photograph the fluorescent emission
from adriamycin.  The Wratten filter absorbed the stimulating blue light
emitted by the light source, allowing only the orange-red fluorescent glow of
the adriamycin to be photographed.  Sunglasses were worn during visual
observations to filter out the interfering light emitted by the stimulating
light source; ultraviolet and blue filtering sunglasses manufactured by Sun
Tiger (Pasadena, CA) block transmission of light below 550 nm and were used in
this research.  To insure constant intensity and maximize sensitivity,
attempts were made to maintain the background light levels at a minimum.  All
photographs were taken with the stimulating light source (projector) and
camera held at 20-25cm from the fluorescent materials, and the angle between
the light source and camera held to less than 45 degrees.  With background
light levels under 10 lux, using Ektachrome 160 tungsten film, exposure times
between 1/4 and 1 second, and a maximum aperture setting of 2.8, the presence
of adriamycin fluorescence on test materials was demonstrated with
concentrations ranging from 2.0 to 0.002 mg/ml placed on various materials
including but not limited to stainless steel, benchtop absorbent padding, a
cotton lab coat cloth and latex glove material.  These materials were felt to
be typical of the types of materials on which antineoplastics might spill or
leak in the clinical setting.  Orange-red fluorescence was observed on all
test materials at all concentrations except for the most dilute which was not
observed to fluoresce on stainless steel or latex.  Monitoring was conducted
on the worksurfaces, protective clothing and exposed skin both prior to and
after handling adriamycin itself or materials possibly contaminated with
adriamycin.
                                 469

-------
                                    RESULTS


Sixty-four separate monitoring sessions for adriamycin exposures in hospital
environments were conducted from June - August 1988, for dermal contact with
adriamycin.  The various jobs monitored for adriamycin exposure included
full-time pharmacists, pharmacy interns, technicians, physician assistants,
nurses, laundry workers and maintenance workers.  The areas and work surfaces
monitored for adriamycin contamination included chemotherapy preparation
areas, outpatient departments, filters in biological safety cabinets and HVAC
systems, hospital laundry areas, and chemotherapy infusion equipment.  No
dermal exposures to adriamycin among the hospital staff monitored were found.
Many of the pharmacists monitored were double gloved and wore protective
smocks when they mixed adriamycin.  Furthermore, all the hospital personnel
surveyed followed good work practices when handling antineoplastic drugs.  In
addition, no adriamycin contamination was documented on any work surfaces.
There were however, several noteworthy findings.

The ability to detect fluorescence of adriamycin (2 mg/ml to 0.002 mg/ml)
applied to some of the disposable gloves, disposable gowns/smocks, toweling,
front covers of the infusion pumps, and several other work surfaces typically
found in the hospitals surveyed, varied according to the material/work surface
and the concentration applied.  For example, the ability to detect
fluorescence on some types of latex gloves and especially orange-red colored
gloves, and on stainless steel, was reduced with the more dilute
concentrations of adriamycin.  Allowing the eyes time for dark adaptation may
play a role in increasing one's ability to detect fainter fluorescence over
smaller areas.  The increased illuminance of background light levels in the
survey area which were more than the optimum range of 10 to 35 lux, provides
for an additional interference problem.  The excitation BG-12 4084 filter
mounted on the projector was chosen due to its availability and because it has
a transmission peak at AOOnm; it passes relatively little of the fluorescence
stimulating energy of 480 nm wavelength.  While the filter was adequate,
fluorescence intensity would likely increase with the use of a filter with
peak transmission at 480nm.  Lastly, some difficulties arose in conducting the
field evaluation using the bulky equipment.  No doubt, miniaturization of the
detection system would provide greater acceptance for its use throughout the
health care environment.
                                  CONCLUSIONS


This  research demonstrates that a unique fluorescent detection system can be
used  to reduce present uncertainties involved in occupational exposure to
antineoplastic drugs.  Fluorescence detection provides a simple means of
measuring the contacted area.  This method is quite useful in assessing the
adequacy of cleanup after the drug has been spilled.  The method is sensitive,
minimal equipment is required; and very little training is needed to enable
personnel to monitor their own work areas and skin.  For less than $100.00,
along with a slide projector, any work area can be monitored for adriamycin
contamination on a continuing basis by visual observations.  Further research
is needed to define the limit of detection of visible light stimulated
fluorescence detection of adriamycin and optimize the method for use in the
field.  The use of visible light to stimulate fluorescence may have broader
applications in industrial hygiene and dermal exposure and surface
contamination studies.  Stimulating light sources equipped with several
interchangeable filters could allow for rapid detection of a number of
compounds.
                                       470

-------
                                  REFERENCES
1.  DeVita, V. ed., Cancer-Principles and Practice  of  Oncology, 2nd Ed.,
    Lippincott, 1985.

2.  United States Department of Labor,  Bureau  of Labor Statistics, "Employment
    by industry and occupation...hospitals." p.1675, 1983.

3.  Sorsa, M. et al.   "Occupational  exposure to anticancer drugs-potential and
    real hazards." Mutation Research,  154:135-149,  1985.

4.  Solimando, D. and  Wilson,  J.   "Demostration of  skin fluorescence following
    exposure to doxorubicin."   Cancer Nursing,  6:313-315, August 1983.

5.  Vaughn, M. and Christensen, W.   "Occupational exposure to cancer
    chemotherapeutic drugs:  a  literature  review." AIHC Jour., 46:B8-B16, June
    1985.

6.  Falck, K.G., Grohn,  P.,  et al.,  "Mutagenicity in urine of nurses handling
    cytostatic drugs." Lancet, 1979.;1:1250-1.

7.  Rogers, B. (1987). "Work practices  of nurses who handle antineoplastic
    agents."  AAOHN Jour.  #35, pp.24-31.

8.  Norppa, Sorsa, Vainio,  et  al., "Increased  sister chromatid exchange
    frequencies in lymphocytes of  nurses  handling cytostatic drugs." Scand.
    Jour. Work and Environ.  Hlth., //6,  pp. 299-301. 1980.

9.  Kleinberg, M. and  Quinn, M. "Airborne drug levels  in a laminar-flow hood."
    Am.  J. Hosp. Phann.,  38:1301-1313,  1981.

10. Neal, A., et al.   "Exposure of hospital workers to airborne antineoplastic
    agents." Am. J. Hosp.  Phar., 40:597-601, 1983.

11. Vanderpool, H.  "The  ethics of clinical experimentation with anticancer
    drugs." in Cancer  Treatment and  Research in a Humanistic Perspective,
    Gross and Garb, eds.  1985.

12. World Health Organization, International Agency for Research on Cancer,
    IARC Monographs on the  Evaluation of  the Carcinogenic Risk of Chemicals to
    Humans, Vol. #10,  IARC  Lyon, France.

13. Van  Raalte, J. "Detection  of Adriamycin by Visible Light Stimulated
    Fluorescence." Master's  Thesis,  University of Cincinnati, June 13, 1986.
                                            BARRIER FILTER
         Figure  £_,  Schematic  of fluorescence detection  equipment.
                                     471

-------
                                                           DISCUSSION
PHIL GREENBAUM: I wondered if you had checked to see if any studies had
been done as far as birth defects related to this drug?

RICH STEPHENSON: Perhaps there's some literature that has been published
on that. I can't recall them off the lop of my head, though.

JUDD POSNER: It seems to me that this red drug had a capacity for being
determined with probably the simplest of the spectrophotometers that involve the
eye, and  how much more sensitive  in general was the  UV than just looking
around for red spots?

RICH STEPHENSON: We didn't use the UV detector because of the hazards
associated with the UV light. We went with something that posed less of a hazard,
the  visible light source.

JLDD POSNER: How much more sensitive was the visible light measurement
than the eye could see? I mean,do you have some ideaabout what kind of increase
in sensitivity that gave you?
RICH STEPHENSON: That wasn't pan of my thesis. It perhaps would be an
interesting topic for additional work.

HARRY SALEM: You slated that there was no dermal contamination, yet I
observed from your slide the pharmacist was wearing double gloves and short
sleeves. Was the potential dermal contamination tested on the bare arms or under
the gloves or protective clothing?

RICH STEPHENSON: Before the pharmacist or intern mixed or applied the
drug, we looked at the hands and the arms, and any exposed skin surface. And
definitely before they donned any gloves. And then we looked at it afterwards.
In prior studies done by Rice and VanRaulty, they did find some contamination.
But I think just  knowing that we were present in the hospital environment and
telling the participants what we were looking for and why we were there, there
was a learning  curve that happened right on the spot. So they took extreme
caution to fol low good work practices and not spi 11 any on their hands or clothing.
                                                                     472

-------
                   REAL-TIME PERSONAL MONITORING IN THE WORKPLACE

                                USING RADIO TELEMETRY
           Ronald J. Kovein

  National Institute for Occupational
           Safety and Health
             Paul A.  Hentz

  National Institute for Occupational
           Safety and Health
 Disclaimer

 Mention of a company name or product does
 not   constitute   endorsement   by   the
 National  Institute   for   Occupational
 Safety  and Health.

 Abstract

 A system used to radio transmit data from
 remote  locations within  a workplace to a
 personal   computer     for     immediate
 interpretation  has  been  developed by
 NIOSH researchers.   The system consists
 of several radio transmitters  and  a base
 receiver that is capable  of multi-channel
 reception.   Exposure data obtained  from
 any  direct-reading  instrument  with  a
 recorder output  signal can be displayed
 and  stored at  the  computer.   The  worker
 being monitored carries the instrument in
 a    backpack,   along   with   a    radio
 transmitter.     Using   telemetry,  the
 concentration of the  airborne contaminant
 under  study  can  then be  plotted  on a
 video monitor  for  immediate assessment.
 If there is  more than one  worker  under
 study at a time, multiple exposure  curves
 can  be displayed on the screen.

 The  base   receiver    is   a   commercial
 (frequency)   scanner  that    has   been
 modified   to   accept   RS-232    serial
 communication.   The  manual  keypad has
 been removed so that channel selection is
 accomplished  through  system  software.
 The  radio  transmitter is  similar  to  a
wireless telephone; there are no wires to
 entangle.  The radio  telemetry system
allows  a  worker  unrestricted movement
within its effective range.  It also can
be used to monitor up to five  individual
workers per program execution.

A  case   study  involving  a  furniture
refinisher's   exposure   to   methylene
chloride is described here to demonstrate
the  utility  of  radio  telemetry.   The
worker carried a radio transmitter that
was  attached  to a  photoionization air
analyzer.  Qualitative methylene chloride
exposures  were remotely monitored  on a
video   monitor   throughout   the  day.
Increases in exposure levels, due to job
tasks,  work   practices,   and  emission
sources  were  immediately  identified  so
that corrective action could be taken at
that time.

It is  important  for  researchers  who are
developing    real-time    monitoring
techniques  to consider  the  procedures
discussed   in  this   report.      Also,
researchers  conducting   field  studies
should be cognizant of  the variety  of
real-time monitoring techniques  and use
them  to  their  advantage in  evaluating
worker exposure.
                                         473

-------
Introduction

In  recent  years,  researchers  at  the
National   Institute  for   Occupational
Safety  and  Health  (NIOSH)  have  used
microcomputers, data  loggers  or coaxial
cables,   and  video-taping   techniques
during  hazard control  studies to  help
acquire  real-time  exposure  data.*1'2'3'
The   data   source   was   a   portable
direct-reading instrument that measures a
worker's exposure  to  a  hazardous  vapor,
gas or  dust.    Such instruments measure
the concentration of airborne pollutants
through  detection by flame  ionization,
photoionization,    electrochemical
reactions,   infrared   and   ultraviolet
radiation, and chemiluminescence. *4)  Most
of  these devices  provide  a  continuous
analog signal through an output connector
that is proportional to  the concentration
detected.  The signal can be routed to a
strip  chart recorder  for  a  continuous
record  of exposure  levels,  or  it  can be
routed  to an  electro-mechanical  device
(e.g.,  a  solenoid  valve)  for  process
control.   Or,  the  analog signal  can be
converted   into   an  audio   signal  and
transmitted  over  a  radio  channel  to  a
distant receiver for immediate processing
by a microcomputer.

The  radio  telemetry system,  discussed
herein,  was developed  to   improve  this
real-time    exposure   monitoring    by
allowing:    (1)     the  acquisition  and
analysis  of exposure data  at  locations
remote  from non-stationary sources and
(2)   immediate   access   to   exposure
information  that   is  not possible  with
data loggers.  The radio telemetry system
overcomes  the distinct  disadvantage of
delay by continuously supplying data to a
personal computer  throughout a monitoring
session.'A*   An interface modification of
a  commercial (frequency) scanner  allows
the computer to selectively tune  up to
five  separate  frequencies  through its
RS-232 serial port (i.e. the system would
permit  monitoring  of  five  individual
workers  per  program execution).    The
 
-------
for transmission over the radio channel.
Using   frequency-shift   keying   (FSK)
modulation,  the  advantages  include  an
error   rate    that    is    essentially
independent of  signal  amplitude,  equal
per-digit error probabilities for a mark
and   space,   and  simple   noncoherent
detection  without need  to process  the
carrier.

The base receiver, a modified Regency Z60
Programmable  Scanner,  contains the  FSK
demodulator   needed   to   reconvert   a
transmitter's   audio-frequency   signals
(FSK tones) back to binary signals.  The
Z60's  normal  keyboard  programming  has
been  replaced with an  RS-232  interface
connection to the  IBM AT*  or compatible
computer.   Programs in BASIC were written
during  system  development  to  evaluate
performance; C-language programs are used
in the field  for data collection.

The   input   section    of   the   remote
transmitter   consists   of   an   8-bit
successive   approximation
analog-to-digital  (A/D)  converter.   The
A/D converter operates continuously in a
free-running  mode.     Each  frequency-
synthesized transmitter can be programmed
to  operate on  any  frequency in  the FM
broadcast  band.*6'    Selection  of  the
carrier  frequency  is  important  since
reception  can  vary  from one  broadcast
region  to  the next.    Engineering firms
that  specialize in radio frequency  (RF)
design can build a transmitter to Part 15
(CFR 47) specifications.  Independent RF
testing  laboratories  can  provide  the
certification  required  by  the  Federal
Communications Commission (FCC).

Field Demonstration

The radio telemetry system's first field
use   was   in  a  furniture  refinishing
facility where the substance under study
was methylene chloride  (NIOSH recommends
that   worker   exposure  to   methylene
chloride  be  controlled to  the  lowest
feasible limit).'7'    A photoionization
air  analyzer  (Photovac  TIP  II")  was
strapped to  a rack which,  in  turn,  was
attached  to  a   tubular-framed  backpack
(Figure 2).   The  TIP  II™ comes  with a
receptacle  that  allows  an  electrical
connection to a portable chart recorder.
Instead of the chart recorder, the analog
signal was  split between  a  data logger
(Rustrak*    Ranger)    and   a    radio
transmitter.  Two air sampling pumps also
were attached to the backpack.  (The data
logger and pumps  shown  in Figure 2 were
unrelated  to the  demonstration  of  the
telemetry system, and will not be further
discussed.)    Although  data  collected
through radio telemetry could be used to
determine   if   exposure   limits   were
exceeded, the data presented here are the
result   of   a   qualitative   approach.
Instead  of  parts per million  (ppm),  DC
voltage   (an  analog   output  of   the
direct-reading  instrument)  was used  to
identify and  minimize peak exposures to
methylene chloride.  For example, 0,5 VDC
may represent a concentration of  50 ppm,
but  it  is  approximately one-half  the
exposure represented by 1.0 VDC.

During   the   field   demonstration,   the
subject  worker  and equipment  backpack
provided qualitative exposure  data from
three  separate  work  areas  within  the
facility.(8)   Figures  3  and 4 show where
furniture     finishes    were    removed
(stripped) and rinsed off, respectively.
In  the  absence  of chemical  leaks  or
spills, the worker was not subjected to a
vapor buildup in the third work area.

A   cart  provided   mobility  for   the
computer,   video   monitor,   and   base
receiver.  The system demonstrated during
test runs  that data transmissions could
be  accurately   received   over  100-ft.
distances within the facility.

Figure 5 shows  the  worker stripping the
finish from a wooden chair, while wearing
an air-purifying respirator.  (It should
be noted that NIOSH recommends the use of
either  a supplied-air  respirator,  or a
self-contained breathing apparatus with a
full   facepiece  and   operated   in   a
pressure-demand  mode  for  any detectable
concentration of  methylene chloride.) <9>
Each time  that he  completed  a  piece of
furniture, a table  or  chair,  the worker
had to  carry it  into the rinse  room to
spray  off  the chemical residue.   After
the  rinse,  the  furniture was  usually
placed outside  the  double doors  (in the
area of the cart) to air dry.  The worker
also sometimes left the  stripping room to
replenish his supply of solution before
                                          475

-------
starting another  piece.   Regardless of
the worker's activity or location, it was
important  to maintain  accurate records
during monitoring.(8)

While  the  video  monitor  displayed  the
worker's exposure to methylene chloride
vapors,  the  computer  simultaneously
stored  the  data  to  disk.    Figures  6
through 8 are reproductions from selected
data that  were stored on the disk.   The
reproductions are similar to  the original
graphs  that  were temporarily  viewed at
the work site as the exposures  occurred.
With few exceptions,  the emission sources
and work practices which led  to  increased
exposure were easily identified.

It  should  be  noted  that  researchers
should consider instrument response time
when viewing work processes.  To pinpoint
the sources and practices that contribute
most to  the overall  level  of  exposure,
the  researcher  should  avoid  choosing
instruments with a response time of more
than  10  seconds  (5  seconds  in  some
applications).   A delayed  response  may
create difficulties  in interpreting  the
relationship    between    instantaneous
exposure  levels   and  the work process
itself.<10>   If the duration  of  a delayed
response is  known,  data  that  have  been
saved on disk can be  offset to reflect
the true time for each exposure level.

Figure   6    illustrates   the   worker's
relative exposure to methylene chloride
during the removal of a table's finish.
When  the  worker was   observed leaning
directly over the table, the particular
exposure level  that  was  seen provided
evidence that worker  participation in the
elimination,  or  reduction,  of personal
exposure is  as important as engineering
controls.  The relationship between  the
exposure of a worker to  solvents and the
work method used  is often a comparison of
actions and consequences.
Dramatic improvements can be obtained in
the job environment by relatively simple
changes  in work  practices.   A  method
called Picture Mix  Exposure (PIMEX)  was
developed  at  the National  Institute of
Occupational Health in Solna, Sweden that
specifically  addresses   the problem  of
employee   awareness.<:I1^      The   method
assumes  that  exposure  depends to  some
extent on the  way the  individual employee
works  at   his/her   workplace.     Good
examples of this  are a  spray  painter's
exposure to solvents,  a welder's exposure
to welding fumes, and many other cases in
which an employee handles  the  source of
the contaminant.   Swedish  field  studies
employing   video   techniques  make   it
possible   to   identify   problems   on
videotape.  Through the use of  a video
mixer, measurements  obtained by  direct-
reading  instruments  are  superimposed at
the edge  of a monitor screen  in  a form
similar to  a  bar chart.    The  height of
the bar is proportional at all times with
the level of the signal.  Since the PIMEX
method uses radio  transmitters to route
the  measurements  to the  mixer,  video
mixing  is  performed on site  so  that
superimposed videotapes can be  available
for the company debriefing.  While NIOSH
researchers  use  a  videotaping  system,
present  video techniques  exclude video
mixing  on  location.
-------
Besides  registering a sizeable  peak in
exposure as a result of the initial water
rinse,  the  contamination is  slow,  in
terms  of breath  cycles, to  dissipate.
The initial  peak,  followed by a gradual
reduction    in    the   relative   vapor
concentration,  was  characteristic  for
both types of furniture.   Figure 8 shows
the  water  rinse   of  another  piece  of
furniture.   The series of smaller peaks
following the initial  one is  a response
to  the  jet  of  water  striking  fresh
solvent   on  unrinsed   areas   of   the
furniture.

There   appeared   to   be  a   potential
ventilation  problem in the rinse booth
that  required  further   investigation.
NIOSH  researchers were  alerted  by  the
exposure levels that were being displayed
on the video monitor (Figure  8).   Once
alerted, a  smoke  tube  was discharged in
the booth and its  trail was followed.  It
was found  that the temperature  of  the
chemical   solution  was   outside   the
chemical    manufacturer's    recommended
range,  resulting   in a  breakdown in  a
paraffin-based   vapor   barrier   which
created  unnecessary fuming and  product
loss.*8)     Through  the   use   of  radio
telemetry, immediate action was taken to
correct the problem.

Conclusion

The  usefulness  of  radio  telemetry,  a
system that instantly transmits exposure
data from direct-reading  instruments to a
microcomputer, has been demonstrated on a
field  survey at  a  furniture-stripping
operation.  The system offers  advantages
over  other  methods of data  collection.
Using telemetry,  workers  under study can
enjoy more natural movement than they can
by  being   tied   with  coaxial  cables.
Unlike   data  loggers,   radio   signals
produce  instantaneous   results.     The
researcher can view exposure  information
received  from  several  workers  and/or
processes on a moment-by-moment basis,
thus  preserving   the primary  advantage
(instant  feedback)  of   direct-reading
instrumentation.   Immediate exposure
(B>   Temperature Range:   60° to 85°F per
product  label  instructions  for  paint
remover #2105 manufactured by Kwick Kleen
Industrial Solvents,  Inc., Vincennes, IN.
determinations   are   needed   in   the
workplace to prevent employee injury and
to    advise     management.        Such
determinations    allow    the    swift
elimination of  emission  sources and the
timely development of exposure scenarios,
proper work procedures  and training aids.

Video   techniques   used   by   Swedish
researchers    have    recognized    the
expediency of radio telemetry.  Although
information that has been saved on disk
can  be later  mixed with videotape  to
produce  training   aids,   there  is  an
advantage  in  video mixing  at  the  job
site.  By combining the  results as they
occur, the  need  to mix  hours of  data
files and videotape at another  time and
place is eliminated.

Finally,    the   application   of   radio
telemetry provides  more  flexibility and
personal involvement while monitoring in
the workplace.   Personnel involved in a
field   study    are   generally    more
productive.   There  is  more interaction,
observation, and discussion.   The focal
point of this activity  centers around the
video monitor.   As  exposure information
is updated on the screen,  comments can be
made and notes  taken.   Researchers have
the   opportunity  to  take   additional
measures     in    response    to
higher-than-expected exposure  readings.
Follow-up  investigations  could  become
unnecessary when corrective  actions are
proven effective before leaving the work
site.

It is important  for researchers who are
developing    real-time    monitoring
techniques  to   consider   the  procedures
discussed   in   this   report.      Also,
researchers  conducting   field   studies
should be  cognizant of  the variety  of
real-time monitoring techniques and use
them  to  their  advantage in  evaluating
worker exposure.

Acknowledgment

We would  like  to thank Martin  Abell  of
NIOSH and  William Metzger, a NIOSH summer
employee,  for  their assistance in  the
development of  the radio telemetry system
software.
                                         477

-------
References
 1. Sheehy, John,  et al., "Methodolgy,"
    Control of  Asbestos Exposure During
    Brake Drum Service,  DHHS  (NIOSH) Pub.
    No. 89-121, U.S. Dept. of Health and
    Human  Services,  Cincinnati,  Ohio,
    1989, pp. 12-13.

 2. Gressel, Michael, et al. , "Advantages
    of  Real-time  Data   Acquisition  for
    Exposure    Assessment,"    Applied
    Industrial  Hygiene,  Vol.  3, No.  11,
    November 1988, p. 316.

 3. Caplan, Paul, et al., "Study Results
    and   Discussion,"   Survey   Report:
    Control Technology of Solid Material
    Handling, NTIS Pub.  No. PB-86-191285,
    National    Technical    Information
    Service, Springfield, Virginia, 1985,
    pp. 22-24.

 4. First, Melvin, "Sampling and Analysis
    of Air  Contaminants:  An Overview,"
    Applied Industrial  Hygiene, Vol.  3,
    No. 12, December 1988, p. F-20.

 5. Beasley, Amy  and Fairfield, Cheryl,
    Survey  Report:     The  Control   of
    Methylene  Chloride  During  Furniture
    Stripping   at  the   Association  for
    Retarded Citizens, ECTB Rpt. No. 170-
    18, U.S.  Dept. of  Health  and Human
    Services, Cincinnati, Ohio, 1990.

 6. Code of Federal Regulations. Federal
    Communications  Commission.    47  CFR
    15.239, rev. October 1, 1989.

 7. Current  Intelligence  Bulletin  #46.
    Methylene Chloride,  DHHS (NIOSH) Pub.
    No. 86-114, U.S. Dept. of Health and
    Human  Services,  Cincinnati,  Ohio,
    1986, pp. 2-3.

 8. Kovein,  Ronald   and  Hentz,   Paul,
    "Real-Time Personal Monitoring in the
    Workplace   Using  Radio  Telemetry,"
    Applied    Occupational     and
    Environmental Hygiene  (in press).

 9. NIOSH   Pocket  Guide   to   Chemical
    Hazards, DHHS (NIOSH) Pub.  No.  90-
    117, U.S. Government Printing Office,
    Washington, D.C. 1990, pp.  150-151.
10. Carlsson,  Jan-Ake,  "PIMEX:    A New
    Method   for  Worksite   Environment
    Control,"   Journal of  the  oil and
    Colour  Chemists'  Association,  Vol.
    72, No. 6, June 1989,  p. 211.

11. Rosen,  Gunnar,  "Video  Filming and
    Pollution Measurement as  a Teaching
    Aid in Reducing Exposure to Airborne
    Pollutants,"  Visualization  Methods
    and Emission Studies as Aids in the
    Control  of  Exposure   to  Airborne
    Contaminants, Vol. l, Arbete & Halsa,
    Solna,  Sweden,  1989,  pp.  inn  -
    111:10.
              ">*IO!I IRHNSMimns
  V
J9L
                    y      -
                  _d     d
          'x *?     IM «3    IM *4
          88 ' MH*    90 5 MHZ    I 01.1 MHZ
                BflSt STRTIDN
                                 11.7 MHZ
                               • S-IO* MH2
                                                   Figure l.  Radio Telemetry
                                                              Block Diagram
                                           478

-------
Figure 2.  Equipment Backpack
Figure 4.  Rinse Room
            Figure  3.  Workstation
                                       479

-------
Figure 5.   Stripping Furniture

                                                200    400    600    800   1000   1200   1400

                                                     ELAPSED TIME  (SECONDS)

                                             Figure  7.   Chemical Stripping/Water
                                                         Rinsing
                                        oo
                                                  25
 50    75    100    125    150   175

ELAPSED TIME  (SECONDS)
                                                                                     200
      WORKER LEANING OVER TABLE
           WHILE STRIPPING
                                              Figure 8,
    Peak Exposures  in
    the Rinse Room
            50    75   100   125    150   175

           ELAPSED TIME  (SECONDS)

      Figure  6.  Personal Exposure as  a
                 Result of Poor Work
                 Practices
                                           200
                                            480

-------
                                                            DISCUSSION
MATT STINCHFIELD: Did you attempt to calibrate your relative reading to
actual air concentrations, and if so, how did you do it?

RONALD KOVEIN: That's something we took for granted. We were not out
to determine the exact concentration in that room. This is like a background sniff
to identify the sources, practices, so on. In other words, along the y axis there may
or may not be 50 ppm. But in a relative fashion we know the air quality of that
room, that rinse booth, was far less superior than. say. close proximity  to the
ventilation system.

MATT STINCHFIELD: Yes. the work procedure and the corresponding spike
that we saw was apparent. Have you considered how you might go  about
calibrating your field instrument?

RONALD KOVEIN: Yes, it is highly doable at this stage. In my mind it is even
secondary to others  because it is so doable. We've tested the hardware. For
instance, the A to D in the front end of that transmitter is highly linear. We can
improve system software. It's still growing. All these things can help make the
system more user friendly and  to allow for actual concentration readings. It is
highly doable.
MATT STINCHFIELD: One final pan to that. The software you were using for
receiving the information through your RS-232- is that a commercially available
type of data acquisition software, or was that something custom?

RONALD KOVEIN: Yes. it is. But once it's on the floppy disk it's part of the
data storage and you can use just about anything you want, similar to a data
logger. It is custom in that everything was built from scratch. We're not aware of
any thing out there commercially that does quite what we do. And in fact, we think
it is quite unique, though it may be considered very basic. You'll notice  it is
simple. It is asynchronous, one-way communicalions. We have nothing sophis-
ticated or as expensive as cyclic redundancy checking, for instance. 1 don't feel
we need it. We have improvements down the way. and it is going to get better in
range. There is so much left to do. The commercial scanner by the way. is rough.
It did a great job for what we wanted it to do. but we can optimize it. We can give
it belter filtration making it much more sensitive and selective. And at the same
time we have improvements to do on the prototype transmitters. We can increase
field intensity for instance. Now with the revision of CFR 47 fora Part  15 device.
it can go from about 50 microvolts per meter up to 250. So as I say,  things are
starting for our system in both performance and quality.
                                                                      481

-------
                   IMPROVEMENTS IN THE MONITORING
                    OF PPM LEVEL ORGANIC VAPORS
                   WITH FIELD PORTABLE INSTRUMENTS
                          Gerald Moore
                       GMD Systems, Inc.


An Overview of the Scope for Monitoring

In the whole field of gas detection and measurement,  organic
compounds represent by far the  largest group of  interest  to users
due to the wide variety of these substances that are  used  in
different applications.  Measurements are made for reasons of
process control, flammability hazard or toxicity and  sometimes
for all of these reasons with the same  substance.

Before focusing on the specific area of toxic  level monitoring,
it is interesting to review the number of organic compounds and
the needs for monitoring.

One of the most widely used reference books, the CRC  Handbook of
Chemistry and Phvsics*T lists a grand total of 17,746 chemical
compounds, both organic and inorganic.  This listing  breaks down
into 4,126 inorganic and 13,620 organic compounds; i.e.,
approximately two-thirds of all the listed compounds  are
organ!cs.

However,  when  one looks for a  more specific listing  of compounds
likely to be of interest for monitoring in industry,  there  is a
vast difference in the numbers.  One of the most comprehensive
list of organic vapors in common use is that published by the
National Fire Protection Association (NFPA) in the U.S.A.2  This
listing covers the fire hazard  properties of liquids,  gases and
volatile solids and has been progressively updated over the
years.  After eliminating synonyms and inorganics, a  total of
approximately 1,250 substances  remains, suggesting that this  may
be a reliable estimate of the number of organic  compounds  in
common use for which organic vapor monitoring  may be  needed.
Clearly this number of compounds far exceeds any other group  of
compounds that require monitoring in air, but  represents  a small
fraction  (less than 10%) of the total organics listing.

When one thinks of toxic hazards, TLV guidelines come to  mind as
the accepted numerical classification.  The American  Conference
of Governmental Industrial Hygienists  (ACGIH)  recently published
a guide for TLVs3 in various countries .  In this listing there
are 972 compounds of which 355  are organic vapors having  a
defined and listed TLV.  The other 6'17 are dusts  inorganics and
organics having no listed TLV  (carcinogens, etc). Since this
paper is concerned with monitoring toxic  organic vapors,  the
analysis will concentrate on the 355 listed TLVs  in an attempt  to
put an overall context on to organic vapor monitoring needs.
                               483

-------
An Overview  of  the  Scope  for  Monitoring  (continued)

Figure  1 shows  the  results  of  this  numerical  breakdown.   From  a
starting point  of 1,250 substances  in  common  use,  those
substances clearly  of  interest  only for  their  flammabi1ity  were
next  identified.  This produced a surprisingly  small  total  of  44
substances,  typified by methane, ethylene  and  substances  such  as
corn  oil, soybean oil, etc.   Taking out  this  group, together with
the 355  listed  by the  ACGIH,  leaves a  total of  851  (68%  of  the
total number) which the author  believes  to be  toxic at some level
even  though  they have  no  formal listing.   Obvious  examples  are
members  of the  same chemical  family or series  such as  ketones  or
aldehydes, where one member of  the  family  has  a TLV listing and
the others do not.  It could  be argued that only certain  members
of the series are in fact toxic, but the author believe  that a
more  reasonable explanation Is  that  the  medical and
epidemiological data has  not  yet been  produced  and formalized.

From  a starting point  of  1,250  organic vapors,  one can only,
therefore, subtract around  4% as being purely  of Interest from a
flammabllity point of  view.  All the rest should be considered as
having either listed or unlisted toxic properties  and  might,
therefore, be of interest for monitoring in the atmosphere.  This
is quite a challenge,   requiring monitors for  over  1,200 different
compounds at toxic levels.


The Need for Sensitivity  In Monitorlng

A breakdown of  the ACGIH  listed TLVs for organic vapors  is shown
in Figure 2.  It will  be  seen that  the numbers of TLV  listings
below 100 ppb or above 1000 ppm are very small and that the
largest single groups  are in the ranges  1-10 ppm and  10-100 ppm.
A flexible direct reading instrument having a range from
approximately 0.1 — >1000  ppm would  therefore be of great utility
in measuring around 90% of the toxic organic vapors and 70% of
the listings could be  covered with  a range of 0.1 — >100 ppm.


Detection Principles in Common Use

The user has available a wide variety of detection techniques to
bring to bear on the problem of organic vapor monitoring.
Figure 3 provides a simplified overview of these techniques in
relation to one another and to the  spectrum of requirements.

Eliminating those techniques that  are highly specific  in nature,
e.g.,  colorImetric paper tapes and  those that are primarily
intended for use as  flammabi1ity monitors,  leaves the following
list  of methods as suitable for portable direct reading monitors:
                                  484

-------
          connoNLv LISTED ORGANIC VAPORS
                       1258
   LISTED ILU

    28X €355)
NON-TOXIC

3.5X (44)
    HOT LISTED
      TOXICS
         C051)
    BENZENE
  ACRVLOMITRILE
METHYL EIHVL KETONE
 METHANE
 ETHVLENE
1,4-DICHLOROBUTEHE

2>4-DICHLOROPHEHOL
          96X LISTED AND NON-LISTED TOXICS
                4X FLftHNftBLE OHLV
                        Figure I
                            485

-------
                              IZBr
00
o>
                       NUHBEB  68
                               B
                                      11
                                   1-9.9
                                                                TOXIC OBGANIC UAPOBS
                                                          DISTORTION OF LISTED TLU-TUft'S
                                                  19
               53
                                                                                               63
                                                            15

1B-99PPB  188-999 PPB  1-9.9 PPH   18-99 Pffl 188-999 PPB  1888 PPB+

             TLH-TW LISTING (flCGIH GUIDE 1996)
                                                                  Figure  2

-------
                                                       SENSITIVITY RfiNGES FDR ORGflNIC VftPDR MONITORING INSTRUMENTS
£
-4
UNITS
PRINCIPflL
CONCERN
METHODS
flVftlLfiBLE
1
1
1
V
PPRTS PER BILLION (10~9) 1 PflRTS PER MILLION
1
0.1 1 10 100 1 10 100



1
<10~6> 1 PERCENT GflS (X)
\
1000 1 10 10ft
1
t
t
1
1
1
1



COLORIMETRIC PflPER TfiPES
1
1
1
1
t
	 __ __ )
1
1
LftSER / COMERft / FTIR
1
DXIDflTION (£)
(PPM RftNGE)


1 OXIDATION <1>
1 (LFL RftNBE)
1
t
- > 1
t
1
1 OXYGEN
DEFICIENCY
< 	 THE RMflL 	 >
CONDUCTIVITY
< — ELECTROCHEM— >
Og CELLS

                                                                        Figure 3

-------
Detection Principles in Common Use (continued)


     * Closed Path Infrared Absorption        (IR)
     * Flame lonlzatlon                       (FID)
     * Gas Chromatography with FID            (FID/GC)
     * Catalytic Oxidation                    (Catalytic)
     * Photoionization                        (PID)
     * Gas Chromatography with PID            (PID/GC)


The last two categories have seen many improvements in the last
five years,  resulting in a new generation of  lightweight direct-
reading instruments for monitoring organic vapors.  The need for
specific compound identification in complex mixtures has
stimulated the development of portable gas chromatographs of high
sensitivity, suitable for field use.

Since the other techniques mentioned have been described
frequently in other published material, this  paper focuses on the
particular properties of the photoionization  detector and the
ongoing development of truly portable instruments based on the
use of this detector in conjunction with gas  chromatographic
t echnique.
General Description of PID Based Instruments

Photoionization is becoming a very popular technique for the
detection  of organic vapors at low ppm levels.  This detection
principle,  especially when combined with gas chromatography can
provide a  powerful tool in Identification and quantitation.  A
photoionization detector is similar to an FID except that the
ionization energy is provided by an ultra-violet emitting lamp.
A photon of UV radiation is absorbed by a molecule of the organic
vapor causing Ionization of the molecule.  An ion flux is
generated  between two high voltage electrodes and a detectable
current results.   The detector usually consists of a sealed
interchangeable UV lamp that emits photons of specific energy.
If this energy level is high enough to ionize the organic
molecule,  this will be detected in the ionization chamber as an
electrical  signal which is electronically converted to a measured
concentration.  If the proper lamp Is chosen, sensitivities can
be as low as 0.1  ppm, but there is a trade-off between high
energy emissions  and lamp life.  The high cost of replacement
lamp (>U.S.$500)  makes this trade-off of importance.  These
direct reading instruments are simple to use, highly portable and
reasonably cheap.  However, the range of relative responses is
very wide  as will be discussed later.  Examples are the Photovac
"TIP"4 and MSA "Photon"5 instruments.
                                 488

-------
General Description of PIP Based Instruments (continued)

As mentioned earlier,  if this type of detector is used In
conjunction with gas chromatography,  specificity improves and
limits of detection can be lowered.  Several instruments having
this configuration have been designed and brought on to the
market in recent years.  A good example is the PHOTOVAC Model
10S708.  One of the little used advantages of a PID based GC is
the possibility of using air as the carrier gas,  thus eliminating
all consumable supplies and the need  for compressed gas
cylinders.   This feature will be discussed in more detail in the
next section of this paper.
Special Features of PID Based Instruments

The versatility of a PID based direct reading portable instrument
and its ease of use has stimulated the development of a number of
commercially available instruments and encouraged their
widespread use in the field.  However, the particular advantages
and limitations of the technique are not always well understood
and this can lead to improper application.

As stated earlier, the extent to which an organic vapor is
ionized depends on the energy level of the photons emitted by the
UV lamp, relative to the level required to ionize the vapor
(lonizatlon Potential).  There are various published tables of
ionization potentials and some of these also give the relative
responses of these detectors to a variety of organic
vapors. »'>1"   UV lamps are available having energy levels
ranging from 8.5 --> 11.7 eV with the higher energy lamps
typically having shorter lifetimes.

Figure 4 provides a breakdown of the number of organic vapors
that can be measured with each of the three most commonly used
lamp types, assuming the same total number of 1,250 as used
earlier.  It will be seen that the use of a 9.5 eV lamp would
drastically curtail the usefulness of this technique and,
thereby, possibly give many "false negatives" in field use.  A
10.2 eV lamp gives a far more comprehensive general purpose
response, allowing around 85% of all commonly used organic vapors
to be measured.  It should be particularly noted, however, that
many of the compounds of most interest from a toxicity point of
view are halogenated organics which require higher energy lamps
to achieve ionization.  Therefore, the improvement in the numbers
detectable with an 11.7 eV lamp is not as superficial as might at
first appear, since the halogenated compounds are heavily
represented in this difference number.  Care should, therefore,
be taken in choosing a PID that is appropriate to the part icular
compounds in a given application.
                               489

-------
                                    1298-
CD
O
                        DETECTABLE   888
                        COHPOUNDS
                       (1258 TOTAL)  688
                                       B
                                                 275 (2ZZ)
                                                                    TOXIC OBGANIC UAPOBS
                                                                  PHOTOIONIZATION DETECTOB
                                                              NUHBEB DETECTABLE UEBSUS HUP TVPE
                                              H ui i u fim M •«> •»» i»t iui( »)
                                                   9.5eU
  IB.ZeU
WKPTffE
                                                                                                  1213 (97Z)
11.7GU
                                                                           Figure 4

-------
Special Features of PIP Based Instruments (continued)

The main advantage of a PID based portable monitor is its
universality of response.  Subject to the careful choice of lamp
as described earlier, the monitor will give reading for an
extremely wide range of organic vapors.  Typically such an
instrument is calibrated on the substance of choice,  or
alternatively on a so-called "typical" substance.  For other
substances,  there will obviously be a range of relative responses
both above and below the nominal calibration.

Unfortunately, some suppliers use an aromatic compound such as
benzene for  their "nominal" calibration.  This is undesirable
since aromatics, in general, are the most easily ionized and,
therefore, respond with most types of UV lamps.  In addition,ff
they tend to give high responses, meaning that most other
substances will have lower relative responses, in some cases as
little as l/100th that of benzene.  Clearly, a more general
purpose instrument would be calibrated on a "median response"
substance so that relative responses would be distributed more
evenly above and below the nominal calibration (this assumes that
an individual calibration for a particular substance is not
avallable).

This approach was researched using the available references, and
the result is shown  in Figure 5.  By listing all available
relative responses and scaling them to the median value, a
compound could be selected  (methyl propyl ketone) and the
distribution of relative responses of such a median-calibrated
instrument is shown.  A useful number of compounds (263 or  21% of
the total) fall within relative response range of ±25% which
could be considered sufficiently accurate for many applications.
However, it  will also be noted that almost half of the total
number are outside the range of ±80% so some care is still
necessary.  Clearly, the PID technique is wide-ranging in  its
uses but some care is necessary if general purpose uses are being
considered.
PID GC Improvements

The addition of capillary column gas chromatography to a PID-
based monitor results in an extremely flexible and portable
organic vapor monitor.  The authors have been involved with the
development of such a system during the last few years and
believe that it represents a significant advance in technique.
This technique was originally developed at the Swedish National
Defense Research Institute (FOA) and was licensed to GMD Systems,
Inc., who will be marketing it under the trade name "Autograph".
The unit is presently at the advanced prototype stage and will be
available on the market during 19916.
                                491

-------
        78--
 2 OF
 TOTAL   58
(1258)
                                       PHOTOIONIZftTIQN
                                     1EMIIUE RESPONSES TO TOXIC
                                                    mm
                             mm COMPOUND:  RETHYL PBOPVL KETONE
532(663)
                                                342(425)
                                                .WA'.V.V.ViVt'.V.'tVtVi


                                                tU**tfH»H*J**HH*
                                                111.11111111 »"l MI At CM
                                                HtiiiHWy«Yntn'H
                                                IIHIIIIIIIIllllllllll
                */-lH2          */-25Z         */-582           +/-8B2

                               BELfiTIHE BESPONSE BfUGE (COHPABO TO
                 472(588)
                                                                                 iVfujVn
-------
PIP GC Improvements (continued)

The Autograph is a fully portable PID based GC system which can
be operated either in direct reading "detector" mode with the
sample flowing directly to the PID; or alternatively in "GC" mode
permitting selectivity and identification of compounds in mixed
atmospheres.

Figure 6 shows the block diagram of the system, emphasizing the
extreme simplicity of the flow path and minimum of functional
components.

It should be  noted that no external supply of carrier gas is
needed, since outside air from the environment is used as the
carrier and cleaned of organics by passing through the sorbent
trap shown, which consists of a small tube containing one or more
synthetic sorbent materials in granular form.

The sorbent trap thus, simultaneously serves to provide a clean
air "carrier" stream and also acts as a preconcentrator for the
pollutants of interest.

The sorbent tube is tightly wound with an electrical heater
winding;  and  after a preset sampling period, the tube is
electrically  heated to around 250° C for a few seconds, thereby
desorbing the organics previously adsorbed.  The desorbed
substances pass on to the heated capillary column (typically 10
meter 0.3 nm I.D.) in which they are separated from each other.
Each component of the mixture emerges from the column after a
fixed retention time and is detected and quantified by the PID.

Since retention times are dependent on flowrate and column
temperature,  these two parameters are tightly controlled via the
microprocessor and appropriate sensors.

This technique of preconcentrating the organic pollutants via a
sorbent with  subsequent rapid thermal desorption, improves the
lower detection limit by about 50 - 100 times as compared to a
PID alone.

The preconcentrat or sorbent tube is also removable and can be
used alone as a diffusion-operated personal sampler.  By
distributing  a number of these samplers and analyzing them later,
all the functions of a normal analytical laboratory can be
performed In  the field with a single Instrument which can also be
used for survey work.

The microprocessor calculates concentration data, identifies
compounds against a known substance library and displays all
parameters on the instrument's LCD graphics display panel.  All
data Is also  stored for each sample period.
                               493

-------
 DIRECT  INLET
D
£
S
0
R
B
POUER
T
E
n
p
s
i
G
N
A
L
               MICRO

               PROCESSOR
                                        R
                                        A
                                        I
                                        E
                                  KEYBOARD
                                  DISPUW
                                  PRINT/PEG!
                                  ALARM
                 Figure  6
        PID BASED GAS CH10HATOGIAPH

             GHD  "AUTOGRAPH"
                    494

-------
PIP GC  Improvements  (continued)

Operating parameters such as column temperature, length of
sampling period, etc., can be programmed by the operator either
via an  Integral keypad or via a serial data link from an external
laptop  or other personal computer, which also serves to download
and analyze data which is stored  in the Instrument's memory
during  field use.

Typical analysis cycle times are  around 4 minutes, and it will be
appreciated that a new sample is  being collected during this
cycle,  since the sorbent continues to act as a filter to the
incoming air.  However, longer sampling periods can be used,
giving  increased sensitivity of detection, since the mass of
sample  on the sorbent  is increased.  Sensitivities down to
0.01 ppb have been achieved in this way.

The versatility of the technique  is enhanced by providing a
sample  bypass valve which diverts the incoming air directly to
the PID.  This permits rapid qualitative screening of the test
atmosphere with the concentration displayed in analog bar-graph
format  on the graphic display.  If organics are detected, the
Instrument can then be put into "GC" mode for Identification  if
requi red.

The complete instrument is compact in size (approximately 4-1/2"
high x  10" wide x 12.0" deep) and weighs approximately 10 pounds.
It is battery operated and runs for up to 8 hours on a single
charge.

The Autograph demonstrates the degree to which it is possible to
simplify and miniaturize these techniques and shows the potential
for carrying out many analytical  functions with a field-portable
inst rument.
Conclus1ons

It has been shown that there are around 1,250 organic compounds
In general use requiring organic vapor monitoring.  Approximately
96% should be considered as listed or potentially toxic and,
therefore, there may be a need for toxic level monitoring.  This
group of compounds represents by far the most numerous family of
substances for which specific monitoring is required.

The range of toxic concentrations is mainly from 0.1 --> 1000 ppm
with only very small numbers outside this range.  Specific
detectors are often available for particular organics at ppb
levels, but the remainder require a good general purpose
monitoring technique.
                                495

-------
            Conclus i ons  (continued)

            Improvement  in  PID-based  instruments have made them more
            available and easy  to use  in recent years.   High sensitivity
            broad-ranging response are  the advantages, while cost  of  the
            detector  and non-uniformity of relative  responses are  the main
            d i sadvantages.
                               and
            PID-based  portable GC  systems are  now widely  available.   This
            combination of  sensitivity,  selectivity  and portability  is
            unique.  However,  not  all  instruments take full  advantage  of the
            technique  as regards true  portability, elimination  of  carrier
            gases,  etc.  The  development  of the  GMD  "Autograph"  has  been
            described  as an  example of  good exploitation  of  this  potential.


                                           REFERENCES
             1.  CRC  Handbook of  Chemistry and  Physics,  60th Edition,
                April,  1979

             2.  NFPA  325M, Firo  Hazard  Properties of  Flammable Liquids,  Gases
                and  Volatile Solids,  19S4 Edition

             3.  Guide  to Occupational Exposure  Vulues- 1990,  American
                Conferenci? of Governmental Industrial  Hygienists,
                Cincinnati,  OH

             4.  Photovac International,  Inc.,  Huntington,  NY 11743

             5.  MSA Corporation,  Pittsburgh,  PA 15230

             6.  GMD Systems,  Inc.,  Bendersonvi11e, PA  15339

             7.  Listing of Molecular  lonization Potentials,  unknown
                textbook source,  GMD  files

             8.  Technical Bulletin  No.   11, Photovac  International,  Inc.,
                Huntington,  NY   11743

             9.  Rittfcldt,  Lars,  The  National Defense Research Institute  of
                Sweden,  personal  communication  of PID relative response
                fact ors

           10.  Langhorst,  M.L.,  "Phot oi orii zat i on Detector  Sensitivity of
                Organic  Compounds",  Journal  of  Chromatographic Science,
                Vol. 19,  February  1981
                                          DISCUSSION
JUDD POSNER: My question relates to using air as a carrier gas. My
recollection way back when I was using PID was that there was a negative beat
for oxygen and that might be a problem with using air as acarrier gas. Does it not,
in fact, lower your sensitivity somewhat?

The second thing is, having spent so much time arguing with people that tubes
are not really a very good design for passive monitoring, I'm appalled that you
should come up here and tell me that you can use that little external tube as a
passive monitor.

GERALD MOORE: You got me with a tube, I have to admit. It's kind of a neat
feature of particular techniques, so why not use it. If I could have made it look
like a passive monitor, I really would have done it out of principle.
As to the more serious point about the oxygen, we haven't actually seen any
interfering peaks that interfere with the identification. I think the more serious
problem that's been raised by several people we've talked to is possible oxidation
of the thermally unstable compounds with an air carrier either in the system, on
the column, or wherever. Not having the advantage of a nitrogen carrier, I think
that is going to be a problem for some compounds. We don't have enough
experience to know how many. We hope that the advantages of the technique
considering the wide field we're dealing with here will be greater than the
disadvantages.
                                                  496

-------
           RAPID ASSESSMENT OF SUPERFUND SITES  FOR HAZARDOUS  MATERIALS
                      WITH X-RAY FLUORESCENCE SPECTROMETRY
               H.H. Cole III, R.E. Enwall, G.A. Raab, C.A. Kuharic
                Lockheed Engineering & Sciences Co.,  Las Vegas,  NV

                           W.H. Engelmann, L.A. Eccles
               U.S. Environmental Protection Agency, Las Vegas, NV
                Abstract

Field-portable X-ray fluorescence (FPXRF)
is nationally recognized as an excellent
screening tool for inorganic contaminants
on hazardous waste sites.  However, FPXRF
is more than  a  screening  tool when used
correctly.     Properly   calibrated  and
monitored,  FPXRF produces  quantitative
data of known quality.  Albeit, the data
are often of lower quality than intensive
laboratory  analytical methods, but  one
must consider the end-use of  the data.
In  many situations,  definition of  the
spatial distribution of the contaminants
can be accomplished  most cost effectively
by  taking  numerous  FPXRF  measurements
rather   than   a   limited   number   of
laboratory  analyses of  higher precision
and accuracy.
INTRODUCTION

Sampling    procedures     traditionally
employed   for   site   characterization
revolve around the analytical laboratory.
Physical  samples  are collected  onsite,
packaged,  decontaminated  if  necessary,
and shipped to a laboratory for analysis
under chain-of-custody restrictions (see
Figure 1).  Laboratory residence time is
typically 20 to 40 days, with analytical
costs of  $150 or more  per sample using
contract   laboratory    program    (CLP)
procedures.    Analytical   results  are
compiled  into  data  packs and  shipped to
site personnel for review.  Clearly, this
approach  is costly,  time consuming,  and
affords   many  opportunities  for  the
incurrence of  errors and data  loss.   It
discourages multistage  sampling  and,  in
some  cases,  has  led to development  of
inadequate  sampling plans  in order  to
avoid the inherent expense.

The   disadvantages   of  the   foregoing
approach have  prompted new emphasis  on
the  development  of rapid,  inexpensive
field analytical methods.   In the case of
inorganic     contaminants,    X-ray
fluorescence   (XRF)   is   particularly
applicable,    especially   for   heavier
elements such  as  the transition  metals.
XRF   spectrometers   capable   of   being
transported to  field sites and operated
thereon are  commercially available.  Some
require operation in a field laboratory,
whereas   others  can   be   transported
manually and  operated  directly  on  the
site surface.   Availability of the latter
has  led  to   development   of a  field
portable  X-ray  fluorescence   (FPXRF)
approach  by   Lockheed  Engineering   &
Sciences Co. (LESC)  as part of the U.S.
Environmental  Protection  Agency  (EPA)
effort  to   evaluate and  develop  field
screening and analytical methods.

The FPXRF approach consists of performing
in situ  analyses with a field portable
instrument  (see  Figure  2).     Maximum
flexibility and minimum  data  turnaround
can  be   achieved  when   the portable
instrument  is   supported  with  a  field
laboratory  containing  equipment   for
sample  preparation, data  analysis  and
display,  and  a  laboratory  grade  XRF
instrument  for   analyzing  calibration
standards and  confirmatory sample.   If
desired,  data  and a   report  can  be
delivered to   site   personnel  prior  to
demobilization.
                                         497

-------
X-MET 880 FIELD SPECTROMETER

Field  implementation  employed  in  LESC
studies to date  is  limited  to the X-Met
880.    This   instrument   is  a  field-
portable, energy-dispersive spectrometer
commercially    distributed   through
Outokumpu Electronics,  Inc., Langhorne,
PA.     It   is  self-contained,   battery
powered,  and  weighs  8.5  kg.    These
characteristics, and the fact that it is
hermetically sealed and can therefore be
decontaminated, allow operation directly
onsite.  X-ray fluorescence  is induced by
a  low intensity 244Cm  or  241Am gamma-ray
source   housed,   along  with    a   gas
proportional  detector, in  the  sampling
probe.  Operational safety  is maintained
by  a  shutter approved  by  the Nuclear
Regulatory Commission.

Analysis  with  the  X-Met  880  consists
simply  of placing  the probe  in  direct
contact  with  the  sampling  medium  and
opening  the   shutter  with  a  trigger.
Fluorescent  x-ray  photons  are  counted
over  a user-specified period of time by a
counting  circuit   and classified  into
discrete energy  levels by a multichannel
analyzer    to   produce    a   spectrum
characteristic of  the elements  in  the
sampling  medium.   Net  intensities  for
each  target  element  are  calculated  by
software    deconvolution   of    the
characteristic spectrum and converted to
concentration  values   by  means  of  a
calibration model.  This model is  derived
empirically    by   measuring   the   net
intensities  of the target  elements in a
set of calibration standards, and  fitting
a   linear  function  that   relates  net
intensity to  concentration  by a multiple
regression procedure.

As is the case with all XRF systems, the
relationship  between  net   intensity and
concentration    varies    with   the
characteristics of  the sample matrix.  In
the   case    of   solid,    inhomogeneous
particulate   media  such   as   soils  or
sludges,   the  concentration-intensity
relationship  is particularly  influenced
by  variability   in  the   grain  size
distribution,   bulk  density,  and  the
geometric relationships between discrete
grains  containing the target element(s)
and the detector.   The geometry  problem
is exacerbated by the very small volume
(roughly 0.04  cubic inch)  measured by the
probe.      Net   intensities  can   be
artificially   enhanced or   absorbed  by
certain  non-target elements that may be
present.  Moisture has also been reported
to affect  intensities(1) .   Data quality
can be significantly influenced by any or
all of  these matrix effects  which  must
therefore  be  taken into account  in the
calibration procedure, a subject that is
discussed  in  greater  detail  in  the  next
section.

ROUTINE FIELD PROCEDURES

Calibration

The   X-Met   880   has   no   fundamental
parameter capabilities which would allow
for standardless  calibration.   It  uses
calibration  curves  based  on  matrices
similar to those of the routine samples.
The instrument has 32 calibration models,
each  of  which  can  contain  up  to  6
calibration   curves.   Therefore,   each
calibration   model   can   simultaneously
quantify up to six analytes.  The number
of  calibration standards  required  for
each  calibration  model  depends on  the
number of analytes of interest; generally
eight to ten standards per analyte.

     Site Typical

A site typical calibration curve is based
on  samples similar in composition,  but
not necessarily matrix matched.   Extreme
caution should be exercised when using a
site-typical  calibration  curve.    The
authors  have encountered the situation
where  increased  iron  levels  in  mine
tailings  relative  to  the  calibration
standards  resulted in anomalously  high
chromium  results  (in excess  of several
wt% Cr!).   Corroboratory analyses found
chromium in the zero to 40 mg/kg range.

     Site Specific

To  minimize  enhancement/absorption  and
spectral interference errors, calibration
standards  should  be collected  from the
specific  site in  question.    These  Site
Specific Calibration (SSC) standards must
closely emulate the physical and chemical
matrix of  the routine  samples.   The SSC
standards  are prepared  as  loose  soils
(screened through 2 mm but unpulverized)
so  that  the  particle  size  bias  of the
routine  samples   is   included   in   the
instrument calibration.

Characterization  of  the SSC  standards
must  be done using  a  total  digestion
procedure   rather    than    a   partial
extraction  (i.e.,  CLP "Total Metals'"21),
because  XRF is a total analyte  method
regardless of phase or speciation.
                                           498

-------
     Co-calibration

If more than one X-Met is used on a site,
the  two   instruments   should  be   co-
calibrated with the  same  SSC standards.
Even  with   co-calibrated  instruments,
there can be inter-instrument bias.  This
should be tracked  and quantified by using
splits   of  the   same  check   samples
(discussed   below)    to   monitor   the
instruments.

The  authors  were involved  in  a  site
screening which was restrained to using 2
X-Mets that were not co-calibrated.   The
instruments  produced two  very different
populations  of data for  chromium,  an
order  of magnitude  apart.   The  higher
concentration   set   exceed  the   state
regulatory  levels.     Despite  the  low
chromium   values   in   the   laboratory
corroboratory  samples,  the  high  FPXRF
values are still an  issue with the state
health agency involved (approximately 1.5
years later).

Sampling

In situ analysis does not require that a
physical  sample  be  removed  from  the
ground.  The FPXRF probe is placed on the
ground and the analysis  mode  is activated
by pulling the trigger.  Acquisition time
can be preset  at  any desired length; 30
to 120 seconds is the most common range.
FPXRF   in   situ   analyses   are   very
beneficial during remediation.  FPXRF can
be employed in iterative passes following
contaminated  soil removal   efforts  and
quickly  produce  the  results  of  each
remediation attempt.

In situ analyses are the quickest way to
obtain soil chemistry data but these data
also   contain   the   highest  degree  of
variability  (error).   A large source of
error  is  the  extreme  heterogeneity of
most  soils.   The radioactive  source in
the probe  is exciting a very small cross
section of the soil  (20 mm diameter by 2
mm  depth) .   When  the probe is  moved
laterally  5  cm, the  detector is 'seeing'
a very different sample.  Another source
of error  is the extremely wide range of
particle  sizes  in  the  sample.    This
produces a known negative  bias(3) that can
be  somewhat  compensated   for by  using
loose,  unpulverized SSC  standards.   A
third   source  of   error   is   surface
microvariability which  is caused by such
physical   factors  as  wind  or  running
water,  vehicles  or  footsteps,  and  by
chemical  alteration  of  the surface.  An
excellent  example of  a rapid  chemical
change   in    surface    phenomena   was
encountered  by the  authors  at  a  mine
tailings site  in New  Mexico.    A white
precipitate of  ZnSO4  (?)  formed surface
crusts  in  the   late   morning   as  the
tailings   piles  dried   out.      Every
afternoon,   thunder  storms  washed  the
precipitate  back  into  the soil.    No
precipitate was apparent the next morning
until  the  sun  began  to dry the soil.
Different  daily spatial patterns could
radically alter concentration maps.

Surface microvariability can be mitigated
by   in  situ  homogenization    or _ by
collecting   intrusive   samples,   i.e.
samples that are physically removed from
the sampling media'41.

DATA QUALITY

The procedures in this  section address
only in situ analyses.  Intrusive samples
have several sources of variability that
do not  occur  with in situ sampling such
as collection, handling, and preparation
errors.

Quality Control Procedures

     Replicate Analyses

All FPXRF routine samples are analyzed in
triplicate and the means are the reported
values.  The three in situ measurements
are made in a 6 by 6  by 6 inch triangular
pattern  around  the   sample  location
marker.  After the third measurement of
every fifth sample location, the probe is
left  in place and analysis is repeated
two   more   times    (stationary   probe
triplicate).

Before the  first  in  situ sample location
and  after  every  tenth  sample  location
low-  and mid-calibration  range  quality
control  (QC)  check samples are analyzed
in  triplicate.   QC check  samples  are
loose soils in a 31 mm diameter  by 2.5 cm
cup,  approximately one half full.  Each
QC   check   sample    is   analyzed   in
triplicate, by removing  the cup from the
detector between  each analysis,  shaking
the soil in the cup,  then lightly tapping
the  cup  on   a   smooth  surface  before
replacing  it on the  detector.   Analyzing
these  samples  periodically  will warn the
technician   of  gain  change   or  other
instrument  problems.   The  authors have
found   this   particularly   helpful  in
detecting a low battery  before  the X-Met
software gives the "low battery" warning.
                                          499

-------
      Confirmatory Samples

 The  number  of  samples  for  laboratory
 confirmation  of  the  FPXRF  instrument
 results are based on the  overall  number
 of FPXRF samples points and the budget of
 the onsite coordinator.  One confirmatory
 in 40 routine samples  is adequate on  a
 site  with   approximately  300   sample
 locations.  Confirmatory sample frequency
 can  decrease   with  increased   sample
 locations.

 Quality Assurance Parameters

      Precision

 Different  levels  of  precision  can   be
 determined.        Minimum    instrument
 variability  in the field is measured from
 the stationary probe triplicate.   A more
 comprehensive  assessment  of precision is
 measured from the performance of  the  QC
 check samples.  Physical agitation of  the
 sample  between measurements yields some
 component  of the soil microvariability.
 The  QC  samples  are analyzed  over  the
 entire   time  span   of  analyses  and,
 consequently,  yield an   overall   FPXRF
 system  precision.

      Accuracy

 Accuracy is also determined from the low-
 and  mid-calibration  range  QC   check
 samples.     QC  values  are plotted   in
 control  charts  (one for  the  low  range
 sample   and   one  for  the  high   range
 samples)   with   concentration   on   the
 ordinate and successive measurements on
 the  abscissa.      Accuracy   and   the
 instrument  bias  from the "true"  values
 can  be  quickly  determined  from   the
 control  charts.   Table 1 compares  FPXRF
 accuracy and  precision  performance  to
 that  of  the  CLP Data Quality Objectives
 (DQOs)<2>.

      Detection Limits

 Detection  limits are  defined  as  three
 times the standard deviation (SD)  of the
 low-calibration  range QC  check sample.
 CLP contract required detection  limits*2*
 are compared to FPXRF detection limits in
 Table 2.

 Data Quality Objectives

Tables 1 and 2 give  some indication  of
levels  for  precision and  accuracy that
can be achieved  under conditions deemed
to  be typical  for  the  types of  waste
sites investigated.  These values should
be  viewed  with some  caution,  however,
recalling that the matrix-specific nature
of    the    intensity-concentration
relationship  dictates  that achievable QA
levels     are    also    site-specific.
Precision, accuracy, and detection limits
may  vary  significantly from matrix to
matrix,  in contrast with "wet chemical"
procedures   in  which  physical  matrix
effects  are eliminated by taking samples
into  solution  prior  to  analysis.    QA
levels   displayed   in   Tables  1  and  2
probably represent neither  the  best or
worst cases. Evaluating the applicability
of  FPXRF to meet  DQOs for  a given site
therefore  requires careful evaluation of
matrix   character  and  variability  when
using empirically  calibrated instruments
such as  the X-Met  880.

In  the  authors' experience,  there seems
to  be some reluctance to consider FPXRF
as  anything more than a screening tool.
Part  of this  problem  stems  from  the
failure  of site personnel to  consider the
end-use  of the  data  when  defining DQOs
and   evaluating   potential   analytical
methods.     Examples   come   from  site
managers who   have  asked   if  specific
detection  limits  can  be  achieved  for
certain  target elements when  those limits
are one  or more orders of magnitude lower
than potential action levels. Perhaps an
even  more  widespread  problem  concerns
spatial  applications.  When determination
of  the   spatial distribution of  target
elements constitutes  the data  end-use,
DQO definition almost invariably focusses
on  errors  relating  to  sampling  and
analysis and  not  on  errors  relating to
spatial  interpolation.   This problem is
so   significant    in   terms   of   the
application  of  FPXRF  that  it  warrants
some detailed consideration.

The inferential  link between samples and
the spatially distributed population they
are intended to  represent is established
through    the   process   of    spatial
interpolation.  This process consists of
estimating   concentration   values   at
unsampled points, usually located at the
nodes of a  regular grid, by  applying an
appropriate algorithm  to sample  values.
It results  in a spatial model  of target
element  concentration, and serves as the
basis of such graphical decision-making
tools as isometric diagrams  and  contour
plots.     The   reliability  of  decisions
based upon a  spatial  model  ultimately
depend on  estimation  errors  incurred at
the grid nodes.
                                           500

-------
 Geostatistical theory shows that spatial
 estimation  errors  consist  of  several
 components related to sample collection,
 preparation,  and  analysis,  and  of  a
 component related to spatial extension of
 the  sample  value   to  an   unsampled
 location'5'.     Extension   error   is   a
 function of  the  spatial variability  of
 concentration  and   of  the   distances
 between samples  and  grid  nodes, and  is
 usually much larger  than  sample-related
 errors1*'.  Thus,  the  largest  source  of
 error can be reduced most  effectively by
 increasing sampling density rather than
 by improving the  precision  and  accuracy
 of the  analytical method.   It  follows
 that   DQO   definition    for    spatial
 applications  should  focus  on   spatial
 estimation   errors   rather  than   on
 analytical errors.   It is possible  to
 meet  or  exceed   high  DQOs  with   an
 inexpensive  method such as FPXRF,  even
 though  its precision and accuracy may be
 less than  those  of  CLP  wet-chemical
 methods.   This is achieved  by employing
 higher  sampling densities,  at  less  cost,
 thereby reducing  the  largest source  of
 error  in   the   spatial    model   that
 represents the instrument  of decision.
 The  proviso,  of  course,   is that  the
 concentrations  of  interest  exceed  the
 detection  limit  of  the  FPXRF  for  the
 target   element  under  the  prevailing
 matrix  conditions.

 To   demonstrate   the   foregoing   points,
 Figure   3   shows  a   variogram   model
 describing  the  spatial variability  of
 lead concentrations  obtained with the
 FPXRF method in  contaminated  soil  at a
 Superfund  site.    Intersection   of the
 model at a  high  value  along  the  GAMMA
 axis  indicates high analytical error.  A
 second variogram in Figure  3 represents a
 model with no analytical error,   clearly
 an   ideal  but  unachievable  situation.
 Using these  models,  the spatial errors
 incurred  in  estimating the average lead
 concentration  of  a  50'  x  50'   x   1'
 remediation  block  were calculated for
 different densities of samples  taken on a
 square grid pattern within  the block (see
 Table 3).  Direct  analytical cost  of $150
 per  sample  was  assumed for  the  ideal
 analytical  case,   and  $13  per  sample
 location was assumed for the FPXRF  case.
The   latter   value  is  based   on   CLP
 laboratory  analysis  of 25  calibration
 samples  at   $150   per  sample,   and  a
sampling  campaign totaling  300  hundred
sample locations.   Applying these values
to  the  number   of  samples  for  each
sampling  density  results  in the total
 costs   shown  in   Table  3.     Spatial
 estimation error,  expressed as  relative
 standard deviation, are plotted against
 sampling costs for each case in Figure 4.
 These  plots show that FPXRF can achieve
 levels  of   spatial   estimation   error
 similar to those of the best  laboratory
 methods,  and that a given  level  of  error
 can be achieved by FPXRF at significantly
 less cost than laboratory  methods.  This
 means  that FPXRF should not be  restricted
 to  screening, but  can  also be  employed
 for  site  characterization and  remedial
 evaluation sampling in many situations.
 As before, these conclusions are based on
 the  assumption  that  concentrations   of
 interest are above the  FPXRF detection
 limits for the given situation.

 CONCLUSIONS

 Based  on results achieved thus  far,  FPXRF
 has   demonstrated  the   capability   of
 providing data necessary  for screening
 and    characterizing    many   inorganic
 contaminants,    both    rapidly    and
 inexpensively.  The authors believe that
 it  will  play an important  and perhaps
 central role in site remediation  in  times
 to  come.   Like   any  other   analytical
 method,  FPXRF has limitations  as well  as
 strengths.    Many   of  these  will   be
 improved    by    recent    and   future
 improvements  in technology.  Sensitivity
 and detection limits,  for instance, can
 now  be  materially  improved  with  high
 resolution detector systems designed for
 field  portable instruments.  Calibration
 constraints   can   be   relaxed   with
 incorporation  of fundamental  parameter
 techniques and better  software  systems.
 However,   several   points   need  to   be
 emphasized for  attainment  of  optimal
 results  when  field  sampling with the  X-
 Met 880 and similar instruments:

  f  Site  specific calibration standards
     are  absolutely necessary  to obtain
     defensible quantitative data.

  »•  Proper  sampling  protocols  must  be
     designed to allow quality assessment
     of the data.

  *  DQOs must be correctly defined.

  »•  Geostatistical    procedures    are
     essential for  proper  definition  of
     DQOs  and  for  QA   evaluation  of
     spatially distributed FPXRF data.

 If these points are  followed, the authors
believe  that  FPXRF can  meet  or exceed
                                          501

-------
 traditional  CLP  procedures  more  cost-
 effectively in  many situations.

 REFERENCES

 (1)   Glanzman,    R.K.    1990.    personal
      communication.     Senior
      Geohydrologist/Geochemist.      CH2M
      HILL, Denver,  CO.

 (2)   U.S.   EPA,   "Contractor   Laboratory
      Program   Statement   of   Work   for
      Inorganic  Analyses,  SOW  No.  788,"
      Attachment  A,  U.S.   Environmental
      Protection Agency, Washington, D.C.,
      1989.

 (3)   Wheeler,   B.D,  "Accuracy  in  X-ray
      Spectrochemical Analysis  as Related
      to    Sample     Preparation,"
      Spectroscopy 3., 1988,  24-33.
 (4)  U.S. EPA,  "Data  Quality  Objectives
      for  Remedial  Response Activities:
      Development   Process,"   EPA/540/G-
      87/004,         U.S.     Environmental
      Protection Agency, Washington, D.C.,
      1987, 154pp.

 (5)  Journal,  A.G.  and Huijbregts, C.J.,
      "Mining   Geostatistics,"   Academic
      Press,  Inc., New York,  1978.

 (6)  U.S.  EPA,   "Soil  Sampling  Quality
      Assurance  Guide,  Second  Edition,"
      EPA/600/8-89/046, U.S.  Environmental
      Protection Agency,  Las  Vegas, 1989
      pp.  26-27.

    NOTICE:  Although the research described  in this article has been
    supported by the United States Environmental Protection Agency through
    Contract No. 68-CO-0049 to Lockheed Engineering & Sciences Company, it
    has not been subjected to Agency review and therefore does not necessarily
    reflect the views of the Agency and no official endorsement should be
    inferred.
Table 1. Comparison of CLP DQOs to FPXRF Performance
for soil samples.
Analytical Method
CLP Spectroscopic (AA/ICP)
FPXRF
Accuracy
+/- 25 %
+/- 50 %
Precision
+/- 35 %
+/- 35 %
Table 2. Comparison of Detection
Limits for CLP and FPXRF.
(all values in mg/kg)
Analytes
Cu
Zn
As
Pb
CLP*
5
4
2
0.6
FPXRF
410 - 470
105 - 200
100 - 250
120 - 513
                           *  Assumed  soil weight of  1 g and  end volume of  200 mL.
                                         DISCUSSION
JIM PASMORE: I have a question on the data you showed for precision. What
was your measurement time on that, or the precision measurements, table 2.

BILL COLE: Thirty seconds.
STEPHEN KNOLLMEYER: I was just wondering why you didn't have any
detection limits for the cadmium? Didn't you take it?

BOB ENWALL: That was just a slight mistake on our part. We haven't really
had any experience with cadmium that would give us a number.
                                                 502

-------
Traditional Pathway for Sampling  and Analysis

                 PACKING & SHIPPING
    FIGURE 1. Traditional Approach to Site Characterization.
                            503

-------
Table 3 . Comparison of Costs for Estimating the Average
Concentration of a 50' x 50' x 1' Remediation
Block Using Different Sampling Densities.

SAMPLE DENSITY
(# SAMPLES)
1
4
9
16
25
36
49
ESTIMATION
ERROR
FPXRF
1.8298
0.8447
0.5510
0.4094
0.3268
0.2709
0.2318
ESTIMATION
ERROR
IDEAL ANAL.
0.9211
0.2977
0.1604
0.1067
0.0825
0.0629
0.0522
COST
$
FPXRF
13
52
117
208
325
468
637
COST
$
IDEAL ANAL.
150
600
1350
2400
3750
5400
7350
                             Screening Data
                              to Computer
                                                       Develop Semivariogram
                                                          from Geostatistics
                  Develop Kriging
               Concentration Isopleth
                Maps,  Strategy and
                Sampling Plans from
                   Geostatistics
3-

2—
                                                      10  20  30  tti  tO  60 70

                                                           Separation Oixtanc*
FIGURE 2.   FPXRF  Approach Employing  In Situ Measurements.
                                   504

-------
   6.0-
   4&-
   4.0-
   &6-
   9.0-
  H
   2.0-
   0.6-
   0.0-
.A.
                         A     A
                 ZT
       I  I I I I I  I I I [  I I I—I  | I I I  I |—I I  I I | I  I I I |
            10     20     80    40    SO     M
                                          i I  i i i i  I i i i  i I i i  i i I i  i i i
                 70     M
 I ' ' '  ' I
80     100
                                DISTANCE (to*)
              aouo LINE: mar MODEL  DASHED UNE OEM. ANALYTICAL MODEL
 RGURE 3.  RELATIVE VAR10GRAM8 FOR FPXRF AND IDEAL ANALYTICAL CASES.
  2.0-f
  1.5-
  ; 14)-
  0.6-
  ao-i
                     1 ' ' M
                         HX>
                                             woo
                              COST (DOLLARS)
            SOUOUNE FPXRF CA9E     DASHED UMB DEAL ANAIYIICAL CASE

RQURE 4. ESTIMATION OF A 50' X 50' X1' REMEDIATION BLOCK. ESTIMATION
          ERROR VS. COST FOR FPXRF AND IDEAL ANALYTICAL CASES.
                                   505

-------
                 A HIGH RESOLUTION PORTABLE XRF HgI2 SPECTROMETER FOR FIELD

                               SCREENING OF HAZARDOUS WASTES.
           J. B.  Ashe
           Ashe Analytics
           Austin,
           Texas   78746

           P. F.  Berry and G.  R.  Voots
           TN Technologies,  Inc.
           Round  Rock,  Texas  78664
                M. Bernick
                Roy F. Weston,  Inc.
                Edison, NJ  08837


                G.  Prince
                U.S.E.P.A.
                Environmental
                Response  Team
                Edison, NJ  08837
 ABSTRACT

 A field portable XRF spectrometer based on
 a mercuric Iodide (Hgl,)  semiconductor x-ray
 detector is described.   Its multi-element
 capabilities  will  be illustrated with mea-
 surements  on  chemically-analyzed  samples
 representing materials collected from sever-
 al  hazardous waste  sites containing differ-
 ent metallic pollutants in a variety of soil
 matrices.

 The range of the analyzer extends from Ca to
 U,  and  a typical configuration provides for
 about  20 elements  which  are simultaneously
 reported together with the analytical preci-
 sion.   Minimum  detection  limits  for  most
 elements are  in  the range  of 50 -200 mg/kg
 with a  200 second analysis time.

 The solid  state detector is  operated  near
 ambient  temperature and affords  an  energy
 resolution of better than 300 eV for the Mn
 K  x-rays.    Intrinsic detector  efficiency
 exceeds  60% for  energies  up  to  100  keV.
 Dual  radioisotopic  source  excitation  is
 provided from the list:  Fe-55, Cd-109,  Cm-
 244, Am-241.  A 6.5 kg battery-powered  Data
 Processing Unit features menu-driven opera-
 tion  and on-board  dual   2000  channel   MCA
 spectrum  display   capability.     Internal
 storage  provides  for  the  retention of  30
 spectra  and   100  multi-element  analytical
 reports.

A  "fundamental  parameters"  based  analysis
 algorithm  is  used  to  compute  elemental
concentrations.    This  computational   ap-
proach,  together  with  the  comprehensive
element  coverage,   permits   "standardless"
measurements over a wide range  of material
compositions.
INTRODUCTION

Soil  contamination by  hazardous  metallic
waste is present at the level  of concern on
more than 50% of the  sites  on the National
Priorities List. Complete evaluation of the
degree  of hazard  and  measurement of  the
spatial extent of those  hazards involves the
analysis of literally millions  of samples.
Although  procedures  have been  approved  by
the Environmental  Protection Agency for use
under  the  Contract Laboratory  Program  for
performing enforcement  quality  analyses  on
environmental  samples,  a  need still  exists
for  rapid,  reliable,  and  cost-effective
assays to expedite the characterization and
remediation of  sites.   Ideally  such  assays
could be accomplished by direct measurement
in the field.

Several investigators  (1-15)  have reported
on  the application  of energy  dispersive
x-ray  fluorescence (XRF)  to  the  assay  of
metal-contaminated soil. Savings in time and
analysis cost  over  the standard EPA-approved
chemical methods are  significant  (1,5,11).
Essentially   a   nondestructive   technique
suited to the  measurement  of almost any kind
of material in powder, liquid  or solid form,
XRF is further distinguished by its ability
to analyze for many elements,  including the
unexpected, in a truly simultaneous fashion.
Thus it offers low  cost  and rapid turnaround
time per analysis.

By means of XRF, two environmental applica-
tion  methodologies -   in  situ  assay  with
field  portable  instruments  and  intrusive
sample analysis  with laboratory grade equip-
ment - are currently  being  pursued and are
reported  to  yield good definition of  the
magnitude and extent of contamination -
                                               507

-------
particularly when coupled with  geostatisti-
cal   sampling  methods  and  data   analysis
(12,14).  Laboratory-grade instruments gener-
ally offer better precision, higher accuracy
and greater  sensitivity than portable equip-
ment since,  with no  weight  and power  re-
strictions,   x-ray  tube  excitation  (as  op-
posed to  radioisotopes)  and high resolution
spectrometers  of the cryogenically-cooled
semiconductor  Si(Li)  detector  type  can  be
used. Portable  instrument designs  have  so
far employed gas proportional detectors  and
their capability for  multielement applica-
tion  has been  somewhat  impaired  by  the
limited  energy  resolution of  that  type  of
detector. The  in-situ measurement capabili-
ty of a portable instrument,  albeit at some
trade-off in analytical  performance,  is  of
advantage in  reducing the time delays  and
data-integrity risks associated with sample
handling procedures. Other benefits are  its
low cost  per analysis, its  utility in delin-
eating hot spots as an aid  in the collection
of samples for enforcement quality assay or
to guide the work of site  remediation.
Quantitative XRF application  always requires
an   appropriate   calibration,   usually   by
measurement  on  a  representative  suite  of
chemically  known standards.  Alternatively,
as  is  now  the option on  most laboratory-
grade instruments using  Fundamental Parame-
ter  (FP) methods  of  analysis,  only pure
elements or a few standards ( which  need not
be  of site-specific  composition)  are  re-
quired. Currently available portable instru-
ments with their lower x-ray  resolving power
are  more restricted to the use  of an empiri-
cally   structured   analysis   algorithm  of
limited  element  coverage. They, therefore,
require  a multi-sample calibration  on site-
specific material  (11).  This  is  a major
drawback to  their  general  use,   and  the
quality  of  the resultant analyses  is  opera-
tor  sensitive and highly  dependent on  the
validity of the  calibration  samples.

The  development of  a high-resolution non-
cryogenic semiconductor x-ray  detector has
made possible a new  field-portable  instru-
ment design which can provide for the appli-
cation of FP analysis of soils  using a site-
 independent calibration based  on  pure  ele-
ments standards. Our paper will present  some
of the  results  obtained with   this  instru-
ment.

 INSTRUMENTATION

The instrument used to evaluate the applica-
 tion of  an FP-based XRF method  for field
 analysis of metal-contaminated soil is shown
 in  Figure  1.  Similar  instruments  are  now
 used industry-wide for  on-site verification
 of alloy materials  (16). The system operates
                          CD CD COB
                 Figure 1
      Field-Portable XRF Spectrometer

off  either  AC-power  or rechargeable  NiCd
batteries and weighs  approx.  17 Ib  (8kg).
Compared to the alloy-analysis  design,  the
soil application unit was modified  only  in
regard to the  type of  isotopic  excitation
sources  contained  in  the hand-held  probe,
and minor revisions to the PROM-based oper-
ating software  in the data processor module.
The  sample  measurements  we  report  were
performed with  an isotope combination of Cd-
109  and  Am-241;  each  of an  effective  3mCi
source activity. The sources are separately
shielded  in a  motorized  turret  and  are
positioned  for measurement  under  program
control. An on-screen  set-up menu allows the
individual  source  exposure  time to  be se-
lected  from 1  to   999  sees.  All  reported
measurements used  a 200 sec. selection.

Probe accessories such as a detachable base,
shown  in  the figure,  and clip-on front-end
attachments, facilitate the  measurement of
contained samples,  but  the  main utility of
the  probe is afforded by its compact, hand-
size,  design for direct application to in-
situ material.  Measurement  is  initiated by
momentary push-button action, either at the
probe  or on the instrument  panel,  so free-
standing long period assay and,  as required,
operation within  an environmentally sealed
plastic  enclosure  are  quite  practical.  A
tough  replaceable x-ray entrance window also
seals  the  probe   face  over a  measurement
aperture of 0.5 x 0.75 inch  (1.25 x  2.0 cm).
                                                 508

-------
 Spectrometer-Analysis  Operations

 The   x-ray   analytical   capability   of  the
 instrument  is  established mainly by  its use
 of a  new-technology, high resolution energy
 dispersive  x-ray detection device based  on
 semiconductor  HgI2.  This detector   is con-
 tained  in  a capsule within  the probe and
 operates  at a  controlled, less-than-ambient
 temperature by   low-power  thermo-electric
 cooling.  The energy resolution, expressed  as
 FWHM, is  of the order of 300  eV  for the  Mn
 K x-ray line.

 The  data  processor  performs  all   of  the
 necessary analog/digital  electronic func-
 tions to translate the  detected x-ray infor-
 mation into quantitative analytical results.
 Pulse-amplitude   records,   for  instance,
 representing the sample x-ray  fluorescence
 spectra are generated and  stored for each
 excitation  source in a 2x2000-channel memo-
 ry. These spectra,  if  desired, can  be pre-
 sented on the  instrument's LCD  panel accom-
 panied with  the  usual control   features of a
 multichannel   analyzer.  Peak   identifiers,
 regions of  interest, and element-line mark-
 ers, for example, are operative in that mode
 and are displayed in  calibrated x-ray energy
 units. Non-volatile RAM storage allows up to
 30 spectra  to  be retained together with the
 analytical  results for more than 100 field
 measurements.  An  RS-232  serial  port  is
 provided for printer  and  computer communica-
 tion.

All operations  are  prompted   by  on-screen
menus which  indicate the available  options
and how to proceed. An example  of the "turn-
on" menu  is  as follows:

            MAIN MENU
            Enter Choice of Operation
             1 SOILS ANALYSIS
            2 RECALL STORED RESULTS
            3 REVIEW/CHANGE SET-UP
            4 STORE/RECALL SPECTRUM
            5 OTHER FUNCTIONS

Routine operation proceeds by  option  #1,
which leads  to  a  set  up of the data acquisi-
tion times per source and initialization of
the probe  controls for measurement. Measure-
ment concludes with  an  audible signal  fol-
lowed by an  on-screen report of  the analyzed
elements.  Results  are  labelled by  element
symbol and,  as  later described,  include both
element concentrations and an  indication of
the computed standard  deviations. Options,
such as the  storage and printout of results,
follow the on-screen report if pre-selected
from  the  Main  Menu. Spectrum  operations,
available as shown,  are  normally by-passed
 in the  routine measurement sequence which
returns directly to a "ready"  status after
the analysis report.

Element  concentrations are  computed using
Fundamental  Parameter derived coefficients
in an algorithm of the  form:

      CONC = R x S x  (1 + SUM{anxCn})

Where,  "R"  is  the  measured  analyte x-ray
intensity relative to the pure element: "S"
is a calculated sensitivity coefficient: and
the quantity  SUM{)  is  a  summation of "n"-
element  absorption-enhancement  terms con-
taining  calculated  alpha-coefficients (17)
and iteratively computed element concentra-
tions. Preparation of the  instrument for the
measurements reported in  the next  sections
entailed  only  a normalization  to  the pure
element  response. No  other calibration was
performed. X-ray  intensities are  processed
for more than  20  elements  but  only those
determined to  be in  excess  of three-times
the standard deviation  are presented  in the
analysis report. All  element x-ray intensi-
ties, however, can be viewed on the screen.
SAMPLE MEASUREMENTS

Although the instrument is  capable of per-
forming in-situ measurements, the effective-
ness of an FP-analytical approach was evalu-
ated by analyzing intrusive samples so that
comparative  analyses  could  be  obtained.  A
total of 55 samples were measured represent-
ing material  from four NPL sites character-
ized by complex metallic contaminations. The
samples were air dried, disagglomerated but
not ground,  and passed  through a 20 mesh
screen  (.84  mm hole).  The  oversize  was
discarded.  The undersize was split so that
replicate  samples  were  available  for  the
chemical  and  x-ray assays.   The  powdered
samples, ranging  in  mass  from 2 to 8 gms,
were placed in 30mm sample cups and covered
with a 0.005 mm  polypropylene x-ray window
for measurement in the upright probe geome-
try.  Sample  thicknesses ranged from 8.5 to
16 mm and bulk densities  of  the loose pow-
ders ranged  from 0.26  to  1.2 gm/cc.   The
chemical assays  of  the sample  splits were
performed by  a commercial laboratory (18)
using flame Atomic Absorption (AA)  analysis
for the elements of interest.

Two of the sampled sites  (noted  as Sites  1
and 3  in the  results)  are  inactive  metal
plating locations. The samples were collect-
ed from  settling  lagoons and consisted  of
plating sludges mixed with the local soils.
Site 2 had been the location of a  smelter;
a metal  working facility;  and, in recent
time,  a Ni-Cd  battery manufacturing opera-
tion.   It is  an estuary location, submerged
except during low tide, and the soil would
                                              509

-------
 be best described  as  contaminated  bay sedi-
 ment.    Site  4 was a scrap metal  storage-
 segregation   facility,  involving  a   wide
 variety of metals  and metallic  compounds  in
 open  areas throughout the  site.

 An example of the instrument-generated pulse
 height  spectrum (excited by Cm-244) for one
 of the  soil samples from a  plating  lagoon  is
 shown  in Figure 2,  overlaid with a spectrum
 for the same sample obtained on  a gas-filled
 proportional-detector instrument.  The supe-
 rior  x-ray  resolving   power  of  the  HgI2
 detector is  obvious and is  seen to provide
 well  for  quantitative  analysis   of   minor
 elements in the presence of adjacent atomic-
 number  elements at high concentration.

o
rr
x
                 6     8   10   12    14
                 X-RAY ENERGY (keV>
16
 Figure 2.   Example  XRF Spectra for  a  HgI2
 and a Gas Proportional  Detector on  the  Same
 Soil  Sample (Cm-244  Excitation).

 The instrument-reported XRF assay results of
 a Cd-109/Am-241  excitation  measurement  on
 the  soil  sample  are given  in  Figure  3.
 Where available,  the results of prior  analy-
 sis by Atomic Absorption  are  tabulated.
DATE:
TIME:
MODE:
07/10/90
10:15:55
STANDARD


COMPOSITION:


C;i
Cr
Fe
Co
Ml
Cu
Zn
Cd
Sn
f'b
CONC.
(PCNT)
5.035
0.513
8.453
.118
1.570
1.661
11.865
.683
.453
.047
ST. DEV.
(PCNT)
0.705
.055
.064
.024
.020
.015
.053
.005
.005
.002
X-RAT
(mg/kg>
50,035
5,130
84,453
1,180
15,700
16,610
118,865
6,830
4,530
470
ATOMIC ABSORP.
(mg/kg)
*
6,600
*
*
16,000
*
110,000
7,900
*
430
 The format of the XRF results  shown  in  the
 left-hand columns  of Figure 3 is that of the
 resident operating software  which  reports
 both  the element  concentrations  and  the
 computed statistical  uncertainties in units
 of weight-per-cent-element.   For comparison
 with the AA  mg/kg  assays in the figure,  and
 for the results discussed in the  next  sec-
 tion,  a  factor  of 104  has been applied  to
 the x-ray  data.    In  this  single-sample
 comparison,  the XRF and  AA results are  seen
 to be  in good  agreement  for the  elements
 measured by  both  techniques.   However,  a
 number of contaminants  not known to be  at
 the site were  detected  at a statistically
 significant  level  by  XRF.

 COMPARATIVE  RESULTS

 Graphical  comparisons  of the  XRF results
 with  AA analyses  for  the  55  split samples
 from the  four sites are presented in Figures
 4-9.   Because  of  suspected   intra-sample
 heterogeneity,  a statistical  comparison   of
 the results   of  the  two measurements on  a
 simple point-by-point basis was  not consid-
 ered appropriate.  Under such  circumstances,
 global  data comparisons  are more valid. Two
 global  methods  were used:   The XRF results
 were  regressed  against  the   AA  results  to
 reveal  any relative biases in  the data sets;
 and the relative  percent  difference (RPD)
 between  the  two determinations  were  calcu-
 lated  to  indicate the  average  disparity
 between  the  analyses. Any disparity would
 reflect  the  imprecision  of each determina-
 tion and real chemical  differences  in the
 sample  splits.  Where the measurements span
 a  range  of values  well  above  the detection
 threshold, the average RPD should be  a good
 indicator of the intra-sample  heterogeneity
 uninfluenced by the analytical precision of
 either  technique.   The  average  RPD's from
 samples  with  contamination levels  greater
 than 10 times the standard deviation were
computed  for  each  analyte  and are  shown in
Table  1  along with the main results of the
regression analysis:

 Table 1 Some XRF vs.  AA Correlation  Data
                               * = Not Analyzed

Figure  3.   X-Ray  Analysis Report  and AA
Assay Data for the  Soil Sample of Figure 2.
ANALYTE

Cr
Fe
Ni
Zn
Cd*
Cd"
Pb

0
0
0
0
0

0
RANGE
(mg/kg)
- 28,000
- 350,000
- 20,000
- 150,000
- 10,000

- 580,000
REGRESSION DATA
SLOPE
1.02
1.36
1.11
0.99
0.84
1.10
1.06
a sip.
.09
.05
.05
.02
.07
.06
.09
K2
0.84
0.98
0.96
0.9S
0.90
0.95
0.85
RELATIVE
% OIFF (RPO)
38
23
30
23
33
17
30
                 Cd uncorrected for bulk density
                 Cd corrected for bulk density (see text)
                                                 510

-------
XR
30000
24000
18000
12000
RfYW
0
F mg/kg CHROMIUM




"*u #
•
ff
*


D
D/"


D
7



/



/


n
Sltel
Site 3
Site 4


0 6000 12000 18000 24000 30000
Cr, Atomic Absorption, mg/kg
XRI
20000
15000
10000
lyyyv
Q.
F mg/kg NICKEL



•D /
"/*

D
/


/


/



D
SHe1
B
Site 2
•
Site 4



0 5000 10000 15000 20000
Ni, Atomic Absorption, mg/kg
               Figure 4
Comparative Assay Results for Chromium
              Figure 6
Comparative Assay Results for Nickel
XRF
500000
400000
200000
100000
QJ
mg/kg IRON




/







/


m
/





/


Site 2
Site 4





0 100000 200000 300000 400000 50OOOC
Fe, Atomic Absorption, mg/kg
XRF
80000
60000
40000
OfWYV
Q|
mg/kg ZINC



a ^
£

a
ef'
<'a
a
a

/'


/
' u '


a
Sltel
•
Site 2
Site 4


0 20000 40000 60000 80000
Zn, Atomic Absorption, mg/kg
               Figure 5
  Comparative Assay Results for Iron
              Figure 1
 Comparative Assay Results for Zinc
                                          511

-------
XRF mg/kg CADMIUM
15000
12000
9000
6000
3000-
0
<



Dens. Corrected
Uncorrected



. . •
J^


/
"
a QD



•
X




] a


/





3000 6000 9000 12000 15000
Cd, Atomic Absorption, mg/kg
XRF
600000
450000
300000
150000'
0
(
mg/kg LEAD


•
...••'
..-•'
v
I


/
/


/
.-'


/
/

n
Srtel
Site 2
Site 4




) 150000 300000 450000 600000
Pb, Atomic Absorption, mg/kg
               Figure 8
 Comparative Assay  Results  for Cadmium
(Illustrating Bulk  Density  Corrections)
                 Figure 9b
    Comparative Assay  Results for Lead
               (High Range)
XF
800
600
400
200-
F mg/kg LEAD


K W
:
_

/
n /
[-3 "
D i
•

n
j ..'-''
	 	 __



a
Sitel
Site 2
Site 4

—

0 200 400 600 800
Pb, Atomic Absorption, mg/kg
               Figure 9a
  Comparative Assay  Results  for  Lead
              (Low Range)
Concerning the density correction on the  Cd
assays (noted in Table .1),  it  is generally
recognized that for XRF analysis of loosely
packed powder material there is always some
influence of bulk density on the measured  x-
ray intensity.  The magnitude of the effect
depends on the excitation and fluorescent  x-
ray energies  and  was  investigated  by mea-
surement  on  a specially prepared  suite  of
"doped" samples covering a density range  of
0.2 to 2.0 gm/cc.   A functional dependence
of the x-ray intensity, relative to a "high
density"  sample, was derived to  be  of the
form;
             (mu*rho)/(mu*rho+l)

 where "rho"  is the bulk density and "mu"  is
 a calculated absorption coefficient for  the
 analyte x-rays.  The value of "mu" decreases
 with  increasing  x-ray energy  and,   in  low
 bulk density material, becomes  a factor  in
 the XRF analysis of elements such as Cd,  Sn,
 Sb,  etc.,  excited  by  the relatively high
 energy emission of Am-241.  Over the density
 range of the measured  samples  (0.26  to  1.2
 gm/cc) the  values of the  calculated x-ray
 intensity correction  were in  the  range  of
 1.2 to 1.9.   The  results are illustrated in
 the data of Figure 8.

 The  XRF  precision data reported alongside
 the  assay  values   (as  previously noted  in
 Table 1)  are based on the calculated statis-
 tical errors associated with each analyte x-
                                           512

-------
ray  intensity  and include  the  compounding
effect  of the  iterative  solution  of  the
multielement analysis algorithm.  The data,
therefore, can  convey a realistic indication
of  analytical  detection limit,  consistent
with all  of  the  conditions  of measurement.
An evaluation of  the  precision was conducted
by repeated independent  analysis on the same
undisturbed  sample over  a  three-day  period
and the results are  shown in Table 2.

  Table 2    Precision Evaluation Data
ANALYTE


Fe
Ni
Cu
Zn
Cd
Sn
Pb
Bi
ANALYTE
CONCENTRATION

-------
expected  that similar accuracy  and precision
values would  be achieved for other analytes
in  the  same part  of  the  periodic  table  as
those reported here.

The  MDL's  reported  in  Table  3 were  deter-
mined under the available but less-than-opt-
imum instrumental  parameters.   Count rates
with the  sources employed were  substantially
lower than  the  practical  limit of the  sys-
tem. The detector head,  shown  in Figure  1,
is   designed  for  optimum  presentation  of
small metallic samples  commonly encountered
in   alloy-sorting   applications.   A   sample
presentation  geometry  of  more  appropriate
design for soil assay should yield a  signif-
icant  improvement  in the measurement effi-
ciency.   Most  data  were acquired  with  the
source  pair of Cd-109 and  Am-241. Chromium,
and  perhaps  other analytes,  could   be  more
sensitively  assayed   with  other   isotopic
sources. Improvements in these areas of the
instrument  design are expected to lower the
MDL of  all  analytes  by at  least a factor of
two, and should lower the MDL  of  Cr  into the
150  -  200 mg/kg  range.

CONCLUSIONS

The results obtained  from this study clearly
show the potential of a Fundamental  Parame-
ter approach  to the analysis of soil  contam-
ination with  a  high-resolution energy-dispe-
rsive XRF analyzer of field portable design.
The operational convenience  of a calibration
based  only on measurement  of  pure  element
standards  is well  demonstrated. Considering
the diversity of soil types  tested  and  their
wide  variation  in the  level   of contamina-
tion,   the  overall   accuracy   is  good  and
certainly adequate for  screening  tests  where
operational  requirements  are  generally set
at  +/-  50%  accuracy  and  +/-   1°% precision
(20).     The  results  reported here  easily
satisfy  those  requirements.

FOOTNOTES  and  REFERENCES

1.  Furst, C.A.; Spittler, T. and Titlinghast, V.  "Screening
    for Metals at Hazardous Waste Sites:   A Rapid Cost-Effec-
    tive  Technique  Using X-Ray Fluorescence"   Proc.of 6th
    National  Conference  on  Management of  Uncontrolled
    Hazardous  Waste  Sites.  Hazardous  Materials  Control
    Research Institute,  Silver Springs,  MO.  (1985)

2.  Grupp,  D.J.;  Everitt, D.A.; Bath,  R.J. and  Spear, R.
    "Use  of a Transportable XRF  Spectrometer  for  On-site
    Analysis of Hg in Soils" American Environmental Labora-
    tories, pp 32-40, Nov, 1989

3.  Grupp,  D.;  Fendler.K.; Mathers, V.;  Bath, R. and Spear,
    R. "On-Site Multi-Element  Analysis  of  Hazardous Waste
    Site Soil  Samples Using X-Ray Fluorescence"  Conf. Proc.
    HAZTECH 90, Houston, TX. pp 8B-887 - 897 (1990)

4.  Jacobus, N.  "Screening of Hazardous  Waste With an Energy
    Dispersive X-Ray Fluorescence Spectrometer"  Advances in
    X-Ray Technology, Vol. 33,  p655-663.  C.S. Barrett et al,
    ed.  Plenum Press, New York (1990)
5.   Meiri.D.; Bradfield, D.G.  and Downs, D.M.  "Delineation
    of  Heavy  Metals  in  Surface  Soil  by Portable  X-ray
    Fluorescence   Analysis  Screening"   Fourth  National
    Outdoor Action Conference on Aquifer Restoration, Ground
    Water Monitoring  and Geophysical Methods, pp 1067-1079.
    Las Vegas (1990)

6.   Mernitz,  S.  and Staible,  T.  "Use of  a  Portable X-Ray
    Analyzer  and  Geostatistical  Methods  to  Detect  and
    Evaluate Hazardous Metals in Mine/Milt Tailings" Proc.of
    6th  National  Conference  on Management of Uncontrolled
    Hazardous  Waste   Sites.   Hazardous Materials  Control
    Research  Institute, Silver Springs,  MD. (1985)

7.   Perils, R.  and Chapin, M.   "Lou  Level  XRF Screening
    Analysis"    Fifth International  Symposium on  Field
    Screening  Methods for Hazardous Waste Site Investiga-
    tions, pp 81-94.  (1988)

8.   Piorek, S. and Rhodes, J.R. "Hazardous Waste Screening
    Using  a  Portable X-Ray  Analyzer"  Symposium  on Waste
    Minimization and Environmental  Programs  within D.O.D.,
    Long Beach, California, April 1987

9.   Piorek, S.  "XRF  Technique as  a Method of Choice  for
    On-site Analysis  of Soil Contaminants and Waste Materi-
    al"  Advances  in X-Ray Technology, Vol. 33, p629-637.
    C.S. Barrett et al,  ed.   Plenum Press, New York (1990)

10. Raab, G.A.;  Faber, H.L. and Simon,  S.J.  "Evaluation of
    a Prototype Field-Portable X-Ray Fluorescence System for
    Hazardous Waste  Screening"   EPA/600/4-87/021.   U.S.
    Environmental Protection Agency, Las Vegas, Nevada.(1987)

11. Raab, G.A.;  Kuharic, C.A.; Cole  III, W.H.; Enuall, R.E.
    and   Duggan,  J.S.  "The  Use  of  Field-Portable X-Ray
    Fluorescence Technology in the Hazardous  Waste  Industry"
    Advances in X-Ray Technology,  Vol. 33,  p629-637. C.S.
    Barrett et al, ed. Plenum Press, Neu York (1990)

12. Raab, G.A.; EnwalL, R.E.; Cole III,  W.H.; Faber, M.L. and
    Eccles,  L.A.   "X-Ray Fluorescence  Field Method  for
    Screening of  Inorganic Contaminants at  Hazardous Waste
    Sites" Hazardous Waste Measurements, M.Simmons, ed. Lewis
    Publishers Chelsea, MI (in press,1990)

13. Sackman, A.R.; Perlis, R.  and Chapman, M. "Applications
    of X-Ray Fluorescence  Spectroscopy for Site Screening"

14. Walsh, J.P.; Harding,  T and Aulenbach, S.  "Using X-Ray
    Fluorescence Spectrometry and Geostatistics for Soil-Me-
    tal  Contamination"  Proceedings of Conference on Plan-
    ning, Rehabilitation and Treatment of Disturbed Lands, pp
    307-317.  Billings, MT (1990)

15. Watson, W.;  Walsh, J.P. and  Glynn, B.   "On-site X-ray
    Fluorescence Spectrometry Mapping of Metal Contaminants
    in Soils at Superfund Sites"   American Laboratory, pp
    60-68. July,  1989.

16. Berry,  P.F. and  Voots,  G.R.  "On-site Verification of
    Alloy Materials  with a Neu Field-Portable XRF  Analyzer
    Based on a  High-Resolution  HgI2  Semiconductor X-ray
    Detector" Proc.  12th   World  Conf. on  Non-Destructive
    Testing (Amsterdam). Vol.1. p737. 1989. Elsevier Science
    Pub.

17. de Jongh  W.  K.  "X-ray Fluorescence  Analysis  Applying
    Theoretical  Matrix Corrections" X-ray Spectrometry 2.
    p151,  1973

18. Atomic  Absorption analyses were performed by  Jayanti
    Patel and Pradeep Gupta of Roy E. Weston's REAC Laborato-
    ry.

19. American Chemical Society Committee  on  Environmental
    Quality   "Principals  of Environmental Analysis." Anal.
    Chem 55 (1983)

20. Raab,  G.A.;  Faber, M.L. and Simon,  S.J.  "Development of
    a Field-Portable  X-Ray Ftuorsescence System for On-Site
    Hazardous Waste  Screening" Lockheed Engineering  and
    Management Services Co, Report  for Contract No. 68-03-3-
    249.
                                                      514

-------
                                                           DISCUSSION


JOHN MORRIS: The fundamental parameters technique assumes that it's      PETER BERRY: The value of the fundamental parameters is  only really
dealing with all the components of the system. This question is twofold. Do you      achieved if you measure all of the elements. And so naturally we do expose all
routinely switch over radionuclide sources to span spectrum for all samples to do      of the sources to obtain the data to apply that model. The light elements are not
this? And do you make assumptions about the light component, or do you      practically measured in the field, so everything is expressed relative to a lighter
calculate that with the Raley-Compton ratio?                                   element like silica. Elements like calcium can be determined and the coefficients
                                                                         derived, but we cannot measure silica and quantify that. So that one is assumed
                                                                         to be the balance of the material.
                                                                    515

-------
                            LOW CONCENTRATION SOIL CONTAMINANT

                            CHARACTERIZATION USING EDXRF ANALYSIS
                                         A.R. Harding
                                    Spectrace Instruments, Inc.
                                      345 E. Middlefield Rd.
                                    Mountain View, CA 94043
INTRODUCTION

Effective assessment and remediation of
hazardous waste sites dictates that
analytical methodologies be developed which
assist in the evaluation of site
contamination and simultaneously make
efficient use of sampling time and resources
(1). Optimally, a technique would provide
on-site personnel with immediate and
accurate information concerning the identity
and concentration of inorganic soil
contaminants (2).

Inorganic pollutants can be readily
determined in contaminated soils with energy
dispersive X-ray fluorescence spectrometry
(EDXRF) using a thermoelectrically cooled
Si(Li) detector (3).  A field mobile
laboratory van or trailer can accommodate
the EDXRF system because the electrically
cooled detector, which provides high
resolution EDXRF spectra, does not require
cryogenic cooling.  Soil sample preparation
for EDXRF analysis is minimal, therefore,
short turnaround times are realized between
sampling and reporting results.

This report will describe an EDXRF method
developed to determine four inorganic soil
contaminants:  lead, arsenic, zinc, and
cadmium at four sampling depths,  the EDXRF
results for approximately one hundred eighty
soil samples will be compared to results
obtained for sample splits submitted for
analysis at an independent laboratory.
Evaluation of low concentration arsenic
detectability with elevated lead
concentrations in these samples will be
discussed.  Accuracy and precision of the
EDXRF method will  also be compared to the
independent methods using a standard
reference material and soil samples
submitted in triplicate to both
laboratories.

EXPERIMENTAL

The field mobile EDXRF spectrometer used in
this work was a Spectrace 6000 (Spectrace
Instruments, Inc., Mountain View, CA).  The
EDXRF system consists of three modules: the
spectrometer, the control/pulse processing
electronics, and the data analysis computer.
The compact size and weight (90 Ibs.) of the
modules permits installation of the system
in a laboratory trailer or van.

The bench top spectrometer module, which can
accommodate a single soil sample, is powered
by 110 V line or generator feed. The
excitation source used is a low powered Rh
anode X-ray tube (50 kV, 0.35 mA (17 W)
maximum output) positioned at a 45° incident
angle to the sample.  Three primary
radiation filters permit optimum spectral
acquisition conditions to be computer
selected.

The thermoelectrically cooled Si(Li) X-ray
detector is mounted at a 45° take-off angle
in an inverted geometry with respect to the
sample.  The 20 mm2 Si(Li) crystal, which is
protected by a 0.5 mil Be window, is cooled
to -90°C for operation using a multi-stage
thermoelectric (Peltier effect) cooler.  The
300 watts produced at the detector heat sink
are dissipated by forced ambient air.
Thermoelectrically cooled detectors provide
typical resolutions of 185 eV (Mn Ko).
                                               517

-------
A card cage module is interfaced between the
spectrometer and an IBM PS/2 or PC/AT series
computer.  The card cage components include
the  detector high voltage supply, the pulse
processing electronics, and the control
circuit board for the EDXRF spectrometer.
The data analysis software executed on the
PC is capable of either a fundamental
parameters or empirical data treatment
scheme using a combination of standard
reference materials and/or site specific
standards.

Sampling of the suspected waste site was
performed using EPA approved protocols in a
9500'x 3500* rectangular area.  Forty three
(43) cores were collected and partitioned
into four depth levels: surface to 2"; 2" to
6"; 6" to 12"; and 12" to 18", and
designated levels 1 through 4, respectively.
At the site, samples were first homogenized
and then split into two fractions.  One was
submitted for EDXRF analysis and the other
sent to an independent lab for analysis.

The independent laboratory used EPA SW 846
(methods 3050 and 6010) methodology to
determine Cd, Pb, and Zn concentrations in
the soil sample splits.  Arsenic was
determined in those splits using SW 846
method 3050 and EPA method 206.4
(spectrophotometric).

Sample preparation for EDXRF analysis
consisted of drying the sample for 4 minutes
in a microwave oven followed by sieving the
dried sample.  Material passing through the
2 mm sieve was collected as sample and was
free of large foreign objects such as
pebbles and sticks.  Drying the sample was
required due to the variable moisture
content in the submitted soils; some surface
samples had the consistency of mud.  The
sieved soil  was then ground in a Spex
shatterbox grinder (Spex Ind., Edison, NJ)
using tungsten carbide cups for 2 minutes.
Grinding cups were subsequently cleaned
using soap and tap water.   The cleaned cups
were rinsed with distilled/deionized water
followed by isopropanol.  Approximately 5
grams of prepared sample were poured into a
disposable 32 mm X-ray sample cup and
covered with a 6.3 ^m polypropylene film.
Five grams of dried sample gave the
equivalent of a 15 mm sample depth in the
cup.   Approximately twenty five samples were
prepared and analyzed per day.
STANDARDIZATION METHOD

Two sets of excitation conditions were
employed to determine seven elements  in the
soil samples, four of which are of  specific
environmental concern: Zn, As, Pb,  and Cd.
Table 1 lists the two sets of spectral
acquisition conditions and which conditions
were used to determine each analyte.  Figure
1 is a mid Z spectrum of a soil sample that
was found to contain 125 ppm As, 1100 ppm
Pb, and 729 ppm Zn.  A multiple linear least
squares peak fitting routine was used for
deconvolution of overlapped peaks.

The soil characterization method was
standardized using four standard reference
materials (SRM): NBS 1648 (urban
particulate); NBS 2704 (river sediment); SO-
1 and SO-3, two soil standards available
from the Canada Centre for Mineral  and
Energy Technology.  Standards labeled NBS
are available from the National Institute
for Standards and Technology (NIST).  These
SRMs have certified concentrations  of Fe,
Mn, Cu, Zn, Pb, and Cd.

A fundamental parameters (FP) method  (5) was
employed as the data treatment scheme and
used certified concentrations of Fe, Mn, Cu,
Zn, Pb, and Cd in the four standard
materials.  To compute instrumental
sensitivity (emission peak counts per second
per ppm), the balance of the standard was
assumed to be comprised of Si05 to  account
for the contribution of the matrix  on the
measured analyte X-ray intensity.   The
balance component Si05 was selected to mimic
the concentration of Si and 0 in typical
soils, approximately 24% Si and 45% 0.
Since none of the selected SRM's contain
arsenic, As sensitivity (cps/ppm) was
determined using a fundamental parameters
theoretical calculation based on the
computed Zn sensitivity.  Table 2 lists the
analyte sensitivities computed by the FP
method.

There are some advantages to using  an FP
method for standardization compared to site
specific soil standards.  The FP method can
use readily available, well characterized
SRMs to measure analyte sensitivities. Site
specific soil standards, by contrast, are
usually collected with a separate sampling
mobilization.  The FP method standardized
with SRM's can provide accurate analyte
                                                 518

-------
concentrations to be determined in samples
with fairly wide matrix variations without
restandardization, unlike methods
incorporating site specific standards.
RESULTS

Table 2 lists the lower limits of detection
determined using the two sets of spectral
acquisition conditions (Table 1). The
pertinent equation is: LLD (ppm)=
3/Ib/m(/T), where Ib is the background
intensity (cps), m is the analyte
sensitivity (cps/ppm), and T is the
acquisition livetime in seconds (6).
Calculated LLD values are dependent upon
spectrum acquisition times, sample matrix,
and excitation conditions.  The conditions
in Table 1 were selected to optimize the Pb
and Cd spectral regions.  Improved LLD's are
possible with EDXRF using longer spectrum
acquisition livetimes and optimized
excitation conditions for selected spectral
regions.

Results for the determination of four
analytes by EDXRF in  180 samples (43 cores
at 4 levels, two SRM's, three samples  in
triplicate) were compared to independent
analysis results in order to evaluate  the
level of agreement between the  two methods.
Table 3 lists the correlation plot data  for
the analytes in terms of actual  slope,
intercept, errors, and the correlation
coefficient of  the fit.  Each analyte
correlation plot included approximately  150
data points.

As shown  in Table 3,  slopes  of  the  plots  for
Pb, Cd, Zn, and As are within 8% of  1.00  and
all correlation coefficients  are greater
than 0.92.  The calculated slope near  1.00
and correlation coefficients greater than
0.90 indicates  agreement between the two
analytical techniques.   Figure  2 is  a  plot
of 94 data points  in  the range  of 0  to 300
ppm Pb.   Figure 3  is  a  plot  of  110  EDXRF and
 ICP analyzed samples  in  the  range of 0 to
 100 ppm Cd and  also  indicates  agreement
between the  results  of  the two  methods.

 EFFECT OF  LEAD  ON  EDXRF  ARSENIC
DETECTABILITY

 Figure  1  illustrates  the  spectral
 interference between  the  emission  lines  of
 lead and  arsenic  in  the  EDXRF  measurement.
 The As Ka (10.53  keV) and  the  Pb La (10.55
 keV) peaks  are  directly  superimposed.   Peak
 deconvolution  software  must,  therefore,  rely
 on  the  relatively  low intensity As  KB  (11.73
 keV) peak for  unobstructed arsenic  peak
shape data.   However,  to the low energy side
of the As KB is the Pb Ln (11.35 keV) and to
the high energy side of the As KB are the Pb
Ln6 (12.14 keV), the Pb Ln4 (12.31 keV), and
the Pb LB! (12.61 keV) peaks which appear as
a single peak shape in the spectrum.  As the
Pb emission lines increase in intensity with
increasing lead concentration the arsenic KB
peak becomes indistinguishable.

The nature of the arsenic/lead interference
in the EDXRF spectrum has a detrimental
effect on the arsenic lower detection limit
(LLD) in soils containing high Pb
concentrations. The magnitude of the
interference effect is directly related to
the resolution provided by the EDXRF
spectrometer.  EDXRF spectrometers with
improved resolutions would exhibit reduced
As/Pb spectral interference.  Reduced
spectral interference thereby reduces the
detrimental effect of elevated Pb
concentrations on the As LLD.

To quantify the effect of As/Pb spectral
interference on the EDXRF arsenic LLD,  148
samples  in this study were evaluated.   Of
the  148  samples, 43 samples were reported as
not  detected for arsenic by EDXRF.   Of  those
43 samples, 31 were reported  by the
independent lab as containing  12 ppm As or
higher.  Arsenic non-detects,  reported  by
EDXRF, were evaluated with respect  to  the As
and  Pb concentrations reported by the
independent lab for the  same  sample.

The  overall findings  of  the  148  samples can
be illustrated  using  analysis  results  of
four samples  (Table 4)  as  examples.  Two of
the  samples have non-detected  As  reported by
EDXRF and  two  had  detected  EDXRF  arsenic
concentrations.    In  sample A, the
 independent lab reported an As concentration
of 12 ppm  while EDXRF reported  a  non-
detected (NO)  arsenic concentration.  Note
that 12  ppm  is  the EDXRF arsenic  detection
limit.   Calculating the  ratio of As to Pb
concentrations,  as determined by  the
 independent  lab, a value of  0.083 was
obtained.   Nearly  the  same ratio  was found
 for  sample B,  again where EDXRF  reported a
ND while the  independent lab  determined 17
 ppm  As.

The  largest  absolute  As  concentration  found
 by the independent lab  that  was  reported as
 ND by EDXRF was 67 ppm  As.   That sample
 contains 1310 ppm  Pb  (1217 ppm Pb determined
 by EDXRF)  which is an As/Pb ratio of 0.051.
 EDXRF reported a non-detected As result for
 all  samples  containing  an As/Pb
 concentration ratio below 0.046.
                                                519

-------
 Table 4 also lists two examples of EDXRF
 successfully analyzing low concentrations of
 arsenic in the presence of lead.  The As/Pb
 concentration ratios for samples C and D
 were 0.046 and 0.053, respectively.

 From the data in Table 4 and the correlation
 data shown in Table 3, three findings
 emerge.  First,  the EDXRF spectrometer used
 here is unable to determine arsenic in
 samples containing an As/Pb concentration
 ratio of less than 0.046. Secondly, arsenic
 determination by EDXRF is unreliable for
 samples containing As/Pb concentration
 ratios in the range of 0.046 and 0.083.
 This is due, in  large part, to the errors in
 counting statistics for EDXRF measurements
 near the arsenic LLD.  Lastly, EDXRF results
 show excellent correlation with the
 independent lab  results for samples
 containing As/Pb concentration ratios above
 0.083.

 ACCURACY AND PRECISION

 To evaluate the  accuracy provided by the
 EDXRF method two SRMs were submitted as
 unknowns for EDXRF analysis as well  as being
 submitted to the independent lab for
 analysis.   Table 5 lists the results for SRM
 SO-2.  EDXRF analysis of SO-2 provides
 results that are in good agreement with
 certified values.   The independent ICP
 analysis of zinc in SO-2,  however, is biased
 low by a factor  of one-half.

 Precision was evaluated by submitting three
 samples a total  of three times for
 independent and  EDXRF analysis.  Table 6
 shows the results  for the  two methods along
 with the calculated standard deviation (in
 ppm)  of the three  replicate analyses.  Note
 that Cd in  sample  C was only reported by
 EDXRF to the nearest 1  ppm and three values
 of  9 ppm Cd were determined,  hence the zero
 standard deviation  for  the three replicates.
 EDXRF precision  is  better  than 10% relative
 standard deviation  in all  but  one  case (As
 in  sample C)  and compares  well with  that
 provided by  the  independent lab.

 CONCLUSION

 Field mobile  EDXRF  analysis  of soils
 suspected of  being  contaminated  provides
 information concerning  the  nature, extent,
 and magnitude of the  contamination.   Due to
 the minimal sample  preparation necessary for
 EDXRF analysis, sampling to result
turnaround time  is  relatively  short  so the
most effective use of sampling resources is
realized.  EDXRF detection  limits  below 20
ppm were obtained for the elements of
 environmental  concern.   The effect of
 increasing  lead  concentration on arsenic
 detectability  was  quantified.  Using the
 EDXRF  method described  here,  reliable As
 results  were found for  those samples
 containing  As/Pb concentration ratios above
 0.083.   Accuracy and  precision for the
 analytes of interest  using  the  EDXRF method
 was  shown to be  comparable  to results
 obtained by independent analysis.
 Comparable  results for  Cd,  As,  Pb,  and Zn
 between  independent and EDXRF methods
 validates the  use  of  EDXRF  analysis for
 hazardous waste  site  investigation and
 remediation.

 ACKNOWLEDGMENT

 The  author  would like to acknowledge James
 P. Walsh and Associates for site sampling
 and  providing  the  independent analysis data.

 REFERENCES

 (1)  Vincent, H.; Field  Screening Methods for
 Hazardous Waste  Site  Investigations
 Symposium Proceedings.  (1988),  61.

 (2)  Perl is, R.;  Chapin,  M.;  "Low Level  XRF
 Screening Analysis  of Hazardous  Waste
 Sites",  Field  Screening  Methods  for
 Hazardous Waste  Site  Investigations
 Symposium Proceedings.  (1988),  81.

 (3)  Madden, N.W.;  Hanepen,  G.H.; Clark,
 B.C.; "A Low Power  High  Resolution
 Thermoelectrically  Cooled Si(Li)
 Spectrometer",  IEEE Trans.  Nuc.  Sci..  33,
 (1), (1986), 303.

 (4)  Data Quality Objectives For  Remedial
 Response Activities.  U.S. EPA,  EPA  540/G-
 87,003, Appendix A, (1987).

 (5)  Criss, J.W.;  Birks,  L.S.; "Calculation
Methods for Fluorescent  X-ray Spectrometry-
 Empirical Coefficients vs.  Fundamental
 Parameters", Analytical  Chemistry.  40,
 (1968), 1080.

 (6) Currie,  L.A.; "Limits for Qualitative
Detection and Quantitative Determination",
Analytical Chemistry.  40, (1968), 586.
                                                520

-------
Table 1. Spectral acquisition conditions for
         the EDXRF analysis of soils.
SPECTRAL REGION
MID Z
HIGH Z
CONDITIONS
35 kV, 0.35 mA, 0.13mm
Rh filter, 200 s livetime
50 kV, 0.35 mA, 0.63mm
Cu filter, 200 s livetime
ANALYTES
Mn, Fe, Cu
Zn, Pb, As
Cd
Table 2.  Sensitivity and lower limits of detection for the
          analytes of interest.
ANALYTE
Mn
Fe
Cu
Zn
Pb
As
Cd
SENSITIVITY (cps/ppm)
0.010
0.015
0.046
0.067
0.084
0.132
0.107
LLD (ppm)
21
19
26
19
7
12
4
Table 3.  Correlation plot data for the four analytes
          of environmental interest.

ANALYTE
Pb
As
Cd
Zn

SLOPE
1.01+0.03
1.08+0.05
1.02+0.03
1.02+0.02

INTERCEPT
10.0+13.8
0.98+3.54
3.09+2.19
63.0+13.6
CORRELATION
COEFFICIENT
0.96
0.92
0.94
0.98
Table 4. Examples of four samples illustrating the
         effect of lead concentration on the arsenic
         lower limit of detection.  All  concentration
         values are in ppm.
SAMPLE
A
B
C
D
As XRF
ND
ND
28
16
Pb XRF
153
200
381
217
As AA
12
17
16
11
Pb AA
144
209
349
209
As/Pb
0.083
0.081
0.046
0.053
                            521

-------
Table 5.  Results of the analysis of SRM SO-2 by ICP
          and EDXRF methods.  All values in ppm.
SAMPLE
SO-2
ANALYTE
Pb
Zn
ICP
19
55
EDXRF
17
123
CERTIFIED
21
124
Table 6.  EDXRF and independent lab results for three soil
          samples each analyzed in triplicate.  All values in
          ppm.
SAMPLE
A



B



C



ELEMENT
As
Cd
Pb
Zn
As
Cd
Pb
Zn
As
Cd
Pb
Zn
I NO. LAB
45 + 4
20 + 2
286 + 28
185 ± 15
17 + 3
80 + 6
141 + 15
556 ± 39
17 + 1
10.0 + 0.9
117 + 8
173 + 26
EDXRF
41 + 3
31 + 3
312 + 12
134 + 10
14 + 1
58 + 4
158 + 3
529 ± 46
19 + 4
9 + 0
142 + 14
128 + 3
                             522

-------
      7
   KBV
r i r r i  ri r i i i i  i i i

?      10     11     12
i i  i i i i  i i i i

    13     14
13    16
   Figure 1. Mid  Z  spectrum of a soil sample containing 1100 ppm Pb,
             729  ppm Zn,  and 125 ppm As.  Full scale on the y-axis is
             2,000  counts.
    SO    100   190    200    250    300

          PB (ppm) BY EDXRF
                                      8
                               20      40      BO     80
                                    CO (ppm) BY EDXRF
Figure 2.   Pb correlation plot
for 94 samples.
                          Figure 3.  Cd correlation plot
                          for 110 samples.
                                   523

-------
                                                           DISCUSSION
ROGER JENKINS: Would you care to speculate on the effect of the microwave
heating on any mercury that might be in the sample. Do you think that would
drive it off or not?
ANTHONY HARDING: I don't have any experience with mercury in soils and
heating. I'm sorry.

ROGER JENKINS: Secondly, the drawings, sieving and grinding adds a
considerable amount of time to the total sample analysis time. How much do you
think that buys you in terms of increased precision or accuracy? In other words,
if we eliminated that, and just sort of mainly chopped up the sample and stuck it
in the x-ray system, what could we get?

ANTHONY HARDING: In terms of precision degradation, it would probably
be a factor of two or three. In other words, your coarsely graded sample would
be about a factor of three more or imprecise relative to the method that we took
here. And of course, if your precision has degraded, your accuracy is likely to
have degraded as well.

MARTY HARSHBARGER-KELLY: I'm  familiar with  medical x-ray
generating devices, and they typically use a tungsten or molybdenum target for
the mammographic units. What is the target material  in your x-ray source?

ANTHONY HARDING: Rhodium.

MARTY HARSHBARGER-KELLY: And you use a rhodium filter to attenuate
the beam too, for the mid-range Z?

ANTHONY HARDING: Yes, we modify the spectral distribution from the x-
ray lube lo minimize the background and produce improved excitation efficiency
for a spectral region.

MARTY HARSHBARGER-KELLY: And the rhodium is used for both the
mid-range and the high Z metals?

ANTHONY HARDING: Rhodium is selected because it's a very good general
purpose anode material. Typically unless there is a specific excitation advantage
to going to a tungsten anode, molybdenum anode, silver anode, most applications
are done quite adequately with the general rhodium anode tube.

RUDOLF GREULICH: I'd like to ask you to comment on the background
labels you  might expect in  your  samples. You are talking  about low  soil
concentrations. Those you have been showing are rather high and rather low.
Some of them might be influenced by the background. Do you know anything
about that?

ANTHONY HARDING: The detection limits were determined as interference
free detection limits in the soil matrix.

RUDOLF GREULICH: You don't know what the background levels might be?
ANTHONY HARDING: The background levels are going to be varying site to
site, in different regions.

MARK BERNICK: I'd like to comment on the, what appeared to be, 50%
recoveries of the AA lab and the zinc. And what that means in terms of the actual
accuracy of the digestive and analytical method.

ANTHONY HARDING: I wish I could comment on that particular effect.
Because it does have some ramifications in terms of our correlation data counted
later, or earlier, whatever. I really don't have a particular reason. All I know is we
submitted that sample lo two different laboratories. They were both using ICP for
thai particular soil, and they both got 55 ppm. So, unfortunately, I can't explain
it.

JOHN MORRIS: On your arsenic lead peak stripping, most of your arsenic data
were fairly low. I was wondering if you had tried MBS SRM 1645. It's the older
river sediment. It's no longer commercially available, but you ought to be able
to find some somewhere in some labs. Arsenic was 66. And the lead was 715,1
believe.

ANTHONY HARDING: That's about 10%.

JOHN MORRIS: Yes. I was just  wondering whether you could do it with the
higher levels.

ANTHONY HARDING: We were able lo do it from up to 125 ppm arsenic. But
that correlation plot data was up to 125.

JOHN MORRIS: Your actual points that you showed were lower.

ANTHONY HARDING: Well, yes. I didn't show the arsenic correlation.

JACK HERNDON: I  was  wondering how long  it takes for the detector to
stabilize after, say, a period of 24 hours or longer if the unit is turned off for that
period. How long does it lake to stabilize before you can start taking readings?

ANTHONY HARDING: From room temperature the detector takes about 45
minutes to an hour to cool down to operational temperature. I'd give it another
30 minutes for temperature  stabilization an the pulse processing electronics.
We're pretty insensitive to temperature variations that are normally obtained in
a laboratory van or trailer.

JACK HERNDON: What is the range of metals that your unit can detect? How
light?

ANTHONY HARDING: We  can delect sodium through uranium, atomic
numbers 11 through 92.

JACK HERNDON: Do you have a vacuum system for lower ranges?

ANTHONY HARDING: Yes.That particularchamberthat I showed isevacuable.
                                                                     524

-------
                 DATA QUALITY ASSURANCE/QUALITY CONTROL FOR FIELD
                         X-RAY FLUORESCENCE SPECTROMETRY
    Clark D. Carlson and John R. Alexander
          U.S.E.P.A ESAT Region 10
          The Bionetics Corporation
          Port Orchard, Washington
ABSTRACT

      Because of the nature of field  screening
with portable x-ray fluorescence (XRF) spectrome-
try, the majority of quality assurance/quality con-
trol is performed prior to and following the field
activity.  Prior to any field  screening activity, the
calibration of the instrument is the most vital area
for QA/QC in the analysis. It is recommended that
a  suite of  site specific calibration  samples be
prepared with soil which is representative of the
site to be investigated.  This soil should be col-
lected  at the site to be  investigated and should
include a set of both clean samples and contami-
nated samples, if possible. The concentrations of
these samples should  be verified by  laboratory
techniques.  These samples are then mixed and
spiked with  the appropriate analytes to give the
suite of calibration samples.  From  this  suite of
calibration  samples  is  selected  one or  more
samples  to  be used  for  initial and continuing
calibration verification  (ICV and  CCV). During  a
sampling activity, the  QA/QC measures are the
assurance of representativeness of  the sample,
particle size  , the  CCV and duplicate analysis.
Samples  must also be collected  periodically,
approximately 10% of  samples  analyzed, for
laboratory verification of the result. The results of
two field  screening activities are presented with
the protocols for quality assurance of the data.

INTRODUCTION

The use of x-ray fluorescence(XRF) spectrometry
as a method of screening hazardous waste sites
for inorganic contaminants has become a viable
option due to the commercial availability of field
portable instrumentation (1).  The usefulness of a
method as a screening tool depends on portability
of the instrument, the speed of the analysis, and
the precision and accuracy of the data.  The size
of the instruments makes them  highly portable
and the results from the analyses can be present-
ed immediately for interpretation. The question of
precision and accuracy of that data is then of the
highest priority.  In studies at mines  and hazard-
ous waste sites, the  usefulness of data obtained
from field portable XRF units has  been reported
(2).

As with any analytical method, the quality of the
data must be maintained to protect  the ultimate
usefulness  of any data obtained.   The  USEPA
defines the parameters of data quality as  the
precision,   accuracy,  representativeness,  com-
pleteness, and comparability (3). The methods for
meeting the requirements of data quality are, for
the most part, universal for  all types of chemical
analysis and field screening techniques, however,
there are some  special  situations in XRF spec-
trometry that require unique forms of quality
control. These special situations include particle
size  affects(4),   and  preparation  of calibration
samples(S).  The other, standard, QA  methods are
calibration of the instrumentation(linear correlation
of the calibration), calibration verifications, dupli-
cate analyses, laboratory verification, and sample
representativeness.  This paper will describe the
quality control protocols used in field  screening of
various hazardous waste sites.

PROCEDURE

The field portable XRF unit  used  in acquiring all
data  was  an   X-MET  880  manufactured  by
Outokumpu  Electronics.  The X-MET 880 is an
                                               525

-------
energy dispersive x-ray fluorescence unit with two
radioactive sources (244Cm and 241Am).   For the
examples which are presented in this paper, only
the low  energy  source (244Cm)  is used.   The
calibration curves are calculated with the instru-
ments linear regression software utilizing all of the
calibration sample.

Prior to explaining the quality assurance/quality
control associated with the field use of the instru-
ment, a rudimentary knowledge of the procedures
used will assist  in the evaluation of the QA/QC
protocols.   Below  is the general outline  used
when sampling with the X-MET 880 X-ray fluores-
cence spectrometer.

1.    Calibration of the instrument.

            a.    only necessary  if elements not
                 previously calibrated for are re-
                 quired, or if a close match be-
                 tween the matrices of the cali-
                 bration standard and the sam-
                 ples is desired.

            b.    calibration can be performed
                 with spiked  samples or previ-
                 ously analyzed  samples.   A
                 minimum of four samples per
                 element  (up to 20 samples)
                 should be used for the calibra-
                 tion.

2.    Check the calibration of each model to  be
      used (A model is a calibration including up
      to six elements).  This  is to be done  by
      measuring a check sample (a control stan-
      dard). If the  values of the elements in the
      sample are outside of one sigma (1  a), the
      STA command on the X-MET 880  can  be
      used to reslope the calibration curve.

3.    Prepare the instrument for field use.  Check
      the charge on the batteries  (each battery is
      good for 8 hours of use), inspect the instru-
      ment to see that all cords are in place and
      in good condition ,  and cover the probe
      window with the polypropylene film to keep
      it clean during use.  Bring extra polypropyl-
     ene film for replacement if necessary.

4.    Upon arrival at the site, the instrument is to
     be turned on and allowed to equilibrate for
     at least 30 minutes.  The instrument also
     needs to  be  allowed  to gain control for 5
     minutes after each 20-25  minutes of use. If
     the surrounding temperature is changing
     rapidly, the  gain  control should be per-
     formed at shorter time intervals.

5.    Prepare the site for sampling

           a.    determine the  frequency  of
                sampling

           b.    make a map of where sampling
                is to occur  (these two  steps
                may be performed  prior to ar-
                rival)

           c.    determine the number of meas-
                urements per point  on the map
                which will give a representative
                value for the points. This num-
                 ber will be between four and
                seven  (seven giving a  confi-
                 dence interval of >90%)

 6.   Preparation of each point for measurement

      Immediately  prior to taking the measure-
      ments at  a point on the map, a  representa-
      tive sample  to be  measured must be ex-
      posed.

            a.    for  surface   studies,  organic
                 matter (stick, grass, bugs, etc.)
                 and any large unrepresentative
                 objects should  be removed.
                 The  surface  may  be  scraped
                 with a rake or a shovel.

            b.    for  subsurface  studies,  the
                 proper amount of surface mate-
                 rial must be removed. This can
                 be done  using a shovel or oth-
                 er digging apparatus and get-
                 ting to the level of interest, or
                                                 526

-------
                  by obtaining  a core  sample
                  (with a coring tool)  and mea-
                  suring at the proper  depth.

            c.     the samples (four to seven) are
                  homogenized by the quartering
                  technique and sieved to 9mm.
 7.    After preparing each  point, the measure-
      ments can be obtained.  One measurement
      is required for each model used the analy-
      sis. A counting time of 50 seconds should
      be used for each measurement. This value
      can be adjusted by the operator if the matrix
      characteristics are such that longer or short-
      er counting times  are  indicated.   Factors
      influencing this  decision  include  particle
      size, moisture content and homogeneity of
      the matrix.

      The data to be recorded for each measure-
      ment are:

           a.    Concentration  response   for
                 each element
           b.    counting statistic for each ele-
                 ment (gives the standard devia-
                 tion due to the measuring time
                 used)
           c.    intensity data for each element

8.    For quality assurance,  the control sample
      should  be measured before the measure-
      ments begin, after each  10 measurements,
      and after the final measurement. As indicat-
      ed earlier, the control  sample  should  fall
      within ten percent (10 %) of the  actual value
      or  the  model should  be  restandardized
      using the standardization function on the
      instrument.  A duplicate analysis should
      also be performed for precision analysis.

9.    Repack the instrument for travel back to the
      lab, inspecting for any problems.

It should be noted that the quality control referred
to in section 8 is that associated with only the field
analysis and not the pre-screening QC.
 The  QA/QC  involved in the  process of  field
 screening for metals with a field portable XRF unit
 can be divided into two main categories, the pre-
 screening QA/QC followed by the field (screening)
 and laboratory (post-screening) QA/QC.  These
 two main categories can then be further separated
 into the individual elements that make up the
 protocols involved in the assurance of the quality
 of the data.

 PRE-SCREENING QA/QC

 Prior to any field screening  activity, a number of
 procedures can be followed to assist in the quality
 control  of the final product. The first of these is in
 the calibration of the instrumentation.  There are
 two primary methods for obtaining a calibration
 curve for a field screening application which will
 produce data of acceptable quality; a site specific
 calibration and a generic calibration. For the site
 specific calibration,  a  number of samples  with
 varying concentrations of interest must be ob-
 tained for each site which will be screened.  For a
 generic calibration, a suite of samples prepared
 by the spiking of a generic, or common, soil with
 various levels of the analytes of concern.   The
 former of these  methods will be more time  con-
 suming and therefore have a greater cost associ-
 ated  with  the  analysis, while the latter  may not
 take into consideration matrix affects from the soil
 on  the  site.  Of the two methods the  one that
 appears to give the best results is the site specific
 calibration.

 The best way of preparing site specific calibration
 samples is to have a series of analyzed  samples
 obtained from the site of interest which contain the
 proper ranges of analytes of interest. This method
 would give a reliable calibration but would defeat
the purpose of having an instrument to screen a
 site  for possible contamination  since  the  site
would already be well characterized. An alterna-
tive involves the spiking  of samples obtained  at
the site to be screened.  In some cases it is not
 possible to obtain samples prior  to a screening
activity and in these cases a generic calibration
will be the only option.
                                               527

-------
When a request for screening at a site is made, a
minimum of two samples from the site are re-
quired to be used in creating the calibration sam-
ples.  One of the samples must be from a part of
the site which  is considered to be uncontami-
nated, or "clean", while the subsequent samples
should be obtained from what  is expected to  be
highly contaminated areas of the  site.   These
samples are dried,  sieved through a 9mm  pore
size sieve,  homogenized and  analyzed  in the
laboratory.  The method for analyzing the samples
to be used  in the calibration will depend on the
type of results are desired. The XRF can emulate
whatever method is called for in the project  plan.
If the analysis is to emulate a total contaminant
digestion, then the calibration samples should be
analyzed using SW846 Method 3051 (6) followed
by ICP-AES. The data can also emulate a TCLP
type digestion or a total digestion (i.e. hydrogen
fluoride). In addition to using this data in prepar-
ing the calibration samples, the analysis may also
give some  idea as to what unexpected contami-
nants may  be present at the site. The clean soil
is used as the  blank, the soil to be spiked, and
the diluent for the contaminated soil.  The contam-
inated soil  is used as the limit for the calibration
curve, unless a greater range of concentrations
than  this will account for is requested.  In this
case, the clean soil can be spiked at higher levels
or spikes can be added to the contaminated soils.
 From the samples obtained at the site, five calibra-
 tion samples are prepared.  These five consist of
 the  clean  soil, the contaminated soil,  a  25/75
 mixture of the two samples, a 50/50 mixture of the
 two samples, and 75/25 mixture of the two sam-
 ples. The next 15 to 25 samples are prepared by
 spiking these five mixtures  with the analytes of
 interest, giving 20 to 30 calibration  samples to
 create the  calibration  curve.  The spiking  of the
 samples is preformed with the oxides and nitrates
 of the analytes of interest in varying ratios as is
 seen in the table below:
                                                  Table 1 Site 1 Calibration Samples
#
1
2
3
4
5
6
7
8
9
10
As
<3.0
<3.0
<3.0
<3.0
10000
6000
5100
2000
1000
5000
Cr
12
136
110
90
6012
4012
1112
10012
2012
5012
Pb
201
2590
2110
1630
4590
12590
8590
3590
6590
7590
Zn
200
6870
5540
4200
7870
9070
16870
10870
12870
11870
Fe
12200
47800
40680
33560
12200
12200
12200
12200
12200
12200
 all concentrations in mg/Kg
The use of the oxides and nitrates is due mainly
to their availability and  ease of handling.  The
above table is a partial list of the values used in
the calibration for the screening of  a superfund
site in region 10 (see results section). The prepa-
ration of the calibration samples normally requires
three to four labor hours. Twenty gram samples
of the dried and sieved  clean soil are measured
out, one for each element  to be analyzed.  To
these samples is added a nitrate or an oxide of
the analyte of concern in a proportion to give  a
sample concentration  of ten  weight  percent.
These samples are then homogenized and used
in the proper ratios with the clean soil to give the
calibration samples.  Three  of these  samples are
analyzed using Total digestion  and  ICP-AES to
verify the concentrations. One of the three sam-
ples is chosen as the control sample for the  ICV
and CCVs.  For this site the sample chosen as the
control sample  was # 10.  Normally one of the
samples which  has  not  been spiked is used as
the control sample but since there was little or no
arsenic or chromium in the sample, it was decided
to use  a sample which contained all of  the anal-
ytes of interest.

When preparing the calibration curves, the corre-
lation coefficient shows the linearity of the calibra-
                                                528

-------
tion.  Since there is a direct correlation between
intensity and  concentration in XRF spectrometry,
the quality of the data will be dependant on the
linear correlation.  The  acceptance limit for the
linear correlation used in this study was 0.990.  If
the correlation falls below the limit, the intensities
for the calibration samples are recollected and if
the correlation is still  low, the samples with low
values (far off the calibration  line) are re-prepared
and reanalyzed.

FIELD AND LABORATORY QA/QC

In the field portion of the analysis, there are a
number of areas where the quality of the data
must  be documented.  As with any instrumental
technique, there are  QC requirements  in XRF
spectrometry  including the initial and continuing
calibration verification  and duplicate analysis. An
aspect of XRF  spectrometry which  can  cause
unique problems is that particle size can affect the
results and so must be controlled.  Finally, the
field analysis  itself gives rise to possible sources
of error, such as how representative the  sample
collected  is  and  laboratory verification  of the
results. All of these factors must have associated
QC/QA to document the quality of the results.

The normal QC which  is followed with any instru-
mental technique include the initial and continuing
calibration verification (ICV and CCV respectively).
The ICV is performed  prior to any sampling and
the CCVs  are performed after every ten sampling
sites.   If any of the verifications  are out of the
control limits,  the control limits being ± 20%, then
the calibration needs to be restandardized. The
X-MET 880  software  has  a   restandardization
function built in  so that all that is required is a
remeasurement of a standard sample. Since there
are no moving parts within the instrument, there is
rarely a need  for  a restandardization during a field
screening activity.   A  duplicate analysis is run to
give an idea as to the precision of the analysis.  A
sample is chosen at  random for the  duplicate
analysis.

Particle size  affects can cause discrepancies in
XRF data and so it is necessary to minimize these
affects.  One  way of proceeding to this end is to
match the field sample particles size to that used
in the calibration samples as closely as possible.
By avoiding a difference in particle size, the affect
of particle size should be a minimum. To accom-
plish this, all samples are sieved with a standard
sieve to less than 9mm.  This size of sieve was
used so that all particles that can be considered
soil are included in the analysis. To get an even
closer match in particle size would require some
sort of particle size reduction which would greatly
increase the time required for the field screening
procedure.

The problem of obtaining representative samples
will be of concern when the site is large and the
sampling areas are spaced some distance apart.
There is  a need to make sure that the readings
obtained  are a  reasonable  reflection of the con-
tamination  at the  sampling site.  To obtain  a
statistically representative sample, seven  unique
portions  are  obtained from  the  site and  then
homogenized  using  the quartering technique.
The  homogenized sample is then  sieved (see
above) and analyzed with the XRF spectrometer.
The final QC requirement for the field analysis is
the collection of samples to be used for laboratory
verification.  The samples are to be collected from
the sieved material at a frequency of approximate-
ly 10% of the total samples. This gives a range of
sample concentrations to help in the interpretation
of the field  data.  These  samples will then  be
analyzed by the USEPA approved method at a
laboratory to determine the  accepted concen-
trations of contaminants at the  site.

RESULTS

This section will give the  result obtained at two
hazardous waste sites using the X-MET  880 to
analyze for a variety of elements. Both of the sites
were analyzed using a site specific calibration. In
both cases, the calibration samples were prepared
using two samples from the site, one contaminat-
ed and the other uncontaminated, with spiking of
the soils.
                                               529

-------
Site 1 is a \unkyard which at one time contained
transformers and lead-acid batteries in addition to
other types of scrap metal. This is an eleven acre
site with many type of soil including fill brought in
from other places.  The element of most concern
at this  site was  lead.  The  three figures  below
(Figures 1, 2, and 3) show the correlation between
the laboratory results and the field XRF spectros-
copy results.  In the preparation of the calibration
curve, the analysis on the preliminary samples
was  performed using  SW846 method  3051  (6) for
the digestion and ICP-AES for the analysis (see
Table 1).  The laboratory analyses on  the verifica-
tion samples used the same  procedures.
Figure 1  Site 1 correlation for lead results
                                                   Figure 2  Site 1 correlation for zinc results.
                                                                   Zinc


^
a.
a.
<2 <"
3 "g
 W
"- 3
sj:
o
.B
Jj


10
9
8

7
6

5
4
3

2
1
0
.






'm

m

" •"

                                                                     5         10
                                                                    Thousands
                                                                 XRF results (ppm)
                                                   F'gure 3 Site 1 correlation for iron results.
      35

      30
                      Lead
  E
  c.
  3 "§20
  05
  u
  O
10

 5

 0
                                                    E
                                                    c.
                                                  70

                                                  60

                                                  50
                                                                         Iron
                                              "3 c 40
             5    10   15  20   25   30  35
                   Thousands
                XRF results (ppm)
                                                    X3
                                                    3
20

10

 0
                                                           0  10  20 30  40  50 60  70 80  90
                                                                      Thousands
                                                                   XRF results (ppm)
                                                530

-------
 Table 2  Duplicate XRF Analysis for Lead
         at Site 1
Sample #
1
2
3
4
5
6
7
8
9
10
Analysis #1
8600
28630
12600
12580
16480
5430
14110
8540
7910
10270
Analysis #2
7270
9370
6970
15200
9500
5870
14370
7250
7290
13100
From the Figures and Tables above, it is apparent
that the analysis simulated the laboratory data
fairly well.  Other elements were analyzed for (As,
Cr, and Cu) but they were all  present below the
detection limit of the calibration. The discrepanc-
ies in the duplicate analysis can be explained as
a problem in the variability of the contamination at
the points  and the  difficulty in obtaining a large
amount during the duplicate analysis.  The large
variability may be due to one small nugget in the
first sample which  contained a large  amount  of
lead.

There  were  no problems encountered in the
continuing  calibration  verification.   All of the
values for the control sample were within 10%  of
that found  in the laboratory analysis so  restand-
ardiation was not required during the analysis.

Site 2 is a former Oil  filtering operation for the
reclamation of used oil.  The  filtercake  material
was buried and the covered over with gravel to
make a parking lot.  The samples were obtained
using a drilling rig.  This was a very small  site with
the sampling occurring at 12 boring holes. The
main element  of concern  was lead.   Table  2
shows the  comparison of the field readings and
the laboratory values for the two samples used  in
the verification.  The ultimate purpose of the XRF
screening  was to determine, on the site, the hot
spot for lead contamination. In the preparation of
the calibration curve, the analysis on the prelimi-
nary samples was performed using SW846 meth-
od 3051 for the digestion and ICP-AES for the
analysis. The laboratory analyses on the verifica-
tion samples used the same procedure.
                                                  Table 3  Results from Site 2

                                                               Hot Spot
                                 Representative

Pb
As
Cr
Lab
(ppm)
15384
18.11
41.57
XRF
(ppm)
15500
<180
<100
Lab
(ppm)
4608
24.99
90.39
XRF
(ppm)
5050
<180
110
The results show that the X-MET 880 XRF spec-
trometer gave results that were close to that found
during the laboratory verification.
No  problems  encountered  in  the  continuing
calibration verification.  All of the values for the
control sample were within 10% of that found  in
the laboratory analysis so restandardiation was not
required during the analysis.
SUMMARY

This study has shown the quality objectives for
utilizing XRF as a screening tool for metals at
hazardous waste sites. The results from the field
screening appear to emulate the data obtained
from the laboratory verification.  A major factor in
the quality of the results would  appear to be the
site  specific  calibration.   The  use  of  the  site
specific calibration appears to give good quality
data without adding a great amount of time to the
pre-screening process. The data presented is the
product of two sites which had very different soil
types which could create difficulties when using a
                                              531

-------
generic calibration.

The  other methods of quality control are  also
responsible for the quality of the data received
from the screening process.   These methods
include the  calibration  verification,  duplicate
analysis,  assurance of the representativeness of
the sample, particle size,  and laboratory verifica-
tion.  These steps are already accepted methods
in the collection and analysis of any environmental
samples.

When the protocols listed  above are followed, the
data obtained from the screening of hazardous
waste  sites for inorganic contaminants  by x-ray
fluorescence  spectroscopy correlates well  with
confirmatory results and require minimal reanaly-
sis in the field.  The use of XRF spectroscopy as
a screening tool will meet the criteria establish for
these tools, those being speed of analysis, accu-
racy of the method, cost effectiveness  and quality
of the data.
REFERENCES

(1)    Chudyk, W.; "Field Screening of Hazardous
      Wastes"; Environ. Sci. Technol., Vol 23, No.
      5, 1989, pg. 504

(2)    Chappell, R.W.; Davis, A.O.; Olsen,  R.L;-
      "Portable X-Ray Fluorescence as a Screen-
      ing Tool for Analysis of Heavy Metals  in
      Soils and Mine Wastes", Proceedings of the
      Conference on Management of Uncontrolled
      Hazardous Waste Sites, U.S. Environmental
      Protection Agency, U.S. Government  Print-
      ing Office, Washington D.C.,  1987,  EPA-
      540/G-87/004, pg. 115

(3)    USEPA, "Data Quality Objectives for Reme-
      dial   Response  Activities:  Development
      Process"  U.S.  Environmental  Protection
      Agency, U.S. Government  Printing Office,
      Washington D.C., 1987, EPA-540/G-87/003,
      P9-4
(4)   Grant, C.L;Pelton, P.A.; "Influence of Sam-
     pling on the Quality of Analysis with Empha-
     sis on Powders" Advances in X-ray Analysis,
     Vol.  17, 1974, pg.44

(5)   Piorek, S.;Rhodes, J.R.; "A New Calibration
     Technique for X-ray Analyzers  Used  in
     Hazardous Waste  Screening" Proceedings
     of the 5th National RCRA/Superfund Confer-
     ence, U.S. Environmental Protection Agen-
     cy, Environmental  Monitoring and Support
     Laboratory: Las Vegas, NV, 1986, pg. 428

(6)   USEPA, "Test Methods for Evaluating Solid
     Waste" Fourth Edition, U.S. Environmental
     Protection Agency, U.S. Government Print-
     ing Office, Washington D.C., 1988, SW846,
     Vol.16, pg. 3051-1
ACKNOWLEDGEMENTS

The authors would like to thank the U.S. Environ-
mental Protection Agency for funding this work.
We would also like to express our thanks to Mr.
Roy R. Jones for his many useful discussions and
assistance throughout this project.
                                              532

-------
                                                            DISCUSSION
RUSS SLOBODA: My question involves the site specific calibration. Most soil
studies usually involve a variety of matrices at a specific site and it would seem
to be rather naive to say that one calibration is specific to the whole site. What
type of protocol do you have to help you decide? Is it a geologist helping you to
decide if these are different mineral matrices? Or do you test in your base lab
every single thing that might be a different matrix before you decide on how may
site-specific calibrations  are  necessary  for the types of matrices you're
encountering.

CLARK CARLSON: What we like to do, of course, is to have enough samples
from the site. We require at least two. But depending on the size  of the sites
(because most of the sites that I 've done haven't been that large), the assumption
that just a few samples will get us a fairly good correlation as to the  matrix over
the whole site, I feel, is a good one. But when you're talking about very large sites,
you may run into some problems with the matrix, and you may have to do more
than one site-specific calibration.

JOHN BARICH: When specifying a job, what is your rule of thumb as to what
percentage of your budgeted dollars and what percentage of your performance
period should be devoted to the QA program?

CLARK CARLSON: As far as the time allotted, it usually takes on the average,
about four or five hours to make our calibration samples. And depending on the
size of the site, of course, that's going to affect the percentage of time that you're
going to use for that particular portion of the QA/QC. But other than that, since
we're doing the duplicate samples and the sample verification, the rule of thumb
that I've been using is roughly between 20% and 25% of the time.
                                                                     533

-------
           A Study of the Calibration of a Portable Energy Dispersive
                        X-ray Fluorescence Spectrometer
                     C.A.  Ramsey,  D.J.  Smith,  and E.L.  BourO  *
                 United States Environmental Protection Agency
                   National Enforcement Investigations  Center
                       Box 25227,  Denver,  Colorado  80225
ABSTRACT

Generation of reliable concentration
information from portable energy
dispersive X-ray fluorescence
necessitates the development and use of
appropriate calibration methods.  The
present study considers some of the
difficulties encountered with the use of
empirical calibration, which is obtained
from the measurement of standards with a
similar matrix to that of the samples
being investigated.  The effectiveness
of two sets of empirical calibration
standards have been investigated, namely
a set of site specific samples analyzed
using a referee method, and a set of
artificially prepared calibration
samples produced by spiking an
uncontaminated soil matrix.  It was
found that sample matrix variation
unaccounted for in the calibration leads
to uncontrolled bias in the analytical
results.

INTRODUCTION

Of major concern in the characterization
of many hazardous waste sites is the
rapid and accurate determination of
metal concentrations in solid and liquid
samples.  Generally, these
determinations are carried out in an
off-site laboratory using techniques
such as atomic absorption (AA)
spectrophotometry or inductively coupled
plasma (ICP) atomic emission
spectrometry.  These methods, while
relatively accurate, are time consuming
and expensive; they also cannot
practically provide on-site, real-time
results.  Thus, there is a need for a
supplementary technique that can provide
adequate information about a site in a
timely and less expensive manner.
Portable energy dispersive X-ray
fluorescence (FPXRF) spectrometry is a
promising method to meet this need.
Instrumentation is commercially
available which utilizes radioisotope
sources and gas-filled proportional
counters.

Problems exist, however, which are
inherent to X-ray fluorescence (XRF)
methods that must be addressed to
provide useful concentration data for
unknown samples.  Some of these
problems, such as particle size and
mineralogical effects, are common to all
types of XRF and can be minimized by
preparation of all samples and standards
to a uniform state.  Other well-known
problems consist of interelement
effects, which include spectral overlap,
primary and secondary absorption, and
enhancement.  Absorption and enhancement
effects are commonly addressed in
laboratory XRF using computational
procedures such as the "Fundamental
Parameters" approach  (Jenkins, 1974).
With contemporary laboratory wavelength-
and energy-dispersive instrumentation,
spectral interferences present little
difficulty in all but a few cases.
Thus, in laboratory implementations of
XRF, results comparable to AA and ICP
are routinely achievable.
To date, the aforementioned problems
have not been adequately addressed in
applications of FPXRF to environmental
monitoring; of major concern is the
relatively poor spectral resolution of
the proportional counter detectors,
which require larger overlap correction
factors for adjacent elements than solid
                                         535

-------
state dectectors.  In addition,
approaches such as "Fundamental
Parameters" are not practical to
implement in FPXRF because of the
presence of many unresolved spectral
interferences, computation requirements,
and the inability to measure all
components in the sample.  Thus,
empirical calibration approaches have
been used in FPXRF; these relate the
measured spectra of a "training set"
(calibration set) to the concentrations
of elements present in the training set
samples, typically using an approach
such as multiple linear regression
(MLR).  Two types of training sets are
common, namely real samples analyzed by
referee methods  (site-specific models),
and synthetically prepared standards
(generic models).  For both types of
training sets, the analyst attempts to
match the matrix of the training set to
the anticipated unknown samples.
Obviously, selection of a proper
training set matrix and the
determination of its applicability to
unknown samples are critical problems.

This study involved the application of
empirical calibration-based FPXRF to an
environmental monitoring situation,
namely, the determination of the
elements chromium, zinc, and lead in a
soil-like waste material.  Samples from
a site contaminated with heavy metals
were collected and analyzed by ICP.
These samples were used to evaluate the
performance of site-specific and generic
training sets.

XRF THEORY

The first step of any X-ray analysis is
the removal of inner shell electrons
from the atom.  A vacancy is then
created which is immediately filled by
an electron from a higher energy shell.
The resulting free energy is emitted as
radiation that is characteristic of the
excited element.  The second stage is
the selection of an emission line from
the element of interest by means of a
wavelength or energy dispersive
spectrometer.  Next is the detection and
integration of the characteristic
emission line of interest, and finally,
the conversion of intensity to
concentration by the use of some
calibration procedure.

Until the late 1960s nearly all X-ray
spectrometers were the wavelength
dispersive type in which wavelengths are
separated by Bragg diffraction from a
single crystal.  Although wavelength
dispersive X-ray fluorescence (WDXRF)
spectrometers  have high spectral
resolution, they are bulky, expensive,
and require a high power x-ray tube as
an excitation source.

More suited for field work are energy
dispersive X-ray fluorescence (EDXRF)
spectrometers.  EDXRF spectrometers,
being more efficient than wavelength
dispersive spectrometers, can utilize
small radioactive sources for excitation
instead of large X-ray tubes.  In
addition, the separation of emission
lines does not require the use of a
large crystal chamber and goniometer as
does a WDXRF spectrometer.  There are
several types of detectors that are
employed in FPXRF spectrometers:
scintillation counters, solid state
detectors, and gas-filled proportional
counters.  The scintillation counter has
very poor resolution and requires the
use of balanced filters to discriminate
between lines.  The need for filters
increases the mechanical complexity  and
limits the flexibility of an instrument.
Solid state detectors use a crystal of
lithium drifted silicon.  Silicon
detectors have very good resolution
(-0.16 KeV full width at half maximum
for the 5.89 KeV Mn-Ka) but require the
use of liquid nitrogen or thermoelectric
cooling to minimize electronic noise.
Gas filled proportional counters have an
intermediate resolution  (-0.77 KeV for
Mn-Kg) between the scintillation counter
and the solid state detector but do not
require cooling.  Thus, the combination
of radioisotopic source for excitation
and gas-filled proportional counters for
detection are more suited for field
portable instruments.

EXPERIMENT

Samples and Standards.  Contaminated
soil samples were collected from a site
bearing potentially hazardous waste
material.  Initially, 25 samples were
taken from soil that was contaminated
with heavy metals from a metal recycling
plant.  Fifteen of these 25 samples were
used as a site-specific training
(calibration) set, referred to hereafter
as TRAIN1.  The TRAIN1 samples were
selected to provide the maximum range of
concentrations for both the analytes and
major elements.  The remaining 10
                                           536

-------
samples were used as an "unknown"
testset; these are referred to as TEST1.
An additional testset (TEST2)  of eleven
samples, taken from a different area on
the same site, consisted of material
from the same process produced at an
earlier time.  ICP analysis showed that
the levels of iron, calcium, and silicon
were higher in the TEST2 samples than in
the soil from the TRAIN1 group (refer to
Table 1).

The generic training set (TRAIN2)
consisted of synthetically prepared
standards obtained commercially.  These
standards had been produced by spiking
an uncontaminated sandy soil matrix with
chromium,  copper, zinc, arsenic,
cadmium, and lead.

Sample Preparation.  Approximately one
kilogram of each soil sample was air-
dried to constant weight and sieved
through a two millimeter nylon sieve.  A
50 gram subsample was selected by
randomly taking approximately 50 one
gram aliquots from the primary sample.
This subsample was then ground in a
tungsten carbide rotary ring mill to
minus 200 mesh particle size.   From the
ground subsample, an analytical sample
was selected for FPXRF analysis, which
consisted of approximately 30-40
increments of about 0.25 grams each,
which were then placed in a
polypropylene cup for the analysis.
Analysis of replicates by ICP verified
the homogeneity of the ground samples
(less than five percent relative
standard deviation between 0.25 gram
aliquots).  These measures minimized
subsampling and particle size effects.

Samples were prepared for ICP by fusing
0.25 grams of ground material with 2.0
grams potassium hydroxide (KOH),
followed by dissolution of the melt in
hydrochloric and nitric acid (HCl-HNOo).
Fusion was selected over commonly used
acid digestions such as EPA Method 3050
because of greater accuracy for critical
elements such as chromium and iron;
procedures such as Method 3050 tend to
yield low results for many elements
because of poor attack of silica-based
minerals.   This fusion procedure
produced reliable analytical results for
thirty elements of interest, as
evidenced by acceptable results for
spiked samples, replicate samples, and
reference materials.  Accuracy of these
results was further verified by
determination of the analytes using a
laboratory-based EDXRF.

FPXRF Measurements.  An Outokumpu X-HET
880 FPXRF, containing a 30 mCi 244Cm
excitation source and an argon
proportional counter, was used in these
studies.  Data was acquired using a 200
second measurement time, and a
consistent sample presentation geometry.
Chromium, zinc, lead, and backscatter
(BS), as well as several potential
interfering elements were measured in
the training set.  The Ka emission lines
for chromium and zinc were selected and
the La line for lead was used in Trainl
because no arsenic was present in the
samples.  The 1^ line for lead was used
in Train2 due to the presence of arsenic
in the standards.  Spectral
interferences were treated using a
Gaussian elimination algorithm
(subtracting the portion of signal due
to the interferant after measuring
interferant intensity and using a pre-
established correction coefficient)
provided for in the X-MET software
(Outokumpu Oy).  Stepwise multiple
linear regression was used to develop
models accounting for absorption and
enhancement effects; interference-
corrected spectral intensities were used
as dependent variables; concentrations
of the elements were used as independent
variables.  The significances of
independent variable effects were
determined using t-tests.  Some
standards were omitted from the model if
significant improvement in r  was
achieved by doing so.  A summary of the
models generated is presented in Table
2.

RESULTS AND DISCUSSION

Examination of Table 3 reveals that bias
for all elements generally increases in
the order (TRAIN1 ; TEST1)
< (TRAIN1 ; TEST2) «  (TRAIN2 ; TEST1 or
TEST2).  Chromium could not be
effectively quantitated in any samples
by the TRAIN2 (generic) model.  Indeed,
chromium was not detected for most
samples even though 400-600 mg/kg Cr was
actually present.  The bias problem was
not due to lack of measurement precision
(see Table 4).  A possible explanation
is as follows:  The Cr-Ka, being of low
energy, is particularly susceptible to
sample matrix effects, such as the
presence of iron at varying levels.  The
analogous effect is not as pronounced
                                         537

-------
for the higher energy Zn-Ka and Pb-La.
Additionally, the Cr-Ka line is subject
to spectral interference from the Fe-Ka
line, which is problematic when the
concentration of the interferant  (Fe) is
high compared to the analyte (Cr).
A difficulty with the TRAIN2 (generic)
calibration model is the lack of
definition of the sample matrix effect
(differing absorption/enhancement
correction coefficients).  Elemental
analysis of the TRAIN2 standards by lab-
based EDXRF revealed very little range
in the concentrations of major elements
such as iron, silicon, and calcium.
Empirical training models containing no
variance in influential parameters (e.g.
matrix, concentration of interferants)
cannot be expected to produce models
which are robust with respect to these
variations.  For TRAIN2, t-values for
iron and backscatter were statistically
insignificant in this set of standards,
indicating that no variances exist for
major elements in the TRAIN2 materials.

Differences in matrices are thus a main
pitfall in FPXRF.  It is unrealistic to
believe that, for field applications,
the matrix will be identical for all
samples encountered.  It is also
difficult to determine, in real time,
the applicability of a specific training
set to a particular sample.  A specific
training set may be inadequate due to
the presence of unanticipated spectral
interferants, or large matrix
differences, or both problems.  In this
study, changes of less than a factor of
two in the concentrations of iron,
calcium, and silicon between TEST1 and
TEST2 strongly influence the biases of
the resulting data.

FPXRF calibration models are frequently
evaluated based upon the value of the
regression coefficient, r .  It is
essential to note that good correlation
alone does not ensure accurate results.
The regression coefficients for all
analytes using TRAIN1 and TRAIN2 were
close to unity,  yet many predictions
were highly biased.  While a model with
a low r2 value lacks any predictive
capability,  a high r  alone does not
guarantee its predictive capability for
a specific test sample.

CONCLUSIONS

FPXRF is an analytical technique which
is not highly robust with respect to
 spectral  interferences  and sample  matrix
 effects.   Even with site  specific
 empirical calibration,  the applicability
 of  a particular  model to  a specific
 sample  cannot be ensured.   In  some cases
 several models might have to be employed
 to  cover  the entire range of analytes,
 inteferents, and matrices.  Quantitative
 analyses  of varying, unknown hazardous
 waste streams present a challenge  of the
 highest order to FPXRF  application;
 substantial possibility exists for false
 positive  and false  negative readings, as
 well as highly biased quantitation.  Low
 atomic  number elements  and lower
 concentrations appear to  be more
 susceptible to quantitation problems.
 However,  FPXRF is presently very useful
 as  a field analytical device for
 problems  such as contamination
 delimination and segregation of waste
 streams.  Qualitative,  semi-
 quantitative, and quantitative
 analytical results  are  all potentially
 achievable on a case-by-case basis.
 Assuming present instrumental hardware,
 additional research  in  FPXRF should be
 directed towards development of more
 robust calibration methods, for example
 chemometric calibration.

 REFERENCES

Jenkins, Ron, An Introduction to X-ray
 Spectrometry, Heyden, London,  1974.

Outokumpu Oy, - "Operation Instructions
X-MET 880 Analyzer Ver.  1", Outokumpu
Oy,  Espoo, Finland.
                                           538

-------
                         Table 1

     Matrix and Analyte Concentration Ranges  (rag/Kg)


       Element        TRAIN1                  TEST2

         Fe        83900-127000           87500-196000
         Ca        17900-30700            27600-49100
         Si        64400-146000          121000-220000
         Cr          497-766                307-567
         Pb         4740-8230              3210-8560
         Zn         9280-20100             5840-26800
                         Table  2

                Summary  of  Calibration  Data


         TRAIN1 (Site-Specific)  Calibration Model

               Interfering        Points       Regression
Analytes         Elements         Omitted      Coefficient

   Cr         Cr, Ti, Fe, BS         2             .990
   Pb         Pb, Fe, BS             3             .990
   Zn         Zn, Fe, BS             1             .991


            TRAIN2  (Generic) Calibration Model

              Interfering         Points       Correlation
Analytes        Elements          Omitted      Coefficient

   Cr        Cr, Mn, Fe, BS          0             .998
   Pb        Pb, BS                  0             .996
   Zn        Zn, BS, Pb              0             .998
                            539

-------
                                           Table 3

                                    Bias Data  for Samples
                              Absolute Percent Relative Error
                          TRAIN1  (Site-Specific)  Calibration  Model

                                      Chromium           Lead            zinc

                 TRAIN1                  7.0              9.3               5.6
                 TEST1                   8.9              5.3               3.5
                 TEST2                 100.0            23.3              13.1


                             TRAIN2  (Generic)  Calibration Model

                                      Chromium           Lead            zinc

                 TRAIN2                 12.0              15.1             8.9
                 TEST1                  97.9              33.8            39.8
                 TEST2                  86.0              38.4            49.6

               NOTES:  The training set  rows refer to the mean magnitude of
               the relative errors produced by re-measurement of  the
               training  set using the developed model.   The testset rows
               refer  to  the mean magnitude  of the relative errors produced
               by measurement of the samples using the  developed  model.
                                           Table  4

                                    Measurement Precision
                            Percent Relative standard Deviation

                         TRAIN1 (Site-specific)  Calibration Model

                       Samples           Chromium     Lead     zinc

                       TRAIN1                 4.2          2.1      1.3
                       TEST1                  8.4          1.7      0.7
                       TEST2                  ND           1.9      1.9


                            TRAIN2 (Generic)  Calibration Model

                       Samples           Chromium     Lead     zinc

                       TRAIN2                 1.3          4.7      0.5
                       TEST1                  ND           2.7      1.8
                       TEST2                  ND           2.9      0.8


              NOTE:   "ND" means  only values  of  zero were  measured;
              precision was therefore not determined.
                                       DISCUSSION


STEVE KNOLLMEYER: I have two questions. One, does the X-MET 880    DONALD SMITH: As far as I know there's no internal calibration available
haveaninternalmeasurementcapability?Thatis,needingnoexternalcalibration.    with that. So, no we didn't.
And two, if so, do you compare errors using that internal calibration method
versus your matrix specific calibration standards?
                                              540

-------
                   Use  of Long-path  FTIR  Spectrometry in Conjunction with
                               Scintillometry to  Measure Gas  Fluxes.
                                            Douglas I. Moore,
                                            Clifford N. Dahm,
                                             James R. Gosz,
                              Biology Department, University of New Mexico,
                                         Albuquerque, NM, 87131
                                                  and
                                             Reginald J. Hill,
                                  NOAA/ERLAVave Propagation Laboratory
                                    325 Broadway, Boulder, CO,  80303
ABSTRACT

Fourier  Transform  Infrared  (FTIR)  spectrometry  is
rapidly becoming  a technique of choice for analyzing
volatile  hazardous  waste  emissions  (1).    We  have
developed  a field  portable system  that  is  capable of
measuring gas concentrations at up to a 1 Km pathlength.
The  advantage  of  such a system  is  that it can  analyze
samples  virtually  in  real  time  for myriad  compounds
simultaneously without introduction of any artifacts from
sample collection.  Because detection sensitivity increases
with  path length, analysis of compounds can often be made
down to  ppb levels.   While we  have concentrated on
measuring  gas  emissions from biologic  sources, we  are
also  capable of monitoring  hazardous gas emissions that
have characteristic infrared absorbance peaks  in regions
of the IR  spectrum that are not dominated by water or CC>2.
Numerous volatile organic compounds such as chlorinated
hydrocarbons, aromatic  hydrocarbons, alcohols, ketones,
esters, ethers and aldehydes fall into this category.

While gas concentrations  are of interest, emission rates
are needed to accurately evaluate waste sites.  Obtaining
such flux  rates has become the focal point of our research.
Recently, the Wave Propagation  Lab  (WPL)  at  NOAA
(Boulder  CO)   has  demonstrated   that  an optical-
scintillation instrument  can  measure  path-averaged
momentum  and   heat  fluxes.     Development   of
scintillometers  by  WPL,  which  is  currently in progress,
will   allow long-path  measurement of water flux as well.
Combining these  long-path  flux measurements with
measurements of gradients of gas concentrations using the
FTIR has the potential to  provide an estimate of flux rates
for numerous gases  simultaneously.  This technique  will
then  have application in  natural, agricultural and human
impacted  areas such as landfills and hazardous waste sites.
 INTRODUCTION

 Fourier  Transform  Infrared  (FTIR)  spectrometry  is
 rapidly  becoming  a technique of  choice  for analyzing
 volatile hazardous waste emissions. A particular advantage
 of FTIR spectrometry is that it can be configured for long-
 path analysis.  Large areas from which potential emissions
 are occurring  can be  analyzed at  one  time.  We have
 developed a  field  portable  system that is  capable of
 measuring a number of infrared absorbing trace gases over
 long paths of  up to a  kilometer (Fig. 1).  Many of these
 gases  have either natural  and/or anthropogenic  sources
 and/or sinks.    Increased  path-length  provides  several
 advantages such as increased detection level (ppb levels)
 and  long-path  averaging  of the  high  variability  in  gas
 emissions that may often be the case on a smaller scale
 (2).
Figure  1.  Schematic of an infrared radiation source and
Fourier transform infrared (FTIR)  spectrometer equipped
with  telescopes  to  allow  long-path analyses.   The  flat
mirror allows rapid manipulation of optical pathlength and
analyses  over undisturbed terrain.
                                                      541

-------
The detection of the presence or absence of a particular IR
absorbing  contaminant  is  diagnostic of  what  volatile
constituents are being released from a site.  An even more
valuable piece of  information  would be an  estimate of the
total flux  of  that contaminant through the site into the
atmosphere  over time.     Such  an  estimate requires
knowledge  of the  micrometeorological conditions of the site
during the  period  when long-path  FTIR  spectrometry
measurements are  being made.   Ideally,  the  micro-
meteorological measurements would also be path-averaged
over  a comparable path  being   sampled  by the  FTIR
spectrometer.  Scintillometry techniques, presently under
development  by atmospheric  scientists  at  the NOAA wave
propagation  laboratory  in Boulder,  Colorado,  provide a
potential method to  link path  averaged  FTIR spectroscopic
profiling and micrometeorological  characterization of the
near surface atmosphere. Such a combined data base may
make estimates of large scale contaminant fluxes possible.

The  purpose of this article is to  describe our long-path
FTIR  spectrometer instrument, to present some analyses of
various IR absorbing trace gases  measured using a  long-
path  configuration,  and  to discuss  means of  combining
micrometeorological data   with   long-path  trace  gas
concentrations to  obtain estimates  of fluxes.  In particular,
combining  of  long-path scintillometry  with  long-path
FTIR  gas concentration data shows  the promise of providing
a better tool for  quantifying gas fluxes from both natural
and polluted sources.

MATERIALS AND METHODS

FTIR Spectrometer

Our  FTIR instrument is a  Nicolet  740  optical  bench
interfaced with a Nicolet 660 work station having a storage
module capable of storing 344 Mbytes of data.  The 740
unit is capable of 0.3 cm-1  wavenumber  resolution (3,4).
The  optical  bench  is mounted  on  a wheeled cart in
conjunction with a 60 cm diameter Cassegrain  telescope.
The  infrared  source (1000-W  halogen  quartz lamp or
globar) is  mounted  at the focal point of a  second 60 cm
diameter Cassegrain telescope.   The infrared beam is
transmitted to a 60  cm square flat mirror and returned to
the receiving telescope which focuses  it on an adjustable
aperture  (Fig.  1).  The  beam  is  then recollimated  and
transmitted into  the optical bench  and  subsequently
through the interferometer to the detector.  The 740 bench
is equipped  with an Hg-Cd-Te (MCT-A)  detector.   All
sample collection  and processing is handled using Nicolet
software. Each  collected sample can be either a single  scan
or a composite of a  series of  scans.  Sampling rates are a
function of scans per sample and the pathlength of the
moveable  mirror.  Most samples discussed in  this paper
were collected at  0.5 cm-1  wavenumber resolution.  Each
scan takes about a second at this resolution.  Generally 16
scans per sample were collected although  in some cases a
1-minute sampling rate was desired which  resulted in 60
scans per sample.  Each sample yields  an interferogram
(Fig.  2) which  must  then  be processed to give  an
absorbance   spectrum   (Fig.  3)   from  which  gas
concentrations can  be calculated.   This  quantification is
performed  using  a multivariate least squares fit (LSF)
program developed  by Haaland and  Easterling  (5).  This
technique allows quantifying of multiple components whose
lines  overlap in a given  spectral  region  with  better
precision and accuracy than can be obtained from a single
peak analysis.
    7'  50  1250   1350  IHSO
                       DflTfl POINTS
Figure 2.   Example interferogram from the FTIR.   The
energy source was a quartz-halogen lamp, pathlength 407
m,  atmospheric pressure  724  torr  and  instrument
resolution  0.3 crrH.
                       WflVENUMBER
Figure 3.  The raw data of the interferogram is processed
to  give an absorbance  spectrum  that allows calculation of
gas concentrations.  This  interferogram was collected  at
Maricopa,  Arizona over a cotton  field, 10:26  hrs, 12 June
1988, pathlength 407 m.
                                                         542

-------
Optical  Scintillometer

Scintillometers  are devices that sense  the scintillation
(intensity  variations)  in propagating   electromagnetic
(EM) waves.  The refractive  index  structure parameter
Cn2 characterizes the magnitude of this  scintillation.   EM
scintillation  results  from turbulent  fluctuations  in  the
atmospheric  refractive index  which  in  turn result from
fluctuations  in  atmospheric temperature  and  humidity
fields.   The scintillometer  currently being  used is an
optical  scintillation inner-scale meter. In the experiment
reported here, it is deployed  on a horizontal  path of 150 m
at a height of 4  m. This instrument measures the variance
of the  log  intensity of  diverging Laser  light  detected
through  a 1mm-diameter hole  as well as the variance of
the logarithm  of aperture-averaged  intensity  from  a
4.4-cm    diameter,   phase    incoherent  uniformly-
illuminated source. The ratio of these variances gives the
inner  scale  of  turbulence  I0-   The   refractive index
structure  parameter Cn2 is also determined  from  this
instrument's data.    An  approximate correction  for
saturation of scintillation  of  the  Laser   variances is
computed to extend the range of validity  of the instrument.
The heat and momentum fluxes are deduced from Cn2 and I0
using Monin-Obukhov similarity relationships.   This is
the "inertial  dissipation" method of determining these
fluxes.    In   this  configuration  the  instrument  can be
considered a fluxes scintillometer.

For comparison purposes, a three-axis sonic anemometer
having  platinum  resistance-wire  thermometer  near its
center is also deployed at a  height of 4 m on a tower  near
the optical scintillation  instrument.  The sonic anemometer
measures all three fluctuating components of velocity  at a
25  Hz data rate.  The correlation of the vertical component
of velocity with  the streamwise horizontal component gives
the momentum  flux (divided by  air density).  The friction
velocity U- is the square root of the negative coefficient of
this correlation.  The correlation of the vertical component
of velocity with  the temperature from the resistance-wire
thermometer gives  the  temperature flux  (heat  flux is
temperature flux  multiplied  by the  air's heat capacity.)

RESULTS AND DISCUSSION

FTIR

Much of our initial work  has involved testing the utility of
the long-path FTIR for monitoring gases from natural as
well as  anthropogenic  origins.  Gases  that  we  most
routinely analyze include H20, COa, CH4l N20, and CO but
many other gases can  also be quantified  simultaneously or
at  some  later time.   This points  up one  of the primary
advantages of FTIR. Spectra collected can be reprocessed at
a future time to  look for oases that mav not have been of
primary interest at the time that  the original analysis was
carried out.  Below  are  examples of some uses  of our
long-path FTIR.

Gas Emissions

The FTIR was set up  to measure a 100 m path over a small
shallow lake on  the  Isleta  Indian Reservation near
Albuquerque, N.M on June 26, 1989.  A series  of samples
was taken  to establish ambient gas  concentrations.  The
bottom sediments were then disturbed by a person walking
around to force degassing of the sediments. A second series
of samples was collected  during  this period  (Fig. 4).
Emission of CH4 is readily apparent. These emissions were
in fact  point  source   emissions  and  the  measured
concentrations denote a mean concentration for the entire
100 m path, most of  which was not undergoing degassing.
This artificially induced gas emission points up the ability
of the FTIR to quantify  such emissions but the  experiment
was designed to test our ability to  measure changes in path-
averaged atmospheric concentrations and was not meant to
estimate gas fluxes.
     i,"d
     •S 2.7-
     e
     1 "^
     i"1
     « 2.1-
     5
1.9-

1.7-
U
              Undisturbed
                                    Disturbed
                                                30
           0           10           20
                          Time (min.)

Figure 4.   Field demonstration of the ability  of long-path
FTIR to detect CH4 emissions from a shallow lake in New
Mexico. Measurements were made  over the lake on June
26,  1989  before  and  after  bottom  sediments  were
disturbed along  a portion of the pathlength.
Gradient Profiling

A primary focus of our study is to develop the capability to
measure gas fluxes over a long path which should average
out the large spatial variability  encountered using small
scale techniques  such as chambers.  Micrometeorological
techniques  such  as  Eddy correlation and  Bowen Ratio
techniques have been shown to be capable of measuring heat
and water fluxes  under certain ideal conditions.  We have
explored the possibility  of  using  the  diffusivity constants
obtained from Bowen stations and eddy correlation stations
to plug into a flux gradient equation:
                                                        543

-------
                      F= k - dc/dz
      d)
where F is the mixing  ratio  flux, dc is the  difference  in a
gas concentration (in ppm) measured at two heights and dz
is the difference in those  two heights.  We experimented
with  this technique  over a tall grass  prairie site near the
Konza Prairie  Long  Term Ecological Research (LTER) site
near Manhattan, KA in July of 1989. The FTIR source and
spectrometer were mounted on the lift gate of a truck so
that  the instrument could  be raised from  near-ground
level at  1 m to a 2 m level. Samples were collected at one
height for 10 minutes then moved to the other height for a
10 minute collection  period.  This procedure was repeated
over a 3 day  period (July 25, 26, 27, 1989 - Julian days
206,  207.  208).

Our  initial  results  were encouraging as a significant
gradient could be detected for most gases.  In some cases,
such as for water vapor, this gradient agreed reasonably
well with that measured by a Bowen Ratio station in the
same vicinity (Fig. 5).  During some of our measurements
the gradient was opposite to that expected.  Presumably,
this  was  the result of the rapid changes in surface layer
conditions that were missed due  to the time required to
move  the instrument  from  one  height to  another  and
realign  it.
-   200

^   10°
3     0

2,  -100

1  -200

fc  -300
           Water Flux - Konza Prairie - July 27,1989
       -400
                  Bowen
                  FTIR
            8
14
                                                 16
                     10        12
                        Time (Hours)
Figure 5.  Water  fluxes as quantified by long-path FTIR
and a Bowen Ratio station at  the Konza prairie research
area on July 27,  1989.  The values are based  on the
concentration gradient at 2 m and 1 m heights above the
ground.  Negative  values indicate that the direction of the
flux is away from the soil and vegetation.

Figure 6 shows the concentration gradients that were
monitored for the 3 days of sampling.  There are  several
noteworthy  items:   1.  Water  gradients  were  always
negative as  should be the case under stable surface layer
conditions (concentration at  2 m  < concentration at 1 m).
                                                         2. Daytime COa gradients always showed a positive gradient
                                                         presumably reflecting the active COj absorption by plants.
                                                         3. Of even greater significance was the positive gradient
                                                         for CH4 coupled with the significant daily decreases In CH4
                                                         concentrations. A rain event on July 23 (Julian  day 204)
                                                         was  followed by a drying  period for the remainder  of the
                                                         sampling days. The sequence is likely due to a high rate of
                                                         oxidation of CH4 by methane  oxidizers in  the local soils
                                                         coupled with a decrease in atmospheric concentrations as
                                                         the general area dried. The regional source of the CH4 Is
                                                         not known.   4.  CO concentrations also demonstrated a
                                                         constant drop through the sampling period.  A shift  in the
                                                         wind away from the direction of an  interstate that ran just
                                                         north of the  site to a more southeasterly direction is  a
                                                         likely explanation for this change as  anthropogenic sources
                                                         far exceed biological sources.
                     6a
                          I6.O-



                          15.5-



                          15.O



                          14.5-
                                                                    14.0
                                                                        H2O
                                                                                           m 2m
                                                                                           01m
                                 206        207       208
                                      Julian Day (1989)
                    6b
                                                               i
                                                               §
                                                               a
                                                              6c
                                                                            C02
                                                    H 2m
                                                    Dim
                                  206       207       208

                                       Julian Day (1989)
                                                                                                2m
                                                                                             El 1m
                                                                           206        207        208
                                                                               Julian Day (1989)
                                                         544

-------
   6d


   S
   I
    g
   a
        ion
               206
                         107
                   Julian Day (1989)
Figure 6.  Concentration gradients for H2O (6a), and trace
gases- CO2 (6b), CH4 (6c) and CO (6d) measured by long-
path FTIR on the Konza Prairie research area during  the
period  July  25-27,1989.

Air Quality Monitoring

In an effort to explore the  possibility of using the FTIR to
monitor  air   quality  in   the  metropolitan   area   of
Albuquerque,  N.M., we set up the instrument to analyze
over a  path of about 500 m.  A  portion of this  path was
just over the level of automobile traffic on one of the main
traffic arteries  in  Albuquerque.   The  sampling  period
extended from rush hour through the  period of peak
fireplace smoke emission.   A  primary  purpose  of  the
experiment was  to determine  if  methyl chloride  (an
indicator of  wood smoke source)  could be  detected.
Concentrations of this gas  were found  to be too low to be
detectable even at the  500 m  pathlength due to both low
concentrations  and  relatively  poor absorption  by  this
molecule. However, examination of the spectra showed that
a number of gases  which are not normally detectable under
clean air conditions were  present in elevated  levels.
Ethylene and  methanol were easily detected while ammonia
and  formaldehyde   were   also   above   detectable
concentrations.

These  tests  were  carried out on  two nights during  the
winter of 1989-90.   The first was on Dec. 8, 1989.  This
had been declared  a   no-burn   night  by the  City's
Environmental Health Department meaning that no wood
burning  in fireplaces  or  stoves  was  permitted.   Gas
sampling began at  about  5:40  PM and  continued until
11:00 PM.  Figure  7  shows the level of gas concentrations
for  N2O,CO,CO2, CH4t H2OandC2H4  Values are means of
10 - one minute  samples.   All of the gases, with  the
exception of N2O,  increased until about 7:30 PM at which
time they began a steady decline.  The second sampling date
was Jan. 12,  1990. This night was not declared a no-burn
night by the city although  it was expected to be marginal
with respect to weather conditions which could assist  with
the  dispersal  of  gases  emitted  by  wood burning  or
automobiles.   Indeed, surface wind  conditions for both
nights were quite similar.  On Jan. 12, most of the data for
the period 7:40 to 9:15 were lost due to problems with the
FTIR  instrument (Fig. 8).   As with the Dec. 8 date,  all
gases tracked each other with the  exception of N2O which
showed an inverse pattern.  However, unlike the December
collection, the gases  showed a steady increase from 5:00
PM until  about 10:00 PM when  they  began to decline.
While the time of peak levels was different, concentration
of peak levels for all gases were generally comparable.
Table  1    gives  the maximum,  minimum,  and  mean
concentrations for all of the gases measured for the two
sampling  dates.  The sampling intervals for the two dates
were  not  the same although  total sample numbers were
similar.
   2
   ^-

   i
   o
   U
  O
       550-
       450-
       350-
       250-
        150
             No-Burn
                                      -+- N20(ppb)
                                      *• C0(ppb/10)
                                      •*• C02(ppm)
                                      -*• CH4(ppb/5)
                                      •*• C2H4(ppb*5)
                                             11   12
          5     6     7     8     910
                      Time (PM) - Dec. 8,1989
Figure 7. Trace gas concentrations measured by long-path
FTIR during a no-burn (no wood  burning allowed) night in
Albuquerque, New Mexico. Values are concentrations from
the means of  10  - 1  minute, samples on  the  night of
December 8 1989.  Pathlength was 500 m.
        550-
                NZO(ppb)
                C0(ppb/10)
                C02 (ppm)
                CH4 (ppb/5)
                C2H4 (ppb*5)
                                               Burn
    _c
    *•*
    2
    5
    o
    o
    U
    i/i
    3
    O
            5     6    7     8     91011
                        Time (PM) • Jan. 12,1990
 Figure 8.  Trace gas concentrations measured by long-path
 FTIR during a night when wood burning was allowed (Jan.
 12,  1990).  Pathlength was 500 m.
                                                       545

-------
Table 1.  Mean, maximum and minimum concentrations of
trace gases measued with long-path FTIR on Dec. 8, 1989
(No-Burn)  and  Jan. 12,  1990  (Burn)  in  Albuquerque,
NM.


Gas Units
N2O
CO
CO2
CH4
H2O
CH3OH
C2H4
NH3
ppb
ppb
ppm
ppb
ppm
ppb
ppb
ppb
Mean
No-Burn
328
3316
426
1659
2037
7.7
61.6
16.7
Maximum
Burn
314
3438
405
1678
2204
15.2
59.2
19.7
No-Burn
337
4980
472
1790
2181
31
120
36
Burn
324
5274
463
1946
2300
30
110
26
Minimum
No-Burn
317
1518
386
1522
1926
0
23
9
Burn
297
1763
353
1494
2127
3
30
14
 While the  small  sample size precludes  any  definitive
 conclusions  as   to   the  relative  contribution  of
 wood-burning and automobiles to the elevated levels of the
 sampled gases, some things are suggested by the data.
 1. The  timing of the peak CO level on the burn  night
 suggests  that wood-burning contributes  a considerable
 quantity of CO to the  city air when  permitted.
  2. Based  on mean and maximum  CO levels for the two
 nights, firewood burning contributes proportionately  more
 CH4, CHaOH and NHa  than  vehicle emissions while
 automobile exhaust shows a greater  contribution of  C02,
 N20andC2H4.
   3.  On  the  burn  night,  the  CO  concentrations  were
 extremely  erratic during the  time of maximum traffic.
 This indicates a nearby source due  to heavy vehicle traffic.
 Later in the  evening, the concentrations were less erratic
 suggesting a less proximate source which  would likely be
 the case for wood  smoke.
         0.30 -,
          0.20
          o.to
         -o.oo -
         -0.10
      «
       t
       f
    ,6'
                      4
                     J>*
                     •au
Fovorobta Wind Variability
Acceptable Wind Variability
OumlianabM Wind Variability
Eilreme Wind Variability
Poor Wind Direction
Unfavorable Cloud Cove
                                              "oja
            -010     -0.00     0.10      0.20
                      temp flux-eddy correlation
 Figure  9.   Temperature  flux  measured  by the  flux
 scintillometer compared with temperature  flux  from eddy
 correlation.  Units are °C m s-1.
                            Scintillometry

                            Comparison  of  the  temperature flux and friction velocity
                            determined  from  the  fluxes  scintillometer  with  that
                            obtained from the point sensors are shown in figures 9 and
                            10.  These data show that the fluxes scintillometer gives
                            good values even in nonideal atmospheric conditions.  Even
                            during periods of intermittent cloud cover and unfavorable
                            wind conditions, when the validity of the Monin-Obukhov
                            similarity is  limited  or unknown, measurements obtained
                            from the  fluxes scintillometer  compared well with the
                            point sensor measurements.
                                    0.70 -


                                    0.60 -


                                    0.50


                                    0.40


                                    0.30


                                    0.20 -


                                    0.10
                                     0.00
                                                                                 • rovoroble Wind Variability
                                                                                 « Acceptable Wind VoriabiMy
                                                                                 • OumtionoMe Wind Variability
                                                                                 • extreme wind Variability
                                                                                 * Poor Wind Direction
                                                                                  Unfavorable Cloud Cover
                                                                0.50  0.60  0.70
          ni  | I i I i | i  i i i | i i i i I i i i i r* '
           0.00  O.tO  0.20  0.30  0.40
                     U. - wind speed
Figure  10.    Friction  velocity measured  by  the  flux
scintillometer compared with  friction  velocity  deduced
from wind speed, surface roughness,  and eddy-correlation
temperature flux.  Units are m s-1

Integration of FTIR and Scintillometry

    Our next  step in the integration of the FTIR system with
the  long-path  measurement  of  water  and  heat fluxes
involves two steps.  The first is  the development of   2
scintillometers  by the WPL    One uses  a   10.6  p.m
wavelength  Laser  source;  the  other  uses  a 3   mm
wavelength Gunn diode source.  The  combination of thes'
two  scintillometers with the fluxes scintillometer allows
long-path  measurement  of  the  fluxes  of  latent  heal
sensible heat and momentum.  Modification of  the FTIi
system to allow  it to measure gas concentrations at  twi
heights  almost simultaneously is the second  proposed
modification. This will be accomplished through the use  of
a  periscope  system  that will eliminate the  lengthy delays
in  realignment  that  are necessary when  the entire
instrument must be moved vertically as was the case in thi
Konza Prairie experiment.

Two related methods will be used to derive gas fluxes from
FTIR and scintillometry measurements.  These are outlined
                                                            546

-------
by Andreas (6).  The first uses the Obukhov length L from
measurement  of  vertical  momentum,   u-  and  the
temperature and water vapor fluxes. These, combined with
the gas concentrations, GI  and GZ measured at the  two
heights zi and zg by the FTIR, will give an estimate of the
corresponding flux scale g-  from the relation
AG = G2. G,=(g./K)[ln
                          • ug (z2/L) - ug(Z! /L)]
(2)
as suggested by Hicks and Liss (7).  The gas flux is then
given by Fg=  -u-g-. The second method will use a modified
Bowen ratio method which  is based on the belief that ug
should be  the same  for all  conservative scalars.   In
particular, the difference in  water-vapor mixing ratio,
AQ, also satisfies equation 2
AQ = QT . Q2 =(q-/K) [In (23/2,) - uq (z2/L) -
                                                   (3)
where q- is the water vapor flux scale.  Since -u-g-= Fg
is the gas flux we are seeking and -u-q- = Fq is the water
flux, and since we assume ug=uq, Equations 2 and 3 yield:
                 Fg =   Fq (AG/AQ)
                                                  (4)
To use equation 4, we obtain AG and AQ from the FTIR and
Fq from the scintillometers.An important assumption of the
application of gradient  profiling is that the  various gases
for which fluxes are being derived behave similarly to heat
and  water vapor in  the atmosphere.  The scintillometry
system measures path averaged fluxes of heat, momentum
and  humidity  with  scintillometers  measuring  at 1  mm,
10|im, and 1u,m.  Gaseous flux estimates are then based on
these measurements of atmospheric conditions.  All gases
might not have identical flux profile  relationships that lead
to equation 2; that is it may be that ug*uq  (8).  It will be
critical  in future  work to  determine  the  flux-profile
relationships for  trace gases.

CONCLUSIONS

Our  research to date with long-path FTIR spectroscopy and
path-averaged scintillometry has resulted  in the  following
conclusions.

1)  Long-path  FTIR  has  the  analytical  sensitivity to
measure  numerous  atmospheric trace  gases   both
anthropogenically  and  naturally  derived at  ambient
concentrations.

2)  Increased pathlength commonly adds  to the sensitivity
with which we can measure  atmospheric trace gases.
3}  Optical scintillometry is a potential tool for measuring
path-averaged fluxes of heat, momentum and humidity in
the atmosphere  over  the  same  path  in  which  gas
concentrations are being determined by long-path FTIR.

4}  Gradient profiling  using long-path,  path-averaged
FTIR spectroscopy has been used to  show distinct vertical
structure of H2O,  CO2 and CH4 in the atmosphere over a
prairie.

5)  Coupling  gradient profiling of atmospheric trace gases
with the  long-path FTIR and path averaged  scintillometry
is a promising means to begin to estimate gas fluxes at
larger  spatial   scales   from  various  landscapes.
Applications are seen for both field screening of hazardous
waste and toxic chemical emissions to the atmosphere and
for many global greenhouse gases.

ACKNOWLEDGEMENTS

This work was supported  by a grant  from the National
Science Foundation (BSR 8618487).   We  also wish to
acknowledge Bill  Herget  of Nicolet Instruments for  his
technical assistance in getting the FTIR instrument set up
and operational, Tom Garrison and  Greg Shore for  help
with collection and processing  of FTIR data and Yorgos
Marinakis for  help  in   collecting  and  interpreting
micrometeorological data.   We would  also thank Shashi
Verma for sharing  his micrometeorological data from the
1989  FIFE experiment at Konza Prairie.

REFERENCES

1.   Spartz,  M.L. M.R.  Witkowski, J.H. Fateley,  J.M.
Jarvis, J.S. White,  J.V. Paukstelis, R.M. Hammaker, W.G.
Fateley, R.E. Carter,  Jr., M. Thomas, D.D. Lane,  G.A.
Marotz,  B.J.  Fairless,  T. Holloway, J.L. Hudson,  D.F.
Gurka.  Evaluation of a mobile FT-IR system for rapid
volatile  organic  compound  determination,  Part  I:
Preliminary  qualitative   and  quantitative  calibration
results.  Amer. Environ. Lab.  1  (2) Nov. 1989,  pp.  15-
30.

2.   Robertson, G.P., M.A. Huston,  F.C.  Evans and J.M.
Tiedje.    Spatial  variability in a  successional  plant
community: patterns of nitrogen availability. Ecology 6 9
(5)  1988, pp.  1517-1524.

3.  Gosz, J.R., C.N. Dahm and P.G. Risser. Long-path FTIR
measurement of  atmospheric  trace gas concentrations.
Ecology  69(5)  1988,  pp.  1326-1330.
                                                       547

-------
 4.  Gosz,  J.R., C.N. Dahm,  D. I. Moore and  S. Hofstadler.
 Field  testing  Long-path   Fourier  Transform  Infrared
 (FTIR) Spectroscopy for measurements of atmospheric gas
 concentrations. Remote Sens. Environ. 32  1990,  pp.  103-
 110.

 5.  Haaland, D.M.  and R.G.  Easterling.  Application  of new
 least  squares  methods  for  the  quantitative infrared
 analysis of multicomponent samples.  Applied Spectroscopy
 36(6)  1982,   pp.  665-673.

 6.  Andreas,E.L Can long-path FTIR Spectroscopy yield gas
 flux  measurements  through a variance technique?  Atmos.
 Environ,  (submitted),  1990.

 7.  Hicks. B.B. and  P.S.  Liss.   Transfer of SC>2 and other
 reactive  gases across  the  air-sea interface.  Tellus  46,
 1976,   pp. 348-354.

 8.  Hill,  R.J.   Implications  of  Monin-Obukhov  similarity
 theory for scalar quantities.   J.  Atmos. Sci.  46, 1989,  pp.
 2236-2244.
                                                       DISCUSSION
JOHN EVANS:  I was just curious how you calibrate the instrument and
secondarily, what sort of precision and accuracy of the path length average
concentration you get for something, like methane, for example.

DOUG MOORE: We calibrated FTIR using white cell, .25 meter white cell in
the laboratory. And for things like methane we get detection limits of about 30
ppb, plus or minus.

JOHN EVANS: What sort of precision and accuracy can you get on normal
measurements say, in the atmosphere?

DOUG MOORE: Well, yes, that's going to be path dependent. I'm not sure I
understand. What numbers do you want it in?

JOHN EVANS: Well, we see a graph up there with some numbers up and down.
Are they 1%,  10%, 50% accuracy precision?

DOUG MOORE: We're better than 5% accuracy. Probably better than that on
the long path.

TOM PRITCHETT: When you were actually measuring this flux, do you have
to shine the beam directly over the source, or do you shine the beam downwind
of the source in looking at any downwind transport?
CLIFF DAHM: Our appl ications are quite a bit d ifferent than many other people
who are looking at point source. We're not looking at point source emission.
We're looking at something that's broadly distributed across the environment.
So, what we're looking at is something where we really need to know something
about fetch length from which the sources are generated. But we're not looking
at a point source. If we are looking at a point source, we would go into point
source analysis, we would have to be downwind and perpendicular to that point
source, or over that area of point source. What  we're looking at,  though, is
landscape emissions of things that tend to be distributed rather, at least reasonably,
uniformly over the environmenl.

TOM PRITCHETT: So, basically you're looking at the flux as coming directly
underneath your beam, essentially?

CLIFF DAHM: That's a very difficult question as to exactly where those gases
are coming. They're coming from downwind. It's very dependent, of course, on
wind field conditions at the time you're making the measurements. But you can
be generating input terms to your vertical structure of the atmosphere that can be
anywhere up to 100 times the height you are above the ground, 100 times upwind
of that direction. Again, it depends very much on meteorological conditions at the
time of the emission.
                                                                  548

-------
                      PATTERN RECOGNITION METHODS FOR FTIR REMOTE SENSING
                 Gary W. Small*

            Department of Chemistry
               University of Iowa
              Iowa City, IA  52242
             Robert T. Krout11

        U. S. Army Chemical Research,
     Development, and Engineering Center
      Aberdeen Proving Ground, MD  21010
ABSTRACT

Digital filtering and pattern recognition meth-
ods are described that Implement an automated
detection algorithm for passive Fourier trans-
form Infrared (FTIR) remote sensing data.  The
detection 1s performed with only a 76-po1nt
segment of the FTIR Interferogram, thereby
enabling a "short-scan" Interferometer to be
used.  Two novel pattern recognition methods are
Introduced that provide for the Intelligent
selection of training set patterns and for the
calculation of collectively optimized plecewlse
linear discriminants.  This methodology Is
evaluated with a large quantity of passive FTIR
remote sensing data and 1s shown to perform 1n a
highly effective manner.

INTRODUCTION

Fourier transform Infrared (FTIR) remote sensors
are environmental monitoring devices that employ
an Interferometer-based optical system to col-
lect Infrared spectral data 1n the outdoor
environment.  The resulting data can be analyzed
for the presence of characteristic spectral
bands corresponding to target analytes.

Infrared remote sensors can be operated 1n two
ways.  The spectrometer can be used with an
external blackbody Infrared source, or the
sensor can be employed 1n a passive mode simply
to collect whatever ambient Infrared background
radiation 1s present 1n the field of view.  The
passive technique 1s the more flexible of the
two Implementations, as the sensor consists of a
single unit.

Specific application environments for passive

'Corresponding author.
FTIR sensors Include monitoring at hazardous
waste sites, leak detection at chemical plants,
and regulatory monitoring of smokestack emis-
sions.  In these applications, the spectrometer
can be positioned 1n a stationary configuration
or mounted 1n a ground or airborne vehicle.

Two fundamental problems limit the applicability
of passive FTIR sensors In demanding monitoring
applications.  First, the sensor must be rugged
enough to operate under the conditions required
for the application.  For example, 1n an air-
borne Implementation, the spectrometer must be
stable enough to allow data collection under
conditions of moderate vibrations or varying G-
forces.  Second, 1n the passive FTIR experiment,
no stable Infrared spectral background exists
for use 1n processing the collected data.
Standard laboratory spectral processing tech-
niques that employ a background or reference
spectrum cannot be used.

The most fragile component of a typical FTIR
remote sensor Is the Interferometer drive mecha-
nism, which must allow the collection of a
stable Interferogram of 1024 or 2048 points.
The required Interferogram length 1s dictated by
the spectral resolution required to detect the
target analyte(s) of Interest.  This relation-
ship between Interferogram length and spectral
resolution derives from an Inherent characteris-
tic of the Fast Fourier Transform (FFT), the
data processing tool used to extract Infrared
spectra from the collected Interferograms.  The
FFT assumes that the Interferogram 1s an Infi-
nitely long waveform that contains zeros for all
points not explicitly collected.  This has the
effect of adding s1n(x)/x components to the
computed spectrum, resulting 1n spectral broad-
ening.  As the number of collected Interferogram
points Is Increased, the s1n(x)/x contribution
                                                  549

-------
1s decreased.  Correspondingly, spectra computed
from very short Interferograms are severely
distorted due to these effects.

One approach to Increasing the potential rugged-
ness of an FTIR remote sensor Is to adopt a
simplified "short-scan" Interferometer design.
The drive mechanism for such a system would
allow only the collection of a 100-200 point
Interferogram segment.  Conceptually, this
system would be much more rugged than a conven-
tional design, as the moving mirror of the
Interferometer would need to maintain optical
alignment for only a very short distance.  The
drawback to such a system 1s that a conventional
spectral-based analysis cannot be performed, due
to the characteristics of the FFT noted above.

Recently, we have Introduced an alternative FTIR
data processing strategy that 1s compatible with
short Interferograms (1).  The approach used 1s
based on the application of bandpass digital
filters directly to short Interferogram seg-
ments.  If the filter bandpass 1s chosen to
coincide with the frequencies of the spectral
band(s) of a target analyte, the application of
the filter has the effect of extracting specific
spectral Information directly from the Interfer-
ogram segment.  This approach overcomes the
limitations of the FFT, but still allows the
data analysis to be based on selected Infrared
frequencies.  Additionally, judicious choice of
the Interferogram segment allows the analysis to
be performed without the use of data describing
the Infrared background.

The principal drawback to this scheme 1s the
difficulty 1n Interpreting the filtered Inter-
ferogram data.  Virtually every application of
FTIR remote sensing requires that the collected
data be Interpreted automatically and a decision
made as to the presence or absence of the target
analyte(s).  In application scenarios such as
leak detection, a positive decision regarding
the presence of the analyte 1s used to trigger
an alarm.  Clearly, 1n such cases, the decision-
making aspect of the analysis 1s critical.

In the work presented here, pattern recognition
techniques are described that allow the Imple-
mentation of an effective decision-making algo-
rithm for use 1n analyzing filtered Interfero-
gram segments.  The utility of this methodology
1s demonstrated through the use of a large
quantity of passive FTIR remote sensing data.

EXPERIMENTAL

The FTIR remote sensing data used for this
research were collected with a passive FTIR
sensor built by M1dac Corp. (Costa Mesa, CA) to
the specifications of the U.S. Army Chemical
Research, Development, and Engineering Center,
Edgewood, MD.  The spectrometer design 1s based
on a linear-drive Mlchelson Interferometer
coupled with a 11qu1d-n1trogen-cooled Hg:Cd:Te
detector that responds 1n the range of 8-12 urn.
The collected data consisted of I024-po1nt
Interferograms, with a corresponding spectral
resolution of approximately 4 cm~l.  The data
collection was performed with the Instrument
placed on a tripod.  Under a variety of Infrared
background conditions, a test analyte, SF§
(Matheson Gas Products, Secaucus, NJ), was
released In the field of view of the spectrome-
ter.  SF5 was selected as a target because of
Its use as a standard test compound 1n pollution
monitoring.  It has a single strong absorption
at 940 crrfl.  Due to the great variety of Infra-
red backgrounds observed, the collected data
contained both SFs absorption and emission
bands.

The data analysis described here was performed
by use of software written 1n FORTRAN-77 and
assembly language.  The design of digital fil-
ters and the selection of the pattern recogni-
tion training set were performed on a Prime 9955
computer system operating 1n the Gerard P. Weeg
Computing Center at the University of Iowa.  The
pattern recognition analysis was performed on  a
Hewlett-Packard Vectra RS/20c, a 20-MHz 80386
IBM PC-compatible microcomputer with 4 Mb RAM
(Hewlett Packard, Inc., Sunnyvale, CA).  The MS-
DOS 3.3 operating system was used.  The compil-
ers, assembler, and operating system used with
the Hewlett-Packard system were manufactured by
Microsoft, Inc. (Redmond, WA).  This software
was executed under the Desqv1ew-386 multi-task-
ing environment (Quarterdeck Office Systems,
Santa Monica, CA).

OVERVIEW OF INTERFEROGRAM-BASED ANALYSIS

Figure 1 displays the action of a bandpass
digital filter 1n the spectral domain.  A sin-
gle-beam spectrum Is displayed with an absorp-
tion band at 940 cm~l from the analyte, SF5.
The Interferogram producing this spectrum was
collected with the remote sensor positioned on
top of a building looking down at a ground
source of SF5, approximately 180 feet away.
Superimposed on the spectrum 1s a Gaussian-
shaped frequency response function of a digital
filter.  The frequency response has a width at
half maximum of 33.0 cm"*-, and 1s centered on
the SFs absorption band.  This filter can be
applied 1n the spectral domain by multiplying
the frequency response function by the single-
beam spectrum.  The resulting filtered spectrum
Is zeroed outside of the filter bandpass, and
the SFs absorption 1s superimposed on the filter
bandpass function.

The same filtering procedure can be performed  1n
the Interferogram domain.  Here, the correspond-
                                                   550

-------
Ing operation 1s the convolution of the Inter-
ferogram and the time-domain representation of
the frequency response function.
                     100  1000  noo
                     FUEQUENCr (em")
Figure 1.  Single-beam spectrum (solid line)
with filter frequency response (dashed line)
superimposed.

In mathematical terms,
          H(f)X(f)  <->  /h(k)x(t-k) dk
(1)
where H(f)X(f) 1s the product of the frequency
response function, H, and the single beam spec-
trum, X.  The Fourier transform pair of H(f)X(f)
Is the convolution of the raw Interferogram, x,
and the 1nterferogram-doma1n representation of
the filter bandpass, h (termed the Impulse
response of the filter).  H and X are functions
of frequency, f, while h and x are functions of
the time variables, t and k.

In the Interferogram, the filtering operation
suppresses those sinusoidal signals whose fre-
quencies^ He outside of the filter bandpass.
The flittered 1nterferogranv1s thereby reduced to
two features:  (1) the Interferogram representa-
tion of the Gaussian frequency response function
and (2) the corresponding representation of the
analyte band.  As the Gaussian feature 1s wider
than the absorption band. Its Interferogram
representation damps at a faster rate.  Thus,
beyond the point 1n the filtered Interferogram
where the representation of the Gaussian feature
has damped to zero, the dominant Information 1s
a sinusoidal signal whose amplitude 1s related
to the height of the analyte absorption band.
        Figures 2 and 3 Illustrate these concepts.
        Figure 2 depicts points 160-235 (relative to the
        centerburst) 1n two unfUtered Interferograms.
        The lower Interferogram corresponds to the
        single-beam spectrum In Figure 1.  The upper
        Interferogram was collected during the same
        experiment, but SFg was not present 1n the field
                                 200    210
                            INTERFEROGRAM POINT
Figure 2.  Segments (points 160-235) of two
Interferograms collected by the remote sen-
sor.  SF5 was present when the lower Inter-
ferogram was collected.
                           INTERFEROGRAU POINT
        Figure 3.   Interferogram segments from Figure
        2 after application of the bandpass filter.
        The SFs Information 1s now clearly seen 1n
        the lower plot.
                                                  551

-------
of view of the spectrometer.  No discernible
difference can be seen 1n the two Interferograms
to Indicate the presence of SF5 Information In
the lower plot.  Figure 3 displays the same
Interferogram segments after application of an
1nterferogram-doma1n digital filter developed to
approximate the frequency response 1n Figure 1.
By suppressing frequencies other than those
associated with the targeted spectral band, the
filtering operation produces a signal that can
be used to detect the presence of the analyte.
The 76-po1nt sinusoidal signals 1n Figure 3 form
the test data used In this work 1n the develop-
ment of an automated detection scheme for pas-
sive FTIR remote sensors.

DIGITAL FILTER DESIGN STRATEGIES

The design of a practical digital filter for use
1n the manner described above requires that the
h(k) values 1n eq. 1 be generated such that the
convolution Integral can be approximated accu-
rately.  Additionally, the approximation must be
truncated to a finite number of terms.  The most
common approach to this approximation takes the
form of
       *'t
hOxt + hlxt-l
nkxt-k
           (2)
where x'^ 1s point t 1n the filtered Interfero-
gram, the h(< are as defined above, and the xt_^
are points In the raw (I.e. unflltered) Inter-
ferogram.  Since the Impulse response function
has been truncated to a finite number of terms,
filters of this type are termed finite Impulse
response (FIR) digital filters.

The most widely used approach to the generation
of FIR filter coefficients was developed by
McClellan and Parks (2).  In this method,  the
Remez-Exchange algorithm 1s used to generate a
series approximation to the frequency response
function of the filter.  The h|< are then comput-
ed directly from this series approximation.

Recently, we have Introduced a design strategy
for FIR filters based on regression analysis
(3).  This approach 1s based on two considera-
tions.  First, some of the terms 1n eq. 2  are
undoubtedly more significant than others 1n
obtaining a good approximation to the convolu-
tion Integral.  Therefore, 1t may be possible to
delete some terms without a significant loss In
filter performance.  Given that eq. 2 1s a
linear model, standard regression analysis
techniques can be used to assess the signifi-
cance of the Individual terms.  In this computa-
tion, a set of Interferograms collected by the
remote sensor Is used to build the regression
models.

Second, a better approximation to the convolu-
tion Integral may be obtained by utilizing a
different set of filter coefficients for each
Interferogram point.  For example, Interferogram
points 160 and 161 would be filtered with dif-
ferent filter coefficients.  Analogously, the
filter for use 1n application to a I00-po1nt
Interferogram segment would contain 100 sets of
filter coefficients.

We have termed these filters FIR matrix (FIRM)
filters, as the filter format defines a two-
dimensional matrix of filter coefficients.  By
tailoring each filter to an Individual point,
smaller sets of filter coefficients can be used
at each point, thereby saving computation time.
In testing, FIRM filters outperformed conven-
tional FIR filters with twice the number of
coefficients.

For the work reported here, a FIRM filter was
generated for points 160-235 based on a set of
2429 Interferograms collected by the FTIR remote
sensor.  Both SFs-conta1n1ng (429) and non-SF5
Interferograms (2000) were present.  The filter
was generated to approximate the frequency
response shown 1n Figure 1.  The region of
points 160-235 was selected as the point at
which the Information due to the Gaussian fre-
quency response function has effectively damped
to zero.  Across this point range, the Gaussian
signal decreases from 0.2/8 to 0.000556 of Its
maximum value.

In the filter calculation, a stepwlse multiple
linear regression procedure was used to select
statistically significant terms from the region
of t-0 to t-100 1n eq. 2.  To be Included 1n the
final model, terms had to meet a significance
level of 99.99%, based on the £ distribution.
The resulting filter averaged 37 coefficients
per point, while the average value of R2 for the
regression calculations was 93,2%.  This filter
was used 1n the generation of Figure 3.

INTERFEROGRAM ANALYSIS BY PATTERN RECOGNITION

The filtered Interferogram segments 1n Figure 3
are easily differentiated.  However, as the
analyte band decreases 1n Intensity to the limit
of detection, the corresponding filtered Inter-
ferogram segments are Indistinguishable from
those arising solely from background noise.  If
a decision 1s to be made regarding the presence
of the analyte, a procedure must be devised for
distinguishing Interferogram segments exhibiting
weak analyte signals from those exhibiting only
noise.

Pattern recognition techniques are numerical
algorithms for use 1n classifying data objects
("patterns") Into categories or classes.  These
methods have been used 1n a variety of applica-
tions In chemistry (4-6).  In the present exam-
ple, Interferograms belong to one of two d1s-
                                                   552

-------
tlnct categories:  (1) SF6-act1ve or (2) SFs-
1nact1ve.  The patterns in this case are the 76-
polnt filtered Interferogram segments.   These
Interferogram segments can be considered as
points 1n a 76-d1mens1onal vector space.  If the
points corresponding to the given categories
cluster together In the data space,  pattern
recognition techniques can be used to assign
unclassified points to the appropriate  catego-
ries.

Two Issues are paramount 1n developing  a suc-
cessful pattern recognition analysis scheme.
First, a representative set of example  data must
be obtained for use 1n developing the data
classification algorithm.   This "training set"
of data must» to the degree possible, encompass
the range of patterns to which the analysis will
be exposed.  Second, the appropriate pattern
recognition algorithm must be selected.  A
knowledge of the data space must be gained 1n
order to make the proper selection.   Both of
these Issues are addressed below for the case of
the filtered Interferogram data.

OPTIMAL SELECTION OF TRAINING SET MEMBERS

When Initially forming a training set,  1t 1s
desirable to select a variety of patterns from a
pool of candidate patterns that 1s as large as
possible.  A training set comprised of  a large
number of patterns 1s not necessarily the same
as one comprised of a large variety of  patterns,
however.   Many of the available patterns may
effectively be duplicates.  The Intelligent
selection of training set members Involves
maximizing pattern diversity while minimizing
the Inclusion of duplicate patterns.

This type of optimized selection of training set
members becomes Increasingly complicated as the
number of candidate patterns Increases  and as
the dimensionality of the patterns themselves
Increases.  In the present example,  31  different
data sets were collected,  consisting of approxi-
mately 14,700 interferograms.  Standard tech-
niques for deducing pattern similarity  such as
the calculation of all palrwlse distances be-
tween patterns are computationally cumbersome
with data sets of this size.  To address this
problem,  we have developed an algorithm which
provides an automated way to select optimal
training sets which have the same characteris-
tics as the starting pool  of candidate patterns.

The intelligent selection of patterns for use 1n
the training set requires some type of  distance
calculation to quantify the relationships among
the data points.  Performing this calculation
with the 76-dimens1onal data 1s undesirable,
however,  due to the high computational  cost.
The selection process can be greatly simplified
by reducing the dimensionality of the patterns.
This can be accomplished through the use of
principal components analysis (7,8).  For n-
dlmenslonal patterns, an optimal ^.-dimensional
representation can be formed simply from the
projections of the n-d1mens1onal patterns onto
the first £ principal components.  To Insure the
accuracy of any subsequent interpoint distance
calculations, £ 1s typically chosen to span a
large fraction (e.g. 9556) of the data variance.

Patterns for the training sets were chosen by
use of an algorithm which divides a £-dimension-
al principal components space Into smaller .p_-
dlmenslonal volumes.  One pattern 1s then se-
lected from each of the smaller volumes, thus
providing equitable, global  sampling of all
patterns 1n the principal components space.  For
ease of conceptualization, this global sampling
strategy 1s Illustrated in three dimensions 1n
Figure 4.  The total number of smaller volumes,
blocks 1n this case* as well as the shape of the
blocks, 1s determined by the number of specified
divisions along each principal component.  The
number of divisions 1s termed the mesh size.
Each of the smaller volumes may contain several
patterns, as shown in the expanded view of one
of the blocks In Figure 4.  To insure that the
pattern selected from that block 1s the most
different from patterns selected from neighbor-
Ing blocks, the pattern closest to the center of
each block is chosen.  The center of the block
1n the expanded view 1s Indicated by a solid
dot.  Applying this procedure to all blocks
results In a smaller set of selected patterns
Figure 4.  Conceptual depiction of the divi-
sion of a three-dimensional principal compo-
nents space Into smaller three-dimensional
volumes.  Patterns are selected from each of
the smaller volumes.
                                                  553

-------
which preserves the overall distribution of
patterns 1n the full data set.  For the filtered
Interferogram data, the spread of patterns for
the two data classes 1s different.  For this
reason, the principal components analysis and
pattern selection were performed Individually
for the two data classes.

After the number of blocks and their dimensions
are defined, each block 1s assigned a unique
number from 1 to the total number of blocks, N^.
M  is defined as
                                              (3)
where tru 1s the selected mesh size for the Jtn
principal component and a Is the number of
principal components being used.  The relative
location of each pattern Is thus defined by
determining Its block number.  Since the block
numbers are positive Integers, the real princi-
pal components coordinate values of each pattern
are also converted to positive whole numbers.
Coordinates along the J^h principal component
are transformed as
mJC(cj -
                                  - Cj,m1n)]  (4)
where cj' 1s the transformed coordinate,  cj 1s
the original coordinate, mj 1s the mesh size
defined above, and cj,m-|n and Cj>max are the
minimum and maximum coordinates along the jtn
principal component.  The computed Cj1  values
are then rounded to the nearest Integers, there-
by creating a new set of coordinates for each
pattern.  The transformed Integer coordinates
are designated as Cj".  The block number, B, of
any pattern can then be computed directly as
B =
             C(C1" ~
         1=2
                                 1-1

                                 IT
                                 j=l
                                      (5)
where c\n 1s the Integer coordinate value of the
pattern along the first principal component, cj"
1s the Integer coordinate value along the 1^n
principal component, and $. and mj are as defined
above.  After computing the block number for a
given pattern, the center of the block and the
distance to the pattern are computed using the
cj1.  This distance 1s used later to select the
pattern which is closest to the center of the
block.  The patterns are then sorted by Increas-
ing block number, and the total number of occu-
pied blocks 1s calculated.  The final set of
optimum patterns 1s selected by sampling each of
the occupied blocks.
                                              This procedure was used  to reduce the  set of
                                              14,700 Interferograms to 4000.   Of the 4000
                                              filtered Interferogram segments  chosen to form
                                              the training set,  2000 contained SF5 signals,
                                              while 2000 contained no  SFs  Information.  Six
                                              principal  components were used  1n the  selection
                                              of the non-SF6 patterns,  while three principal
                                              components were used In  the  selection  of the
                                              SFg-act1ve patterns.   This training set was used
                                              1n the development of a  pattern  recognition
                                              scheme for the automated detection of  SFg sig-
                                              nals 1n the filtered Interferogram data.

                                              PIECEWISE  LINEAR DISCRIMINANT TECHNIQUES FOR THE
                                              AUTOMATED  DETECTION OF SF6

                                              The selection of an appropriate  pattern recogni-
                                              tion technique for the SFs detection problem 1s
                                              keyed by an Investigation of the manner 1n which
                                              the SF5 and non-SF5 data classes cluster 1n the
                                              76-d1mens1onal  space.  Principal components
                                              analysis can be used to  explore  these  relation-
                                              ships visually.   Figure  5 1s a plot of the
                                              projections of the 4000  training set patterns
                                              onto the first three principal components of the
                                              data.   SFs-actlve Interferogram  segments are
                                              Indicated  by open circles, while non-SF5 Inter-
                                              ferograms  are Indicated  by solid triangles.  All
                                              of the non-SFs points are clustered at the
                                              center of  the plot.  To  provide  a better view of
                                              the Interface between the data classes, Figure 6
                                              1s an expanded view of the boxed region 1n
                                              Figure 5.   It 1s clear from  an  Inspection of
                                              Figure 6 that the data classes merge at the
                                              limit of detection of
                                                                    PC2
                                                 PCI
                                             Figure 5.  Principal  components plot depict-
                                             ing the relationships among the 4000 patterns
                                             1n the training set.   SF5~act1ve patterns are
                                             depicted as open circles.   The non-SFs pat-
                                             terns (solid triangles) are all located 1n
                                             the boxed region.
                                                    554

-------
Figure 6.  Expanded view of the boxed region
1n Figure 5.  The SFg-act1ve patterns are
again Indicated as open circles, while the
non-SF5 patterns are Indicated as solid
triangles.

Consideration of the class distributions 1n
Figures 5 and 6 suggests that plecewlse linear
discriminant analysis (9,10) 1s the pattern
recognition method of choice for the filtered
Interferogram data.  This technique 1s based on
the construction of boundaries or separating
surfaces between the data classes.   The separat-
ing surfaces are termed discriminants, as they
define boundaries 1n the data space that allow
the classes to be discriminated.  The plecewlse
linear discriminant consists of multiple linear
surfaces which collectively form a plecewlse
approximation of a nonlinear separating surface.
The need for a nonlinear discriminant Is clearly
motivated 1n Figures 5 and 6 by the circular
distribution of the SF5~act1ve points around the
        points.
Each linear surface comprising the plecewlse
linear discriminant 1s defined by the locus of
points orthogonal to an n-d1mens1onal vector
termed a weight vector or discriminant,  where n
1s the dimensionality of the pattern data (76 in
the present example).  Each weight vector,  w, is
calculated such that
                    wx
(6)

(7)
where xa represents a SFs-act1ve pattern,  and xn
represents a non-SFg pattern.   The dot products
1n eqs. 6 and 7 are termed discriminant scores.
The Individual weight vectors comprising the
plecewlse linear discriminant are calculated
sequentially, with each discriminant separating
a portion of the patterns 1n the training set.

The algorithm used for this work calculates
discriminants that have a pure-class subset on
one side of the discriminant and a mixture of
the two classes on the other side.  A discrimi-
nant of this type is termed "single-sided".
After a single-sided discriminant has been
calculated, those patterns on the pure-class or
single-side of the discriminant are removed from
the calculation, and another discriminant 1s
computed In the same manner.  The result 1s a
set of discriminants 1n which each discriminant
separates a different pure-class subset.  Col-
lectively, the set of discriminants defines a
separating surface.

To classify an unknown pattern, each discrimi-
nant 1s applied to the pattern, producing one of
two possible results:  (1) the pattern falls on
the pure-class side of the discriminant; or (2)
the pattern falls on the mixed-class side of the
discriminant.  The first discriminant which
classifies the pattern onto the pure-class side
determines the class of the pattern.  The last
discriminant determines the class If the unknown
pattern is never classified on the pure-class
side.

A multi-step procedure was devised to calculate
and optimize the set of discriminants comprising
the plecewlse linear separating surface.  The
keys to this algorithm are. the use of Simplex
optimization techniques (11) to position each
discriminant and a novel discriminant recalcula-
tion procedure to perform a collective optimiza-
tion of the discriminants.

The Simplex algorithm computes a new weight
vector w by moving the previous w In an optimal
direction in the data space.  The algorithm
consists of a set of rules which governs this
movement based on a numerical response function
that reflects the performance of the weight
vector.  For the current work, many variations
of response functions were Implemented and
evaluated.  To be effective, the response func-
tion must encode several characteristics related
to the performance of each weight vector, In-
cluding the number of patterns separated, and
whether the discriminant Is single-sided.  In
addition, the response function should define a
continuous surface along which the optimization
can travel.  The response function,  R, used 1n
this work 1s defined as
                                              (8)

                                              (9)
                       R =  Cl.O  - S/f]  S

                         S  =  (Ns/Nt)a  Ns
                                                  555

-------
                f = 2 [1og(Ns)+l]
(10)
tors.
S 1s termed the single-sided response, where Ns
1s the number of SFg-act1ve patterns separated,
Nt 1s the total number of patterns placed on the
single-side of the discriminant, and a 1s an
exponent that penalizes discriminants that are
not single-sided.  Appropriate values of a have
been determined empirically to be 1n the range
of 10-200, depending on the magnitude of Ns.
For a single-sided discriminant, S 1s equal to
the number of SF5~act1ve patterns separated by
the discriminant (I.e. (Ns/Nt)a = 1.0).  R 1s
made a continuous function by the use of s., the
standard deviation of the discriminant scores
for the non-SFs patterns.  A smaller standard
deviation value produces a larger (I.e. more
optimum) value of the response function.  It was
hypothesized that by minimizing the variation of
the non-SFg discriminant scores, the resultant
discriminant would be more nearly aligned with
the Interface between the two data classes.  For
the data used here, the value of the standard
deviation 1s typically on the order of 10~5, and
consequently must be scaled to reduce Its Influ-
ence on R.  The scaling factor, f, 1s used for
this purpose.  Thus, the value of R can be
Interpreted as the value of S (I.e. number of
SF6-act1ve patterns) that has been penalized
based on the degree of variation among the
discriminant scores for the non-SF5 patterns.

The Simplex optimization described above 1s an
effective technique for optimizing each weight
vector.  However, optimizing each weight vector
Individually may not produce the optimum piece-
wise linear discriminant, since the discriminant
consists of a set of weight vectors.  To address
this problem, a collective optimization algo-
rithm was developed for this study.  The proce-
dure used here 1s motivated by considering that
the calculation of the Initial set of weight
vectors 1s hierarchical 1n nature.  The calcula-
tion of each weight vector 1s Influenced by the
performance of weight vectors that have been
previously computed.  Each of these vectors 1s
computed such that 1t separates as many of the
remaining patterns as possible.  In order to
effect a collective optimization, a method must
be developed to allow subsequent weight vectors
to Influence the calculation of previous weight
vectors.

The recalculation 1s performed Identically to
the single weight vector calculation described
above,  but the data set of patterns 1s altered
to reflect the presence of other vectors 1n the
set.  Prior to recalculating a given weight
vector, those patterns classified by later
weight vectors are removed from the data set.
This simple procedure allows the earlier weight
vectors to be reposltloned based on the classi-
fication performance of the later weight vec-
         Employlng the 4000-member training set,  a  piece-
         wise linear discriminant was computed consisting
         of eight weight vectors.  The Simplex optimiza-
         tion algorithm was used to optimize each of the
         vectors, and the recalculation procedure was
         applied to optimize the set of weight vectors
         collectively.  The recalculated discriminant
         classified 3945 of the 4000 patterns correctly
         (98.655).

         To evaluate the prediction performance of  the
         discriminants, two data sets were employed that
         were not represented among the 4000 Interfero-
         grams 1n the original data set.  The two data
         sets each contained 1000 Interferograms.   The
         application of a plecewlse linear discriminant
         to a set of unknown patterns Is performed  by
         computing the discriminant score for each  pat-
         tern.  In a graphical representation,  the  re-
         sults can be displayed as a plot of the  discrim-
         inant scores vs. pattern number.   Since  multiple
         weight vectors are used, there are multiple
         discriminant scores that could be plotted.   For
         the purposes of this analysis, the largest
         discriminant scores obtained by applying all
         weight vectors to each pattern were used.   For
         SF6~act1ve patterns, the signal 1s then  maxi-
         mized, and for non-SFs patterns,  the plotted
         values then reflect the distance from the  pat-
         tern to the nonlinear separating surface.

         Figures 7 and 8 show the resulting plots of
         discriminant scores for the two prediction data
         sets.  The discriminant scores greater than zero
         In the plots correspond to detections of SF6.
         An Inspection of the figures and plots of  trans-
         formed spectra Indicate that the detections are
         highly accurate.  The rate of false alarms 1s
         less than 1/6.  These results suggest that  the
         combination of an Intelligent training set
         selection algorithm along with the calculation
         of an optimized plecewlse linear discriminant
         produces a sensitive, effective detection  scheme
         for passive FTIR data.

         CONCLUSION

         The results presented here confirm that  a  short
         Interferogram segment can be used for the  reli-
         able detection of target analytes from passive
         FTIR data.  The combination of digital filtering
         and pattern recognition techniques allows  this
         detection algorithm to be Implemented.  This
         achievement makes possible the design of a new
         generation of passive FTIR sensors based on the
         "short-scan" Interferometer concept.

         These results also Introduce two new general-
         purpose algorithms for use 1n pattern recogni-
         tion analyses.  The training set selection
         algorithm described here can be used to  select
                                                   556

-------
training sets for use with any pattern  recogni-
tion method.  Results obtained 1n testing this
algorithm Indicate clearly that the method
outperforms pattern selection strategies based
on random sampling of a pool of candidate pat-
terns.
                    W^*^I|M4
        100   200   300  400   500   600   700   100   000   1000
                    INTERFEROGRAM NUMBER
Figure 7.  Plot of discriminant scores for the
first prediction data set.  None of these 1nter-
ferograms were Included 1n the calculation of
the plecewlse linear discriminant.
        100   200   JDO   400   500   (.00   700   BOO   BOO   1000
                    INTERFEROGRAM NUMBER
Figure 8.  Plot of discriminant scores for the
second prediction data set.  None of these
Interferograms were Included 1n the calculation
of the plecewlse linear discriminant.
The multi-step procedure described above  for
optimizing the placement of plecewlse  linear
discriminants Is also a general approach  that  is
not limited to the remote sensing application
used here.  The techniques developed 1n this
work are applicable to any pattern recognition
problem 1n which the Interface between the data
classes 1s complex.  The optimized discriminants
are particularly suited to problems 1n which 1t
1s Important that the discriminants define the
limit of detection of a species.

Work 1s continuing 1n our laboratory on the
overall problem of collective optimization of
the weight vectors comprising a plecewlse linear
discriminant.  We are currently exploring the
possibility of operating the simplex optimiza-
tion with a response function based on the
performance of all weight vectors simultaneous-
ly.

REFERENCES

 (1) Small, G. W., Kroutil, R. T., Ditillo, J.
     T., Loerop,  W. R.,  "Detection of Atmospher-
     ic Pollutants by Direct Analysis of Passive
     Fourier Transform Infrared Interferograms",
     Analytical  Chemistry. 6Q» 1988,  264.

 (2) McClellan,  J. H.,  Parks, T. W.,  "A Unified
     Approach to the Design of Optimum FIR
     Linear-Phase Digital Filters", IEEE Trans-
     actions no Circuit Theory. CT-20, 1973,
     697.

 (3) Small, G. W., Harms, A. C., Kroutll, R. T.,
     Ditlllo, J.  T., Loerop, W. R.,"Design of
     Optimized Finite Impulse Response Digital
     Filters for Use with Passive Fourier Trans-
     form Infrared Interferograms", Analytical
     Chemistry.  £2, 1990, 1768.

 (4) Jurs, P. C., "Pattern Recognition Used to
     Investigate Multivarlate Data 1n Analytical
     Chemistry",  Science. 232, 1986,  1219.

 (5) Derde, M. P., Massart, D. L., "Supervised
     Pattern Recognition:  the Ideal  Method?",
     Analytica Chimica  Ac±a., 121,  1986, 1.

 (6) Varmuza, K., "Pattern Recognition 1n Ana-
     lytical  Chemistry",  Analytica Chimica .Ada.,
     122,  1980,  227.

 (7) Retelling,  H.,"Analysis of a Complex of
     Statistical  Variables into Principal Compo-
     nents",  Journal M Educational Psychology,
     2A, 1933, 417.

 (8) Martens, H., Naes,  T., "Methods for  Cali-
     bration", Multivariate Calibration. Wiley,
     New York, 1989, p.  111.
                                                   557

-------
  (9)  Lee,  T.,  Richards,  J. A., "P1ecew1se Linear
      Classification Using Seniority Logic Com-
      mittee Methods, with Application to Remote
      Sensing",  Pattern Recognition. 1Z,  1984,
      453.

(10)  Duda, R.  0.  and Possum,  H.,  "Pattern Clas-
      sification by Itenatively Determined Linear
      and P1ecew1se Linear Discriminant  Func-
      tions",  IEEE Transactions on Electronic
      Computers. 15., 1966, 220.

(11)  Routh,  M. W.; Swartz, P. A.; Denton, M. B.,
      "Performance of the Super Modified
      Simplex1', Analytical Chemistry. 42, 1977,
      1422.
                                                    DISCUSSION
DONALD GURKA: Can you visualize a digital filter analog to the Hadamard,
to eliminate the multi channel disadvantage on the transparent spectral component?
For example, use a series of digital filters which would only let through the
channels of information that you want to transform.

GARY SMALL: It's a nice idea. The problem would come in the construction
of the filters that would have many, very narrow individual band passes. The
problem that you get in the design of filters is that the narrower you want the band
pass, the more difficult it is to actually implement the filter that will work in the
time domain. The problem that you would come into would be having to have
either many individual filters, or to have a very complex filter that would have
multiple band passes. So, I think the key question would really be going back to
the electrical engineering techniques that one uses in designing filters to see
whether that would be viable. Our experience is that if you really want very
narrow band passes, it's a difficult problem in filter design. That might be too
tough, actually.

DONALD GURKA: So, the answer is yes, but it won't be easy?

GARY SMALL: The answer is it's conceptually a nice idea. I think implementing
it would be difficult.
                                                              558

-------
                REMOTE VAPOR SENSING USING A MOBILE FTIR SENSOR
R.T.  Kroutil,   J.T.  Ditillo,  R.L.
Gross, R.J.  Combs, W.R. Loerop; U.S.
Army Chemical Research, Development
and  Engineering  Center,  Aberdeen
Proving Ground,  Maryland 21010
G.W.  Small;  Department  of  chemistry,
University  of  Iowa,   Iowa  City,   Iowa
52242
 (A) Introduction
     The remote-passive detection of
hazardous  chemical  vapors  is  an
important application  for both the
military  and  civilian communities
interested in environmental issues.
Remote  Fourier  Transform Infrared
[FTIR] Spectrometers are  capable of
detecting absorptions and emissions
of low-concentration chemical vapor
clouds using an ambient temperature
atmospheric  background.    For many
pollution-monitoring  problems FTIR
spectroscopy  represents  the  only
viable approach for the detection of
many environmental pollutants.

      Remote     FTIR    technology,
developed to  detect chemical warfare
agents,  is directly applicable for
compliance assurance for many  of the
chemicals listed  in the U.S.  Clean
Air Act.   FTIR spectrometers have
the   potential   to  monitor  stack
emissions, hazardous  components in
wood  smoke,   auto  emissions,  and
industrial releases.    By using  a
FTIR    one    can    detect   vapor
concentrations  from chemical  leaks
or  spills.   With  advanced warning
provided  by   an  FTIR,  residents
located in a  surrounding area might

be  given  enough  time   to  safely
evacuate.  In these applications an
FTIR   would   be   mounted  on   an
emergency response team helicopter
to   give   an   identification   of   a
particular chemical species.

     An  infrared  remote  chemical  sensor
consists   of   a   sensor   and   signal
processor  that operate  in parallel  to
give an  indication of  the  presence of a
pollutant.     The sensor   detects  the
signatures  of  all  chemical vapors  and
backgrounds, while the signal processing
algorithms   discriminate   between  the
spectral  features associated with  the
pollutant and background emissions.  The
typical   instrumentation  required  for
remote  infrared chemical  vapor  sensing
consists of a two wave number resolution
interferometer with a specialized  set of
collimating    optics.       The   signal
processor  detects a  chemical cloud in a
fixed-site application by measuring the
background radiant emission profile as a
function of time.  When a target cloud
moves    into    the    field   of   an
interferometer a  specific change  in the
background radiant emission  profile is
detected.

       Recently,   a   number   of   new
applications  using  a  remote  chemical
 sensor   have  been   developed   using
 interferometers    operating    from
helicopters,    aircraft,   and    earth-
 orbiting satellites.   The  remote sensing
 problem  for  these  cases  is  severely
 complicated   because    the  background
 emission  profile changes  rapidly with
 respect to  time.   When operating in a
 rapidly moving scenario,  the change  in
 emission profile versus time cannot  be
                                         559

-------
used for  detection of the presence
of a vapor cloud.   This change  in
the  radiant background  can  be  an
order of magnitude greater than the
spectral  emission  profile  of  the
vapor cloud.  In order to remove the
changing     background    spectral
features    a   signal    processing
technique  is needed.

     The   collection  of  data  from
mobile  FTIR applications is  further
complicated by  the  extreme _data-
processing requirements  in which an
interferometer may collect up to 30
two  wave  number  scans  per  second.
For    many   pollution   monitoring
applications, size, power, cost, and
weight limitations  require   that a
low-powered signal-board computer be
used for  real-time data  analysis.  A
recent development   in  the  signal
processing  hardware  area  is   the
advent   of  the   Digital   Signal
Processing (DSP)  chip.   Current DSPs
are  capable of processing up to 33
million floating point  instructions
per  second. Remote sensors operating
 in mobile  environments  can  benefit
 from   the   high    computational
 throughput of single-board computers
 using DSP technology.

 (B)  THE BACKGROUND OF THE PROBLEM

      One  can consider the radiance
 incident   on  a  remote sensor  as
 combinations  of  energy  from  the
 background, the  target vapor  cloud
 of  interest,   and  the   intervening
 atmospheric gases.  One can describe
 an  integral  equation consisting of
 infinitesimal layers  of atmosphere.
 In  this   case,   a radiance emission
 source will be absorbed in the layer
 by  both   the  target  cloud  and  the
 intervening    atmospheric
 constituents.  The radiance measured
 at  the  detector  is given  by  the
 following equation,  where
  (1)
s:
                          - KT(X)
                 - KN'(X)  }dx
where kT and  k.  are the extinction
coefficients of  the target gas  and
the atmosphere
                                          of a blackbody at the temperature on  the
                                          infinitesimal layer.  N1  is the  radiance
                                          incident  on  the  infinitesimal  layer
                                          traveling to  the sensor,  and  x is  the
                                          length variable  that is integrated  for
                                          the  length  of  the  target cloud   and
                                          intervening atmosphere.

                                               Assuming  homogeneous  atmospheric
                                          and   target    cloud    species,    the
                                          integration of  equation  (1)  gives   the
                                          power incident  on a  passive  sensor as
                                          shown in reference [1],
                                          (2)
             [TATTNBQ
                                                                  - TATT) NT
                                          where    TA     is    the    atmospheric
                                          transmittance,  TT  is  the  target cloud
                                          transmittance,  NB  is the radiance of the
                                          background,  NT  is the radiance of  a
                                          blackbody at the ambient temperature,  A
                                          is  the collector  area, and  SL  is  the
                                          solid angle  of  acceptance of the sensor.
                                          The  atmospheric cloud  transmittance  is

                                          (3)     T. = e -KAR
                                         where   R   is   the   distance  of   the
                                         intervening  atmosphere.     The  target
                                         cloud transmittance  is
                                          (4)
          = e-l
-------
1 to 20 ppm-m.   The exact detection
level   is    dependent    on    the
absorptivity of each compound.

(C)   SIGNAL  PROCESSING  OF  REMOTE
SENSING DATA FOR DETECTION AND
    ALARM

    (1) BACKGROUND

     Signal  processing  of  remote
sensing data is required to extract
background  spectral  features  from
those of a  vapor  target cloud.   It
has   been   shown   that   digital
filtering   used   in   either   the
frequency or the interferogram space
can  be used  to   extract  spectral
background features.[2,3,4]  Signal
processing  algorithms used  in  the
interferogram space have advantages
with  application  to  detection  and
alarm algorithms for remote sensors.
First,   the    signal   processing
algorithm   does   not   require   a
conversion into the frequency domain
by  a  Fast  Fourier  Transformation
(FFT).  This reduces the number of
computations    required   for    a
detection   algorithm.      It   also
reduces some resolution  degradation
caused by  the  apodization function
of  the   FFT  in   the   transformed
spectrum.    Second,  the broad band
spectral  background  features  are
somewhat separated by point number
in the interferogram  space.  In the
time domain, the  central fringe of
an  interferogram  is  at  the  zero
retardation of the moving mirror of
the   interferometer  and  contains
information   from    all   spectral
wavelengths.    For  remote  sensing
data the  central  fringe contains a
disproportionate contribution of the
broad-band   blackbody   radiation
curve.[5,6]   As  one moves further
out    into   the    ends   of   the
interferogram,   the  broad spectral
components constructively  interfere
with each other more than the narrow
band spectral features.  The effect
is   a   severe   damping  of   the
contribution of the broad spectral
components.   Signal  processing in
the  interferogram domain  can take
advantage of this  effect for remote
sensing  by  processing  only  short
interferogram   segments   located
adjacent to the center-burst of the
interferogram.   Digital filters used
away  from  the  center-burst are not
required to operate  over the entire
 16-bit   dynamic  range   of   the  data.
 Because   of   these  reasons,   digital
 filters can be used more efficiently for
 signal   detection  algorithms   in  the
 interferogram    space   than   in   the
 frequency domain.

      (2) FINITE IMPULSE RESPONSE FILTERS

     The  most   commonly  used  digital
 filter    for    signal    processing
 applications  is  known  as  the  finite-
 impulse  response   (FIR)  filter.    The
 basic  form  of  the equation  is  shown in
 the following equation.
                  N
 (3)
 U
1=1
b
where  Y  is  the  filtered  data  point
resulting  from the  application of  the
filter, X is the  input  raw data values,
and b, are the filter coefficients.

     The purpose  of an FIR  filter used
for signal detection in a remote sensing
interferogram  is   to   generate  narrow
bandpass  responses  to  eliminate  the
background information.  One of the most
widely  used techniques for  generating
coefficients  for  an FIR  filter is  the
process  known  as  the  Remez  Exchange
procedure.   In this method,  the  filter
coefficients,  b,  are generated through
the use of the attenuation approximation
theorem.    In  this  theorem,   if  the
approximation  error  of  the  frequency
response outside of a passband response
is uniformly distributed,  the resulting
narrow  bandpass  filter response  error
will be minimized.

     The Remez  Exchange procedure  must
satisfy the condition where the response
is defined as P,  where,
                  M
(4)
b  cos (wn)
                 n=o
    A weighted error function is defined
as the difference  to the true  response
from that of P where,
(5)    E (e'w) = W(e'H)  [  D  -  P(6JW)  ]
                                        561

-------
     In  this  equation  W   is   the
passband   to   stopband    weighing
function, and  D  is the actual  real
frequency response of the function.

     The alternation theorem states
that   for   any   selected   set   of
extremal    frequencies,    w,    the
alternation   condition   must    be
satisfied where,
 (6)
E(e'JH)   = -I
     The Remez procedure is based on
the fact that  an  iterative  solution
can  be developed  in a  convergence
procedure between  equations  (5)  and
(6) .   When  the correct values of  b
are generated, then  value  of E(e"j")
in the two equations will converge.

    Generating    narrow    bandpass
digital  filters  to extract  out  the
signal of interest has  been applied
to the analysis of collected remote
sensing  interferograms  that contain
a   spectral   absorptions   of   the
simulant Sulfur  Hexafluoride (SF6) .
In  this  example  a narrow  bandpass
digital filter using 40 coefficients
was  developed  that  had  a  center
frequency corresponding to  940 wave
numbers. At the modulation  frequency
of the interferometer  this  spectral
frequency corresponded to a bandpass
of 50 Hz with  a center frequency of
2450 Hz.   Figure 1 shows two short
segment   interferograms   collected
from an  interferometer  mounted on  a
UH-1 Army  helicopter travelling  at
120  knots  and 1000  feet  altitude.
The    bottom   interferogram    was
collected  when  the   interferometer
was travelling past the target cloud
of SF6.   Figure 2  shows the  results
after filtering the segment with  the
40-term digital filter.  The bottom
interferogram  segment  shows   the
fundamental  frequency  in which  the
SF6 was present.   It  should be noted
that the digital  filter strategy  can
detect  either  the  absorption   or
emission case.  The result  of a case
of  the  target cloud  being  warmer
than  the  background  is  that   the
resulting filter output will be  180
degrees   out-of-phase   from   the
absorption case.

     The   result    of   collecting
successive   interferogram   segments
while moving is  shown in figure  3.
In this figure a magnitude  response
 of the  output  for the  40-term digital
 filter is shown as a  function of time.
 During this  run an  interferometer  was
 mounted on a UH-1 helicopter and flown
 around  a  source  of   SF6.     As  the
 helicopter passed by  the  target cloud
 (three times)  the instrument alarmed to
 make  a detection.   The x-axis  in this
 figure  corresponds  to  approximately
 three  minutes of  collected data.   The
 target  cloud   was   released   at  the
 beginning of  the data run.   During each
 helicopter pass the  response  of  the
 digital filter became  weaker due to the
 fact    that   the   target   cloud   was
 dispersing in a 10 mile per hour cross
 wind.

      (3)   INFINITE   IMPULSE   RESPONSE
 FILTER

     Infinite  impulse   response  (IIR)
 digital filters  are generally feedback
 loop filters  in  which  additional filter
 coefficients  are used.   The basic form
 of  the   equation   is   shown   in  the
 following equation.
         N
         C
                           M
     The coefficients,  b,  are identical
to the FIR case.  The only difference is
in  the  feedback  response which  is  a
weighted sum of the  past output values.
The  weighted sum  of  present and  past
input values  are added  to  the  feedback
response.  The  major advantage  in  using
a   narrow   band   IIR  filter   for   an
interferogram   is  that  the  number  of
coefficients  is  reduced  making  it  a
highly  efficient filter.   This effect
can  be  illustrated  in  figure  4.    The
attenuation  response  in   this figure
shows that the  IIR   case  has a better
attenuation than for the FIR  case.   The
two   filters   compared   are   on    a
logarithmic  scale  and  have   roughly
equivalent numbers of  computations  for
an  interferogram  segment.    The  major
disadvantage  of IIR filters  are   that
they can be unstable over  large dynamic
ranges   and   can   have   phase   non-
linearities.   A practical problem in the
implementation  of narrow-bandpass  IIR
digital   filters    is   the    required
interferogram    segment   length    for
feedback response  to stabilize and  the
output result to become constant.  This
                                        562

-------
 requirement   is   currently   being
 studied    in    order   to   develop
 alternate methods for implementation
 of  IIR filters for  analyzing  short
 interferogram segments.

 (D)   INSTRUMENTATION   FOR  REMOTE
 SENSING

      Instrumentation currently being
 used  for the mobile-remote detection
 of  chemical  vapors  consists of  an
 interferometer,     an    infrared
 detector,  an  analog signal  module,
 and   a  digital   signal   processing
 module.    The   interferometer  and
 detector   collect   the    infrared
 background  spectral  radiance  and
 convert  it into  an analog  signal.
 The analog signal processing module
 filters and  amplifies the  detector
 signal.  The analog module has  a 16-
 bit  analog-to-digital  converter  to
 convert the signal  to  digital  form.
 The digital signal processing module
 can  analyze  the data  using a  wide
 variety    of    signal    processing
 techniques.

      The interferometer constructed
 for the U.S.  Army CRDEC by the  Midac
 Corporation,  Costa Mesa,  California
 occupies   0.3   cu    feet,   weighs
 approximaely  15  pounds,   and   uses
 only  28  watts  at  12  volts.    The
 interferometer consists of  a linear
 drive  mechanical  mechanism  capable
 of   collecting   two  wave  number
 spectra at speeds of up to  11  scans
 per second.  The interferometer has
 a Helium-Neon 10 milliwatt  laser to
 provide a  reference signal  for  the
 analog-to-digital  converter.    The
 mechanical mechanism is  controlled
 by two small electronic servo  cards
 in  which  one  card  contains   the
 analog electronics   and  the second
 card    contains    the    digital
 electronics.   The infrared  detector
 used   in   the   interferometer   was
 purchased  from  Judson Electronics,
 Costa Mesa, California,  and is. a 2
mm square  Mercury Cadmium  Teluride
 (MCT)  infrared  detector.   A narrow
band  detector   is   used  in   this
 application  since  the atmospheric
transparent  spectral   window    for
 remote sensing is only from  8 to  12
microns.    A  Zinc  Selenide (ZeSe)
beamsplitter  allows  the  instrument
to give a measured noise  equivalent
spectral  radiance (NESR)  of
approximately 1.5 x  10"8  Watts/cm2  *  sr
* cm'1.

(E) DIGITAL  SIGNAL PROCESSING  HARDWARE
FOR REMOTE SENSING

     The  requirement to  perform real-
time  data   analysis  for  the  mobile
chemical sensor can easily surpass  that
of today's conventional  microprocessors.
This    is    particularly   true    when
applications demand several of  the  time
series analytical methods. Conventional
processors  are  designed  to  perform a
wide variety of functions, resulting  in
lackluster    performance     during
multiplication  and  summing operations.
To  overcome  this  shortcoming,   many
microcomputer users purchase an  optional
numerics coprocessor.

     The coprocessors are microcomputers
that have  been  optimized to  perform a
variety   of  mathematical   functions.
These   functions  include integer   and
floating  point  arithmetic  as  well  as
some algebraic  functions.  Coprocessors
can  increase   the   performance  of  a
microcomputer   dramatically;    however,
even the  slowest of the  DSP  chips  can
outperform   the  processor-coprocessor
combination  by   a  factor of  ten   for
common  functions  required for signal
processing applications.

     DSP processors are  much faster  than
conventional  processors  because    of
differences   in    the    chip   design
architecture.  To perform the multitude
of  functions required  by the   desktop
microcomputer, the internal architecture
of the  general-purpose  processor is not
tailored to any particular application.
Most microprocessors use a   single-bus
architecture  in   which  both   program
instructions  and  data  flow  across  the
same   set   of   data   lines.      This
architecture,   known  as  von   Neumann
architecture,  can  result  in   a   data
bottleneck caused by the path  of flow on
the data bus.

     The  size  of  the  general   purpose
registers can also have  a serious effect
on computational  performance.   Intel's
8088 and 80286 processors have  only 16-
bit-wide  registers,  while   the 80386
possesses 32-bit registers.   To perform
math  operations,   the   microprocessor
breaks  the   numbers   into   manageable
portions  and   performs  a   series   of
software   operations   to  obtain   the
desired result.    This  process  requires
                                         563

-------
many  machine  cycles  to  complete.
Numeric processors reduce the number
of  required  cycles   by  employing
larger  registers.    The Intel  8087
numeric  processor  has  80-bit-wide
internal   registers;   however,   it
still  requires multiple cycles  to
perform    even    the    simplest
mathematical computation.

     DSP   chips  are   distinguished
from the  general-purpose processor-
coprocessor  by  their  ability  to
perform  instructions  in  a  single
cycle.   Internal  architecture  of a
DSP is  optimized to perform single-
cycle computations that allow faster
performance  of  the  sum-of-product
calculations   required    by   many
digital    signal-processing
algorithms.    The  performance  is
obtained through the use of hardware
multipliers-accumulators,    Harvard
architecture, or pipelining.

     Hardware multipliers and adders
of  the DSP  eliminate the  software
overhead  required  by conventional
processors    in     mathematical
operations.   These  units allow  the
DSP  to  perform  operations  in a
single  cycle  and  insure  sufficient
register width for accurate results.
These  multipliers  and accumulators
are   arranged   to    optimize   the
multiplication followed  by  addition
type operations.

     To take  advantage of the  high
speed  advantage  of  the  multiplier,
the DSP must insure  a  steady flow of
data into it.   To achieve this,  many
different  techniques  are  employed;
however, most manufacturers use some
variation    of    the     Harvard
architecture.     Unlike   the   von
Neumann   approach,     the    Harvard
architecture uses  separate  program
and data  memories,  each having  its
own bus or  buses.     In  a  digital
filtering    operation,     this
architecture allows the  data and a
corresponding  coefficient   to  be
fetched from memory along  separate
buses  and loaded into  the multiplier
simultaneously while an  instruction
is fetched on the program bus.

     Pipelining  is  another  scheme
used to insure  an adequate flow of
data to the multiplier-accumulator.
In a  pipelined architecture, each
instruction is composed  of  several
steps such as  FETCH,  DECODE,  MULTIPLY,
and ADD.   Each  subsequent instruction is
likewise  divided;   however,   it  always
follows  one  step  behind  the previous
instruction in the  pipeline.   In other
words, while  the first  instruction is
decoding  the  instruction   fetched  one
cycle before, the second instruction is
being  fetched   from   memory.    While
requiring multiple cycles to complete an
entire instruction,  once the pipeline is
filled,  a  result  is  obtained  every
cycle.

     The   general-purpose   DSP   is   a
single-chip integrated circuit designed
to  allow  the  greatest  flexibility as
well as provide good overall throughput.
These integrated circuits range from 16-
bit    to    32-bit    floating   point
architectures  and  are capable of real-
time  signal processing on  signals of up
to  200  KHz.   Commonly offered  features
include   zero   overhead   looping,  bit
reversed   addressing,   and    external
interfaces   to  serial   and   parallel
devices.   Numerous combinations of on-
board and  off-board  memory  are  also
available  from  several manufacturers.
The   general-purpose  DSP  architectures
make  them  ideal for  a wide range of
applications.

      The   general-purpose   DSP  can be
programmed to perform any of the digital
signal processing algorithms much like  a
conventional  microprocessor.    They are
generally programmed  in  their  native
assembly code;  however,  many  of the DSP
chips  have   a   high   level  language
compiler available.  The  use of  a  high
level language makes
software    conversion    from   the
microprocessor-based     systems    much
easier.

      Popular  DSP   chips   include  Texas
 Instrument's 320 family  (32010,  32020,
 320C25,   and   the   32030),   the  AT  &  T
 DSP32C,   the  Motorola   56001,  and  the
 Analog  Device  ADSP2100.     Development
 boards    for    the   popular    IBM
 PC/XT/AT/compatibles  are often available
 from either the manufacturer  or a third-
 party source.   These boards  are either
 used for the development  of  stand-alone
 design or as a high speed  digital signal
 processing  coprocessor   for   a  host
 computer.       Assemblers,   compilers,
 simulators,  and   debuggers  are   often
 included  with many  of  the  development
 boards.
                                         564

-------
     The  Remote  Sensing  Group  at
U.S.   Army    Chemical   Research,
Development and  Engineering Center
(CRDEC)   has  selected  the  AT  &  T
DSP32C as the target processor.  The
DSP32C is  a  CMOS  32-bit  floating-
point processor based on a piplined,
von Neumann architecture.   This 25-
MIPS   (million   instructions   per
second)   device  contains  twenty-one
16-bit fixed-point registers for use
in  control,    address,   and  logic
functions and,  in addition,  four 40-
bit accumulators  to  perform 32-bit
floating-point   mathematical
operations. On-chip memory  includes
2 K of read-only  memory and 4 K of
random access memory.  The  DSP has
an off-chip memory capability of 16
MB.  The  DSP32C also supports serial
I/O  and  a  parallel  I/O  channel
designed  for   easy interfacing  to
either   an  8-bit   or  a   16-bit
microprocessor.

(F) CONCLUSIONS

      Infrared    interferometer
hardware, signal processing computer
hardware, and the application of new
mathematical algorithms have rapidly
advanced   the    remote    sensing
technology  during  the  last  four
years.        Lightweight,     small
interferometers   exist  that   can
withstand    severe    mechanical
vibrations   while   operating   on
rapidly moving  helicopter platforms.
Signal  processing  algorithms  are
available which can extract  infrared
background information in  order to
give  an  automatic alarm  indication
for  the  presence  of  a  particular
chemical   vapor species.    Finally,
digital  signal  processing  hardware
has  been constructed  which allows
infrared remote  sensors  to  process
data  in  real-time.    This  advance
eliminates the need to collect data
for later analysis in a laboratory.
York (1990)
 W    Small,   A.C.
 "- j "
Chan.. 62
                           Harms,   R.T.
                          "-"-   Loerop'
   ,  Chemical  Analysis Monograph
Wiley,  New  York (1975) .

      E.G.   Codding,   G.  Horlick,
      ^ ,27,85 (1973).
REFERENCES:

[1]  S.   Chandrasekhar,   Radiative
Transfer,  Dover, New York (1960).

[2] G.W.  Small,  R.T.  Kroutil, J.T.
Ditillo, W.R.  Loerop,  Anal.  Chem..
60, 264-269  (1988) .

[3] R.T. Kroutil, J.T.  Ditillo, G.W.
Small,  Computer-Enhanced Analytical
                                         565

-------
 50
 25--
  0
-25-
 50
   150
200
250
300
   150
200
250
300
                       Figure 1

-------
               150
200
250
300
s
              150
200
250
300
                                   Figure 2

-------
                             MAGNITUDE RESPONSE OF FIR FILTER
en
O>
oo
W.U 1 W
uj 0.008-
o
a.
ft 0.006-
UJ
a
JD 0.004-
<
^ 0.002-













i


I
p












*r*u, tf^rm^^f^

Iff
i
\ •
M






R


it
Jw\
l*K>*rv^WW '" Wv^u-w-^V/^^SrV^^TT
0
                            100    200    300      400    500

                             COLLECTED INTERFEROGRAM NUMBER
                                           Figure 3

-------
            ATTENUATION OF FREQUENCY RESPONSE
                                 100 COEFFICIENTS
                                       FIR
                               12 ORDER IIR
-140
    750  800   850  900   950  1000  1050 1100  1150
                      WAVENUMBERS
                           Figure h

-------
                                                             DISCUSSION
DONALD GURKA: Can you tell us something about the range and payload of
your drone?
JOHN DiTILLO: Right now, the payload is about 25 pounds, and that includes
everything, the video cameras, the interferometer, the gas, the whole bit. Just the
electronic portion or the sensor portion of it has to be limited to about 25 pounds.
It just happens that the specification for the aircraft the contractor had. You can
build them as large as  you want, and  the  military drones  that are  under
development have very large payload capacities. This just happened to be the one
that we fell upon. A little bit about the aircraft: it flies at about 100 knots with the
gas on board, it can fly for about an hour, and it has autopilot capabilities so you
can send it out on a pre-planned mission and have it fly lazy eights or whatever
over a specific area. Some of the efforts  we have  this year are to tie a global
positioning system into that,  so you can not only get video information back,
which isn't very realistic from a military standpoint, particularly if you look at
the scenario we have now in Iraq where the ground features aren't very distinct.
Video is not going to tell you a whole lot. So, we think it would be a much better
idea if  you actually had grid coordinates as well as a response out of the
algorithm. So, there's some more effort that's going into the aircraft itself as far
as its capabilities, but that just happened to be what was available at the  time.

TOM PRITCHETT: I'm familiar with calibrating the active FTIR units. How
do you calibrate a passive unit?
JOHN DiTILLO: I don't know that much about the optical end of things. The
XM-21 had an internal black body calibration that it went through on start up.
Beyond that I don't know that much about it.

CHIP MILLER: Seems like  from the days of show pair,  low pair days, I
remember problems with confusing silicates and absorption of silicates with the
phosphoryl absorption of the organophosphonates that you're interested in. Is
that still considered a problem, especially with regard with the desert scenario
and the silica?

JOHN DiTILLO: That's the montmorillinite and kaolin problem. A lot of time
and effort was spent on that problem. That was a serious problem early on. As you
can imagine in a military scenario, false alarms can be devastating. And through
years and years of testing and refinement of the algorithm, the XM-21 is virtually
fool proof.  The instrument has been trained to eliminate a lot of those early
problems with dust or compounds that are similar to nerve agents and pesticides.
Just from a military standpoint, if a unit were to get a false alarm out of an XM-
21, the first thing they would do is go into mob gear. As soon as a unit does that,
their fighting efficiency goes down to about 10%. So, you can imagine if your
enemy knows that it hasn't  hit you with agent, and it's looking across the field
and you're in mob gear, you can imagine what kind of ramifications that's going
to have. So, the army goes through great pains to eliminate any  kind of false
alarms due  to dusts and dirts and burning tires, and that type of stuff. And the
instrument has been trained to eliminate those problems.
                                                                         570

-------
                   USE OF WIND DATA TO COMPARE POINT-SAMPLE AMBIENT AIR VOC
                     CONCENTRATIONS WITH THOSE OBTAINED BY OPEN-PATH FT-IR
Ray E. Carter, Jr.,  Dennis D.  Lane, and
Glen A. Marotz
Department of Civil  Engineering
4002 Learned Hall, University of Kansas
Lawrence, KS  66045
Mark J.  Thomas and Jody L. Hudson
U.S. EPA,  Region VII
25 Funston Rd., Kansas City, KS  66115
ABSTRACT

The technique of open-path FT-IR  spectrometry
is being used  increasingly to measure VOCs in
ambient  air.   Since the  FT-IR technique
produces a path-integrated  concentration and
most  other  techniques  produce  point
concentrations, some  method of interconversion
is often desirable.  In the  case of  a plume
generated by a single point source, a solution
to the interconversion problem  can  be found
through the use of wind data.  A quantitative
relationship  was  developed between wind
direction frequency  and concentration.  This
relationship  was  used to  predict  path-
integrated  concentrations,  given point
concentrations.  The same principle was used to
predict  point  concentrations, given path-
integrated concentrations.

The interconversion technique involved the use
of one-minute means  of wind direction as inputs
to a Gaussian dispersion model.  The one-minute
concentration-to-emission rate  (C/Q)  ratios
produced by the model were integrated over the
sampling period of the  test  to yield a  C/Q
ratio that was  based on the  wind directional
frequency distribution, rather than on  the
overall mean wind direction.  These integrated
C/Q ratios for selected points  were  used to
develop  predictive methods for both point
concentrations  and path-integrated
concentrations; this process is described in
detail within the body of the paper.

The interconversion  technique was tested, using
data from simulated field  tests  during which
VOC  releases  were  made.   The VOC  plumes
generated were monitored along a line normal to
the projected plume  centerline, using the FT-IR
technique and  also by  collecting  whole-air
samples in evacuated stainless steel samplers
for subsequent GC/FID analysis.  During  the
final  test set, no FT-IR measurements were
made;  an observed path-integrated concentration
was  produced  by  using  the  mean  of
concentrations from point samples  collected
five  meters  apart along the  path, thus
providing an evaluation of the  technique free
of any bias that might  exist  between the  two
analytical methods.

Correlations  between  the observed
concentrations for the point samples  and  the
corresponding  integrated  C/Q ratios were
assessed  for  each test  and  found  to be
significant at  the  0.1  level  in most cases.
Although a bias between the two analytical
methods was seen, the predicted  path-integrated
concentrations were  strongly correlated with
the observed values.  For the test set in which
only point samples were  collected,  excellent
agreement between  the predicted and observed
path-integrated concentrations  was seen.   The
predictive method for point concentrations  did
a good job of predicting  both the location  and
the magnitude  of  the highest  concentrations
from each test, and reflected the general shape
of the concentration-versus-crosswind curve
well.

INTRODUCTION

The technique of open-path Fourier  transform
infrared (FT-IR)  spectrometry is being used
increasingly  to measure volatile organic
compounds  (VOCs) in ambient air.  Depending on
the circumstances  (nature of the source,
receptors affected, etc.)  this technique may
complement, or  in  some  cases  replace,  the
collection of  multiple  whole-air samples with
subsequent laboratory analysis.
                                              571

-------
The FT-IR technique  produces  a concentration
integrated over the  length  of  the  path from
source  to detector for  a  given  compound,
whereas the  whole-air  method  produces  a
concentration  for  only  those  points sampled.
In order to use the data  produced by  the two
techniques  optimally,  some method  of
interconversion is desirable.

Solution of the  interconversion problem would
be difficult in many  cases.  However,  in the
case of  a plume  generated by  a single point
source  with  relatively  constant emission
characteristics, a solution  can be found
through the use of wind  data.  The location and
movement of the plume centerline (and therefore
the maximum concentrations) are determined by
the wind direction.  Thus,  the concentration of
plume-derived  compounds  found in point  samples
should be primarily a function of the amount of
time that the  wind blew  from  the source toward
those points during the  sampling period.

If  a quantitative  relationship between wind
direction  frequency and concentration  can be
derived, that relationship can be used to
calculate  the concentration  at any point along
the path from  the  IR  source  to the detector.
The mean of those values for  points closely and
evenly  spaced along the path should then
provide an  approximate  path-integrated
concentration.  The same principle  can be
applied to  the conversion of  path-integrated
concentrations to point  concentrations.

APPROACH

Sampling and AnaJLys.is_Framework_Overvi_ew

Plumes  consisting of both single compounds and
mixtures were  generated  from  a stack two meters
above the ground.  Point samples were collected
downwind from  the source along a line normal to
the  projected plume centerline for subsequent
analysis.  During the first three of the four
test sets  performed,  the plume was also
monitored along the same path using an open-
path FT-IR  spectrometric method developed at
Kansas State University  (1).

During  the fourth test  set,  point samples were
collected at  five-meter intervals along the
path;  no FT-IR measurements were made.  This
closely-spaced  network  of point samplers
provided a more detailed characterization of
the concentration-versus-crosswind distance
curve.  It also provided an observed path-
integrated concentration based on the mean of
concentrations from those  samples, allowing
further evaluation of  the  interconversion
technique free of  any bias that may exist
between the  two analytical methods.
Wind data collected during the sampling period
were then used in conjunction  with  a  Gaussian
dispersion model  to  determine  a  calculated
concentration-to-emission rate ratio (C/Q)  for
selected points along  the path for  each minute
of the test.  The  C/Q values for  each point
were then summed  over the  sampling  period of
the test  to  produce an  integrated C/Q  (or C/Q*)
for each  point.

The above data and calculated  values were then
used in the following ways,  to  test  the
validity and usefulness of  the conversion
technique:

1)   Correlations between the observed
     concentrations for the  point  samples and
     the corresponding C/Q* values were
     performed for each test.

2)   The  observed point concentrations and the
     calculated  C/Q* values were  used to
     predict a path-integrated concentration.
     This  value  was  then  compared  to  the
     observed path-integrated  concentration for
     the same time  period.

3)   Using  the observed  path-i n t e gr ated
     concentration  and the wind  data, an
     equation  was developed to  predict
     concentrations at selected  points. These
     predicted  values were then compared to
     observed concentrations present  in samples
     collected at  those points.

 Did ivi dual _T es t _Runs

 Four sets of field tests were performed, one  on
 each of  the following dates:  February 24,
 1989;  October 20, 1989; April  25,  1990; and
 October 22, 1990.   VOC releases were made  from
 a stack  two meters above the ground in  an  open
 field near Lawrence,  Kansas.  During the first
 three   test sets,  the  general  spatial
 relationship of the VOC source and  the  sampling
 devices  was as shown in Figure 1.   Specific
 parameters  (distances, number of point samples
 collected,  etc.),  in addition to the field
 deployment  used  in the  fourth test set, are
 described in the succeeding paragraphs.

 Figure  1

 On February 2U,  1989, a  20-minute release  of
 toluene was  made. The  release was monitored
 along a  line normal to the  projected  centerline
 of  the  VOC plume 50  meters from the source,
 using the open-path FT-IR methodology developed
 at  Kansas  State  (1) and  also by collecting
 whole-air samples  in evacuated stainless steel
 samplers for subsequent  GC analysis.  Sampling
 and analysis of whole-air  samples  followed
 protocols  developed at the  University of Kansas
                                                572

-------
                                                          S = VOC Source
                                                          O = Point Sampler
                                                          C = Centerline Sampler
                                                          FD = FT-IR Detector
                                                          Fs = FT-IR Source
                                 Figure 1.   Sampling Network
(2).   Nine whole-air samplers were deployed  at
ground level along the path, with a = 30x.   The
FT-IR  path length was 120 meters.

On October 20,  1989, six  30-minute  releases
were made.  Only  1 ,1 ,1-trichloroethane was
released during the first three of these tests.
During the final three tests, various mixtures
of the following  six compounds were released:
n-pentane, methylene chloride, methyl ethyl
ketone,  tert-butyl   alcohol,  1,1,1-
trichloroethane, and toluene.  The releases
were again monitored along a line normal to the
projected plume centerline  50  meters from  the
source,  using both the FT-IR and whole-air
techniques.   Five whole-air samples were
collected at ground level  during  each  test,
with a = 30x.  The FT-IR path  length was  100
meters.  During  the final  two tests  of  this
set,  the wind had shifted  approximately  30
degrees, making  it questionable whether  the
plume centerline was within the  sampling
network.  Data from those  tests  are not
reported.

On April 25, 1990, ten 12-minute releases were
made.   Mixtures of the following five compounds
were  released during the  first nine tests:
methylene chloride, methyl ethyl ketone, tert-
butyl  alcohol, 1,1,1-trichloroethane, and
toluene.  Only 1 , 1 ,1 -trichloroethane was
released  during the  final test.  The plumes were
again monitored along a  line  normal to the
projected plume centerline, this time  40 meters
from the source, using  both  the FT-IR and
whole-air techniques. A minimum of five whole-
air samplers were deployed along  the path
during  each test at  one meter above the ground,
with a = 20x during  Tests 1-3  and a = 30x
during Tests 1-10.   The IR path length was  50
meters during  Tests  1-3,  200  meters during
Tests 4-6, and 100 meters during Tests 7-10.

On October 22, 1990,  three  12-minute  releases
were made and again monitored  along  a  line
normal  to the projected plume centerline  40
meters  from the source.  Fifteen  whole-air
samplers  were collected at  ground level  along
the path  at five-meter intervals in each of the
the three tests, as  shown in Figure 2.   No FT-
IR measurements were made.  Only toluene and
1,1,1-trichloroethane  were released.

The network of samplers was shifted five meters
to the  left for Test  3 to account for  what was
perceived to be a slight wind shift.  Samplers
were thus arrayed from 40 meters left  to  30
meters  right of the  projected centerline during
this test.
                                                           S = VOC Source
                                                           O = Point Sampler
                                                           C = Centerline Sampler
                            Figure 2.   Sampling Network, 10/22/90
                                             573

-------
The  following meteorological data were
collected during  all  releases:  one-minute
means of  temperature and relative humidity,  and
one-minute means  and  standard deviations of
wind direction and wind speed.

EXPLANATION OF METHOD

Determination_of_C/Q_VaUi es_

An  algorithm  was written to  calculate
integrated  concentration-to-emission rate
ratios (C/Q»)  for  selected points.  As  it is
now  written,  the algorithm can produce a C/Q*
value for up  to  ten points per execution.   The
following inputs are required:

      - wind direction means for each minute of
       the test
      - wind speed  means for each minute  of  the
       test
      - overall  standard deviation  of wind
       direction for the  test
      - location  of  the  centerline  point
       (distance and direction from the source)
      - location  of  the  remaining  points
       selected (distance left or right of the
       centerline  point)

 Inputs  were  used  in  conjunction with  a
 dispersion model to produce  C/Q* ratios  for
 selected points, in the following manner:

 1)    For  each minute of the  test, x-  and y-
                         2)
   Observations

    S = VOC source

    C = centerline point

    D = designated point

   SC = 50 meters

   DC = 25 meters
   network centerline direction
      180C
  meanwind direction  (•»)
  B - 180° - 160°  = 20°
160C
coordinates  were determined  (with  the
source as  the  origin and the mean wind
direction  for  that  minute providing the
direction of the x-axis) for each  of  the
selected points.  Each of  the points was
assumed to  be on a line perpendicular  to
the line from the source to  the centerline
point and the appropriate trigonometry was
then performed.   An example for one such
point is shown in Figure 3.

The  standard  deviation of  the wind
direction was then used to determine  the
stability class.  Based on that class,  and
on the  x-coordinate for a given point,  a
value  for  o   and o   was determined  for
each point (3)-     An example  of this
process is shown  below,  using the point
discussed  in  part   1  and assuming  a
directional  standard  deviation  of  10
degrees and a wind speed of 5 m/sec.

- direction  standard deviation  of  10°
  indicates stability class D
- for that  stability  class,
                             o  = 0.08 x  (1 + O.OOOIx)
                                  0.08(55.5m)(1
                                  14.43m

                                  0.06 x  (1 +• 0.0015x)
                    0.0001(55.5m))

                       -1/2
                               3.20m
Trigonometry

DA - DC  - AC = DC - SC tanB

SA = SC/cose

AB = DA  sing = (DC - SC tanB) sinB

 x = SB  = SA + SB

   = SC/cosB + (DC - SC tanB) sinB
   = 55.5 meters
                                                            DB
         DA cosa - 6.4  meters
                      Figure 3.   Determination of x-  and y- coordinates
                                                574

-------
3)  These  values for  o   and o  are then used
     in the dispersion  equation  to  calculate
     C/Q,  as shown  below,  also for the point
     discussed in parts  1 and  2.
     C/Q = — ~
                 1
                        -(6.4)2/2(4.43)2
4)
       2Tr(4.43~)(3.22)(5)'

                -4
     =  7.907 x 10

The C/Q values for each point were summed
 over the sampling  period  for  that  test,
 which  yielded an integrated C/Q (or C/Q*)
 for each point.  C/Q*  values for the  seven
 points at  which point  samples  were
 collected during Test  10 on April  25,  1990
 are as follows:
         Location


            30°L

            20°L

            10°L

         centerline

            10°R

            20°R

            30°R
                             C/Q*
                             3.206 x  10

                             2.020 x  10
                             4.
                                   x  10
                             4.112 x  10

                             4.016 x  10

                             8.565 x  10

                             2.840 x10
-4

-2

-2

-2

-3
-6
Correlation of Observed PointConcentrations
Correlations  between  observed concentrations
from point samples  and corresponding  C/Q*
values were assessed.   This operation produced
a correlation coefficient  (r) and an associated
probability (p).  The latter value represents
the probability that  a  correlation coefficient
of this magnitude could occur if there were no
relationship between  wind  directional frequency
and concentration.  For  Test 10 on April 25,
1990,  the following  concentrations of 1,1,1-
tr i chloroet hane were found  in the  point
samples':
Location

   30°L
   20°L
   10°L
centerline
   10°R
   20°R
   30°R
                    Concentration (ppb)

                            215
                           1056
                           1163
                           1188
                            175
                             46
                              0.0
These  concentrations  produce a correlation
coefficient of 0.968 with the corresponding
C/Q* values for the same  test  (shown in part 4
of the  preceding section),  with a probability
of <0.001.

Prediction_gf__Path-I_ntegrated Concentrations

The  observed  point concentrations and the
calculated  C/QX ratios were used to produce a
predicted  path-integrated concentration for
each test,  in the following manner:

1)   A  linear regression  was  performed, with
     the  model-derived C/Q* values used as the
     independent  variable  and the observed
     concentrations in  point samples  the
     dependent variable,  yielding a functional
     relationship  between wind direction
     frequency and concentration.  A regression
     equation  of  this form   (y  -  ax  +  b)
     normally contains a non-zero intercept
     value, which can produce relatively high
     predicted concentrations  corresponding to
     extremely low C/Q*  values  (an unrealistic
     situation),   or even  negative
     concentrations  if  the   intercept  is
     negative.

     Two solutions to this   problem  were
     attempted.  The first was  to constrain the
     regression line to  pass  through the origin
     (intercept = 0). This solution eliminates
     the problems  caused  by the  intercept,
     although it does not produce as good a fit
     between predicted and  observed values.
     The second solution was to perform a log-
     log regression.   The equation produced
     generally yielded a slightly better fit
     than the zero-intercept  method, while
     still  allowing  predicted concentrations to
     asymptotically  approach zero  for
     decreasing values of C/Q*.  Shown below
     are regression  equations produced from the
     4/25/90  (Test  10) data,  using  both methods
     described in this paragraph.  Also  shown
     are F-ratios,  indicating  the  significance
     of the regression, and predicted  versus
     observed values.

          Zero-Intercept Method

          C = 33804  (C/Q*)
          F = 88.0   (p <0.001)

      Location   Observed C  Predicted C
               30°L
               20°L
               10°L
               centerline
               10°R
               20°R
               30°R
                  215
                  1056
                  1163
                  1188
                  175
                   U6
                    0.0
  11
 683
1652
1390
 136
   0.
                                                                                  0.0
                                               575

-------
         Log-Log Method

         In C - 0.392  (ln(C/Q*)+8.28
            F = 280    (p <0.001)

     Location Observed_C    Predicted_C

      30°L        215          168
      20°L       1056          852
      10°L       1163         1204
      centerline  1188         1124
      10°R        175          452
      20°R         46           41
      30°R          0.0          0.3

2)    Once the regression  equation was derived,
     it  was used to predict concentrations at
     evenly spaced  points  along the path.
     Predicted concentrations  (based on a log-
     log regression) for  the 4/25/90 test  are
     as  follows:

  Location  Concentration  Location Concentration
                                                   2)
50m L
45m L
40m L
35m L
30m L
25m L
20m L
15m L
10m L
5m L
centerline
0.0
0.0
0.2
2
17
97
359
818
1146
1236
1129
3) The mean of the
50m R
45m R
40m R
35m R
30m R
25m R
20m R
15m R
10m R
5m R

predicted
0.0
0.0
0.0
0.0
0.0
0.1
2
34
219
658

concen
     for  evenly  spaced  points  produced  a
     predicted path-integrated concentration.
     The predicted and observed path-integrated
     concentrations are shown below for  the
     4/25/90 test.

     observed path-integrated  concentration =
     185 ppb

     predicted path-integrated concentration =
     272 ppb

 Prediction ^_Polnt_Concentrations

 The  observed path-integrated concentration and
 wind data collected for the  same time period
 were  used to produce predicted concentrations
 for  selected points along the  path  in  the
 following manner:

 1)   Wind data and a dispersion model were used
     to produce C/Q* values for evenly spaced
     points  along  the  path,  as  outlined
     earlier.
                                                   3)
    A conversion factor was determined, based
    on   the  observed  path-integrated
    concentration and the mean of C/Q* values
    for  evenly spaced points along the  path,
    using the following equations:
       PIC
              n
         F x (l
              i
(C/Q*)i)/n
and
 (Cpt). = F x (C/Q»)1
and
where F = conversion factor
    PIC = path-integrated  concentration
(C/Q*)  = integrated C/Q for the ith point

      n = total  number of evenly  spaced
         points
 (C  ). = predicted concentration at  the
         ith point.

 Based on the data collected during Test 10
 on 4/25/90,

      F = 185/7.955 x 10~3
      F = 23255.8

 The  conversion  factor was used  in
 conjunction with  C/Q*  values to predict
 concentrations for selected points,  which
 were then  compared  with observed
 concentrations from samples collected at
 those points.  For the centerline point on
 4/25/90  (Test  10), the  predicted  1,1,1-
 trichloroethane  concentration  is  as
 follows:
                                  ~2
         C t = 23255.8 x 4.112 x 10

         C   - 956 ppb
     Shown  below are 1 , 1 , 1-trichloroethane
     concentrations predicted for all points at
     which samples were collected during the
     4/25/90 test.  Also shown are the observed
     concentrations for those points.
Location

  30°L
  20°L
  10°L
centerline
  10°R
  20°R
  30 °R
              Concentration  (ppb)
          Predicted          Observed
               8
             470
             1137
             956
              93
               0.
               0.0
            215
           1056
           1163
           1188
            175
             46
              0.0
                                                 576

-------
 RESULTS, DISCUSSION

 Correlation of  Observed  Point Concentrations
 with C/Q* Values  "

 Table  1  shows  correlations (and associated
 probabilities) between observed  point
 concentrations  and corresponding C/Q* values
 for each test.  For  releases that consisted of
 mixtures of two or more compounds, only 1,1,1-
 tr ichloroethane (1,1,1-TCA)  and  toluene
 concentrations were used  in  the analysis.

 In  all  but three  tests, the  correlation
 coefficients produced were associated  with
 probabilities  of less than 0.1.  In  at least
 one of  those three cases,  Test  3  on 10/20/89,
 it appears  that  additional valid data points
 would  improve the correlation.   These
 statistics  would support  the belief that there
 is a strong  relationship between wind direction
 frequency and concentration.
f££^i£t?S_Y.ers!i3 Obse_r_ved_P£th-Integr a ted
Concentrations
Table  2  shows  observed  pa t h-i nt egr a t ed
concentrations  for all  field tests.  Again,
only results for 1,1,1-trichloroethane and
toluene  are reported.  Also shown are the
             Table  1.  Correlation of Concentrations with C/Q* Values
                            corresponding  predicted path-integrated
                            concentrations, based on observed  point
                            concentrations  and  wind data.   The values
                            reported are those produced  using log-log
                            regression of concentration  against  C/Q*.   At
                            the time of  this  writing,  FT-IR data was not
                            available for 2/21/89.

                            It  should  be  noted  that observed  path-
                            integrated concentrations reported for the test
                            set of 10/22/90 were  produced by using the mean
                            of concentrations from point samples  collected
                            five meters  apart along  the  path.  Values of
                            predicted path-integrated  concentration  for
                            this  data set  were produced  using  log-log
                            regression of concentration against C/Q*  for
                            the following five  of  the  fifteen  points at
                            which  samples were collected:   the  projected
                            centerline point, points  10  meters left  and
                            right  of  that point,  and points 20 meters left
                            and right of that point.
                            Excellent agreement is seen  in the observed and
                            predicted values  from  the  10/22/90 test set,
                            with the  largest  percent difference being 2.5%.
                            Values shown for the 10/20/89 and 4/25/90 test
                            sets indicate the possibility of a bias between
                            the two analytical  methods, especially  in the
                            analysis  of 1,1 ,1-trichloroethane,  making an
                            evaluation of the prediction  technique more
                            difficult in this case.  In order to  gain some
             Date
             2/24/89
     Stability  Number of Valid  Correlation Coef^ and Probabi_lUy_
Test _C_lass_     Data Points    1 ,1,1rfCA"vs C/Q* fofuene'vs ~c7Q*~
10/20/89
10/20/89
10/20/89
10/20/89
4/25/90
4/25/90
4/25/90
4/25/90
4/25/90
4/25/90
4/25/90
4/25/90
4/25/90
4/25/90
1
2
3
4
1
2
3
4
5
6
7
8
9
10
                              B
                              B
                              B
                              B

                              D
                              D
                              C
                              C
                              C
                              D
                              D
                              C
                              C
                              D
             10/22/90   1      B
             10/22/90   2      B
             10/22/90   3      C
                    5
                    4
                    4
                    5

                    5
                    5
                    5
                    5
                    5
                    5
                    5
                    5
                    5
                    1

                   15
                   15
                   15
                                                           NR
                                                r - 0.889  p<0.01
r = 0.900
r - 0.970
r = 0.885
r = 0.868
    0.946
    0.943
    0.968
    0.853
    0.878
    0.960
    0.898
    0.663
    0.739
r = 0.968

r = 0.942
r - 0.968
r - 0.918
r
r
r
r
r
r
r
r
r
p<0
p<0
p<0
p<0
p<0
p<0
p<0
p<0
p<0
p<0
p<0
p<0
p<0
p<0
p<0
p<0
p<0
.05
.05
.2
.1
.02
.02
.01
.1
.1
.01
.05
.4
.2
.001
.001
.001
.001
NR
NR
NR
r -
r =
r =
r «=
r -
r -
r =
r =
r -
r •

r -
r =
r »
0
0
0
0
0
0
0
0
0
0

0
0
0
.858
.940
.944
.982
.895
.887
.916
.910
.675
.759
NR
.942
.969
.954
p<0.1
p<0.02
p<0.02
p<0.01
p<0.05
p<0.05
p<0.05
p<0.05
p<0.4
p<0.2

p<0.001
p<0.001
p<0.001
             NR - Compound not released during this test
                                              577

-------
                  Table 2.  Predicted vs Observed Path-Integrated Concentrations
                                    1,1,1-TCA Cone,  (ppb)      Toluene Cone,  (ppb)
                  Date     Test     Observed    Predicted     Observed    Predicted
                  2/24/89
NR
10/20/89
10/20/89
10/20/89
10/20/89
4/25/90
4/25/90
4/25/90
4/25/90
4/25/90
4/25/90
4/25/90
4/25/90
4/25/90
4/25/90
10/22/90
10/22/90
10/22/90
1
2
3
4
1
2
3
4
5
6
7
8
9
10
1
2
3
                                       185
                                       181
                                       199
                                       78

                                       56
                                       138
                                       38
                                       16
                                       34
                                        7
                                       36
                                       75
                                       18
                                       185

                                       334
                                       277
                                       282
      284
      297
      307
      106

       85
      197
       44
       37
       62
       10
       71
      107
       25
      272

      340
      280
      278
 NA
 NA

 35
 85
233
 20
 30
 42
 18
116
146
333
284
288
      NR
      NR
      NR
      NR
                              235
 64

 31
142
204
 12
 45
 55
 28
 76
118
341
279
282
                  NR - Compound not released during this test
                  NA - FT-IR data  not available
insight  into the performance  of the technique,
correlations  between observed and  predicted
values were assessed for the  4/25/90  test  set.
For  1 , 1 , 1-trichloroethane, the correlation
coefficient was  0.994, with an associated
probability  of  much  less than 0.001.   For
toluene, the correlation coefficient produced
was 0.915, with  an associated probability of
less than  0.001.  These results indicate  that
the technique presented for predicting path-
integrated concentrations  is potentially  a
sound one  and  warrants further study.

Predicted  vs.  Observed Point  Concentrations

Table 3 (on following pages) shows observed
concentrations for all point  samples  collected
in the four test sets.

Also shown are the corresponding  predicted
concentrations for the  points,  based  on
observed path-integrated concentrations and
wind data.  At the time of  this writing, FT-IR
data was not available for 2/24/89.

As seen  in Table 3,  the technique presented for
predicting point concentrations  does  a
reasonably good  job  of predicting the general
shape of the  concentration-versus-crosswind
distance curve,  although concentrations 15 or
more degrees from the centerline are  generally
predicted  less accurately than are those nearer
       the  centerline.  Given the bias present between
       the  two analytical  methods, both the location
       and  the magnitude of  the highest concentrations
       from each test are predicted quite accurately.

       CONCLUSIONS

       The  following conclusions are warranted,  based
       on the data presented:

       1)   The use  of one-minute  means  of  wind
            direction  as   inputs  to a  Gaussian
            dispersion model produce concentration-to-
            emission rate ratios that  are strongly
            correlated with  observed concentrations.

       2)   The  predicted p a t h - i n t e g r a t e d
            concentrations  show good  agreement with
            the observed values,  given the bias  seen
            between the two  analytical methods.

       3)   The predicted point concentrations reflect
            the  general shape of the  concentration-
            versus-crosswind distance curve well.  The
            location and the relative magnitude of the
            highest concentrations from each  test are
            predicted accurately.

       4)   The  principles  underlying   the
            interconversion  methods are sound, and the
            methods themselves warrant further testing
            and development.
                                                578

-------
Table 3-  Predicted vs Observed Point Concentrations
Sample 1,1,1-TCA
Location Obs.
Cone.
Pred.
Toluene
Obs.
Cone.
Pred.

2/24/89 - Test 1
30° L
20° L
10° L
5° L
centerline NR
5° R
10° R
20° R
30° R
10/20/89 - Test 1
30° L 67
15° L 71 1
centerline 966
15° R 749
30° R 65
10/20/89 - Test 2
30° L 223
15° L 884
centerline X
15° R 283
30° R 47
10/20/89 - Test 3
30° "L "x
15° L 925
centerline 1043
15° R 241
30° R 43
10/20/89 - Test 4
30° L" ~ 48
15° L 272
centerline 278
15° R 297
30° R 20
4/25/90 - Test 1
20° L 33
10° L 182
centerline 295
10° R 134
20° R 44
4/25/90 - Test 2
20° L 78
10° L 525
centerline 737
10° R 277
20° R 28











206
336
405
304
34

181
376
386
232
88

236
372
410
294
68

65
98
141
176
57

6
170
191
63
0.8

8
286
715
73
0.1

214
891
1009
718
364
351
333
114
28



















27
149
148
158
11

12
65
106
50
19

68
357
502
193
23





NA







NR





NR





NR





NA



4
106
119
39
0.5

5
176
440
45
0.1
Sample 1,1,1-TCA
Location Obs.

4/25/90 - Test 3
20° L 29
10° L 88
centerline 123
10° R 83
20° R 16
4/25/90 - Test 4
30° L 130
15° L 150
centerline 346
15° R 90
30° R 17
4/25/90 - Test 5
307 L 179
15° L 439
centerline 381
15° R 130
30° R 4
4/25/90 - Test 6
30 °T " 20
15° L 66
centerline 116
15° R 18
30° R 2
4/25/90 - Test 7
30° L 37
15° L 310
centerline 329
15° R 22
30° R 6
4/25/90 - Test 8
30° L 39
15° L 746
centerline 308
15° R 111
30° R 12
4/25/90 - Test 9
30° L 96
15° L 95
centerline 49
15° R 5
30° R 4
Cone.
Pred.


8
71
135
63
14

3
34
179
83
0.3

21
220
325
82
0.2

0.0
27
93
7
0.0

0.8
67
161
2
0.0

5
243
385
88
0.9

24
106
37
0.6
0.0
Toluene
Obs.


117
374
613
396
79

26
57
11 1
34
8

149
293
284
100
13

102
402
561
94
16

22
118
130
12
7

31
494
211
73
3

463
478
228
23
21
Cone.
Pred.


48
438
830
387
84

4
42
224
103
0.4

19
194
286
73
0.2

0.0
161
559
41
0.0

0.4
34
81
1
0.0

8
376
595
136
1

195
855
301
5
0.0
                                    579

-------
      Table 3.  Predicted vs Observed Point Concentrations  (cont'd)
Sample 1
Location
,1 ,1-TCA Cone.
Obs. Pred.
Toluene
Obs.
Cone.
Pred.
Sample
Location
1,1,1-TCA
Cone.
Obs. Pred.
Toluene
Obs.
Cone.
Pred.






4/25/90 - Test 10
30° L
20° L
10° L
centerline
10° R
20° R
30° R

10/22/90 -
35m L
30m L
25m L
20m L
15m L
10m L
5m L
centerline
5m R
10m R
15m R
20m R
25m R
30m R
35m R
215
1056
1163
1188
175
46
0.0

Test 1
6.1
21
61
1 44
479
909
789
691
607
574
482
232
8.0
2.8
1.8
8
470
1137
956
93
0.2
0.0


5.7
21
68
194
458
818
1027
911
650
429
249
116
43
13
3.4









5.7
22
63
148
486
920
785
686
591
575
470
224
9.9
2.8
4.5



NR





5.7
21
68
193
457
816
1024
909
648
428
248
116
43
13
3.4
centerline
5m R
10m R
15m R
20m R
25m R
30m R
35m R

10/22/90 -
40m L
35m L
30m L
25m L
20m L
15m L
10m L
5m L
centerline
5m R
10m R
15m R
20m R
25m R
30m R

610
479
553
333
229
80
25
42

Test 3
0.6
1.7
1.5
1.5
33
59
189
395
740
826
778
621
391
122
64

520
456
416
335
217
113
49
18


0.0
0.0
0.0
0.1
2.0
32
200
527
900
1127
777
404
188
56
9.8

592
506
561
344
237
83
27
44


0.5
0.6
0.7
1 .0
34
61
198
420
768
846
787
627
379
123
66

533
468
427
343
222
116
50
18


0.0
0.0
0.0
0.0
2.0
33
204
538
919
1151
793
412
192
57
10

10/22/90  - Test 2
35m
30m
25m
20m
15m
10m
 5m
1.2
17
24
89
342
641
690
5.5
24
85
228
448
620
619
1.3
19
28
90
358
643
725
5.6
25
87
234
459
636
635
NR - Compound not released during this test
NA - FT-IR data not available
 X - Whole-air sampling error, no data
REFERENCES

(1)   Spartz,  M.L.,  M.R.  Witkowski,  J.H.
     Fateley, J.M.  Jarvis,  J.S.  White, J.V.
     Paukstelis,  R.M. Hammaker,  W.G. Fateley,
     R.E.  Carter, Jr., M. Thomas,  D.D.  Lane,
     G.A. Marotz, B.J. Fairless,  T.  Holloway,
     J.L. Hudson, D.F. Gurka.  "Evaluation of a
     mobile FT-IR system for rapid volatile
     organic compound determination,  Part I:
     Preliminary  qualitative and quantitative
     calibration  results,"  American
     Environmental Laboratory, v.  1,  no.  2,
     November, 1989,  PP 15-30

(2)   Marotz, G.A., D.D. Lane, R.E.  Carter, Jr.,
     R. Tripp, J. Helvig.  "Preliminary  results
     from  a rapid deployment field study  of
     heavy gas detection and dispersion  using a
     whole-air technique," 87-103-7, EPA/APCA
     Symposium  on Measurement  of  Toxic Air
     Pollutants,  Raleigh, NC, 1987.
(3)   Hanna,  S.R., G.A. Briggs,  R.P.  Hosker,
     Jr.,  "Gaussion plume model for continuous
     sources , "  H_£rid_book_on_A tmospheric
     Diffusion ,  Technical" Information ~Center7
     U.S. Department of Energy, 1982, pp 25-31!
                                               580

-------
                                                            DISCUSSION
DONALD GURKA: How certain are we that there is no physical gradient across
the plume? That is, we're not looking at part gas and part aerosol?

RAY CARTER: We have found that, based on the studies that we've done over
a five-year period, the plume is well dispersed as it comes out of the stack. It is
in fine liquid droplets as it comes out of the stack, but the droplets are so dispersed
that they vaporize almost immediately. We have data from other tests that would
support this.

DONALD GURKA: I guess my point is that the physical gradianl might have
a differential effect on the open path concentration approach versus the canister
concentration approach.

RAY CARTER: I think I see your point, and there did seem to be a bias between
the two methods. And that is one good possibility if it is not all  in vapor state. We
believe that it is, but since there is a bias it's still worth looking into.

TOM PRITCHETT: Since in essence you're doing a controlled release of
essentially a known emission rale, and in the case of your long path, open path
monitoring, you're calculating emission rate versus concentration ratio. Have
you used the open path monitoring to calculate ancxperimental emission rate and
compared that against the generation rate of vapors? And also, use, let's say, the
transect method of data reduction on the canister data, to  also calculate an
emission rate and see which one of those methods wasgiving you an experimental
emission rate  that was closest to your  generation rate?

RAY CARTER: We did not have real good control on the FT1R methodology.
It was primarily done by the people from Kansas State, and so we haven't really
done anything with their data. I think your point is a good one,  though. We could
use our data and the method that you suggest to see if we accurately predict the
emission rate  that we did measure.

TOM PRITCHETT: Just as a follow up to that, if you looked at both the canister
data and the open-path monitoring data, since you have a consistent bias, you
might be able to resolve who's causing the bias by checking the  two experimental
emission rates versus your generation  rate.

RAY CARTER: Yes, I would agree with that. And if we can obtain all of data
from Kansas State that would be a good test to undertake.

ERNIE TUAZON: Would il be better to use a dye compound, like something
that is more evenly distributed in the atmosphere which can be measured by the
FTIR while  it's doing its measurements, and also being sampled by your
canister? Like a compound as simple as methane in the air—it's almost relatively
constant. Even nitrous oxide or carbon dioxide. This will be in the data already.
You will have sampled it, and the FTIR will have measured it already. So, if
you're looking for that bias then, part of your answer may lie there.

RAY CARTER: You're suggesting using compounds that do naturally exist?
Then you would not have the gradient across the path?

ERNIE TUAZON: That's correct, this is not a test of your model, but it tells you
which one is producing a lower reading in that general direction.
RAY CARTER: Yes, since our main purpose was to test this model, then we
preferred to actually have a gradient across the path, but I see your point. That
would be another good method of determining which is the biased method. That
was not really our intent. When we discovered that there was a bias, the reason
we did the fourth test set was to Jry to test the method independent of the bias.

ERNIE TUAZON: I am aware of that. Also, one thing that might also affect your
comparison is the way you sample with the FTIR. There is a dead time between
FTIR measurements if you're calculating rightafteryou collect the interferograms.
In other words, there will be segments of interferograms being collected and then
dead time while you're calculating, and then you again collect interferograms.
In the canister, you're continuously sampling while that's occurring, aren't you?

RAY CARTER: I'm not sure I understand your question, but if I do, I believe
that was taken care of by the sequencing of the sampling. We attempted to
cooperate as much as possible with the people doing the FTIR measurements,
merely adjusted our sampling to fit whatever schedule they preferred.

DONALD GURKA: Yes, it seems to me that the RTPGroup with, I think that's
Bill McClenny, also saw this negative bias, but it seems to me that the bias was
within the combined experimental error. Is that correct?

BILL McCLENNY: The tests that we did were in the Delaware Site  Program.
And the tests there were done  with a plume that was originating from a nearby
industrial plant, and consisted of two primary emissions, paradichlorobenzene
and chlorobenzene. All tests that we did were by moving a canister along the path
next to the path of the augus — the line of sight for the beam. We had the system
set up with a source receiver at one end and a retroreflector at the other end. We
were carrying the canisters back and forth between those two locations. By
moving over a period of one-half hour, we gel an integrated canister sample, or
a sample that's  integrated over time. And then the  spectra from the Fourier
Transform System, were co-added over the same period of time. The  two were
compared. The comparison was based on a common standard, i.e., the GC/MS
standard that was used to look at the paradichlorobenzene and the chlorobenzene,
also used for the FTIR System. By using a common standard we had a common
basis on which to compare, and we compared those measurements directly. And
those measurements were very close with paradichlorobenzene, even though we
were depending on a plume that was dispersing from a point source over which
we had no control. But, in our case we had to locate it at the right position, which
was an inconvenience. So, this type of comparison in which you have a control
source has advantages. For us, we were in the field. We had to locate downwind
of the source, and so our efficiency of taking these comparisons was reduced
because we had  to wait for the right experimental conditions.

DONALD GURKA: But, the negative bias was within the combined experimental
error for the chemistry of these compounds?

BILL McCLENNY: We didn't have any bias  that was discernable for  the
paradichlorobenzene. But, we were dealing with concentrations that varied from
150 ppb down to about 11 ppb. And over that range, because paradichloroben-
zene has a very high absorption coefficient, we can see it easily with the FTIR
system, and therefore we had,  I think, a good comparison.
                                                                     581

-------
           REMOTE DETECTION OF ORGANICS USING FOURIER TRANSFORM INFRARED SPECTROSCOPY*
        Jack C. Demirgian and Sandra M. Spurgash
                 Analytical Chemistry Laboratory
                    Chemical Technology Division
                     Argonne National Laboratory
                          9700 South Cass Avenue
                              Argonne, IL  60439
                                   (708)972-6807
ABSTRACT

Fourier transform  infrared  (FTIR) spectroscopy
is an ideal  technique  for remote detection of
organic emissions.  There is an atmospheric
window in  the  1200 to  800 cm"1 region, which
corresponds  to  the "fingerprint" region for
organic molecules.  Virtually all organic
molecules  have  a unique absorption/emission
pattern in the  fingerprint  region.  A remote-
passive FTIR relies on ambient emission of
infrared energy from organics to obtain
spectra.   The  instrumentation consists of inlet
optics, an interferometer,  a mercury cadmium
telluride  (MCT) detector, and an on-board
computer.  The  transportable unit measures 40
cm by 50 cm and has been used to collect data
while mounted on a helicopter or ground
vehicle.   Through  the  use of this FTIR combined
with least squares software, it is possible to
analyze qualitatively  and quantitatively for
organic vapors  from either  the air or ground.

The data presented will include quantitative
releases of common organics present in
incinerator stacks, hazardous wastes, and
illegal laboratories.  Data will be presented
for pure compounds, mixtures, and target
analytes in  the presence of interfering
compounds.  The sensitivity, reproducibility,
and the potential  of the technique will be
discussed.
INTRODUCTION

The emission of organic vapors is a concern for
environmental, health,  safety, and regulatory
reasons.  Sources of organic vapors include
industrial leaks, incinerator emissions, motor
vehicle exhausts, evaporation from contaminated
areas, leaking storage tanks, petroleum
refineries, and even illegal drug laboratories.
The EPA-certified procedures for organic
emissions consist of sampling a fixed amount of
air through a sampling apparatus such as a
volatile organic sampling train (VOST), which
traps the organics.  The sample is then
transported for gas chromatographic (GC) or gas
chromatographic/mass spectroscopic (GC/MS)
laboratory analysis.  Results are available in
a period of weeks or months.  The entire
procedure is costly and time consuming.  Also,
the GC or GC/MS analysis is a one-time
procedure.  If the concentration of organics is
outside the acceptable instrumentation limits,
the sample cannot be reanalyzed.

Remote-passive FTIR offers  the potential to
detect, identify, and monitor emissions in real
time in the field.  FTIR spectroscopy is ideal
for remote detection.  There is an atmospheric
window in the 1200 to 800 cnr1 region, which
corresponds to the "fingerprint" region for
organic molecules.  Virtually all organic
molecules have a unique absorption/emission
pattern in the fingerprint  region.  Through the
use of FTIR combined with classical least
squares (CLS) or partial least squares (PLS)
software, it is possible to analyze
qualitatively and quantitatively for organic
emissions from either the air or ground.

The FTIR consists of infrared optics,
interferometer, MCT detector, on-board
computer, and external data collection system.
The remote FTIR used to collect the data
presented in this report was designed by the
U.S. Army and is designated as the XM21.  It is
extremely rugged and can collect data mounted
on a helicopter or tank or ground mounted.  It
can be programmed for target analytes and will
set off an alarm as soon as they are detected.
                                                 583

-------
 The instrumentation emits no energy but detects
 the natural emissions of organics due to the
 difference in temperature of the organic
 molecule from the background.  For example,
 organics emitted from an incinerator will be
 thermally warm relative to the sky or ground
 near the stack.   A solvent exiting from an open
 window or exhaust pipe will be warmer or colder
 than the building from which it is emitted.  On
 a sunny day when the background is warmer than
 the solvent vapor, a standard transmittance
 spectrum is obtained.  During the evening when
 the background is cooler than the solvent
 vapor, an emission spectrum is obtained.
 This paper will focus on two potential
 applications for remote FTIR spectroscopy: (1)
 the monitoring of incinerator emissions and (2)
 the detection of solvents emitted from a
 building,  such as an illegal drug laboratory or
 production facility in which solvents are used.
 RESULTS AND DISCUSSION

 Data are presented for three pure liquids,
 methanol (MEOH),  chloroform (CHC13),  and carbon
 tetrachloride (CC14),  to determine their
 detection levels  under laboratory conditions.
 The  liquids were  released in front of a
 blackbody that was maintained at 40°C.  The
 liquids represent common laboratory solvents,
 two  of which (CHC13  and CC14)  are also
 principal organic hazardous components (POHCs)
 monitored in incinerator emissions.
 1.    Analysis of Pure Components

 Methanol data were collected at flows
 corresponding to concentrations of 3.7, 7.6,
 13.9,  27.2,  and 34 ppm-m.  The IR spectra for
 the  data are presented in Fig. 1.  The
 quantitative results are presented in Table 1.
 EXPERIMENTAL

 1.   Method

 In  a typical experiment,  the distances from the
 target (brick wall,  blackbody,  field release)
 to  the FTIR were accurately measured.   A series
 of  releases of single components and mixtures
 was performed using  a vaporizer designed at
 this facility.  The  vaporizer was capable of
 converting a liquid  flow  to vapor,  which was
 released in front of the  appropriate
 background.  The flow from the vaporizer was
 determined using a hot wire anemometer.   The
 design of the vaporizer is described
 elsewhere.x

 2.   Concentration Units

 Data are reported in concentration-pathlength
 Units of ppm-m.   For example,  a concentration-
 pathlength  release of 1 ppm-m is equivalent to
 a release of 1 ppm over a width of  1 m.   Our
 release  width was 10 cm,  the width  of  the
 vaporizer.   Hence, when we release  10  ppm over
 a width  of  10 cm (0.1 m),  the pathlength
 concentration is 1 ppm-m.   A 1  ppm-m release is
 equivalent  to a  concentration of 1  ppm emitted
 from  a stack 1 m in  diameter while  the remote
 FTIR  is  collecting data across  the  plume.   The
 parts per million level for the data presented
herein was  determined by  converting the  liquid
 flow  to  cubic  centimeters  of vapor  and dividing
by the cubic  meters  of  air  released.
      Table 1.  Quantitative Data for MEOH Using a
              Blackbody Background.  Concentration-
              pathlength units are ppm-m.
Cone.

34.0
27.2
13.9
7.6
3.7
Detected
2 std
STD
27.7
15.7
9.9
STD
Detected
3 std
STD
27.6
STD
8.8
STD
Agreement  is  excellent  for  the 27.2 ppm-m
sample, with  deviation  less  than 2%.   The
deviation  was less  than 13%  for the 13.9 ppm-m
sample.  However, when  only  two standards were
used,  the  error  for the 7.6  ppm-m sample was
30%.   This error decreased  to  16% when a third
calibration standard was used.

Both CHC13  and CC14  are  of special  interest
because they  are monitored during an
incinerator trial burn.   Monitoring them in
real time  enables the determination of on-
stream destruction  removal efficiency  (ORE).
On-stream  ORE determination  would eliminate  the
need for a trial burn.

CHC13 and  CC14 are strong infrared absorbers in
the 800-700 wavenumber  region.   This is  beyond
the optimum region  for  the detector in the
XM21.  Hence, the sensitivity of the instrument
is lower for  detecting  these two components
than would  be the case with a detector
optimized  in  this region.
                                                   584

-------
Analysis of CHC13 consisted of  five  different
flows corresponding  to  concentration
pathlengths of 2.87, 5.12, 6.49,  7.85,  and  8.87
ppm-m.  The first and last values were  used for
the calibration curve.  The IR  spectra  obtained
are shown in Fig. 2.  All absorbances in  the
region 830-720 wavenumbers were used in
quantitation.  The quantitative results are
shown in Table 2.
  Table 2.  Remote Detection of Chloroform Using a
          Blackbody Background.  Concentration-
          pathlength units are ppm-m.
     Cone.
                   Detected
                                X Error
5.12
6.49
7.85
4.85
6.15
7.10
5.3*
5.2
9.6
The relative intensity  of  the  two  peaks
associated with  chloroform are clearly seen.
The absorbance in  the 1220 wavenumber region
was not used for the data  calculations shown  in
Table 1.  With a two-point calibration curve,
the percent error  was within  10%.   The
intensity of the 2.89 ppm-m absorption is
sufficiently strong  to  demonstrate sensitivity
in the high parts  per billion  concentration
range.

Data for CC14 showed similar sensitivity.   Five
different flows  of liquids were analyzed,  which
correspond to concentration pathlengths  of
3.47, 4.33, 5.77,  6.92,  and 8.36 ppm-m.   The
first and last values were used for the
calibration curve.  The IR spectra obtained are
shown in Fig. 3.   All absorbances  in the region
810-784 wavenumbers were used  in quantitation.
The quantitative results are  shown in Table 3.
                                The quantitative data are accurate  to  within
                                10%, as were the chloroform data.   The
                                detection levels are similar.  The  absorption
                                of CC14 is much sharper than that for  CHC13.
                                The relatively low resolution of  the
                                instrumentation gives the CC14 its  sharp
                                features.  At 2 wavenumber resolution,  the  fine
                                structure of most organic absorbances  becomes
                                evident.   At 4 wavenumber resolution,  the fine
                                structure is lost,  and the absorption  degrades
                                to a curve or straight lines.   The 4 wavenumber
                                resolution of the equipment employed in this
                                preliminary study limits the ability of the
                                software to identify and align peak absorbances
                                by their fine structure.  The absorbance in the
                                780-760 region is background carbon dioxide.
                                This absorbance is also present in, and
                                overlaps with,  chloroform.  Higher  resolution
                                data would show the carbon dioxide  as  sharp
                                bands superimposed on the chloroform
                                absorbance.

                                The laboratory work with CHC13 and CC14
                                demonstrated the potential of FTIR  for remote
                                detection of organics.  The next phase
                                consisted of determining the efficiency of  the
                                instrumentation in the field.   An experiment
                                was set up to detect MEOH released  in  front of
                                a brick wall.  This experiment simulates the
                                detection of organics emitted from  production
                                facilities or illegal drug laboratories.  Data
                                collection is made more difficult because the
                                temperature of the brick wall and,  hence, the
                                intensity of infrared energy being  emitted  are
                                changing during the day.

                                Data collected for MEOH are presented.  The
                                three MEOH flows used correspond  to
                                concentrations of 8.2, 18.3, and 27.8  ppm-m.
                                The spectra for these three concentration
                                ranges are shown in Fig. 4.  Because the brick
                                wall was cooler than the released vapor
                                emission, spectra were obtained.  The
                                concentration of the second sample  was
                                calculated to be 22.5 ppm-m, or approximately
                                23% above the actual value.
      Table 3.  Remote Detection of Carbon
              Tetrachloride Using a Blackbody
              Background.  Concentration-
              pathlength units are ppm-m.
      Cone.
      4.33
      5.77
      6.92
                     Detected
3.92
6.28
7.58
                                 Error
9.5Z
8.8
9.5
2.   Analysis of Mixtures

A critical issue in demonstrating the potential
of remote FTIR  is  the  ability  of  the
instrumentation to function  in complex
environments.   The technique must be  able to
identify and quantify  components  in mixtures
under difficult and changing backgrounds.
Usually, components in mixtures absorb infrared
radiation at different energies.   The
difference was  readily observed in a  simple
experiment in which 10 uL  of MEOH and 5 uL of
ethyl ether were injected  into an evacuated 10
cm cell placed  in  front of a blackbody
background.  The spectral  data are presented  in
Fig. 5.
                                                   585

-------
a.  Laboratory Release

Qualitative and quantitative data were obtained
for a mixture of CHC13 and CC14.  These tvo
analytes have partially overlapping absorbance
peaks.  This work simulates monitoring two
POHCs being emitted from an incinerator.  The
blackbody temperature initially was set at 41°C
and allowed to slowly increase  to 44° C by the
end of the measurements.  This  changing
background better simulates background
conditions found in an actual remote  situation,
where data collection begins in the morning and
as the day progresses the temperature
increases.

Data were collected at six concentrations:
0.40, 0.81, 1.45, 2.02, 3.03, 4.04 for CHC1,
and 0.34, 0.68, 1.23, 1.71, 2.56, 3.41 for
CC14.  CHC13 and CC14  were mixed 50:50 by
weight.  The lowest concentration level (<0.5
ppm-m for each analyte) was below threshold
detection level and is not replotted.  The IR
spectra of the other five solutions are shown
in Fig. 6.  The quantitative data are shown in
Table 4.  The 820-784 wavenumber region was
analyzed for CC14 and the 784-720 wavenumber
region was used for CHC13.
DEM in MEOH:  0.67, 0.98, 1.12 ppm-m DEM in
6.06, 8.82, and 9.72 MEOH, respectively.  This
was immediately followed by  three more flows,
which contained only MEOH:   5.93, 9.88, and
19.69 ppm-m.  The objective  was  to determine  if
the totally overlapped DEM could be identified
and quantified in the first  three mixtures and
not misidentified (false positive) in the last
three flows.  Figure 7 shows  the IR spectra of
pure DEM, pure MEOH, and the  mixture of both
components collected remotely.  One cannot
visibly detect the presence  of DEM in the
spectra of the mixture.  The  quantitative data
are presented in Table 5.  The 1090-1000
wavenumber region was analyzed for DEM, and the
1100-975 wavenumber region was used for MEOH.
   Table 5.  Remote Detection of a Mixture of MEOH and DEM.
           Concentration-pathlength units are ppm-m.

Cone.
8.82
5.93
19.69
MEOH
Detected
8.30
7.11
16. 46

Error
5.9X
19.9
16.4

Cone.
0.98
0
0
DEM
Detected
0.91
ND
0.17

Error
7.1Z


 Table 4.  Remote Detection of a Mixture of CHC13 and CC1,.
         Concentration-pathlength units are ppm-m.
Cone.
0.40
1.45
3.03
CHClj
Detected
0.70
1.17
3.06
Error
75*
19
1
Cone.
0.34
1.23
2.56
CC1,
Detected
0.59
1.00
2.58
Error
74*
19
1
 The second, fourth, and sixth samples were used
 as calibration standards.  The software was
 able to identify the CHC13 and CC14 and
 quantify their concentrations with the
 calibration curve.   The large deviation for the
 lowest level is not unexpected.  One cannot
 visually identify CHC13 and CC14  in this
 spectrum.   The capability of the software to
 identify the analyte under these conditions is
 encouraging.  The steadily increasing
 background temperature did not result in a
 degradation of the data.
 The absorbances of CHC13 and CC14 only
 partially overlap.  It is necessary to study a
 system in which the absorbances of both
 components completely overlap.  A diethyl
 malonate (DEM) and MEOH mixture was studied at
 six different flows.  First, three flows were
 studied, which contained a low concentration of
 The first,  third, and fifth samples were used
 as standards.   The software was able to
 correctly quantify the low-concentration DEM
 sample.   The first pure MEOH sample was also
 correctly identified, although the quantitative
 data showed significantly more error.  The
 highest  concentration MEOH sample, 19.69 ppm-m,
 showed a low concentration of DEM present
 (false positive).  However, the concentration
 detected was below the threshold detection
 level of DEM.

 b.   Field Release

 Field data were obtained at Aberdeen Proving
 Ground.   The XM21 was  placed approximately 500
 ft  (~200 m) from the region where SF6 was
 released.  The angle of view, low sky
 background,  is the most difficult to work with
 because  of the infinite pathlength and greater
 amount of atmospheric pollutants.
 A methanol-DEM mixture was released while an
 SF6 release was in progress.  The raethanol
 peaks were observed as emissions because they
 were released at 42°C, which  was above  ambient
 temperature.  The SF6 was observed as an
 absorbance spectrum because it was released
 from a pressurized tank and the gas was below
 ambient temperature.
                                                    586

-------
The spectral  data are presented  in  Fig.  8,  and
the quantitative data are presented in Table 6.
The spectral  data show the steadily increasing
concentration of the methanol-DEM spectral
features and  the reduction of  the SF6
absorption  as the gas disperses.  The  primary
band for DEM  is completely overlapped  by the
MEOH emission (Fig.  8).  The secondary bands at
1200-1150 wavenumbers are readily observed.
   Table 6.  Methanol-DEM Release with SF6 Dispersion
           in Lov Sky Background.  All concentrations
           are in ppm-m.
                    MEOH
         Cone. Released   Detected  Z Error
4.6
11.7
17.4
23.5
STD
8.2
STD
25.5

29.9
-
9.4
                     DEM
        Cone. Released
                       Detected  X Error
2.2
5.8
8.5
11.6
STD
3.9
STD
12.5

32.8
_
7.8
 The quantitative  data in Table 6 were obtained
 using only two standards.   The percent error
 was approximately the same as that obtained  for
 pure methanol.  However, the analysis was  more
 difficult because only the spectral range  of
 1125-975 wavenumbers  was used.  The DEM  and
 methanol completely overlap in this region.   No
 pure components were  entered into the
 calibration file.  Hence,  these data indicate
 that the analysis of  mixtures is no more
 difficult than the analysis of pure components.
 SUMMARY AND CONCLUSION

 This study has demonstrated that remote
 infrared detection  is  a precise and reliable
 technique for monitoring organic emissions.
 The equipment is  capable of detecting  SF6
 releases at 500  ft  (-200 m) and low
 concentrations of pure components and  mixtures
 released in the  environment.  Quantitation was
 within 30% for these releases.  Mixtures were
 no more difficult to analyze than pure
 components.
 Several areas  still must be addressed.   The
 limitation of  the PLS software is  the  large
 number of spectra required to reduce
quantitation error.  For the data presented
here,  it was not possible  to obtain the number
of spectra  that would reduce the  quantitation
error.  The classical least squares method
requires fewer library data and should improve
data quality.   We hope to  expand  this work to
include advanced signal processing using
digital filtering in the time  domain so that
variations  in background and the  need for
calibration spectra are eliminated.  A
limitation  of the existing equipment is the
lack of front-end optics,  which are required
for analysis at distances  of 1 km.
BIBLIOGRAPHY

1.   J.  C.  Demirgian and  S.  M.  Spurgash,
     "Remote Detection of  Chemical Agents by
     Infrared Spectroscopy," ANL/ACL-90/1
     Progress Report 1989.
*Work  supported by  the  U.S.  Department of
Energy under Contract W-31-109-Eng-38 and the
Chemical Research Development  and Engineering
Center.
                  I I 00            10IIII
                      IfAVENUMDEUS
     I in I. llctnolu III S|h.-clri\ i.f MROII ni»lainr,.'7'-', am! :n ,.|.iii..ii.

            KKMOTK IIKTK*-Mi IN 111-' i;ill.OKOI--(iKM
          I ;•(!()     I Mill     IIIIIO     !l(l(l      IIIKI
                      1VAVKNUM OEKS
    iK 2 llmiiol- Ill Spi-Un ..fCHCI, OliUliml ll.lnu .. Ill.u-Llulr |]i..k|[r..ii,,il. C.»«-onlc»l»i
       ..i run-, w,-i< ».ii. r,.\,11.5, T.ti, ..n-i «•> ,T...,.
                                                    587

-------
         RKMUTK  MKI'Kt "I'll IN  i 11-' UAKIMlN  TKTRACI M.tlKl I)K
                                    n;:u          MOD
                                  IV,\\'I-;NUMMKIC:S
                                                                                                                                       -rrKt TIHM  ••!•• ,\   MIXTMUK UK cnci.n  AMU  i:i;i.
                                                                                                                         Fi,;. -,    llcm.iU- 1 i,  i.-!,., ,  ,.| ., r [,.i.,i,-  >! Cl.i'l,  ..ml CCI j nl 1'ivc i1,,,,,,-,,',,,!,,,
                           lion                       I nnil
                                   WAVKNIIMMKltS
[ .,   I.    Hi-iM-ili: in S|«-Llfn . I  .' II r l[l OliliiiliL-il  lu,,,,; » 111 irk. \Vlill ll.i< LI;..,-.H.|  Ciil
         .il Mi:nn w. i. a .', ia..i, ,,i,.i '2~.n ,,|..n n,
            Rcniule Dclctlion uf Pure MCOII. DEM, and i Mijlore  ot Dt.M tnj MEOII.
                                                                                                                     M  FIELD,  Sl'O  RULI2ASE  ITITII  2:L  MUOII/DliM  .|-20  PPM-h
      (;	,,,,i,,,,, ,,f iKiiniK III S|.Tlr	C.,i,l.»:,iic.|( III /.I. «' MKOII
      MliOll wllli « 5 ,1. A	I «( lill.jl lil.« Ad.tal 1.1 .  10 c... O.ll
                                                                                                                                            1100                  JOOO
                                                                                                                                                  WAV1SNUMDERS
MD30                        ROS=  -I
Fit. «    MEOII-DEM Hclf»^ .III, Uw Sly  U
         [foul Llif liiilruinrnuiioii.
                                                                                                  588

-------
                                                             DISCUSSION
EMILE HYMAN:  Can you comment  on the potential for stack emission
monitoring?

JACK DEMIRGIAN: We think there is tremendous potential forstack emission
monitoring, and that's one reason we ran carbon telrach lore form as an early set.
We are currently collecting, with the DOE Program, regular on-stream incineration
monitoring data. Once we get that data we will be giving it to Kroulil so that he
can calculate the coordinates for it. And then we can start to test this technique
on incinerations. There is a municipal incinerator on-site or near on-site, at the
Aberdeen Proving Ground. We have already collected some data there, before
some of the signal processing work was completed, and we wilt be analyzing that
data. We will be definitively moving, and hopefully we'll be presenting some on-
stream incineration monitoring  data at the incineration conference this  May.
Hopefully, after that we will present the remote equivalent of the on-stream
monitoring. So, stay tuned.

EMILE HYMAN: How about SOx and NOx and that kind of thing?

JACK DEMIRGIAN: Again,  SOx and most of the NOx absorb out of the
detector window for this particular detector. SOx  and  NOx are a little  more
difficult because they're between CXh and water in their absorption region. And
so the digital filler has to be done to eliminate water, or certainly reduce the effect
of water. This is not undoable. The technology is right there now. The new plateau
I discussed by Kroutil and Ditillo and Small, gives the potential to do that. 1 think
it's just more a function of time and staff than it is technologically difficult. Right
now the emphasis has been on organics as opposed to the SOx and NOx. There
are several good SOx and NOx monitors and we don't really want to compete. We
want to open up new areas. The new Clean Air Act has opened up a lot of areas
in which I think this remote passive technique will fill the bill. And I think FTIR
is right now  about  the only technique  available that can  do some of the
requirements of the new Clean Air Act.

DONALD GURKA: Those who know me  know me to be skeptic. You said no
false positives and no false negatives? That suggests that the systems that you
looked at thus far are simple. Is  that your conclusion?

JACK DEMIRGIAN: Yes, well, first off,  keep in  mind that the sophisticated
algorithms have just been developed this past September. What we have been
doing now is  with conventional techniques. You're absolutely right. The first
system we  worked with were pure components. Then we  worked with two
component mixtures. We've worked with two component mixtures now with
some varying background. In this year we are funded to go ahead with  more
complex backgrounds and more complex  mixtures. In fact, we have ordered the
equipment so that we can now make multi-component  mixtures and better
characterize the complex mixtures that would put more of the false positives and
false negatives to the lest, but your point is quite well taken. We presented data
with one and twocomponenl mixtures. The fact that we' ve chosen a difficult two
component mixture is a good sign of things to come. Had we failed miserably
with the methanol DEN, then I think you would have a  very valid statement. I
think when we had succeeded with the very difficult methanol DEN case we have
justified going on to more complex mixtures. Again, the quantitative remote
passive as we're doing now is very new. And I guess if I'm here at the next
conference, you'll know that it worked.

DONALD GURKA:  Can you visualize slanting this approach to all false
positives or all false negatives? Can you adjust your approach so that you screen
out only negatives or you screen out only positives?

JACK DEMIRGIAN: That's an interesting question. And I think that the key
on doing the false positives and false negatives is probably going to reside with
Kroulil's ability to digitally filter these things out. They have specific expertise
in filtering out false negatives, and they have a very great interest in filtering out
false positives. In order to do that properly with very complex mixtures, it is
better to ask Ditillo or  Kroutil, because that's their specific expertise.

ERNIE TUAZON: You allude to methanol being a simple system. It isn't.
Underneath methanol,  if you look at it very closely, there would be interferences
by ammonia. So, if you are in an area where there's fertilizer or a factory, or even
ammonia producing cows are around, it will be an interference. Also, CO2 is an
interference. These are the so-called lacing lines of CO2 that you don't see in the
laboratory, but at long parts you will see them. It's underneath those. Also, if you
are in the Los Angeles  atmosphere, ozone will be an interference. So, it's not a
simple one as far as that is concerned, anyway.

JACK DEMIRGIAN:  We  specifically addressed the CO2 by  collecting
chloroform data which totally enveloped the CO2. The Army has been working
on the ozone problem  for quite a bit of time, and that's within their coordinate
system to filler out. Now, I don't think we have done ammonia yet or high
concentrations of...

ERNIETUAZON: No, I don't mean that. I mean just in analyzing that particular
band that you see, there are a lot of interferences underneath that.

JACK DEMIRGIAN: Yes, ammonia itself has relatively sharp bands and the
algorithm  is able to discern a sharp band versus a smooth band. And I can show
you some data.  If we get into the THAMA data  where we're looking for
explosives which are nitrates, they  absorb smack in the middle of the water
region. So, if you've got an interference, it's water in the soil samples. And we've
spiked the soil samples  with 10% water, and looked at ppm explosives. Even with
the very sharp water bands versus the relatively broad explosive bands it does not
affect the  algorithms'  ability to quantify. We were very satisfied with these
results. A lot of the atmospheric gases that are small molecules have very sharp
bands, and the  organics have  much broader bands and that makes quantitation
and identification a good deal easier. So,  as I said, methanol was relatively
straight forward. Most of the atmospheric  gases are not going to be as big a
problem. I would guess in an  application such as treaty verification where you
have someone deliberately trying  to fool your system and put out components
that are very, very similar, that might be a tougher test. But, atmospheric is not
as bad as you would think.
                                                                      589

-------
           INTERPRETATION OF PPM-METER DATA FROM LONG-PATH OPTICAL MONITORING
             SYSTEMS AS THEY WOULD BE USED AT SUPERFUND HAZARDOUS WASTES SITES
                                               Thomas H. Pritchett
                                       U.S. Environmental Protection Agency
                                                    Edison, NJ
                              Timothy R. Minnich, Robert L. Scotto and Margaret R. Leo
                                              Blasland, Bouck & Lee
                                                    Edison, NJ
 Recently, several groups have been evaluating the use of
 long-path optical monitoring systems such as remote sensing
 UV and remote sensing FTIR. Because of the potential
 power of this field analytical technique, several of the more
 active groups have been attempting to compare the results
 from the long-path systems to Summa canister results. While
 these comparisons have generally demonstrated that the two
 techniques are  indeed comparable, they generally have not
 addressed how  the long-path (or path-integrated) data could
 be interpreted to meet the site manager's needs. Unfortu-
 nately, many in the air toxics field are not familiar with the
 downwind dispersion equations which can be used in con-
junction with path-integrated concentrations to define the
emission source to downwind receptor relationship. Instead,
most are usually used to interpreting only average concentra-
tion data from single locations.
The fundamentals of downwind transport illustrate how
ppm-meter data can be interpreted in three of the most
appropriate applications of long-path optical monitoring:
assessing of the baseline air emissions from an inactive site,
assessing the air impact of future cleanup operations during
pilot scale testing and fenceline monitoring during actual
cleanup operations. In all cases, the path integrated data can
be used to predict "worst-case" air concentrations for the
surrounding area — even for areas not covered by the
original monitoring study.
                                                      591

-------
                                                             DISCUSSION
DONALD (JURKA: My question is on regulatory aspects that are driven by the
technological slate of the art. As far as emissions go, the state of the art is usually
or always point sampling approaches. The question is, are there any gaps in the
point sampling approaches which are covered by open path  IR, thus that open
path IR in that situation is now the state of the art?
TOM PRITCHKTT: There are two types of regulatory approaches. There are
air  toxics  people who work  with point monitoring. Those point monitoring
techniques assume  either a  time averaged,  a long term  exposure average
concentration,  or they assume  a maximum concentration. Well, the point
monitoring approach is not a trivial matter to ensure that you're taking your point
monitoring sample at the maximum point in the plume. You have to know where
that maximum is before you take your sample. I've actually been out with a
mobile mass spec and was driving down the highway and watched a regulatory
agency take a measurement, a grab sample, that they're going to use to slam on
a company. Well, I drove another 50 yards down the road and found the plume
from (he facility, which they were actually trying to go after. They were taking
a sample in the upwind plume. So you've got to be very, very careful when you're
doing point monitoring to determine regulatory compliance. You can very easily
miss the maximum plume.

At  this time, if you're trying to regulate a particular emission  source based upon
a long-term average point monitoring concentration,  there  is a major gap in
guidance  in how to interpret that data. Particularly, what do you do  with
nondetects. Do you use a source receptor relationship  and treat the upwind
nondetects as zeros, or do you use detection limits? That's particularly important
for risk assessment because the one-lo-two order of magnitude difference
between ihc instrument detection limits and the concentration of concern tends
to mask its risk.

The other type of regulatory situation is the one that is used in the Clean Air Act.
They regulate in terms of emission rates, grams per second. You cannot exceed
so many grams per second. I don't know what approach they're going to take in
the Clean Air Act, but I think ultimately they're going to have to go to grams per
second. To calculate an emission rate using point monitoring techniques under
simple gauging conditions takes anywhere from seven to ten Summa canisters,
and that's to get one reading which you then have to do in triplicate. So, that's 21
to 30 samples just to  get one measurement you feel that you can use as an
emission rale.

With open path monitoring you can do that in probably about  three minutes. You
may very well be driven not so much by the detection limit — how many spectra
you have to co-add to gel your detection I im il, as much as how many spectra  must
you require before you have the gauging conditions which you're using to
interpret the data.

TOM PRITCHETT: The answer is that there's a lot of gaps in point monitoring
techniques which make it not as ideal for, let's say, health assessment purposes
related to specific source, and for determining emission rales related to a given
source that I think open path monitoring can solve.

GARY ROBERTSON: Tom, while I was listening to you I noticed that you were
doing most of your measurements perpendicular to the plume. It looked to me
that if you took those measurements at various angles to the plume, you could get
a lot of information about defining the plume and perhaps even the shape of the
plume.
TOM PRITCHETT:  Under the straight gauging condition, if you're  near
perpendicular, say  within about 30 degrees perpendicular, you can use the
meteorological datalodefine the plume. It's a lot easier, believe it or not, lodefine
the plume using your meteorological conditions, as was shown by the Kansas
Slate study, than it is to try to multiplex by burying the angle that you're shining
the beam. It's also logistically a lot easier lo have one beam set up and just filter
out the data where you're essentially perpendicular, than it is to sit there and
continuously move your beam to try to gel your multiplex beam orientation.

CLIFF DAHM: I wanted to explore a couple of things. One is you talk about
meteorological measurements and conditions. What do you recommend routinely
be measured meteorologically, for example at a fence line? And secondarily, I
also want to know, whether, on any of these studies that you've been talking
about, if there's been vertical structure determined at some of the plumes that
you've been monitoring?

TOM PRITCHETT: In answer to the first question, you typically look for wind
speed, wind  direction and sigma theta to calculate the stability class. And in
answer to the second question, looking at the vertical component of the plume,
no, we haven't. Again, we're looking downwind of small point sources. The one
thing we have done in that relationship is we actually used controlled releases.
Let me go back to one of my questions here; look at the bottom equation, equation
5.1t you have a control led release, and you're measuring the path of concentration,
and you know wind speed, the only thing that's unknown is sigma 9. What we've
done  is used controlled releases at different distances to see whether or not the
sigma O's were, consistent with the predicted sigma 6's of the Clifford Path —
I guess it was Clifford that measured it or Pascal. But anyhow, we found  very,
very good agreement for that stability class of the sigma 6's. So, in essence we
have not directly measured vertical dispersion, but we've shown that the vertical
dispersions being used in the gauging equations were experimentally confirmed.
                                                                         592

-------
                                                     AWARDS  CEREMONY
The sponsors of Second International Symposium — Field Screening Methods for
Hazardous Wastes and Toxic Chemicals were pleased to include an awards program.
Mr. John Koutsandreas, Florida State University and the Symposium Executive
Secretary organized the program and assembled the review panel that evaluated
nearly 60 platform presentations and over 60 poster session papers. The members of
the awards committee included:

Mr. Robert Booth, former director of U.S. EPA's Cincinnati EMSL
Dr. Steven Levine. University of Michigan, School of Public Health
Dr. David Nelson, Vice President.  Perkin-Elmer Corporation
Dr. Roy Herndon, Director, Chemical, Biological and Toxicology Research, Florida
      Stale University
Dr. Michael Dellarco, U.S.  EPA, Office of Research and Development
Dr. Russell McAllister, U.S. EPA, Office of Solid Waste and Emergency Response
Dr. Joseph Leone Hi, Associate Director, Applied Electromagnetic and Optics Lab,
      SRI International
Mr. David Bottrell, Department of Energy, Office of Technology Development
Dr. Richard Tinlin, Geraghly & Miller, Inc.

The Symposium organizers and sponsors arc grateful to this awards committee for
their time and effort expended in evaluating the presentations.

The panel judged the two best private sector (i.e..non-Federal) papers and the two
best Federal papers and awarded U.S. EPA engraved plaques to:

Private Sector
Susan Eberlein for "Space Technology for Application to Terrestrial Hazardous
Materials Analysis and Acquisition"

Hui Wang for "Comparison of Aqueous Hcadspace Air Standard Versus Summa
Canister Air Standard for Volatile Organic Compound Field Screening"

Federal
Donald Smilh for"A Study of the Calibration of a Portable Energy Dispersive X-Ray
Fluorescence Spectrometer"

Tom Spinier for "The Use of Field Gas Chromulogruphy to Protect Groundwater
Supplies"

U.S. EPA engraved plaques were awarded to Ihc two bcsl poster presentations as
determined by the Awards Committee. They were:

"A Field-Portable Supercritical Fluid Extractor for Characterizing  Semivolatile
Organic Compounds in Waste and Soil Samples" B.W. Wright  and J.S. Fruchter

"Real Time Detection of Biological Aerosols" P.J. Stopa. M.T. Good, W. Zulich,
D.W. Sickenburger. E.W. Sarver, R.A. Mackey

Mr. Larry Cottran from Hewlett Packard presented two eagle trophies, donated by
Hewlett  Packard, for overall outstanding technical contribution and quality of
presentation. Hewlett Packard pays  considerable  attention to these two critical
elements: improving quality and increasing technical contributions.  The winners
were:

For best technical contribution:
Fred Milanovich for "A Fiber Optic Sensor for the Continuous Monitoring of
Chlorinated Hydrocarbons"

For best presentation:
Steven Levine for "Fourier Transform Infrared Speclropholomctry for Monitoring
of Contaminant Gases and Vapors  in the Workplace Air"
Certificates were also presented that recognized the most outstanding paper in each
of the ten platform sessions. They were:
Session 1
CHEMICAL SENSORS
Fred Milanovich for "A Fiber Optic Sensor for the Continuous Monitoring of
Chlorinated Hydrocarbons"

Session 2   ION MOBILITY SPECTROMETRY

Suzanne Ehart Bell for "Hand-Held GC-Ion Mobility Spectrometry for On-Site
Analysis of Complex Organic Mixtures in Air or Vapors over Waste Sites"

Session 3   ROBOTICS

Susan Eberlein for "Space Technology for Application to Terrestrial Hazardous
Materials Analysis and Acquisition"

Session 4   QA AND STUDY DESIGN

John Mateo for "A Quality Assurance Sampling Plan for Emergency Response
(QASPER)"

Session 5   AIR PATHWAY MONITORING AT SUPERFUND SITES

Steven Levine for "High Speed Gas Chromatography for Air Monitoring"

Session 6   FIELD MOBILE GC/MS TECHNIQUES

Mary Cisper for "Field Measurement of Volatile Organic Compounds by Ion Trap
Mass Spectrometry"

Session 7   PORTABLE GAS CHROMATOGRAPHY

Hui Wang  for "Comparison of Aqueous Headspace Air Standard Versus Summa
Canister Air Standard for Volatile Organic Compound Field Screening"

Session 8   FIELD SCREENING METHODS FOR WORKF.R SAFETY

Gerald Moore for "Improvements in the Monitoring of PPM Level Organic Vapors
with Field  Portable Instruments"

Session 9   X-RAY FLUORESCENCE

Donald Smith for"A Study of the Calibration of Field Portable X-Ray Fluorescence
Instruments"

Session 10  FOURIER TRANSFORM  INFRARED SPECTROMETRY &
OTHER SPECTROSCOPY METHODS

Gary Small for "Pattern Recognition Methods for FTIR Remote Sensing"
                                                                    593

-------
              CONCLUDING REMARKS BY SYMPOSIUM CHAIRPERSON,
                                  DR. LLEWELLYN WILLIAMS

As I lay in bed this morning, I looked back over the week as each of my senses awakened. First came my sense of touch, and
I felt that the Symposium was a success, and I was touched by the quality of the papers and of the posters. And I was almost
"touched" by a number of technology developers looking for Federal funds. And I recall the pain of stabbing my upper lip with
a toothpick holding two Swedish meatballs.
Next came my hearing. I heard a broad range of useful information from the bureaucrats. I heard of breakthroughs and research
advances from our researchers and technology developers. Fortunately, I heard few complaints about the papers. And I heard
the sound of two thousand Swedish meatballs being poured  into a silver chafing dish.
Next to return was my sight. I saw a lit entry way and a table full of awards. I saw dim images on the screen during the Plenary
Session. I saw colorful exhibits and the sharp graphics of poster and platform sessions. And I saw two thousand Swedish
meatballs being poured into a silver chafing dish.
The next of my senses to return was that of smell. I detected the sweet smell of success that could be attributed to the enthusiasm
and energies of you, the participants. I smelted the various emergency deodorizing measures used on Wednesday morning when
the shower water didn't work in the hotel. And I smelled the odor of over two thousand Swedish meatballs simmering in a silver
chafing dish.
The last sense to return was my taste. I recall the good taste displayed by the exhibitors during our reception period. I can taste
a consistently good coffee that was provided on our breaks. And I fear I'll continue to taste the Swedish meatballs for days to
come.
The results of all of this sensory input was a series of visions. The first is a vision of us all returning to Las Vegas two years
from now to do it all again. The second vision is of the widespread acceptance of field methods and the data derived therefrom.
And finally the vision of fifty large Lutheran women feverishly molding Swedish meatballs.
At this time, as unprepared as I am, someone asked for another poem.
  Now that you have seen it all and will set upon your way.
  We'd love to get your feedback as we plan for number tres.
  How'd you like the balance and the papers and the rest?
  What things would you change?
  What did you like the best?
  Thanks to all the many folks  who made this whole thing happen,
  And bailed out the Chairman every time they caught him  napping.
  And thanks for sponsor monies, and support from all the brass.
  It helped us build a program that was nothing hut first class.
  My special thanks to Eric andJoAnn, and yes, to Kouts,
  For pulling things together so give your horns some toots.
  And if we had success in our attempts to make it work,
  It was your participation here that really made it perk.
  So, looking to the future 1 suspect we'll meet again.
  As we catch up on developments in monitoring, and then
  We'll see if the technologies have made it to the play off
  Can bear the fruit, and stand the test, and over time will pay off.
Ladies and gentlemen, thank you  very much.
                                                    595

-------
                                     CALIBRATION OF FIBER OPTIC CHEMICAL SENSORS
           W. F. Arendale and Richard Hatcher
         Laboratory for Inline Process Analyses
           Kennelli E. Johnson Research Center
        The University of Alabama in llunlsville
                 HuiiLsvillc, AL l.vS'jy
                                                 and
                                            Bruce Nielsen
                                         H(| AFESC/RDVW
                                    Tyndall AFB, FL 32403-6001
Fiber oplic chemical sensors to be used for monitoring envi-
ronmental pollutants have been extensively researched lor at
least ten years. Although excellent research results have been
presented, few if any systems have met the quality assurance
requirements such that they are now in production and avail-
able for general use.  Many fiber optic sensors for monitoring
physical parameters as temperature,  microslrain, and accous-
tics are available.  Sensors for monitoring chemical processes
have been used successfully.

Sensors that meet QA/QC manufacturing requirements are
usually the result of careful modeling.  For the past several
years a part of our research activity has been related to the
formulation of appropriate models for FOCS. Our studies
have shown that the  chemistries of the materials placed on the
distal end of the fiber are adequately modeled.  The deficien-
cies appear to be related  to the  physics and engineering ol the
optics, lack of quality control during the  manufacturing proc-
ess, and/or lack ol sufficient information being collected to
assure reliable information in the presence of intcrlcmils.

Optical fibers are used as:

        (1)  Carriers ot optical  signals - photons travel simul-
            taneously in many directions.
        (2)  Sensor/Carriers  - Optical properties of
            provide the sensing medium
liber
        (3) Components of integrated diagnostic systems.

Photons pass down the core of a fiber in several ways,  hi
addition to the rays passing down the center of the fiber some
of the. photons are reflected at the boundary between the core
and the cladding. The cone that  includes rays that pass
through the fiber is shown in Figure 1A. Energy can be lost to
the cladding when a bend occurs in the liber  as shown in
Figure IB. To avoid loss due to microbends and to protect the
fiber from stress, the clad fiber is bundled into a protective
cable shown in Figure 1C.
                           Figure I. Light Transmission Through Fibers

                   The communication industry has spent millions of dollars on
                   obtaining  high purity silica core, selection and application of
                   the cladding, and cabling of the liber.  Some of the problems
                   that must be overcome are shown on Figure 2.
         Figure 2.  Sources of Energy Loss in Fibers

 Our recommendation for quality sensors arc:

         (1) Use the purist high quality fiber
         (2) The cladding lor high quality liber is added to
            the  core  while  the fiber  is still  in the  inert
            atmosphere  of the  drawing furnancc.  Do  not
            attempt to change the cladding except doing the
            manufacturing process.
         (3) If possible use only cabled  or rigidly supported
            fiber.
         (4) Standardize the sensor using a sufficient number
            ol measurement parameters such that all mean-
            ingful variance is represented. It is not necessary
            to  quautitate all sources of variances,  but a
            measurement parameter must be  included.

l-iguie 3 shows a test chamber that we have used successfully
lor determining the required number of measurement parame-
ters and standardizing our sensors.
                                                               597

-------
                       .    	    ,
                       c n; n :ir z. s>n. cn< j

Figure 3. Inline MuHivariate Analysis Flow Apparatus
                                                                               598

-------
       GAS-CHROMATOGRAPHIC  ANALYSIS OF SOIL-GAS SAMPLES AT A GASOLINE-SPILL
      R.J. Baker, J.M. Fischer, N.P. Smith, S.A. Koehnlein, and A.L. Baehr
        U.S. Geological Survey, 810 Bear Tavern Rd, W. Trenton, NJ 08628
ABSTRACT

The U.S. Geological Survey is studying
remediation processes at a gasoline-
spill site in Galloway Township, New
Jersey.  A field-laboratory trailer was
equipped with a gas chromatograph  (GC)
configured to analyze soil-gas samples
for gasoline hydrocarbons and inorganic
gases, such as oxygen, nitrogen, carbon
monoxide, and carbon dioxide.  This
instrument was selected over other
analytical options because of its
versatility; it can be used to monitor
all significant organic and inorganic
components of unsaturated-zone gases.
Each of two chromatographic columns is
equipped with a vapor-sample injection
valve fitted with a sample loop.  The
sample-loop volume determines the
injection size.  A chromatography data
system and a micro-computer are used for
data acquisition, processing, and
storage.

A thermal-conductivity detector is used
in conjunction with a 3.3-meter-long
molecular sieve column for analysis of
inorganic gases.  A flame-ionization
detector is used with a 30-meter-long
fused silica capillary column with a
dimethylpolysiloxane stationary phase
for analysis of vapor-phase gasoline
hydrocarbons.  Inorganic and organic
species are identified by retention time
and quantified by linear-regression
standard-curve analysis.

A method for evaluating hydrocarbon
chromatograms that does not require
identification of specific peaks was
developed.  Chromatograms are divided
into retention-time increments, each of
which contains peaks of compounds that
have the same carbon number (number of
carbon atoms).  A sample can then be
described semiquantitatively in terms of
the number of compounds of each carbon
number, total mass of each carbon
number, or percent of mass represented
by each carbon number.  The method is
based on the relation between the carbon
number of a compound and its boiling
point, and between boiling point and GC
retention time.  By using this method,
retention time can be used to determine
the boiling point and most probable
carbon number of an unidentified
hydrocarbon compound.  The margin of
error of the method was established by
determining the carbon numbers of 167
compounds from their boiling points.
Correct carbon-number assignments were
made for 131 compounds (78.4 percent),
and carbon-number was underestimated for
16 compounds (9.6 percent)  and
overestimated for 20 compounds (12.0
percent).  All over- and underestimates
were in error by one carbon atom.

The GC and the data-evaluation methods
used are providing excellent soil-gas
characterization during this field
study.  Chromatogram analysis by carbon-
number determination can be used in
other studies where hydrocarbons are
detected but not specifically
identified.
                                         599

-------
      SIGNIFICANT  PHYSICAL EFFECTS  ON SURFACE  ACOUSTIC WAVE (SAW)  SENSORS'
                                  David L. Bartley
               National  Institute  for  Occupational  Safety and  Health
                                4676  Columbia  Parkway
                               Cincinnati. Ohio 45227
 Surface acoustic wave (SAW)
 devices are presently being
 developed for applications in
 chemical sensing as  well  as  for
 polymer characterization.  SAW
 gas and vapor sensors have the
 potential for miniaturization
 and high sensitivity to a  wide
 variety of substances.  Polymer
 characterization is  applicable
 to  such diverse  fields as
 protective coating design  and
 decontamination  of polymers.
 Research was  conducted to  better
 understand  the physical
 mechanisms  behind SAW response.
 Practical problems as to film
 uniformity, thickness
 measurement and environmental
 controls such as temperature  and
 gas flow rates necessary in such
measurements were considered.
 The effects of elastic
 properties in comparison to mass
 loading of polymer coatings on
 SAW substrates were
 investigated.   A theoretical
 basis  for  the effects of vapor-
 induced swelling or of thermal
 expansion  was  established.
 Compressive tension and its
 effect  on  SAW frequencies were
 found to be simple to describe.
 if  there is no film slippage or
 polymer flow.   The response of
 quartz-substrate SAW crystals
 coated  with polycarbonate and
 polyimide  (glassy polymers)  upon
 exposure to  toluene  and methanol
was measured and  was  found
consistent with theory in
predicting effects of the order
of the ratio of coating to
substrate elastic constants.
                                Bartley.  D.L.  and Dominguez. D.D.:
                       of Polymer Coatings on  Surface Acoustic Waves."
      Anal.Chem. 62:1649  (1990).
                                       601

-------
       AN EVALUATION OF FIELD PORTABLE XRF SOIL PREPARATION
                                      METHODS
  Mark Bcrnlck
  Roy F. Weston, IncVREAC Contract
  GSA Raritan Depot
  Building 209 Annex
  2890WoodbridgcAve.
  Edison, NJ 08837

  Donna Idler
  Roy F. Weston, IncTREAC Contract
  GSA Raritan Depot
  Building 209 Annex
  2890 Woodbridge Ave.
  Edison, NJ 08837

  Lawrence Kaelln
  Roy F. Weston, IncTREAC Contract
  GSA Raritan Depot
  Building 209 Annex
 2890 Woodbridge Ave.
 Edison, NJ 08837
Mark Sprenger
USEPA/ERT
GSA Raritan Depot
Building 18
Edison, NJ. 08837

Dave Miller and Jayanti Patel
Roy F. Weston, IncTREAC Contract
GSA Raritan Depot
Building 209 Annex
2890 Woodbridge Ave.
Edison, NJ. 08837

George Prince
USEPA/ERT
GSA Raritan Depot
Building 18
Edison, NJ 08837
 INTRODUCTION

 The USEPA Environmental Response Team
 (ERT) has been using field portable X-ray Fluores-
 cence (XRF) spectrometers to characterize Super-
 fund and hazardous waste sites. An Outokumpu
 Electronics Inc.  (OEI) X-MET 880 XRF
 spectrometer equipped with a surface probe con-
 taining Cm-244 and Am-241  radio-isotopes was
 selected. Field portable XRFs have enabled the
 ERT to estimate the extent of metal contamination;
 support biological 'plant stress assessment" studies
 in tidal wetlands; and, support a health and safety
 assessment of the extent of contamination and pos-
 sible human exposure of a network of hiking trails
 and campsites contaminated by a smelting opera-
 tion.

 BATTERY BREAKAGE SITE

 In September, 1989 the ERT deployed a XRF unit
 to perform an extensive post  remedial site lead
 survey at an abandoned battery reclamation site.
 Analyses were planned  for both surface and sub-
 surface soil samples to determine the extent of lead
 contamination.  The  OEI X-MET 880 XRF
spectrometer was calibrated with a suite of 18 site
specific lead standards by an analytical chemist.
   An in-situ analysis method was preferred since it
   would reduce sample preparation time and the risk
   of personal exposure to the contaminants. A litera-
   ture search failed to find any studies demonstrating
   the correlation of results from an in-situ XRF
   analysis and the accepted method of preparing
   soils by drying and sieving. The analyses included
   29 in-situ and prepared soil samples in an effort to
   evaluate the  sample preparation methods.
   Samples  that were analyzed by the XRF  in-situ
   method were dried and sieved and reanalyzed by
   XRF.  The paired-difference  t-test was used to
   evaluate  the results of both sample preparation
   methods using a significance level of 5 percent.

   PROCEDURE FOR IN-SITU SURFACE SOIL
   LEAD XRF ANALYSIS

   All organic matter and large rocks were removed
   from the  area (8" by 5") to be analyzed. The area
   was then rendered flat with a stainless steel trowel
   The XRF instrument was initiated for a 60 second
   measuring time for the lead analysis while the sur-
   face probe was held flush  against the soil surface.
   The sample area number, location (A)  and XRF
   lead result were logged into a field notebook. The
   analysis was  then repeated selecting a different
   analysis location (B) within the prepared area. The
   results of the two analyses were  averaged and
   reported.
                                           603

-------
 PROCEDURE FOR DRIED AND SIEVED SOIL
 XRF ANALYSIS

 The soil within the prepared area was removed to a
 depth of a quarter inch. Large stones and organic
 matter were removed prior to drying.  The  entire
 sample was dried in an oven and sieved through 10
 and 20 mesh stainless steel sieves with a stainless steel
 spoon. All organic matter and stones were removed
 and discarded. The sample was homogenized for one
 minute by dividing  the sample  into quarters and
 mixing opposite sides together.

 A 31-mm, polyethylene X-ray sample cup was labeled
 and Tilled with soil. The cup was sealed with 0.2-mm
 thick polypropylene, X-ray window film.  Prior to
 XRF analysis, the sample cup was gently tapped
 against the table top three times to  pack  the soil
 evenly against the polypropylene window film. The
 sample cup was placed directly on the XRF detector
 window and the instrument initialized for a 60 second
 lead analysis time.  The result of the analysis was
 reported.

 CHEMICAL ANALYSIS PROCEDURE  FOR
 METALS IN SOIL

 The XRF sample cup was submitted to the laboratory
 for chemical atomic  absorption (AA)  analysis. Ap-
 proximately 0.5 g of  sample, weighed  to 0.001 g ac-
 curacy, was thoroughly mixed with 10 ml 1:1  nitric
 acid:water, digested  according to SW-846, Method
 3050 and analyzed according to Method 7000.

 RESULTS OF THE XRF AND AA METHODS
 ANALYSES

 Seven of the 29 samples analyzed had XRF results
 below the XRF detection limit of 123 mg/kg or quan-
 titation limit (QL) of 410 mg/kg. The XRF and AA
 results of these samples are presented in Table 1.
 The AA results show that all of the sample lead
 concentrations fell below the XRF quantitation limit
 of 410 mg/kg. The results of both XRF methods were
 all below the XRF QL except for sample 5469B's
XRF in-situ result that was 543 mg/kg lead. This was
considered acceptable since priority  samples with
XRF results at or near the QL would be sent to the
lab for AA analysis.  A high frequency of false  nega-
tive XRF results  would have caused either  XRF
method to be questioned.
 Three of the 29 samples analyzed had XRF results
 above the XRF linear calibration  range (  5300
 mg/kg). The XRF and AA results of these samples
 are presented in Table 2. The AA results show that
 all of the sample lead concentrations are near the end
 of the XRF linear calibration range.

 Nineteen of the samples had XRF results above the
 XRF quantitation limit and below the end of the XRF
 linear calibration range. The XRF and AA results of
 these samples are presented in Table 3. These 19
 samples along with samples 5464B and 5469B from
 Table 1 (n - 21) were used in a paired-difference
 t-test analysis.

 RESULTS OF THE XRF AND  AA METHODS
 PAIRED-DIFFERENCE T-TEST EVALUA-
 TION

 The goal of the paired difference t-test is to deter-
 mine  if the mean difference of two populations of
 paired results, is different from zero at the 5-percent
 significance level. In other words, the analyst is will-
 ing to accept a 1 in  20 chance of saying that the
 average difference of the two populations is sig-
 nificantly different from zero when in fact it is not.
 Additionally, the test makes two assumptions. First,
 that each pair of measurements is independent of the
 other pairs. Second, that the differences are from a
 normal distribution. The populations used in this test
were normalized with a square root function.

The in-situ and dried & sieved XRF results were
 analyzed by the paired- difference t-test  The prob-
 ability value for this test was 0279 and is greater then
0.05 (which is associated with a 5-percent significance
 level). Therefore, the average difference of the
 paired results of the two XRF analytical methods is
 not significantly different from zero.

The in-situ XRF and AA results were analyzed by the
paired-difference t-test.  The probability value for
this test was 0.671 and is greater then 0.05. Therefore,
the average difference of the paired results of the two
analytical methods is not significantly different from
zero.

The results of this statistical test enabled the project
manager to conclude  that the two  methods of
preparation were not significantly different and that
the in-situ XRF method and the AA lead analytical
 methods were not significantly different.  In-situ
 analysis was then performed on 500 surface and sub-
                                                604

-------
surface samples. Additionally, portable XRF was
used to support selection of soil samples for use in a
treatability study.

RESULTS OF THE SITE IN-SITU XRF AND AA
EXTENT OF CONTAMINATION ANALYSES

Seventy-one of the 500 samples analyzed by in-situ
XRF were collected and submitted to the REAC
laboratory for AA lead analysis. Twenty-six of these
samples had XRF results below the XRF QL of 270
rag/kg (the XRF lead calibration curve was modified
using site-specific standards from the method evalua-
tion work that resulted in lower XRF lead detection
and quantitation limits  and an extended  linear
calibration range). The XRF and AA results of these
samples are presented  in Table 4.  Four of these
samples had AA results above the XRF QL of 270
mg/kg.

Five of the 71 samples  had XRF results above the
XRF linear calibration range of 12,000 mg/kg lead.
The XRF and  AA results of these samples are
presented in Table 5. All of these samples had AA
results above or near the end of the  XRF linear
calibration range.

Forty of the samples had XRF results above the XRF
quantitation limit and below the end of the XRF linear
calibration range. The XRF and AA results of these
samples are presented in Table 6.  These forty
samples and the six samples in Table 4 with "J" XRF
values (n = 46) were used in a paired-difference t-test
analysis. The probability value for this test was 0.872
and is greater then 0.05 (which is associated with a
5-percent significance level). Therefore, the average
difference of the paired results of the two lead analyti-
cal methods is not significantly different from zero.

CONCLUSIONS

The paired-difference t-test can be used as a decision
tool in the evaluation of  XRF soil lead  analytical
methods. It showed that the average difference be-
tween these two methods was not significantly dif-
ferent from zero at the 5-percent  significance level.
At also showed the average difference between the
in-situ XRF and the AA lead analytical methods was
not significantly different from zero at the 5-percent
significance level for both the XRF method develop-
ment data and the site extent of contamination data.
Additionally, the portable OEI X-MET 880 sup-
ported the following:

*Soil lead analysis in a densely wooded area of the site
initially assumed to be uncontaminated . Investiga-
tion of the area located battery casings mixed with soil
under the leaves. XRF lead analysis confirmed that
the area was contaminated with high mg/kg levels of
lead

'Selection of soil samples for a treatability study.

*320 XRF surface lead analysis results were used to
develop a site contour map.

OTHER XRF APPLICATIONS

In June,  1990, the ERT deployed a XRF unit to
perform cadmium and nickel analyses on sediment
samples from a tidal wetlands contaminated by a
battery manufacturing facility.  The biological "plant
stress assessment" work plan called for the investiga-
tion of a minimum of one plot of vegetation for each
anticipated nominal cadmium  concentration range.
The results of XRF analyses of the plot sediments
enabled the project manager to select the appropriate
plots.   Selected samples were submitted for
laboratory AA analysis. The paired-difference t-test
evaluated the results of both XRF and AA analytical
methods  and found the  average  difference  of the
paired results of the two cadmium analytical methods
was not significantly different from zero (at the 5-per-
cent significance level).

That same month, an XRF unit performed lead and
zinc in-situ surface soil analysis in a network of hiking
trails and campsites contaminated by a smelting
operation for a health and safety assessment of the
extent of contamination and  possible human ex-
posure. Samples were submitted for AA analysis and
the results of both analytical methods were evaluated
by the paired-difference t-test.

The average difference of the paired results of the zinc
analytical methods (most lead values were below the
XRF QL) were significantly different from zero  (at
the 5-percent significance level).  Almost all of the
XRF zinc results were higher then the AA analysis. It
is suspected that the contamination from the smelter
is concentrated in the top layer of the soil and was
diluted when the top quarter inch of soil was removed
for laboratory AA analysis.
                                               605

-------
                                       TABLE 1
                               BATTERY BREAKAGE SITE
              RESULTS OF IN-SITU AND DRIED & SIEVED SOIL XRF Pb ANALYSES,
                              AND AA Pb ANALYSIS IN mg/kg
              SAMPLES WITH RESULTS BELOW THE XRF QUANTITATION LEVEL

        SAMPLE NO.   IN-SITU XRF     DRIED & SIEVED XRF
        5461B         ND             ND
        5464B         125-J            148-J
        5469B         543             125-J
        5470B         ND             ND
        4281B         190-J            ND
        4499B         ND             ND
        4495B         138-J            ND

        XRF detection limit = 123 mg/kg. XRF quantitation limit = 410 mg/kg.
        AA detection limit = 5 mg/kg.
        ND denotes not detected.
        J denotes the sample concentration is between the detection limit and the quantitation limit.

                                       TABLE 2
                               BATTERY BREAKAGE SITE
             RESULTS OF IN-SITU AND DRIED & SIEVED SOIL XRF Pb ANAYLSES,
                              AND AA Pb ANALYSIS IN mg/kg
           SAMPLES WITH RESULTS ABOVE THE XRF LINEAR CALIBRATION RANGE

    SAMPLE NO.  IN-SITU  XRF         DRIED & SIEVED XRF            AA LAB
    4332B        13800 *              6040 *                           5000
    4496B        7680 *               6520 *                           5300
    3319B        6590 *               7210 *                           5100

    XRF detection limit= 123 mg/kg. XRF quantitattion limit= 410 mg/kg.
    AA detection limit- 5 mg/kg.
    *- denotes sample cone, is above the XRF linear calibration range (> 5300 mg/kg).


                                      TABLE 3
                               BATTERY BREAKAGE SITE
RESULTS OF IN-SITU AND DRIED & SIEVED SOIL XRF Pb ANALYSES, AND AA Pb ANALYSIS IN mg/kg
SAMPLES WITH RESULTS ABOVE THE XRF QUANTITATION LIMIT & IN XRF CALIBRATION RANGE
SAMPLE NO.
5460B
5462B
5463B
5465B
5466B
5467B
5468B
5471B
4280B
4331B
4498B
5590C
4500B
5489B
5490B
5491B
5493B
5494B
5492B
XRF detection limit=
IN-SITU XRF
1950
1390
3180
838
771
540
416
1000
1170
1030
532
655
1730
2530
1430
1390
1370
910
1420
123 mg/kg. XRF quantitation
DRIED & SIEVED XRF
2320
1500
3080
1080
946
615
1450
2010
3040
1310
644
716
1990
680
2610
1350
1250
883
1270
limit= 410 mg/kg. AA detection limit=
AA LAB
1800
1200
3200
880
820
530
950
1100
2800
1300
470
460
1600
1000
1900
880
1200
700
1100
5 mg/kg.
                                          606

-------
                                         TABLE 4
                                BATTERY BREAKAGE SITE
   RESULTS OF IN-SITU XRF AND AA Pb ANALYSIS IN rag/kg ON SURFACE AND BORING SOILS
              SAMPLES WITH RESULTS BELOW THE XRF QUANTITATION LEVEL
   SAMPLE NO.  IN-SITU XRF  AALAB     SAMPLE NO.  IN-SITU XRF AA LAB
   B2-1          ND          21           B4-3         ND          32
   B6-1          ND          15           B8-2         ND          15
   B19-2         ND          32           B20-2        ND          120
   B20-4         ND          16           B22-2        ND          19
   B24-1         ND          90           B25-1        ND          51
   B26-2         ND          33           B28-2        ND          100
   B29-3         ND          22           B30-4        ND          11
   B32-6         ND          25           B34-6        ND          25
   B39-0         ND          50           B45-0        ND          54
   MW10-2      ND          24           MW11-2      ND          23
   B4-1          81J          98           B42-6        152 J         370
   B13-1         176 J         160          B2-2         190 J         790
   B39-12        249 J         480          MW9-1       261J         1400

   ND-denotes not detected. XRF detection limit=81 mg/kg' AA detection limit-5 mg/kg.
   J-denotes concentration is between the detection and quantification limit. XRF quantitation limit = 270 mg/kg.

                                         TABLES
                                BATTERY BREAKAGE SITE
   RESULTS OF IN-SITU XRF AND AA Pb ANALYSIS IN mg/kg ON SURFACE AND BORING SOILS
        SAMPLES WITH XRF RESULTS ABOVE THE XRF LINEAR CALIBRATION RANGE
   SAMPLE NO.  IN-SITU XRF  AALAB
   B13-2A       22500 *       53000
   B25-2         69500«       120000
   B28-3         73000 *       11000
   B30-2         54900 *       110000
   B30-3         15700 *       170000

   XRF detection limit= 81 mg/kg. AA detection limit= 5 mg/kg.  XRF quantitation limit= 270 mg/kg.
   J-denotes concentration is between the detection and quantification limit
   •-denotes sample cone, is above the XRF linear calibration range (>12000 mg/kg).

                                         TABLE 6
                                BATTERY BREAKAGE SITE
   RESULTS OF IN-SITU XRF AND AA Pb ANALYSIS IN mg/kg ON SURFACE AND BORING SOILS
 SAMPLES RESULTS ABOVE THE XRF QUANTITATION LIMIT AND IN XRF CALIBRATION RANGE
SAMPLE NO.  IN-SITU XRF  AALAB      SAMPLE NO. IN-SITU XRF AALAB
B2-4          555           750          B4-2         1010         780
B5-2          412           1600         B6-0         2990         3100
B10-1         1290          2700         Bll-0        10700         6700
Bll-1         4230          4100         B13-2B       449          290
B14-1         1120          740          B14-2        11700         6100
B15-1         991           570          B17-1        1740         2500
B19-1         862           1400         B20-1        2460         4500
B20-3         648           610          B22-1        744          1300
B26-1         11000        8900         B31-2        4870         3200
B32-0         8390          12000        B34-0        1120         1500
B36^         3710          2700         B39-18       844          1100
B41-0         10500        11000        B41-6        292          360
B44-30        730           800          MW9-2       945          810
MW11-1       1920          2700         SS165        579          380
SS170         1630          4400         SS173        1040         970
SS175         3080          2600         SS180        3010         2200
SS183         6930          5500         SS189        5060         2100
SS195         2220          1700         SS199        5690         8300
B5-1          301           140          B12-3A       657          140
B18-3         1303          270          B12-2        2797         170

XRF detection limit = 81 mg/kg. XRF quantitation limit = 270 mg/kg. AA detection limit » 5 mg/kg.
                                           607

-------
     DEVELOPMENT OF A FIELD SCREENING TECHNIQUE

                    FOR DIMETHYL MERCURY IN AIR
      Brian E. Brass, Environmental Scientist
          Roy F. Weston / REAC Contract
       GSA Raritan Depot, Building 209 Annex
                Edison, NJ. 08837
          Lawrence P. Kaelin, Chemist
         Roy F. Weston / REAC Contract
      GSA Raritan Depot, Building 209 Annex
               Edison, NJ. 08837
                                Thomas H. Pritchett, Chemist
                          USEPA, Environmental Response Team (ERT)
                                GSA Raritan Depot, Building 18
                                     Edison, NJ. 08837
Many forms of organic and inorganic mercury are per-
vasive in the environment; with both natural and in-
dustrial sources contributing to the total environmental
mercury burden. Mercury can be biologically metabo-
lized to form various organomercurials.  One of these,
Dimethyl Mercury (DMM)  represents a potential
health risk via air exposure because of its high volatility
and toxicity.

Mercury contaminated soils and sediments are fre-
quently biologically active and have been demonstrated
to contain  DMM.  If left undisturbed, emissions of
DMM will be related to the biological generation rate
of DMM as well as the emission rate through the soil or
sediments as regulated by porosity, temperature, pres-
sure, and other physical-chemical factors. However,
when the soils are disturbed, the potential for elevated
emissions of DMM increases.

Research conducted by the USEPA's Environmental
Response Team (ERT), with the support of Roy F.
Weston, Inc. through the Response, Engineering, and
Analytical Contract (REAC), has resulted in a potential
real-time monitoring technique.
BACKGROUND INFORMATION PROMPTING
RESEARCH

The Army Corps of Engineers had been conducting a
cleanup of mercury contaminated soils; however, site
remediation  was suspended due to the potential for
DMM emissions and the lack of a real-time air monitor-
ing method.

A preliminary investigation conducted by ERT and
Weston/REAC at the request of the Army Corps  of
Engineering and USEPA  Region I, indicated that
DMM could be present in the soils, especially in areas
where anaerobic activity was prevalent.

REVIEW OF REAL-TIME PORTABLE INSTRU-
MENTS FOR DETECTING DMM

The criteria established for selecting a real-time instru-
ment for monitoring DMM were quite restrictive. First,
the instrument had to be portable and permit operation
by non-technical personnel. Second, it had to have
real-time or semi-real-time monitoring capabilities.
                                           609

-------
 The third requirement was that it be specific to or-
 ganomercurials.
 A  literature review indicated that the primary
 methods for detecting mercury were atomic adsorp-
 tion, gas  chromatograph, infra-red analyzers, and
 gold film technology.  Atomic Adsorption instru-
 ments are generally not portable, require technical
 expertise to operate and provide only semi-real-time
 results. Gas chromatography requires technical ex-
 pertise and provides semi-real-time results.  Infra-
 red analyzers have  too  many interferences from
 organic compounds.  Gold film detectors are cross-
 sensitive to sulfide compounds, however, the use of
 an  internal sulfide trapping pre-filter negates this
 cross-sensitivity. The gold film technology thus ap-
 peared to have the greatest potential for the required
 application, and a gold film mercury vapor analyzer
 was selected for detailed review.

 GOLD FILM TECHNOLOGY

 The Arizona Instruments Model 411, Gold Film Mer-
 cury Vapor Analyzer was selected as the instrument
 of choice. The Model 411 was  originally developed
 for monitoring elemental mercury in air, and operates
 on  the principal that mercury will form an amalgam
 when it contacts a gold film.

 The Model 411 detects the presence of mercury by
 passing a stream of air across a thin gold film. As the
 mercury in the  air contacts the film  an amalgam is
 formed. The amalgamation causes an increase in the
 electrical resistance of the film proportional to the
 mass of mercury in the sample.  The change in resis-
 tance is compared to a sealed reference gold film and
 processed by a  microprocessor to provide a digital
 read-out in milligrams of mercury per cubic meter
 (mg/m3).

 To  eliminate the necessity of thermally desorbing the
 gold film after each use, the Model  411 employs a
 microprocessor which allows the  instrument to
 operate over a wide range of resistances while
 remaining balanced with the reference film. Heating
 the  film to approximately 250°C, and subsequently
passing a stream of mercury free air across the film
desorbs the mercury and restores the film to its
baseline resistance.

EXPERIMENTS WITH SILVER-COATED
CHROMOSORB
 Since DMM may be metabolized from elemental mer-
 cury, the hypothesis was  made  that both DMM and
 elemental mercury might be present during air monitor-
 ing. Elemental mercury would interfere with the detec-
 tion of DMM in monitoring air, therefore a means to
 remove the elemental mercury from the sample without
 affecting the DMM concentration was required.  Silver-
 coated Chromosorb was tested for this purpose.

 The first test involved monitoring for elemental mercury
 with the Model 411 using a silver-coated Chromosorb
 tube pre-scrubber within a test vessel containing elemen-
 tal mercury in  which the vapor pressure had reached
 equilibrium. Thirty samples of the mercury-saturated air
 were collected without breakthrough occurring from the
 silver-coated Chromosorb tube.

 The second test of the silver-coated Chromosorb pre-
 scrubber involved determining  if DMM would pass
 through it. This was accomplished by preparing a DMM
 standard and measuring the concentration  with and
 without the silver-coated pre-scrubber.  As Table 1 indi-
 cates, the test results for the analysis with and without the
 pre-scrubber are essentially the same.

 PREPARATION OF DMM STANDARDS

 DMM standards are not commercially available, there-
 fore, it was necessary to prepare them in-house in Summa
 passivated canisters by injecting a measured volume of
 DMM and methanol solution  into the canister. Due to
 uncertainties in this procedure, DMM standards were
 confirmed by select ion gas chromatography and mass
 spectra analysis.

 USE OF THE  MODEL 411  FOR DETECTION OF
 DMM

 Initial experiments conducted with the Model 411 (con-
 figured as per manufacturer's specifications)  provided
 erratic results and an inadequate detection  limit for
 DMM.

The Arizona Instrument's Model 411 was therefore
 modified as follows:

 l.The detector resistance was  increased from ap-
proximately 60 ohms to approximately 98 ohms.
ZThe instrument's sample flow rate was increased from
720 cubic centimeters per minute (cc/m) to 866 cc/m.
3.The sampling duration was doubled from 10 seconds to
20 seconds.
                                                 610

-------
    TABLE 1. COMPARISON OF ARIZONA INSTRUMENTS MODEL 411 RESPONSE WITH
AND WITHOUT A SILVER-COATED CHROMOSORB PRE-SCRUBBER TO 4.8 ppb-V and 24.6 ppb-V
                            STANDARDS OF DMM
DMM
Standard
Concentration
ppb-V
4.8
4.8
4.8
4.8
4.8
4.8
4.8
4.8
4.8
4.8
4.8
4.8
4.8
4.8
4.8
4.8
4.8
4.8
4.8
4.8
4.8
4.8
4.8
4.8
24.6
24.6
24.6
24.6
24.6
24.6
24.6
24.6
24.6
24.6
24.6
24.6
24.6
24.6
24.6
24.6
24.6
24.6
24.6
24.6
24.6
24.6
24.6
24.6
Date
Run
7/25/89
7/25/89
7/25/89
7/25/89
7/25/89
7/25/89
7/25/89
7/27/89
7/27/89
7/27/89
7/28/89
7/28/89
7/28/89
7/28/89
7/28/89
7/28/89
8/01/89
8/01/89
8/01/89
8/01/89
8/03/89
8/03/89
8/03/89
8/03/89
7/25/89
7/25/89
7/25/89
7/25/89
7/25/89
7/25/89
7/25/89
7/27/89
7/27/89
7/27/89
7/27/89
7/28/89
7/28/89
7/28/89
7/28/89
7/28/89
8/01/89
8/01/89
8/01/89
8/01/89
8/03/89
8/03/89
8/03/89
8/03/89
Model
411
Response
(unitless)
without
Pre-Filter
0.003
0.003
0.003
0.003
0.003
0.003
0.004
0.003
0.004
0.004
0.002
0.001
0.002
0.002
0.002
0.002
0.002
0.002
0.003
0.002
0.002
0.002
0.002
0.002
0.012
0.011
0.010
0.015
0.016
0.014
0.012
0.013
0.013
0.012
0.011
0.011
0.009
0.009
0.010
0.008
0.010
0.011
0.011
0.010
0.009
0.008
0.008
0.009
Model 411
Response
(unitless) with
silver-coated
Chromosorb
Pre-Filter
0.003
0.002
0.002
0.003
0.002
0.003
0.003
0.003
0.003
0.003
0.002
0.002
0.002
0.003
0.002
0.002
0.002
0.002
0.002
0.002
0.002
0.002
0.002
0.003
0.016
0.012
0.009
0.011
0.011
0.011
0.011
0.013
0.011
0.011
0,011
0.009
0.009
0.008
0.008
0.008
0.010
0.010
0.010
0.009
0.008
0.008
0.008
0.009
                                    611

-------
4 j\ silver-coated Chromosorb tube was utilized as
a pre-scrubber to remove elemental mercury.
S.The  calibration switches were adjusted to
calibrate the instrument to a known concentration
ofDMM.

SENSOR  STATUS DRIFT IN THE MODEL
411

It was observed during method development that
the Model 411's sensor status would first increase
after the instrument detected DMM, then
decrease after a period of time. The sensor status
is an indication of the percent gold film saturation.
The  increase  and subsequent downward drift in
sensor status was not encountered with elemental
mercury. Sensor status drift resulted in the instru-
ment indicating readings lower than actual con-
centrations and was corrected for by allowing the
instrument to  balance the Wheatstone bridge be-
tween the sample gold film and the reference film
prior to monitoring another sample.  This was
accomplished by drawing four 20-second samples
into  the instrument through a iodized charcoal
filter. The filter effectively adsorbs organic and
inorganic mercury resulting in mercury free sweep
air.  The number of mercury free air sweeps re-
quired to  permit the instrument to re-establish
baseline resistance was determined empirically to
be four.
LINEAR RANGE OF THE MODEL 411 FOR
THE DETECTION OF DMM

The linear range of the Model 411 for the detec-
tion of DMM was determined by diluting a 13.70
ppb-V DMM standard down to approximately
one half the Threshold Limit Value (TLV) of 0.01
mg(Hg)/M3. Dilutions were made to 0,0.64,6.40,
and 13.70 ppb-V and were validated by GC/MS
analysis.

The DMM dilutions were measured with the
Model 411 and the data utilized to generate a
calibration curve (Figure  1).  The calibration
curve was found to be linear, with a critical cor-
relation coefficient (R ) of 0.9S. This was deemed
an acceptable linear range for this work effort.

DISCUSSION AND CONCLUSIONS

This study indicates that gold film mercury vapor
detectors have definite potential in monitoring for
DMM. The Model 411 appears promising be-
cause of its simplicity, stability and effectiveness
as a screening tool. However, as with any screen-
ing device, it should not be relied upon exclusively,
rather, it should be incorporated into a multi-
tiered sampling and monitoring program.
                                            612

-------
a>
u
                o£  E
                        0. 15
                        °-10
                        0.05-
                        0.00
             FIGURE  1
   MODEL 411 RESPONSE VERSUS
DIMETHYL MERCURY CONCENTRATION
             (ppb-V)
              GRAPH
                               I I I | I I I I | I I I I | I 1 I I | ''' I	I ' ' ' ' 1^" ' ' 7 ' rT 1 I J T T I | I I I I | I I I I | 1 I I I | I I I I |
                           0.00    2.00    4.00    6.00    8.00   10.00   12.00   14.00
                                            DMM CONC.  (ppb-V)
                             Note:  1 ppb—V is  approximately 0.01  mg/m3  for  DMM
                   DMM CONC.
                     (ppb-V)

                      0
                      0.64
                      6.40
                      13.70
   METER RESPONSE
       (mg/m3)
          o
          0.008
          0.043
          0.126
LINEAR REGRESSION
      VALUES:

   R2  = 0.98
   y Intercept = 0.00220
   Standard error of
    y = 0.010048
   Slope = 0.008959
   Standard error of
    x = 0.000911

-------
           APPLICABILITY OF THIN-LAYER CHROMATOGRAPHY
                      TO FIELD SCREENING OF
             NITROGEN-CONTAINING AROMATIC COMPOUNDS
           William C.  Brumley and Cynthia M.  Brownrigg
              U.S. Environmental Protection Agency
      Environmental Monitoring Systems Laboratory-Las Vegas
       Quality Assurance and Methods  Development Division
            P.O.  Box 93478,  Las Vegas, NV  89193-3478
BACKGROUND

Nitrogen-containing aromatic
compounds (NCAC's)  are toxic
and mutagenic environmental
contaminants of widespread
occurrence.   Often, their
presence in soil or sediment
is a result of wood preserving
activities involving creosote.
Previously,  we have used thin
layer chromatography (TLC) to
effect compound class
separations of NCAC's from
polynuclear aromatic
hydrocarbons (PNA's).  Prior
to the TLC separation,
contaminants were extracted
from the soil sample and
divided into an HCL and
neutral fraction.  The neutral
fraction in particular was
subjected to preparatory TLC
to isolate neutral NCAC's
without interfering
polynuclear aromatic
hydrocarbons (PNA's).

It seemed feasible to apply
TLC to field screening of
NCAC's and other compound
classes such as PNA's.
EXPERIMENTAL SECTION

TLC

E. Merck silica gel 60
preparative plates were used
that were 1 mm thick and 20 X
20 cm size with pre-
concentration zone.
Aldrich silica gel 60 plates
were used with 0.25 mm
thickness and 5 X 20 cm in
size.
Primary solvent systems were
30:70 methylene
chloride:hexane for neutral
NCAC's, 30:70:10 methylene
chloride:hexane:propanol for
the basic NCAC's, and 30:70:10
methylene chloride:
hexane:isopropyl ether for
combined fractions.

GC/MS

A Finnigan-MAT 4021 was used
in the electron ionization
mode with source temperature
270°C.  The mass range scanned
was m/z 50-450 in 1 sec.  A 30
m DB-5 column was used and
temperature programmed from
60-300'C
at 20*C/min.
                                615

-------
RESULTS AND DISCUSSION

Fractionation

Samples were available from
Soxhlet, sonication, or
supercritical fluid extraction
of soils and were analyzed in
methylene chloride solution.
A fractionation scheme was
used to separate NCAC's from
PNA's  (Fig. 1).  This scheme
afforded two fractions:  the
basic compounds called the HC1
fraction and the neutral
compounds called the neutral
fraction.  Both fractions were
free of interfering PNA's as
determined by GC/MS.

The neutral fraction had been
subjected to preparative TLC
in order to free the
cyanoarenes and
indole/carbazole derived
molecules from interfering
PNA's.  The Rf  range of 0.05-
0.32 was scraped (Fig. 2).
This region therefore provides
screening capability for
neutral NCAC's in soil.  The
HC1 fraction could be
subjected to TLC determination
(Fig. 3) and was free of
interfering PNA's as
determined by GC/MS.

Validation by GC/MS

Fig. 4 and 5 provide the total
ion current chromatograms for
the HC1 and neutral fractions
of NCAC's from a soil heavily
contaminated by creosote.
Selected compound classes are
labeled in order to facilitate
comparison of retention
behavior and relative amounts
of NCAC's present.

Advantages of TLC

TLC offers a low cost, rugged,
simple, and efficient method
to screen for target compounds
such as NCAC's.  Greater
selectivity can be achieved
than that illustrated by
incorporation of a third
solvent such as isopropyl
ether in place of propanol.

A great advantage of the use
of TLC is the multiple sample
capability.  Up to 40 samples
could be run on a 20 X 20 cm
plate.  This is a clear
advantage over HPLC methods.
By going to greater
complexity, automation can be
built into the methods.
Automatic spotting,
densitometry, and multiple
development are some of the
options available.

CONCLUSION

TLC offers a simple and
economical way to do field
screening of multiple samples.
This technique is by no means
limited to NCAC's.
Generalization to PNA's and
other aromatic compounds is
obvious.  Non-aromatic
compounds can also be
determined through the use of
visualization reagents or in
situ derivatization reagents.

NOTICE

Although the research
described in this report has
been funded by the U.S.
Environmental Protection
Agency, it has not been
subjected to Agency review
and, therefore, does not
necessarily reflect the views
of the Agency and no official
endorsement should be
inferred.  Mention of trade
names or commercial products
does not constitute
endorsement or recommendation
for use.
                                 616

-------
 Figure  l.   Fractionation scheme
 to  separate NCAC's (both basic
 and neutral compounds)  from PNA's.

M *_•<.: 1 v
(neul





nil

McCl dr led on A 1 pi-.i
|



o pll 14
*nd oxir.ict
McCl




T
McCl/HfK (




01 / 70%) Kxlr.u-ti'd w 1 1 li McCl

Scrape band
Analytes (Nl t riles, etc. ) 1
Ext racted with Hc-Cl

                 TLC
      (Neutral  Soil Sample)
      Ml IHYlCMitiA/OI !•-
              OA2OLE-
     CYANOPHENANTHni Nl -
      CYANONAPHIMAI I 'J!
                        I'NA1'.
                        DANCil. Sl.UAI'l..()
                        •CNA':-,
TLC
(HCI Standard Sample)
13ENZOOUINOLINE

METHYLOUINOLINC

V
o
-o o
o

ACRIDINE
OUINOLINF

Figure 2.  Preparatory  TLC isolation
of neutral NCAC's  in  the  presence
of PNA's; solvent  system  70:30
hexane:methylene chloride.
Figure 3.  TLC  of  the HCL fraction
(basic NCAC's)  illustrated with
standards; solvent system 70:30:10
hexane:methylene chloride:propanol,
                                       617

-------
                       ALKYL OUINOLINES
Figure  4.   GC/MS total  ion trace  of the HCL fraction (basic
NCAC's),  m/z 50-450,  electron ionization.
      T
      280
      3:2!
                      CYANONAPHTHALENES
                                  CVANOPHENANTHBENE


                                  eit
16H8
         i2ea sew
         2«:B3 TIME
Figure  5.   GC/MS total  ion trace  of the neutral  fraction
(NCAC's),  m/z 50-450,  electron  ionization.
                                618

-------
                   ASSESSING THE AIR EMISSIONS FROM A CONTAMINATED AQUIFER
                                            AT A SUPERFUND SITE
                                      Sella Burchette and Thomas H. Pritchett
                                    U.S. Environmental Protection Agency - ERT
                                                   Edison, NJ
                                                 Steven Schuetz
                                                 IT Corporation
                                                   Edison, NJ
                                                 Kristen Harvey
                                               Roy F. Weston, Inc.
                                                   Edison, NJ
The Environmental Response Team was asked by Region II
to assess the degree, if any, that vapors were migrating from
a contaminated aquifer through the vadose zone to the air. If
such migration was occurring, the Region also wanted to
know what would be the worst case long term average air
exposure to the surrounding residents. Even though the
emissions may not have been large at any point, the fact that
these emissions could be occurring over a large area raised
the possibility of an overall significant, long-term air emis-
sion problem.
The ERT's sampling approach involved the taking of flux
measurements over several transects. The flux measurements
were taken using a modified  sampling system that would
easily switch from purging the system to filling a Tedlar bag
without any changes in the flowrates. The results from the
portable GC analyses of the Tedlar bag samples were then
used to compute flux rates (ug/sec/m2) at each point.
These flux values were then kriged, and the results were
plotted in order to determine the overall area of concern. The
intermediate kriging output file was then used to calculate an
average flux rate for the area of concern. This average flux
rate was then converted to an overall area source emission
term (total g/sec) that was then plugged into a long-term
exposure air dispersion model in order to estimate the long-
term average exposure of the nearby residences. This final
set of numbers were given to the Region for a subsequent
quantitative risk assessment.
                                                       619

-------
                 CALCULATION AND USE OF RETENTION INDICES FOR
                IDENTIFICATION OF VOLATILE ORGANIC COMPOUNDS
                     WITH A MICROCHIP GAS CHROMATOGRAPH
                            K.R. Carney, E.B. Overtoil, R.L. Wong
                            Institute  for  Environmental  Studies
                                Louisiana State University
                                     Baton Rouge, LA
Introduction

A  major difficulty in using elution time data
for component  identification  is the high
variability of such data with changes in
chromatographic conditions.   The  use of
retention indices improves  the  situation
somewhat but a degree  of  variability
remains.   Reproducing  Kovat's retention
indices  with  different  instruments in a
laboratory setting can  be difficult and  such
difficulty  are increased  substantially during
field  operations.

The  Microsensor Technology Inc.  model
M200 gas chromatograph is  a
microprocessor  controlled  instrument
constructed  from  micromachined  injector
and  detector assemblies  along  with
microbore capillary  columns.   Independent
column  heaters  are capable of controlling
the temperature of each column  to  within
0.1 °C over a  range of 30° to 180°C.  The
resulting instrument  is  not  only  highly
portable  but  is  capable of generating highly
reproducible  retention  data  on both a
"between  days"  and a  "between
instruments"  basis.

The  dual  column  capability  reduces the
likelihood of coincident  elution  times for
different compounds  and  thereby increases
the reliability of component identifications.
A  software package  developed  at LSU uses
a  two  tiered standardization technique  to
provide even  more  consistent  retention
data and then uses that  data to  generate
qualitative  information.
Discussion

A  time  series  of 220 separate  analyses of
pentane,  hexane and  heptane   was
performed over a period of 4  weeks.
Approximately  20 samples  per day were
run for  three one week periods having a  4
day  interval between weeks.  Variations in
retention time  were  below  one percent for
the four  weeks were  on  the order of 0.2
percent in any  one week.   This kind of
stability justifies  the  use  of an external
retention  index standard.

The  corresponding  retention index  data
show extremely consistent values over a
several  week period.   Consequently,
retention indices  may be determined  in the
laboratory and  used  in  the field without
bringing  authentic standards of all  possible
sample components.   Variations in Kovat's
retention indices  over the  four week  period
had  standard deviations of  less than 0.5
units.
                                             621

-------
An  M200 equipped  with an OV-73  column
and  an  OV-1701  column  provides
multidimensional  elution time  data  .   The
rather  narrow  dispersion  of compounds
about a  straight  correlation line indicates
that  the  retention indices  on these two
columns  are moderately correlated; a
considerable  area in the detection  space  is
empty.   These  two  columns do  not  differ
greatly in  polarity,  thus  one might  expect
such a correlation.  As  a  result  the
dimensionality  of the  space is closer to 1.3
than 2.  Note  however, the substantial
increase  in resolving  power versus  that  for
either of the columns singly (i.e., the
projection of the space onto one axis).  Using
a conservative estimate of a  peak  capacity
of 70 for a typical 100 second
chromatogram  and  a  dimensionality of  1.3,
this  system has  a peak capacity  of
approximately  250.   Even with  the
increased resolving  power of  two
columns,however, an  unresolvable  pair of
compounds  is easily found.  Such cases
could  be further  minimized by  optimizing
the  choice of  which two  stationary phases
used to  obtain the  detection space.

Estimation of  the retention index from
retention time  data requires  an  estimate of
the  column dead  time  or  gas  holdup time.
The accuracy of this  estimate can
substantially bias  the resulting  retention
index  values  for  early  eluting compounds.
A variation of  ±2.5%  in the dead time
estimate  for column temperatures of 40°C
results in  greater than 10% error in
retention index  calculations  at  retention
indices  of approximately 200.   The value of
I at which the  departure  from  linearity
becomes  significant increases  with
temperature  (approximately  100  units  per
20°  change).   Typical  variance  we have
observed in dead time estimates.have been
approximately  2.5%.

Significantly,  using  the elution time of the
air   peak consistently  overestimates  the
dead time.   Comparing the elution  times of
air  and  hydrogen suggests that  air  is
actually retained  to  the extent of 0.1 to 0.2
seconds.   This  means that using air  as to
estimate  dead  time  consistently results in a
nonlinear function  that  systematically
overestimates  retention  indices.
Consequently,  we use  an iterative method
(1) which estimates to by linearizing the log
(t-to) vs I function.  The result a linear
function  which in  essence recalibrates the
retention  index library  to current
conditions.

Temperature dependence of  retention
indices  were observed  to range  from near
zero for some  nonpolar  compounds  to
approximately  2  units  per degree for
alcohols.  Oxygen  containing  compounds
showed  negative  correlations  with
temperature,  in contrast  with  non oxygen
containing compounds  regardless of
polarity.   This  leads to  the possibility of
using  temperature  dependence  information
as a tool for identifying  compound classes, i:
not specific  compounds.

Conclusions

The use of  retention index  library concept
with the  M200 should  provide a reasonably
reliable  screening tool  for sample
component identification.   The  high degree
of reproducibility in retention  data  for  one
instrument  should  make libraries  prepared
in the laboratory field  deployable.   Based or
our experiences with the  M200,  there is a
good  possibility  that the 120  component
library created  at LSU  will work  with a
large  proportion  of  the  M200  instruments
built  to  date.

An inherent limitation in  the  technique
presented here is a  lack  of  dimensionality
due  to  the  similarity of the  two  stationary
phases used.   While the resolving power is
quite good  as  is, improvements can  be  mad<
by optimizing  the  choice of stationary  phase
and/or   implementing   temperature
programming.

Implementation  of  temperature
programming  is  problematic  with the
                                               622

-------
micro-thermal conductivity  detector  used in
the  M200.  Baseline drift has been a
significant  problem.   We have recently  been
successful  in obtaining  temperature
programmed  chromatograms  with the
M200  with thermal conductivity  detection.
We expect this to result in a tremendous
increase  in peak  capacity and overall range
of analytes.

               References

      1.  R.E. Kaiser, Chromatographia,2,
         217 (1969).

      2.  G. Guiochon, C.L. Guillemin,
         Quantitative  Gas  Chromatography,
         Elsevier, New York (1988).
                                              623

-------
             Determination of PCB's  bv  Enzyme  Immunoassav
                           MaryAnne Chamerlik-Cooper
                                 Robert E. Carlson
                     ECOCHEM Research,  Inc., Chaska,  MN

                                Robert O. Harrison
                     ImmunoSystems, Inc.,  Scarborough,  ME
 ABSTRACT
       A competitive inhibition Enzyme
 ImmunoAssay (EIA) has been developed for
 the  determination  of  PolyCJilorinated
 Biphenyls (PCB's).  The test is capable of
 analyzing for PCB's in  the field in  15
 minutes (from prepared sample), using no
 specialized equipment. The test specificity is
 restricted to PCB's, with high sensitivity for
 Aroclor's 1016,  1232, 1242, 1248,  1254,
 and  1260, and moderate   sensitivity for
 Aroclor  1221.    Matrix   and  solvent
 interferences are minimal. The test is capable
 of direct analysis of PCB's at low ppb levels
 in water. A rapid extraction technique using
 DMF,  DMSO, or methanol  gave a mean
 recovery of 70% of Aroclpr 1248 spiked into
 non-oily soils.  Soils spiked with Aroclor
 1248  in  transformer oil  (to a  final
 concentration of 50 ppm) and extracted with
 DMSO or 4 other solvents  showed much
 lower recoveries, though  still adequate to
 produce a strong test response. These spiked
 soils were easily distinguished from mock
 spiked soils (transformer oil only).  Semi-
 quantitative estimates of PCB levels were
 made using an approximate correction factor
 based on the oil-DMSO partitioning behavior
 of Aroclor 1248.  Improved soil extraction
 methods are being developed, but the present
 rapid extraction and EIA should be suitable
 for PCB screening of soils in many field and
 laboratory situations.

METHODS

Reagent Development

       The  development  of the EIA  for
PCB's  followed  these steps:   1)  PCB
derivatives were synthesized for conjugation
to proteins;  2) one of these PCB derivatives
 was conjugated to a carrier protein and the
 resulting conjugate was used to immunize
 animals, which then produced antibodies
 recognizing both the PCB derivative and
 PCB's;  3) a PCB derivative was conjugated
 to horseradish peroxidase (HRP) to make a
 conjugate which can be captured by anti-PCB
 antibodies; 4) the PCB-HRP conjugate was
 used to screen and select antibodies;  5) the
 selected system was optimized for sensitivity
 and matrix tolerance and characterized for
 specificity. Also required but at present only
 partially completed are the following steps:
 6) develop sample preparation methods for
 specific sample types; 7) validate using field
 samples.

 PCB EIA Procedure

 The following procedure  was used for  the
 analysis of samples containing PCB's:   1)
 rabbit antibodies which recognize the PCB
 structure are  immobilized on the walls of
 plastic test tubes or microwell strips;   2)
 samples or calibrators are added  to Assay
 Diluent in tubes or wells, allowing PCB's to
 be captured by the immobilized antibodies.
 PCB's are retained on the  solid phase when
 the rest  of the sample is washed away;  3)
 PCB-enzyme conjugate is  added to tubes or
 wells and bound in the same manner as in
 step 2.  The unbound conjugate is washed
away and the  amount retained by the
 immobilized  antibody  is  inversely
 proportional to the amount of PCB bound in
 step 2; 4) enzyme substrate and chromogen
are added to  the tubes or wells for color
 development  by the bound enzyme. The
intensity  of  color is  also  inversely
proportional to the amount of PCB bound in
 step 2.   Therefore, more  color means
less PCB.
                                       625

-------
Field Soil Extraction

Soil samples were extracted for analysis by
the following procedure:  1)  place soil into
syringe fitted with plastic frit prefilter in the
bottom of the barrel and a 0.2 uM filter. Tap
to allow soil to settle, insert plunger and
press lightly to tamp surface. Fill to 1.5 mL
mark for 2 g soil.  Remove plunger and place
Luer cap on filter tip; 2)  Add2mLDMSO
or other solvent, re-insert plunger, and shake
to break up soil plug.  Time one minute from
plug breakup;  3) Remove Luer cap and
express solvent from  syringe. Only a small
volume of filtrate is required since the extract
will be diluted  for EIA.  Capture filtrate in
clean glass tube or drip one drop (30 uL)
directly into Assay Diluent in antibody coated
tube (step 2 of EIA procedure).

RESULTS AND DISCUSSION

Matrix and Solvent Tolerance

       The EIA interference of methanol
extracts of seven PCB-free soils was tested
using two EIA  formats.  The extracts were
diluted into Assay Diluent for EIA.  The
sequential test was performed as described in
the EIA Procedure  section above.   The
simultaneous test combined the  first and
second incubations (sample and PCB-HRP
conjugate on antibody-coated tube at the same
time). At the end of that incubation, the tubes
were washed and the  normal procedure was
resumed at this point.  The sequential test
was unaffected (>80% of control) by 1:4
dilutions of extract, while the simultaneous
test was strongly  affected at 1:10.  Similar
data were obtained for DMSO extracts of the
same  soils.  Additionally, the sequential
assay format tolerated DMSO up to 50% in a
similar experimental design.
       These data show that the sequential
EIA described offers  excellent resistance to
the effects of concentrated sample extracts,
superior to the simultaneous test.  This  in
turn means that  extracts of low PCB samples
can be assayed with minimal  matrix and
solvent effects, by increasing  the amount
added to the test  in the sample incubation
step.
Assay Precision

       Standard solutions of Aroclor 1248 in
DMF were diluted 1:100 into Assay Diluent
for EIA analysis. The test was performed as
described in the EIA Procedure section
above. Data were calculated as a percent of
the control for each calibrator, then means
and standard deviations were calculated for
each calibrator.  Precision estimates were
made based on 14 runs over 11 days.  For
three calibrators of 0, 7, and 50 ppm, diluted
1:100 (final concentrations of 0, 70, and 500
ppb), the means and standard deviations were
respectively 100+/-4, 35+7-2, and 15+/-2; all
data are expressed as percent of the mean of
all negative control absorbance values.  This
result shows that the EIA described offers
excellent reproducibility.

Test Specificity

       The  crossreactivity of the test for
seven  commonly detected Aroclor's was
examined. Standard solutions of 200 ppm in
methanol (Supelco) were used to make serial
dilutions  in  methanol.   These standard
solutions were diluted 1:100 into an aqueous
diluent for EIA analysis. Figure 1 shows that
the test recognizes most  of the Aroclor's
nearly equally. Based on this 1:100 dilution,
the 500 ppb points (final assay concentration)
correspond to an initial extract concentration
of 50 ppm for soils. Thus,  this test will
easily detect all seven of these Aroclor's at 50
ppm based on a 1:100 extract dilution.
       Specificity  was  also  tested  for
selected  specific congeners  in the same
manner as for the Aroclor's of Figure 1. The
congeners most  strongly recognized were
2,2',5,5'  tetrachlorobiphenyl,   2,3',4,4',5
pentachlorobiphenyl,  and  2,2',4,4',5,5',
hexachlorobiphenyl.  These data show that
the Aroclor specificity reflects the congener
specificity. Biphenyl and several chlorinated
single ring compounds were also tested for
crossreactivity in the EIA.  All of these
compounds  demonstrated less than 0.5%
crossreactivity compared to Aroclor 1248:
1,2-dichlorobenzene, 1,3-dichlorobenzene,
1,4-dichlorobenzene, 1,2,4-trichlorobenzene,
biphenyl,  2,4-dichlorophenol,   2,5-
dichlorophenol,  2,4,5-trichlorophenol,
2,4,6-trichlorophenol,      and
                                           626

-------
pentachlorophenol.  This means that more
than 200 ppm of any of these compounds
would be required to give the same test
response as 1 ppm of Aroclor 1248.

Test Sensitivity

       Standard solutions  in  DMF were
diluted 1:100 or 1:250 into reverse osmosis
purified water for EIA analysis.  The test was
performed as described in the EIA Procedure
section  above.   Figure 2 shows a typical
standard curve; data  points represent the
means of two tests run on  the same day.
Similar results  were obtained for repeated
runs of Aroclor 1248 diluted in the standard
Assay Diluent.   These data  show that this
EIA is capable of direct analysis of PCB's at
low ppb levels in water.
Spike Recoveries Using Field Soil Extraction

       Soil samples were spiked, extracted,
and analyzed by EIA to determine the ability
of the test to detect PCB's in soil. Standard
solutions in DMF, hexane, or transformer oil
were used as noted below.
       Spikes of 1  mg/mL Aroclor  1248 in
hexane or DMF were made into 2 g samples
of 4 PCB-free soils, giving  a final PCB
concentration  of  50 ppm.    Soils  were
extracted using 2 mL of DMF, DMSO,  or
methanol and recoveries were determined by
EIA  using   1:100  extract  dilutions.
Recoveries averaged 70+/-20% for a total of
15 samples. Mock spikes were performed in
the same manner, using  the same soils and
solvents.  For sixteen  samples,  the mean
percent of control was 102+/-15% (100%  of
control means no PCB and no interferences).
These data indicate adequate recovery for soil
screening and minimal interferences from the
soils tested.
       Mock spikes were performed in the
same  manner  as above, except  the spike
material was clean transformer oil in a ratio of
100 uL/2 g soil (5%   v/w).  For seven
samples extracted with DMSO or methanol,
the mean percent of control was 77+/-8%
(100% of control  means no PCB  and no
interferences).  This indicates interferences
from the soils tested greater than those non-
oily soils described above, but still giving a
signal approximately equal to 1 ppm of
Aroclor 1248 in the original extract (based on
the EIA Procedure step 2 dilution of 1:100).
Spikes  of  1  mg/mL Aroclor  1248 in
transformer oil were made into 2 g samples
of 4  PCB-free soils, giving a final PCB
concentration of 50 ppm and a 5%  v/w ratio
of oil to soil. Soils were extracted using 2
mL  of DMSO  and  recoveries   were
determined  by EIA using  1:100 extract
dilutions. Recoveries averaged 12+/-3% for
a total of 4 samples. These values are similar
to those obtained for DMSO extractions of
the spiked  transformer  oil with no soil
present. Methanol, DMF,  THF, and N-
methylpyrrolidone gave slightly less effective
extraction  from  oily soils than DMSO.
Significant reductions in recoveries of 50
ppm Aroclor 1248 spikes were observed at a
0.1 % v/w ratio of oil to soil.
      The extracts of oily spiked soil and
oily mock spiked soil described above were
analyzed by EIA using extract dilution factors
ranging from 1:100 to 1:5.  At a dilution
factor of 1:20, or 5% extract in PCB Diluent,
the mock spike response equated to much
less than 7 ppm in the original sample (the
color was much higher than the diluted 7 ppm
calibrator).  At the same dilution, the spiked
soil response equated to nearly 50 ppm in the
original sample (the color was similar to the
diluted 50 ppm calibrator).
      Based on the above results, oily soils
or suspected oily soils can be analyzed using
a direct DMSO or methanol extraction and
EIA of an  increased concentration of the
extract,  such as 5%, to partially correct for
inefficient partitioning from the oil phase.
Using this  technique, samples could be
confidently screened in the field at the 50
ppm level.  Oily soils containing 50 ppm
PCB  would give a strong EIA response,
while oily soils with no PCB's would behave
as described for the mock spiked oily soils
described above.

CONCLUSIONS

1. The test  is capable of analyzing  for
   PCB's in the field in less than 20 minutes
   (15  minutes from  prepared sample),
   using no specialized equipment.
2. Test specificity is restricted to PCB's.
                                        627

-------
Aroclor's  1248,  1254,  and  1260 are
recognized best; 1242, 1232,  and  1016
are recognized  nearly as well; 1221 is
recognized significantly less well, but can
still be detected easily at 50 ppm.
Congener specificity of the  test reflects
the Aroclor specificity.
The test is capable of  direct analysis of
PCB's at low ppb levels in water.
Soils which are not oily can  be analyzed
using a solvent extraction and direct EIA
of the diluted extract.
Oily soils can be analyzed using a solvent
extraction and direct EIA of an increased
volume of the diluted extract to correct for
inefficient partitioning from the  oil phase.
8.  Further work  in this area will include
    improved extraction from oily soils, oil
    analysis, sediment analysis, biological
    sample analysis, field testing for  the
    above, and quantitation of PCB's using
    the strip-well method for lab analysis.
ACKNOWLEDGEMENT

      The initial phase of development of
this  PCB  immunoassay  was  partially
supported by the US EPA through a sub-
contract to ECOCHEM from Mid-Pacific
Environmental Laboratories, Inc.
                   Figure 1.  Crossreactivitv for 7 Aroclors
                   0        10        100       1000    10000
                       AROCLOR  Concentration (ppb)

                    Figure 2.  EIA of Aroclor 1248 In Water
               100
                90

             o'°
             c  70
             8  6o
             •5  so
             C  40
             O  30
             |i  20
                1 0
                 0
                         0.1      1      10    100   1000  10000

                             ppb AROCLOR 1248
                                      628

-------
                           Practical  Limits in Field Determination  of
                             Fluorescence Using Fiber Optic Sensors
                                Wayne Chudyk
                                Carol Botteron
Kenneth Pohlig
Rose Najjar
                                   Civil  Engineering Department
                                          Anderson Hall
                                         Tufts University
                                         Medford,  MA  02155
                                         Phone (617)381-3211
                                          Fax  (617)381-3819
Purpose
The long-term aim of our research program is
to develop an instrument useful  for the field
determination of aromatic organic contami-
nants.  Concentration ranges of  interest are
in or below the ppb range.  This approach may
be applied to ground water analysis, in both
the saturated zone and in the vadose zone,
and for water or wastewater treatment process
monitoring.

Our earlier work reported on the usefulness
of fiber optic sensors in detection of
aromatic organic ground water contaminants
such as the benzene, toluene, ethylbenzene,
and xylenes (BTEX) fraction of petroleum
fuels.  We use a laser fluorimeter with fiber
optic sensors for in-situ measurements.  The
lower limits of detection observed for com-
pounds excited in the ultraviolet (266 nm)
appear to be restricted by the optical
qualities of the field instrument, including
the sensor.  For example, the dynamic
response range of fluorescence signal versus
concentration is narrow when excitation
occurs in the ultraviolet as opposed to
visible (532 nm) with dye tracer studies
(1, 2, 3).

Optical noise limiting the dynamic range of
the instrument could have sources inside or
outside the instrument.  Optical evaluation
of the system has involved fluorescence life-
time analysis, which has produced results in
two areas.  The first area is in determination
of the source of instrumental spectral noise
(1).  The second area is in application of
fluorescence lifetimes as an identification
and measurement tool (4, 5).
 Scope

 This report outlines the suspected sources  of
 spectral  noise that limit the dynamic range
 of the instrument,  and presents  some  applica-
 tion of the use of  fluorescence  lifetimes
 analysis  as a measurement tool.

 Methods

 In-situ measurement of ground-water contami-
 nants using unmodified fiber sensors  has  been
 the focus of our work.  Our research  group
 has built and field tested a second generation
 prototype instrument.   The prototype  instru-
 ment's theory, construction, and testing
 results have been presented elsewhere (1, 2,
 3, 6).  In summary, it uses a Nd:YAG  pulsed
 laser (Laser Photonics MYLA laser operating
 at 0.5 Hz), with an internal frequency
 doubling  crystal, as a 532 nm light source.
 The 532 nm light is again frequency doubled
 to 266 nm with an appropriate crystal (BBO,
 Quantum Technology) and coupled  to an optical
 fiber (Superguide UV 600N, Fiberguide
 Industries) that is placed in a  protective
 tefzel sleeve along with an identical emission
 detection fiber.  Both fibers terminate in  a
 stainless steel sensor, which holds the fibers
 in the water to be  analyzed.

 Fluorescence and scattered excitation light
 collected by the emission detection fiber is
 carried back to the surface for  analysis.
 Either a  monochromator (H-20, Instruments SA,
 Inc.) or a set of glass cut-off  filters can
 be used for the requisite fluorescence light
 isolation and detection.  Light  intensity is
 measured  either using a photomultiplier (PMT)
 (Hamamatsu).  A measurement of scattered
                                                 629

-------
excitation light is also made to use for power
normalization.  When measuring fluorescence
intensity, the PMT current outputs are con-
verted to voltages using an electronic box-
car integrator (EG&G/Princeton Applied
Research), and the voltages are stored in a
portable personal computer (Compaq) used
as a data logger.  For fluorescence lifetime
measurements, signals from the PMT's are fed
to a digitizing oscilloscope {LeCroy Model
9450, dual channel, 350 Mhz, 400 megasamples/
second per channel) with 50 ohm input imped-
ance at a sampling interval of 2.5 nanoseconds.
The oscilloscope is triggered by a photodiode
(Hamamatsu 1722 PIN photodiode) which is
illuminated by a fraction of the 266 nm
excitation laser light just before focusing
into the excitation fiber.

Sensor lengths, which correspond to useful
well depths, of up to thirty meters have
been used in the field, with most useful
data obtained with 10 m sensors.  Laboratory
investigations have typically used shorter
sensors, with vapor analysis experiments
using fiber lengths of 1.85 m, while solution
analysis experiments usually use sensor
lenghts of 10 m.

Vapor phase analyses were performed by sus-
pending a sensor in a sealed glass dessicator
or flask over a solution of known concentra-
tion of the analyte.  Henry's law could then
be used to predict the vapor phase concentra-
tion above the liquid (7).

Results

Fluorescence lifetime measurements on gasoline
samples show that such measurements may be a
useful means of determining solution concen-
tration.  A log-log plot of fluorescence
lifetime values versus concentration for
unleaded gasoline shows a straight line for
data from less than 0.01 to more than 500 ppm.
A best-fit line through the data is described
by the equation log (lifetime) = 1.66 + 0.181
log (ppm).

Vapor phase analysis of phenol shows linear
response of fluorescence for phenol vapor
concentrations of 0.1 up to 1000 micrograms
per 1iter of air.

Interferences from turbidity are indicated by
the decreasing fluorescence seen from a
1  mg/1  phenol  solution in water when increas-
ing amounts of silt were added to the
solution.
Conclusions

1.  Spectral noise in the instrument, origi-
nating in the flashlamp of the laser in
addition to some fiber fluorescence, limits
the dynamic range of the instrument.

2.  Fluorescence lifetime analysis may be an
additional parameter useful  in determining con-
centration.

3.  Vapor analysis suggests  that vadose zone
analysis may be possible by  performing analyses
on air instead of water.

4.  Turbidity interferes with fluorescence of
aqueous solutions, and its scattering effects
may contribute to observed dynamic range
1 imitations.

References

1.  Chudyk, W., and K. Pohlig, "Dynamic Range
Limits in Field Determination of Fluorescence
Using Fiber Optic Sensors,"  SPIE Annual
Conference, San Diego, CA, September 17-20,
1990.

2.  Chudyk, W., K. Pohlig, L. Wolf, and
R. Fordiani, "Field Determinatoin of Ground
Water Contamination Using Laser Fluorescence
and Fiber Optics," Proceedings Volume 1172:
Chemical, Biochemical, and Environmental Fiber
Senors, SPIE Annual Conference, Boston, MA,
September 6-7, 1989.

3.  Chudyk, W., K. Pohlig, N. Rico, and
G. Johnson, "Field Screening for Aromatic
Organics Using Laser-Induced Fluorescence
and Fiber Optics," EPA/EMSL  First Inter-
national Symposium: Field Screening Methods
for Hazardous Waste Site Investigations. Las
Vegas, NV, October 11-13, 1988.

4.  Birks, J.B., and I.H. Munroe, "The
Fluorescence Lifetimes of Aromatic Molecules,"
Progress in Reaction Kinetics, 4, 1967, pp.
239-300.

5.  Pohlig, K., Design, Construction, and
Testing of a Remote Laser-Induced Fluores-
cence Fiberoptic Groundwater Contaminant
Detector, MS Thesis, Tufts University,
Medford, MA, February, 1991.

6.  Nirmalakhandan, N.N., and R.E. Speece,
"QSAR Model for Predicting Henry's Constant,"
Environmental Science & Technology, £2, 11,
November, 1988, pp. 1349-1357.

Acknowledgement

The authors thank the continuing support of
the Alexander Host Foundation.
                                                   630

-------
                                  The Colloidal Borescope.
                           A means of assessing local colloidal flux
                          and groundwater velocity in porous media
               T. A. Cronk
    Health and Safety Research Division
      Oak Ridge National Laboratory
                P. M. Kearl
      Environmental Science Division
      Oak Ridge National Laboratory
The colloidal borescope is a waterproof video
camera capable of viewing indigenous colloids
in  a  monitoring well.    Through  optical
magnification, the movement and density of
these   colloids  is  easily  assessed.    The
instrument shows promise of  providing  an
improved  methodology for determining both
local groundwater flow velocity and colloidal
transport potential.

Because   observations  are  taken  at  the
microscale, data obtained are indicative of the
local transport parameters of the subsurface
flow   system.     By   taking   numerous
measurements  at different spatial locations,
heterogeneities of the flow system in both the
vertical and horizontal sense may be defined.
Preferential flow zones and fractures may thus
be  located.   The ability to determine the
spatial  variability   of   relevant  transport
parameters is  also necessary for the effective
use of stochastic transport models.

Field use of the colloidal borescope indicates
that colloidal parameters are very sensitive to
external   perturbations.     The   slightest
disturbance  induces  pressure   waves  and
turbulent  flow which greatly affect colloidal
density and migration rates.  An interesting
field experiment confirming this sensitivity
involved dropping a  slug into  a well some 5
meters away from a well in which the
instrument   was   recording   steady-state
movement.  The results were dramatic as the
colloidal migration pattern literally exploded
into a  turbulent  flow pattern.    One  can
imagine  the affects  even the  most  gentle
pumping techniques must  have on  turbidity
and colloid  density during sampling.  These
observations indicate the instrument provides
a much  more accurate method of  assessing
natural colloidal densities and migration rates.

Observations to date indicate a steady, laminar
flow field  in the  borehole  which  has  an
excellent directional correlation with specific
discharge   in   the   surrounding   aquifer.
Knowledge  of groundwater  flow  direction
allows   direct  assessment  of  the  affects
pumping and  injection systems  have on the
natural   flow  field.    The  effectiveness  of
groundwater extraction/injection systems may
thus be  investigated  by observing  the flow
direction near the radius of influence of the
system to see if the  designed  flow field  is
being achieved.

Groundwater  flow    velocities    in   the
surrounding   aquifer   are   inferred   from
potential flow theory relationships  between
flow through a wellbore and specific discharge
in the surrounding aquifer.  While both field
and laboratory observations confirm a direct
relationship between  these two parameters,
these same  observations  indicate  wellbore
                                             631

-------
velocities  are  consistently higher than those
predicted   from  potential   theory.     The
observed relationships suggest potential theory
may not be adequate to characterize the  flow
patterns at  the scale  of observation.   The
development  of  a  consistent  theoretical
explanation for these observed relationships is
sure to provide  new  knowledge  about the
underlying physical processes  involved.   The
colloidal borescope thus provides an exciting
new  means  of  investigating the  physical
phenomena affecting flow in porous media at
the pore  scale.   These observations  offer a
means of enhancing conceptual understanding
and of developing improved transport models.
                                              632

-------
                      FIELDABLE ENZYME IMMUNOASSAY KITS FOR DRUGS AND
                                      ENVIRONMENTAL CHEMICALS
                            Peter H. Duquette, Patrick E. Guire, Mclvin J. Swanson,
                          Martha J. Hamilton, Stephen J. Chudzik and Ralph A. Chappa

                                           Bio-Metric Systems, Inc.
                                       9924 West Seventy-Fourth Street
                                        Eden Prairie, Minnesota 55344
Abstract

Immunoassays (e.g., RIA, EIA) have been demonstrated to
be   useful   for  rapid,   convenient  detection   and
semiquantitative   analysis   of   drugs   and   various
environmental pollutants.  Bio-Metric Systems, Inc. (BSI)
has developed a rapid, sensitive, self-contained, disposable,
EIA device  designed  to  allow  untrained  personnel  to
perform in field situations. This format has been developed
for drugs in  urine or on surfaces and for  environmental
contamination in  soil or  water.  The  analyte in  the test
sample competes  with an enzyme-analyte conjugate for a
limited number  of  immobilized  antibody  sites.   This
AccuPress™  Test format  can detect analytes at 10 ppb in
biological fluids, water,  and  soil, with positive results
indicated by  clearly  visible color development  within  10
minutes.    This   format  is  designed to  have  all dry
components and  to have  an ambient shelf  life  of greater
than one year. The format is readily adaptable for use with
numerous low molecular weight analytes.

Introduction

Chemical pollution and the use of illicit drugs in the U.S.
are two of our most important social problems. Monitoring
environmental samples  for various man-made  hazardous
chemicals has become  necessary in order  to protect  the
populace from the carcinogenic and toxic effects  of such
materials. Of particular concern are certain polychlorinated
organics  such as pentachlorophenol (PCP), chlorinated
dibenzo-p-dioxins (CDD's), and chlorinated dibenzofurans
(CDF's) which are known to be extremely carcinogenic.
Because of the known  association of CDD's and CDF's
with PCP and the possibility that food-producing  animals
are exposed to these compounds through treated wood, one
must be aware  of the possibility that food and  water
supplies  can  become  contaminated by the chemicals.
Likewise, despite all the recent publicity regarding drug
abuse, evidence seems to indicate that the importation and
use of illicit drugs in the United States continues unabated
[1-5].  Contributing factors to this problem are  the large
financial rewards generated by the illicit drug trade and the
seemingly insatiable demand by the American drug user.
Because of these factors, law enforcement agencies need to
know  which  drugs  of  abuse are  most  prevalent  in
trafficking; and they must possess the analytical capabilities
for the detection of these illicit materials.

Most  of the  present  day  methods   for  analysis  of
environmental hazards and illicit drug samples have been
based mainly on conventional chromatography techniques
such as gas chromatography (GC) [6,7], high performance
liquid   chromatography   (HPLC)   [8,9],   and  gas
chromatography  -  mass spectrometry (GC-MS) [10,11].
Although these methods give excellent resolution and are
highly sensitive, they exhibit certain disadvantages such as:
1) the need for extensive cleanup treatment before analysis;
2) expense of solvents and instrument maintenance; 3) need
for trained personnel in a laboratory setting; and 4) length
of time for analyses.

The development of immunochemical techniques has  added
a new dimension to the detection and identification of low
concentrations of pollutants and drugs. These methods are
based  upon the displacement of,  or competition  with,
labeled analyte from an antibody-hapten complex by sample
analyte and the subsequent detection of the labeled analyte
by instrumental methods.  These techniques provide high
sensitivity,  good reliability, relatively fast results (usually
within a few hours), and require less expensive equipment.
At present, two different types of immunoassays have been
                                                      633

-------
developed: radioimmunoassay (RIA) [12-15] and enzyme
immunoassay (EIA) [16-19] for the detection of cocaine,
heroin, and environmental toxins.  Both assays are quite
sensitive, require minimal sample pretreatment, and allow
several  assays  to  be run daily.   However,  the  major
disadvantage  of  most  EIA's  is  the  requirement  for
expensive laboratory equipment in a laboratory setting.
Furthermore,  with the radioactive technique there are the
additional problems  of radioactive  waste disposal  and
specialized handling.

Experimental  Approach

Bio-Metric Systems,  Inc. (BSI)  has developed prototype
AccuPress EIA test kits for both the  analysis of PCP in
water and soil, and cocaine and heroin on surfaces. The
tasks  required to produce the  kits were: 1) obtain  and
evaluate antibody specific to PCP, cocaine, and heroin; 2)
prepare enzyme-hapten conjugates; 3) investigate various
extraction and sampling methods; 4) optimize reagents; 5)
perform component and system stability testing; and 6)
conduct simulated field trials using fortified samples.

Antisera Procurement

Antibodies specific for PCP were purified from antiserum
that had been prepared by immunizing rabbits with either
keyhole limpet hemocyanin (KLH)-PCP or bovine  serum
albumin   (BSA)-PCP  immunogen.     The   PCP-KLH
immunogen was produced by direct coupling of 2,3,5,6-
tetrachloro-4-aminophenol to KLH by diazotization.  An
alternate  immunogen was  prepared  by  coupling  PCP
through the phenol portion of the moiety  to the BSA. This
was accomplished by coupling PCP-valeric acid with BSA
either by:  1)  direct attachment of the free carboxylic acid
moiety to the  amines  on the protein by use of l-ethyl-3(3-
dimethylaminopropyl) carbodiimide (EDC) or 2) coupling
the N-oxysuccinimide ester (i.e.,  NOS) of  the acid by
established methods [20]. Polyclonal antibodies which were
specific to cocaine  and heroin were  induced utilizing
immunogens prepared by similar procedures to those used
for PCP.

Enzvme-Hapten Conjugate Preparation

One of the most critical steps in the development of our
assay is the formation of the enzyme-hapten conjugate. The
appropriate enzyme-hapten conjugate must be able to:  1)
be bound by the immobilized specific antibody; 2) compete
with the  analyte for  antibody bonding;  and 3) maintain
sufficient enzyme activity to generate a signal when low
concentrations of analyte are present in test samples.  We
have developed  assays  which  utilize the enzyme pair-
glucose oxidase (GO)/horseradish peroxidase (HRP).  In
previous work we have used GO-hapten conjugates  and
immobilized enzymes; however, recently we have prepared
GO-biopolymer conjugates to which were added various
haptens. We have found that this step allows better binding
of  the  enzyme-hapten  conjugate  to the  immobilized
antibody.  The GO-biopolymer conjugate can be prepared
by  first oxidizing  the  biopolymer with  periodate,  then
adding the GO followed by reduction of the resulting Schiff
bases with  NaBH,  to  yield  a  stable GO-biopolymer
conjugate.  The carboxylic acid derivatives of the hapten
analogs prepared for the production of immunogens were
coupled to the enzyme (GO) or modified  GO (i.e.,  GO-
biopolymer) by one  of  the  following methods:  direct
attachment of the analogs (i.e., free carboxylic acid moiety)
to the amines on the protein by use of EDC, or by coupling
of the N-oxysuccinimide ester (i.e., NOS) of the hapten by
established methods [20].

Sampling and Extraction Procedures

One of the biggest challenges to the development of EIA's
for the analysis of PCP in soil and drug residues on hands
and surfaces is to develop suitable sampling procedures for
extraction of the desired analytes.  Although  hexane or
toluene are used for extraction of PCP residues from soil,
these  solvents are  not compatible with an immune  test
system,  such as our AccuPress test  format  (Figure 1).
However,  since PCP is highly soluble  in MeOH [21], and
MeOH is compatible  with our AccuPress reagents in  less
than a 40% concentration [22], MeOH was chosen as the
extraction solvent for  use in our assay. The basic protocol
as developed for PCP was the following:  Ten gram  soil
samples were spiked with 0,20 ppb, 100 ppb, and 500 ppb
PCP in methanol. The PCP was thoroughly mixed into the
samples which were then dried at  37°C for two hours in
order to mimic naturally occurring contaminated samples.
The soil samples were extracted by vigorous shaking for
one minute  with 40%  MeOH/PBS.  The extracts were
filtered through a 1.2 pn filter and the filtrate (1.0  ml)
applied directly to  the sample well of the  AccuPress test
module. The antibody disk was washed with five drops of
PBS/3% PEG, after which one drop of conjugate was added
and incubated for two minutes.  The  two  segments were
then pinched together  and released.  The color development
was read after five minutes.  Based upon an arbitrary color
scale designation of 0-5, 20 ppb of PCP in soil gave a color
rating of 3. A positive response was easily distinguishable
from the 0 ppb sample which resulted in a colorless readout
(rating of 0).

During the prototype development of the AccuPress test for
drugs on surfaces,  BSI developed  a sampling vial which
served both as a sampling device and also as an application
device.   As depicted  in Figure  2,  the  sampling   vial
consisted of a sample swab attached  to the dropper  top.
Also attached to the  dropper top was a filter that would
                                                       634

-------
remove any extraneous materials from the sample solution
during application of the test (A).  The user would simply
remove the dropper top with the attached swab from the
sample vial and swab the area to be tested (B). Next, the
dropper top was  replaced and sampler vial shaken.  If
testing a solid, a small amount of the solid was transferred
to the sampler vial, followed by replacing the top, and
shaking the sampler vial.   The  dropper cap was then
removed and 5-10 drops of the sample to the AccuPress test
(C). The development  of this sampler made the prototype
AccuPress tests very convenient to use.

AccuPress Test Format

The AccuPress test format (Figure 1) consists of four parts:
1) antibody disks  (A); 2) read-out disks (B);  3) absorbent
blotting reservoir  (C);  and 4)  a  crush  vial containing
lyophilized antibody, while  the read-out disks (B) contain
an  immobilized enzyme (HRP),  the chromogen  (ABTS),
and a substrate (glucose) for the enzyme-hapten conjugate.
Also, a reservoir (C) containing an absorbent pad is located
beneath  the  antibody  disks,  and  the   enzyme-hapten
conjugate   corresponding   to the  desired   analyte  is
lyophilized in a small crush vial (D).   The user simply
reconstitutes the lyophilized conjugate (D) by squeezing the
tube,  crushing the ampule, allowing the conjugate enough
time for complete dissolution (ten seconds). Next, five to
ten drops of sample are applied to the antibody disks and
the disks  are rinsed with wash  solution to  remove any
extraneous material.    One  drop of  the reconstituted
conjugate is added to the antibody disks (A) and incubated
for one to two minutes. The user then  folds  the top plate
containing  the read-out disks (B) over the bottom plate
containing  antibody disks (A), pinches  for approximately
three  seconds, and the results are read in  five to ten minutes
with a positive result being indicated by color formation.

This  assay format exhibits distinct advantages over other
enzyme immunoassay  formats (e.g., ELISA).  First, the
sample size is less limited since by exposing the  antibody
disk to a large volume of sample (i.e.,  up to 800 |il), the
analyte can be concentrated on the disk,  thus increasing the
sensitivity of the  assay (Figure  1).  Second, a wash step
with PBS allows  any possible interfering  substances still
present in the environmental sample to  be washed off the
antibody disk. Third, the enzyme-hapten conjugate can be
added after  the  sample,  which  should also  increase
sensitivity. Finally, this format allows one to use controls
more  easily and could  also be easily adapted as  a  multi-
analyte assay.

Reagent Optimization

Since the reagents for both the AccuPress test for PCP and
the test for drugs  on surfaces were prepared by similar
methods, only  generalized procedures  will be  described
using the PCP test as an example.

In order to carefully control the amount of immunoglobulin
(IgG) coupled  to the antibody disks, the  antiserum was
purified to >95% IgG by standard methods. The antiserum
was fractionated with saturated ammonium sulfate (SAS),
pH 7.8, by addition of an equivalent volume of SAS to neat
antiserum (50% saturation). After stirring for two hours at
room temperature,  the antiserum was  centrifuged.  The
pelleted material was redissolved to one-half the original
volume with 20 mM phosphate buffer, pH 7.2, and dialyzed
exhaustively  against 20  mM phosphate, pH 7.2.  The
dialysate was then  purified over Whatman DE-52 anion
exchanger.  The IgG peak was pooled and characterized  for
total protein  (Pierce BCA Protein Reagents) [23] and  for
total IgG (ICN Rabbit IgG radial immuno diffusion kits)
[24]. The  IgG was then  prepared for coupling  to paper
disks or stored in  PBS,  pH 7.2, at -70°C, in  100  mg
aliqdots.

Antibody disks were prepared and evaluated to determine
the amount of specific antibody needed to be coupled to  the
disk in order to obtain the desired sensitivity.   Antibody
disks were prepared using ratios of DE-52 purified antibody
per 50 disks.  The levels investigated were 20 mg/50 disks,
10 mg/50 disks, 5 mg/50  disks, and 2.5 mg/50  disks.  A
study was done to determine which load of antibody could
be used in our test format  to optimally achieve the desired
sensitivity.  The coupling efficiency or the optimal amount
of specific antibody/disk was determined by measuring  the
protein concentration [25] before and after coupling and by
radiolabeled uptake experiments using radiolabeled analyte.
This radiochemical procedure involved the incubation of the
antibody disk  with [14C]-analyte for 45 minutes.  The
amount of  radioactive analyte bound to  the  antibody
allowed one  to calculate the pmoles of analyte bound per
antibody disk.

As previously noted, one  step in the development of our
assay is the  production of the  enzyme-hapten conjugate.
The amount of hapten on  the enzyme-hapten conjugate is
critical  for  achieving the required characteristics.   If
insufficient or excessive hapten groups  are coupled to  the
enzyme, the conjugate either will not bind adequately to  the
antibody, or it will bind so well that native analyte cannot
compete effectively.  Thus, hapten coupling experiments
had to be performed to determine the range of hapten
groups needed on the enzyme to achieve the desired binding
characteristic.   Several  conjugates  were  prepared   by
coupling hapten-NOS to GO using various molar ratios of
hapten to enzyme  (e.g.,  20,  50, 100X)  in  the  reaction
mixture.  Analysis of the number of remaining amines  by
a standard 2,4,6-trinitrobenzene sulfonic acid (TNBS) assay
[26] after coupling of the  hapten-NOS  to the enzyme, as
                                                       635

-------
compared to unmodified enzyme, indicated the approximate
load of  the  hapten on  the  enzyme (i.e.,  50% amine
reduction).  The enzyme-hapten conjugates prepared using
100 mg of hapten derivative to 50 mg GO consistently
produced the best conjugates for our EIA development with
regard to enzymatic and immunological activity.

Read-out  disks  were  prepared using  HRP covalently
coupled to chromatography paper disks through a diamine
spacer.  The modified HRP was prepared by first oxidizing
the carbohydrate portion of the HRP with periodate, then
adding the diamine, followed  by reduction  of the Schiff
base with NaBH^ to yield a stable HRP-diamine derivative.
Enzymatic  activity was measured before and after amine
modification  (Table 1).   After  immobilization of the
enzyme, the  disks were  incubated  with  a  solution  of
glucose, chromogen (2,2'-azinobis[3-ethylbenzothiazoline-6-
sulfonic acid], ABTS) and proprietary stabilizer in PBS/1%
polyethylene glycol (PEG) 4000 for 60 minutes. The liquid
was decanted and  the excess moisture was wicked away.
The disks were placed in trays, frozen at -70°C for  20
minutes, and lyophilized overnight. The disks were stored
at room temperature in a  humidity controlled room (19%
R.H.) until used.   Accelerated stability studies at elevated
temperatures  (i.e.,  37°C, 55°C) indicated that the read-out
disks are quite stable with only a 20 to 30% loss of enzyme
activity  after  34  days at 55°C (Figure 3).   The  disks
performed well   in the  assay,  showing   little  or  no
background in the  absence of free analyte, while giving an
easily observable response when analyte was present.

Component and System Stability

Our AccuPress EIA test format (Figure 1) consists of three
components:  antibody disks, read-out disks, and lyophilized
enzyme-hapten conjugate, all containing biological reagents
whose activities can be affected by environmental factors
(e.g., temperature  and humidity).  Because  of  this, it is
difficult to prepare a rapid enzyme immunoassay suitable
for field use which would exhibit a minimum shelf life of
one year when  stored at room temperature.  Therefore,
stability  testing   was  carried  out  on the individual
components and  whole test  kits at various time  points
during  storage at  four different temperatures.  Although
both AccuPress test kits and components for both drugs and
PCP were evaluated, only the PCP test kit data is presented,
which is quite representative of both test kits.

Antibody disks were  prepared as  previously reported,
lyophilized, and stored for two weeks before the stability
study was  initiated. The disks used for the radiolabeled
uptake study were  packaged in bilaminar foil pouches with
a desiccant pack and a N2 infusion immediately before heat
sealing.   The packages were stored in the appropriate
temperature controlled environments for the duration of the
study.  The testing protocol was as follows:  A standard
stock solution of 10 dpm/nl of [14C]-pentachlorophenol was
prepared and stored at -20°C. On test days, an aliquot of
stock solution was diluted in 0.01 M phosphate, 150 mM
NaCl,  pH  7.2.   The  antibody  disks stored at various
temperatures were   allowed  to  equilibrate  to  room
temperature.  One ml of diluted  radiolabeled  PCP per
antibody disk was incubated in 1.5 ml microcentrifuge tubes
(five PCP  disks  and  five control-normal  rabbit serum
antibody disks) for two hours with shaking on an orbital
shaker.  After incubation, 500 ^1 of the supernatant was
removed from each tube  for liquid scintillation  counting.
The quantity of bound PCP was expressed in pmoles/disk.

The stability of the enzymatic activity of the GO-hapten
conjugates was also investigated. A stock conjugate reagent
was packaged in crush  vials for use in the stability testing.
The appropriate dilution of the conjugate was determined
using the appropriate antibody disks.  A concentrated form
of the stock conjugate was added to the stabilization media
and aliquoted into polyethylene tubes for lyophilization.
After lyophilization, a reconstitution buffer, encapsulated in
an  onion skin  glass vial, was  added to the  tube.  The
dropper top with filter  was  applied to the top  of the tube.
The vials were packaged in foil pouches with a desiccant
packet and the test  module, and  stored at the selected
temperatures until tested by direct enzymatic analysis. Four
separate crush  vials  per storage  temperature   were
reconstituted and tested by  the  Worthington Kinetic
Glucose-Oxidase  assay.  The same vials were  used  for
performance testing in the  whole kit stability evaluation.
The mean of the determined rates for four vials (per temp-
erature per time point)  were converted to specific activity.
At each time point, an aliquot of native GO was assayed in
quadruplicate as a control. The  stock of native GO was
prepared at 1 mg/ml, aliquoted and stored at -70°C.

The stability of the enzymatic activity on the read-out disks
at various elevated temperatures was also investigated.
Read-out  disks  were  incubated  with  stabilizers  and
lyophilized.  The dry  read-out disks were  stored in foil
pouches with a  desiccant packet and flushed with N2 before
sealing.    The  pouches  were  stored  at the  selected
temperatures until tested  by either direct application of a
standard amount of enzyme or were assembled into kits for
performance testing.  For the actual  evaluation, the disks
were allowed to equilibrate to room temperature and were
stored in a  desiccator until tested. A standard aliquot of GO
(the same as the control  used for the Worthington kinetic
assay) was diluted to 1  |ig/ml. The disks to be tested were
laid out on a white sheet of paper and 15 Hi of 1 Jig/ml GO
was added to each disk. Color development was monitored
and recorded at 1 minute and 5 minutes compared  to an
arbitrary color chart having five spots of progressing color
intensity.
                                                        636

-------
 Assembled AccuPress test kits were prepared and evaluated
 in an accelerated temperature study.  The preparation of the
 kits and testing protocol are presented as follows:
 Preparation of Kits: Antibody disks were prepared using
 ammonium sulfate preparation of antisera at an IgG level
 corresponding to 50  disks per 5 ml of whole antisera with
 four different stabilization formulations.  These disks were
 then evaluated for performance.   Read-out disks used for
 the whole kit testing were prepared as previously described.
 Conjugate was  titered  to match  antibody  disks  and
 lyophilized in polyethylene tubes.  Glass onion skin vials
 with premeasured aliquots of PBS were placed in the tubes
 and they were sealed with a Porex filter and dropper top.
 Kits  were  assembled  in  the dry  room  (14%  relative
 humidity) and sealed in  bilaminar foil  pouches after N2
 flush.

 Testing Protocol:  Stock standard PCP was diluted  into
 40% MeOH/PBS to  a concentration of 100 ppb (= to 100
 ng/ml). Negative control was 40% MeOH/PBS. The testing
 protocol was as follows:
   -   10 drops (0.25 ml) of positive control were added to
       the positive well.
       10 drops  (0.25 ml) of negative control were added
       to the negative well.
       8-10 drops of neutral pH wash solution were added
       to all wells. (Allergan Lens Plus®).
       Conjugate was reconstituted by squeezing crush vial
       and shaking vigorously.
   -   One drop of conjugate was added to each well.
       It was incubated for 2 minutes.
       The module was pinched together for 2-3 seconds.
       Color development was monitored for 10 minutes,
       and recorded  at  5 and 10 minute intervals.

The results of our  component  and whole  test  module
stability  studies  using the  PCP format  as  a  model
demonstrated that we had excellent stability for both the
components and  the whole test module.  For example, we
were able to demonstrate a 95% retention of PCP antibody
activity after storage at 55°C for one month (Figure 4).
Similarly, the enzyme-hapten conjugate and read-out disk
also were  stable when stored  at 55°C for  one month
(Figures 5  & 6).  Examination of the performance of the
whole  test modules (Figure 7) indicated  that the modules
were quite stable at 55°C for 35  days. It is apparent that
the components and the whole test modules are stable  and
exhibit a shelf life of  at least one year when stored at room
temperature.

Reliability Testing

During  the  in-house  testing,   we  investigated  the
reproducibility of the  test device in detecting PCP residues
in soil.  The four types of soil  (sand,  clay, black dirt,
Minnesota river sediment) were collected and dried. Clay
was heat  dried in  a vacuum oven.  Black dirt, sand and.
river sediment were air dried. All soil samples were sieved
through a #14 mesh screen. The soil samples were weighed
out at 10 gms per vial and spiked  with  PCP at varied
concentrations in 100 (0.1 of methanol to equal 0, 10 ppb, 50
ppb, 100 ppb, 1000 ppb.  Ten samples were prepared for
each PCP level per soil type. Each soil extract was tested
in duplicate in a blind study.

Extraction & Assay Procedure;
       To each 10  gram sample, add 10 ml  of  40%
       MeOH/PBS.
       Shake the vials vigorously for 1 minute.
       Allow the sediment to  settle for a minimum of 5_
       minutes.
       Filter the supernatant using a syringe, through an
       ED-141 prefilter placed in the syringe barrel and a
       1.2 \i S&S Uni-Ro filter.
   -   Add 0.5 ml of filtered extract to the module sample
       well.
   -   Wash    with    8-10    drops   of   neutral   pH
       buffer.(Allergan Lens Plus).
       Crush conjugate ampule and shake to reconstitute.
   -   Discard the first  drop of conjugate  in  a waste
       container.
       Add 1 drop of conjugate to each well.
       Incubate the conjugate for 2 minutes.
       Pinch the module together for  2-3 seconds.
       Monitor the color development for 10  minutes,
       recording the 5 and 10 minute color. A chart with
       5 spots of increasing green color intensity is used as
       a reference.

Since the AccuPress test is intended as a qualitative screen,
we have  used the following  definition of positive  and
negative results. If a color less than or equal to 1.0 on our
5 step color chart develops in 5 minutes, then the results are
classified  as negative.  Any test which  develops a color
greater than a  1.0 in 5 minutes  is classified as  positive.

When applying this rule to the above samples, the following
conclusions were drawn:

   1.   For sand: No zero analyte samples generated color
       greater than 0.25; there were no false positives; 90%
       of the  10 ppb samples were positive (by definition);
       and all samples >10 ppb were  positive.

  2.   For Mississippi river sediment:   No zero analyte
       sample generated color  greater than 0.25; 90% of
       the 10  ppb  samples were positive (by definition);
       and all samples >10 ppb are positive.

  3.   For black dirt:  No zero analyte  sample generated
                                                       637

-------
      color greater than 0.25; 100% of the 50 ppb samples
      generated color of 1.0 or greater.

  4.  For clay:  No zero analyte sample generated color
      greater than 0.25; 60% of the  10  ppb samples
      generated color of 1.0 or greater and 100% of the
      50 ppb samples generated a color of 1.0 or greater.

Our results (Figure  8)  suggest  the greater  extraction
efficiency for sand and river sediment compared to black
dirt and clay type soils at low levels of PCP contamination
(between 10 and 500 ppb). When testing soils suspected to
have high levels of contamination (>1 ppm) the results for
all soil types converge and the color generated with this test
is maximized.

Summary

Data presented  demonstrates  that BSI has developed an
easy-to-use enzyme immunoassay that  can be used to
measure PCP in soil at concentrations  of 10 ppb or greater.
The test kit has  many distinct  advantages over other
screening tests which are currently commercially available
for other small molecular weight analytes:  1) the assay has
a positive  read-out system;  2) the  use of wash steps
eliminates   interfering  substances;  3)   no  laboratory
equipment is needed, eliminating the purchase, calibration,
or maintenance of any equipment; 4) the assay is fast (less
than a total of ten  minutes is needed for the results); 5) all
of the necessary reagents for the assay are present in the
assay kit, consequently the assay  is very easy  to use by
unskilled personnel; and 6) the enzyme immunoassay has
been miniaturized  to maximize speed, portability, and ease
of use.  Also, we were able to obtain evidence,  through
accelerated time studies, that the test components and whole
kits  were   stable  for  one year  when  stored  at  room
temperature.

Acknowledgments

The authors wish to acknowledge the following  support:
NIDA SBIR Grant No. 5 R44 DA03553; NIDA SBIR Grant
No. 1  R43 DA04372; NffiHS SBIR Grant No. 2 R44
ES04148;  FBI  Contract No. J-FBI-89-154; and  USEPA
SBIR Contract No. 68D80035.

                     References

 1.     Morganthau, T., Miller, M. and Contreras, J., "Now
       It's Bush's War," Newsweek.  September 18, 1989,
       22.

2.     Morganthau, T., Miller, M., Sandza, R., Contreras,
       J., Lane, C. and DeFrank, T.M.,  "Hitting the Drug
       Lords," Newsweek. September 4, 1989,  18.
3.     Baker, J.N., "The Newest Drug War," Newsweek.
      April 3, 1989, 20.

4.     Morganthau, T. and Miller. M., "The Drug Warrior,"
      Newsweek. April 10, 1989, 20.

5.     Hackett, G., "On the Firing Line," Newsweek. May
      29, 1989, 32.

6.     Heikes,  D.L.  and  Griffitt,  K.R.,  "Gas-Liquid
      Chromatographic  Determination   of
      Pentachlorophenol in  Mason  Jar Lids and Home
      Canned Foods," J.  Assoc. Off. Anal. Chem.  63,
       1980, 1125.

7.     Edgerton, T.R.,  Moseman, R.F., Lores, E.M.,  and
      Wright, L.H., "Determination of Trace Amounts of
       Chlorinated  Phenols  in  Human  Urine  by Gas
       Chromatography," Anal. Chem. 52, 1980, 1774.

8.     Daniels, C.R. and Swan, E.P., "Determination of
       Chlorinated Phenols in Surface-Treated Lumber by
       HPLC." J. Chromatogr. Sci. 17. 1979, 628.

9.     Ugland, K.,  Lundanes, E.  and  Greibrokl,   T.,
       "Determination  of Chlorinated Phenols by High-
       Performance   Liquid  Chromatography,"   J^
       Chromatogr. 213. 1981, 83.

10.    Paul, B.D., Mitchell, J.M., Mell, Jr., L.D., Irving, J.,
       "Gas   Chromatography/Electron   Impact   Mass
       Fragmentometric Determination  of  Urinary  6-
       Acetylmorphine, A  Metabolite of Heroin," J. AnalL
       Toxicol. 13.  1989,2.

11.    Chen,  B.H., Taylor, E.H.  and  Pappas,  A.A.,
       "Comparison of Derivatives for  Determination of
       Codeine   and   Morphine   by   Gas
       Chromatography/Mass  Spectrometry," J.  Anal.
       Toxicol. 14.  1990, 12.

12.    Albro,  P.W., Luster, M.I., Chae, K., Chaudhary,
       S.K., Clark, G., Lawson,  L.D., Corbett, J.T.  and
       McKinney,  J.D.,   "A  Radioimmunoassay   for
       Chlorinated  Dibenzo-p-Dioxins,"  Toxicol.  Appl,
       Pharmacol. 50. 1979, 137.

13.    Newsome,   W.H.   and   Shields,   J.B.,
       "Radioimmunoassay of PCB's in Milk and Blood,"
       Intern.  J. Environ. Anal. Chem. U). 1981, 295.

14.    Mule, S.J., Jukofsky, D., Hogan, M., DePace, A.,
       and   Verebey,   K.,  "Evaluation  of   the
       Radioimmunoassay for Benzoylecgonine (A Cocaine
                                                       638

-------
15.
16.
17.
18.
19.
20.
Metabolite) in Human Urine," Clin. Chem. 23.1977
796.

Spector,  S.   and  Parker,  C.W.,   "Morphine:
Radioimmunoassay, Science 168. 1970, 1347.

Newsome,  W.H.,   "An   Enzyme-Linked
Immunosorbent  Assay for Metalaxyl in Foods," L
Aerie. Food. Chem 33, 1985, 528.

Kelley, M.M., Zahnow, E.W., Peterson, W.C., and
Toy, Stephen T., "Chlorsulfuron Determination in
Soil  Extracts by Immunoassay," J. Agric.  Food
Chem. 33. 1985, 962.
Rubenstein, K.D., Schneider, R.S. and Ullman, E.F.,
"Homogeneous  Enzyme  Immunoassay:  A  New
Immunochemical Technique," Biochem.  Biophvs.
Res. Commun. 47(4). 1972,  846.

VanDyke, C., Byck, R., Barash, P.O., and Jatlow,
P., "Urinary Excretion of Immunologically Reactive
Metabolite(s) After Intranasal  Administration of
Cocaine, as Followed by Enzyme Immunoassay,"
Clin. Chem. 23(2). 1977, 241.

Anderson, G.W., Zimmerman, J.E., and Callahan,
P.M., "The Use of Esters of N-Hydroxysuccinimide
in Peptide Synthesis", J.A.C.S. 86, 1964, 1839.
 21.    Benvenue,  A.,   and   Beckman,   H.,
       "Pentachlorophenol: A Discussion of its Properties
       and  its Occurrence as a Residue in Human and
       Animal Tissues," Residue Reviews .19, 1967, 83.

 22.    Swanson, M.J., "Field Test Kits for Chemical and
       Biological Warfare Agents," U.S. Army AMCCOM,
       Contract No. DAAK11-83-C-0090, SBIR-Phase II,
       Final Report, March, 1985.

 23.    Smigh, P.K., Krohn, R.I., Hermanson, G.I., Mallia,
       A.K., Gortner, F.H., Provenzano, M.D., Fugimoto,
       E.K., Goeke, N.M., Olson, B.J., and Klenk, D.C.,
       "Measurement  of  Protein  Using  Bicinchoninic
       Acid." Anal. Biochem. 150. 1985, 76.

 24.    Mancini,  G., Carbonara, A., and Heremans,  J.,
       "Immunochemical  Quantitation  of Antigens  by
       Single Radial Immunodiffusion," Immunochemistry
       2, 1965, 235.

 25.    Itzhaki,  R.F.  and Gill, D.M.,   "Micro-Biuret
       Reaction," Anal. Biochem. 9.  1964, 401.

26.   Habeeb, A.F.S.A., "Determination of Amino Groups
      in Proteins by Trinitrobenzene-sulfonic Acid," Anal.
      Biochem. 4. 1966, 326.

27.   Worthington  Enzyme  Manual, L.A. Duker, ed.,
      Worthington Biochemical Corp., p. 37, 1977.
                Table  1.   Modified HRP Activity  after  Purification
               HRP-DADPA (Pre-purification)

               HRP-DADPA (Post-purification)
                                                               Activity
                                                          Units/nq protein

                                                               700 ± 63

                                                               365 ± 49
                                                                          n =  6
                                                 639

-------
NOTE: All components should be at room temperature.
                                1.  Open foil package and remove test
                                   module, color development tube, and
                                   wash tube. (Just before use.)
                                2.  Sample application: Remove red cap
                                   from sample  bottle and  apply  10
                                   drops (±5 drops) to the sample well of
                                   the module.
       A
D-
                  3.  Wash application: Twist tab off wash
                     tube and squeeze entire contents into
                     sample well.
                                4.   Color development tube application:
                                    Hold tube upright and squeeze tube
                                    where  indicated  to  crush  ampule
                                    inside.   Shake  vigorously for  10
                                    seconds.

                                    Carefully apply ONE  DROP of color
                                    development solution to sample well.

                                    Incubate for 1-2 minutes.
                                5.   After incubation, press module closed
                                    for 2-3 seconds. Release and open.
                                    (Press only once.)
                                6.   Open the module and monitor color
                                    development. Record the result at 5
                                    minutes.
 A POSITIVE RESULT WILL SHOW A GREEN COLOR AS DARK OR DARKER
      THAN THE REFERENCE COLOR.
 A NEGATIVE RESULT WILL REMAIN WHITE OR BE LIGHTER THAN THE
      REFERENCE COLOR.
               FIGURE  1.  AccuPress™ Test
Neg.
Pos.
                                 640

-------
     DROPPER CAP
     DROPPER TOP
     FILTER


     SAMPLE SWAB


     SAMPLE BUFFER
                     B
FIGURE 2.  SAMPLER/EXTRACTION DEVICE
                   DAY
    FIGURE 3. READ-OUT DISK STABILITY
                    641

-------
 to
       40
       30
       20
       10
b
        4°C.
        R.T.
        37°C.
        45°C.
        55°C.
             5    10   15   20   25   30   35   40
                  STORAGE TIME IN DAYS
          FIGURE 4. POP ANTIBODY DISK STABILITY
                       RADIOLABEL UPTAKE
                                                       4°C.
                                                       R.T.
                                                       37°C.
                                                       45°C.
                                                       55°C.
                          20         30
                    STORAGE TIME IN DAYS
             FIGURE 5.  POP CONJUGATE STABILITY
 40
                                                       R.T.
                                                       37°C.
                                                       45°C.
                                                       55°C.
                 10        20        30
                 STORAGE TIME IN DAYS
           FIGURE 6. READ-OUT DISK STABILITY
40
                           642

-------
                                                            4°C.

                                                            R.T.

                                                            37°C.

                                                            45°C.
                                                             o
                                                            55 C.
             0
10
20
30
                        STORAGE TIME IN DAYS

                 FIGURE 7. STABILITY TESTING
                          POP ACCUPRESS TEST KIT
tr
3
8
UJ
3
i
                                 RIVER SOIL

                                 SAND

                                 CLAY
                                 BLACK DIRT
                                                600
                                   800
                                   1000
                  FIGURE 8. POP ACCUPRESS TEST
                           VARIOUS SOIL TYPES
                           (N-10 EACH DATA POINT)
                                  643

-------
               XUMA EXPERT SYSTEM FOR SUPPORT OF INVESTIGATION AND EVALUATION OF

                                      CONTAMINATED SITES
V. Eitel, R.  Hahn

Landesanstalt fiir Umveltschutz
Baden-WUr 11 emberg
Abteilung Boden,  Abfall,  Altlasten
Griesbachstr. 3
D-7500 Karlsruhe 21,  Germany
                                                     V.  Geiger,  R.  Veidemann

                                                     Kernforschungszentrum Karlsruhe
                                                     Institut filr Datenverarbeitung
                                                     in  der Technik
                                                     Postfach 36 40
                                                     D-7500 Karlsruhe 1,  Germany
 1.   INTRODUCTION
    In Baden-Wiirttemberg,  programmes are
 carried out to investigate and to regi-
 ster the contaminated sites and to eva-
 luate  their environmental hazard.  The
 expert  system will help  in  this vork.
 The XUMA (German acronym  for expert sy-
 stem on environmental hazards of conta-
 minated  sites) expert system  is being
 developed  within  the  framework of  a
 joint  research project of the Institut
 filr Datenverarbeitung in  der Technik of
 the Kernforschungszentrum Karlsruhe and
 the  Abteilung Boden, Abfall, Altlasten
 of  the Landesanstalt fOr  Umveltschutz
 Baden-Viirttemberg  [1]. It is being im-
 plemented on a Texas Instruments Explo-
 rer II  with the Inference ART develop-
 ment  environment and the RTMS database
 system.  The  programs are written  in
 LISP and ART.
                                                   ports  the  user  when a  case-specific
                                                   analysis plan for the contaminates site
                                                   is prepared.  The third function is used
                                                   for the input of the results of the
                                                   chemico-physical   analyses  into   the
                                                   system.  The  fourth  function supports
                                                   the assessment of a case, i.e.a comment
                                                   in form of an expert opinion. The fifth
                                                   function helps the user reconstruct the
                                                   derivation  of the statements  inferred
                                                   and   the  last  function  enables  the
                                                   authorized  experts to modify  and com-
                                                   plete  the  domain  knowledge  acquired
                                                   within the knowledge base.
                                                        In  the following, the central ap-
                                                   plication  function of the  system, the
                                                   assessment function, as well as the ex-
                                                   planation  facility and first experien-
                                                   ces  gained  with  these functions  are
                                                   described in further detail.
                       of  analysis  re-
2.    SURVEY
   The1 following functions are covered
by the system:
     1.  Evaluation
     2.  Preparation   of  an  analysis
         plan
     3.  Acquisition
         suits
     4.  Assessment
     5.  Explanation facility
     6.  Knowledge acquisition
   The  evaluation  function  primarily
deals  with the determination of a num-
erical  value for  a first  comparative
estimation  of  the environmental hazard
of  contaminated sites.  This value  is
then used for setting priorities during
the  investigation  and  sanitation  of
waste  sites. The second  function sup-
3.   ASSESSMENT
   The  basis  of  the assessment  of a
contaminated  site are  the results  of
investigations. The chemical and physi-
cal  investigations are very important.
The expert has the problem to valuate a
lot of analysis data.
   XUMA  will help the expert   in doing
this with the function "Assessment". On
the   basis  of  the  analysis  results
statements  are devired for the assess-
ment  of the hazard  level. Indications
of  further investigations are  given as
well  as  other  assessment  statements
like  indications of inconsistencies in
the analysis data, control of plausibi-
lity. In an other function, that is not
yet  realized, the local  situation in-
cluding  hydogeology will be considered
                                                645

-------
[2].   The function  "preparation of  an
analysis  plan",  helps to find an indi-
vidual  analysis  plan with the investi-
gation  parameters,  that are of concern
for  example  on   a  contaminated  soil
within  an  industrial  plant. Contami-
nated  soil,  wastes,  eluates  of  the
wastes,   leakage  water,  groundwater,
surface  water, air, soil will  be ana-
lysed.  For the assessment the analysis
data,  information taken from the samp-
ling  records and informations from the
record  of the chemical analysis are of
concern.
   XUMA  contains rules for the assess-
ment of
   - parameters
   - one chemical analysis of a sample
   - all  chemical analysis of one sam-
     ple,  for  example  a contaminated
     soil and the eluate
   - all  analysis results of sample of
     discrete areas
   - all  analysis results and informa-
     tions of the case.

   More  than 25  tables of limit values
or threshold values for water, soil and
air  with more than 100 different para-
meters  are  taken  into the  knowledge
base of the system as well as rules ex-
plaining  the special scope of a table.
Values  of this limit values tables and
threshold values  tables were associated
with 6 quality classes. Quality class I
is  corresponding to the background va-
lues,  quality class II is "tolerable",
class  III means  "further investigation
necessary",  IV,  V and VI  medium, high
and very high hazard level. Example: If
the  analysis result of a  parameter of
groundwater  is  smaller  or equal  0.2
times  the  limit  value of  the german
drinking  water quality table,  then it
is  quality class I. The rules of asso-
ciating  the  concentration  values  to
quality  classes  are the result  of ex-
periences with the risk assessment of a
great number of contaminated sites.
  Assessment of the Analysis; 201.83
           25.07.1983 Eluate
           Analysis Results
Colour, qualitative        » yellow
Electric conductivity      - 430   uS/cm
Ammonium                   =» 0.200 mg/1
Chloride                   < 10    mg/1
Cyanide, total             - 0.750 mg/1
Phenol, total              - 0.900 mg/1
Dry matter                 - 1146  mg/1
Residue on ignition (550 C)= 1112  mg/1
Hydrocarbons (IR)            3.400 mg/1
Mineral oil                - 3.400 mg/1
Loss on ignition at 550 C  - 34    mg/1
Assessment Results
Assessments  on the Basis of  Limit Valu
Tables;
'Dry matter'     is  put  into  quality
                 class II - permissible
                 (TVO).
'Ammonium'       is  put  into  quality
                 class II - permissible
                 (TVO).
'Chloride'       is  put  into  quality
                 class I  - within   the
                 range   of  background
                 values (EG-TW).
'Cyanide, total' is  put  into  quality
                 class  interval IV   to
                 VI (TVO).
'Mineral oil'    is  put  into  quality
                 class   IV  -   medium
                 hazard       potential
                 (NDL-GW).
'Phenol, total'  is  put  into  quality
                 class  V - high hazard
                 potential (NDL-GV).
'Electric        is put into quality
conductivity'    class II - permissible
                 (TVO).
Definite Statements;
Theportion oforganic  matter in  the
dry matter is about 2 % (calculated). A
considerable  portion of the substances
contained is not analyzed.
The value of 'dry matter' is normal.
The  value of 'residue on  ignition'  is
normal.
The  parameter  'colour,  extinction  at
436 nm' should be analyzed.
There   are  indirect  indications    of
'crude  tar' to contained.  Reason:  co-
lour.
Potential Statements;
There  is some indication that the orga-
nic  portion in the dry matter might  be
high.  Reason: dry  matter »  electric
conductivity.
Total Result;
The  analysis is put into quality class
interval V  to VI.
                                                   FIGURE 1:  Example of the assessment of
                                                   an analysis
                                               646

-------
   The tables were classified in groups
for  groundwater,   surface  water,  soil
etc.   Within one  group,  .rules are de-
fined,  for which  purpose and with  what
priority  the tables are  to be used. If
there is no possibility to find a value
for a parameter in one group of tables,
e.g.    groundwater  tables,  rules   are
given  to use  other  groups of tables,
e.g.  drinking water tables. If there is
no  value in any of the  tables for one
parameter,  XUMA proposes different pa-
rameters  with similar chemical charac-
ter  (e.g. o-Xylol for p-Xylol). Examp-
les  of rules for  the  summarization of
the  assessment statements are shown in
figure 2.
Rules
les:
with regard to limit  value tab-
If leakage water is to be assessed,
   and  the value of at least one para-
   meter is quality class IV - VI,
then the sewage tables are to be used.

If  a measured value x is compared with
the Dutch soil table
   and  B Value < x <  C value is valid
   for  the B and C values of the para-
   meter,
then the measured value belongs to qua-
lity class III.

Rule  of assessing an individual analy-
sis parameter:

If pH value < 5,
then  the solubility of heavy metals is
increased.

Rule  of summarizing the results on the
sample level:

If  the turbidness of a water sample is
   clear in the sampling record and not
   clear in the laboratory analysis,
then the sample has  changed chemically
   after sampling.

Rule for summarizing the results on the
case level:

If  'cyanide,  total'  or 'hydrocarbons
   (IR)'  is high or very high in leak-
   age water samples,
then  the ground water should be analy-
   zed.
                                               The  assessment of a case gives sta-
                                            tements  on the  hazard level  (quality
                                            class),  the need of further investiga-
                                            tions,  statistic and definite  and po-
                                            tential statements.
                                               XUMA  can give some help for present
                                            technical  investigations  or  remedial
                                            actions. The field screening analytical
                                            data  will be transmitted to the expert
                                            system which gives an assessment for an
                                            actual  case. So the  following actions
                                            e.g. further sampling or remediation of
                                            the waste will have a better basis.
4.   EXPLANATION
   It  is of particular  importance for
the  acceptance of the system  that its
conclusions are clear and may be recon-
structed by the user. For this purpose,
an explanation facility has been imple-
mented   [3].  The  derivation  of  the
statements  is explained to the user by
means  of texts written in  the natural
language.  Each statement displayed  is
mouse-sensitive.  The explanation faci-
lity  is called by clicking on a state-
ment with the mouse.  Now, the user can
choose  between  the  local  or  global
justification  of the statement. In lo-
cal  justification (Fig. 2), the state-
ment itself is listed together with the
last  rule that has led  to this state-
ment  and with the conditions fulfilled
(premises).  In  global  justification,
the  complete  tree  of  derivation  is
shown,  i.e.,  the  derivation  of  the
statement from the analysis results and
the  facts  and  rules included  in the
static  knowledge base is  represented.
The derivation structure is represented
by indentations.
FIGURE 2: Examples of assessment rules
                                               647

-------
          Local Justification

The fact to be explained is:

The  portion of  organic  matter in  the
dry  matter, calculated from residue on
... ignition, is about 38 %.

It  vas deduced by  the rule G2-RESIDUE-
ON-IGNITION-2:

   If  loss on ignition and  dry matter
   are known,
   ... then the portion of organic mat-
       ter  can be  calculated to be ap-
       proximately:
   ... loss on ignition/dry matter.

The following premises are fulfilled:

   The  analysis '207.83 25.07.83  Elu-
   ate'   resulted  in:  dry  matter  =
   210 mg/1.
   The  analysis '207.83 25.07.83  Elu-
   ate' resulted in: loss on ignition =
   80 mg/1.
    FIGURE 3: Example of the local just-
    ification of a statement.
 5. EXPERIENCES
   XUMA  was tested in the  LfU. It was
 surprising how easy the system is to be
 used even for a user without experience
 in computers. The system is a good ass-
 istant   to help the expert in risk ass-
 essment.  Maybe that some of  the rules
 seem  to be simple  or trivial if  they
 are  seen isolated. If rules are combi-
 ned and used without any exception, the
 statements  are very  helpfull for  the
 expert.
   Even  inconsistent  assessments  can
 help to find errors in the analysis da-
 ta  or in the rules. The expert is then
 able to create better rules or modified
 quality  classes. A risk  assessment is
 not only given on the basis of one dis-
 crete  value as a yes-no decision it is
 furthermore the result of comparing va-
 rious  standard value tables with defi-
 ned  scope.
   So  the expert will have the necess-
 ary  tolerance for the special assesse-
 ment  of individual cases.  Risk assess-
 ments  are transparent,  standardized as
 far  as possible and  reproducible. The
 system can be used only by risk assess-
ment experts.
6. REFERENCES

[1] Veidemann, R.     and    Geiger, V.
    (1989).  XUMA -  Bin Assistent   fdr
    die  Beurteilung von Altlasten.  In
    A. Jaeschke,  V. Geiger and B. Page
    (Eds.), Informatik im Umweltschutz,
    Informatik-Fachberichte 228, pp  385
    - 394. Berlin: Springer-Verlag.
[2] Ministerium filr Umwelt Baden-Wurtt-
    emberg   (Ed.)  (1988).  Altlasten-
    Handbuch,   Teil 1, Altlasten-Bewer-
    tung.  Wasserwirtschaftsverwaltung,
    No. 18.
[3] Huber,  K.-P. (1988) Erklarungskom-
    ponente fur das Expertensystem XUMA
    unter  Berlicksichtigung verschiede-
    ner    Benutzerklassen.    Kernfor-
    schungszentrum Karlsruhe, KfK 4478.
                                                648

-------
                 A RAPID RESPONSE SAW-GC CHEMICAL MONITOR FOR
                          LOW-LEVEL VAPOR DETECTION

                       JOHN A.  ELTON and JAMES F. HOULE
                            EASTMAN KODAK COMPANY
                             ROCHESTER, NEW YORK
             INTRODUCTION

Chemical vapor monitors  (CVMs) are
generally not sufficiently sensitive,
selective,  or reliable enough to
detect a multiplicity of vapors in
less than 2 min.  There  is,
therefore,  a need for a  CVM which can
simultaneously detect a  variety of
vapors in the presence of
interferents.  In addition, the
detection of a specific  vapor must be
conclusive so that false alarms are
minimized.   Detection in less than 2
min requires either highly selective
multiple detection methods if several
vapors are present or separation so
that each vapor can be detected and
identified.  The sensitivity  for each
vapor must also be sufficient to
allow detection at desired or
required levels.

SAW sensors have been used to detect
vapors at low concentrations  [1].
However, the SAW detection limits
reported to date for agents such as
GD and HD are much higher than the
limits other devices are capable of
reaching.  Detection limits of
approximately 100 ppb  (0.6 mg/m^) for
GD and 5 ppm  (32 mg/m^)  for HD have
been reported.  It will  be shown in
this paper that much lower levels may
be obtained  for GD and HD when the
system described herein  is used.  In
addition, results on the detection of
methyl benzoate and phenyl acetone
using the same system as used for CW
detection will be provided.

              BACKGROUND

The CVM unit contains major
modifications which allow
significantly improved response
times.  Ambient vapors are collected
on a thermally desorbed  type
concentrator by pumping  air through a
glass tube packed with concentrator
material (Figure 1; concentration).
At the end of a fixed 20 sec
interval, the concentrator is heated
and the collected vapors desorbed
onto the GC column.  Desorption
occurs in about 6 sec and provides
chromatographic peaks that are
compatible with the SAW detector
(Figure 1; injection).  An additional
4 to 8 sec is typically needed,
however, in order to obtain complete
injection of the vapor plug onto the
GC column.
    CARRIER GAS
     GENERATOR
           AMBIENT AIR
            INLET
   VAPOR
CONCENTRATOR
       CONCBJTHATOR
         VENT
   CONCENTRATION

   NJECTON
       ANALYSIS

       VALVES
Fig. 1. Kodak's Chemical  Vapor
        Monitor Showing
        Concentration,  Injection,  and
        Analysis  Stages of Operation.

The GC column greatly enhances  the
selectivity of the  system by
separating the vapors  (Figure  1;
analysis).  Each  vapor  plug which
elutes from the GC  column at a
different time is immediately
injected onto a SAW sensor.  A
second, uncoated  SAW sensor located
nearby is used as a reference.   When
combined with a frequency mixer, this
configuration provides  a  frequency
                                       649

-------
difference (Af)  that  is  easily
measured and relates  to
concentration.

The CVM has several other  subsystems.
The sequencing of valves,
concentrator, pumps,  and the
acquisition of SAW sensor  data is
controlled by a Macintosh™ computer.
A second subsystem provides clean
air/carrier gas to the GC  column with
a small pump that draws  ambient air
through molecular sieve  and charcoal
scrubbers.  A solid-state  mass flow
controller is used to guarantee a
stable carrier gas flow  under varying
conditions of pump and scrubber
aging.  A typical output
of the CVM is given in Fig. 2 and
shows the concentration, injection,
and analysis characteristics of the
device.

             EXPERIMENTAL

Each SAW sensor was first  tested as
an individual sensor  with  each vapor
of interest at  one or more
concentrations.   The  sensor was then
                                 incorporated into  the CVM and system
                                 testing performed.

                                 Vapor Generation and Verification

                                 Vapors were generated using a
                                 Microsensor Systems, Inc. VG-7000
                                 Automatic Vapor Generation System.
                                 All vapors supplied to the CVM were
                                 monitored using a  Hewlett Packard
                                 5890 Gas Chromatograph containing an
                                 FID detector.  For the SAW sensor
                                 testing,  periodic  checks of the vapor
                                 concentration  were made every 11
                                 minutes.   For  the  CVM tests, vapor
                                 concentration  was  determined by
                                 sampling the final portion of vapor
                                 which impinged upon the CVM
                                 concentrator.  This procedure was
                                 performed in order to ensure
                                 verifiable vapor concentrations.

                                 SAW Sensor Preparation

                                 Selective coatings of ethyl cellulose
                                 (ECL)  and fluoropolyol (FPOL) on SAW
                                 sensors were prepared using
                                 proprietary thin film coating
                                 techniques.  All coatings were
                                 observed under a microscope to
                     CHEMICAL VAPOR MONITOR
                     RESPONSE TO CEES VAPOR
   F
   R
   E
   Q

   S
   H
   I
   F
   T
      700
500
300
100
     -100
             CONCENTRATION
                 6.4 mg/m3
                         INJECTION
                            10 sec
               ANALYSIS
                                                                140
                                                                -20
                    10
                        20
    30
TIME (sec)
40
50
Fig.  2.  Typical  CVM Output
                                      650

-------
determine uniformity of the coating.
Only uniform and well adhered
coatings were used in this study.
The thickness of a coating was
determined by recording the frequency
shifts of the device both before and
after coating.

SAW Sensor Testing

Individual SAW sensors were tested
using various concentrations of the
vapors of interest.  The SAW sensors
were exposed to the vapors for a
minimum of 20 minutes.  A signal to
noise ratio of at least 3:1 was
chosen as a criterion for acceptable
data.

CVM Testing

The general operation of the CVM was
described above.  Standard operation
is a 20 second concentration period,
a 12 second injection period, and an
88 second analysis time.  Deviations
from this standard will be indicated
in the results section when
appropriate.

               RESULTS

Results were obtained using two types
of SAW sensors.  The first tests used
a modified version of an established
type of sensor (Type I).   This was
followed by extensive testing of a
new type of SAW sensor (Type II).
Both SAW sensor test results and CVM
test results are reported below.

SAW Sensor Testing

Table 1 provides the frequency shifts
observed from selected experiments
when the ECL-I, FPOL-I, ECL-II, and
FPOL-II sensors were tested with
various concentrations of GEES, DMMP,
methyl benzoate,  and phenyl acetone.

Chemical Vapor Monitor Testing

During CVM testing the vapor flow was
connected to the concentrator input
of the system; flow rates through the
concentrator were monitored.  The
results of selected tests using the
first type of SAW sensor are shown in
Table 2.
               TABLE 1
          SAW Sensor Results

        Vapor Concentration   SAW
                 (mg/n»3)     Response*
                              (Hz)
CEES**
DMMP**
CEES
DMMP
MB**
PA**
CEES
DMMP
MB
PA
2100
91
17.8
20.6
2.4
1.7
516
2.3 ±
19.2
6.48
2400
4,710
58
362
75
88
248
1.1 773
222
265 ± 60
Sensor
ECL-I
FPOL-I
ECL-II
FPOL-II
*80 Hz noise level  (Type I);
  5 Hz (Type II)
**CEES - chloroethyl ethylsulfide;
  DMMP - dimethyl methylphosphonate;
  MB - methyl benzoate; PA - phenyl
  acetone

               TABLE 2
    Chemical Vapor Monitor  Results

Sensor  Vapor Concentration   CVM
                 (mg/m-')    Response*
                              (Hz)
ECL-I**
          CEES
FPOL-I*** DMMP
ECL-II    CEES
FPOL-II
          DMMP
          MB
          PA
 6.1
17.6
 5.27
23.2
27.2
11.0
 570
1600
  82
 495
 213
 150
*Response obtained in less than two
 minutes; 2 Hz noise level unless
 otherwise specified
**5 Hz noise level; 10 second
  injection period
***10 Hz noise level; 14 second
   injection period

              DISCUSSION

These results indicate that detection
limits for GD and HD using the new
type of SAW sensor should be
considerably less than previously
reported [1-2].   Table 3 gives
extrapolated detection limits and
response times for both the
SAW sensor experiments and the CVM
experiments.  The extrapolated
detection limits are determined from
                                      651

-------
the values reported herein using a
3:1 signal to noise ratio.

               TABLE 3
    Extrapolated Detection Limits

Type    Vapor  Extrapolated  Response
                Detection     Time
               Limit  (mg/m3)  (min)
SAW Sensor

Type I  CEES
        DMMP
Type II CEES
        DMMP
        MB
        PA

CVM

Type I  CEES
        DMMP
Type II CEES
        DMMP
        MB
        PA
210
  4.6
  4.6
  0.05
  0.48
  0.29
  0.16
  0.33
  0.38
  0.28
  0.77
  0.44
20
40
20
40
20
20
It should be noted that optimization
of coating thickness was performed
much more extensively with the Type I
sensor.  Significant improvements in
the Type II sensor are expected in
the future as further optimization of
coating thickness and subsystem
parameters are performed.  The much
lower noise level of the Type II
sensor is the principal advantage of
using this technology since both
sensors should provide approximately
the same response when the same
thicknesses and types of coatings are
utilized.

The results also reveal that the
greatly increased sensitivity of the
Type II SAW sensor is not carried
over to the detection of DMMP using
the complete CVM unit.  FPOL coated
sensors do not equilibrate as quickly
with DMMP as with the other vapors.
The peaks observed during all DMMP
testing were much broader than for
all other cases.  Significant tailing
of peaks was observed.  Different GC
columns and higher temperature
operation of the SAW sensors may help
to narrow the peak width and improve
the detection limit for DMMP.
The results also show the response of
the two sensors to methyl benzoate
and phenyl acetone.  It is believed
that this is the first time that
detection of such vapors with SAW
detectors has been reported.  The
extrapolated detection limits
indicate that relatively low levels
of these vapors can be detected with
SAW sensors.  No effort has yet been
made to develop special selective
coatings for these two vapors.

              CONCLUSION

Our studies have shown that SAW
technology can be used to detect DMMP
and CEES at concentrations below 1
mg/m3 in less than two minutes. We
have also reported for the first time
the behavior of methyl benzoate and
phenyl acetone to SAW sensors
typically used for chemical agent
detection.  Concentrations below 1
mg/m3 are also indicated for these
two vapors.  Because of the
preliminary nature of some of the
data presented herein, we anticipate
even lower detection levels in the
future as operational parameters and
selective coatings are optimized.

           ACKNOWLEDGEMENTS

The authors would like to thank
Arthur Snow of the Naval Research
Laboratory for providing us the
fluoropolyol used in these
experiments.
              REFERENCES

1.  Dennis M. Davis, Leon J. Schiff,
and John A. Parsons, "Detection Of
Chemical Warfare Agents Using A 158
MHz SAW Microsensor,"Proceedings of
the 1987 U.S. Army Chemical Research,
Development and Engineering Center
Scientific Conference on Chemical
Defense Research,  2., 939-945, 1988.
2.  Dennis M. Davis, Raymond E.
Miller, and John A. Parsons, "Surface
Acoustic Wave Detection of
Organophosphorus Compounds,"
Proceedings of the 1987 U.S. Army
Chemical Research, Development and
Engineering Center Scientific
Conference on Chemical Defense
Research,  2, 917-923, 1988.
                                        652

-------
                            PASSIVE CRYOGENIC WHOLE AIR  FIELD  SAMPLER
S teven J. Fernandez
Bill G. Motes
Joseph P. Ougan, Jr.
Susan K. Bird
Gary J.  McManus
Westinghouse Idaho Nuclear Company
P.O. Box 4000,  Mail Stop 2202
Idaho Falls, Idaho 83403
The portable, passive cryogenic sampler has
been   designed  by   the   Idaho   National
Engineering   Laboratory   (INEL)   for  the
collection of whole air samples without the
loss  or  concentration  of  any atmospheric
constituents.  The principle of operation is
the   collection  by  bulk   gas   flow  and
condensation  of a whole air sample  into a
previously evacuated cylinder held at liquid
nitrogen temperature using a reservoir.  The
ability of the  sampler  to collect a highly
compressed gas  sample without concentration
of condensable  gases permits a large number
of gas  constituents  to be  analyzed  from a
single  sample,  even  when  analytes  vary
widely in their boiling points.

Design  criteria for the  portable,  passive
whole  air sampler are  listed in  Table 1.
The sampler,  constructed in-house of readily
commercially available components, is shown
in Figure 1.

The  sampler   evaluation was performed  in
three phases.   The first  phase  determined
sample  flowrate,  sample  size,   resultant
sample  pressure,  and   sample  collection
lifetime  as  a  function  of  the  liquid
nitrogen  additions.     The  second  phase
analyzed  simulated whole  air samples  for
bulk  composition,  noble  gases,   selected
chlorofluorocarbons,  and tritium before and
after  collection  in   the  cryogenic  air
sampler.   In the third phase actual field
samples were  collected and analyzed for bulk
composition and chlorofluorocarbon content.
These   samples   were   then  concentrated,
separated, and  analyzed  for noble  gases.
Also  included in the  third phase was  the
analysis  of  altered  whole  air  samples,
blindly  and  randomly  introduced into  the
sample  analysis  scheme  as   a  means   of
detecting sample tampering.

Results  of the  evaluation  of the  design
criteria for  the sampler are listed in Table
2.  A partial listing of past customers  and
their application of the sampler is found in
Table  3.   The need  of a low  cost  passive
cryogenic  sampler  that  can  collect many
whole air  samples at remote  locations with
minimal  logistical   support   will   become
widespread in the future.

We have found the sampler to meet or exceed
all of the characteristics intended for  it.
The sampler is capable of the collection of
samples without concentration or loss of any
sample  constituents  regardless of  boiling
point.

The required sample  volume of  100 L at  STP
has been successfully achieved, and samples
as  large  as  131.2  L have been collected.
Most samples are between 70  and 90  L.   The
volume of sample collected is dependent upon
the  sample  duration  and  flowrate.    By
selection  of  the  proper  combination   of
sample  duration and flowrate, samples  of
accurately known size  from a  few  to  100
liters may be collected unattended within 30
minutes or over a period of time of 2 hours.

Laboratory   tests   on   known   standards
demonstrate that no  concentration  or loss of
atmospheric constituents occurs.

No   electrical  power   is   required   for
operation of the sampler, which would enable
it to operate in hazardous environments such
as where potentially explosive mixtures  of
hydrogen and oxygen are found.

The sampling lifetime can extend to 4 hours
with  refilling   of   the   liquid  nitrogen
reservoir.
                                                 653

-------
                    TABLE 1
     Desien Criteria for Whole Air Sampler
                     TABLE 2
  Development of Cryogenic Whole Air Sampler
1) Sample Volume of 100 Liters
2) No Concentration or Loss of Constituents
3) Sampling Lifetime Greater Than 2 Hours
4) Small (50cm x 15cm) & Lightweight (20kg)
5) No Electrical Power
6) Operator Safety
7) Ease of Operation
   Parameter

Sample Flowrate
Sample Volume

Sample Pressure

Sampling Lifetime
Sample
Concentration
Sample Loss Tests
       Results

Controllable, 1 cc to 3 L/Min
Nominally 100 Liters, 130 L
Maximum
Nominal 2000 psi, 3650 psi
Maximum
30 Min to 10 Hours
Noble Gas Ratios, Unaltered

Chlorofluorocarbons, 100Z
Recovery
Tritium, 100Z Recovery
                                              TABLE 3
Customer
DOE- Office
of Materials
US Air Force
DOE- Office
Waste Mgt
US Air Force

DOE- Defense Programs
DOE-Office
Arms Control

US Air Force
Facility
ICPP
Proposed for
White Sands,
KA-III Series
ICPP
TREAT Pulse
Reactor INEL

Advanced Test
Reactor INEL
INEL Research

ICPP
e vji.yugen.n; WIIUJ.B ftir aampiei
Application
Hydrogen-Rich
Off-Gas Study
Fuel/Air
Explosives
Environmental
Sampling
Environmental
Sampling

Off-Gas Studies
Evaluation of
Arms Control
Verification

Fission Products
in Ar Carrier Gas
Analvtes
Permanent
Gases
Combustion
Products,
Oxygen
85Kr,N2,02
Ar , C02
Kr, Xe,
Freon-11
Me thy 1-
chloroform
41Ar,Kr,Xe
CFC13,CH3CC13
Freon-113
XKr.Xe.He,
N2,02,Ar,H2,
Freon-12,
C02, Freon-113
Fission
Product Gases
                                                  654

-------
 Variable  Set-Point
Pressure Releif Valve
   Fixed Set-Point
 Pressure Relief Valve
      Plate  L1d
   Sample Cylinder
     Base Plate
V
                                                                      Rotameter
                              Sample Inlet/Isolation
                                      Valve
                                                                   Hexagonal Nipple
                                                                      SS Tubing
                                                                    Liquid Nitrogen
                                                                       Reservoir
                                                                      ICPP-A-16925
                                                                         (1-91)
                                    FIGURE 1

                    PASSIVE  CRYOGENIC  WHOLE AIR FIELD SAMPLER
                                           655

-------
                            Effectiveness of Porous Glass Elements for Suction Lysimeters
                                   to Monitor Soil Water for Organic Contaminants

                                                      by
                                        Stanley M. Finger, Hamid Hojaji,
                                    Morad Boroomand, and Pedro B. Macedo
                                           Vitreous State Laboratory
                                         Catholic University of America
                                             Washington, DC 20064
ABSTRACT

       The objective of this effort is the development of
a porous glass suction  lysimeter which can be used to
sample organic contaminants associated with unsaturated
soil  matrices.   Current ceramic  suction lysimeters  are
ineffective in sampling hydrophobic compounds since their
surface  chemistry is hydrophilic,  effectively  repelling
organic species.

       Methods for preparing porous glass elements with
controlled porosity have been developed.  Elements with
air entry values (as  measured by the bubbling pressure
method) corresponding to effective pore sizes as small as
2 microns  with high saturated hydraulic conductivities
have been achieved.

      The  performance of porous glass elements  in
sampling organic contaminants in aqueous media is being
evaluated.      Aliphatic   (1-octanol)  and    aromatic
(ethylbenzene) compounds dissolved in water were used
as the tests solutions. Tests are also being performed with
inorganic constituents in the test water to determine  the
ability of the test elements to sample inorganics.  Initial
results indicate that the porous glass elements are able to
effectively sample organic and  inorganic constituents in
the test solutions.  These data indicate that analyte concen-
trations  in the water sampled through the porous glass
elements are within about 10% of the test solution concen-
trations.
 BACKGROUND

       The U. S. Environmental Protection Agency
 (EPA) requires vadose zone monitoring at active
 land treatment and disposal facilities for hazardous
 wastes. The state of California has extended this
 requirement to practically all active and closed
 storage, treatment, and disposal facilities for haz-
 ardous waste, solid waste, and underground storage
 tanks. Routine analysis of samples collected with
 suction lysimeters is considered an important ele-
 ment in the vadose zone monitoring requirement.
 Most of the suction lysimeters in use now were
 developed for the agricultural industry to monitor
 leachate from crops. These data are used to pro-
 gram the application of fertilizers and soil amend-
 ments. Another device, the tensiometer, is used in
 conjunction with the suction lysimeter to monitor
 soil moisture; this information is used to program
 irrigation. This same equipment is now being used
 to monitor the land treatment of certain hazardous
 wastes, e.g. refinery separator sludge and wood
 preservative waste.  Many of the components of
 interest in these wastes are organics and heavy
 metals.

       The suction  lysimeter's porous element,
 through which soil water is drawn under vacuum,
 has been purposely designed to be hydrophilic to
 facilitate the transport of  the aqueous phase.
Porous elements currently in use are most frequent-
ly ceramic. However, TFE-fluorocarbon, nylon
mesh and alundum  have also been  used.  The
porous element is typically treated with acid and
water to remove contamination and enhance hydro-
                                                      657

-------
philicity.  Unfortunately, the resulting hydrophilic nature
of the porous element presents  an effective barrier to
sampling of non-polar components.  Organics, whether
dissolved in the aqueous phase or existing as a separate
phase are significantly under-sampled by existing suction
lysimeters.  One recent field study of soil-pore water
sampling systems showed no correlation between organics
found by sampling  compared with  analysis  of soil cores
(1).  Additional studies have shown that xylene (2), DDT
(3), and fecal coliform (4) are not effectively sampled by
ceramic  suction lysimeters.   In  addition, a  number of
inorganic parameters, heavy metals in particular, are also
attenuated by ceramic suction lysimeters. Simultaneously,
a  number of inorganic constituents are leached  from
ceramic suction lysimeters into soil water samples. While
TFE-fluorocarbon  porous   elements are less prone to
significant adsorption or desorption of inorganics,  they
also under-sample organic components.  Additionally, the
large pore sizes of TFE  media restrict their  range of
operation to wetter soils than can be sampled by ceramic
suction lysimeters.

       Ideally,  a  suction   lysimeter should  provide a
sample which accurately represents the soil  liquid phases
at the sampling location.  This would include all compo-
nents, organic as well as inorganic, dissolved in the soil
water and any non-aqueous, i.e.  organic, phases.  While
sampling all components representatively, the  lysimeter
should  also be inert so it does  not leach any chemical
species into the sample. To achieve this ideal goal, the
porous element must be very stable over a wide range of
aqueous and organic conditions and be neither hydrophilic
nor hydrophobic.   Such  a perfect porous  element  is
probably unachievable.   However, elements  made of
porous glass  could form the basis for approaching this
goal. Porous glass elements can be formulated which are
inert to organic and aqueous media over wide ranges of
pH and dissolved components.  Also, the surface structure
can be controlled to moderate its hydrophilicity/hydropho-
bicity.  This  control can be achieved by modifying the
composition of the glass, modifying the thermal process-
ing of the glass, and, if necessary,  by chemically treating
the glass to incorporate desired  chemical species on the
surface.  This paper describes the results  of laboratory
studies aimed at the development  of porous glass elements
for use in suction lysimeters  to provide more accurate
sampling of organic as well as inorganic species.
EXPERIMENTS WITH HIGH SILICA POROUS
GLASS ELEMENTS

  Preparation of the Elements

       A series of porous glass discs were prepared
from powdered high silica borosilicate glass by
sintering.  The solid state sintering mechanism for
different glass systems is  well-known and, to a
large degree, applies to porous glasses. Densifi-
cation and the resultant reduction in porous volume
occurs in two separate regimes when high silica
porous  glasses, as used  in  this research,  are
sintered. The onset of the first stage starts above
750°C, at which point the micro-pores start to
disappear. The driving force for this process is a
reduction in the surface energy. Above 950°C, the
second stage of sintering begins. In this stage, neck
formation occurs between the individual grains of
glass, affecting the macro-pores. It is important to
control the overlap of the two stages, with more
emphasis on the second  stage since this stage
controls the macro-porosity of the system.

       A systematic study was conducted to evalu-
ate the degree of densification when sintering
powdered porous glass. The objective was to gain
control over the pore structure of the elements for
the porous glass suction lysimeter.

       The porous elements were prepared by firing
at different peak temperatures.  The glass was held
at the peak temperature for various times ranging
from 30 to 90 minutes. Heating and cooling rates
were maintained constant for all the samples.
Densities of the resulting glass discs were measured
and normalized against the density of the solid glass
having the same composition (the density of solid
high silica glass with 4-5 % boron oxide is approxi-
mately 2.25 g/mL). The densification and fraction-
al porosity as a function of firing temperature (60
minute firing time) are shown in Figure 1. As can
be seen in this graph, porous glass powder sintered
at 1200°C for 60 minutes achieves an 85 % densifi-
cation. A series of scanning electron micrographs,
 showing the structure of porous glass elements
prepared at temperatures of 1100 and 1150°C for
 60 minutes and 1200°C for 90 minutes are shown
 in Figures 2a-c. These micrographs visually show
 that the pore size and fractional volume decrease
 with increasing firing temperature and firing time.
                                                          658

-------
Pore Size and Hydraulic Conductivity Measurements

       Bubbling pressure, or air entry value, measure-
ments  were performed on  the porous  glass elements
prepared. At first, a lucite disc holder was used.  This
worked well at low pressures but leaked  at higher pres-
sures.   A second holder,  made of stainless  steel was
prepared which worked well  over the full  range of
pressures studied.   Figure  3 shows both the lucite and
stainless steel holders.

       The pore size corresponding to the air entry value
was calculated by the following equation:
       d = 30 Y/P
(1)
where d is the pore size in microns, P  is the bubbling
pressure (the pressure at which air first comes through the
porous disc) in mm Hg, and Y is the surface  tension of
water in dynes/cm at the temperature of  the experiment.
At room temperature, Y is 73.05 dynes/cm.  It should be
noted that the pore size measured by this  procedure is an
effective pore  size; the actual pore sizes  vary  as can be
seen in the scanning electron micrographs (Figure 2).

       Air entry value  measurements were performed on
a number of porous discs prepared over a range of sin-
tering temperatures and times. The results are plotted in
Figures 4 and 5.  In Figure 4, the effective pore size is
plotted as a function of firing time at three different firing
temperatures (1050, 1150, and 1200°C).  Figure 5 shows
the effect  of firing temperature  on effective pore  size
when the  firing  time  is  held constant  at  60 minutes.
These graphs clearly show that the pore size can be varied
down to 2  microns (firing at 1200°C for  60 minutes).

       The flow rate through a series of the porous glass
elements was  also studied.   These data were used to
calculate the hydraulic conductivity and determine the
relationship between effective pore size and hydraulic
conductivity.  The same holder used to  measure  the air
entry value was used to measure the flow rate.  For these
experiments, the  flow was  induced by maintaining  a
vacuum on the porous  glass disc.  The experiments were
performed using a vacuum of 63.5 cm  (25 in.) of Hg.
The fluid used for these experiments was deionized water,
which was drawn from a burette able to measure volume
to 0.1 mL. The flow through each disc was measured for
at least two runs and the results averaged.

       Figure 6 shows the  measured  flow  rate  as  a
function of sintering temperature (60 minute firing  time).
It also plots the pore size against the same abscissa.  This
graph shows that as the sintering  temperature increases,
                  the flow rate decreases along with the pore size, as
                  would be expected.

                         The hydraulic conductivity was calculated
                  using the following equation:
                         K = (Q/t) * (L/A) * h
                                         (2)
where Q is the volume of water flowing through the
element in time, t, L and A are the thickness and
cross-sectional area of the element, respectively,
and h is the pressure differential across the element.
Figure 7 plots the hydraulic conductivity against
pore size.  The data indicate a linear relationship
between hydraulic conductivity and effective pore
size.

       The hydraulic conductivity of the porous
glass elements appear larger than that of ceramic
suction lysimeter elements of the same pore size.
For example, the saturated hydraulic conductivity of
a Soilmoisture Corporation ceramic suction lysi-
meter with an air entry value of 1 bar (pore size 2.1
microns) is 3.36E-7  cm/s.  Figure 7 shows the
saturated hydraulic conductivity of a  2 micron
porous glass element to be about 1E-5 cm/s, almost
two orders of magnitude greater than the ceramic
element.

       A series of nominal 2 micron  pore size
elements were prepared. The measured pore size
and hydraulic conductivities of the elements are
reported in Table 1. The average pore size was 2.1
microns with a standard deviation of 0.8 micron.
The average hydraulic  conductivity was 1.8E-6
cm/s with a standard deviation  of 0.6E-6 cm/s.
The pore size and  hydraulic conductivity of the
elements range by a  factor of about 2.    The
hydraulic  conductivities of these elements are
almost an order of magnitude higher than that of a
comparable  Soilmoisture Corporation ceramic
suction  lysimeter,  although not as high as the
porous glass element reported in Figure 7.
                    Sampling Efficiency

                         Experiments were performed to determine
                   the permeability of the porous glass discs to inor-
                   ganics dissolved in water. Inorganic test solutions
                   contained sodium chloride, barium chloride, lead
                   chloride, and potassium chromate. The results of
                   these tests are shown in Table 2. For all tests, a
                   61.0 cm (24 in.) Hg vacuum was maintained across
                                                        659

-------
the elements.  The concentration of the inorganics was
measured by Direct Coupled Plasma (DCP) Spectroscopy.
The table shows good correlation between the concentra-
tions  in  the  sample  solution as compared to  the  test
solution.  The average ratio of sample to test concen-
trations was 0.88 with a standard deviation of 0.16.

  Problems Observed

       Several problems were observed with the borosili-
cate porous glasses used in the first set of experiments.
One problem was that when the porous glass powder used
to make the elements was exposed to air for extended
periods (hours), the resulting elements were very fragile
(they tended to  crack easily).  It was hypothesized that
this could be due to the formation of internal cracks
caused by drying or by formation of silica gel within the
pores. This problem  was resolved by keeping the porous
glass powder in  water until it  was used to  form the
elements.

       A second,  more serious, problem was clogging of
the elements over time. It was hypothesized that compo-
nents of the glass were leaching into and precipitating in
the interstices of the elements.  To alleviate this problem,
the porous glass was modified by the addition of zirconia
to produce a more durable glass matrix.  The results with
this zirconia glass are reported in the following sections.
 EXPERIMENTS WITH  ZIRCONIA GLASS POROUS
 ELEMENTS

  Preparation  of the Zirconia Glass Elements

        The composition of the porous glass powder used
 to make the test elements was modified by the addition of
 4-5% zirconia.  This modification was made to produce
 a more durable glass which would be more consistent and
 less likely to  clog.  The glass was prepared by  sintering
 the powder at 1150°C for 60 minutes.
  Pore Size and Hydraulic Conductivity

       Table 3  provides  the pore  size  and hydraulic
 conductivity measured on several elements of the zirconia
 glass. The consistency, in terms of pore size and hydrau-
 lic conductivity, among elements was much better than the
 earlier test elements.  However, the pore size was approx-
 imately  3  microns.  Revised heat treatments should  be
 able  to  lower the effective pore size to the 2 micron
 range.
  Sampling Efficiency

       Experiments were conducted to determine
the permeability of the porous glass discs to or-
ganics dissolved in water. Organics used in the test
solutions were ethylbenzene and 1-octanol. For all
tests, a 61.0 cm (24  in.) Hg vacuum was main-
tained across the elements. The concentration of
the organics was measured with a Total Organic
Carbon (TOC) analyzer.

       The test organic solutions were prepared by
carefully placing a layer of the organic chemical on
top of a large beaker of water.  The liquids were
allowed to equilibrate over several days. The water
in the bottom of the beaker was periodically sam-
pled (without disturbing the interface between the
two phases) and its TOC content measured.  When
the TOC content of the water became constant, it
was carefully removed from the beaker so that no
droplets of organics  were entrained.

       The evaporation of the organic component
from the test solutions under vacuum was studied.
Figure 8 shows the significant decrease in the TOC
of the test 1-octanol solution as a function of time
when the solution was kept under a 61.0 cm Hg
vacuum. The data is linear when plotted against the
square root of time, indicating that the rate of
evaporation is controlled by the diffusion of organic
to the surface of the liquid. A similar experiment
conducted with ethylbenzene showed no decrease in
TOC as a function of time. The difference in the
rate of evaporation of the two compounds is due to
the (a) their volatilities, and (b) their polarity.
Since ethylbenzene is less volatile than 1-octanol, it
evaporates  at a slower rate.  Also, since ethyl-
benzene is  more polar than  1-octanol, it  forms
stronger hydrogen bonds with water molecules,  also
retarding its rate of evaporation.

       This observation is very important in the
development of a suction lysimeter for sampling
organics in soil water. Organic components which
tend to volatilize easily from aqueous solution could
be lost due to evaporation. This problem can be
corrected either through capture of the evaporated
organics on an adsorbent, such as carbon. Alterna-
tively, the TOC could be corrected mathematically
using calibration data such as Figure 8.  Capture
and subsequent analysis of volatilized organics
would obviously be a more desirable approach.
                                                         660

-------
       The performance  of the zirconia porous  glass
 elements in sampling organic solutions is summarized in
 Tables 4 and 5.  The ability of the zirconia porous glass
 elements to sample the ethylbenzene solution was excel-
 lent. The difference between the TOC in test and sample
 solutions was always less than 3 ppm, a error of about
 4%.

       The tests  conducted with the 1-octanol solution,
 Table 5,   showed the effects  of octanol  evaporation.
 However, when the TOC measurements are corrected for
 the  octanol evaporation using Figure 8, the results are
 quite good. For  the 1 hour suction period used in these
 experiments, the  correction factor is 1.36.  This  correc-
 tion factor was used to generate the column of corrected
 TOC's in Table 5.  The average value of the corrected
 TOC's is 305.0 ppm compared to 298.9 ppm TOC in the
 test  solution. This represents only a 2% error.

       Thus, these data, while limited, demonstrate an
 excellent ability  to sample  organic  compounds  in  soil
 water.
2. Barbee, G., "A Comparison of Methods for
Obtaining Unsaturated Zone Soil Solution Samples,"
M.S. Thesis, Texas A&M University, College
Station, Texas, 1983, p. 78
3. Stearns, R., R. Morrison and T. Tsai, "Validity
of the Porous Cup Vacuum/Suction Lysimeter as a
Sampling Tool for Vadose Waters," University of
California Engineering Laboratory, Report CE313,
1980, p. 11
4. Dazzo, F. and D. Rothwell, Appl. Microbiol..
27, 1172(1974)
 CONCLUSIONS

       This paper  documents  the  significant progress
 being made toward the development of a porous glass
 suction  lysimeter capable of sampling organic and inor-
 ganic constituents in  soil water.   The ability to make
 porous  glass elements with pore  sizes as small as  2
 microns with  high hydraulic  conductivity  has been
 demonstrated.  Also, initial experiments indicate that the
 elements can accurately sample organics and inorganics in
 water.  Work is continuing to optimize the preparation,
 including the composition and thermal treatment, of the
 porous glass elements and to develop a comprehensive set
 of data  on  the ability of the optimized porous glass ele-
 ments to accurately sample soil water.

       Future work will evaluate optimized porous glass
 elements with simulated  and real  soils, leading to  the
 development of  a suction lysimeter using porous glass
 elements.
REFERENCES

1. Brown, K. W., "Efficiency of Soil Core and Soil-Pore
Water Sampling Systems," EPA Report No. EPA/600/2-
86/083 (September 1986)
                                                      661

-------
            Table 1
     Pore Size and Hydraulic
Conductivity of Porous Glass Elements
                 TABLE 2

TESTS WITH INORGANICS IN AQUEOUS MEDIA
Element
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Mean
Std.
Dev.
Pore Size,
microns
2.3
2.1
3.5
3.5
2.3
1.6
1.1
2.3
1.9
1.3
3.5
1.2
1.6
1.6
2.1
0.8
Hydraulic
Conductivity
cm/s
1.9 x 10*
1.6
2.8
2.6
1.2
1.2
1.2
1.2
1.6
2.5
2.6
2.6
1.2
1.4
1.8 x 10*
0.6 x 10*
Inorganic
NaCl
BaCl2
PbCl2
K2Cr2O7

PH
7.0
7.0
1.8
1.8
4.4
4.4
6.2
6.2
Test
Concentration
(ppm)
196
196
69.4
69.4
102
102
94.7
94.7

Sampled
Cone.
(ppm)
206
198
67.5
66.1
66.4
63.6
81.9
91.1
MEAN
STD.
DEV.
Ratio
1.05
1.01
0.97
0.95
0.65
0.62
0.86
0.96
0.88
0.16
                                    662

-------
           TABLE 3

PORE SIZE AND HYDRAULIC CONDUCTIVITY
OF ZIRCONIA POROUS GLASS ELEMENTS
Pore
Size,
microns
3.0
3.0
2.8
3.0
3.0
Hydraulic
Conductivity,
(cm/s)
7.4E-06
7.6E-06
7.4E-06
5.4E-06
5.3E-06
TABLE 4

SAMPLING OF AQUEOUS
ETHYLBENZENE SOLUTIONS
Porous
Glass
Element

1
2
3
4
5
6
7
8
Total Organic
Carbon, ppm
Test
Solution
72.95
72.95
72.95
72.95
79.07
79.07
79.07
79.07
Solution
Sampled
74.83
72.02
75.92
78.48
76.83
76.11
78.68
78.94
            Table 5

      Sampling of 1-Octanol Solution!
Porous
Glau
Element

1
2
3
4
5
6
7
Total Organic
Carbon, ppm
Ten
Solution
298.85
298.85
298.85
298.85
298.85
298.85
298.85
Solution
Sampled
214.2
211.6
209.6
247.0
220.8
230.4
236.2
Sample Corrected
for Evaporation
291.3
282.8
285.1
335.9
300.3
313.3
321.2
                                       663

-------
           Densification, and Volume Porosity
                     of Porous  Glass
                vs. Sintering Temperature
      % Densification
                                        % Volume Porosity
     900    950    100O    1050    1100   1150    1200    125O
                      Temperature [C]
              —•— % Densification   —°— %Vol. Porosity
FIGURE 1

-------
  FIGURE 2A
  Scanning Electron Micrographs
  of Porous Glass Lysimeter
  Elements

                                                        Firing Temperature: 1100°C
                                                        Firing Time: 60 minutes
Firing Temperature: ]]50°C
Firing Time: 60 minutes

Firing Temperature: 1200° C
Firing Time: 60 minutes


                                        665

-------
 FIGURE 2B
 Scanning Electron Micrographs
 of Porous Glass Lysimeter
 Elements
                                                                                       i

Firing Temperature: 1150°C
Firing Time: 60 minutes
Firing Temperature: 1100°C
Firing Time: 60 minutes
                                         666

-------
FIGURE 2C
Scanning Electron Micrographs
of Porous Glass Lysimeter
Elements


 Firing Temperature: 1200° C
 Firing Time: 60 minutes
                                         667

-------
                      FIGURE 3
Lucite and Stainless Steel Holders for Porous Glass Elements
                               (Scale: heavy grid lines are one inch apart)

-------
cr.
to
                                         PORE SIZE VARIATION
                                               as a (unction of
                                   TEMPERATURE and TIME of SINTERING
                            PORE SIZE (microns)
                                     20
                                               40        60
                                           SINTERING TIME (min)
                                                                   80
                                           • 1200
                                                    • 1150
                                                             •1050
                        FIGURE 4.  Pore size was measured by the
                        technique.
              Ave.Pore Size as a Function
              of  the Sintering Temperature
                   Sintering Time: 60 min
                                                                                                            Pora Size (microns)
                                                                                                          900    950
Figure 5
                 10OO    105O    11OO    115O    1200    1250

                     Temperature |oC]


                    ~m~ pore size In microns
Relation of Pore size
and Hydraulic Conductivity
with Sintering Temperaure
Pora SIZ8 [microns]

8 	 ~~~*— — I 	
4 • ^--^S
.* ,
.
.

Hydraulic Cond Icm/s
:
3; j
*5ss*v^
:

950 10OO 1050 11OO 1150 1200 12
1.000E-O4
1.000E-05
1.00OE-06
1.00OE-O7
SO
Sintering TemperaturelC]
— •— Av». Por» Slzeluml
FIGURE 6
••— Hydraulic Cond. cm/s

                                                                                                                           Hydraulic Conductivity
                                                                                                                                 as a Function of
                                                                                                                             Average Pore  Size
                                                                                                                  Hydraulic Cond. [cm/s]
                                                                                                           1.OOOE-O4F
                                                                                                                           2468
                                                                                                                               Pore Size [microns]
                                                                                                                           • Porous Glass
                                                                                                                                         • Porous Ceramic
                                                                                                        FIGURE 7

-------
                 Octanol-Water System
          Selective Octanol Evaporation Rate
                During Suction Lysimetry
     TOO of the Liquid Sample [ppm]
             0.5        1        1.5        2
                (Suction Time] "I/2 Ihr]'1/2

                        —- TOC
                                                  2.5
FIGURE 8

-------
                     COMPARISON OF MOBILE LABORATORY XRF AND CLP SPLIT SAMPLE
                   LEAD RESULTS FROM A SUPERFUND SITE REMEDIATION IN NEH JERSEY
                                        Jon C.  Gabry,  Ph.D.
ABSTRACT

A mobile laboratory X-ray  fluorescent spectro-
photometer  (XRF) was utilized to determine soil
lead concentrations  1n 2,725  samples obtained
during  a  Superfund Site  remediation  In  New
Jersey.   These  sample  results   assisted  In
guiding  remedial  excavation  activities  at the
site.    One   hundred   twenty-five   site   soil
samples  were  split  and  analyzed  for  lead  by
the  on-slte mobile laboratory  utilizing  a XRF
and  by  a  USEPA   Contract  Laboratory  Program
(CLP)  laboratory  utilizing  atomic  absorption
(AA)  and/or  Inductively  coupled  argon plasma
(ICAP) methodologies.  In  general, XRF results
were usually  higher than the  CLP  split sample
results.    Although   unknown,  XRF    spectral
emission Interference  and/or  Incomplete   homo-
genlzatlon  of  the sample  prior  to  splitting
are  the most   probable  causes  of these  dif-
ferences.   The  XRF  generated  duplicate  and
split  sample mean  RPDs that  were comparable or
better   than   those  obtained   from   the  CLP
laboratory and  another EPA funded  study.

INTRODUCTION

The    use   of   XRF   spectrophotometers   for
elemental   analysis   of    soil    samples  In
analytical   field   screening   programs   at
hazardous  waste   sites   1s   Increasing.   The
purpose  of  this   paper   Is   to  present   a
comparison  of  on-slte  mobile  laboratory XRF
and  CLP split  sample  results  obtained from  a
Superfund  site remediation  In New Jersey.  At
this  site,  a  XRF was  utilized  to  determine
soil  lead  concentrations  In  2,725   samples
obtained   to   guide   remedial     excavation
activities.    The   site,  a   former   used oil
reprocessing   facility.   Is   situated  on the
coastal  plain  with a  uniform sandy soil  type
across the entire site.

METHODS

A portable X-MET 840  XRF spectrophotometer was
used  In   an   on-slte   mobile   laboratory  to
determine  soil  lead concentrations.   Prior  to
the analysis of any site soil  samples, the XRF
was configured  to  the on-slte  soil  matrix and
calibrated.     This    was    accomplished    by
obtaining  a  composite  of  clean  native  site
soil  that  was sent to a CLP  laboratory which
subsequently  generated  10  spiked native  soil
calibration  standards  verified  by  AA  and/or
ICAP   CLP  methods   encompassing   soil   lead
concentrations  ranging  from  20 ppm  to  1000
ppm.    The   on-slte    XRF   was   subsequently
calibrated  with   these   standards   using  the
L-beta  spectral  line  of  lead  to  avoid  any
possible    Inter-element    Interferences    by
arsenic  present  within  the   site   soil.   All
soil  samples  were dried  and  ground  with  a
mortar  and pestle  prior  to  XRF analysis which
followed    the    Instrument    manufacturer's
Instructions  and  utilized  a  counting  time  of
300   seconds.   Quality  assurance   protocols
performed  during  sample  analysis  Included the
analysis  of native soil  blanks and continuing
calibration    verification    standards,   and
duplicate  sample  analysis at  a frequency of  1
per   20  samples.   Based  upon  the  analytical
data    obtained,    a    detection   limit   of
approximately   20  ppm  lead  In   soil   was
estimated  for  the  XRF.

One  hundred twenty-five site  soil samples were
split and  analyzed for lead  with  the  on-slte
XRF   and   by   a  CLP   laboratory utilizing  AA
and/or  ICAP methodologies.   All split  samples
                                                  671

-------
were  homogenized   1n   the   field  prior   to
splitting.    Additionally,    multiple    split
samples    were   submitted   blindly   to   both
laboratories  as  part  of the  quality assurance
program.

RESULTS

Split sample  relative  percent difference (RPD)
values  for  mobile  lab  XRF versus CLP  results
ranged  from 5.2 to 173.2 with  a mean  RPD of
77.4  ±  48.7  (n-76).    These   results-  were
comparable  to an  EPA  funded  study  (1)  which
exhibited  XRF versus  CLP  RPD  results  ranging
from  16.6 to 131.5 with  a mean  RPD of  76.5 ±
45.7  (n-6).

For     Intralaboratory    duplicate    analyses
performed   on  the  split  samples,   the   CLP
laboratory  exhibited a  mean RPD  of 20.0  ± 26.7
(n-9) whereas the  on-slte XRF had a mean  RPD
of  6.6  ± 5.2  (n-4).   Multiple  split  samples
submitted blindly  to  the laboratories  had mean
RPD's  of   163.7  ± 15.7  (n=3)  for  the  CLP
laboratory  and  65.7  ±  48.7  (n-7)  for  the
on-s1te  mobile   laboratory  utilizing the  XRF.
Duplicate analysis  performed  by an Independent
laboratory  contracted  by  the  state  had  a  RPD
of   171.8  with   the   two  analytical   results
differing by  a factor of 13.2.

In  general,  XRF  results  were  usually  higher
than  the CLP split sample results  by  factors
ranging  from  1.09  to 13.91 with  a mean  factor
of  3.01  ±  3.05 
-------
               Screening of Groundwater for Aromatics by Synchronous Fluorescence
                       R. B. Gammage, J. W. Haas, III, and T. M. Allen*
                              Health and Safety Research Division
                                Oak Ridge National Laboratory
                           P.O. Box 2008, Oak Ridge, TN 37831-6383
BACKGROUND, PURPOSE AND SCOPE

Pollution by  petroleum,  oil,  and lubricants is a
ubiquitous  national   problem.     The   aromatic
constituents  contained in these pollutants   can
generally be induced to fluoresce.  The problem of
identifying individual compounds, in what is often a
complex mixture of fluorescing constituents, can be
enhanced  by resorting to a technique known as
synchronous fluorescence (SF) (1).

The  first  application  of  the  SF  technique  for
screening polynuclear aromatic (PNA) contaminants
in groundwater was described in the proceedings of
the First International Symposium on Field Screening
Methods (2).   In the interim, we  continued to
develop the  technique  and  continued  long-term
examination of groundwater taken from specific wells
on the Department of Energy (DOE) reservation.
Our eventual purpose is to show that SF screening is
an acceptable field screening method at Levels I and
II (3).  Our efforts in  achieving this end have been
slowed by limited funding. Nevertheless we are able
to report worthwhile progress; reference spectra and
minimum detection levels (MDLs) were determined
for  17 PNAs, a rapid solid-phase extraction and
concentration method was developed and multi-year
screening of groundwater from a specific well was
continued.

EXPERIMENTAL

A detailed description of the method for making SF
measurements is contained in reference 1.  In making
*Gordon College, Wenham, Massachusetts
the currently reported SF measurements, we used a
Perkin Elmer LS-50 spectrometer.  In order to
optimize  resolving power, compound selectivity and
sensitivity, slit widths of 2.5 nm were used for both
excitation and emission light beams.  The wavelength
difference  between  the  excitation and  emission
monochromators was set at the minimum possible
value for this spectrometer, which was 5 nm.  The
scanning speed used to obtain the reported data was
300 nm/min.

A solid-phase extraction and concentration procedure
was devised for lowering the MLD to 10 ppb or less
of each of the 17 PNAs investigated. A home-made
cartridge  packed with C18 bonded-phase material
was  employed as the solid extractant.  The PNA-
containing water sample (250 mL)  was  first forced
from a syringe through the cartridge.  The  exiting
PNL-free water was discarded. N-propanol (2 mL)
was  next  passed through  the cartridge to elute the
adsorbed PNAs which were now concentrated 125-
fold.  The selection of n-propanol as organic solvent
was based largely one our having at hand, n-propanol
with a low fluorescence background between 250 nm
and 500 nm.

RESULTS  AND DISCUSSION

After extraction  from water and concentration by
125-fold  into  n-propanol,  calibration  curves  of
concentration versus SF response were determined.
The data for the 17 PNAs are summarized in
Table 1.  The noted SF is the wavelength at which a
single peak or  the major  one  of multiple  peaks
occurs; 9  of the 17 PNAs  produced  a single peak at
a A A. of 5 nm.  The MDL for each PNA  is based on
a signal strength three  times  that of the standard
                                                   673

-------
deviation in the background. The MDLs range from
about 1 ppt for benzo(k)fluorathene to about 5 ppb
for pyrene.

A complete screening measurement can be made in
about 5 minutes;  the extraction and concentration
step  takes about  3 minutes and  the spectroscopic
measurement takes about 2 minutes. An example of
this concentration and SF measurement is shown in
Fig. 1 for a sample of groundwater spiked with 5 ppb
of naphthalene.

The  EPA Contract Laboratories  Program requires
participants to be able to quantitatively analyze PNAs
on the EPA  priority list at 10 ppb (4).  We have
shown that the SF method with a concentration stage
(125 x) is capable, in principle, of matching this strict
requirement.

The capability  of the  SF method for qualitative
screening  of  groundwater over a period of 4 years
can  be visualized by referring to Fig. 2.   Between
 1988 and 1990, a constituent fluorescing at about 500
nm has appeared in the  groundwater taken from well
GW15;  the fluorescence in the region of 280  nm
stayed essentially unchanged.  It remains to identify
the composition of the entities fluorescing at 280 nm
and 500 nm and determine  whether  they  are of
 natural or anthropogenic origin.

A field screening method based on UV fluorescence
 has been described by Popp et al. (5) and is listed as
Method FM-25 in reference 3.  A measure of  the
total PNA  concentration  is  made  using  two
wavelength pairs. The method was practiced at two
wood treating sites; some samples were analyzed by
both the UV-fluorescence screening method and the
conventional  Contract  Laboratory Program (CLP)
GC/MS method.  There was an order of magnitude
relationship  between  the  results  of  the UV-
fluorescing screening and the conventional CLP
GC/MS analysis for PNAs.

It remains to apply and compare the SF screening
with the UV-fluorescence screening and CLP GC/MS
techniques. One could then evaluate the advantages
that  would accrue from making  more compound
specific screening measurements using SF.  The SF
technique should have the greater compatibility with
the CLP GC/MS method because the SF screening
can be tailored to measuring the sum of the 18 PNA
compounds on the hazardous substances list.
CONCLUSIONS

Progress  continues  in  developing  SF as  a field
laboratory, quick-screening technique for Level I and
Level II analysis of PNA in groundwater.  An easy
concentration step permits analysis of individual PNA
at concentrations of 10  ppb or less.  Direct  SF
measurements of groundwater taken from a specific
well over a period of  several years show  that
qualitative changes  in fluorescing constituents can
readily be followed.  The next phase of development
should  include  comparative testing  against the
conventional UV-fluorescence screening and CLP
GC/MS methods.

ACKNOWLEDGEMENTS

Research sponsored by the Office of Health and
Environmental Research and Office of Technology
Development, U.S. Department of Energy under
contract DE-AC05-84OR21400 with Martin Marietta
Energy Systems, Inc.

REFERENCES

 1.     Vo-Dinh,  T.,   Gammage,   R.   B.,
       Hawthorne, A. R.  and  Thorngate,  J. H.,
       "Synchronous Spectroscopy  for Analysis  of
       Polynuclear Aromatic Compounds," Environ.
       Sci. Technol., Vol. 12, 1984,  p. 477.

 2.     Gammage,   R.   B.,  Haas,  III,   J.   W.,
       Miller, G. H., and Vo-Dinh, T., "Improved
       Luminescence   Technique  for  Screening
       Aromatic  Contaminants  in  Environmental
       Samples," Proc.,  1st Intl  Symp.  on Field
       Screening Methods for Hazardous Waste Site
       Investigations," Las  Vegas,  NV, Oct. 11-13,
       1988, p. 51.

 3.     U.S.  Environmental Protection Agency, "Field
       Screening Methods  Catalog User's Guide,"
       PB89-134159, EPA/540/2-88-005, Office  of
       Emergency   and   Remedial  Response,
       Hazardous Site Evaluation Division, USEPA,
       Washington, DC, September 1988.

 4.     U.S.    Environmental  Protection   Agency,
       Contract Laboratory Program, "State of Work
       for Organics Analysis: Target Compound List
       and  Contract Required Quantitation Limits,"
       Document  Number  OLM01.0, U.S. EPA,
       Washington, D.C., October 1986.
                                                   674

-------
Fig. 1.  Synchronous fluorescence spectrum of naphthalene
           at 5 ppb after 125-fold concentration
        240 260  280 300 3K  340 360  380 400 420  440 460  460 500 i?0 MO 560  MC
  Fig. 2.  Qualitative SF screening of fluorescing constituents
       in groundwater from a well (GW15) on the DOE
                   Oak Ridge Reservation
       I
              GW15       AA =5
                               nm
               300         400

                    Emission Wavelength (nm)
500
                                                 1990
                                                 1988
                            675

-------
5.
Popp,  S.A.  and  Motwani,  J.  N.  "UV
Fluorescence Field  Screening  Technique
Developed and Utilized Under the Superfund
Program," Hazardous Wastes and Hazardous
Materials  Conference,  New Orleans,  LA,
April 12-14, 1989.
              Table 1.  Synchronous fluorescence maximum and minimum
                      level of detection after 125-fold concentration
                       of 17 PNAs on the hazardous substance list
Compound

Fluorene
Naphthalene
Acenaphthylene
Acenaphthene
2-Methylnaphthalene
Phenanthrene
Chrysene
Pyrene
Anthracene
Benzo(a)anthracene
Dibenzo(a,h)anthracene
Benzo(b)fluoranthene
Fluoranthene
Benzo(h)fluoranthene
Benzo(a)pyrene
Benzo(g,h,i)perylene
Indeno( 1 ,2,3-c,d)pyrene
SF Maximum
(nm)
299.5
311.5
316.0
316.2
319.5
339.5
355.0
366.1
374.0
383.0
392.7
392.7
396.6
400.5
402.5
433.1
459.1
MLD*

3ppt
1.8 ppb
1.6 ppb
22ppt
0.4 ppb
0.4 ppb
0.8 ppb
5.2 ppb
13ppt
0.6 ppb
23ppt
4.2 ppb
5.5 ppb
Ippt
lOppt
3.8 ppb
30 ppb
               *PNA in propanol
               Perkin Elmer LS 50 Spectrometer
               2-1/2 minute scan
               2.5 nm slitwidths
                = 5 nm
                                             676

-------
                IN SITU DETECTION OF TOXIC AROMATIC COMPOUNDS IN GROUNDWATER
                                    USING FIBEROPTIC UV SPECTROSCOPY
                                 J. W. Haas HI, T. G. Matthews, and R. B. Gammage
                                         Health & Safety Research Division
                                           Oak Ridge National Laboratory
                                      P.O. Box 2008, Oak Ridge, TN 37831-6113
INTRODUCTION

Contamination of groundwater with organic compounds is a
common problem at Department of Energy (DOE) facilities
and other sites. Among the more prevalent contaminants
are benzene, toluene, ethyl benzene, and xylenes (BTEX)
which are used individually as solvents and are also major
components of gasoline and other fuels.  Leaking
underground fuel storage tanks are significant contributers
to groundwater pollution.  Polycyclic aromatic hydrocarbons
(PAH) are also possible groundwater contaminants,
originating from fuel leaks or other sources.

Because aromatic contaminants are so ubiquitous, two needs
related to their detection have arisen:
       1. Rapid, cost-effective screening methods are
needed as an alternative to slow, expensive conventional
analyses.
       2. The fate of aromatic contaminants in groundwater
needs to be determined.  An understanding of paths and
rates of migration or biodegradation is crucial to the design
of effective remediation strategies.

One approach  that can meet both of these needs is a sensor
that can detect aromatic compounds directly in groundwater.
Our previous experience with derivative ultraviolet
absorption spectroscopy, DUVAS (1-3), suggested that this
might be a useful tool for this application.  At the first
meeting of this Symposium, we presented results
demonstrating the feasibility of performing groundwater
analysis in situ  using a fiberoptic probe (4). Here we report
on the development of a field-portable spectrometer and a
fiberoptic probe based on that earlier work.

EXPERIMENTAL

Spectrometer construction. An Sin x 14in x 6in aluminum
box was used for the spectrometer (see Figure 1). Although
a more powerful xenon lamp was used for the light source
in previous benchtop experiments, a deuterium lamp
(Hamamatsu, Bridgewater, NJ) was employed in this work.
Higher UV output was obtained from the monochromator
as a result of the superior focal characteristics of the
deuterium lamp.  A homemade power supply was used with
the lamp.  An additional supply was included in the
spectrometer to power the CVI (Albuquerque, NM) Model
DK120 (Albuquerque, NM) monochromator (110mm focal
length) and the photodiode detector housed in the probe
(see description below). The monochromator was stepped
via an external DK-1200 controller, also manufactured by
CVI.  Both power supplies were powered from a small car
battery linked through an inverter.  Light from the
deuterium lamp was coupled into the monochromator
through an f/1  fused silica lens. Light emerging from the
monochromator was coupled into the optical fiber through
another f/1 lens.  The signal voltage returning from the
probe to the spectrometer was sent to an IBM-compatible
personal computer containing a Data Translation
(Marlboro, MA) Model DT-2811 A/D board. Data
collection and processing was handled with modified
SpectraCalc  (Galactic Industries, Salem, NH) software.

Fiberoptic probe. Previous experiments (4) showed that a
one-fiber, detector-in-probe design could increase
significantly the maximum sensing distance of a probe using
UV light over a more traditional two-fiber design. The
design was used to produce the in situ probe diagrammed in
Figure 2. The probe will be described in detail elsewhere,
however a brief description follows. A 25m, 600um core,
high-OH, all-silica optical fiber (Polymicro Technologies,
Phoenix, AZ) was used to bring UV light to the  probe.  The
light emerging from the fiber was focused through a 1cm
optical sample path onto a photodiode detector (United
Detector Technology, Hawthorne, CA) protected by an
optical flat.  Water entered the optical path through slits in
the side of the probe.  Photodiode power and the detected
signal were transmitted through 5-conductor cable. Both
the optical fiber and electrical cable were protected by a
                                                        677

-------
thick-walled air hose connected between the probe and the
spectrometer.  The probe body was constructed of stainless
steel and was about 6in length and l1Ain diameter, fitting
easily into 2in or 4in diameter groundwater monitoring
wells.

Samples. Benzene standards were prepared by diluting
concentrated methanol solutions with distilled water. The
contaminated groundwater samples were collected from well
GW-15 at the Bear Creek Burial Grounds on the Oak
Ridge Reservation.

RESULTS AND DISCUSSION

Analytical capabilities.  The probe was first tested under
laboratory conditions to determine its capability for long-
distance measurements. Improved signal processing and
reduced electronic noise allowed detection of benzene at
concentrations down to less than lug/mL when a 50m fiber
was used.  PAH can be determined even more sensitively,
with lower detection limits ranging to below Ing/mL for
compounds such as anthracene.  For benzene, a linear
calibration of absorbance vs. concentration was obtained
 regardless of fiber length.

 Figure 3 is a spectrum of lug/mL benzene spiked into
 uncontaminated groundwater.  A combination of Fourier
 filtering, smoothing, and second derivative signal processing
 was used to produce the characteristic benzene "fingerprint"
 (Figure 3B) from the almost featureless transmission trace
 (Figure 3A).  One advantage of the second derivative
 approach is that it tolerates reasonable levels of sample
 turbidity without need for a double-beam design, which
 would be difficult to incorporate into our probe. However,
 if a sample is so turbid that little or no light reaches the
 detector, then a measurement can not be made.

 Groundwater samples.  The Bear Creek Burial Grounds on
 the Oak Ridge Reservation has a significant groundwater
 contamination problem.  Pollutants include chlorinated
 solvents and light aromatic solvents.  Over a period of 5
 years we have been monitoring benzene and toluene
 concentrations in groundwater wells at the site. Figure 4
 tracks benzene concentrations in one well  (GW15); the
 trend is similar in other wells.  Although benzene levels
 dropped by a factor of 5 in the first 30 months, they have
 changed little since 1989, levelling off at a disconcerting
 20ug/mL.

 In order to evaluate the performance of the fiberoptic
 DUVAS system, we compared analytical results for GW15
 water with results obtained using a laboratory DUVAS
 instrument and  independent gas chromatographic analysis
 (Purge Method 5030 and GC Method 8000).  The
 comparison is summarized in Table I.
                        Table I
       Benzene in GW15 - Comparison of Methods
Method

In situ DUVAS

Ex situ DUVAS


GC
1
2

1
2
Cone. (ug/mU)

     18.8

     17.9
     17.2

     24.2
     17.4
Because the GC method involved several sample
preparation steps and relied on a one-point calibration for
quantitation, it is not surprising that the greatest variation
was observed in the GC data.  Total GC analysis time was
about 3 hours for the two samples (including the standards).
In contrast, it took approximately 30 minutes to run a five-
point calibration curve and the two samples using the
DUVAS instruments (no sample preparation was required).
Clearly, either DUVAS approach offers considerable cost
savings over the GC method.  Again, the laboratory
DUVAS method also demonstrated better precision than
the GC method. The fiberoptic DUVAS results were close
to those of the laboratory instrument, clearly demonstrating
that analytical performance is  not compromised in the field
instrument.  The results also suggest, at least for this well-
mixed shallow well, that samples collected with a bailer and
properly contained are representative of actual groundwater
concentrations.  It is also notable that an exceptionally short
holding time (about 20  hours) was used prior to laboratory
analysis of the samples. This,  of course, is not  typical.

CONCLUSIONS

A fiberoptic DUVAS probe and field portable  spectrometer
have been fabricated and tested. The instrument provides
reliable measurement of aromatic contaminants in
groundwater, as demonstrated at a local groundwater well.
In future work, we plan to use the device at a jet fuel spill
site where current analytical results are ambiguous.  Depth
profiling within undisturbed wells will be conducted to help
locate the fuel, believed to be in a narrow subsurface zone.

ACKNOWLEDGEMENT

This research was sponsored by the office of Health and
Environmental Research and office of Technology
Development, U.S. Department of Energy, under contract
No. DE-AC05-84OR21400 with Martin Marietta Energy
Systems, Inc.
                                                           678

-------
 REFERENCES

 1. Hawthorne, A.R., Thorngate, J.H., Gammagc, R.B., and
 Vo-Dinh, T., "Trace Organic Analysis Using Second-
 Derivative UV-Absorption Spectroscopy," Proc. 9th
 Materials Research Symposium, Gaithersburg, MD, April
 10-13, 1978, p. 719.

 2. Hawthorne, A.R., Thorngate, J.H., Gammagc, R.B., and
 Vo-Dinh, T, "Development of a Prototype Instrument for
 Field Monitoring of PAH Vapors," Polynuclcar Aromatic
 Hydrocarbons, Ann Arbor Science, Ann Arbor, MI, 1979, p.
 299.
                   3. Hawthorne, A.R., Morris, S.A., Moody, R.L., and
                   Gammage, R.B., "DUVAS as a Real-time, Field-portable
                   Wastewater Monitor for Phcnolics," J. Environ. Sci. Health,
                   A19(3), 1984, 253.

                   4. Haas III, J.W., Lee, E.Y., Thomas, C.L., and Gammage,
                   R.B., "Second-Derivative Ultraviolet Absorption Monitoring
                   of Aromatic Contaminants in Groundwater," Proc.  1st Intl.
                   Symp. on  Field Screening Methods for Hazardous Waste
                   Site Investigations,  Las Vegas, NV, Oct 11-13, 1988, p. 105.
                           TO BATTERY
    A
   8in
    T
                   LAMP
                  POWER
                  SUPPLY
                    TO COMPUTER
            POWER SUPPLY
                DEUTERIUM
                LAMP,	
D-
LENS
MONOCHROMATOR
TO PROBE
                                      14in
Figure  1.  Portable spectrometer for use with fiberoptic  DUVAS  probe.
                                               679

-------
                                        PHOTODIODE DETECTOR

                                          AND AMPLIFIER
                           FOCUSING

                             LENS
OPTICAL FIBER
    PROTECTIVE

       HOSE
                        OPTICAL    PHOTODIODE POWER

                          FLAT     AND RETURN SIGNAL
 Figure 2.  Cut-away view of fiberoptic DUVAS probe.
g

w
OT
I
  200
   250


WAVELENGTH (nm)
300
                                 LLJ
                      m
                      DC
                      o
                      CO
                      m  o
                                  UJ

                                  Q
                                    200
    250


WAVELENGTH (nm)
                                                              300
Figure 3.  (A) Transmission and (B) second derivative spectra

of benzene in groundwater using the fiberoptic DUVAS.
                                680

-------
   200-
0)
ID
LU
O
"Z.
O
O
   100-
     0
     1986
1988
1990
1992
                             YEAR
Figure 4. DUVAS monitoring of benzene in groundwater well GW15.
                            681

-------
                           DEVELOPMENT OF FIELD SCREENING METHODS FOR
                                TNT AND RDX IN SOIL AND GROUND WATER
                Thomas F. Jenkins and Marianne E. Walsh
                    U.S. Army Cold Regions Research and
                                  Engineering Laboratory
                               Hanover, New Hampshire
Martin H. Stutz and Kenneth T. Lang
U.S. Army Toxic and Hazardous Materials Agency
Aberdeen Proving Ground, Maryland
INTRODUCTION

   One of the most serious environmental problems facing the
Army is the presence of soil contaminated with residues of high
explosives at sites where the munitions were formerly manu-
factured, stored, used or demilitarized. TNT and RDX are the
two residues most commonly encountered because these ex-
plosives were extensively produced and do not rapidly decom-
pose. Since TNT and RDX leach through the unsaturated zone
with downward percolating water, they pose an immediate
problem to ground water; thus  contaminated soil must be
treated or isolated. Though laboratory methods for analyzing
munitions residues in soil and water are now available (1,2),
reliable field methods are also desirable so that zones of high
contamination can  be located during initial surveys and the
interface between clean soil and  contaminated soil identified
during  cleanup.
DESCRIPTION OF METHODS

   The procedures for the soil (3,4) and water methods are
similar (Fig. 1). For the soil method about 20 g of soil is shaken
with 100 mL of acetone to extract the munitions residues and
the extract is filtered using a disposable syringe filter. The
methods then depend on  the production of colored reaction
products when separate aliquots of these extracts are subjected
to two simple reaction sequences (Fig. 2). For TNT, a portion
of the extract is reacted with a strong base, and if TNT is present,
the reddish colored Jackson-Meisenheimer anion is produced.
Several other trinitroaromatics also produce reddish anions and
hence are potential interferences (3,5). For RDX another por-
tion of the extract is passed  through a disposable anion ex-
change cartridge to remove  any nitrate or nitrite. Then the
extract is acidified and reacted with powdered zinc (4). This
converts RDX to nitrous acid, which is detected by adding a
Hach NitriVer 3 powder pillow (Fig. 2). The development of a
red or orange color is indicative of the present of RDX or one
of several other military explosives that are potential interfer-
ences (HMX, nitroglycerine, PETN  or nitrocellulose).  The
intensity of the color produced can be measured with a battery-
operated spectrophotometer. The absorbances at 540 nm for
TNT and 507 nm for RDX are linearly related to concentration
(3,4). Detection limits are about 1 ug/g for both TNT and RDX.
                      SOIL SAMPLE
                    Extract With Acetone
TNT PR<
>CEDURE
| Obtain Initial Absorbance (540 nm) |


| Add KOH + Na2 SO3 |

| Fil

ter |
| Obtain Absorbance (540 nm) |
Janowsky Reaction (1 886)
I
RDX PROCEDURE
| Pass Through Anion Exchanger


| Add Zinc and Acetic Acid |


| Filter |


I Add Nitri Ver 3 Powder Pillow


| Obtain Absorbance (507 nm)
Griess Reaction (1864)
                      WATER SAMPLE
           [Pass Through Solid Phase Extraction Cartridge |
                  |   Elute With Acetone  |
                         ^P
                 Process Extract as Shown Above

         Figure 1. Flow diagram for field methods.
                                                        683

-------
 TNT Method
         CH3
                          CH3
    20%) of
water are present, the absorbance is also low. At intermediate
concentrations of water in acetone (1-17%), however, similar
absorbances (±15%) are obtained. If a 20-g sample of wet soil
is extracted with 100 mL of acetone, the 1-17% range of water
in acetone would correspond to soil moisture contents ranging
from 4-83% (on a wet weight of soil basis). This range of
moisture content is typical of the large majority of surface soils.

  3. Reagent contact time. Several experiments were con-
ducted to determine if reagent contact time (1-20 min) had an
effect on the absorbance obtained. TNT solutions were  pre-
pared in acetone containing 3.8% water. The  results indicate
that the absorbance first increases and then declines with con-
tact time, the times being somewhat concentration dependent.
One interpretation is that the rate-limiting step is dissolution of
the solid reactants. If this is true, the concentration of water in
the acetone is also likely to have an  effect, and the optimum
reagent contact time will be sample specific. The rate of decline
of absorbance for excess contact time, however, is relatively
slow. The reason for the reduction in absorbance for longer
contact times is discussed elsewhere  (3,6).

  Based on these tests, a reagent contact time of three minutes
was selected. For high TNT concentrations, three minutes may
be  insufficient to attain the  maximum absorbance,  but the
absorbance will exceed 1.0 A.U. for this case and extracts will
have to be diluted anyway. For very low TNT concentrations or
solutions with less dissolved water, the measured absorbance
will be reduced but the reduction will be very  small. For field
measurements, the ambient temperature can have an influence
on the proper reagent contact time. Observation of the rate of
color development for the standard will assist  in selecting the
most appropriate time for a given temperature.

  4. Stability of filtered solutions. A test was conducted to de-
termine if the colored anions formed from TNT were stable
with time after filtration. Filtration removes the colored anions
from further contact with the solid reactants and results indicate
that absorbance measurements are reliable for at least two hours
(3).

  5. Comparison of TNT concentration estimates for soil ex-
tracts. A series of field-contaminated soils were extracted with
acetone and the extracts were analyzed using the field method
                                                          684

-------
                    Table 1. Comparison of colorimetric and RP-HPLC analysis of soil extracts.
TNT concentration TNB concentration
Colorimetric
Field Method
Sample origin (Hg/g)
Vigo Chemical Plant (Ind.)
Hawthorne AAP (Nev.)
Nebraska Ordnance Works (Neb.)
Nebraska Ordnance Works (Neb.)
Hastings East Indus. Park (Neb.)
Weldon Springs Training Area (Mo.)
Sangamon Ordnance Plant (II.)
Weldon Springs Training Area (Mo.)
Hawthorne AAP (Nev.)
Nebraska Ordnance Works (Neb.)
Raritan Arsenal (N.J.)
Nebraska Ordnance Works (Neb.)
Lexington-Bluegrass Depot (Ky.)
Chickasaw Ordnance Works (Tn.)
Hawthorne AAP (Nev.)
Weldon Springs Training Area (Mo.)
Hawthorne AAP (Nev.)
Raritan Arsenal (N.J.)
Nebraska Ordnance works (Neb.)
Nebraska Ordnance works (Neb.)
Nebraska Ordnance works (Neb.)
Nebraska Ordnance works (Neb.)
Nebraska Ordnance works (Neb.)
Nebraska Ordnance works (Neb.)
13.5
5.49
2.39
592
85.3
4.02
32.7
145
8.67
146
85.3
0.38
15.0

-------
6y the fab method for both RDX alone and the sum of RDX and
HMX (Table 1). The estimates of RDX concentration obtained
by the field procedure were not significantly different from
those obtained by the HPLC procedure for RDX alone or for the
sum of RDX and HMX.

Field Testing
  The soil methods have been field tested at Umatilla, Oregon,
Newport, Indiana, Camp Shelby, Michigan, and Eagle River
Flats, Alaska. The methods were found to be usable under field
conditions and the estimates of analyte concentrations corre-
late well with estimates obtained by the standard laboratory
procedures (3,4).
REFERENCES

I.Jenkins, T.F., Walsh,  M.E., Schumacher, P.W., Miyares,
   P.M., Bauer, C.F., and Grant, C.L., Liquid chromatographic
   method for the determination of extractable nitroaromatic
   and nitramine residues in soil. Journal of the Association of
   Official Analytical Chemists. 72: 890-899 (1989).
2. Jenkins, T.F., Miyares, P.H. and Walsh, M.E., An improved
  RP-HPLC method for determining nitroaromatics and nitra-
  mines in water. U.S. Army Cold Regions Research and En-
  gineering Laboratory, Special Report 88-23, Hanover, New
  Hampshire (1988).

3. Jenkins, T.F., Development of a simplified field method for
  the determination of TNT in soil. U.S. Army Cold Regions
  Research and Engineering Laboratory, Special Report 90-
  38, Hanover, New Hampshire (1990).

4. Walsh, M.E. and Jenkins, T.F., Development of a  field
  screening method for RDX in soil. U.S. Army Cold Regions
  Research and Engineering Laboratory Special Report, Ha-
  nover, New Hampshire (in press).

5. Bost, R.W. and Nicholson, F., A color test for the identifica-
  tion of mono-, di- and trinitro compounds. Industrial Engi-
  neering Chemistry (Analytical Edition). 7:190-191  (1935).

6. Terrier, F., Rate and equilibrium studies in Jackson-Meisen-
  heimer complexes. Chemical Reviews. 82: 77-151  (1982).
                                                          686

-------
                          QUANTIFICATION OF PESTICIDES ON SOILS BY
                                  THERMAL EXTRACTION-GC/MS

                                          T. Junk, T. R. Irvin
                                   Institute for Environmental Studies
                                       Louisiana State University
                                        Baton Rouge, LA 70803

                                       K. C. Donnelly, D. Marek
                                      Agronomy Field Laboratory
                                        Texas A & M University
                                       College Station, TX 77943
 Introduction

 Site  investigations and  cleanup activities often
 require the rapid analyses of  soil  samples for
 semivolatile  environmental  toxicants.    The
 widespread  use   of   pesticides   makes   the
 development    of    rapid    field-deployable
 quantification   methods   for   this   class   of
 compounds  particularly  desirable.     Recently,
 thermal   extraction   techniques   have   been
 investigated as a rapid alternative to classical soil
 analyses by solvent extraction-GC/MS.   Samples
 are   heated  according   to  a  preprogrammed
 temperature profile and evolvig volatiles analyzed
 by in-line GCMS methods. Thus,  tedious wet
 extraction procedures are completely eliminated.
 However, the quantification of  toxicants by this
 procedure  poses  problems.  Thermal  extraction
 efficiencies for toxicants do not necessarily reflect
 solvent extraction  efficiencies.   Indeed, they are
 typically lower and depend on  the physical and
 chemical properties of matrices (soils) as well as
 those of analytes. Furthermore, toxicant extraction
 and  analysis are combined into one procedural
 step and  cannot be monitored  seperately  using
 internal   standards  and  surrogate   standards.
 Isotopic dilution mass  spectroscopy, on  the other
 hand, postulates the free exchange of labeled and
 non-labeled analyte molecules  in  a sample, an
 assumption that is  generally correct for solutions,
 but  questionable  for  solid samples.   Thus, a
 comparison  study  of thermal and wet extraction
 procedures  was  undertaken   using   pesticide-
 -containing soil standards with pesticide contents
 ranging from  1 ppm  to  1000 ppm.   Analytical
 results obtained with a thermal extraction system
 were  compared to those  obtained  by  Soxhlet
extraction  and  subsequent  analysis   using  a
conventional gas chromatograph coupled to an ion
trap mass spectrometer.
Instrumentation

Thermal extractions were performed on a Pyran
Thermal Chromatograph  coupled tob a Finnigan
ion trap mass spectrometer.  The system features
an  all-quartz analytical  flow  path  to  minimize
catalytic   sample  decomposition and  is  fully
automated   to   eliminate   operator-induced
variations. Soil  samples were weighed into porous
quartz crucibles  and  heated  in  the  pyrocell
compartment of  the  analyzer  according  to a
preprogrammed temperature profile. All volatile
components released  during the heating  phase
were  flushed   through  a splitter  assembly  by
helium    carrier    gas    and    subsequently
cryo-condensed onto a fused silica GC column.
Analyte   identification/quantification   followed
conventional GC/MS  procedures.  The  analyte
mixture was separated on a  fused silica capillary
column   and   analytes   were   identified  and
quantified by an  in-line  Finnigan ion trap mass
spectrometer. For this  study,  the  system  was
equipped with a 0.32 mm x  15 m x  2.5 urn DB-5
column.

Soil extracts were analyzed on  a  Varian 3500
Series gas chromatograph using the  same type of
column as before. The identical Finnigan ion trap
mass   spectrometer  was  used  for   analyte
identification     and     quantification      (by
disconnecting-reconnecting the transfer line).

Experimental

A total of  18  soil  samples was prepared and
analyzed  in triplicate  by  both  methods:  six
samples each of Bastrop, Padena, and Weswood
soils contaminated with 1000, 500, 100, 10, 5, and
1 ppm pesticides and spiked with two surrogate
standards.   Both   methods  were  optimized
independently;   different  column   temperature
                                                    687

-------
programming  was  used  along  with  different
tuning parameters for  the  mass  spectrometer.
Quantification for the thermal extraction system
was achieved independently  by using  the  two
surrogate  standards  added to the soil  samples
during   their   preparation   and  by   spiking
isotopically labeled pesticide analogs onto the soil
samples. Quantification for GC/MS was achieved
with surrogate and internal standards in the usual
manner.  All quantifications were  based  on peak
area ratios of analytes to standards for selected ion
chomatograms.    The  following  analytes  were
chosen:  a-hexachlorocyclohexane, y-hexachloro-
cyclohexane,     aldrin,     endosulfan,     bis-
(chlorophenyl)trichloroethane  (DDT),  and  bis-
(chlorophenyl)dichloroethane   (DDE).      For
internal  standards, 3,4,5,6-tetrachloro-2-xylene  ,
9-bromophenanthrene,   2-bromofluorene,   and
4-bromobiphenyl.     For    isotopic     dilution
quantification,     hexachlorocyclohexanes-13C6,
adrin-13C5,   endosulfan-D4,   DDE-D8,   and
DDT-D8. Three soils were chosen for this series
of   experiments,   Weswood   soil,   a   sandy,
organic-lean  soil,  Padena soil, organic-lean with
high  clay content, and Bastrop soil,  a clay  rich
topsoil. The variation of pesticide contents in the
samples   over   three   orders  of   magnitude
necessitated   the use  of variable  split ratios
between  1:10   and   1:40  and  variable  sample
weights   between  10   and  200  mg.   Isotopic
standards   were    spiked   as    solutions   in
dichloromethane directly onto  the soil  samples
immediately  before  thermal  extraction.   The
thermal  extraction  system was   calibrated  by
determining  the  peak area  ratios  for equal
concentrations   of  analytes  and  standards as
average  over four runs. Quantifications for the
conventional  GC  system  were  based  on  the
average of three six  point calibration curves (50 -
550 ng pesticides injected).  Mass spectra were
acquired in full scan  mode, 64-400 amu.

Conclusions

The Pyran system was shown  to  be  capable of
providing  rapid  (35 min)  analyses of  different
soils for  most of the pesticides included in this
study.   Virtually no  background signal from
organic  materials contained in  the  soils  was
observed and clean total ion chromatograms were
obtained. Problems with pesticide  decomposition
were encountered for dieldrin, endosulfan, DDT,
and to a lesser extent for ODD. While DDT and
DDD underwent dehydrochlorination to  alkenes,
endosulfan suffered loss of sulfur dioxide  with
subsequent ring  closure  to the  corresponding
isobenzofuran  derivative,  which  was  thermally
extractable and quantifiable.   Dieldrin  was not
thermally  extractable;   most likely  due to its
conversion to the  corresponding diol by traces of
 water contained in the soils.  DDE formed by
 dehydrochlorination of DDT is indistighuishable
 from DDE contained in the sample. As a result,
 recovery values found for DDE in the presence of
 DDT   can  be  regarded  as  artificially high.
 Comparison of recoveries  and percent  standard
 deviations of recovery based on isotopic dilution
 quantification and internal standard quantification
 allowed  us  to distinguish  between deviation of
 found  from actual  pesticide  content  due  to
 variations of the thermal extraction process in the
 pyrocell  (e.g. uneven packing of the soils in the
 sample crucibles, variable  helium flow through
 the  soils etc., which will  effect pesticides  and
 internal standards in the same manner, since both
 are uniformly distributed throughout the soils)  and
 variations due  to  differences  in  the  thermal
 desorption behavior for different chemicals (wich
 will not show for pesticides and their chemically
 identical  isotopic  analogs).  The results showed
 that no free exchange exists of adsorbed pesticides
 and their isotopic analogs spiked onto  the soils
 before analysis.  Calculated recoveries based on
 internal    standard    quantification    generally
 decreased as pesticide  contamination decreased,
 as shown for Weswood soil samples. However,
 this  did not  hold for all pesticides: poor thermal
 extraction of standards  in combination with high
 thermal  extraction  of  analytes may  result  in
 calculated recoveries in excess of 100 percent.  A
 pronounced dependence of extraction efficiencies
 on  the type  of soil  analyzed was demonstrated.
 Soils with high clay content will generally allow
 lower  recoveries than  sandy soils for  internal
 standard quantification.

 Conventional solvent extraction was  clearly  less
 dependent on the chemical nature of the extracted
 analytes and  their concentrations in the soils than
 thermal extraction. These results etablish thermal
extraction as analytically useful tool for the rapid
semiquantitative,  in  some  cases  quantitative,
analysis   of  soil   samples   for   semivolatile
pesticides; however, with somewhat lower analyte
recoveries and  higher  deviations  than those
obtained in conventional procedures.
                                                       688

-------
 A PORTABLE GAS CHROMATOGRAPH WITH AN ARGON IONIZATION DETECTOR

                  FOR THE FIELD ANALYSIS OF VOLATILE ORGANICS
               Lawrence Peter Kapllnr rt.»mllf
               Roy F. Weston, REAC Project
               GSA Raritan Depot, Building 209 Annex
               2890 Woodbridge Avenue
               Edison, NJ 08837
 Thomas Henry Prichett, Chemist
 US EPA, Environmental Response Team
 GSA Raritan Depot, Building 18
 2890 Woodbridge Avenue
 Edison, NJ 08837
  ABSTRACT

  The Environmental Response Team of the US EPA
  (ERT-EPA) has been deploying field portable gas
  chromatographs (CGs) for the characterization of Su-
  perfund sites and landfill throughout the country. Port-
  able GCs allow rapid determination of volatile organic
  compounds (VOCs) and can both identify and quantify
  these compounds.  Several researchers as well  as the
  ERT have used data  generated via portable GCs to
  estimate and model contamination plumes, fugitive
  emissions, and to direct remediation activities (1,2).

 The Sentex portable GC,  equipped with a high energy
 ionization detector, has been shown to be invaluable for
 determining the levels  of VOCs at several EPA super-
 fund sites.

 The argon ionization detector (AID) has a high ioniza-
 tion energy and will yield a response form a wide variety
 of compounds whose ionization potential is at, or below,
 11.7 electron volts (11.7 eV). These will include many
 aromatic, chlorinated alkanes, chlorinated alkenes, and
 nitrogen and sulfur containing compounds. Of par-
 ticular interest arc the halogenated alkanes which do not
 respond well using photoionization detector based port-
 able GCs (3). The same AID can be easily converted to
 a Electron Capture Detector (BCD). The ECD is very
 responsive to halogenated compounds. The ECD works
 best using nitrogen instead of argon as the carrier gas.
 Carrier gases can be easily switched in the field to take
 advantage of the dual detector capabilities of the Scntcx
 Scentograph GC. For general screening on  a wide
 variety of compounds the ERT has used the Sentex GC
 in the AID mode for most of the Superfund sites inves-
 tigated.

 INTRODUCTION

TheSentexSensingTechnologies,Inc.(Ridgciield,NJ.)
model Scentograph  is a totally portable GC operating
with a 11.7 Ev AID. Internal cylinders of argon carrier
gas and calibration gas,  as well as a 12 volt DC battery
pack allows the Sentex GC to operate without external
support from six to fourteen hours, depending on flow
and oven temperature. All operations are controlled by
the portable lap top computer (PC) interfaced with the
GC. The-PC permits the Scentograph to be automated
and therefore can be set up to run unattended. An
optional communications software package and modem
can be used to control and operate the GC remotely via
phone  links. The PC will also archive all raw and
processed data as well as initial operating parameters.
The Sentex Scentograph GC software can identify a total
of 16 compounds stored in the current operating calibra-
tion library. A one point, one for each of the 16 possible
compounds, calibration is used to quantify the identified
compounds. A  post analysis software routine allows
sample run to  be compared to additional libraries ,
thereby allowing identification / quantification against
hundreds of compounds.

THE SENTEX GAS CHROMATOGRAPH

The GC system  itself consists of  three major com-
ponents: the programmable sampling pump and adsorp-
tion trap, the temperature programmable gc column and
detector block, and the PC data system.

The programmable sampling system and trap consists of
an internal sampling pump which can be programmed
via the PC to draw a sample for various periods of time.
Pump duration ranges from 1 to 999 seconds at a typical
flow rate of 100 cubic centimeters per minute. The
sample is drawn onto an adsorption trap of either Tenax
or Carbosieve where the sample components are con-
centration on the surface of the trapping material. The
trap is then heated from 1 to 4 seconds and backflushcd
to thermally dcsorb  the concentrated sample com-
ponents off the adsorption trap and onto the GC separa-
tions column. Various trapping materials besides Tenax
and Carbosieve are available.

The temperature programmable block provides a stable
heated  zone, from 30 ° C to 140 ° C for both the GC
analytical separations column and the GC detector. The
                                                   689

-------
Sentex block heater yields very stable temperatures,
reducing peak retention time drift present in other
field portable GCs. Since the heater is of a high mass
block design cool down times between run cycle can
be prohibitively long. The temperature ramping
routines available in the software have been found to
be impractical for most of the rapid screening needs
of the ERT. Consequently, all GC field operations, to
date, have used isothermal oven temperatures, typi-
cally 30 to 80° C. The GC oven is very small at 3" high,
3" wide, and 6" long. This constraint has made only
packed GC columns usable. Recent modifications of
the oven dimensions has allowed for the use of mega-
bore capillary columns. At present capillary columns
usable with the Sentex GC are only available through
Sentex.

The dual AID / ECD detector system will respond to
most compounds of environmental interest in either
one  mode or the other. The AID has been used
predominately because it detects both aromatic and
chlorinated hydrocarbons down to the low parts-per-
billion, volume (ppbv) range. The AID has been
found to be very stable and equilibrates within one
half hour after initial setup. A grossly contaminated
AID can be easily reconditioned by baking out the
system at an elevated temperature for a short period
of time. Field experience with the  GC configured in
the ECD mode has found the detector to take several
hours to stabilize. The ECD is also more sensitive
than the AID to the compound it responses to. It can
be more easily contaminated and may take several
hours at  an elevated  temperature to recondition a
contaminated ECD. Both the AID and the ECD have
a linear dynamic range of only 2 to 3 orders of mag-
nitude and be easily saturated at the higher parts-per-
million, volume (ppmv) concentration range. The
dual detector system operates best at the 10 to 1000
ppbv range. For most of the field screening needs of
the ERT this range is suitable. The AID / ECD detec-
tor uses a radioactive tritium foil (Ha) as a beta energy
source. A modified NRC license  available through
Sentex is required No wipe tests are required and air
shipping is not a problem since the activity of the foil
is below DOT restrictions.
CONCLUSION

The Sentex Scentograph gas chromatograph has been
used for the analysis of volatile organic compounds at
various EPA Superfunds by the US EPA Environ-
mental Response Team. Detection limits have ranged
from  5 to 50 ppbv for various aromatic and
chlorinated compounds, when using the argon ioniza-
tion detector. Screening applications include ambient
air analysis, indoor air, stack emissions, and soil gas
surveys. A wide variety of aromatic and halogenated
hydrocarbons have been investigated ( Table 1). The
Sentex GC has yielded data that compared well to
other conformational analysis, such as GC / MS
(Table 2). In several cases the Sentex GC was the only
field  portable GC that could detect chlorinated
alkanes, in the field, at the low ppbv range.

At present only vapor phase sample matrices have
been  sampled by the ERT. Optional equipment can
be used to dynamically purge volatiles from soil and
water matrices for subsequent GC analysis. Initial
evaluation of this optional "purge and trap" apparatus
has shown detection limits for benzene, toluene and
total xylenes in soils to be in the low to mid ppbv range,
depending on soil matrix and GC operating condi-
tions.

 REFERENCES

 1) Spittler, Thomas and W.S. Clifford. " A New
 Method for  Detection  of Organic Vapors in  the
 Vadose Zone." NWWA Conference Proceedings on
 " rharacterJ7atinn and  Mnnitnrinp of the Vadose
 Zone.", Denver, CO, 1985
 2) Clay, P.F. and T.M. Spittler. " The Use of Portable
 Instruments in  Hazardous Waste Site Charac-
 terization." Proceedings of the  National Conference
 nn " Managpnipnt of Uncontrolled Ha7ardniis Waste
 Sites.", HMCRI, Silver Springs, MD, 1985
 3) Kaelin, Lawrence and T.H. Pritchett. " Analytical
 Protocols for Portable Gas Chromatographs as Used
 by the US  EPA Environmental Response Team."
             nf the " 198th ACS National Meeting.
 Fnvirnnmental Chemistry Division.". Miami Beach,
 FL.1989.
                                                 690

-------
             Table 1
COMPOUNDS DETECTED VIA SENTEX AID
IN THE FIELD BY THE US EPA / ERT
      Benzene
      Toluene
      o - Xylene
      m,p - Xylene
      Methyl chloride
      Ethyl chloride
      Vinyl chloride
      Methylene chloride
      1,1 Dichloroethane
      1,2 Dichloroethane
      1,1 Dichloroethene
      trans 1,2 Dichloroethene
      1,1,1 trichloroethane
      Bis 2 chloroethyl ether
      Trichloroethylene
      Tetrachloroethylene.
           Table 2
COMPARISION OF FIELD DATA FOR VINYL
     CHLORIDE SOIL GAS SAMPLES

  SENTEX GC  GC / MS (tube)
  1.15 ppmv    0.54 ppmv
  1.01 ppmv    0.82 ppmv
  2,45 ppmv    3.27 ppmv
  ND(<0.005)   ND(<0.01)
  0.20 ppmv    0.42 ppmv
  0.18 ppmv    0.79 ppmv
  ND(<0.005)   ND(<0.01)
  0.82 ppmv     1.38 ppmv
  7.43 ppmv    4.0  ppmv
  0.006 ppmv   ND(<0.01)
                                          691

-------
            SEAMIST — A Technique for Rapid and Effective Screening of
                              Contaminated Waste Sites
Carl Keller
                                                                            Bill Lowry
The SEAMIST system was developed to allow
the insertion and removal of absorbent
collectors in long drillholes of marginal
stability.  However, the technique has
such attractive attributes that its use
is being extended to many other aspects
of instrumentation and sampling from
drillholes.  The name SEAMIST is an
acronym for Science and Engineering
Associates Membrane Instrumentation and
Sampling technique.  The technique is
simple though not obvious.

The principle feature is a hole liner
made of a tubular fabric or film called
an "impermeable membrane" (Figure I).
The membrane lines the drillhole and is
pressed against the hole wall by a modest
internal pressure  (1-3 psi).   The bottom
of the membrane is gathered together
(inside out) and tied with a cord, "the
tether", which extends up the center of
the hole to a reel, in a canister, at the
surface.  The top of the membrane is
attached to a short pipe extending from
the canister.  The function is simply
that turning the reel winds up the tether
and inverts the membrane, peeling it
outside in from the hole wall.  The
entire membrane can be wound onto the
reel, inside out.  Reversing the reel
allows the membrane to reverse its
motion, extending down the hole under
pressure and everting as it descends to
re-1ine the hole.
                                               CRANK
                                               CANISTER
                                               BASE PIPE
                      Figure 1.   Components of the SEAMIST system.
                                           693

-------
As the everting membrane descends/ it
provides support of the hole wall.  It
also lines the hole like a continuous
packer and prevents flow into the hole.
The primary utility of that function is
that the membrane can carry instruments
into the hole by their being fastened to
the membrane  (e.g./ thermo-couple/
absorbent pads, fiber optics, tubing,
electrodes, etc...) or, the larger
instruments can be carried down on the
tether (e.g., gamma logs, neutron logs,
resistance logs or a video camera (using
a clear membrane).  The interior of the
membrane is isolated from the exterior,
except where ports and tubing allow
access to the geologic medium.

Since the membrane supports the hole
wall, a casing and backfill is not
required in many holes.  Therefore,  one
has access to the entire hole wall for
collection of water or gas samples or for
in situ measurements while the membrane
is supporting the hole wall and sealing
it against flow.  The membrane insertion
into a drillhole can proceed as quickly
as 20 ft/min or faster.  Since the
insertion supports the hole and
simultaneously carries instruments into
place, one can actually case and
instrument a 50 ft hole in under five
minutes.

For long term installations, the interior
of the membrane can be filled with water
or sand (even "dirty sand", since it
doesn't contact the medium to be
measured).  Later, the sand or water can
be flushed or blown out of the hole, and
the membrane and instrumentation can be
removed or replaced.

The obvious utility for field screening
purposes is that the SEAMIST is fast,
relatively cheap, and removable.  What is
also an advantage is that the membrane
nestles around each instrument or
sampling port forming a membrane blister
on the hole wall.  The interior of that
blister,  and the associated instrument or
port, is isolated from other such
blisters at other elevations in the hole.
In fact,  one side of the hole is isolated
from the other side.   In principle,  a
reactive covering on the membrane can be
emplaced and pressed against the entire
hole wall to provide a two-dimensional
map (azimuth and elevation) of
contaminants in the wall material.
The instrument array shown in Figure 2
was designed for monitoring of a steam
flood experiment yet in the planning
stage.  It is an example of instruments
that can be emplaced by SEAMIST.  The
results are yet to come.  The concept is
young and in need of field testing.

Current research of this concept is
funded by DOE (Argonne National
Laboratory) for vadose measurements and
by DOE (Sandia National Laboratory) for
geothermal drilling applications.  Since
the SEAMIST system functions equally well
horizontally and in constricted and
crooked holes,  that is probably its best
application yet to be developed and
tested.
                                            694

-------
     Electrical leads
*

\ Geologic
) medium
10'
2\
'•'.••'.••'.'•S«v



•
•^
.
&
V




Membrane
StainlesssteeK
screen section
   -Electrical contact
        (typical)

(•<—Gas sampling port
    and thermocouple
        (typical)
                                                                      Gas sampling ports
                                                                            Tubing to
                                                                             surface
                                                                         Port through
                                                                          membrane

                                                                       ^-Glass screen
                                                                            spacer
        Figure 2.   Membrane lined monitoring hole  design  for steam flow
                    experiment.
                                         695

-------
                   PORTABLE GAS CHROMATOGRAPH FIELD MONITORING OF PCB LEVELS
                                    IN SOIL AT THE ELZA GATE PROPERTY
                                    Marty R. Keller and Gomes Ganapathi, Ph.D.
                                               Bechtel National, Inc.
                                               Oak Ridge, Tennessee
ABSTRACT

Bechtel National, Inc. (BNI) conducted radiological and
chemical surveys of the Elza Gate property in Oak Ridge,
Tennessee, as part of the U.S. Department of Energy
(DOE) Formerly Utilized Sites Remedial Action Program
(FUSRAP).

Based on site history and preliminary characterizations at
the site, it was determined that polychlorinated biphenyls
(PCBs) were present across the site. Because PCB  analysis
with the use of a portable gas chromatograph (GC)  is
relatively fast and inexpensive, soil sample analysis results
could be made available to help direct the field sampling
program.

This paper provides a discussion of the manner in which
PCBs were monitored in the field during ongoing sampling,
the cost of these analyses, and a comparison of portable GC
screening results with Contract Laboratory Program (CLP)
laboratory results (1).
SITE DESCRIPTION AND HISTORY

The 8.1-ha (20-acre) Elza Gate property is located in the
eastern portion of the city of Oak Ridge, Tennessee, now
known as Melton Lake Industrial Park. Access to the site is
off Melton Lake Drive, near its intersection with the Oak
Ridge Turnpike (Figure 1).

In the early 1940s, the site was developed by the Manhattan
Engineer District (MED) as a storage area for pitchblende
(a high-grade uranium ore from Africa) and ore processing
residues. Five warehouses were constructed on the site,
three of which were used to store radioactive materials.
The Atomic Energy Commission (AEC) used the site until
the early 1970s, when it was vacated. After a radiological
survey and appropriate decontamination activities were
conducted in 1972, the site was deemed acceptable for use
with no radiological restrictions (2). At that time, title to
the property was transferred to the General Services Ad-
ministration and then to the City of Oak Ridge. The prop-
erty was subsequently sold to Jet Air, Inc., and used for the
operation of a fabrication and metal plating facility.

In 1987, at the request of the Tennessee Department of
Health and Environment, Oak Ridge Associated Universi-
ties (ORAU) conducted a survey at the site because of the
possibility of contamination from the metal plating facility.
This survey confirmed the presence of heavy metals and
PCBs at the site.

In October 1988, a preliminary radiological survey of the
site was conducted by Oak Ridge National Laboratory
(ORNL) for DOE. The survey indicated that residual
radioactivity exceeded the criteria for declaring a site
eligible for remediation under FUSRAP. As a result, on
November 30,1988, the entire Melton Lake Industrial Park
was designated a FUSRAP site (3).

In 1988, ownership of the property was transferred to
MECO, a development company. The site is presently
under further development for use as an industrial park. In
addition to the five MED warehouses previously men-
tioned, smaller structures also may have been on site.  None
of the original structures remain, but the concrete pads on
which the warehouses were built are still in place.

One building currently on the property was erected on an
existent concrete pad. A second pad adjacent to this build-
ing is used as a vehicle parking lot and material storage
                                                  697

-------
        J	 2
     FEET (Thousands)
    SCALE APPROXIMATE
Figure 1  Location of the Elza Gate Site

-------
pad. The site has undergone considerable modification
since 1987, and the building is currently occupied by a
manufacturer of storage containers. Modification of the
property is expected to continue as the parcels are sold or
leased.

SAMPLING LOCATIONS

Because PCBs were previously detected at low concentra-
tions over the site area, all samples collected during the
chemical characterization effort were analyzed for PCBs.
Both systematic and biased locations  were sampled. Sys-
tematic samples were collected from the comers and center
of each 61-m (200-ft) grid block.  Using the data from
previous characterizations and information from the prop-
erty history, biased sampling locations were selected. A
hand held auger was used to collect three samples from
each location for analysis. The samples were collected at
0.3-m (1-ft) intervals to a depth of 1 m (3 ft).

CHARACTERIZATION METHODOLOGY

Since conventional characterization using CLP laboratory
protocols is costly and turnaround time required for CLP
analyses is approximately 30 days, on-site screening of
PCBs in soil samples using a portable GC was considered
useful in making real-time decisions on the rationale for
additional sampling locations during the ongoing chemical
characterization.

A Hewlett-Packard 5890 portable GC equipped with a
capillary column and an electron capture detector  for
monitoring PCB levels in soil was used on FUSRAP during
the  Elza Gate site characterization. The ability to  detect
PCBs on site while sampling is taking place is one of the
key advantages of this field screening method (4)  which
was refined by Twomey, Turner, and  Murray (5).  The need
for additional samples can be evaluated using this strategy
while the sampling crew is still in the field. Another
advantage is that this method permits comparison  between
the reproducibility of field data and that of CLP data
because similar equipment and techniques are used.

The extraction procedure used for the Modified Spittler
Method consists of placing 2 g of soil in a test tube and
adding 0.5 ml water, 2.0 ml methanol, and
2.5  ml hexane (6).  The sample is then vigorously  shaken,
and aqueous and organic phases are allowed to form layers.
The hexane layer containing PCBs is  withdrawn from the
top  of the mixture and injected into the GC.

While this extraction method is less efficient than  the CLP
prescribed procedure, it is very rapid and cost-effective.
Using commonly available laboratory equipment,  one
analyst can easily extract 20 samples in less than 2 hr.
The cost of analysis using this screening method (including
sample preparation, analysis, and data evaluation) is be-
tween $50 and $100, compared to $300 for the equivalent
CLP analysis. The savings in cost, coupled with the time
savings (25 min for the field screen vs. 30 days for the CLP
analysis),  warrant the consideration of this screening
method to complement CLP analyses.

The results BNI obtained using this method correlate well
with CLP  laboratory results from the same samples
(Figure 2). The field screening results, while generally
lower than values obtained by the CLP laboratory, give an
excellent indication of locations where PCB concentrations
are elevated and where additional samples should be
collected for laboratory analyses. A comparison of Modi-
fied Spittler and CLP predictions of PCBs in soils is shown
in Figure 3.

Major reasons for variability in results include the
following:

• Since percent moisture was not determined for the screen-
  ing samples, these results were not calculated on a dry
  weight basis.

• Even with the best efforts to homogenize the sample,
  concentrations of PCBs vary within the same sample.

• The extraction technique used with the screening method
  is less efficient than the CLP procedure in extracting
  PCBs from the soil matrix.
                       REGRESSION FIT
     0.10  0.30  050  0.70  0.80   1.10  1.30   1.50  1.70   1.80  2.10
                    SCREENING RESULTS (PPM)

  Figure 2 Correlation between Screening Results and CLP
         Laboratory Data for PCBs at the Elza Gate Property
                                                   699

-------
 10000
  1000
   100
10000
                1        10       100      1000
                    Modified Spinier Value (ppm)

      Figure 3  Comparison of Modified Spittler and CLP
              Predictions of PCBs in Soils and Sediments
              (Adapted From Fowler and Bennett 1987)
CONCLUSIONS

The Modified Spittler Method (5), originally developed by
ABB Environmental Services chemists, has been refined to
determine PCB concentrations in soil that represent excel-
lent comparisons with results generated by CLP proce-
dures.

The Modified Spittler Method has proved to be a fast,
accurate, and cost-effective procedure for determining PCB
concentrations in soil at the Elza Gate FUSRAP site. It
permits collection and analysis of a larger number of
samples during a field characterization and provides direc-
tion during the sampling effort, indicating to field person-
nel where additional soil samples should be collected for
analysis. The result is a more thorough characterization
requiring fewer field sampling efforts.
REFERENCES

1. U.S. Environmental Protection Agency, 1985. "State-
   ment of Work for Organic Analysis, Exhibit D:
   Analytical Methods," USEPA Contract Laboratory
   Program, 7/85 Revision, K1-D134.

2. U.S. Department of Energy, 1972. Letter from
   S. R. Sapirie to J. L. Smith, "Disposal of Excess Real
   Property-Parcel 1228" [Attachment: Authorization for
   Remedial Action at the Melton Lake Industrial Park
   (Former Elza Gate Warehouses)], BNI CCN 057470,
   Oak Ridge, Tenn. (February 3).

3. U.S. Department of Energy, 1988. Letter from
   J. J. Fiore (DOE-HQ) to P. J. Gross (DOE-ORO),
   "Authorization for Remedial Action at the Melton Lake
   Industrial Park (Former Elza Gate Area Warehouses),
   Oak Ridge, Tenn."  [Attachment: Summary for Desig-
   nation of Melton Lake Industrial Park Site], BNI CCN
   057470, Oak Ridge, Tenn. (November 30).

4. Spittler, T. M., 1984. "Field Measurement of Polychlo-
   rinated Biphenyls in Soil and Sediment Using a Portable
   Gas Chromatograph," Environmental Sampling for
   Hazardous Wastes, American Chemical Society,
   pp. 37-42.

5. Twomey, D. M., Turner S. A., and Murray W. A., 1990.
   "The Modified Spittler Method for Fast, Accurate and
   Low Cost Determination of PCB Concentrations in
   Soils and Sediments," Proceedings of the Second
   International Conference for the Remediation of PCB
   Contamination, April 2 and 3, 1990, Houston, Texas,
   pp. 83-89.

6. Fowler, B. A. and Bennett, J. T., 1987. "Screening for
   Characterization of PCB-Containing Soils and Sedi-
   ments," Proceedings of the National Conference on
   Hazardous Wastes and Hazardous Materials,
   March 16-18, 1987, Washington, D.C., pp. 204-207.
                                                   700

-------
                           Real Time Monitoring of the Flue of a Chemical Demilitarization Incinerator
                     S.N. Ketkar
                  Exlrel Corporation
                  575 Epsilon Drive
                Pittsburgh, PA 15238
                  S.M. PENN
                Extrel Corporation
                575 Epsilon Drive
              Pittsburgh, PA 15238
Introduction

Public Law 99-145 directs the Secretary of Defense to destroy
the nations stockpile of lethal unitary chemical warfare agents
and munitions stored throughout the continental United States
by September 30, 1994 [1].   The U.S.Army  has selected
incineration as the best available  technology for destroying
chemical warfare agents [2].  The National Research Council
has  endorsed incineration  as the  method of choice for
chemical agent  destruction.   Maximum  protection  of the
environment, the  general public and the personnel involved in
this destruction is required.  For this reason very stringent
requirements  are imposed for the  maximum  allowable
concentration of the chemical warfare agents in the effluent of
these incinerators. The allowable stack concentrations (ASC)
for the two nerve  agents, GB and VX, are 0.0003 mg/m^, while
the ASC for the blister agent HD is 0.003 mg/m3.  Moreover
analytical instrumentation  is needed  that can detect these
levels in real or quasi real time.

Normal  analytical techniques employed to detect such low
concentrations  use   preconcentration  and separation
techniques and are  very time consuming  [3].  The extreme
sensitivity of atmospheric  pressure  ionization makes  it  a
suitable technique  to detect  low  concentrations  of
contaminants in air [4].  Moreover, the ability of atmospheric
pressure ionizers  to handle very large sample flow rates makes
it possible to use this technique for real time detection.   The
specificity achieved  by tandem mass  spectrometry makes a
system  based on atmospheric pressure ionization tandem
mass spectrometer (API-MS/MS) very attractive for monitoring
low concentrations of pollutants in complex  matrices like stack
effluents. We report here the use of a commercially available
API-MS/MS system to monitor for chemical warfare agents GB
and VX, at concentrations near the ASC levels, in the flue of a
demilitarization incinerators.

The  API-MS/MS system  was tested on the flue  of  a
demilitarization incinerator at the Chemical Agent Munitions
Disposal System (CAMDS) at Tooele  Army Depot in Tooele,
Utah. The incinerator was the liquid incinerator (LIC) which is
used to burn the liquid chemical  warfare agents themselves.
Experimental

The  system used  was  a  commercially available EXTREL
Automatic Stack Sampling Mass Spectrometer (ASSMS). This
system  uses an atmospheric pressure  ionization  source
coupled to a triple quadrupole mass  spectrometer.  This
system is described elsewhere in detail [5], so only a brief
overview will be presented here. A corona discharge operating
at atmospheric pressure, with a discharge current of about 5
uA, is used as a source of primary ions. A low pressure region,
operating at a pressure of about 1 torr is used to break up the
weakly bound water clusters which are always present in a
discharge operating at high pressures.  The declustered ions
are injected into the entrance of  a  triple quadrupole mass
spectrometer. The mass spectrometer has three quadrupoles
each with 3/4" round and 6" long cylindrical rods.  The middle
quadrupole is housed in a collision cell having end plates made
from a leaky dielectric material to improve transmission [6].  A
counting  channel  electron multiplier  together with  a
scalar/counter and a threshold discriminator serves as the
detection system.  The triple quadrupole mass spectrometer
was  used in a multiple  reaction monitoring (MRM) mode to
monitor for the chemical agents.

A heat traced teflon transfer line was  used to connect the inlet
ol the ASSMS system to  the flue of the liquid incinerator. A felt
pad impregnated with silver fluoride was placed inside the stack
end of the transfer line to convert chemical agent VX to it's G-
analog.  This was necessary because vapors of VX can not be
quantitatively transferred  through a transfer line.  The G-analog
of VX is structurally similar to GB and  can be quantitatively
transferred through the  transfer line.   For detecting  blister
agent HD, benzene  charge exchange  was used in  the
atmospheric  pressure  ionization source to  produce  the
molecular ion of HD[7].  A  mechanical pump is used to move
the stack effluent, at rates of up  to  5  L/m,  through the
ionization source.  A syringe pump is used  to  introduce
solutions of the chemical agents in the transfer  line,  for
calibration purposes.

This system was used to  monitor the two nerve agents GB and
VX and  the blister agent  HD. For the case of chemical agent
GB the transitions m/z=141  -* m/z=99 and m/z=141 -* m/z=81
were monitored.  For the case of VX  (in reality G-analog) the
transitions m/z=127 -> m/z=99 and m/z=127 ->  m/z=81 were
                                                          701

-------
daughter ion spectra. For added specificity the transition to the
minor daughter ion has to be monitored. The minor daughter
ion is at m/z = 99 for the nerve agents GB and VX while for
blister agent HD it is at m/z = 63.  For the case of the two nerve
agents the minor daughter ion is less than 10% as intense as
the primary daughter ion.  Consequently this system has a
higher detection limit when monitoring this minor daughter ion.
For the two nerve agents the detection limit of the system for
the minor daughter ion is about 20 ASC. For the case of the
blister agent the minor transition is only slightly weaker than the
primary transition and consequently the detection limit for this
transition is 1.43 ASC.

This system was also tested on the exhaust ot a filter stack.
The filter  stack provides a very clean matrix compared to the
matrix provided by the flue of the liquid incinerator. The results
of the tests performed on the exhaust of the filter stack  are
summarized in Table 2.

                         Table 2

          Results of Statistical Analysis ( Filter Stack)
Chemical Agent
GB
VX
HD
Decision Limit
0.08 ASC
0.12 ASC
0.05 ASC
Limit of Detection
0.15 ASC
0.23 ASC
0.13 ASC
Conclusions.

It has been demonstrated that a system based on atmospheric
pressure ionization tandem quadrupole mass spectrometry can
detect, in the flue of a chemical demilitarization incinerator,
nerve agents GB and VX and blister agent HD near the
allowable stack concentrations.  In the absence of any matrix
effects, this system can delect these agents at  concentrations
below 0.25 ASC.  This has been demonstrated  for the case of
the filter stack exhaust.

Credits

This work was supported by The Program Manager for
Chemical Demilitarization, Aberdeen  Proving Grounds,
Edgewood, MD. under contract no. DAAA15-86-C-0107. We
thank Lanny Davis of Chemical Agent and Munitions Disposal
System, Tooele Army Depot, Tooele, UT for his assistance
during the course of this work.
monitored.  For the case of HD the transitions m/z = 158 —> 109
and m/z = 158 —> 63 were monitored.

Results and Discussions

Calibration runs were performed on GB, VX and HD on four
consecutive days, to obtain the detection limit of the system for
detecting these agents in the flue of the liquid incinerator.  On
each day the system was challenged with six different
concentrations of the chemical agents.  We used six challenge
concentrations in the range of .5  ASC to 20 ASC. The ASSMS
system response at each challenge concentration was
measured in triplicate.  The above procedure was repeated on
four days. The ASSMS system response thus obtained was
converted to a found concentration. Regression analysis was
performed on the resulting data to obtain statistical parameters
pertinent to describing the performance of the system.

We followed the procedure used  by U.S. Army Toxic and
Hazardous Materials Agency (USATHAMA) to determine
certified reporting limits[8]. This  procedure consists of
performing a weighted linear regression of the fount v/s target
concentration. Both upper and lower confidence limits, at any
desired confidence level, can then be obtained.  In this work
we used a confidence level of 95%.  Based on this analysis
statistical parameters like the limit of detection (LOD) and
decision limit (DL) can be calculated. LOD is the smallest true
concentration that will be consistently detected.  If the analyte
is present in the sample stream at the LOD concentration level,
the probability that it will be detected is at least 95%. True
concentrations above the LOD are deemed detectable.  DL is
the maximum found concentration that will result, with a
probability of 95%, from a stream containing no analyte.
However, since the DL is usually  less than the LOD this will not
constitute a false positive. These two statistical parameters
contain all the information needed to assess a systems
detection performance.  The results of this statistical analysis is
given in Table 1.

                        Table 1

    Results of Statistical Analysis ( Liquid Incinerator Flue)
Chemical Agent
GB
VX
HD
Decision Limit
0.15 ASC
0.84 ASC
0.6 ASC
Limit of Detection
0.6 ASC
1.79 ASC
1.15 ASC
                                                                The results in Table 1 refer to the primary transition in the
                                                              702

-------
References

1.     Public Law  99-145, "  Department of  Defense
       Authorization Act 1986", Nov. 8, 1985, Title XIV, Part B
       -  Chemical Weapons, Section 1412 -  Destruction of
       Existing Stockpile  of  Lethal Chemical Agents and
       Munitions.

2.     Chemical Stockpile  Disposal Program Report SAPEO-
       CDE-IS-87005, Sep. 1987.

3.     S.J.  Smith, " Detection  methods  for highly toxic
       organophosphates",7a/an/a, 1983, 10,725.

4.     S.N.  Ketkar, S.M.  Penn, W.L.  File, "  Real time
       detection of part per trillion levels of chemical warfare
       agents in ambient air using atmospheric  pressure
       ionization  tandem quadrupole mass spectromelry ",
       Anal.  Chem., To be Published, Feb. 1991.

5.     S.N.Ketkar,  J.G. Dulak, W.L. File, J.D. Buchner  ,
       Seksan  Dheandhanoo,  " Atmospheric pressure
       ionization tandem mass spectrometer system for real
       time detection of low level pollutants in air  ",  Anal.
       Chem., 1989, 61, 260.

6.     S.N. Ketkar and W.L. File," Transmission
       characteristics of a triple quadrupole mass
       spectrometer with leaky dielectric endplates", Rev.
       Sci. Instrum.. , 1988, 59, 387.

7      S.N. Ketkar, J.G. Dulak, S. Dheandhanoo and W.L.
       File," Benzene charge exchange at atmospheric
       pressures for low level detection of pollutants in air",
       Anal. Chimica Acta,  To be published.

8.     Department of the Army, U.S. Army Toxic and
       Hazardous Materials Agency, "Sampling and Chemical
       Analysis Quality Assurance Program", April 1982.
                                                        703

-------
                       FIELD EVALUATION OF THE BRUKER MOBILE MASS SPECTROMETER
                                      UNDER THE U.S. EPA SITE PROGRAM
             S.M. Klainer, M.E. Silverstein,
             VJL Ecker, and D.J. Chaloud
      Lockheed Engineering A Sciences Company
                 Las Vegas, Nevada
                        S. Billets
          U.S. Environmental Protection Agency
      Environmental Monitoring Systems Laboratory
                  Las Vegas, Nevada
INTRODUCTION

The Mobile Environmental Monitor (MEM), afield-deployable
gas   chromatograph/mass   spectrometer   (GO/MS)
manufactured by Bruker Instruments, Inc., was demonstrated
to assess the ability of this technology to perform in-field
analyses of organic contaminants in soil and in water.  The
demonstration, conducted under the U.S.  Environmental
Protection Agency (EPA) Superfund Innovative Technology
Evaluation (SITE)  Program, took place at two Superfund
sites in Massachusetts during August and September 1990.
Detailed studies and quality assurance designs provided the
structure for  each field  demonstration.  Real-world  and
performance-evaluation samples were analyzed by the MEM
and  by equivalent, standard EPA methodologies.   Data
generated by the MEM were compared to that obtained from
the EPA  methods and  were  used  to  assess specific
performance characteristics.
BACKGROUND

The Superfund Program was established by Congress in
1980  to  identify,  prioritize, and  remediate the nation's
uncontrolled hazardous waste sites. Because the problems
associated with hazardous waste sites have proved to be far
more  complex and diffuse  than  anticipated,  Congress
enacted the Superfund Amendments and Reauthorization
Act of 1986 (SARA). Under SARA,  EPA was charged with
effecting  more  timely  and  cost-effective  remedies for
Superfund site remediations.  The SITE Program satisfied
the requirement (SARA, Section 311[b]) that EPA establish
a  program designed  to  accelerate the  development,
demonstration, acceptance, and use of promising alternative
or innovative technologies targeted to meet the objectives of
the overall Superfund Program.
Two categories of technologies are recognized in the SITE
Program:  (1)  treatment technologies that may serve as
alternatives to land  disposal of hazardous waste,  and
(2) monitoring and measurement technologies for identifying
contaminants. Monitoring and measurement technologies that
are accepted into the SITE Program are evaluated as part of
the Monitoring and Measurement  Technologies Program
(MMTP). Under the SITE Program, the MMTP is administered
by the EPA Office of Modeling, Monitoring Systems and Quality
Assurance (OMMSQA) through the Environmental Monitoring
Systems Laboratory in Las Vegas, Nevada (EMSL-LV).  The
Bruker MEM was demonstrated under the MMTP.

The primary purpose of the MMTP is to provide developers with
the means to demonstrate innovative technologies that could
be used as alternatives to the current systems of detecting and
assessing the extent of pollutants at hazardous waste sites.
The  focus of these demonstrations is to evaluate  fully-
developed technologies, thereby making performance and cost
effectiveness data available to interested parties. Superfund
decision makers will thus have the information that is necessary
to consider whether or not these technologies are of potential
use in future site characterization or remediation projects.  The
developers of the monitoring and measurement technologies
are identified from as  many sources as possible, including
solicitations in relevant trade journals, periodicals, seminars,
and professional conferences. Once the developers reply to a
solicitation, the SITE Program representatives begin  an
evaluation process to determine the feasibility, utility, and need
for each technology.

Bruker Instruments, Inc., of Billerica, Massachusetts, responded
to one of these solicitations and its MEM was identified by EPA
as a promising candidate for a field demonstration under the
MMTP. The MEM, designed for the on-site analysis of organic
contaminants, is a mobile mass spectrometer (MMS), optionally
coupled to a gas chromatograph (GC) or a thermal desorption
                                                       705

-------
sampling probe. Currently, full-size (therefore, nonmobile)
laboratory gas chromatography/mass spectrometry (GC/MS)
has been the preferred EPA approach to  identifying and
quantifying organic contaminants at Superfund sites.  This
technology  analyzes  compounds  on  the  basis  of  the
molecular  weight,  retention  time,   and  characteristic
fragmentation patterns of their chemical components. The
primary disadvantages of conventional GC/MS systems are
instrument  size, power demand, and sensitivity to external
factors  (e.g., temperature, humidity, and vibration).  The
development of an MMS rugged enough  to withstand a
variety  of field conditions is  of considerable interest to
parties  responsible for  contaminant monitoring  and  for
remediation of Superfund sites.  Newly developed mobile
systems, such as the Bruker MEM, appear to have attained
satisfactory   levels   of   stability,  power  usage,  and
compactness for field applications.
 MEM SITE DEMONSTRATION

 The purpose of the demonstration was to evaluate the ability
 of the MEM to analyze polychlorinated biphenyls (PCBs) and
 polynuclear aromatic hydrocarbons (PAHs) in soils and to
 analyze volatile organic compounds (VOCs) in water, under
 field conditions at Superfund sites.  The demonstration
 focused on the capability of the  instrument to generate
 rapid, cost-effective, and reliable PCB, PAH, and VOC data
 from real-world samples.  The demonstration was used to
 compare MEM  performance to similar analytical method
 performance as required under the EPA Contract Laboratory
 Program (CLP) or the Resource Conservation Recovery Act
 (RCRA). Detailed project and quality assurance plans were
 prepared which defined the sampling and analysis protocols,
 the experimental design, the quality assurance and quality
 control (QA/QC) requirements, the  data base management
 system, the health and safety considerations, and proposed
 data analysis and  evaluation methods.

 For this demonstration, real-world  samples were collected
 from two National Priorities List (NPL) sites in Massachusetts
 (EPA Region 1). These sites were  selected on the basis of
 documented (Record of Decision)  presence of analytes of
 interest: e.g., PCBs in soil and VOCs in ground water at one
 site and PAHs in soil at the other.  A screening analysis by
 the MEM identified the collection points (i.e., low, medium,
 and high  concentration levels) for five samples  in each
 compound class (PCB, PAH,  VOC).  Bulk samples were
 collected, homogenized, and split  (bottled) into replicates.
 For each compound class, each of  the five distinct samples
 was split into seven  replicates for analysis on site (or near
 site) by the MEM, and off site by standard EPA methods. In
 addition, standard reference materials (SRMs), and blank
 samples were sent to all analysis  locations for variability,
 detection, and other data quality assessments. This process
worked well for the PCB and  PAH soil samples;  however,
remediation activities  at the chosen site precluded the
collection of VOC-contaminated ground-water samples.
Instead, surface water collected from the other site was spiked
with different concentrations of VOCs.
ANALYSIS METHODS

The MEM analytical methodologies for field analysis were
developed by the Trace Analytical Chemistry Laboratory of
Tufts University. The off-site confirmatory laboratories used
EPA-approved methods for analyzing demonstration samples.
This process minimized intermethod biases because the EPA
methods were chosen for their similarity to the MEM field
methods. A brief overview of each method is provided below.

PAHs in soils:  For the MEM, soils were first extracted with
methy lene chloride. The extracts were then thermally desorbed
onto a 3-m chromatography column interfaced with the mass
spectrometer. The data were collected and interpreted in a
manner similar to that used for EPA methods. For the off-site
laboratory, samples were first extracted by RCRA Method 3550
(methylene chloride, sonicatton extraction). The extracts were
then analyzed by RCRA Method 8270 (GC/MS analysis for
semivolatile organic compounds).

VOCs in water: For the MEM, analytes were purged from the
samples  onto  Tenex tubes.   The tubes  were thermally
desorbed onto  a 30-m  fused  silica  capillary column  for
compound separation.    Compound  identification and
quantification were performed using quadrapole MS. The off-
site laboratory employed RCRA Method 8260 (capillary column
GC/MS for volatile organic compounds).

PCBs in soil:  For the MEM, soils were first extracted with
hexane. The extracts were then thermally desorbed onto a 3-m
chromatography column connected to the mass spectrometer;
the final  concentrations were  calculated by  quantitating
individual  chlorination  levels   (mono-  through
octachlorobiphenyl, only). For the off-site laboratory, samples
were first extracted according to RCRA Method 3550 followed
by GC/MS analysis in  concordance with  the CLP  high-
concentration protocol. Like the MEM  method, congener
counting  was used in sample quantification. RCRA Method
3640 (gel permeation chromatography cleanup) was used
when necessary for high-concentration  samples.  (NOTE: The
conventional GC method for PCB analysis was not used; this
method  measures arochlors and, thus, would  not  have
provided  proper intermethod comparison.)
MEM EVALUATION

Data Analysis: The data from all analysis sites were compiled
into  one fully documented data base.   Data were then
subjected to a detailed verification process.   Following
                                                         706

-------
verification,  a variety  of  data analyses were  performed,
including intermethod comparisons (between the MEM and
the off-site laboratory results), reproducibitty estimates (from
replicate analyses on the  same  instrument),  and the
evaluation of data quality indicators. Direct comparison plots
and a variety of statistical routines were used to interpret the
data. Figure 1 and Table 1 represent selected demonstration
results. Figure 1 presents the comparisons of real-world and
SRM samples forthe PCS trichlorbiphenyl congener. Table 1
presents precision, accuracy, and bias information forPAHs
based on SRM  analyses by the MEM and by the off-site
laboratories.
       Figure 1.  Linear regression of trichlorobiphenyl.

                    250H	
                    200-
                   '150-
                  <  100-
                      50-
                                                                       Plot Symbol*
                                                              A- Real-World Samplt   Q- SRM
                                   50
                                             100
                                                       150
                                                                 200
                                                                           250
                                                                                     300
                                                  OFF-SITE ANALYSIS (ppm)
       Table 1. Precision and accuracy data for PAHs based on SRM analysis.
                                                                                                  -250
                                                                                                  -200
                                                                                                  -150
                                                                                                  -100
                                                                                                  -50
                                                                                                  -0
                                                                                                350
Analyte
(Theoretical SRM Value)
Naphthalene
(50.7)
Acenaphthylene
(46.4)
Fluorene
(45.9)
Pyrene
(50.6)
Chrysene +
Benzo(a)anthracene
(97.9)
Lab
MEM
Off-Site
EMSL-LV
MEM
Off-Site
EMSL-LV
MEM
Off-Site
EMSL-LV
MEM
Off-Site
EMSL-LV
MEM
Off-Site
EMSL-LV
Mean*
28.6
32.8
29.7
10.9
20.4
24.6
29.2
31.1
28.0
40.7
34.6
32.4
86.8
66.1
63.3
%RSO
55.1
11.3
1B.S
39.1
11.9
20.1
25.1
8.0
19.6
16.2
9.6
23.4
26.5
9.0
23.5
%Bias from
Theoretical Value
-44.8
-34.9
-40.8
-76.3
-56.9
-46.1
-36.8
-32.5
-39.0
-19.0
-30.8
-36.8
-11.1
•32.6
-34.6
%Bias from
Off-Site Value
•15.2
-
•9.1
•45.0
-
+25.0
•6.5
-
•9.7
+ 17.1
-
-8.6
+31.8
-
-3.0
%Bias from
EMSL-LV Value
-6.7
+ 10.0
-
•56.0
•20.0
-
+3.6
+ 10.7
-
+28.1
+9.4
-
+35.9
+3.1
—
                 •Means bated on analysis ol: 40 replicates lor MEM; 30 replicates lor oil-site; 7 replicate* tor EMSL-LV. Units are in ppm.
                 %RSO » percent relative standard deviation
                 SRM - Standard Reference Material
                 EMSL-LV - Environmental Monitoring Systems Laboratory, Las Vegas. Nevada
                 MEM • Mobile Environmental Monitor
                                                           707

-------
Instrument Characteristics: The primary advantages of the
MEM are its portability  and ruggedness.   Rechargeable
batteries supply  all  power  required by  the MEM, and
logistical requirements are minimal and easily fulfilled. The
use of purified ambient air as the carrier gas eliminates the
need to  transport  compressed   gas  cylinders.   The
demonstration plan called for the analysis of 13 samples per
day; the analysis team had difficulty meeting this sample
throughput requirement.  Although the MEM is easy to
operate under normal conditions, a skilled operator is
required to correctly diagnose and repair malfunctions.
NOTICE

Although the research described herein has been funded
wholly or in part by the U.S. Environmental Protection Agency
under Contract Nos. 68-03-3249 and 68-CO-0049 to Lockheed
Engineering & Sciences Company, it has not been subject to
Agency review and therefore does not necessarily reflect the
views of the Agency, and no official endorsement should be
inferred. Mention of trade names or commercial products does
not constitute endorsement or recommendation for use.
FUTURE MOBILE GC/MS WORK IN THE SITE PROGRAM

On the basis of  data  collected and observations made
during  this  demonstration,  several  issues  have been
identified that must be addressed before the MEM or other
MMS instruments  can  be employed in Superfund  site
monitoring,  characterization,  and  remediation  activities.
These  issues  include: (1)   method  development and
procedural documentation, (2) development of standardized
QA/QC requirements and  limits,  (3) development of data
reporting standards for field analytical measurements, and
(4) development of a detailed troubleshooting  guides and
training programs.   These  issues  do   not  necessarily
represent  problems with the technology itself;  several are
external factors or policy issues that require attention before
the EPA can use MMS instruments as reliable field analytical
devices.

EPA is considering additional mobile mass spectrometer
evaluations at EMSL-L.V under the SITE Program.  Future
laboratory evaluations will concentrate on (1) separating
variability  associated with  the instrument from variability
associated  with  the   method,   (2)  formalizing   QA/QC
procedures, and (3) establishing consistent data reporting
procedures   for  field   applications.     Additional  field
demonstrations and evaluations will follow.
ACKNOWLEDGMENTS
The EPA would like to thank the following organizations for
their participation in the demonstration and evaluation of the
Bruker MEM under the SITE Program:

  EPA Region 1
  Bruker Instruments, Inc.
  Tufts University
  Lockheed Engineering & Sciences Company
  S-CUBED
                                                         708

-------
                                THE  DITAM  ASSAY
                       A FAST, FIELDABLE METHOD TO DETECT
              HAZARDOUS WASTES, TOXIC CHEMICALS, AND DRUGS
                                   Dr. Cynthia A. Ladouceur
             U.S. Army Chemical  Research, Development and Engineering Center
                         Aberdeen Proving Ground, MD 21010-5423
PURPOSE AND SCOPE OF ASSAY

The DITAM (Diffusion Through A Membrane) assay
is   fieldable,  fast,   extremely  easy  to  use,
inexpensive, and can be used to detect one or
several substances simultaneously.  The DITAM
apparatus  was developed along with the DITAM
assay. This new assay is in the initial stages of
development for the detection of several small
molecular weight substances. Model systems have
been tested for the detection of progesterone and
Ricinus communis. a toxin.  Based on the initial
experimental results, it appears that the DITAM
assay will be useful for the rapid detection of a
wide variety of substances in the field.  Examples
of these substances include small molecular weight
hazardous wastes, toxic  chemicals,  and drugs.
Continuing research involves further modifications
of this assay to  enable the detection  of large
molecular weight substances such as proteins from
infectious  organisms  and  antibodies  directed
against these organisms.

MATERIALS AND METHODS

There are  two version of the  DITAM assay  and
apparatus. The original DITAM apparatus consists
of a hand-held polystyrene "box" with two to four
chambers.   The  chambers are separated  by
semipermeable  membranes.    Because of  the
difficulty  in producing  these  "boxes"  in  the
laboratory, the design of the DITAM apparatus was
modified. The modified version consists of a "bag"
in a 12x75 mm capped test tube or an alternate
pocket-sized vial.    The  "bag"  is   actually  a
cylindrical semipermeable membrane which is filled
with reagents and tied or clamped at both ends.
These  reagents  consist of specific antibodies,
enzyme-labeled antigens, and a buffer solution. To
date, several membrane types have been tested for
their usefulness in this assay.  The membranes
must  be flexible  and have pores  which allow
molecules of a  specific size to diffuse through
easily and  rapidly.   The appropriate molecular
weight cutoff of the membrane must be selected in
order to retain the antibody molecules within the
"bag" and enable the enzyme-labeled antigens and
antigens  in the test sample to  pass  through the
"bag."  All  reagents and  supplies needed for  a
modified  DITAM assay are illustrated in the figure
on the following page.

When performing a modified  DITAM assay, an
individual is supplied with  a  test  tube  which
contains the reagent-filled "bag." To perform this
assay in the field, an individual only needs to add
the test sample to a fill line marked on the tube,
shake the  tube  for  approximately  one to two
minutes, add the enzyme substrate, shake the tube
again, and observe the tube for a color change. All
instructions can be printed on the tube.

The test sample can be  liquid or solid. Solid test
samples,   such   as  dust  particles,   can  be
concentrated on cotton  swabs  and placed in test
tubes along with a buffer solution.

The color of the reaction solution depends on the
enzyme and the degraded enzyme substrate. When
horseradish peroxidase is used to label the antigens
and tetramethylbenzidine plus hydrogen peroxide is
used as the enzyme substrate, a turquoise colored
solution inicates  a  positive reaction  and a clear
solution indicates a negative reaction.
                                               709

-------
                                              REAGENTS AND SUPPLIES NEEDED FOR
                                                  A MODIFIED DITAM IMMUNOASSAY
                                             0
                                           © 0
                                             ©
                                           © ©
                                             ©
    semi-permeable
      membrane
test tube
                                                   specific
                                                   antibody
                                                   enzyme-labeled
                                                   antigen
                          enzyme
                          substrate
                                                       specific
                                                       antibody
                                                       bound to
                                                       enzyme-labeled
                                                       antigen
                                                       antigen from
                                                       test sample
 •     degraded
> •   enzyme
 •     substrate
RESULTS AND CONCLUSIONS

Based  on  the  initial experimental results,  the
modified DITAM assay can be completed in three to
five  minutes.    Thus,  it  satisfies  an  initial
requirement of speed in obtaining  assay results.
This assay is extremely easy to perform in both the
field and in the laboratory.  Since all of the assay
instructions can be printed on the tube, little or no
training is required in order to perform this assay.

In order to achieve the maximum contrast between
the positive and negative reaction solution  colors
(turquoise for the former and clear  for  the latter),
the appropriate concentration of reactants must be
employed.    The   use   of  an   inappropriate
concentration of antibody  molecules  inside  the
"bag" can result in  false positive reactions (darker
blue coloring in negative test samples).  Although
there may  be pale blue coloring in the  negative
samples due to background reactions, this can be
                           kept to  a minimum  if  the  concentrations  of
                           reactants are carefully calibrated for the assay.
                           This procedure is standard when developing any
                           new immunoassay.

                           Due to  the  versatility of the  assay,  it can  be
                           developed for use by many government agencies
                           and by the private sector as well. Possible future
                           applications include the following: (1) detection of
                           chemical warfare agents in the field and chemical
                           warfare treaty verification, (2) detection of drugs in
                           humans and animals, (3) protection of humans from
                           environmental contaminants (i.e., pesticides and
                           toxic chemicals in dust and water supplies), and (4)
                           detection of hormones (i.e., determining levels in
                           hospital patients and athletes). Continuing research
                           and further modifications of the DITAM assay and
                           apparatus should enable  the detection of large
                           molecular weight substances such as biological
                           warfare  agents  and  proteins  from  pathogenic
                           organisms (i.e., in food, humans, and animals).
                                                 710

-------
                  RAPID SCREENING OF GROUND WATER CONTAMINANTS
                     USING INNOVATIVE FIELD INSTRUMENTATION
            Amos Linenberg
    Sentex Sensing  Technology,
         Ridgefield,  NJ  07657
Inc.
        David Robinson
Sentex Sensing Technology, Inc.
     Ridgefield, NJ  07657
With the increased use of on-site
monitoring equipment at hazardous
waste locations, an emphasis has been
placed on development of rapid
screening instruments designed
specifically to provide quick and
accurate ground water analysis.

Two instruments have been developed to
accomodate those needs.  The first one,
the "TOP", provides quick assessments
of total concentrations of volatile
hydrocarbons in water.  The second
instrument, the AQUASCAN, provides an
on-line analysis of individual volatile
hydrocarbons in water.

The "TOP"
The "TOP" (Total Organic Purgables) is
an instrument which monitors total
concentrations of volatile hydrocarbons
in water utilizing purge and trap
technology.  Figure 1 is a block
diagram of the "TOP".  The "TOP" is
designed around an argon ionization
detector (AID) and internal sample
purge and trap system.  A computerized
control system activates an internal
sample pump, drawing water into the
purge cell.  Argon gas is purged through
the purge cell, stripping the purgable
hydrocarbons from the water into the
purge gas stream.  The gas containing
the purged hydrocarbons is routed
through an adsorption trap, where the
hydrocarbons are collected.  Once
collected, the trap is heated, desorbing
the hydrocarbons into a blank capillary
tube and into the AID chamber for
quantification.  The results are
OET.


&
CO
Y
-
••
TRAP
mmtrn
1
PURGE
CELL


20PPB
display
PUMP
— • mil
                    Figure 1. Block diagram of TOP instrument

              automatically displayed on a LED screen
              and stored on a memory chip.  This data
              can be transferred to a computer or a
              printer for a permanent record.

              Calibration of the system is carried out
              by sampling a water sample containing
              known concentration levels of hydro-
              carbons.  Figure 2 shows a typical trace
              of a total purgable run.  The dashed line
              line is a calibration trace while the
              solid line is the analysis trace.  The
              integrated area of the analysis is
              compared to that of the calibration,
              and a concentration value in the ppb
              or ppm level is automatically calculated
              and displayed on the LCD screen.

              An AID is used because of its relative
              uniform response to a broad range of
              purgable hydrocarbons (Table 1.)  This
              includes halomethanes and haloethanes
              which are not easily detected by other
              total hydrocarbon detectors, such as
              the photoionization detector (PID).
                                          711

-------
       X
              Time (seconds)

         Figure 2. TOP concentration trace
30
onizing Energy
etected cmpds.


Relative response
among detected
cmpds.
AID
11.6 eV
aliphatics
aromatics
halomethanes
haloethanes
1-10
times
PID
10.6 eV
aliphatics
aromatics


1-1000
times
      Table 1. Comparison of AID and PID
The more uniform  response  of  the AID
assures more accurate  concentration
readings regardless  of the compounds
chosen for calibration.

The AQUASCAN

The AQUASCAN is used for  continuous
monitoring of a water  source  or stream
by purge and trap gas  chromatography.
The instrument consists of an on-line
purge and trap sampling module attached
to a gas chromatograph (GC)  (Figure 3).
The purge and trap module  contains an
internal pump to  draw  calibration,
analysis, or clean water  samples into
the purge cell.   The sample is then
purged with inert gas  with the
resultant vapor swept  into an adsorption
tube.  The trapped volatile organics
are then thermally desorbed and injected
into the GC column where  they are
separated.  Each  compound  is  then
identified and quantified.  The
resultant chromatograph  (Figure 4) is
displayed on the  computer  screen and
stored on disk for later  review.
The AQUASCAN is  automated so that a
permanent operator  is  not needed.  An
internal modem allows  for remote
operation of the  instrument.

Both of these instruments will aid in
the screening and analysis of contam-
inated water sources.   The "TOP" can
rapidly and accurately determine total
hydrocarbon values  in  water.  The total
                                                                    PRINTS)
                                                                    CIRCUIT
                                                                    BOHR)
                                                                          PIKE
                                                                          OLi
                                                                               PUHP
            Figure 3. Block diagram of Aquascan; 1-drain;

                   2-ririse; 3-cal.; 4-anal.

         analysis  time  from calibration to sample
         acquisition  to  concentration display  is
         approximately  two (2)  minutes.

         The AQUASCAN allows  for complete, on-
         line,  automated,  and accurate
         chromatographic analysis of purgable
         hydrocarbons in water.   Each component
         is accurately  identified and quantified.
         The AQUASCAN can serve  as a continuous
         monitoring  system for  traces of VOC's
         in water,  such  as waste water streams
         or water  purification  systems.



























PUK
1
2
3
4
5



WE
two.
BEKZBE
ftt
H-m.Bf
WPIH



BT
%
SO
73
183
i300
1


cut.
875 p(t
1-3 H«
i 'H»
93? ^ti
1.8 pp.
S 3ZSP2IOO; «D. 95 'C


            Figure 4. Aquascan chromatograph trace
                                            712

-------
        IMPROVED DETECTION OF VOLATILE ORGANIC COMPOUNDS IN AIR BY ON-LINE
              SAMPLE CONCENTRATION IN A MICROCHIP GAS CHROMATOGRAPH
       Aaron M. Mainga and Edward B. Overton, Institute for Environmental Studies, Louisiana
                            State University, Baton Rouge, LA 70803.
Abstract

Pre-concentration of dilute gas samples was
performed by adsorption on a 1.5" glass-lined
stainless steel Tenax trap, interfaced on-line, in a
Model 200 Portable Microchip Gas Analyzer
(Microsensor Technology Inc., Fremont,
California). Adsorption onto Tenax (2,6-diphenyl-
p-phenylene oxide polymer) was achieved by
passing 5 mL of dilute gas sample, at room
temperature, through the trap placed between the
sample loop (in a solid-state injector) and the
switch valve. The adsorbed analytes were
desorbed from the Tenax by rapid heating at
200°C followed by venting of the unconcentrated
analytes to the atmosphere before injecting a
concentrated plug of analytes into the  analytical
column. Concentration ratios of up to 30:1 were
achieved for some analytes. Sample recovery
was affected by several factors, such as, the
amount of sample purged through the  trap at
room temperature, amount of sample injected,
rate of desorption heating, final desorption and
cooling temperatures of the trap, sample volatility.
Sample recovery also varied according to
whether the trap contained Tenax alone or Tenax
with Spherocarb as the adsorbent.

Introduction

Throughout the history of modern
chromatography, there has been a consistent
trend to work with ever-decreasing amounts of
analysed materials and at increasing demands on
detection sensitivity. Only slowly has this direction
been translated into smaller chromatographic
columns and corresponding instrumentation (1).
Difficulties are frequently encountered in
attempting to directly analyze organic compounds
of interest, which are often below the part per
billion level. Despite the use of highly sensitive
instruments, detection of trace amounts of
substances in this range presents a technical
challenge, especially as regards to the use of
portable instruments (2). A good example is the
analysis of volatiles from human expired air to
seek distinctive differences between "normals"
and those afflicted by disease. More recently,
awareness has grown to the fact that minute
concetrations of chemical pollutants can have  far
reaching effects as health hazards, further
underscoring the need for reliable analytical
techniques (3).

It is often stated that one of the possible
applications of high speed gas chromatography
(using a Microchip Gas Analyzer)  is in the fields of
breath gas analysis (4), and on-site analysis of
hazardous volatile organic compounds (5). At  the
part per billion level, it has almost always been
necessary to use some off-line cumulative or
concentrating technique to obtain measurable
amounts of sought-after compounds. Ideally, it is
preferable to elliminate as much as possible, the
unwanted background compounds (usually water
and air) while  accumulating the desired
substances quantitatively. For most techniques,
the result is a compromise between these two
goals.

In off-line concentration applications with the
Microchip Gas Analyzer, sample components
trapped in a separate Tenax (a porous polymer of
2,6-diphenyl-p-phenylene oxide)  concentrator are
manually collected in a gas tight syringe and
introduced to the Microchip Gas Analyzer for
separation (6). This off-line technique is generally
time consuming, operator-intensive, and difficult to
automate. In our work, we sought to put the Tenax
GC concentration trap on-line in a Microchip Gas
                                              713

-------
Analyzer, by placing it between the silicon
injection wafer and the injection switch valve.

Experimental

GC System
This consisted of a model 200 Microchip Gas
Analyzer (Microsensor Technology Inc., Fremont,
CA ) which was  equipped with a solid-state
sample injection system; two vacuum pumps
connected in parallel; a 4 m long x 0.10 mm i.d x
0.4 |im phase thickness DB-1701 capillary column;
a miniaturized thermal conductivity detector; and
an apple computer, using M2001  software.

Zero helium was used as carrier gas at a column
head pressure of 20 psi corresponding to a
flowrate of 1 mL/min (an average velocity of about
36 cm/sec.). An  isothermal column temperature of
40°C was used in our experiments.

Sample Mixture
100 ppmv mixture of n-propane to n-octane
hydrocarbons in zero  nitrogen.

1 ppmv mixture  of acetone, benzene, toluene,
chlorobenzene,  and bromobenzene  in zero
nitrogen.
Traps

 A 1 .5" long x ^" o.d.x 0.7 mm i.d. glass-lined
 stainless steel tube containing 2.0 mg Tenax GC,
 60/80 mesh, Applied Science Laboratories
 lnc.,PA.
 1.5"
           1
    long x yg" o.d.x 0.7 mm i.d. glass-lined
stainless steel tube containing 1.5 mg of Tenax
GC and 1.0 mg of Spherocarb 80/100 mesh,
Analabs, Norwalk, CT.

Trap  Loading  and Sample Injection
Five mL of sample mixture was adsorbed onto the
adsorbent in the trap by passing it through with the
aid of two vacuum pumps using a sampling  time of
140 seconds. This was followed by rapid heating
of the trap to a temperature of 190°C for 90
seconds. With the trap at 190°C the switch valve
was turned to the carrier gas position followed by
venting of the the unconcentrated sample in the
sample loop to the atmosphere through the
sample valve and the injection valve opened for
200 msec., allowing about  0.2 jiL of the
concentrate to be injected  into the analytical
column. Delay times of 400 msec, were applied
prior to venting and sample injection. After
injection, the trap was prepared for the next
sample by purging with carrier gas, while hot for
two minutes and allowed to cool to room
temperature with carrier gas  purging continuing.

Results  and Discussion

The main purpose of this work was to investigate
the possibility of placing an on-line Tenax trap in a
Microchip Gas Analyzer. For a glass-lined
stainless steel trap with its many advantages, it
has so far proved practically  applicable (7).With a
Tenax GC or Tenax / Spherocarb trap on-line, the
two sample pumps were drawing the hydrocarbon
sample at the rate of about 2.5 mL/min.

For a trapped sample volume of 5 mL, an injection
volume of 0.2 uL (200 msec, injection time), and a
100 |xm i.d. analytical column, concentration
factors of 10 are quite common, depending on the
amount of venting done before the injection.
Concentration factors are dependent on the vent
times used.

For a trapped volume of 5  mL on a Tenax
adsorbent and an injection volume of 0.2 ^L (200
msec, injection time), the following were some of
the data obtained:

Cpd Name     Area   %RSD  Increase
Propane         Trace
Butane         411813       2          4
Pentane         783870      11           5
Hexane         3164633      8        17
Heptane         7152900      2      Large
Octane         5451000      4      Large

For a trapped volume of 5  mL (the rest same as
above):

Cpd  Name     Area   %RSD  Increase
Acetone         2205500      5          2
Benzene        13518667    4        15
Toluene         7751567      7        41
Chlorobenzene  1760633     14      Large
Bromobenzene  3075167     17          5

Note: An increase of large implies that the
compound was not detected before concentration

In general, the factors found to increase the
concentration factor include: the use of an
analytical column with an  internal diameter larger
than 100 jam i.d.; use of a higher desorption
temperature (> 200°C); venting of an appropriate
amount of dilute sample in front of the adsorbent in
                                                714

-------
amount of dilute sample in front of the adsorbent in
the sample line; use of the fastest rate of
desorption heating  and use of an appropriate
adsorbent for the group of compounds under
investigation.The last factor which is still being
optimized is very crucial more so for the lighter
hydrocarbons which have a very low breakthrough
volume in a Tenax trap as compared to
Spherocarb.

Conclusions

Though still under investigation, our preliminary
results, so far, have indicated that placing a Tenax
or a Tenax / Spherocarb trap on-line in a
Microchip Gas Analyzer can enhance the
detectability of volatile organic compounds.
However, for better reproducibility and
quantitation,  the stability of the trap heater still
needs to  be improved and also a better
combination  of adsorbents in the trap  have to
optimized for better concentration of all the
analytes  under investigation.  According to Zlatkis
et al, (2) Tenax, with a surface area of 18 m2 g-1,
has a low retention for low molecular weight
compounds, especially water. Higher molecular
weight compounds with relatively low polarity can
be trapped and thermally desorbed (at 300° C) at
high efficiency. At this temperature Tenax does not
contribute to detectable artifacts, due to its unusual
thermal stability. Like any other chromatographic
stationary phase, the Tenax or Tenax /
Spherocarb trap must be evaluated as regards
partitioning of a particular compound  between
adsorbent and carrier gas. As a consequence, our
resuts apply to the  specific amount of adsorbents
employed in the trap. In this case, the beakthrough
volumes  for compounds of interest, which are
proportional  to the  amount of adsorbent would
have to be determined for a particular trap before it
is put to use.

REFERENCES

1. Angell, J.B.; Terry, S.C.; Earth, P.W., Scientific
   American, 1983, 44,248.

2. Zlatkis, A.; Weisner, S.; Ghaoui, L; Shanfield,
   H., J. Chromatogr. Lib., 1985, 32, 449-459,
   edited  by Bruner, F.

3. Krotoszynski, B.; Gabriel, G.; O'neill, H., J.
   Chromatogr.  Sci.,1977,15, 239-244.

4. Saadat, S.; Terry, S., C.; American  Lab.,1984, 5,
   90-101.
5. Lee, G.; Ray, C.; Siemers, R., and Moore, R.,
  Am. Lab., February 1989, 110-119.

6. Backhouse, T., "An Investigation of the
  Performance of a Portable Concentration
  Device for the  Analysis of Air Samples, M.S.
  Thesis, Institute for Environmental Studies,
  Louisiana State University,  1989.

7. Kirk, B.D.; Zaffiro, A.D.; Westendorf, R.G., "Use of
  Glass-lined Tubing in Purge and Trap
  Concentrators",presented at the 39th Pittsburgh
  Conference,February,1988, New Orleans, LA.
                                                715

-------
            ON-LINE SCREENING ANALYZERS FOR TRACE ORGANICS
                UTILIZING A MEMBRANE EXTRACTION INTERFACE

                            Richard G. Melcher and Paul L. Morabito
                              THE DOW CHEMICAL COMPANY
                                  ANALYTICAL SCIENCES
                                       1602BLDG.
                                    MIDLAND Ml, 48667
A unique membrane extraction interface has been
developed which enables automatic extraction of
selected trace organic compounds from aqueous
streams and samples.  By selecting the type of
extractant flowing through the tubular silicone rubber
membrane, various classes of organic compounds
can be selectively extracted and concentrated, with
the exclusion of others. In addition to the advantage
of selectivity, the interface can be used for streams
with high dissolved solids and paniculate content.
Several types o.f on-line  monitors have  been
developed based on this interface which would be
suitable for screening and analysis of wastewater,
leachate, ground water and surface water.

The simplest membrane monitor is based on a flow
injection system where the analyte, which is extracted
from the injected sample by the membrane, flows
directly to the detector. This approach is useful in
screening for a major component or for the sum of a
class  of compounds.  For complex samples  the
membrane/detector selectivity may not be sufficient to
isolate  the single  component and  additional
separation is necessary.  In one on-line application,
low ppb levels of chlorophenols are determined by
extraction through the membrane into a basic
extractant.  This extract is then injected into a LC
system for analysis.   In another application, a
capillary gas chromatograph is used to determine low
and sub-ppb levels of organic compounds extracted
through the membrane into a hexane extractant.  An
automated large volume (25-250  n\.) injection
technique, which couples a retention gap with an air
actuated rotary valve, was developed to make  on-
column injections.
Membrane.  The membrane used in the systems
described was SILASTIC brand medical grade tubing
(Dow Corning, Midland, Michigan),  a seamless
silicone rubber tubing designed for  clinical and
laboratory applications. Silicone rubber  is chemically
and mechanically stable and has a high permeation
rate for a large variety of organic compounds. A
single membrane has been used in a continuous
wastewater analyzer for over a year with no apparent
change. Depending on the membrane size,
pressures above 10-20 psi will cause the membrane
to expand and possibly rupture. Tubing of various
sizes can be obtained from medical supply houses.
The two membrane sizes used were 0.013 inches I.D.
by 0.025 inches O.D. (Dow Corning Catalog No. 602-
105), and 0.020 inches I.D. by 0.030  inches O.D.
(Dow Corning Catalog No. 602-135).

General principles of membrane extraction. The
permeation of compounds through a nonporous
polymer membrane occurs by a "solution-diffusion"
mechanism.  The term "permeation" designates the
overall mass transport of the compound across the
membrane, whereas the term "diffusion" designates
only the movement of the penetrant molecules inside
the polymer matrix.  For the compounds we have
worked with, the diffusion coefficients are similar and
the solubility of the compound in the membrane and
the extractant appears to be the major parameter for
selection.(1,2) The three major steps in the process
are:
    1.  The compound of interest in  the sample
contacts the membrane and, depending  on  the
solubility parameters, extracts into the membrane.
                                             717

-------
    2.  A concentration gradient forms and the
compound diffuses across the membrane.
    3.  When the compound contacts the extractant
on the other side of the membrane, it partitions into
the extractant depending on the solubility parameters.

Since the silicone rubber membrane has a similar
solubility parameter as hexane, membrane extraction
can be thought of as a combination of a two-step
liquid-liquid extraction with hexane.  However,
because  the organic phase is solid, many difficult
extraction procedures are now possible:

    1.  Samples which form emulsions can be
extracted.
    2.  Very small volume organic/sample extractions
can be performed.
    3.  Solvents such as acetonitrile, acetone,
isopropanol and methanol, which would normally be
miscible  with the aqueous sample matrix, can be
used as extractants.
    4. Automated on-line extractions interfaced with
analytical instrumentation can be developed more
easily.
 Membrane/Flow Injection Analysis. This is the
 simplest of the membrane systems.  The detailed
 parameters of this system have been reported (2,3)
 and a brief description is given below.
In the Membrane/FIA system the membrane is
connected directly to an LC detector (Figure 1).  The
membrane is contained in a glass flow-through cell.
For determination of phenol, a dilute caustic solution
is pumped through the tubular silicone rubber
membrane and water or a buffered carrier solution is
pumped around the outside of the membrane.  The
sample loop is filled with sample and when the valve
is rotated, the sample is carried past the membrane.
Some of the phenol in the sample permeates the
membrane and forms a phenate salt when it reaches
the caustic. The phenate salt is no longer soluble in
the membrane and concentrates in the caustic.  The
carrier flows the extracted phenol to the  detector
where it is detected as a peak.  Selectivity  depends
on the membrane parameters and the detector
response. Since many neutral and basic compounds
are not very soluble in the caustic solution, they
prefer to remain in the membrane.  Selectivity can
also be obtained between phenols if their pKa values
differ by 2 units or more.  By making  the sample a pH
of 9, phenol (pKa=10) will extract while phenols with a
pKa of less than 8 will show very little extraction.  By
using a basic carrier and an organic extractant such
as methanol or acetonitrile, neutral compounds are
extracted with the exclusion of phenolics. Therefore,
by adjusting the pH of the sample  and the
composition and pH of the extractant, the membrane
acts like a "chemical switch" which can select various
chemical classes.
                                        Extractant
    Detector
                         Sample
 FIGURE 1.  Membrane/Flow Injection System
Membrane/Liquid Chromatoaraphv.  Although the
Membrane/FIA system will work in many situations,
complex samples require more selectivity.  The
Membrane/LC system (1) is similar in principle to the
FIA system except the extract flows into a sample
loop of an LC valve as shown in Figure 2. The
contents of the loop are then injected into the LC
system where the extracted components are
separated and detected. Higher concentration factors
can be obtained by using a stop-flow extraction
technique. If the stop-flow valve shown in FIGURE 2
is rotated, the extractant is trapped in the membrane
and analyte continues to permeate and concentrate.
When the valve is rotated to the initial position, the
concentrated extract flows to the LC sample loop.
Phenols and neutral compounds can be determined
at low and sub-ppb levels using this technique.
                                                 718

-------
                                       HPLC
                                      Column
                       Sample Loop
FIGURE 2. MEMBRANE/LC SYSTEM WITH A
STOP-FLOW VALVE.

Another Membrane/LC has been developed (4,5)
which uses only one pump and only one solution that
serves as both the extractant and the LC eluent.  An
eight-port valve is used to isolate the membrane from
the back pressure of the LC column and it is rotated
for a short period to allow the concentrated extract to
flow to a sample loop  connected to the same valve.
Although this system is not as versatile, its simplicity
makes it useful for dedicated, on-line analyzers.

Membrane/Gas Chromatoaraphy.  The
Membrane/GC systems  use the same type of
membrane, however,  the cell design is modified to
allow for the swelling of the membrane when
extractants such  as hexane are use.  Two types of
Membrane/GC systems have been developed.(6)

One design, shown in FIGURE 3., combines
membrane cell technology with a pneumatically
operated pressurized injection valve (POPSI).  A
hexane extractant  flows through the tubular
membrane and extracts permeated compounds. This
concentrate then flows to the injection valve with an
internal 1-3  
-------
FIGURE 4. MEMBRANE/GC LARGE VOLUME
ON-COLUMN  INJECTION:(a) membrane cell; (b)
extractant pump; (c) sample pump; (d) carrier gas Inlet; (e)
sample loop; (f) extractant waste; (g) oven; (h) retention gap; (i)
capillary union; (j) capillary analytical column; (k) detector.
6.   Morabito, Paul; Melcher, Richard; Hiller, Joseph
and McCabe, Terrenes, "Method for On-Column
Injection Gas Chromatography," United States Patent
4,962,042, October 9, 1990.

7.   Melcher,  Richard  and  Morabito,  Paul,
"Membrane/Gas Chromatographic  System for
Automated Extraction and Determination of Trace
Organics in Aqueous Samples," Analytical Chemistry,
Vol. 62, No. 20, October 15, 1990, 2183-2188.

8.  McCabe, Terrence; Hiller, Joseph and Morabito,
Paul, "An Automated Large Volume  On-Column
Injection Technique  for Capillary Gas
Chromatography," Journal of High  Resolution
Chromatography, Vol. 12, August 1989,517-521.

9.  Morabito, Paul; Hiller, Joseph and McCabe,
Terrence, "A Method for Automated Large Volume
On-Column Injection Technique for Capillary Gas
Chromatography with Solvent Diversion," Journal of
High Resolution Chromatography, Vol. 12, May 1989,
347-349.
References.

1.  Melcher, Richard and Bouyoucos, Spiros,
"Membrane Interface for Automatic Extraction and
Liquid Chromatographic Determination of Trace
Organics in Aqueous Streams," Process Control and
Quality, Vol. 1, Issue 1, November 1990,63-74.

2.  Melcher, Richard, "Flow-Injection Determination
of Membrane Selected Organic Compounds,"
Analytica Chimica Acta, 214,1988, 299-313.

3.  Melcher, Richard, "Membrane Assisted Flow
Injection Analysis," United States Patent 4,819,478,
April 11,  1989.

4.  Melcher, Richard and Cortes, Hernan, "Method
for Membrane Assisted Chromatography," United
States Patent 4,775,476, October 4,1988.

5.  Melcher, Richard  and Cortes,  Hernan,
"Apparatus for Membrane Assisted Chromatography,"
United States Patent, granted May, 1990.
                                               720

-------
            Candidate  Protocols  for  Sampling and Analysis of Chemicals
                        from the Clean A1r Act List
R. G.  Merrill,  Jr.,  J.  T.  Bursey,  D.  L.
Jones,  T.  K.  Moody,  and  C.  R.  Blackley,
Radian   Corporation,  Box  13000,  Research
Triangle Park,  North  Carolina  27709

W. B.  Kuykendal,  Office of Air Quality
Planning and Standards, U. S.  Environmental
Protection Agency,  Research  Triangle  Park,
North Carolina   27711
      Clean Air  Act  (CAA)  amendments  of
1990 renew and intensify  national  efforts
to reduce air pollution.   Title III of the
Amendments   lists   189    hazardous   air
pollutants   (HAPs)    and    requires   the
Environmental  Protection  Agency (EPA)  to
promulgate new control  standards  for  the
principal sources of such emissions.   The
189  HAPs  are  chemicals   not  previously
regulated under  the National  Ambient  Air
Quality Standards.   These HAPs  listed  in
the CAA are not expected  to be  found  in a
large  number  of  areas   nor   in  large
quantities.   However,  health effects  may
occur  at  low  concentrations  because  of
their  high  acute  or  chronic  toxicity.
Measurement  of   pre-  and   post-control
emissions from a wide variety of stationary
sources  will   be  required  in  order  to
determine  success  in reducing  emissions.
Prior to  testing  at a source,  however,  a
written  sampling  and analytical  protocol
must  be   available  to  ensure  that  data
acquired during source testing are accurate
and of known quality.  In order to optimize
the yield  of information  from  any  given
field effort, the sampling  and analytical
methodologies which  are  selected should be
applicable to the broadest  possible range
of compounds.  Some  sacrifice of accuracy
and precision  of the methodology may  be
necessary   to   extend    the    range   of
applicability.     This  study  reports  the
results  of  a  review  and  evaluation  of
existing    information   on  sampling   and
analytical    methods     presently    and
potentially   applicable   to   toxic   air
pollutants.      Generic   Methods   which
simultaneously  yield  information  for  a
large number  of HAP  compounds have  been
emphasized.   The  goal   of using  methods
which cover a broad range of compounds may
initially  require  the  sacrifice  of  some
method accuracy.   Many  analytes have been
assigned to generic methods  on the basis of
physical   properties   since   individual
compound  validation   data   are  often  not
available.  Such validation data must come
from  well  designed   programs  using  data
obtained  by  techniques such  as  dynamic
spiking  of isotopically-labeled  compound
analogs while sampling  operating  sources.
Extension  of  an  existing   methodology  is
completely    valid    only    when    the
applicability  of  both   the sampling  and
analytical  methodology  to  the  specific
analyte has been established.  If  complete
field  method   validation   has  not  been
performed, it is possible that the compound
can  be   analyzed  using  the   analytical
methodology    but    not   collected
quantitatively,  or vice versa.

      Some of the methods cited are written
specifically  to  address gaseous  emissions
from  stationary  sources.    Sampling  and
analytical  methods  may be divided  into
combustion and noncombustion stack methods,
since  CAA  requirements will  cover  both
types  of   stationary   sources.     For  a
stationary   source  not   related   to   a
combustion process, sampling and analytical
methodology used for  ambient air monitoring
may  be  applicable.     For  some   of  the
compounds listed as HAPs, there is a choice
of methodology.   Generic methods have been
emphasized.        Specialized
sampling/analytical     methodology    for
Individual   compounds   or   classes   of
compounds has also been summarized when it
was available.

      The  following  methods were  selected
as having  the broadest  possible  range of
application:
                                               721

-------
Volatile  Organic  Sampling  Train  (VOST)
Methodology1
      The VOST methodology is described in
OSW-SW846  Method  0030  for  sampling  and
Methods   5040  and   5041   for  analysis.
Samples  in the  field  are  taken  using  a
specialized   sampling  train  to  collect
volatile  organic  compounds  on  a sorbent.
The   sorbent   sampling  tubes   are  then
returned  to  the   laboratory for  thermal
desorption  through water,  collection  and
concentration of  the  vapors, and ultimate
analysis    by   gas   chromatography/mass
spectrometry.   The  sampling  methodology
specifies   that   organic  compounds  with
boiling points between 30*  and  100'C may be
sampled  with  the  methodology.  Compounds
with  boiling  points   below  30°C  may  be
sampled  using special  care,  and selected
compounds with  boiling points  above 100'C
may also  be sampled in carefully selected
situations.  Polar, water-soluble compounds
represent   specific   problems   with  this
methodology.   The literature  presents  a
method  for  optimizing  recovery of  such
compounds ,  but  even  with  the  optimized
methodology, many  problems remain.

Semi volatile	Organic   Samplino   Train
(SemiVOSTl Methodology1
      The SemiVOST methodology is describe
in OSW-SW846  Method 0010 for sampling  and
Method 8270 for analysis.   Samples in  the
field  are   taken  using   a  specialized
sampling  train  with sorbent to  collect  a
range  of semi volatile  organic  compounds.
The train  is  recovered  and  components  of
the train  are returned  to  the  laboratory
for   extraction,   concentration   of   the
extract,     and    analysis    using    gas
chromatography/mass   spectrometry.     The
sampling methodology specifies only that it
is  applicable to  compounds  with  boiling
points above  100'C.    Although no  upper
limit on  boiling point Is posed, the method
is  limited  to  compounds  which  can  be
solvent extracted  and analyzed  by GC/MS.
Semivolatile  organic   HAPs   in  the   CAA
Amendments  constitute an  extensive  list,
including single entries that are groups of
compounds.  Some  members of these groups
will    be   observed   with    the   Semi VOST
analytical methodology with poor detection
limits,  but  if   higher  specificity   or
accuracy   is  required   for  polychlorinated
biphenyls or  polycyclic  organic material,
there are specialized  applications  of  the
SemiVOST   methodology   available.    Also,
since the range of compounds on the CAA is
so wide, the sample preparation methodology
will  not   be   optimum  for  all   of  the
compounds simultaneously.  The broad range
of  the  method  must  yield  to  specific
optimization  of  the  methodology  for  a
particular  class  of  compounds,  such  as
derivatization  for  carboxylic  acids,  or
adjustment  of  pH  during  extraction  to
optimize recoveries  of particular classes
of compounds.

Hultl-Hetal Sampling Train1
      The    multi-metal     sampling    and
analytical  protocol  is described  in  OSW-
SW846   Method   0012   for   sampling   and
analysis.     The   methodology   for   the
determination    of    multiple    metals
incorporates a  stack  sampling  train using
specialized   aqueous    solutions,    with
ultimate analysis according to a series of
digestion  and  analytical  methods  which
include  final   quantitation  with  either
atomic    absorption    spectroscopy    or
inductively    coupled    argon    plasma
spectroscopy.    The  source  samples  are
withdrawn  isokinetically  from  the  stack
through  a  heated probe,  with  particulate
emissions collected on a  filter  in a heated
filter holder located  outside the stack and
after the  probe of  the    sampling  train.
The  analytical   methodology  detects  and
quantifies  metal   ions,   so  an  inorganic
compound  is not detected as  a  molecular
species.        For    example,    titanium
tetrachloride   would   be   detected   as
titanium;   no  identity  of the  molecule
would be retained  in  the acidic digestion
process by which the sample  is prepared for
analysis.

Aldehyde/Ketone Sampling Methodology
      Sampling   and  analytical  methodology
for a variety of aldehydes  and  ketones is
described in OSW-846 draft Method 0011 and
Method  8315.    Gaseous  and  particulate
pollutants  are  withdrawn  isokinetically
from an  emission source  and are collected
in    aqueous    acidic    2,4-
dinitrophenylhydrazine   (DNPH)   solution.
Aldehydes   and  ketones   present   in   the
emissions  react  with   DNPH  to   form  a
dinitrophenylhydrazone derivative which is
extracted,  concentrated, solvent-exchanged,
and  then  analyzed  by  high  performance
liquid chromatography.
                                                722

-------
      The sampling and  analytical  methods
described above  will  provide  a means  of
sampling and analyzing approximately 80-90
percent of  the entries  on  the  CAA List.
However, for many of the compounds listed,
a field  test under controlled  conditions
has not been performed  to demonstrate and
evaluate   the   combined   sampling   and
analytical  methodology.     Demonstration
tests  need  to be  performed to  determine
recoveries of the compounds from stationary
source emissions exhibiting  various matrix
conditions,    which    will    cause   the
performance  of   the   methods   to  vary.
However,  a  relatively  broad coverage  of
compounds  can  be  achieved  with  these
generic  methods,   saving the  expense  of
applying individual methods but potentially
losing   some  of   the  specificity   and
sensitivity of individual methods.

     Some of the entries on  the  Clean Air
Act   List   will   require   specialized
methodology  because   of   problems  with
reactivity  or other  difficulties.    For
these    compounds,    some     potentially
applicable methodologies are:
      acetonitrile by  Method 18 (GC/NPD),
      bis(chloromethyl) ether by Method 18,
      1,3-butadiene by Method 18,
      carbaryl, by Method 0010  and Method
      632,
      carbonyl  sulfide, by Method 15,
      chloramben,  by Method 0010 and Method
      515/615,
  •   2,4-D salts and  esters,  Method 0010
      and Method 515/615,
  t   dimethyl   carbamoyl    chloride,   by
      Method 0010 and  Method 531,
  •   4,6-dinltro-o-cresol    and    salts,
      Method  0010  and   Methods  8270  and
      515/615,
  •   ethylene oxide,  CARB Method 431,
  •   hexamethylphosphoramide,  Method 0010
      and Method 632,
  •   hydrazine by Method 18,
  •   methanol  by Method 18,
  •   propoxur by  Method 0010 and Method
      632,
  •   2,3,7,8-tetrachlorodibenzodioxin  by
      Method 23,
  •   asbestos by CARB Method 427,
  •   chlorine by  modified  Method  26 and
      OSHA Method ID-101,
  •   coke oven emissions, Method 0010 and
      Method 8310,
  •   cyanide compounds,  by modified Method
      6 and  NIOSH Method 7904,
  t   hydrochloric add by Method 26,
  •   hydrogen fluoride  by Method  13A or
      13B,
  t   mineral fibers by CARB Method 427,
  •   polycyclic  organic  matter by  CARB
      Method 429 or Method 5 G,
  •   radionuclldes, by Method 114.

      Many of these entries on  the CAA list
represent  a very  broad  category,  and no
single  analytical   methodology   will   be
equally effective for all possible members
of the  category.   Some  of the categories
are undefined or  poorly  defined.   In such
cases, the analytical methodology specified
will  serve for some representative members
of the category.

     No applicable sampling and analytical
methods could  be found  for  the  following
Clean Air Act List entries:
  t   diazomethane,
  •   phosgene,
  •   calcium cyanamide.

     Many  of the entries  on the CAA List
are   extremely   reactive.     Because   of
physical properties such as boiling point,
these entries may be initially assigned to
a specific  methodology.   However, testing
will   be  required to demonstrate  that  the
compound   can   exist   without   serious
decomposition under the conditions of heat,
high  water  content,  and  possibly  acid
content which may exist in a stack.  Also,
a given compound may react completely when
it is present in  emissions at trace levels,
but significant  amounts  may  survive to be
sampled and analyzed  if the  compound is
present at  levels  of parts per million in
the emissions.

     Most    of    the   numbered    Methods
Incorporate  specific guidance  for quality
control (QC) and quality assurance (QA) to
ensure  that data  obtained  are  of  known
quality.    Those  who  wish  to apply  the
methodology must establish their capability
and    continuously   train    staff    and
demonstrate  the quality  of their results.
Most of the guidance  in the areas of method
performance  relates  to  requirements  that
the users of the methods:

  •   perform an  initial  demonstration of
      capability   with   the   method   and
      conduct  ongoing   demonstrations  of
      capability,
                                              723

-------
  •   maintain  accurate  records,  follow
      Chain of Custody procedures,
  •   demonstrate  control  of  Instrument
      parameters,
  •   demonstrate  that  equipment  1s  not
      contaminated prior to use,
  •   perform appropriate QC daily for all
      Instrumentation,
  •   establish  the  ability  to  generate
      acceptable accuracy and precision,
  •   locate/correct   any   problems   in
      instrument operation,
  •   design  and  execute  an  appropriate
      scheme  of  blanks  of  various types,
      duplicates, matrix spikes, and matrix
      spike duplicates,
  •   determine the accuracy and  precision
      of the methodology,
  •   qualify all  data appropriately when
      QC criteria not met and,
  •   participate in performance evaluation
      studies, as available.

      If  the stationary  source   Is  not a
combustion source, useful data can usually
be obtained  from  stack methodologies, but
methodologies  originally  developed  for
ambient  monitoring  may  also be useful,
with  appropriate allowance  for  the fact
that emissions from a  stationary source may
contain  significantly different  matrices
and higher than ambient levels.

      Selection of sampling and analytical
methods is governed by many considerations.
Regulatory    requirements    dictate   the
selection in many instances.  The  detection
limits   which   are   required    for   the
analytical methodology dictate a  selection
in many  instances.   Cost  is  frequently a
major   factor    in   determining   which
methodology will  be used.   Depending upon
cost considerations, for example,  it may be
feasible to add an air toxics component to
a  source  test  program  with a   different
primary mission.    Selection  of  the best
applicable methodology from a wide variety
of potentially applicable methods  is  a very
difficult choice.
reflect  the  views  of the  Agency and  no
official endorsement should be inferred.
                References

1.  United States Environmental Protection
Agency, Office of Solid Waste and Emergency
Response.    Test Methods  for  Evaluating
Solid Waste.  Third Edition.  Report No. SW-
846.  Washington, D.C.:   1986.
2.    M.  H.  Owens,   S.
Lachajczyk,   Development
Analysis   Protocol    for
 A.
of
 Mooney,   T.
VOST  Sample
   Water-Soluble
Volatile POHCs and PICs. EPA-600/8-87-008,
U.  S.  Environmental  Protection  Agency,
Research Triangle Park:  1987.

3.  W. T. Winberry, Jr., N. T. Murphy, and
R. M.  Riggin,  Compendium of Methods for the
Determination of Toxic  Organic Compounds in
Ambient  Air.   EPA-600/4-84-041,  U.   S.
Environmental Protection  Agency,  Research
Triangle Park: 1988.
                Disclaimer

      Although the information described in
this article has  been  funded wholly or in
part by the Environmental Protection Agency
under contract number 68-02-4286 to Radian
Corporation,    it   does  not  necessarily
                                               724

-------
                       THE INVESTIGATION OF SOIL SAMPLING DEVICES AND SHIPPING AND
                        HOLDING TIME EFFECTS ON SOIL VOLATILE ORGANIC COMPOUNDS
J. R. Parolinl, V. G. King, T. W. Nail, and T. E. Lewis
Lockheed Engineering & Sciences Company
Las Vegas, Nevada
INTRODUCTION

Volatile organic compounds (VOCs)  are the  most often
encountered class of compounds at Superfund and other
hazardous  waste sites.   Many VOCs are  considered
hazardous because  they are mutagenic, carcinogenic, or
teratogenic and commonly are the controlling contaminants
in site remediation projects. Because decisions regarding
the extent of contamination and the degree of cleanup have
far-reaching effects,  it is essential that these decisions be
based   on   accurate   measurements   of  the  VOC
concentrations present.  VOCs, however, present sampling,
sample handling, and analytical difficulties, especially when
encountered in soils and  other solid matrices.   Sample
collection and handling  activities can often introduce large
sources of random and  systematic errors compared to the
analysis itself. Negative  bias (i.e., measured value less than
true value) is perhaps the most significant and most difficult
error to delineate and control. This error is primarily caused
by volatilization losses during soil sample collection, storage,
and handling.  Currently, no standardized procedures exist
for sampling  soils for  VOC analysis.  Several different
samplers are available for  collecting intact and disturbed
samples.  Samples are usually removed from the sampler,
which often  disturbs intact samples.   Samples  are then
placed in glass jars or  vials and sealed with Teflon-lined
caps. Practical experience and recent field and laboratory
research, however, suggests that procedures such as these
may lead to significant loss of VOCs (1,2).

EXPERIMENTAL FEATURES

Experiments were conducted to evaluate  sampling and
sample handling techniques for the collection of soil for
volatile organic analyses  (VOA).   Because natural soil
systems can  be extremely heterogeneous,  experiments
were performed by using large (18 in. i.d.) reconstituted soil
columns. The soil was contaminated by the upward diffusion
of VOCs from a glass-bead  layer beneath the soil.  This
approach produced very homogeneous material for the
evaluation of sampling devices and various sample handling
scenarios.  Figure 1  shows the  horizontal  and vertical
homogeneity in bulk density and moisture content obtained by
this column packing procedure.

Four different sampling devices (treatments) were evaluated:
(1) acetate liner (4 cm i.d.) the contents of which were emptied
out and disturbed, (2) split-spoon sampler with a brass liner
(4 cm i.d.), (3) acetate liner (4 cm i.d.), and (4) acetate liner
(2.5 cm i.d.). Samples from each device were placed in either
a 40-mL VOA vial or a 125-mL wide-mouth jar.

Treatment 1 (disturbed vial sample) exhibited the largest VOC
concentrations (Figure 2). The disturbance resulted in a
homogenized material that had a higher concentration than the
original sample because the shallow, low-VOC-level soil was
combined with the deep, high-VOC-level  soil.  A vertical
concentration gradient in the soil column was the cause of the
elevated VOC levels in the disturbed sample. The vial-held
disturbed sample yielded greater VOC concentrations than the
jar-held disturbed samples, which indicates that VOC losses
occurred during the homogenization and separation into
aliquots as specified in EPA Method 8240.  Of the undisturbed
samples, the jar-held samples collected with larger diameter
samplers  (brass or acetate  liner) exhibited  higher VOC
concentrations than the jar-held samples collected with the
small-diameter, acetate-lined sampler. Samples collected from
the large and small diameter intact cores, using a subcorer,
yielded essentially the same VOC levels. Collection of a small
sample from an intact core with a subcorer and extrusion into
a 40-mL VOA vial maintained the integrity of VOCs betterthan
jar-held samples. The vertical concentration gradient, however,
                                                       725

-------
 made direct comparisons difficult. Differences between jar-
 held and vial-held samples were probably caused by sample
 pretreatment rather than by leakage  of VOCs from the
 containers.
     1.20
   o
   o<
     1.15
     1.10
     1.05
    1.00
     fet
              0-4   4-11  11-19 19-27
                     depth (cm)
90*
                         180'  I   I  270*
                                             center
    10.0

     9.8

     9.6
     9.2
   o
   0  9.0
   i)
   a  8.8

  '5  8.6

   E8.4

     8.2

     8.0
             0-4   4-11  11-19 19-27
                     depth (cm)

 Figure 1.  Horizontal and bulk density  and moisture
 content obtained by column packing procedure.
Air shipment of soil samples held in commercially available
sample containers was investigated.  Samples shipped by
different air carriers underwent changes In pressure and
temperature.   The results of pressure and temperature
monitoring on three commercial air carriers are presented in
Figure 3.   This  is obviously  a  small sampling of the
environmental conditions that  occur in aircraft cargo holds.
These conditions will vary with the  type of aircraft, the
altitude at  which the aircraft flies, and the time of year. The
shipment container the monitoring devices were housed in
was   not  insulated,   so  the observed  pressures  and
temperatures are the actual ambient conditions inside the
                                              cargo hold. In-flight intervals are indicated by negative spikes
                                              in pressure. A pressure differential of as much as 2 psi was
                                              exerted upon sample containers. The integrity of VOC soil
                                              samples may be jeopardized when subjected to decreased
                                              pressures in the cargo holds of aircraft.
i; 0.08
V 0.06
C 0.04

8 0.02
                                                                           vial
                                                                           a
                                                                           jor«
                                                           1234
                                                                      treatment  number
                                               Figure 2.  VOC concentrations in vial-held  and jar-held
                                               samples collected with four different sampling devices.
                                              The objective of the shipping effects study was to evaluate the
                                              stability of VOC in soil samples shipped and held in various
                                              commercially available containers.  Five soil columns were
                                              reconstituted and samples from each column were taken in the
                                              following manner:

                                              TjreatmentJM. - A1 -5 g aliquot was extruded into a tared 40-mL
                                              I-CHEM amber-glass VOA vial and sealed with a Teflon-lined
                                              septum cap.

                                              Treatment #2 - A1 -5 g aliquot was extruded Into a taiwi 40^1
                                              I-CHEM amber-glass VOA vial and sealed with a modified
                                              purge-and-trap cap (Associated Design & Manufacturing Co.,
                                              Alexandria, VA- ADMC). Prior to analysis the sample with an
                                              ADMC cap was attached directly to the purge-and-trap unit by
                                              pushing the sparger tube into the cap thus dislodging the
                                              Teflon boili ng ball lodged in the bottom of the cap into the vial.
                                                          726

-------
Thus the sample was exposed to the atmosphere for only
fractions of a second.
                                                           Treatment #4 - A1 -5 g soil was extruded into a tared 40-mL
                                                           QORPAK amber-glass VOA vial and sealed with an ADMC cap.
Treatment #3 - A1 -5 g soil was extruded into a tared 40-mL
QORPAK amber-glass VOA vial  and sealed with a Teflon-
lined septum cap.
o
0

(N
15


14


13

12


11


10


15
                                    Airborne
                                       Onlorls
            pressure
            temperature
           i    i   i    i
                                            100
                                            90
                                            80
                                            70
                                            60
                                            50
                                            40
                                            30
                                            20
                                            10
                                            0
       -50  5   10  15  20  25  30  35 40  45
  n
  a
     12

     ,,
     '  '

     10
         Federal
         Express
         lot Vigot  .!
                                      Ut Vigo
         •  pressure
         •  temperature
          i    i    i
60

50

40
                                             30 3
                                            10
       -50   5   10  15  20  25  30  35  40
     15

     14

     13

     12

     11

     10
                                      UPS-
                                     l«l VlfOI
        •  pressure
        •  temperoture
         i    i    i   i
100
90
80
70
60
50
40
30
20
10
       -50   5   10  15  20  25  30  35  40  45
                   elapsed  time  (hr)

Figure 3.    Preliminary   pressure  and  temperature
measurements.
                Treatment #5 - the entire contents of the middle liner section
                was extruded directly into an Eagle-Pitcher 125-mL wide-mouth
                jar (Eagle Pitcher Industries, Inc.) and sealed with a solid
                phenolic cap lined with Teflon.  (Prior to GC analysis the
                contents of the jar samples were prepared as per EPA Method
                8240 specifications.)

                Treatment #6 - the entire contents of the middle liner section
                was extruded directly into an Eagle-Pitcher 125-mL wide-mouth
                jar (Eagle Pitcher Industries, Inc.) and sealed with a solid
                phenolic cap lined with Teflon.

                For the shipping effects study one set of duplicates was placed
                in an ice chest and held in the laboratory.  Another set of
                treatment duplicates was placed in an ice chest with several
                Freeze Gel packs and shipped on Federal Express. When
                samples returned to Las Vegas, NV, after two days, both held
                and shipped samples were removed from the ice chests and
                placed in the freezer until analysis.

                All the containers evaluated adequately withstood the negative
                pressure differentials exerted by air shipment (Figure 4). Jars
                and vials may be equally suitable for shipping samples via air
                carrier.  The greatest VOC loss occurred when soil samples
                were transferred from the sampling device to the container and
                when the samples were prepared in the laboratory for purge-
                and-trap analysis.
                                                        SUMMARY

                                                        The optimum soil sampling procedure reduces VOC losses by
                                                        minimizing sample disturbance during collection and transfer
                                                        to a container.  The optimum scenario for maintaining the
                                                        integrity of VOCs in a sample was found to be collection of an
                                                        undisturbed sample with a tube-type sampler (split-spoon or
                                                        zero contamination sampler) that has a precut liner. The soil
                                                        In the middle liner section was used for sample collection
                                                        because it represented the least disturbed material.  A 2-g
                                                        aliquot was taken from the center of the exposed soil surface
                                                        in the liner by using a subcorer. The contents of the subcorer
                                                        were extruded directly into a tared 40-mL VOA vial and the vial
                                                        was sealed with a modified purge-and-trap cap. The vial was
                                                        connected to the purge-and-trap unit without  exposing the
                                                        sample to the atmosphere.
                                                           ACKNOWLEDGMENTS

                                                           The authors gratefully acknowledge the University of Nevada
                                                           at Las Vegas Environmental Research Center for the use of
                                                           their facilities. We are grateful to the following employees of
                                                           Lockheed: Neil Amick for assistance in sample analysis; Phil
                                                        727

-------
c
o
o
£
c
V
u
c
o
u






1234
.1U
0.08

0.08
0.04

0.02
o nn
UIBK
-

,"?~"—
" — ••—


1234
20
IS
12
8
4
a
!>CE 1
•
'
' il , f\ '
itiiHiv
                       .loh
                       en
                       Jon
            234
                                                           NOTICE

                                                           Although the research described herein has been funded
                                                           wholly or in pan by the U.S. Environmental Protection Agency
                                                           under Contract No. 68-03-3249 to Lockheed Engineering &
                                                           Sciences Company,  it has not been subjected to Agency
                                                           review and therefore does not necessarily reflect the views of
                                                           the Agency, and no official endorsement should be inferred.
                                                           Mention of trade names or commercial products does not
                                                           constitute endorsement or recommendation for use.
                   treatment number
Figure 4. Shipping effects in containerized samples.
Malley, Mark  Sweeney, and Heath  Havey for  sampling
support;  and Dick Hannah for supplying the  pressure
transducer used in these experiments. We thank to Roger
Shura of EPA EMSL-LV for instrumental support during the
shipping experiments, and we thank Bill Ahlert of Lawler,
Matusky,  and Skelly Engineers for receiving and returning
air-shipped samples.
REFERENCES

1.    U.S. EPA, Investigation of Techniques for the Analysis
     of Volatile Organic Compounds in Soil, EPA/XXX, EPA
     Environmental Monitoring  Systems  Laboratory,  Las
     Vegas, NV, 1991, 320 pp On preparation).

2.    Siegrist,  R, L  and  P.  D. Jennsen. 'Evaluation of
     sampling method effects on volatile organic compound
     concentrations in contaminated soils," Env. Sci. Tech.
     24:1387-1392,1990.
                                                         728

-------
       DEVELOPMENTAL LOGIC FOR ROBOTIC SAMPLING OPERATIONS
       Michael D. Pavelek
       Micren Associates, Inc.
       863 Tallyho Drive
       Hershey, Pennsylvania 17033
       (717) 533-8281
                Chris C. Fromme
                RedZone Robotics, Inc.
                2425 Liberty Ave.
                Pittsburgh, Pennsylvania 15222
                (412) 765-3064
In the past decade great strides have been made in the
application of remotely controlled mobile robots. Uses for
one type of mobile robot known as a sampling robot has
been successfully demonstrated in the nuclear and space
industries.  Two examples of successful applications of
sampling robots are the Remote Core Sampler used at
Three Mile Island to determine the depth and severity of
radioactive contamination in containment walls and the
Viking explorer which took and analyzed  soil samples
from the surface of Mars. Both of these applications had
very similar developmental driving forces: the cost of
performing the operations with humans was prohibitive
and the environments were too hazardous to even consider
it. In the case of the TMI robot, existing  technologies
were  utilized in a cost effective manner to perform
required operations while for Viking, new  technologies
were developed especially for the mission. In both cases,
however,  the  beneficial returns outweighed  initial
development costs.

The most important factors to consider when undertaking
the development of  robotic equipment for sampling
operations for field screening of hazardous wastes and
toxic chemicals are cost, options,  and the scope of the
applications.
The  primary factors that  have an impact on the
development of robotic sampling operations include:

  • Does a robotic system yield an obvious advantage?

  • What are the specific tasks the robotic sampling
    system must perform?

  • Is it possible to perform these tasks  with one
    transporter equipped for bolt-on tooling?

  • Where will controls be located?  Is portability an
    issue?

  • What other tasks could the robotic sampling system
    perform, or support through reconfiguration,  which
    would yield positive benefits?
  • What unique or  unevaluated conditions will  be
    encountered  which  may require  engineering
    evaluations,  materials or component testing,  or
    safety systems?   Who will put  together  the
    expertise to properly address these issues?

  • How will the robotic system be integrated into
    operations?  Who will  provide overall project
    management, design and fabrication and training of
    operators?

  • What  precautions must be taken  to  control
    contamination encountered by the robotic device?
The answers to these questions will begin to put the major
scope of operation into perspective. However, there are
many other  items  that will require attention both
throughout  the  evolution of  the operation  and
afterwards.

Cost Considerations
Cost must be evaluated thoroughly. The amount of money
saved through the elimination of protective clothing, the
improvement  of worker safety, the reduction of manual
staffing requirements, and the reduction of waste
generated can rapidly reduce the initial estimates of the
cost of a robotic operation. Conversely, application  of
new technology may be hindered by the unanticipated
behavior of materials in untested environments, the
failure to adequately  research known  limitations  of
existing  technology,  and  uncalibrated  engineering
solutions to assure satisfactory performance. Any one of
these factors can drive the cost beyond acceptable limits.
In some cases, however, cost is not a factor.  For example,
when a particular operation is required and there are no
existing methods to perform that operation, cost may
cease to be a primary factor in the decision making.

Technological Options
One of the first decision  points in the development
process  is  whether to use existing  technology  or
                                                 729

-------
techniques  or to develop new  technology for the
application. This decision must be made in light of the
capabilities  of the  available  technology, whether
existing equipment or techniques could be modified to
accomplish the task, the projected capability and cost of
new technology, and the risk of developing the  new
technology.
The primary objective of most operations is to accomplish
a task reliably, safely, on schedule, and within budget.
The confidence in the capabilities of the equipment and
techniques to be used  typically drives the  time and
budget projections.   The sensitivities of budget and
schedule make the technology  development decision
particularly difficult.   In  some cases, a competitive
advantage  in the marketplace can be established with
the development of a new technology, offsetting the
initial  development costs.  However, when  the use of
existing techniques is  possible and feasible with respect
to budget and schedule projections, it is difficult and risky
to choose to develop a new technology and place the
operational goals at  risk.   The development of  new
technology becomes an attractive alternative  when the
costs for existing techniques are exorbitant, the available
technology is insufficient for the task, or a competitive
advantage can be established in the marketplace.

Development Strategies
There are two strategies available when introducing new
technologies:  the ideal  strategy  and  the  pragmatic
strategy. The ideal strategy presents the opportunity to
develop the application of a new technology in parallel
with the application of a proven  technology.  In this
case, the application of the new technology is gradually
phased into operations  without  creating disruption or
putting objectives at risk.  With time, the new technology
permits the operation to become more effective and
increases   the  competitive  position  of developers,
rewarding  them for their patience and vision through
economic savings and gain.
Operational reality is more pragmatic.  New technology
should be seriously considered when necessary operations
cannot be completed  with available technology, when
the necessary operational risk poses safety concerns that
cannot be overcome, or when regulatory agencies prohibit
the use of  personnel  or current equipment.  Thus the
developers of successful technologies  are rewarded for
their risk and vision through completion of operational
needs often preventing economic penalties and potential
losses.   This  defines the  pragmatic strategy for
introducing a new technology.  Note that the results
achieved by the two strategies are similar.
Development Approach
Idealistic and pragmatic development strategies require
different approaches.  In  the idealistic scenario, the
development should attempt to  provide the greatest
possible gain with the  least risk to the operation.  For
example, a manufacturing operation  that has several
similar plants located in key geographic areas should
attempt to identify problem  areas in all plants, then
develop and prove robotic technology in one plant. After
the technology has been proven it may be applied in
other plants with much less risk and a greater potential
for gain.  Each of the plants could be selected to develop
one  application of new  technology thoroughly, then
introduce it to the other plants and staff, achieving rapid
improvements in productivity.
Pragmatic development approaches generally do not
require  extensive evaluation to  determine the most
immediate area of need. For example, at Chernobyl, an
operation was required  to move unshielded nuclear fuel
from the roof of one facility  into the void left by the
disaster at an adjacent facility. Similarly, at Three Mile
Island, there was a need to remove the damaged core and
radioactive sediments from within the reactor facility.
At both of these accident sites, the operation staff knew
exactly what  needed to be accomplished, the question
was how it could be accomplished.
A potential disadvantage of  the  pragmatic approach,
and of the development of new technology /equipment in
general, is the lack of clearly defined, ongoing objectives
for the new equipment.  All too often, new technology or
equipment is developed  for  a very specific, one-time
operation without planning for other applications.
Developers should always  look beyond  the immediate
operational needs and attempt to tailor the development
process so that the new technology or equipment can be
easily reconfigured to meet future operational needs.
An example of this development foresight is provided by
Niagara Mohawk Power Corporation. This New York
electric utility recently developed a robotic device to
disassemble a conveyer system and to clean an area where
equipment malfunction resulted in stored radioactive
materials which could not be easily retrieved by plant
personnel.  Although it was possible but undesirable to
utilize  personnel to perform this task,  the developer
decided to develop a specialized  robotic system to
accomplish the task. They required that the design of
the device be reconfigurable to accomplish both the
specific task at hand and additional future operations.
The result of this endeavor is that the robotic device has
successfully completed  the specific work required,  and
will  later  perform several  other planned tasks.  This
example illustrates how a pragmatic situation was used
                                                     730

-------
as an opportunity to develop robotic equipment and
technology which achieved the immediate objective and
will  provide ongoing benefit to the utility through
increased safety of their personnel, reduced requirement
for protective devices, and improved operational
effectiveness. W

Lessons from Experience
The  design  of  robotic  devices  should incorporate
functional requirements for  all  conditions that can
reasonably be expected to be encountered during the
deployment  of the device.  Any additional capabilities
should be evaluated in light of their cost, the level to
which  they  would  enhance system performance, the
extent  to  which the added  capability would  avert
catastrophic  system failure, and the potential cost of a
catastrophic failure.
The  following  is  an example  of  when   enhanced
performance was justified. In the design of the Remote
Core Sampler,  a break-off actuator was specified to
assure safe   return  of  the device and  teleoperated
transporter in the event the core did not break and the
drill became embedded in the concrete. During sampling
operations the drill was embedded in the wall at full
depth, the core could not be broken, and the drill could not
be removed.   The  break-off actuator  allowed the
retrieval of  the robotic sampling  device and the
transporter without difficulty, effectively averting the
catastrophic loss of the robot.  This was a cost effective
enhancement.
On another  mobile  robot  developed  for TMI,  a 10
horsepower electric hydraulic pump powered the onboard
systems. A redundant pump was specified to assure total
operability in the event of pump failure. Only one fluid
reservoir could be fitted into  the space available.  The
redundant pump would also have been disabled in the
event of a  fluid loss from the primary system. A  much
smaller pump with a separate  reservoir dedicated to the
operation of the primary propulsion units at  a reduced
speed would have been a better choice.
The lesson from these examples is that redundant, or
backup, systems may be very effective and actually
salvage some operations and equipment. It is important
to have operations personnel and designers work closely
to achieve realistic  functional requirements for the
robotic equipment in light of the environment in which it
will be deployed.
Another important lesson from our experience identifies
the immense value of the use of transporters with bolt-on
tooling instead of multiple dedicated tooling systems.
The Remote Reconnaissance Vehicle (RRV), used during
the clean-up of TMI was  specifically designed and
constructed to permit attachment and manipulation of
teleoperated and robotic payloads to perform the entire
scope of operations required for completion of the project.
In addition, this equipment has general application to
other future needs. O)
Bolt-on tooling used with the RRV included:

  •Radiation Survey Equipment

  •Core Sampling Equipment
  •Kraft Undersea Manipulator
  •Sludge Sampling System

  •Sludge Vacuuming and Pumping System

  •High Pressure (Water) Flushing Equipment

  •Ultra High Pressure (Water) Concrete Scarifier

  •Abrasive Cut Off Wheels

  •Rotary Impact Drills
It would have been impossible to equip the RRV with
dedicated systems for each of these tooling functions.
Given  the  performance  record  of  the system,  the
transporter concept with bolt-on tooling was a valuable
asset. We assert that the use of bolt-on tooling is also
appropriate for field screening robots — making them
capable, flexible, and reliable.
With a highly reliable transporter, the tooling does not
have to be infallible. In the event of tooling malfunction,
the transporter returns the equipment to a controlled area
for decontamination, repair, or replacement. In the event
that multiple sample types are required, tooling for each
type of sample could be exchanged after completion of
each sampling step and the program continued using the
same transporter.

Requirements for Robotic Sampling
One of the most frequent problems observed which results
in poor sample results has been the lack of attention to
basic technique. Common examples of conditions which
destroy the results of samples before they ever reach the
laboratory are the collection of gas samples by vacuum
pump  into  sealed vials not adjusted  to  standard
temperature and pressure for analysis, the collection of
soil samples which were deposited in the same pouch
without being individually sealed, and the use of the
same scoop for all samples collected.
To achieve  accurate  characterization of a  site it is
imperative that samples are taken under conditions that
guarantee their integrity.  This requires that individual
containers be maintained dean prior to  sample collection
and be kept sealed with the  contents isolated during
                                                  731

-------
transportation and storage. Also, control samples must be
taken frequently enough to verify the integrity of the
sampling system and the identity of the samples and
their location must be accurately maintained.  This is a
tall order for a robotic sampling system and requires
intense attention to detail.
Some reliable techniques must be used to plot the exact
location  of the individual samples.  For a specific site,
position  readings could  be taken from fixed markers by
camera,  sonar,  or  laser  technology.   In a  more
sophisticated setting with larger distances needing
evaluation by a mobile  robotic device, location may be
documented by satellite  triangulation.
It is also critical that the robotic device does not cross
contaminate  the sampling  site.   Specifically,  the
tracking  of surface materials from one location to another
or the loss of sample materials during collection could
potentially contaminate previously uncontaminated soil.
The specific  contaminants to be investigated and  the
degree of their toxicity determine what support will be
required. Some hazardous wastes and toxic chemicals
may be evaluated in simple field facilities while others
may require more sophisticated evaluation. The degree
of hazard present, the ease of analysis, the  number of
samples  to be analyzed, the requirements and cost for
transportation, and the size and needs of the project are
all factors which will determine support requirements.

Operational Safety
It is important to consider the safety of personnel in the
proximity  of robotic devices.  The  area  should be
restricted to exclude personnel not directly working with
the equipment. However, since robotic devices often draw
attention from admiring  spectators, it should be expected
that people not working directly with the robots will
often be present. Measures must be undertaken to ensure
the safety of all those present.

Summary
A decade ago several people were faced with a unique
problem at  the Three Mile  Island  reactor.   They
envisioned employing a mobile robot that would perform
all of the operations  necessary to solve the problem. It
took  three  years  to  make that  vision  a reality.
Similarly,  last year a  group of people at Niagara
Mohawk envisioned  using mobile robots to  help solve
some of the problems they had encountered. In this case
it  only took  six months to transform that vision into
reality.
The amount of time necessary to make a reality of such
visions has decreased tremendously during the past
decade. Mot surprisingly, the technology available has
had a tremendous increase during the same timespan.
Relevant applications of such technology are increasing
each day with the enforcement of  more stringent
regulations and with an increased public awareness of the
effects of exposure to hazardous substances.  Currently,
the nuclear and space industries are at the forefront both
of robotic technology  and  the  applications of that
technology. Robotic technology is now at a point where it
can be effectively applied to the characterization and
remediation of hazardous waste sites.
Applications  with  a  greater volume  of  repetitive
operations will be more effective than those which have
a once and done scope. @) The use of personnel involved in
previous projects provides a continuity of  experience
which also increases effectiveness,  even  for the once and
done applications.
For  companies  with  several  sites  which  require
characterization,  the greatest effectiveness of mobile
robotic operations is projected  through use of a dedicated
team which provides  management, supervision, and
continuity of all sites  from a central location.

References
RedZone Robotics, Inc., Design and Operations Manuals
for the Tethered Remote Operating Device, Unpublished
(1989)
Pavelek II M.D., Giefer  D., and  Hine R., "Remote
Reconnaissance Vehicle Program,"  NP-4265, Research
Project 1544, Electric Power Research Institute (Sept '85)
Vallem RJ.  and Jobe E.G.,  "Decontamination Using
Ultra High Pressure Water at  TMI-2," Trans. Am. Nucl.
Soc.,54,90(1987)
                                                     732

-------
              PRACTICAL  PROBLEMS  ENCOUNTERED  IN  REMOTE  SENSING
                        OF ATMOSPHERIC CONTAMINANTS
       Kirkman R. Phelps
       Michael S. DeSha
           Chemical  Research,  Development,  and  Engineering  Center
                           Detection Directorate
                     Aberdeen Proving Ground, MD 21010
    Recent sensing technology is now
ready and able to play a significant
role in Environmental Protection
Agency (EPA) programs.  The U.S. Army
Chemical, Research, and Development
Center (CRDEC) has been involved in
remote sensing of environmental
contaminants since 1951. During this
time much practical information has
been gathered concerning designing,
building, and testing remote sensing
systems.  This paper briefly examines
the practical development of remote
sensing systems which could benefit
the EPA in its mission to detect
potential environmental contaminants.

     Perhaps the least exciting and
most tedious of any remote detection
program is data base development.  But
for an instrument which will be used
to detect and discriminate thousands
of specific chemical species among a
plethora of natural and manmade
interferents this is the most
important first step in any standoff
detection program.  The detection and
discrimination of environmental
pollutants is an exceedingly difficult
problem because a clear base line
measurement is almost impossible to
get.  This makes a complete data base
very important.

     There are two types of data bases
used in the research and development
of any remote detection instrument.
The initial working data base consists
of spectra, interferents, and
backgrounds acquired either from
in-house laboratory spectrometers
or purchased from spectral data
base houses.  The second type is an
instrument specific data base
developed with a first generation
sensor (a crude, working instrument
designed from first principles).
The initial data base is used to
determine the type of instrument,
the spectral band in which this
instrument will work, the spectral
resolution and some initial
detection and discrimination
parameters.  This data base need
not be quantitative (i.e.
calibrated to a specific instrument
response) but quantitative data
will save some steps in future
development.  The instrument
database is used to fill in details
of instrument development and
"finetune" the instrument for the
work for which it is being
designed.

     For the most part two
categories of remote detectors
exist, active and passive.  Active
systems are based on the Light
Detection And Ranging (LIDAR)
concept.  The detection scheme,
whether it is differential
absorption (DIAL) differential
scattering (DISC), Raman, or laser
induced fluorescence (LIF), is
dependant on the detection
                                          733

-------
requirements of the system.  The LIDAR
emits LASER radiation, at frequencies
appropriate for the chemical under
investigation, and the radiation is
scattered back through a telescope to
the detector for analysis.  Passive
systems are based on either a grating
or an interferometrie spectrometer.
Interferometric systems are usually
employed for portable field
instrumentation because of the  size
and weight advantage over grating
systems of comparable sensitivity.
Passive systems use the ambient
radiation emitted or absorbed by the
chemical vapor under investigation as
the basis for detection.  The
detection analysis is similar to that
used in a laboratory Fourier transform
infrared (FTIR) spectrometer.   The
choice of either a passive or active
system is a consequence of the
spectral band of interest, resolution,
physical state of the contaminate, and
the use concept.

     The selection of spectral  band
and resolution is, at least initially,
a consequence of the chemical species
you want to identify.  In order to
detect a specific chemical you must
select a band and resolution which
affords the best chance of
identification of the chemical.  You
must also consider all possible
interferents against all backgrounds
you may encounter in your use concept.
This a particularly important decision
since all subsequent development will
proceed from this decision point.  You
must also determine what resolution
you require to completely discriminate
the chemical among possible
interferents encountered in your use
concept.

     We have used the phrase "use
concept" twice in the above paragraph
- what is use concept?  Use concept is
nothing more than a notion as how,
where, and under what conditions you
will use your remote sensor.  You can
make your initial conditions something
along the line of; I want to use the
detector everywhere, under any
conditions, and operated by an
untrained chimpanzee.  You then use
some computer modeling and knowledge
of current technology to back away
from this initial use concept.  This
is where a quantitative data base
comes in handy.  If you have
absorptivity coefficients or
scattering cross-sections available
for your chemical species and
interferents and spectral
emissivities for the backgrounds
then you may "build" computer
models of detectors and estimate
their sensitivities.  Computer
models are very convenient for
estimating performance for
instruments with widely different
design concepts including changes
in fieIds-of-view, detector element
response, LASER power, LASER
frequency agility, etc.  The
results of these models can then be
incorporated into a first
generation instrument which can be
used to develop a database that is
instrument specific.

     Unless you have ample
resources in electro-optics, opto-
mechanics, optics and system
integration you will run into some
difficulty building your first
instrument.  You have four choices,
design and fabricate the instrument
in-house; design in-house, have the
parts fabricated by contract and
integrate the instrument in-house;
design in-house but fabricate by
contractor; or find a contractor to
do the whole job.  The first option
requires considerable in-house
expertise and fabrication
facilities but usually very little
money.  The second option requires
the same in-house expertise but
almost no fabrication capability
and only slightly more funds.  The
third option requires some in-house
technical capability and more money
and the last requires in-house
technical capability, a good
contracting officer and lots of
money.  For a research/development
operation we suggest one of the
first two options and of these we
think that the second choice is the
best.  Our reasoning goes like this
all of these options require at
least some in-house technical
knowledge and if you don't have any
in-house fabrication capability
then the only other requirement is
some integration capability.  In
addition you can rely on some of
                                           734

-------
the expertise you will obtain from the
various parts vendors.  This
integration capability can be
initiated with a minimum of startup
time and allows your in-house
personnel to become intimately
familiar with the instrument they are
integrating.  This intimacy will
become vitally important in the later
stages of development.

     You've got your first instrument
in-house and you've worked out most of
the electronic and mechanical bugs.
What's next? - calibration,
characterization, and collection!

     Calibration is a basic precept to
understanding instrument response and
most of the operational attributes of
the instrument.  An unknown instrument
response hinders the ability to make
any type of confident detection and
completely destroys any discrimination
capability.  The initial calibration
method is dependant on the type of
instrument.  For example most passive
FTIR based instruments are calibrated
against some reference source which,
within some practical error, mimics a
blackbody, this permits an
understanding of total instrument
response.

     Instrument characterization is
essentially a "calibration" taking
into account use concept, optical
parameters (i.e. fieId-of-view, field-
of-regard, etc.), measurements against
contaminants and interferents in a
controlled environment, and a
measurement of how instrument response
changes in the operating environment.
These measurements permits the
operator to gain a complete
understanding of the instruments
capabilities or lack of capabilities,
the logistics of maintenance and
operation, and its usefulness as a
contaminate detector.

     Data collection in order to
define an instrument database is a
time consuming, tedious, and expensive
endeavor but is vitally important to
complete instrument development.
There are two stages of data
collection, open air testing of
interferents, simulants, and natural
backgrounds and controlled chamber
testing of obnoxious and dangerous
contaminants.  The chronological
development of this instrumental
database is of little technical
importance and is based solely on
an established or changing use
concept.  Chamber testing of the
obnoxious contaminants is necessary
to establish a sensitivity to known
quantities and is performed in
conjunction with simulant
sensitivity measurements to
corroborate instrument responses to
open air tests.  Open air testing
of interferents, backgrounds, and
simulants establishes instrument
performance in the operating
environment.

     Although detection and
discrimination have distinctly
different definitions in theory
they are practically impossible to
separate in practice. They are
intimately linked simply because
you cannot make a proper detection
of any contaminate without the
ability to discriminate it from
background clutter, for this reason
we will discuss them together.

     The initial ability to detect
and discriminate particular
contaminants or class of
contaminants rests solely on the
human perception.  The operator
must separate the contaminate from
the interferents and backgrounds
based on the use concept, the
operator's knowledge base, the
depth of the database, and the
operator's ability to understand
and adapt to changing environmental
conditions.  By the time instrument
development reaches this point the
use concept has, hopefully, been
established and this leaves us with
the task of developing some type of
detection and discrimination
algorithm based on the remaining
parameters.  You can, given an
infinite amount time, money and
manpower, develop an empirical
solution to this problem.  For all
practical purposes this is
impossible,  and any solution based
on a subset of parameters has at
best a very limited success rate.
Then what is your alternative?
Statistics!  The saviour of the
                                          735

-------
working scientist and the bane of the
absolutists.  Unfortunately there is
no single statistical method which
defines every detection and
discrimination problem - indeed there
are as many methods as there are
problems.  The method is defined by
all of the ingredients in the
development recipe, the use concept,
instrument type, knowledge base, and
database.  It is just a matter of
finding a method that fits (e.g.
filtering, database matching, etc.) or
developing a new method from a
combination of previously defined
methods.  What ever your choice we
strongly urge you employ the services
of an experienced statistician from
the beginning of your development
effort.  It will save you
considerable time and frustration.

     There you have it.  A
practical, albeit brief, recipe for
the practical problems encountered
in the development of remote
sensing instrumentation.  CRDEC has
developed several remote sensors
since 1951, some have successfully
managed the development cycle and
some have not.  But despite all the
requirement and funding vagaries,
CRDEC has acquired an extensive in-
house capability based on years of
practical experience.  This
experience and capability is
available to the EPA for its own
detection programs.
                                            736

-------
             A SI/LI BASED HIGH RESOLUTION  PORTABLE X-RAY ANALYZER
                     FOR FIELD SCREENING OF HAZARDOUS HASTE
                     Stanislaw Piorek and James R. Pasmore
                          Outokumpu Electronics, Inc.
                          P.O.Box L1069,  Langhorne,  PA
INTRODUCTION

Only  four  years have passed  since
the first publication describing the
application  of  a  portable  x-ray
analyzer,    (XRF),    for    on-site
chemical characterization of
contaminated soil [1].

During that  period,  field portable
x-ray   fluorescence   (FPXRF)   has
established  itself   as   the   most
useful technique for  a  broad  range
of environmental applications.  Its
well   known   attributes  such   as
ruggedness,    nondestructiveness,
minimal sample  preparation and speed
of  analysis  are  indisputably  the
factors contributing  to  its  growing
success.    However,     it     was
technological   advances   in    the
proportional    detector    (high
resolution)  and in  microprocessor
technology   (computing  power  and
portable architecture) which really
made   feasible  a   small,    truly
portable,  battery  operated  device
with analytical capabilities similar
to the laboratory XRF systems.
FPXRF ANALYZER CONCEPT

The most  successful  implementation
of  the   FPXRF  for   the   on-site
screening and analysis of inorganics
in hazardous waste  is  based  on the
aforementioned     microprocessor
controlled analyzer  connected  to  a
hand-held probe.
The probe contains an x-ray
source(s), a detector and a means of
reproducible  presentation  of  the
sample    for   measurement.    The
electronic unit accepts  the  signal
from  the probe,  processes  it  and
displays   the  result.    It   also
contains    power     supplies    and
interfaces for communicatiion  with
the operator and peripheral devices.

A   sealed   radioisotope   capsule
emitting x-ray or  low  energy gamma
rays  is  a   preferred  source  of
primary   radiation   for   portable
instruments.    Such   sources   are
rugged,  compact,  light weight  and
drift free.

A  high   resolution,  gas   filled
proportional detector  has  been for
years an integral part of  the  most
successful FPXRF analyzer available,
the X-MET 880.   Its much  improved
energy resolution of   10 to  12% as
compared    with    conventional
proportional  counters  (20%),  made
possible abandonment of  mechanical
means  of  element  separation  (so
called nondispersive XRF,  using  a
pair  of  balanced  filter   for  each
measured element)  in favor of  energy
                                      737

-------
dispersive XRF, based on electronic
separation of elements according to
their characteristic x-ray energies.
More  recently,  the  probe  of  the
analyzer has been modified to accept
two excitation sources and thus has
extended  the  range  of  elemental
analysis of the probe.
Fig. 1.  FPXRF Analyzer X-MET 880
         with a gas  filled detector
         probe.
QUANTITATIVE ANALYSIS

Quantitative    analysis        is
accomplished by employing empirical
calibration  methods.  Usually  a set
of 15 to  20  samples  is required to
develop calibration curves  for up to
six analytes  per calibration program
(model). The  instrument can quantify
six  elements  in  each  of  its  32
calibrationh models.   Availability
of calibration  samples may pose  a
problem  especially  in  situations
where not  much is known  about  the
site  to  be  analyzed.   Since  XRF
                      CLP   analyzed
                     the site to be
                      cal1ed    site
                      samples).   An
                     is calibration
                     i set of spiked
technique, it is important that the
calibration samples match in matrix
composition the  unknown  samples to
be  analyzed.    This condition  can
rarely  be met,  although the  most
accurate   results  have  always  been
obtained  when   the  analyzer   was
calibrated  with
samples collected on
investigated    (so
specific  calibration
alternative solution
of the analyzer with
soil samples,  so called site typical
samples [2].  This approach results
usually in a systematic error (bias)
in the XRF measurements. However, it
can be  easily  corrected  as  it  is  a
common practice to submit 10 to 20%
of all samples measured on the  site
with the  FPXRF  for  verification by
contract  laboratory program  (CLP)
analysis.    By correlating  the  XRF
with the CLP results one is able to
correct   for   the   bias   in   the
remainder of the XRF results.
This approach has been successfully
used  for screening and preliminary
evaluation of levels of contaminats
on  a  number  of  sites where  FPXRF
could  be   accepted   as   a  Level  I
analytical    method    (that     is
inaccuracy  up  to +/-  50%  relative
and   precision  up   to   +/-   10%
relative)  [3].
                                          HIGH  RESOLUTION  SI/LI  PROBE
While     the
configuration
enables   one
limits down to
elements  such
etc.    [2],
calibration
                 FPXRF    analyzer
                 described    above
               to   reach  detection
               100 to 200 mg/kg for
               as  Cu,  Zn,  Pb,  As,
               it    has   demanding
               requirements    when
handling the diverse sample matrices
common  in  analysis  of  hazardous
waste.   To address this  problem  a
new,  Si/Li  based, hand-held  probe
was  designed.   The probe combines
unsurpassed  energy  resolution  with
portability and ease of operation.

The  heart  of the probe  is  a  Si/Li
detector  featuring  30  mm   active
area   and   an   energy  resolution
better than  170  eV for the K-alpha
line of manganese at 1000 cps.
The  detector  is  cooled  by  a  small
                                        738

-------
LN2 capacity with a holding time  of
8 hours.  Dewar  construction enables
operation   of   the   probe  in  any
position  making it  truly portable.
There   were  no   adverse  effects
observed  due  to thermal cycling  of
the probe. The probe can accommodate
two  radioisotope sources  to  cover
the elemental range from K to U.
Fig. 2.  A prototype Si/Li probe.

The    probe    is    equipped   with
interlock mechanisms  which prevent
source  exposure  and  high  voltage
supply  to  the  detector,   whenever
the amount of LN2  in a dewar is not
sufficient.  The probe can be easily
set-up directly on the soil surface
for true in-situ measurements, or it
can be, after turning it over, used
as a sample probe  to measure samples
presented in cups.

Perhaps the  most  important feature
of the probe is that it can be used
directly    with    the    existing
population of X-MET  880's. The Si/Li
probe is therefore a useful addition
to the many  types that  already are
used with this analyzer.

Fig.  2  shows  a photograph  of  the
prototype Si/Li probe.
PROBE PERFORMANCE

The  advantage  of the state-of-the-
art  energy  resolution of the  probe
can  be  seen in Fig.  3 .   The  figure
shows two simulated spectra  as  would
be generated in  a sample with  Cr  to
Fe concentration ratio  of  1 to 20.
                                                COMPOSITE SPECTRUM OF 1:20 Cr/Fe RATIO
                                                      HIOH PFS. WOP Dr. AND SI/LI DrTTCTO*
                                                             / u V.
                                                          CHANNEL NUMBCR
                                                                 SI/LJ; rWHM -16D.V
                                           Fig.  3
        Comparison
        resolution.
of detector
It  is  clear  that  with  a  Si/Li
detector    it    is    possible   to
distinguish a minor  Cr peak from a
massive Fe  peak,  whereas even with
a   high  resolution   proportional
detector such  a faint Cr  peak can
hardly be seen.

Fig.  4  illustrates a  typical  soil
spectrum  excited   with   a  Cd-109
source and collected with the Si/Li
probe  connected to  the X-MET  880
FPXRF analyzer.

As  expected, all  peaks are clearly
resolved except for  the notorious
pair  of As  K-alpha  and Pb L-alpha.
However, it is  important   to  note
that the resolution of the detector
is    not    the    only   parameter
determining its  overall performance.
For  example,   a  gas  proportional
detector has much  higher detection
efficiency than  a small Si/Li diode.
This  is due  to  the  fact  that  a
typical     proportional     counter
collects  radiation   from   a  much
larger  solid  angle than  a  typical
Si/Li detector.  However, a
                                      739

-------
   HEISTC 2HETUIH091«IOI HOOK SHED KUGI1 8-NUIIKIEM1HUII

Fig. 4. Spectrum of  soil sample.

proportional  counter  will  usually
exhibit  also  a higher background
which    somewhat    offsets    its
efficiency    advantage.        The
improvement   in   sensitivity   and
detection  limits  achievable with a
Si/Li detector comes mainly  from the
low background of this  detector.

Although  the  proportional  detector
exhibits excellent performance with
conditions of optimal separation of
more than Z+2 atomic number spread,
when    adjacent     elements    (or
overlapping   spectral   lines)   are
present,  enhanced  resolution  is of
importance. In such  cases of severe
spectral  overlap  and  unfavorable
ratio  of  analyte  concentration to
interfering   matrix   element   the
resolution  factor  plays a critical
role.

Another  important   implication  of
superb  energy  resolution  is  the
ability  to  separate  coherent  and
incoherent   backscatter  peaks  of
primary radiation.  This enables one
to  implement  a  more sophisticated
data treatment, such  as those based
on    a    fundamental    parameters
approach,  which can  better handle a
diversity of sample  matrices.

At present, the  Si/Li  probe can be
used directly with the  X-MET 880 in
an empirical  calibration  mode.   An
extensive  development  program  is
being  completed   to  implement  a
fundamental parameters  based mode,
initially in  a  PC  connected to the
FPXRF analyzer via its  serial port.
                                           Typical  detection limits   obtained
                                           with   a   Si/Li   probe    for    a
                                           multielement  matrix  such as Cu,  Zn,
                                           As,  Pb are on the order of 30 to 80
                                           mg/kg as  opposed to  a 100  to  200
                                           mg/kg with a gas filled proportional
                                           detector.
                                           Further  work  is  in  progress   to
                                           further  refine  the   final   probe
                                           design and mathematical  algorithms
                                           for  data treatment.    These results
                                           will be  reported  in the near future.
[1] Chappell  R.W.,  Davis A.O.,  and
Olsen   R.L.   -   "Portable   X-Ray
Fluorescence as a Screening Tool for
Analysis  of Heavy Metals  in Soils
and Mine Wastes", Proc.  Natl. Conf.
on   Management   of   Uncontrolled
Hazardous Haste  Sites,  Washington,
D.C.,   pp.  115-119,  HMRCI,  Silver
Spring, MD, 1986.

[2] Piorek  S.  and Rhodes J.R. - "A
New Calibration Technique for X-Ray
Analyzers  Used in  Hazardous Waste
Screening", Proc. 5th Natl. Conf. on
Hazardous   Hastes   and   Hazardous
Materials,    pp.  428-433,  HMRCI,
Silver Spring, MD, 1988.

[3]    "U.S.    EPA   Data   Quality
Objectives  for  Remedial  Response
Activities Development Process",
EPA/540/G-7/003, US EPA Washington,
D.C.,  1987.
                                        740

-------
                              Measurement  and  Analysis  of
                    Adsistor  and  Figaro  Gas  Sensors  Used  for
                    Underground   Storage  Tank  Leak  Detection
                Marc A. Portnoff, Richard Grace, Alberto M. Guzman, Jeff  Hibner
                               Carnegie  Mellon  Research  Institute
                                    Carnegie  Mellon University
                                          4400  Fifth  Ave.
                                      Pittsburgh,  PA  15213
ABSTRACT
Gas sensor properties are measured with the purpose
of comparing  two  sensor technologies  used  for
underground storage tank leak detection. Four types of
Figaro gas sensors and the Adsistor gas sensor were
tested   in  simulated   underground  storage tank
environments using the Carnegie Mellon  Research
Institute (CMRI) automated gas testing facilities. This
automated system monitored the sensors' responses
while dynamically exposing them to various mixtures
of methane, butane and xylene. The sensors were also
tested to determine the effects of humidity on their
responses.  Sensor responses were characterized by
sensitivity,  selectivity,  and  speed of response and
recovery to selected test concentrations of methane,
butane and xylene.  The test results are presented as a
list of sensor specifications to allow  the potential end
user a  direct comparison of these two different types
of sensors.

INTRODUCTION
This study  was  initiated  to  help  the  users  of
underground storage tank (LIST) vapor phase product
leak detectors to better understand the capabilities and
limitations of commercial vapor sensors.  The study
was limited to characterizing two types of commercial
vapor sensors,  the Figaro [1] sensor and the Adsistor
[2] sensor.

Four types of Figaro gas sensors, models number 812,
813, 822, 823, and the  Adsistor gas sensor were
tested in simulated UST environments using the CMRI
automated gas testing  facilities.  The characterization
of these sensors resulted in a set of specifications that
allows direct comparison between the different sensor
types.    The  Figaro  sensors  are  metal  oxide
semiconductor  devices that  operate  at  elevated
temperature [1].   The  Adsistor sensor operates  at
ambient temperature and it works on the principle of
gas adsorption  [2] in a  polymeric material.
The selection of test gases was based upon a study
performed by Geoscience Consultants, Ltd  in 1988
[3]. This  study detailed the  hydrocarbon vapor
concentration at 27 gasoline service  stations from
three diverse geographic regions in the  United States.
Their findings indicated that:

•  all  the surveyed  locations  had some evidence of
   underground methane and gasoline vapor products.
•  methane  existed  in  high concentrations  at many
   locations.
•  tracking butane concentrations would be  useful in
   detecting recent gasoline leaks or spills.
•  m-xylene was a large  component of gasoline product.

Based on this study, methane was chosen as a potential
interference that may cause  false alarms  for UST
monitors. Also iso-butane and m-xylene were chosen
as tags because  they represent major chemical
constituents in gasoline.

The sensors were tested to determine their sensitivity
and cross sensitivities to methane, butane, and xylene
and  humidity  to   help the  UST  leak  detector
manufacturers to better understand how to use these
sensors. For  example, 1) if  a sensor responds to
methane but the instrument's user  is unaware of  this
sensitivity,  then, this instrument placed in  the field
could  produce  false  alarms due   to  methane
interference.  2) The humidity level underground at
UST sites is considered to be near saturation  [4],  If a
monitor  is calibrated with dry gas,  and  the sensor is
placed in the damp underground  environment,  this
also could lead to  false alarms, or worse, no alarm
will be set when a real leak is occurring.

Response time is not a  critical sensor  parameter for
this application since leaks in USTs generally occur
slowly and site monitoring is done  on  time scales of
days and not minutes.  However, recovery time can be
important in  situations  where  an accidental spill
                                                    741

-------
occurs.  In this case, if a  sensor  takes too long to
recover from the spill, the detection of a true leak
could be masked.

Sensor responses were characterized by sensitivity,
selectivity,  and speed  of response and recovery to
selected test concentrations of methane, butane and
xylene.  The test results are presented  as  tables of
sensor specifications to show the potential  end user
the advantages and disadvantages of using various
sensor types for monitoring  underground  storage
tanks.
 EXPERIMENTAL

 The data presented was  collected using  the  CMRI
 automated gas sensor characterization facility.  The
 facility has been designed to study the behavior of gas
 sensors and  characterize their response in terms of
 sensitivity,  selectivity,  speed of  response  and
 recovery, and  stability.  A computer  controlled gas
 delivery and data acquisition system (GDS) creates the
 test atmosphere  in the sensor test chamber  and
 records the corresponding sensor responses. The GDS
 controls  and sets proper levels of oxygen, nitrogen,
 and  water  vapor to create  a  clean  baseline
 environment through a network of mass flow dilution
 modules.  This clean air can then be contaminated with
 up to five different vapor compounds.  For this study,
 the   facility  was  modified  to  independently  set
 concentrations  for methane, (CHU), butane (C4Hs),
 and m-xylene  (CsHio).  The GDS  maintained  a
 constant flow rate of  1 liter/minute.

 A second gas system, delivering clean humidified air,
 was used to maintain the sensor atmosphere when the
 sensor chambers were not connected to the GDS.

 An on-line gas  chromatograph was used to verify the
 delivery of gases to the test chamber both during and
 in between tests.

 Three test chambers were built  to house the sensors.
 One chamber was built to test 9 Adsistor sensors and
 two chambers to house 12 Figaro sensors, 6 of each
 type.  All the materials used in the construction of the
 chambers were chosen to minimize contaminating the
 test atmosphere.  The chambers were built to power
 the sensors and monitor their responses in accordance
 with manufacture literature.  The volume of each test
chamber  was 1.2  liters.

The test  chamber temperatures were  monitored
during  testing.  The recorded room temperature and
that of the Adsistor test chamber temperature was
22°C ±. 1°C.   The  temperatures of the  Figaro test
chambers  were  33°C ± 1°C.
TEST  DESCRIPTIONS

Several types of tests were performed to characterize
sensor response. These tests include:

• Gas  concentration  ramp tests to determine sensor
   sensitivity and selectivity to individual test gases.
• Target  gas excursion test  to  determine  sensor
   response to the presence of multiple test gases.
• Water vapor  excursion tests  to  determine  sensor
   humidity response in the presence of multiple  test
   gases.
• Response and recovery time tests to determine how
  fast a sensor responds to changing concentrations of
  test gas.

Gas Concentration  Ramp Test
Ramp tests expose the sensors to a single test gas at a
time.   The sensors are exposed to five different  test
gas concentrations for each test gas. The ranges were
50,150, 500, 1500,  5000 ppm  for  methane and
butane and 10, 30,  100, 300,  1000 ppm for  xylene.
Each  concentration was  held for 30 minutes before
preceding  to the next  level. The sensors were exposed
to clean air for two  hours between each ramp.

Each of the ramp tests was performed at two humidity
levels.  The first set was conducted at  15,000  ppm of
water vapor.  This  level was chosen to represent the
humidity present at underground storage sites (97%
RH at 55 °F). The second set was done in dry air (less
than  50  ppm  water  vapor)   to  simulate   sensor
response when exposed to dry calibration gases.

Target Gas  Excursion Test
This test was designed  to show sensor behavior in the
presence of all  three test gases.  The sensors were
exposed to relatively  small concentrations of the three
gases, as a background  level.   Then  each gas was
separately  raised to  10 times  its background level.
The background gas concentration level was set to 500
ppm CH4, 500 ppm C^Q, and  100 ppm CsHio in air
containing 15,000 ppm of water vapor.

Water  Vapor Excursion  Tests
This test  was  designed  to show  how changes in
humidity effect sensor response in the presence of all
the three test gases.  The background level used was
the same as in the mixture excursion test.  The water
vapor concentration was then changed in thirty  minute
steps from 15000 ppm  , to 5000 ppm, to 1667 ppm,
to 0 ppm  water vapor, and then set back to  15,000
ppm.
                                                       742

-------
 Response and  Recovery Time Tests
 These tests were performed to determine the speed of
 response  and recovery to set levels of target gases.
 The tests were performed in air humidified to 15,000
 ppm water vapor. The sensors were measured at one
 minute intervals  during the  test.    The  xylene
 concentration  changed in thirty minute steps from  0
 ppm, to  1000 ppm,  to 100  ppm, to 1000 ppm and
 back to 0 ppm.

 The response time is defined as the time from when
 the new gas  concentration is first introduced into the
 chamber  until the sensor reaches  95% of  its final
 reading.    The recovery  time is  defined as the time
 from when  the new gas concentration  is  first
 introduced into the chamber until the sensor reaches
 95% of the total change  in the sensor reading.  The
 final reading  for both  recovery  time  and response
 time is defined  as  30  minutes after the new gas
 concentration has changed.

 SENSOR  MODEL EQUATIONS

 To  simplify  direct  comparison  of these  sensors,
 mathematical  model's were  used to convert sensor
 resistance (ohms) into gas concentration (ppm).  The
 model chosen  for the Adsistor is the one suggested by
 the  manufacture  [2].  The  model selected for the
 Figaro sensors is commonly used in the  literature [5].

 Adsistor  Sensor Model Equations
 Adsistor data was collected by measuring the sensor
 electrical  resistance.  The resistance  is related  to
 concentration  for most gas vapor, concentrations by
 equation 1.

 Eqn. 1  R-RD10c/k

 where  R  - Measured resistance  Rb - Resistance in
 clean air,  k = Gas constant  at ambient  temperature,
 and c = Gas concentration (ppm)

The Adsistor sensor resistance versus concentration is
 reported to be a  straight line when plotted on a semi-
 log  graph [2].

 For this paper, because the sensors did not respond to
the  lower  test concentration, a two point fit between
the  100 and 1000 ppm xylene were used  to determine
 Rb and k in equation  1.  Solving equation  1 for c yields
equation  2 which is  used to translate  the  measured
Adsistor resistance into a measured gas concentration.

Eqn. 2     c - k log10(R/Rb)
 Figaro Sensor Model Equations
 Figaro  sensor  data  was  collected  according  to
 manufacturer's recommendations  and converted to
 sensor resistance using equation 3.

 Eqn. 3    R = R| (VB - VR)/VR

 where R  -  Resistance (ohms). R|  = Load resistor
 (3920 ohms), VB -  Voltage  bias  (10  volts), and
 VR - (10- VB) = Sensor voltage

 The resistance concentration curve was observed to be
 approximately linear on a log - log plot. Therefore a
 power law model was adopted for these sensors as seen
 in equation 4.

 Eqn.  4   (a) Log(R) - log(Ro) = Blog(c)

          (b) R/Ro = CB

 where R =  sensor resistance, c - gas concentration
 (ppm),  B = power  law slope, and Ro •  sensor
 resistance when c-1

 The two  parameters RO  and B are determined  by
 considering  measurements taken at  c  =100, and
 c=1000 ppm for the  gas  in question.   Once  the
 parameters  are determined, the sensor resistance is
 translated into concentration by inverting equation 5
Eqn. 5

RESULTS and DISCUSSION

For this abstract only data comparing the Figaro 823
sensor and the Adsistor will be presented. The poster
board data presented shows that the Figaro 812, 822,
and 823 sensors all have comparable responses. The
Figaro 813 sensor is more sensitive to methane than
butane or xylene  and is of questionable  use  for UST
product monitoring.

The  test results  are  presented in terms of sensor
specifications  related  to  sensitivity,  selectivity,
response time,  and  reproducibility.    The  data
presented in this paper are shown in Tables 1-3. The
data is the average of nine Adsistor sensors, and six
Figaro 823 sensors.  The data are reported as  the
average  measured  sensor response  along  with  the
standard and percent standard deviations.

Sensitivity
The ramp tests were used to determine the test gas to
which the sensors were most sensitive.  The  sensors
were then modeled for this target gas.
                                                     743

-------
  The Adsistor sensors and the Figaro 823 sensors were
  clearly  more  sensitive  to  xylene  than either the
  methane or butane vapors.  Thus, these sensors were
  all modeled and calibrated for xylene.

  Selectivity
  The Figaro 823 sensors  respond  to both butane and
  xylene,  but are more sensitive to  xylene that butane.
  The ramp and excursion tests indicate that  these
  sensors are basically insensitive  to methane at the
  levels tested.

  The Adsistors are sensitive to  xylene levels greater
  than 100  ppm.   These  sensors  are  basically
  insensitive to the tested levels of methane and butane.

  Water  Response
  The response of Figaro 823  sensors  is strongly
  affected by changes in humidity.  Changes in reading of
  more than  50% were  observed when the humidity
  varied from dry to wet conditions. This  is  seen both in
  the ramp tests and the water excursion tests.

  The Adsistor sensors readings show little effect due to
  short term changes in humidity.

  Speed  of Response and Recovery
  Both the Figaro 823 and Adsistor sensors respond and
  recover  more quickly when changing from one xylene
  concentration  to another than  from clean  air to a
  xylene concentration level.

  Reproducibility
 All the Figaro 823  serssors tested in this study showed
 wide variations  in  the sensor parameters   and
 responses. The spread in response ranged from 15%
 to 100% of each other.

 The Adsistors sensors tested had model parameters and
 sensor responses with in 11% of each other.
 CONCLUSIONS

 Sensor specifications for direct comparisons of the
 two different sensor types, the Figaro MOS sensor and
 the Adsistor adsorption sensor, has been presented.

 Both sensor types appear to have sufficient properties
 to be used for UST leak detection. Both respond well to
 xylene, with the Figaro sensor being more sensitive to
 lower levels than the Adsistor.  Both sensor types are
 relatively insensitive  to methane,   which is  the
 primary  interfering compound underground.   The
 observed butane response for the Figaro sensor is not
 a serious problem since butane  js also a component of
gasoline.   The Adsistors  as  a  group were  more
reproducible,  and  had  a much  smaller  humidity
interference in comparison  to  the Figaro  sensors.
These two properties  make  the  Adsistor easier to
calibrate  and work with from  an instrumentation
point of view.  However,  the Adsistors  were observed
to have longer xylene recovery times than the Figaro
sensor.

Stability is  a major sensor specification  not  yet
studied.   It plays an important role in determining
how a sensor is  employed in UST monitoring.   If a
sensor changes with time, independent of the actual
conditions, it could lead  to false alarms and/or  not
being able to detect a leak. It is recommended that
stability test be  undertaken  to determine  the
calibration periods of the sensors and how their
characteristics change with time.
ACKNOWLEDGEMENTS
This research was funded by the U. S. Environmental
Protection  Agency,  Environmental  Monitoring
Systems Laboratory, Office of Underground  Storage
Tanks, Las Vegas, Nevada.
REFERENCES
1) Figaro Taguchi sensors are a product of Figaro
    Engineering of Japan represented by  Figaro USA,
    Inc., P. O.  Box 357, Wilmette,  IL  60091.

2) Adsistor Vapor  Sensors aro products of Adsistor
    Technology, P. O. Box 51160, Seattle, Washington
    98115.

3)  Schlez,  C., "Background  Hydrocarbon  Vapor
    Concentration Study for Underground Fuel Storage
    Tanks", Draft Final Report for U.S. EPA, Contract
    No.  68-03-3409, February  29,  1988.

4)  Personal communication with Philip B.  Durgin,
    PhD, U. S. Environmental Protection  Agency,
    Environmental  Monitoring  Systems  Laboratory,
    Las Vegas, Nevada, November 1990.

5) Grace, R., Guzman, M., Portnoff, M., Runco, P.,
    Yannopoulos, "Computational Enhancement of MOS
    Gas Sensor Selectivity", P-33,  Proceedings of the
    Third International Meeting on  Chemical  Sensors,
    Cleveland, OH,  September,1990
                                                       744

-------
Table #
Model P
Xylene F
Calibrati
Xylene F
Cross S
Methane
Butane
1: Figaro £
arameters
B
Ro
123 Sensor Specifications
Calibrated @15K ppm H20]
Average
0.56
9.1E+04
Std. Dev.
0.12
3.6E+04
% Dev.
21 .4%
39.6%
leadings (ppm) @ 15K ppm H2O
id at 100 and 1000 ppm Xylene
Actual Cone
10
30
100
300
1000
leadings (p
Actual Cone
10
30
100
300
1000
ensitivlty (p
5000 ppm
5000 ppm
Average
10.7
43.5
100.0
239.9
1000.0
Std. Dev.
5.8
10.0
0.0
36.1
0.0
% Dev.
53.8%
23.0%
0.0%
15.0%
0.0%
am) @ 0 ppm H2O
Average
0.2
1.4
5.9
38.8
437.8
Std. Dev.
0.3
1.4
4.5
21.5
136.7
% Dev.
141.3%
100.0%
75.0%
55.4%
31.2%
>pm Xylene) @15K ppm H2O
Average
23.5
793.4
Std. Dev.
8.6
792.9
% Dev.
36.6%
99.9%
95% Response Time (Minutes) <§> 15K ppm H20


0 to 1000 ppm
100 to 1000 ppm
95% Recovery Time

1000 to

100 ppm
1000 to 0 ppm
Average
15.30
10.18
Std. Dev.
6.7
7.4
% Dev.
42.3%
68.7%
(Minutes) @ 15K ppm H2O
Average
3.33
4.08
Std. Dev.
1.0
0.9
% Dev.
31 .0%
23.1%
Table #2
Model Pa
Xylene Re
Calibratec
Xylene Re
Cross Se
Methane
Butane
Adsistor Sensor Specifications
rameters [Calibrated @ 15K ppm H20]

K
Rb
Average
2987.72
3.5E+02
Std. Dev.
308.26
3.5E+01
% Dev.
10.3%
10.0%
adings (ppm) @ 15K ppm H2O
1 at 100 and 1000 ppm Xylene
Actual Com
10
30
100
300
1000
Average
61.5
67.9
100.0
233.3
1000.0
Std. Dev.
2.8
2.3
0.0
3.7
0.0
% Dev.
4.6%
3.4%
0.0%
1.6%
0.0%
adings (ppm) @ 0 ppm H20
Actual Coru
10
30
100
300
1000
nsltlvity (pp
5000 ppm
5000 ppm
95% Response Time

0 to 1000

ppm
100 to 1000 ppm
95% Recovery Time


1000 to 100 ppm
1000 to 0
ppm
Average
118.9
126.4
139.0
251.3
997.6
Std. Dev.
13.1
12.5
12.1
10.7
9.7
% Dev.
11.1%
9.9%
8.7%
4.3%
1 .0%
m Xylene) @15K ppm H2O
Average
62.9
61.8
Std. Dev.
4.0
3.2
% Dev.
6.3%
5.2%
(Minutes) @ 15K opm H20
Average
7.29
7.86
Std. Dev.
1.5
1.8
% Dev.
18.6%
20.5%
Minutes) @ 15K ppm H20
Average
>30
>30
Std. Dev.
0.0
0.0
% Dev.
0.0%
0.0%
Table #3: Figaro 823 and Adsistor Sensor Response to Target
Gas Excursion Test and Water Vapor Excursion Test
Calibrated for Xylene @ 15 K ppm H2O
Actual
H20
(Dom)
15002
15002
15002
15002
15002

15002
4999
1667
0
15002
Actual
Methane
(ppm)
500
4999
500
500
500

500
500
500
500
500
Actual
Butane
Jppm)
500
500
4999
500
500

500
500
500
500
590
Actual
Xylene
(ppm)
100
100
100
1000
100

100
100
100
100
100












F
Average

306,7
321.2
1042.6
1720.5
284.7

272.8
157.4
100.1
57.6
318.0
Igaro 8:
Std. Dev.

219.2
234.9
1086.7
696.1
223.5

186.2
102.4
59.3
30.3
247.0
3
% Dev.

71.5%
73.1%
104.2%
40.5%
78.5%

68.3%
65.0%
59.3%
52.7%
77.7%












i
Average

141.2
142.2
142.3
940.6
213.4

134.4
137.1
131.0
127.1
111.3
idsistor
Std. Dev.

8.3
8.6
8.5
10.3
12.1

11.1
13.0
13.9
14.5
10.8
5
% Dev.

5.9%
6.1%
6.0%
1.1%
5.7%

83%
9.5%
1 0.6%
11.4%
9.7%
745

-------
          Extraction Disks for  Spectroscopic Field  Screening Applications
              Edward J. Poziomek
        Environmental Research Center
        University of Nevada, Las Vegas
        Las Vegas, Nevada 89154-4009
       DeLyle Eastwood, Russell L Lidberg,
                 and Gail Gibson
     Lockheed Engineering and Sciences Co.
            Las Vegas, Nevada 89119
Introduction

Field screening methods for hazardous waste site
investigations need to be rapid and low cost to
support on-site monitoring and characterization
activities.  The challenges are enormous because of
the number and variety of chemicals that could be
encountered.  Detecting  and monitoring
contamination  of water is one scenario which could
benefit from the availability of a relatively simple field
screening method.  The data could be used to decide
whether to apply more rigorous analytical methods in
the field and/or to send samples back to the
laboratory.

The research results described in this paper bring
out the potential  of utilizing solid phase extraction
membranes as part of a field screening method.

Concept Description

The idea involves using  commercially available solid
phase extraction membranes to preconcentrate
pollutants onto the membrane by sorption from
aqueous solution followed  by nondestructive
spectroscopic  measurements on site using man
portable or fieldable instruments. Depending on the
analytes being sought and the systems' parameters,
the measurements could involve ultraviolet/visible
luminescence  directly, colorimetry/fluorometry with
appropriate  reagents, X-ray fluorescence analysis,
and/or radioactivity determination.

The solid phase  extraction membranes normally
serve as alternatives to column chromatography in
preconcentrating analytes from dilute solution.  The
use of solid-phase extraction techniques to replace
conventional liquid-liquid extraction for isolating
analytes has gained much popularity.  Two reviews
on water analysis cite various examples (1, 2). The
usual approach is to use short columns or cartriges
containing various solid sorbents.  Such columns are
prepacked and readily available from a number of
manufacturers. The use of solid phase extraction
membranes for preconcentrating analytes is also
gaining popularity.  The type of sorbent used to
concentrate trace materials can vary widely
depending on the analyte and the medium.  The
sorption theory behind the process relates to removal
of components from both gases and liquids.


Organic analytes preconcentrated on the supports
are usually extracted with an appropriate solvent.
The extract is then analyzed using an appropriate
laboratory method such as gas chromatography
(GC), liquid chromatography (LC) or some
hyphenated technique, e.g., GC-mass spectrometry
(MS).  The concept pointed out in this paper involves
examining the extraction membrane directly using
solid state spectroscopy.  Laboratory analysis would
be an available option after field screening.

Method  Description

A variety of information is available in the literature
on solid phase extraction methodology.  For
example, the use of solid phase membranes in the
form of 25- or 47- mm disks for the extraction of
pesticides, polychlorinated biphenyls (PCBs), and
phthalates at the microgram per liter level was
reported recently (3).  The purpose of the work was to
replace liquid-liquid extraction with a more rapid and
less labor intensive technique.  Standard filtration
equipment (a laboratory suction flask)  was utilized.
Groundwater, surface water, and laboratory tap water
were used for pesticide, PCB, and phthalate
analysis, respectively. Adsorbed organic species
were eluted from the disks with a small volume of an
                                                 747

-------
appropriate solvent for subsequent chromatographic
separation. Recoveries usually exceeding 80 to 90%
were obtained for the classes of compounds
examined.  The membranes were obtained from
Analytichem International under the trademark
Empore with a typical composition of 90% (by
weight) of octyl (C8)- or octadecyl (C18)- bonded
silica particles and 10% polytetrafluoroethylene
(PTFE).

The concept described in the present paper extends
the application described above by examining the
solid phase extraction disks in a nondestructive
manner utilizing solid state spectroscopy before the
elution step.  Sufficient information may be obtained
from the spectroscopic examination to often eliminate
the need for any further work thus saving additional
time  and resources.
 Further savings of time and costs are possible if the
 filtration step was also eliminated. For example, the
 solid phase extraction disks could be used in a dip
 stick mode. Alternatively, tabs of the extraction disks
 could be placed into a sample of the water being
 examined.

 The concept is illustrated in Figure 1. The surface of
 the tab,  modified with long alkyl chains, attracts the
 analytes. After a specified amount of time the tab is
removed, allowed to dry, and examined with an
appropriate portable instrument such as a
spectrofluorometer, depending on the analytes being
sought

The use of solid phase extraction media in a static
configuration in which the analytes must diffuse to
the surface has not been reported previously.
However, we have determined that this is not only
feasible but can also provide semiquantitative
information.  An experiment is describe below which
simulates scenarios in which a dip stick is used with
a water sample or in which a tab is inserted into a
well.

Experimental

C18 Empore (TM) solid phase extraction disks were
examined  for the sorption of anthracene from water
and then analyzed nondestructively using solid-state
fluorescence spectroscopy.  Tabs (1 cm x 2 cm) were
cut from the disks and suspended without stirring in
40 ml of aqueous solutions containing ppb
concentrations of anthracene at room temperature.
The tabs were alowed to stand in the solutions for
given time intervals at different concentrations of
anthracene. The tabs were then withdrawn, allowed
to dry in air, and examined front surface using solid-
state fluorescence spectroscopy. A Spex laboratory
spectrofluorometer was utilized.  Figure 2 shows  a
                                                                   SPECTROSCOPIC
                                                                FIELD MEASUREMENT
         Figure 1.  Illustration of the concept of using tabs from solid phase extraction disks to sorb
         analytes from aqueous solution followed by nondestructive solid state spectroscopic
         examination.
                                                    748

-------
CO
UJ

UJ
u
g
o
UJ
c
o

u?
                         350
                                             EMPORE C18
                                              100ppb
                                              /SO ppb
                                               30 ppb
                                              ',10 ppb
                                                1ppb
                                              /BLANK
                                                              2 HOUR EXPOSURE TO
                                                              ANTHRACENE IN WATER
                                                              EXCITATION= 254 nm
            380   395  410  425  440  455  470  485  500
                    WAVELENGTH (nm)
            Figure 2.  Solid-state fluorescence emission curves of Empore C18 tabs that had been
            allowed to stand for two hours in water containing 1-100 ppb of anthracene.  (Excitation
            wavelength 254 nm; band passes 4 nm and 1 nm for the excitation and emission
            monochromators, respectively)
series of solid-state fluorescence curves of Empore
C18 tabs that had been allowed to stand for two
hours in water containing 1-100 ppb anthracene.
The intensity of the emission peak at 380 nm versus
anthracene concentration, was found to be linear.
Various relationships were also found in other
experiments, e.g., a linear increase in solid-state
fluorescence intensity was observed of tabs taken at
various time intervals (minutes to a day) from
solutions containing 10 ppb of anthracene.

Discussion

Though the results reported in this paper are
preliminary, the basic idea of using solid  phase
extraction disks in combination with solid-state
spectrqscopy is attractive to pursue for field
screening applications.  The individual technologies
have strong scientific bases and do not need
extensive development work, although the use of
solid phase extraction membranes in a dip-stick
mode is new.
                               Attractive features are listed below:

                               0  The method is nondestructive.

                               0  Extraction disks are commercially available.

                               0  The potential exists for at least semi-quantitative
                               analysis.

                               0  The method is relatively simple.

                               0  The opportunity exists for screening a variety of
                               organic and inorganic compounds.

                               0  The method is readily adaptable to decision-
                               making in the  field.

                               There are no apparent barriers to overcome in
                               extending the  techniques to  environmental
                               monitoring in aqueous media for a variety of
                               analytes.  Nevertheless, the combination of
                               extraction and nondestructive spectroscopic analysis
                                                  749

-------
using solid phase membranes has not been
examined sufficiently to allow limitations to be
defined thoroughly.

Future studies will focus on concept validation.
Various analytes (organic and inorganic) will be
examined using solid phase extraction
disks/membranes both in dip-stick and filtration
modes.

Notice

Although the information in this paper has been
funded wholly or in part by the U.S. Environmental
Protection Agency under Cooperative Agreement
No. CR814702-01 with the University of Nevada -
Las Vegas, and under Contract 68-CO-0049 with
Lockheed Engineering and Sciences Co., it does not
necessarily reflect the views of the Agency and no
official endorsement should be inferred.  Mention of
trade names or commercial products does not
constitute endorsement or recommendation for use.

References

1.    MacCarthy, P., R. W. Klusman, and J. A.
      Rice, "Water Analysis," Anal. Chem.
      61(12), 1989, 269R-304R.

2.    MacCarthy, P., R. W. Klusman, and J. A.
      Rice, "Water Analysis," Anal. Chem.
      59(12), 1987, 308R-337R.

3.    Hagen, D. F., C. F. Markell, and G. A.
      Schmitt, "Membrane Approach to  Solid-Phase
      Extractions," Anal. Chim.   Acta, 236,1990,
      157-164.
                                                  750

-------
            FIELD ANALYTICAL SUPPORT PROJECT (FASP) DEVELOPMENT OF HIGH-PERFORMANCE
                LIQUID CHROMATOGRAPHY (HPLC) TECHNIQUES FOR ON-SITE ANALYSIS OF
                          POLYCYCLIC AROMATIC HYDROCARBONS (PAHS) AT
                                  PREREMEDIAL SUPERFUND SITES
                     Andrew Riddell, Andrew Hafferty, and Dr.  Tracy Yerian
                                 Ecology and Environment, Inc.
                                   101 Yesler Way, Suite 600
                                  Seattle, Washington  98104
INTRODUCTION

Active and inactive voodtreating facilities
employing creosote are one of the classes
of industry most often investigated  during
the preremedial or site assessment phase of
hazardous waste investigations and cleanups
in the Pacific Northwest.  Creosote is
composed almost exclusively of polycyclic
aromatic hydrocarbons (PAHs).  This group
of organic compounds is listed in the U.S.
Environmental Protection Agency (EPA)
Target Compound List (TCL), and significant
numbers of samples are submitted annually
to the CLP for semivolatile (GC/MS)
analysis, which includes the PAH fraction.
Turnaround time between sample collection
and receipt of validated data is generally
7 to 9 weeks.

The Field Analytical Support Project (FASP)
program developed by Ecology and Environ-
ment, Inc. (E & E) is utilized when project
data quality objectives (DQOs) include any
of the following as goals:

o  Rapid turnaround of data results;

o  Extensive sampling for site
   characterization;

o  Optimization of sampling location
   selection while investigators are
   on-site; and/or

o  Prioritization of samples for more
   expensive CLP analyses.

FASP data are utilized routinely to
supplement and enhance the more rigorously
analyzed CLP results.  Use of FASP during
site investigation activities has been
demonstrated to provide both significant
project cost savings and improved
descriptions of contaminant distribution.

E & E's previously developed gas chroma-
tographic method with flame ionization
detection (GC/FID) for analyses of PAHs in
contaminated soil has been demonstrated to
provide results of good comparability with
samples analyzed through the CLP.  However,
high performance liquid chromatography
(HPLC) with in series ultraviolet/visible
(UV/Vis) and fluorescence detectors offers
numerous advantages over early FASP
methodologies:

o  HPLC instrumentation requires fewer
   gases for field analysis;

o  Two detectors provide real-time
   confirmation of target analytes;

o  HPLC allows injection of larger sample
   volumes, yielding lower method
   quantitation limits; and

o  HPLC methodology provides better
   resolution than the GC methodology for
   comparable analysis times.

The HPLC method developed for analysis of
PAHs in contaminated soil utilizes small
volumes of sample and solvents, and
disposable glassware to minimize the
generation of investigation-derived waste
in the field laboratory.  Rapid extraction
                                                751

-------
and analysis techniques are employed to
allow the shortest possible turnaround time
for on-site samples.

SYSTEM SELECTION

Five commercial systems were evaluated for
FASP use:  Hewlett Packard, Shimadzu,
Spectra-Physics, Dionex, and Waters.  The
primary considerations for purchase were
ruggedness, size, and simplicity (ease of
operation and maintenance).  Secondary
considerations were cost, warranty,
compatibility with Nelson analytical data
processing system, and technical training/
support.  Finally, potential future analy-
tical uses of the system (other analytes of
interest that may be analyzed with the
chosen HPLC) were investigated.

The Spectra-Physics system was chosen as
the most appropriate and cost-effective
instrumentation for field applications.
The physical space constraints for field
laboratories are met by the system, and the
components operate with standard llOv
power.  The system is equipped with a
universal system organizer, which facil-
itates securing the instrumentation during
mobilization for field use.  All mainte-
nance (except electrical) is performed
through front entry into the pumps and
detectors, which in E & E experience, is a
critical necessity for field repair or
maintenance.  The Spectra-Physics SP8800
gradient pump is equipped with an automatic
maintenance log, automatic cleanup cycle,
and self-diagnostic information on elec-
tronics and flow performance.  In a cost
comparison of price quotations, the Spectra
Physics system was the least expensive
overall, with a total system cost of
$30,399.00.  The warranty on the Spectra-
Physics pump and the UV/Vis detector is
5 years.  Shimadzu, Dionex, and Waters each
offered a 1-year warranty, and Hewlett
Packard offered a 90-day warranty in the
base purchase quote.  The field technical
representative for the Spectra-Physics
system is based in Portland, Oregon;
technical support is also available through
an '800' telephone number.  Technical
support includes system installation and
on-site training for all chemists.
Specific applications support is also
available.  Finally, this system is
currently in use at the National Oceanic
and Atmospheric Administration and the
Federal Drug Administration laboratories;
both laboratories require instrumentation
of rugged, durable quality.

EXTRACTION AND ANALYSIS

One ± .01 gram of soil was weighed into a
12 mL disposable culture tube.  The sample
was then extracted twice.  Consecutive 5-mL
volumes of acetonitrile were repipeted into
the culture tube, vortexed 1 minute, cen-
trifuged 10 minutes, and combined in a
10-mL graduated centrifuge tube.  The
sample extract was evaporated to 1 mL under
a gentle stream of nitrogen.  Aliquots of
the concentrated extracts were injected
into the HPLC column for analysis.

The PAHs were analyzed with a Spectra-
Physics Gradient HPLC System equipped with
in-series fluorescence and UV/Vis detec-
tors.  A stainless steel chromatography
column (25 cm x 4.6 mm, 5 mm octadecylsilyl
stationary phase) under isothermal
conditions was employed.  Analyte separa-
tion was achieved using an acetonitrile/
water mobile phase with initial flow
conditions of 35:65 v/v acetonitrile:water
for 2 minutes followed by a 14 minute
linear gradient to 100% acetonitrile.  The
mobile phase composition was then held at
100% acetonitrile for 9 minutes.  Flow rate
during the analysis was 1.5 mL/min and the
analytical run time was 25 minutes.
Samples were quantitated based on a
five-point initial external calibration of
all target analytes.  The linear regression
coefficients of all analytes routinely
exceeded .995.  Samples with large
interfering areas were diluted and re-
analyzed.  Samples were analyzed for the
following PAHs:

Naphthalene
Acenaphthylene
Acenaphthene
Fluorene
Phenanthrene
Anthracene
Fluoranthene
Pyrene
Benzo(a)anthracene
Chrysene
Benzo(b)fluoranthene
Benzo(k)fluoranthene
Benzo(a)pyrene
                                               752

-------
Dibenzo(a,h)anthracene
Indeno (1,2,3,-c,d)pyrene
Benzo(g,h,i)perylene

RESULTS AND  DISCUSSION

Method quantitation limits for HPLC/UV and
GC/FID are presented in Table 1.  For
routine use, HPLC/UV/Vis results vere used
for identification and quantitation of the
PAHs; fluorescence detection allows prac-
tical quantitation limits approximately 10
times lower  than the reported UV quanti-
tation limits, and was used primarily for
confirmatory analysis of the quantitative
data.

For this study, a representative number of
samples from a site previously investigated
(and known to be contaminated with PAHs)
were split for sample analysis, matrix
spike analysis, and duplicate analysis, by
both HPLC and GC methodologies in order to
compare the  analytical results.

HPLC sample  results showed reasonable
agreement with GC comparison analyses.
HPLC data and GC data for three soil
samples are  presented in Table 2.  Analytes
listed in the method that are not reported
in Table 1 were not detected above the
method quantitation limits by either
analytical system.

To illustrate the efficiency of the HPLC
extraction technique, results from three
matrix spike events are summarized in
Table 3.  Three aliquots of a PAH-free soil
sample were  spiked and subsequently
analyzed by  HPLC to generate the matrix
spike recovery data.

HPLC matrix  spike results showed consis-
tently higher recoveries than the matrix
spike analyzed by GC.  This difference is
probably due in part to the loss of
analytes during the cleanup procedure
performed as part of GC sample preparation.

Duplicate analysis results of three
contaminated soil samples are reported in
Table A.  Duplicate sample results from
both HPLC and GC displayed substantial
variability.  This phenomenon was due
primarily to the non-homogenous nature of
the soil matrix at the site; the GC results
from the initial site investigation also
demonstrated this variability.  Variability
of the GC results also could be influenced
by the cleanup step of the GC sample
preparation.

CONCLUSION

Recent developments in HPLC, including
gradient elution and dual in-series
detectors, have been introduced into
E & E's FASP arsenal of instruments and
techniques.  Chemists may now provide
reliable data of known and documented
quality on PAHs in a near real-time mode to
site investigators.  Use of multiple
detectors provides supplemental information
regarding the accuracy of both the quali-
tative identification and quantitative
measurement of target analytes.  The data
presented in this study document the
accuracy and precision of the method for
both standards and real world samples.
FASP analyses employing HPLC for PAH
measurements are designed to meet the DQOs
and data use guidelines for the needs of
preremedial site investigators.  With
appropriate alterations, this method also
can be an effective analytical option for
other types of investigations involving
screening activities.
                                             753

-------
                               A  FIELD COMPARISON OF MONITORING METHODS
                             FOR WASTE ANESTHETIC GASES AND ETHYLENE OXIDE
     Stanley A. Salisbury, National  Institute for Occupational  Safety and Health (NIOSH),  Atlanta
     Region.  Atlanta.  GA; G.E.  Burroughs.  NIOSH. Cincinnati.  OH;  William  J.  Daniels.  Charles
     McCammon. Steven A. Lee,  Denver Region,  Denver,  CO.
Purpose  and Objectives
Several electronic  direct reading  Instruments,
that have been  or could be used  for  monitoring
toxic gases and vapors in hospital environments,
were evaluated by  Investigators from the National
Institute  for  Occupational  Safety  and  Health
(NIOSH) in a series of three field studies.   The
selected Instruments were used to measure waste
anesthetic  gases  and  vapors  released  during
surgical procedures  1n  operating rooms,  and  to
monitor  ethylene   oxide   (EtO)   concentrations
during   the   operation   of  gas   sterilizers.
Instrument readings were compared with results
from conventional  Industrial hygiene air sampling
methods.  The objectives of these field studies
were to:  (1)  compare calibration  and  operating
techniques for  several  types of  direct reading
Instruments,   (2)  compare  Instrument  operational
advantages and disadvantages during actual field
survey applications and,  (3) compare accuracy and
precision  of   direct   reading   instruments  to
conventional  a1r sampling methods.  The locations
chosen included two separate field surveys  at a
small community hospital, and a  third,  and  more
extensive survey  at a large medical  university
teaching hospital.

Methods and Procedures
Battery powered air sampling pumps,  configured
with bag filling  outlet  ports, and 40  Liter (L)
Tedlar* sampling  bags  were  used  to collect air
samples.   The  bagged   air  samples  were  then
comparatively   analyzed   by   direct   reading
Instruments   and    conventional   sampling   and
analytical methods.  Sampling pumps modified for
bag filling were  typically set to  a 1  liter per
minute (Lpm)   flow rate.   During  actual  surgical
procedures,  breathing  zone  air  samples  were
collected either  from the "Scrub  Nurse"  or  from
the  anesthetic  cart.    Samples  were  collected
through  vinyl   plastic  tubes connected to  air
sampling  pump  and  bag  assemblies.    Area  air
samples were  collected  near the operating  room
exhaust vents.

Sterilization gas samples were also collected in
40L  and  SOL  Tedlar*  bags  using  bag  filling
sampling pumps.   Air samples were collected from
several points  near EtO gas  sterilizers  during
gas  purge  cycles and  after hospital  personnel
cracked  open   sterilizer   doors   to  dissipate
residual  gas  before  removing  a  load  to  an
aeration  chamber.    Other   air   samples  were
collected from  inside mechanical  enclosures and
near  floor  drains.    Air  samples  collected  In
Tedlar* bags were then analyzed on-site using the
selected  direct  reading  instruments.    Direct
reading Instruments  were  also used for continuous
and  sequential  real-time monitoring of EtO and
Freon 12 (d1chloro-d1fluoromethane) during purge
cycle  operations and  during  unloading  of  EtO
sterilizers.

Where sufficient sample remained in sample bags,
and after direct analysis with instruments  showed
measurable concentrations  of  either  halogenated
anesthetics (isoflurane  or halothane), EtO,  or
Freon  12}  air  samples were  withdrawn from  the
bags  using   conventional   NIOSH   sampling  and
analytical  methods'1"3'   for  the  analytes   of
Interest.  To obtain analytical  precision data.
three  samples   from  each  selected  bag  were
collected and subsequently analyzed by the NIOSH
contract laboratory. A three-outlet manifold was
connected to the bag valve  and each outlet of the
manifold  was  connected  to  the  Inlet  of  the
appropriate sorbent  tube.   Known volumes  of the
sample were  pulled   from  the  bags  through  the
sorbent  tubes   using    pre-calibrated   battery
powered air sampling  pumps.   NIOSH  recommended
sampling rates were  used.  Based on results from
direct  reading   measurements,  sufficient   air
sample  volumes   were  pulled  from the  bags  to
ensure  the  analytes  collected on the  sorbent
                                                 755

-------
tubes were above the  NIOSH  published  analytical
limits of quantitation.  Results from laboratory
analyses of the  air samples collected  from  air
sample bags were compared to instrument readings
from direct analyses of those same bag samples.

Direct  Reading Instruments Tested
Real-time instruments evaluated were the Briiel  &
Kjaer  (B  &  K)  Multi-gas Monitor Type  1302,  the
Mi ran 103 Specific  Vapor analyzer  and  Mi ran  1B2
Portable Ambient Air Analyzer manufactured by  The
Foxboro   Company,   the  Photovac   Model   10S50
Portable Gas Chromatograph  (GC),  and  the  Summit
Interests, Model SIP-1000 Portable GC.

The  Bruel  and  Kjaer  (B  &  K) Model  1302  gas
monitor   uses    a   photoacoustic   spectroscopy
detection technique to measure simultaneously up
to five gases or vapors, plus  water vapor,  down
to    the    part-per-billion    (ppb)     range.
Microprocessor control  allows  the  instrument to
compensate  for  water  vapor  and  other  gaseous
interferences  such  as  carbon  dioxide.     The
photoacoustic  spectroscopy  technique  uses   an
infrared  (IR)   light  source  focused  through  a
chopper which  pulses  the IR  light  beam through
one of six optical  filters rotated into position
on a  filter carousel.   Light  transmitted  by  the
optical filter  at  the  predetermined wave  length
is   selectively  adsorbed   by  the  gas   being
monitored.      The   gas   sample   analyzed   is
automatically pumped  into  a hermetically  sealed
analysis  cell.    The  modulating  expansion  and
contraction of  the gas in the  cell  caused  by
heating  and  cooling   of   the  gas as  it   is
irradiated  by   the  pulsed  infrared light  beam
generates  pressure  waves in  the  cell   that  are
detected  by  sensitive  microphones  mounted  on
opposite  sides the cell.  The amplitude of these
pressure   waves    is    proportional    to   the
concentration of the measured  gas.   After  the
first  analysis, the  filter  carousel   turns  to
bring  the  next  optical filter  into  position so
that  other  gases  in  the  cell   which  adsorb
infrared  light  at  different wavelengths  can be
subsequently analyzed.   The  1302  is operated by
using the push-buttons  and  the two-line digital
display on its front panel.  Measurement results
are  automatically   stored   in  the  instrument's
"display memory" and can be permanently stored in
one   of   ten   "background   memory"  locations.
Display memory data also can be transferred to a
printer or personal computer.

The  Mi ran  103  and  Mi ran  1B2  analyzers  are
single-beam  infrared  spectrometers.    The  air
sample analyzed  is pumped through  the  analysis
cell at a flow rate  from 25-30  Lpm.  Quantitative
analysis of the gas in  the  cell  is accomplished
by  electronically  detecting  and   comparing  the
energy  of an  infrared  light  source  with  the
energy of the  light after passing through the gas
in  the  cell.    Infrared  energy  lost  through
absorption  by   the   gas  is  proportional   to
concentration of the gas in the cell.   The Mi ran
103 Specific Vapor Analyzer can  monitor  several
gases and vapors.   However,  to change from  one
gas or vapor to  another, a different  filter  and
meter scale must be  installed.   The  use  of  the
Mi ran 103 for monitoring nitrous oxide has  long
been  the established   sampling  and  analytical
method used by NIOSH investigators.'4'   The Mi ran
1B2 Portable Ambient Air Analyzer  is  a portable
microprocessor-controlled infrared  spectrometer
configured  with  an  internal   library  of  116
precalibrated compounds and  ten user  selected
compounds.   It  uses interactive programming  to
prompt the operator through available choices  and
functions.

Both the Summit  and  Photovac portable  GCs  use  a
photoionization  detector, and  each  was equipped
with a Carobopak BHT packed column.   The  carrier
gas used was ultrapure air.  Both GC columns were
operated  at  ambient temperature.   Samples  and
calibrations  standards   were   injected   using
gas-tight  syringes.    Injection  volumes  ranged
from 10 to  500  microliters.   Gas concentrations
were detected by measuring the peak height of an
injected  sample  from   a  recorder  output,  and
comparing the   result  to  a  calibration  curve.
Standards  were  periodically  injected  between
sample injections.

During the surveys all instruments were subjected
to   many    span   calibrations    using    known
concentrations of the gases measured.   The B & K
1302 was configured with filters to measure Freon
12,   halothane,   isoflurane,   ethylene   oxide,
nitrous oxide,  carbon dioxide, and  water vapor.
Before use,  the B & K  1302 was  zero  calibrated
and humidity interference calibrated according to
manufacturers recommendations.  The 1302 was then
subjected   to   a   single   point   span   and
cross-interference calibration  for each  gas  or
vapor to be  measured  before each series of sample
measurements  were made.    The Mi ran  1B2  user
library  parameters   for  nitrous  oxide, EtO,  or
isoflurane  were  used to set up  the  instrument.
To  optimize accuracy of  the  1B2  and 103  when
monitoring  EtO  or nitrous  oxide,  a  five-point
span   calibration   was   performed   using   a
closed-loop calibration system.  Pre-calibration
data  stored  in  the  user library of the  1B2  was
used  for  measuring  isoflurane  concentrations in
spiked  samples.   Calibration of  Photovac  and
Summit portable  GCs  was done  through  microliter
injections  of known  concentrations.   Throughout
the  surveys,  considerable time  and effort  was
devoted  to  calibrating  and verifying  instrument
accuracy  through testing of  prepared  standards.
Nitrous  Oxide,   EtO and Freon  standards  were
prepared from dilutions of pure gases mixed with
clean air or nitrogen  in gas  sampling bags.   A
purchased  cylinder   containing  9.8 ppm  EtO  in
nitrogen was also used.   Halogenated  anesthetic
                                                   756

-------
standards were  prepared from  liquid  anesthetic
agents  supplied  by   the   hospitals   surveyed.
Measured amounts  of  liquids were  injected  into
gas sampling bags and mixed with mete red volumes
of  clean  air or  nitrogen  to  prepare  standards
that   were    then  diluted   to   the   desired
concentrations.   A mixed Standard of 9.8 ppm EtO
and  30 ppm  Freon  12  was  also   used  to  test
instrument accuracy  when  measuring  EtO in  the
presence of Freon 12.

Using   the   appropriate   concentrations    for
performing  span calibrations  was critical  for
obtaining accurate  results  when   measuring  EtO
with   the   1302.     Freon  interference   over
compensated EtO readings when the instrument was
span calibrated  with a 1 ppm  EtO standard.  B & K
recommends using a span calibration  standard of
at least 100  times the detection limit, which for
EtO is 0.2 ppm.    When  the  1302 was recalibrated
using  a 20  ppm  EtO standard,  over compensation
effects were  eliminated.   When calibrating  for
analytes that are (^ compensated, room air could
not be used for preparing standards because of CO,
build-up in indoor air.  Although the  Miran 1B2
can measure both isoflurane and nitrous oxide, to
switch  from  one  gas  to  the  other  required
time-consuming rezeroing of the Instrument.   The
relatively large  volume of air  sample  required
for analysis  by  the Miran 103 and  1B2 (about  20L)
permitted only one measurement  from each sample
bag.   It  was  therefore not  possible  to  measure
both  halogenated anesthetic  and   nitrous  oxide
concentrations from the  same  bag sample using the
Miran  1B2 or 103.  To allow both the 103 and 1B2
to obtain a reading from the  same sample bag,  a
tube  from  the  sampling outlet of  the  1B2  was
connected to  the  inlet  of  the 103.   Of  all  the
instruments evaluated, the  only instrument tested
that could make  simultaneous measurements of  more
than one gas  or vapor  from the same  sample  bag
was the 1302.   Neither  the  Photovac  nor Summit
GCs   would    respond    to   samples   containing
halogenated anesthetics, and the Summit GC  would
not detect nitrous oxide.  Difficulty identifying
the EtO peak  detected  on the Photovac GC rendered
all EtO  readings from  this  instrument  invalid.
An interference  peak from Freon 12 or some  other
source  made  quantitative  analysis of  low-level
EtO concentrations difficult with the  Summit GC.

Measurement Results

Nitrous Oxide
All the  instruments  used  for  measuring  nitrous
oxide, which included the B & K 1302,  Miran  1B2,
Miran   103,   and  Photovac  GC,   gave   similar
readings.  For nitrous oxide concentrations  above
10 ppm,  instrument  responses relative to  Miran
103 readings  were within ±  5% for all  instruments
used.   At  concentrations below 10 ppm,  nitrous
oxide  readings  on  the  1302  averaged  1.82  times
higher  than  Miran   103   readings.     Photovac
readings were within ± 25%  of  readings obtained
on the Miran 103 for concentrations ranging from
5-110 ppm.  The Mi ran 1B2 did not detect nitrous
oxide   in   sample   bags    containing   Miran
103-detectable  concentrations  of  less than  10
ppm.

Halogenated Anesthetics
Laboratory results  from  nine Isoflurane samples
collected during surgical procedures ranged from
0.09 to 0.95 ppm.   B & K 1302  results from gas
bag samples collected side-by-side with charcoal
tube samples were within ±0.3 ppm of the lab
results.   The  average  relative  response of the
B &  K   1302   when  compared   to   the  average
laboratory results was 0.98.  B & K 1302 readings
from  two  sample   bags  spiked  with  isoflurane
averaged 0.16 ppm  higher than laboratory analysis
of  those  same  samples.    Miran  1B2  readings
average  0.23  ppm  lower  than  the  laboratory
results.  More comparisons made from two gas-bag
collected  samples  on  a  follow-up  survey  at
another  hospital   showed  B  &  K  13O2  readings
averaging  0.13  ppm  lower  than the  laboratory
results.  Less satisfactory results were obtained
for halothane when comparing B & K 1302 readings
with laboratory results.   The  average response
from the B &  K 1302  analysis of three sample bags
was 3.8 times lower than the laboratory results.
Laboratory results  ranging  from 1.1 to 1.5 ppm
halothane ranged from 0.3  to 0.5 ppm on the B & K
1302.   No other  direct  readings for  halothane
were   measured   or   detected   on   the   other
instruments.

EtO/Freon 12
The Miran  1B2 and  103 EtO readings  from gas bag
samples collected near operating gas sterilizers
were consistently  higher than the  EtO readings
from the  B & K  1302, Summit GC,  and analytical
laboratory.  Of the nine  bag samples collected,
only four were subjected to follow-up laboratory
analyses.  Laboratory results for one of the four
samples was  4.9 ppm EtO.  Direct  readings  from
analyses  of  this  sample  were  4.2  ppm for  the
B & K 1302, 5.3  ppm for the Summit GC,  11 ppm for
the Mi ran  1B2,  and 9.1  ppm for the Miran  103.
The other three laboratory analyzed samples were
compared  only  with  the  1302  and  the  Summit.
Average response to EtO  for  the  1302  and  Summit
relative  to  laboratory  results  was 1.2 and 1.1
respectively.   The  1302 gave  a  1.3  relative
response to Freon 12 when compared to laboratory
results for four samples ranging in concentration
from about 1 to 80 ppm.

Concurrent monitoring of EtO and Freon 12  during
real-time   measurements   and   analyses    from
collected  bag    samples   showed   considerable
EtO/Freon  ratio  variations  in  both  analytical
laboratory results  and  B  & K  1302  readings.
Although   an    88/12    mixture    contains   a
volume-to-volume ratio of 73%  Freon 12 and 27%
                                                  757

-------
EtO, only two of 23 samples tested came close to
this  ratio.     Most   samples  showed  the  Freon
component well above the expected 73% level.  In
two of the samples the EtO component was greater
than 70% of the total mixture.

Conclusi ons
Any   of   the   instruments   tested  will   give
satisfactory  performance  for  monitoring nitrous
oxide.   Until  additional  field testing  shows
consistent   accuracy   and  comparability   with
laboratory results,  direct reading  instruments
may  not  yet  be  suitable  for  monitoring  all
halogenated  anesthetic  gases.    Both  the B  & K
1302 and Summit GC gave satisfactory performance
for the  monitoring of both short term and  long
term exposures  to  EtO.  The Photovac GC has been
shown  to  give  satisfactory  performance   for
monitoring EtO,'5' but unknown operational problems
caused the instrument  to  fail  during  this  field
testing.  Freon 12 interferences from  the  88/12
sterilization gas will likely give false positive
EtO  readings or  readings  with a high positive
bias on  IR spectrometers  like  the Mi ran 1B2 and
103.   The considerable  variation  in  EtO/Freon
ratios  noted  from various  locations  near  and
during the operation of gas sterilizer equipment
should prohibit the use of  88/12 sterilization
gas as a calibration  standard.
References
1.  National  Institute  for Occupational  Safety
    and  Health.    NIOSH  Manual  of  Analytical
    Methods,   3rd.   Ed..    (with   supplements).
    (NIOSH Publication No. 84-100).  Cincinnati,
    OH: 1984.  (Method 1003).

2.  National  Institute  for Occupational  Safety
    and  Health.    NIOSH  Manual  of  Analytical
    Methods,   3rd.   Ed.,    (with   supplements).
    (NIOSH Publication No. 84-100).  Cincinnati,
    OH: 1984.  (Method 1614).

3.  National  Institute  for Occupational  Safety
    and  Health,    NIOSH  Manual  of  Analytical
    Methods,   3rd.   Ed.,    (with   supplements).
    {NIOSH Publication No. 84-100).  Cincinnati,
    OH: 1984.  (Method 1018).

4.  National  Institute  for Occupational  Safety
    and  Health.    NIOSH  Manual  of  Analytical
    Methods,   3rd.   Ed.,    (with   supplements).
    (NIOSH Publication No. 84-100).  Cincinnati.
    OH: 1984.  (Method 6600).

5.  Cummins,   Kevin,  G.E.   Burroughs,   Julie
    Tremblay.    Field  Comparison  of  Sampling
    Methods  for  Ethylene  Oxide.   (manuscript in
    preparation).
                                                   758

-------
         On-Site and On-Line Spectroscopic Monitoring of Toxic Metal Ions Using Ultraviolet
         Absorption Spectrometry
                      Dr. Kenneth J. Schlager             Bernard J. Beemster
                      Biotronics Technologies, Inc.         Beemster & Associates
                      12020 W. Ripley Ave.               10062 N. Sunnycrest
                      Wauwatosa, Wl 53226               Mequon, Wl 53092
                      (414)475-7653                     (414)242-9101
                              I. The need for on-line monitoring of heavy metals

Heavy metals are common by-products in Industrial operations and can thus enter the environment from wastewater
discharges or  as leachate from industrial  wastes.[1]  Wastewater discharges from industrial operations must be
periodically tested for compliance with permit requirements, including limits for several heavy metals.[2] Groundwater
and drinking water quality testing also includes measurement of several heavy metals.[3,4]

Although Atomic Absorption Spectrometry Is the standard method of analysis required for compliance reporting, it is
not a method that is easily  adapted for  on-line monitoring in factory or field screening applications. Reliable and
affordable methods are needed to detect and measure specific heavy metals In multi-constituent effluents and to detect
specific heavy  metals in surface or ground waters.

                                     II.  Detection of absorption spectra

Heavy  metals  tend  to form anions that bond with water molecules into compounds known as ligands. These
compounds contain bond structures where  electrons can become excited upon exposure to electromagnetic energy
of a specific frequency,  resulting in absorption of light in the ultraviolet-visible wavelength range (200 nm to 800
nm).[5] Chemical analysis of liquids using uv-vis absorption spectra does not rely upon detection of a single peak
wavelength as  with other forms of spectroscopy, but Instead makes use of an absorption signature across a range of
wavelengths. This signature is a function of all absorbing components in the solution. Special apparatus and techniques
are required to detect the spectra and interpret the information.

                          III. Apparatus required for detection of heavy metal spectra

Absorption spectra attributable to individual elements can be observed by  recording the signature for the element
dissolved  in a transparent solvent, such  as  pure water. Spectra for metals such as chromium, copper, iron, mercury
and zinc have been recorded for the applications discussed in this paper. Figure 1 represents the spectra for several
concentrations of iron, ranging from 0.1  to  2.0 ppm (the actual spectra being mathematical values for absorption at
numerous wavelength intervals). It is possible to characterize an unknown substance In pure water as  iron if the
absorption signature matches the pattern observed for Iron. Furthermore, it is possible to estimate the concentration
of iron  by comparing the Intensity of the signature for unknown concentrations to the relative intensity of the signatures
for known concentrations.
                                                     759

-------
The apparatus required to perform absorption spectroscopy in the laboratory is well known.[6] Basic elements include
a light source for the wavelengths of interest, a transparent cell to hold the sample, a detector to measure the light
remaining after transmission through the sample, and a means to process analysis models for interpretation of the
detected information. Many simple laboratory analyzers such as colorimeters use an optical system that is limited to
one or a few specific wavelengths, which limit the instrument to detection of a specific substance. Other laboratory
instruments have a wider wavelength range, but only look at one or a few wavelengths at a time, requiring mechanical
adjustment to the optics  in order to step  through  a wide range of wavelengths. These  instruments are slow and
unsuited for  use outside of the laboratory.
ABSORB














1






I
11 ^
|
r
\




^V— -—

.
"> •••
• v'



!.0 pp. fa
X
1.0 	 • ft
0.5 mm re
0 . 1 1>J"» Fo
,.^' 	 	 --,





\^







:-:^









                                          130       L8Q       230
                                                      ELEMENTS
330
                                  Figure 1. Iron in pure water.
                          IV. Technology advances for on-line absorption spectroscopy

On-line spectroscopy for field use must be able to rapidly detect a wide wavelength range in a flowing sample or in
a dynamic environment. Several recent technology advances make this possible:

       FIBER  OPTICS make it possible for there to be distance between the analyzer and the liquid to be
       analyzed, with the light source and detector remaining in the analyzer. Transmission of light through
       the liquid occurs  in a device known as an optrode, which may be immersed in a process tank or flow
       stream, or may be designed to permit a sample line to flow through a special optical cell.

       ARRAY DETECTORS contain a series of photodiodes, each connected to its own storage capacitor. Each
       element in the detector is responsible for a specific wavelength interval, with as many as 1024 intervals
       possible in the most advanced version.  A fixed grating is used to separate the detected light  into
       wavelength intervals and to project to light onto the detector. The system  used for ultraviolet-visible
       absorption spectroscopy (UVAS) can simultaneously scan 1024 intervals from 200 nm to 800 nm.

       CHEMOMETRICS  is the name collectively given to the statistical and mathematical models used for
       chemical analysis of multi-component liquids. These models make it possible to perform qualitative and
       quantitative analysis by establishing the contribution that an individual chemical constituent makes to
       the overall absorption spectra of the liquid.

                                 V. Chemometric analysis of absorption spectra

Heavy metals must often be analyzed in waters that contain numerous components,  resulting in overlapping or closely
grouped spectra.  The overall absorption spectra for the  liquid is a smooth pattern that results from the effects of
absorption by these individual components. There are three basic steps involved in the  process of using absorption
spectra for chemical analysis:
                                                        760

-------
       QUANTIFICATION involves converting detected spectra for calibration solutions and unknowns into
       numerical values that can be processed using mathematical and statistical procedures.

       PREPROCESSING of raw data reduces the effects of noise and transforms absorption information into
       forms that permit more efficient analysis.

       ANALYSIS of absorption  values identifies individual components and calculates an estimate of their
       concentrations in the liquid.

These three steps are the result of a process that is performed at the beginning of a monitoring project to select the
combination of wavelengths, preprocessing techniques and analysis models that  are capable of providing the most
accurate analysis of the analytes of interest in a specific application. This process uses information from several site
specific samples that contain known concentrations of the target analytes. These samples, known  as a "learning set"
are used to perform a parallel calculations using combinations of techniques to find the model that produces the lowest
error when actual and predicted values are compared. Several "test sets" are then processed to verify the model.

The quantification step is fairly straightforward. Absorption of light is governed by Beer's Law, which  relates absorption
to the absorptivity of the media, path length through the media, and concentration of the absorbing components within
the solution. When all of the absorbing components in the media are known, total absorption at each wavelength is a
function of the sums of all of the absorbing components. A series of simultaneous equations can be used to calculate
absorption. Most often, however, all of the absorbing components are not known, in which case an  inverse technique
that defines concentration as a function of absorbance must be used.[7]

Preprocessing of spectra is often done for multi-component solutions or to adjust for noise or drift. Typical techniques
include the use of first or second derivatives of the absorption  spectrum, the use of Fourier or Walsh transformations,
and the use of Principal  Components Analysis (PCA). PCA uses  statistically determined quantities to  rotate the
coordinate system such that the  original information that may have been  aligned on several axes becomes aligned on
only a few axes. In effect, the variables that are highly correlated with one another can be treated as a single variable,
thus simplifying the analys!s.[8,9]

The analysis techniques currently used  include multiple linear regressions (using least  squares techniques) and
discriminant analysis. Discriminant analysis is a clustering process which defines linear decision boundaries between
information clusters for known concentrations of analytes, and assigns unknowns to an appropriate cluster based upon
detection of significant characteristics for the unknown.[10]

Emerging techniques for analysis include experimental methods such as inductive learning and neural networks,
especially for problems that cannot be simplified through principal components analysis. A technique that shows great
promise is the Lattice-K Nearest Neighbor technique, where known values for variables are organized into the nodes
of a lattice. Predicted values for an unknown are based upon relative distances of variables for the unknown with those
of the nearest neighbors in the lattice.

                         VI. Application of Chemometrics for Analysis of Heavy Metals

Several recent applications have demonstrated the ability of ultraviolet-visible absorption spectroscopy (UVAS) to detect
various heavy metals in multi-component solutions.

Industrial process (boiler) water was analyzed for the presence of iron and copper. Copper was detected over a range
of 1.0 to 5.0 ppm with an  error of 0.047 ppm, while iron was detected over a range of 0.5  to 10.0  ppm with  an error
of 0.014 ppm. These were the lowest errors achieved, using Walsh transformations and discriminant analysis.

Iron was analyzed over a range of 0.0 to  10.0 ppm  in a complex nutrient solution  containing random concentrations
of copper, nitrates, phosphates, calcium, magnesium, sodium, chlorides and other compounds.  Figure 2 shows several
spectra for iron in the nutrient solutions. Figure 3 plots actual versus predicted iron values for 20 samples, using linear
regression of untransformed absorbance values which produced an error of less than 0.03 ppm.  Nitrates were also
successfully analyzed for this application.
                                                      761

-------
    inf:  WCSnR  Fa solutions
    i.eeee
                                 mm pith length      8 to 18 pp«
                                                                  3/1/98
                                188.8      isa.a      see

Figure 2. Iron in nutrient solutions.
                                                                                               PREDICTED VALUE. PPM F«
                                                                                   Figure 3. Actual vs. predicted iron values.
Other applications to date include trace levels of mercury in wastewater (range: 0.0001 to 0.01 ppm), molybdate in
cooling water (range:  1.0 to 2.2  ppm), zinc in wastewater (range: 0.85 to 3.65 ppm), and chromium in wastewater
(range: 0.85 to 4.45 ppm).

                                                     VII. Conclusion

Ultraviolet-visible absorption spectroscopy (UVAS) is an emerging technology that is currently being demonstrated for
on-line analysis of heavy metals and other chemical substances to monitor water quality in complex multi-component
solutions without the need to chemically alter samples prior to analysis.
 REFRENCES

 [1] Lund, Herbert, editor, "Industrial Pollution Control
    Handbook", McGraw-Hill, Inc., New York, 1971

 [2] Clean Water Act, 40 CFR 121 to 135 and 403

 [3] Water Pollution Control Federation, "Understanding Hazardous
    Waste Management", Operations Forum, Volume 8, No. 1,
    January 1991, pages 22-23

 [4] "National Drinking Water Regulations: A Summary of the
    Latest Listings", Water Technology, Volume 13, Number 10,
    October 1990, pages 30-32

 [5] Thompson, Clifton, "Absorption of Radiation",
    Ultraviolet-Visible Absorption Spectroscopy, Willard Grant
    Press, Boston, Mass., 1974, pages 17-27

 [6] Thompson, Clifton, "Instrumentation", ibid, pages 29-45

 [7] Thompson, Clifton, "Spectroscoplc Applications", ibid,
    pages 47-64

 [8] Jolliffe, IT., "Principal Component Analysis",
    Springer-Verlag, New York, 1986

 [9] Zupan, Jure. "Transformations". Algorithms for Chemists,
    John Wiley & Sons, Cichester, 1989, pages 87-142

 [10] Goldstein, Matthew, et al, "Discrete Discriminant Analysis",
    John Wiley & Sons, New York, 1978
                                                                762

-------
            RAPID SCREENING OF SOIL SAMPLES FOR CHLORINATED ORGANIC COMPOUNDS
                                  H. Schlesing, N. Darskus, C. Von Hoist, R. Wallon
                  Biocontrol Institut for Chemische und Biologische Untersuchungen Ingelheim GMBH
                                                  West Germany
Cleanup of an industrial site contaminated with chlorinated
organic compounds requires methods for the rapid assess-
ment of many soil samples. An estimate of soil content of
chlorobenzenes, chlorophenols, and hexachlorocyclohexanes
is the EOX value. This determination requires solvent
extraction of the soil, which generally takes at least 2 h and
is therefore too lengthy for the present purpose. In the present
work we have compared this method with the following more
rapid ones:
  •Thermal desorption of organic compounds from soil,
   followed by combustion in an oxygen atmosphere
   (Organochlortest A-P-E; supplier: Burger)
  • Measurement in the headspace over a soil sample with a
   photoionisation detector (supplier: TIS) test kit based on
   extraction and reduction of chlorinated compounds from
   soil ("Chlor-N-Soil"; supplier: Dexsil)
The main characteristics of these four methods were compared.
Nine soils of different type and degree of contamination
were examined with  the results.
Our provisional method of thermal desorption, which is
still under development, almost always yields higher values
than the EOX method, even though the former have been
corrected for ionic chloride in the soil. Possibly thermal
desorption is more efficient than soxhlet extraction for the
compounds in question. However, except for sample 92/03,
both methods yield the same relative order for the degree of
contamination. This result suggests that the thermal desorp-
tion method merits  further development.
No such correlation was obtained for the PID. For the Chlor-
N-Soil test kit, results were obtained for only three samples
because of limited availability of reagent sets. In principle,
this test appears to be applicable within the limited scope of
its specification, but a correlation of the colour change
(violet: "little", yellow-brown: "strong") with approximate
contamination has yet to be established.
                                                       763

-------
          DEVEDDIMENT OF A MZCRDBORE CAPIIIARY COLUMN GC-FOCAL PIANE MASS SPECTROGRAPH
                         WITH AN ARRAY DETECTOR FOR FHU) MEASUREMENTS
                                          M. P. Sinha

                                   Jet Propulsion laboratory
                               California Institute of Technology
                                       Pasadena,  CA 91109
A gas chzomatograph-mass  spectrograph (GO-MS)
system using a microbore  capillary colunn (50
fan i.d.), and a miniaturized focal plane mass
spectrograph  (Mattauch-Herzog  type)  with  an
array detector has been  developed.  The ex-
tremely, snail carrier gas flow rate (0.05 atm
cmmin   of helium) through the colunn permits
its direct coupling to the ion source,  and
reduces the pumping needs of the MS.  The mass
spectrograph with  an array detector  measures
the intensities of all  masses  simultaneously.
Analysis of mixtures of compounds, each  at a
concentration of  1  ppmv has been performed with
high signal-to-noise ratio.   The Tnirmaim de-
tectable quantity  of benzene is  determined to
be 7.5 x 10   g which corresponds to a concen-
tration of 40 ppb for an injected sample volume
of 0.5 /il.  Lower analyte concentration can be
determined  by increasing the  sample  volume
and/or the signal integration time.  The system
is .found  to have  a linear  dynamic range  of
>10 .   Because of  its  low weight,  power,  and
high sensitivity,  the combination of  a  micro-
bore GC column and a miniaturized  plane mass
spectrograph  is uniquely  suited  for  field
analysis.
INTRODUCTION

The combination of a gas  chromatograph with a
mass spectrometer  (GC-MS)  is one of the  most
powerful  instruments  for  the  analysis  for
complex mixtures.   GC-MS is eminently suited
for  the  measurement  of  environmental  pol-
lutants.  However, in its present  form it has
remained  largely  confined  to the  laboratory
because of  its mass  and power  requirements.
Our own interest lies in  the development  of a
field-portable  GC-MS  instrument.    Such  an
instrument is much needed for the real-time,
on-site measurement  of pollutants,  e.g.,  at
toxic waste dump sites and for fugitive emis-
sions from various sources.   This instrument
should also be fast and possess high efficien-
cy and  sensitivity  in order to  analyze com-
pounds  present at lew concentration  levels.
In  the hyphenated  technique of GC-MS,  the
speed of  analysis  is determined  by  the  GC
separation time.   Fast separation with high
efficiency can be achieved by  the use  of a
narrow-bore  capillary column  (e.g.,  50  fan
i.d.) of short length.      Also, the carrier
gas  flow rate through such a column  is very
low which offer the advantage of reducing the
pump-size  (often requiring large  mass  and
power) needed to maintain the proper operating
vacuum conditions in the MS.

Such microbore columns, however, put important
restrictions on the sample size for analysis,
and  on  the detector:  used for measuring the
eluted  compounds.  '    Extremely narrow and
closed spaced peaks are produced from the use
of  microbore  columns,  particularly  in  the
early part of the chromatogram.   The detector
must, therefore, have a high  sensitivity and
a low-time  constant for signal  measurement.
To maintain  the column efficiency, the dead
volume needs to be minimized.  These consider-
ations  have  prohibited  the  application  of
columns of  <100 /on i.d.  in commercial  GCs.
The fast rate  of data acquisitions needed to
measure peaks from a microbore column makes it
incompatible with a scanning type mass spec-
trometer.

The aforementioned problems in exploiting the
advantages of a microbore column can be over-
come by the use of a mass spectrograph (non-
scanning) .   The capability of a  mass spectro-
graph for  measuring the  intensities  of all
masses  at  the  same time  confers on it  an
almost  unlimited speed  for  obtaining  mass
                                               765

-------
spectra.   Its sensitivity also  is inherently
greater that of a scanning-type MS because the
latter measures the signal at a given mass peak
only for a short dwell time.  However,  in the
past, the lack of a sensitive ion detector has
been an important reason  for not using a nonsc-
anning MS for measurements that required high
sensitivity.  Recently, an array detector known
as an electro-optical ion detector (EOID)  has
been developed  in  our laboratory  for a focal
plane   mass   spectrograph   (Mattauch-Herzog
type). '   The BOID possesses the simultaneity
of a photoplate (used in focal plane MS)  and
the high gain of an electron-multiplier.   The
EOID can integrate  signals continuously for a
wide range of time  (25 ms - 30 s)  and,  by an
appropriate  selection  of  integration  time,
multiple mass spectra from transient  samples
(like a narrow GC peak) can be obtained without
sacrificing sensitivity.

Our approach towards the development of a high
performance  field-portable  GC-MS  instrument
consists of combining a short microbore  column
and a  miniaturized  focal plane mass  spectro-
graph.   In this paper,  the  new GC-MS  system
developed in our laboratory is described.  Some
of the results obtained on this system for the
analysis of  a mixture of priority pollutants
are also reported.
 II.   EXPERIMENTAL
                                                  B.
                                                  Mass Soectroaraoh
A.
Gas Chromatooranh
The experimental arrangement is shown schemati-
cally in Fig. 1. The fused silica microbore GC
column (3.0 m, 50 /an i.d.) with a 0.2 /an bonded
DB-5  stationary phase  (J.  & W.  Scientific,
Folsom, Ca.)  was housed in a temperature pro-
grammable oven.   The outlet end of the column
was directly led into the ion source  of that
mass  spectrograph.    A  sample injector valve
(Valco Instruments) with an internal volume of
0.5 nl was used to inject the sample onto the
column.  A pneumatic actuator along withjpilot
valves  and  a  digital  valve interface   was
incorporated  into the sample injector for fast
injection.    Samples could  thus be  injected
reprcducibly  in less than  14 ms.    GC-grade
helium was_used as  a carrier gas at a flow rate
of 40 on s  .  Because of the small volume flow
rate of the carrier gas (0.05 atm on  min  ),
it was possible to connect the GC  column and
the MS without any  interface.  The direct inlet
of the  column effluents  into the  ion source
eliminated the dead volume  that  usually arise
from  GC-MS   interfaces  and  allowed  for  the
complete utilization of the analyte sample.
                                                  Two miniaturized  focal plane mass  spectro-
                                                  graphs, one with 2.0" long focal plane and the
                                                  other with a 5.0" long  focal  plane have been
                                                  designed and  fabricated at JPL.   The  2.0"
                                                  focal plane covering a  mass range  of  40-250
                                                  amu is destined to be used for field measure-
                                                  ments.  A photograph of this MS is  shown in
                                                  Fig. 2a. The magnetic sector of this analyzer
                                                  was fabricated  from  new magnetic  materials
                                                  having  high energy  product  value,  and  high
                                                  magnetic flux permeability for reducing the
                                                  mass of this sector. The 5.0" focal plane MS
                                                  covers a mass  range  of  28-500 amu.
                                                  C.
                                                  Array Ion Detector
The  details of the  EOID have  been reported
previously. '   In short,  it  consists of  a
microchannel  electron  multiplier  array,  a
phosphor-coated (P-31) fiber optic window, and
a photodiode array (PDA). In the BOID, an ion
exiting the magnet impinges on the microchan-
nel array and initiates an electron cascading
process along the channel  length.  The elec-
trons coining out at the other end  of the chan-
nels  produce photon  images of their parent
ions  on  the phosphor window  (shown  in Fig.
2b).  The intensities of these images are then
measured by the photodiode array (2.0" long
active  region)   having a  center-to-center
distance of 25 /on between its two adjacent
diodes.

The photodiodes are integrating detectors and
accumulate the photon signal (proportional to
the  ion  signal)  for the  desired period of
integration.  The position of  the photodiode
along the focal plane determines  the mass of
the ions producing the ion  image at that loca-
tion.  The  signal stored  in the  photodiodes
are read (at a rate of 220 kHz) serially by a
computer  after a predetermined  integration
time.  Each readout,  called a frame, provides
a mass spectrum of all the ions  accumulated
during the  integration period.   Each diode
accumulates the signal continuously except for
its read-out time ("4 jus) when it  is reset and
resumes signal integration.  This allows for
the complete mass spectral measurement of GC
effluents at a high frequency without any loss
of sensitivity in the process.

Both of the  mass spectrographs described above
are equipped with their own array detectors.
The  computer interface electronics for  the
small MS has not  been cotpleted at this time
and, therefore,  the  results reported  in the
paper were obtained on the 5.0 in. focal plane
MS.  For  laboratory measurements, this did not
create any complications and demonstrated the
analytical capability of the MS-EOID system.
                                                766

-------
Moreover, it is expected that the new 2.0-in.
array detector  will have  better performance
because  of the minimization  of  the  signal
losses at the PDA-fiber-optic window interface
in this design.

A mixture having a concentration of  1 ppmv in
air of each of the compounds listed in Table 1
was  prepared.    The internal  volume of  the
injector valve was filled with this mixture and
injected on the GC column for analysis.
RESULTS AND DISCUSSIONS

The  mass chromatogram  of  a mixture  of  the
compounds listed in Table 1  is shown in Fig. 3.
Each component in the mixture had a concentra-
tion of 1 ppmv  in air.    The  GC column  was
maintained at the room temperature and a signal
integration time of 250 ms for the array detec-
tor was used in the measurement.  Complete mass
spectra of the components eluting into the ions
were recorded every  250 ms. In obtaining the
mass chromatograph, the sum of the intensities
of all masses (>45 aim)  in each record (frame)
is  plotted  against  the corresponding  frame
member  (time).
The  chromatogram  shows  that  the  components
(dichlorodifluoromethane, chloromethane, bromo-
methane and chloromethane)  correspond to peaks
2-5 are narrow and closed spaced.  For example,
the peak-to-peak separation between 2 and 3 is
less than 700 ms  and  the full width of peak 2
is about 300  ms.   Quantitative  measurement of
such GC peaks are made possible by the simul-
taneous  measurement  of  all ions  and by  the
proper  selection  of  the  signal  integration
time.

The continuous measurement  by the EOID with a
short integration time (>25 ms)  can be used to
perform time-resolved mass spectral measurement
and can be applied to resolve otherwise over-
lapping GC peaks.  Figure  4  demonstrates the
effect  of  measurement time on resolution of
compounds by the microbore column.  It is seen
in Fig. 4a that bromomethane and chloromethane
corresponding to  frame numbers  89  and  95,
respectively, are well separated when  an in-
tegration time of 100 ms is  used for their mass
spectral measurement.   For  250  ms integration
time, the chromatcgraphic separation is barely
adequate (Fig. 4b) but  the  separation is lost
when spectral measurements  are  made every 500
ms (Fig. 4c). The time resolution capabilities
of the MS-BOID make it particularly useful for
short  columns  of moderate resolving  power.
Their combination reduces the analysis time and
renders it suitable for a field-portable GC-MS
analyzer.

It should be noted that the quantitative nature
of  measurement  is not  conpromised  by  the
number of mass spectra (frames)  obtained from
a GC peak because of the continuous and simul-
taneous  measurements  of   ion   intensities.
Figure 5  shows that some of  the intensities
contained  in all  the frames of  a GC  peak
(corresponding to dichlorodifluoromethane) is
independent of the integration time used in
recording these  frames.  The  sum of intensi-
ties determines the amount of the compound.

The  mass  chromatogram (Fig.  3)  demonstrates
that  this GC-MS system  can  readily  analyze
mixtures  of  compounds present at the  1 ppmv
level without  preconcentration of the analyt-
ical  sample.   From  these data, the  minimum
detectable quantity  (MDQ) was calculated for
each compounds.  For benzene  this amounts to
7.5 -10   g, which corresponds to a concentra-
tion of 40 ppb for an injected  volume of 0.5
/il (results of 100  ppb mixtures of benzene and
chloroform are included in Fig. 6).   lower
analyte concentrations (<40 ppb)  can be deter-
mined by  increasing  the  sample  volume and/or
the signal integration time.  However, larger
volumes  (>2  /il) cannot  be injected  without
degrading column resolution.  The problem can
be  overcome  by  sweeping the sample  from an
injector valve and cryofocusing the volatile
organic compounds  at the head of the column,
thus, removing the air.   The temperature of
the  column can then  be programmed for subse-
quent analysis.

A series of mixtures  of chloroform and benzene
of various concentrations (0.1 - 100 ppmv) in
air  was  prepared to study  the  dependence of
mass  spectral  intensity  on  concentration.
These  mixtures  were  injected   onto  the  GC
column and their mass spectra were measured.
In  Fig.  6,  the  sum  of the intensities of a
single  mass  (m/z  =  83, characteristics of
chloroform) and also of a group of masses  (76-
78 amu characteristics of benzene) contained
in frames of the respective GC peaks have been
plotted.  The intensity  is found to increase
linearly  with concentration showing a linear
dynamic range of >10. This is the range with
a constant integration time of 250 ms.  It is
possible  to  further  extend the  dynamic range
by  suitably  adjusting the signal integration
time.  The straight lines  in Fig.  6  are the
least square  fit through the  data points.  A
linear-correlation coefficient  equal  to 0.99
is  found  for mass  spectral measurement of
benzene   showing  an  excellent  correlation
between concentration and intensity.
CONCLUSIONS

A GC-MS system using a microbore column  (50 pm
i.d.)  and a  miniaturized mass spectrograph
                                                767

-------
with an array detector has been developed.  The
performance of this system in the analysis of
mixture of priority pollutants has been demon-
strated.  A short microbore column (50 /an i.d.,
3.0 in. long), when combined with the MS-EOID,
resolves the early eluted gases satisfactorily.
The GC-MS system described above possesses hioh
sensitivity and a linear dynamic range of >10  .
The  minimum  detectable   quantity   (MDQ)  for
benzene is found to be 7.5 x 10   g which cor-
responds  to a  concentration  of  40  ppmv  in  a
sample volume of 0.5 /Ltl.   larger  sample volume
can allow measurement of  lower concentrations.
The  combination of a  microbore  column  and  a
miniaturized focal plane MS is eminently suited
for  field measurements.   The  extremely small
carrier gas  flow rate drastically reduces the
mass and  power needs of the mass  spectrograph.

ACKNOWLEDGMENTS

The work  described  in this paper  was performed
at the  Jet Propulsion  Laboratory,  California
Institute of  Technology  and was  supported in
part  by  the  U.  S. Environmental  Protection
Agency (Grant No. R-814410-0-01-0).

REFERENCES

1.    Guiochon,  G.  Anal. Chem., 50,  (1978)
      1812.

2.    Schutes, C. P. M., Vermeen, E. A.  Rijks,
      J.   A.,  and Cramers,  C.  A. "High  Speed
      Profiling of Complex Mixtures by Means of
      Gas Chromatography  in Narrow Bore Capil-
      lary Columns"  in  Proceedings  of the 4th
      Symposium  on Capillary  Chromatoqraphy.
      Kaiser, R. E.  (Ed.), Institute of Chroma-
      tography, Bad Durkhein,  Germany, p.  687.

3.    Trehy, M. L.  Yost,  R. A.  and  Dorsey,  J.
      G.,  Anal. Chem.,  58, (1986) 14.

4.    Holland, J. F. Enke, C.  G.,  Allison,  J.,
      Stuffs,  J. T.  Pinkston, J.  D., Newcome,
      B.,  and Watson,  J.  T.  Anal.   Chem,  55,
      (1983)  998.

5.    Leclercq, P.  A.  Schutjes,  C.   P. M,  and
      Cramers, C. A., J. Chromatofr. Libr.,  32.
      (1985)  55.

6.    Boettger,  H.   G.,   Giffin,  C.  E.,   and
      Norris,   D.   D.,   "Electro-optical   Ion
      Detector  in  Mass  Spectrometry:  Simul-
      taneous Monitoring of All Ions Over Wide
      Mass  Ranges,"  in   Multichannel   Image
      Detector.  Talmi,  Y.  (Ed.) ACS Symposium
      Series No. 102,  ACS, Washington,  D.  C.
      (1976) p.  292.
      Sinha, M. P., in "Particles in Gases and
      Liquids  II:  Detector, Characterization
      and Control," Mittal, M.  L.  (Ed), Plenum
      Publishing Corp.,  1990  (in  press).

      Harvey, M. C. and Stearns,  D. D., Anal.
      Chem 56  (1984) 837.
Compounds
TABLE 1

     Peak No.  (Fig.
air
dichlorodifluoromethane
chloromethane
bromomethane
chloroethane
dichloromethane
1, 1, 1 - trichloroethane
chloroform
benzene
trichloroethylene
                 1
                 2
                 3
                 4
                 5
                 6
                 7
                 8
                 9
                10
                                                  768

-------
           ELECTRIC
           SECTOR
                        MAGNETIC
                        SECTOR
MICROCHANNEL
ELECTRON
MULTIPLIERS-
                                  PHOSPHOR

                              FIBER OPTICS
                               PHOTODIODE
                    SAMPLE
                    INJECTOR
                    VALVE
                                                ELECTRO-OPTICAL
                                                ION DETECTOR
•MICROBORE
 COLUMN
 (50 urn ID)
                                 WASTE
                                 SAMPLE
                        CARRIER GAS (HELIUM)
Fig. 1   Schematic of the microbore capillary column gas chromatorgraph and the local plane mass
       spectrograph assembly. The sample injector is pneumatically actuated and is provided with
       pilot valves and a digital valve interface for fast sample injection.

-------
                         -ELECTRIC SECTOR  ELECTRQ~OPTICAL ION DETECTOR7
                                                  -MAGNETIC SECTOR /
                                                                   /
                        5*  V *«A
Fig. 2a:   Photograph of the focal plane (2.0-in) mass spectrograph
          with an electro-optical ion detector.
               Fig. 2b:   Photograph of ion images
                                  770

-------
    2100
                    50
100
      150         200

FRAME NUMBER
250
300
Fig. 3:   Total ion chromatogram obtained from a mixture of compounds listed in Table 1. Each component
        in the mixture has a concentration of 1 ppmv. A sample volume of 0.5 \i I was injected, and a
        signal integration time of 250 u.s was used for each frame.
                                         771

-------
  2400
  1600
Z 1200
         (a)
             20
                              60
                          SO
                     40       60

                     FRAME NUMBER
                                          89
                                      80      100
                                                         2000
                                                         1600
                                                      Z  1200
                                                         400 -
                                                                (b)
                                                                                         28
10          20

      FRAME NUMBER
                                                                                           30
                     30
                            2400
                            2000
                         I  1200
                            800
                            400
                                  (c)
                                                   10
                                                9     12

                                               FRAME NUMBER
                                                            15
                                                                 17
                                                            15     18    21
      Fig. 4   Effect of signal integration time on resolution of GC peaks. Integration times of 100,250, and
              500 us were used fora frame in (a), (b), and (c), respectively.  The peaks corresponding to
              dichlorodifluoromethane and chloromethane, and bromomethane and chloroethane are not
              resolved with 500 us integration time.
                                                      772

-------
   2
55
z
LLI
                    I
Fig. 5   Sum of the intensities of the
        frame comprising the last two
        peaks in Figs. 4 a, b, c are
        plotted against their frame
        integration time. The sum is
        found to be independent of the
        integration time.
                   200              400

                     INTEGRATION TIME (ms)
                                                  600
                                                            Fig. 6   The straight line plots show a
                                                                    linear dynamic range of >103
                                                                    Os represent the sum of intensities
                                                                    in various frame of mass 83
                                                                    (characteristic of chloroform) where
                                                                    as As represent the sum of
                                                                    intensities for a group of masses
                                                                    76-78 (characteristic of benzene).
                  CONCENTRATION (ppm)
                                                  773

-------
                 APPLICATION OF A RETENTION INDEX APPROACH USING INTERNAL STANDARDS TO A
             LINEAR REGRESSION MODEL FOR RETENTION TIME WINDOWS IN VOLATILE ORGANIC ANALYSIS
                                                Russell Sloboda
                                   NUS Corporation, Wayne, Pennsylvania 19087
The retention time (RT) of an analyte can be predicted by
measuring  its fractional  distance between the RT of the
internal standard (IS) eluting before the analyte and the RT of
the IS eluting after the analyte.  The development of RT
windows using the retention  index method  involves
calculation of a prediction  interval that is derived using linear
regression statistics.  This approach can also be adapted to
the relative retention time (RRT) method, which  uses one
rather than two internal standards to predict an analyte's RT.
Linear regression equations were derived and software was
developed for linked  operation with a gas chromatography
(GC) data system. RT windows were established to compare
the performance of the RT index method versus the RRT and
external standard techniques.  Data  sets were generated
under a variety of conditions for  purge and trap analysis of
calibration standards  using megabore capillary and  packed
columns with a dual detector system. RT windows generated
using the RT index method were much narrower than those
obtained using the RRT  method in variable-temperature
environments and slightly  narrower than the RRT method in a
controlled-temperature environment.  The RRT and the RT
index methods generated  much narrower  windows than the
external standard method  under all conditions. All methods
were successful in terms of speed of calculation, minimal QC
failures, and few interferences. Because the RT index method
achieves  the  narrowest prediction windows of the three
techniques, it  offers increased specificity  of analyte
identification without changing the probability of missing an
analyte that is present.

                    INTRODUCTION

When GC analysis is performed in a mobile  laboratory,
fluctuations in ambient temperature and  other factors can
cause greater variations  in analyte RTs  than under more
controlled laboratory conditions.  The use  of ISs can improve
the accuracy of RT predictions under such conditions. The RRT
method successfully compensates for RT shifts when the
analyte elutes very close (within ± 20  percent) to the IS but
provides a less accurate estimation of RTs for analytes that
elute farther away.  Whereas the RRT method assumes that an
analyte's  RT will be increased  or decreased in proportion to
the ratio  of the RTs of the IS  in the sample divided by the
standard, often  analyte behavior  is intermediate between a
proportional RT shift and  a constant, additive shift.  The RT
index method offers the  advantage  of compensating for
either type of shift by means of a linear interpolation in the
predicted magnitude of RT  shifts in the region from the IS that
elutes before the analyte to the IS following the analyte.  If
the general 2 parameter linear equation for prediction of RTs
is employed (ISi • A + IS2 • B s predicted RT), then it can be
shown that a range of predicted RTs will occur as the fitting
parameters A and  B are varied, with the predictions based
upon the RRT method using the first IS (B = 0) or the second IS
(A = 0) at either extreme and with the  RT index method
(having the side constraint A + B = 1) yielding a prediction that
is intermediate between the two RRT methods.  In addition,
the RT index method is more practical than other enhanced
prediction techniques.  This is because it is only a one-
parameter method and so  does  not require multiple
calibration runs to compute the prediction  coefficients as is
necessary with the true two-parameter linear model
referenced above.

GC conditions were selected  for  rapid and  simultaneous
separation and identification of 33 analytes using purge and
trap sample preconcentration, DB-624or 1%  SP1000 analytical
columns, and a  PID/ECD dual  detector system.  Three  IS
reference peaks were used for each  detector.  Standard
operating procedures (SOPs) and computer programs were
written so that the width of the RT window for each analyte
was computed as plus or minus the square root of the variance
of the observed minus  the predicted RTs within a set of
standard  analyses, multiplied by the student's t-value using a
99% confidence level. The width of RT windows calculated in
this manner was considered valid for up to 60 days, as long as
GC conditions  remained constant.   The  center  of each
analyte's  prediction window was calculated every day using
the RTs from a single standard run in the 12-hour period prior
to the sample.  Predicted RTs were computed for all three
techniques (external standard,  RRT, and RT  index).  The RRT
technique and the RT index technique were designated as the
primary methods for analyte identification, with the external
standard  technique applicable only to those (rare) situations
where interferences precluded the use of ISs for RT prediction.

The GC system  was operated  in the  Region  3 EPA Field
Investigation Team (FIT) mobile laboratory and also in the FIT
base-support facility.  The effect of ambient temperature
fluctuations was investigated by analyzing a series of
standards and calculating and plotting RT window widths
under  several  different temperature conditions and
environments.  In  addition to demonstrating  the overall
superiority of the RT index method over the  RRT and external
standard  methods,  this investigation provided insight into
several factors that influence RT variance.
                                                       775

-------
                        THEORY

All three RT prediction techniques discussed above utilize RT
windows that are confidence intervals for the predicted minus
the observed values of the retention time. The RT window
width is computed as a 2-tailed 99% confidence interval using
the t-distribution coefficient multiplied by the square root of
the variance of the observed minus the predicted RT:

 (1) RT window = ±t0.995,df  x ^Var(RTobs-RTpred)

The above equation is valid provided that the observed RTs
exhibit constant variance and observed minus predicted RTs
exhibit a normal distribution. These assumptions could not be
strictly verified because of the small data  sets  (typically, n = 6
standards) employed to calculate the  variance in each case;
however, moderate departures from  normality should  still
produce reasonable estimations.

The derivation of the variance of the observed minus the
predicted RT is outlined below for the RT index method.(1)
The variance for the RRT and the external standard methods
can be derived in an analogous  fashion.(2,3)  The RT index
prediction formula is as follows:
 (2)
      Where X = sample, S = standard, C = the analyte, and I,
      and \2 are the bracketing internal standards.

 (3)   Algebraic rearrangement of equation 2 leads to:
                                       bs-hs
 (4)   Introduce the change of variables Z1 and Z2:

       Zi  =
 (5)  Substituting (4) into (3):  Cx = ZiX * 22X
                                             Z2S
 (6)   Equation (5) happens to be the  linear regression
      solution for the linear equation with one calibration
      run only: Cx = ZiX + Z2XA

(7)  To derive the slope of the regression line (A), the sum
     of the square of the errors in the regression (observed
     minus predicted values) is minimized by taking the
     first derivative  and setting this equation  equal  to
     zero. This yields:
                       NP
             A   =
                       NP
                       Z  Z2,2
(8)   To derive the  variance of the observed minus the
     predicted retention time,  it is necessary to derive the
     variance of "A."  This is illustrated in reference  1 and
     yields:
                       Var(Q)
          Var(A) =
                        NP
                        Z
(9)   Var [RTobs-RTpred]    =  Var(Cr Z1X-Z2X A)

(10) Because Z]x and Z2X are treated as constants:

     Var [RTobs-RTpred]    =  Var (Q)  + Z2X2 Var (A)

(11) Substituting equation (8) into (10):

                               Z2X2
     Var [RTobs-RTpred]
                               NP
                                          Var(Q
(12)  From chapter 11, section 11.3 of reference no. 2, the
     unbiased estimate of the variance of the  observed
     (retention time) value about the regression  is related
     to the sum of the residuals from the fitted regression
     by the following equation:
                                                              (13)
(14)
         Var(Cj)=
                     1
                               
-------
     Where:      N

                  I  :(l2i-lii)(2XRTc(l)-ltl-l2,)
        A  =    J±J	
                         N
                        .1  d2,-hi)2


 (15) Because one standard run is used to determine the
     center of the prediction window in a sample,
     equations (1) and (2) can be combined to yield the RT
     index prediction window as follows:
     RT window =  RT|a(spL)
            (RTlb(SPL) - RTla(SPL))


                     - RTla(STD))
                                   Var [RTobs - RTpred]
     Where:
             = RTof IS eluting before Gin sample.
             = RTof IS eluting after C in sample.
             = RTof IS eluting before C in standard.
             = RTof IS eluting after C in standard.
     Var[RTobs-RTpred] = equation (14)

 (!6)  For determination of the RT window width using the
     RRT method, the formula for the variance of the
     observed minus predicted (RT) value can be extracted
     from reference nos. 2 and 3, yielding.
     Var[RTobs-RTpred] = {1 +

             N
      1/(N-1)  I   (RTC(i)-RT|(i)
     Where:
 N

 I
i= 1
                             RTc(i))
     M   =
                 N
     N   =   number of  standards  in set used to
            determine RT window width

(17) The formula for the variance of the observed minus
    predicted (RT) value for the external standard
    method is also found in reference nos. 2 and 3:
Var [RTobs - RTpred] =
    Where:   N
           N              N
          J   [RTC(0-I/N  I
                                             RTc(j)]2
                    =  number of standards used to
                       determine RT window width

                    =  RT of C in standard i
                 EXPERIMENTAL SECTION

 GC Configuration.  A Varian 3300 gas chromatograph was
 equipped with a Tracor photoionization detector (PID) and a
 63Ni electron capture detector (ECD) arranged in series with a
 splitter mounted ahead of the ECD to shunt 97 to 99 percent
 of the PID effluent away from the ECD. The GC was connected
 to a Tekmar LSC 2000 sample concentrator for purge and trap
 analysis of soil or water. Data acquisition and reduction were
 accomplished through a  Nelson  Analytical  PC Integrator
 System that consisted of an AT-compatible computer
 connected to a 10,000 data-point-capacity interface module.

 Computer Programs. Nelson Analytical software (rev. 5.0) was
 used to tentatively identify peaks and for quantitation. In the
 first phase of this project, RT window widths were calculated
 using an HP-15c calculator  program.  This interim procedure
 was used while BASIC  programs were developed for linked
 operation with the GC data system.  The BASIC program
 developed to calculate the width of RT windows reads area/RT
 files created by the Nelson system for several standards  run
 over  a 7- to 12-hour  period,  calculates the width of RT
 windows using all 3  techniques (external standard,  RRT
 method, and RT index method), and outputs these windows to
 a disc file.  (This program also allows pooling of RT windows
 over several 12-hour periods.)  For each sample, the Nelson
 software produces a disc file containing concentration results
 and tentative peak assignments.  Next,  the second BASIC
 program reads this sample file, the method file containing  the
 daily standard RTs, and the RT window width file.  Two
 standard outputs  are produced:   The first consists of
 compounds tentatively identified  as present by the Nelson
 software and tabulates the observed minus  predicted  RTs
 versus the width of  RT windows for all three techniques.  The
 second printout lists all compounds present in the RT window
 file (not just "hits"), and a printout of predicted RTs and low
 and high limits of the  RT prediction window is tabulated  for
 each  compound using  all  three  techniques.  This  BASIC
 program substitutes a minimum RT window width if the width
 of the calculated window is less than a specified value.  (This
 avoids  use of  RT windows that are narrower than the
 prediction error caused  by the step size of the integrator.)  For
 this work, the integrator step size was set to 0.01 minute, and
 the minimum allowable width was set (based upon worst-case
 scenarios) to 0.02 minute. This program also allows a constant
 to be added to the calculated width if two compounds  co-
 elute  in the standard and require a wider window to ensure
 detection of both components.

 Analyte Selection and GC Separation.   EPA Region 3's pre-
 rernedial program required  determination of analytes from
 the EPA Contract Laboratory Program's target compound  list
 (TCL).  All volatile  TCL  analytes  except the four gases were
 included in the SOPs  for  analysis.   GC conditions were
 optimized for the  fastest analysis  times that would  not
 degrade resolution between adjacent analytes.  For  the
 capillary column, helium flow was 7 ml/min. The GC was held
 at 40°C for 7 minutes, followed by a 4°C per minute ramp to
 75°C, followed by a  10°C per minute ramp to 138°C. For the
 packed column, helium flow was 40 ml/min. The GC was held
 at 45°C for 3 minutes, followed by an 8°C per minute ramp to
 215°C.  The ECD split ratio was about 65:1 in all cases.

 Selection of Internal Standards.  Three  IS peaks were desired
 for each column and detector so that most analytes could  be
 bracketed by IS peaks on either side. Based upon trial testing,
a final set of ISs was selected for use on each column. These ISs
are listed in table 1 along with target analyte RTs.
                                                      777

-------
         Table 1: Analyte and IS Retention Times
 IS) 1-Br-3-CI-2-mepropane
 trans-1,2-dichloroethene
Retention Time QC.  QC requirements were developed in SOP
format by FIT for Region 3 EPA's pre-remedial program.(1)
These SOPs required determination of external standard-type
RT windows for IS monitoring each time analyte RT windows
were computed.  The standard run in the 12-hour period prior
to each sample was used in conjunction with sample internal
standard  RTs, to predict the center of the RT window for each
analyte.  A second standard was run at the end of each 12-
hour analysis period and was required to exhibit RTs for all
analytes that were within the windows based upon the prior
standard.  In addition, ISs in all  analytical runs were  required
to fall within RT windows.  If  IS RT window criteria were
exceeded,  the analyst was to  check  for  co-eluting IS
interferences via a decision scheme in the SOP.  Co-elution
problems required data evaluation  using external  standard
quantitation or re-analysis on the second column. Conversely,
re-analysis of the sample on the same column was required if
there was no evidence of IS co-elution problems.

Method Evaluation. Base laboratory testing normally was
done under a controlled temperature environment (±2°C),
although on one occasion, ambient temperatures  were
deliberately varied by 4°C to demonstrate the resulting effect
on RT window width.   In-field testing was conducted inside
the FIT mobile laboratory.   Ambient temperature  control
within the mobile laboratory  was achieved via a central,
ceiling-mounted HVAC unit that  was manually operated.
Monitoring of the ambient temperature surrounding the GC
was accomplished via a thermometer mounted on the side of
the GC, away from all heated zones and GC blower outputs.
The  GC was operated in the mobile laboratory under hot
weather conditions in which the air conditioner had difficulty
achieving temperatures  less than 29°C. Operations were also
conducted under cool weather conditions in which ambient
temperature control was achievable but,  because of the
manual heater control, variable over a 21° to 25°C range over
the course of a typical day.

                RESULTS AND DISCUSSION

Table  2 indicates the various conditions under which RT
windows were generated and references the figures that
depict associated RT window performance.  Each of  these
graphs illustrates the plus or minus width of RT windows for
all three techniques, for each analyte,  plotted against the
mean RT for the analyte. The external standard method's RT
window widths are represented by circled  points on  these
plots and are generally  greater than  the RRT window widths
(denoted with an asterisk) or the  RT index window widths
(represented by squares). Vertical lines are located at the
elution times of each IS and are  drawn with heights that
indicate the width of IS RT windows (which are calculated
using  the external standard formula).   The minimum
allowable RT window width (set at 0.02 minute) is represented
on these graphs by a horizontal line.

            Table 2: Key to RT Window Figures
Figure
No.
la
1b
2a
2b
3a
3b
4a
4b
5a
5b
GC Column
DB-624
OB-624
DB-624
DB-624
DB-624
08-624
1%SP1000
1%SP1000
1%SP1000
1%SP1000
Detector
PID
ECD
PID
ECD
PID
ECD
PID
ECD
PID
ECD
Location
base lab
base lab
mobile lab
mobile lab
mobile lab
mobile lab
base lab
base lab
base lab
base lab
Ambient
Temp. Range
Z3-Z4.5"C
23-24.5°C
24-29.5eC
24-29.5°C
21.5-25°C
21.5-25'C
20°C
20°C
20 - 24°C
20 - 24°C
 Figure nos. 1, 3, 4, and 5 each represent the RT windows from
 analysis of six standards on a given day, using a t-value with
 five degrees of freedom. Figure no. 2 was calculated with four
 degrees of freedom using pooled variances from standards
 run on two adjacent days. (This t-value is only 15 percent
 greater than that used in other figures.)

 General Trends.  Comparison of figure nos. 2 and 3 to figure
 no. 1 and comparison of figure no. 5 to figure no. 4 indicate
 that  the  width  of RT windows for all  three  techniques
 increases  substantially as the span of ambient temperatures
 increases.  The level of significance of these comparisons is
 graphically illustrated by a plot of the square root of F-test
                                                          778

-------
critical values (expressed as horizontal  lines) superimposed
over plots of the  ratio of RT window widths obtained at
variable versus controlled temperatures. Figure nos. 6, 7, and
8 compare the RT windows of figure no. 2 to figure no.  1,
figure no. 3 to figure no. 1, and figure no. 5 to figure no.  4,
respectively.  (Critical values in figure no. 6 were corrected for
the different t-values used.)   The F-tests indicate that many
analyte RT  windows cannot be considered  to be from
populations having identical variance at a 10 percent level of
significance.  Therefore,  RT windows  obtained  under
controlled temperatures should not be applied to analytical
runs under variable temperatures and vice versa.

Narrower RT windows, which indicate the superiority of the
RT  index method over the RRT  method, are much more
pronounced  under the variable temperatures in  figure nos. 2,
3, and 5 than under the controlled temperatures in figure nos.
1 and 4.   Capillary column  behavior under variable
temperatures revealed that 94 percent of RRT windows were
wider than corresponding RT index windows, with 56 percent
displaying at least a factor  of  2 ratio.   Under controlled
temperatures, capillary RRT windows were wider in 92 percent
of cases, with 33 percent 2-fold wider.  Packed  column
behavior under variable temperatures revealed that 83
percent of RRT windows were wider, with 46 percent 2-fold
greater.  Under controlled temperatures, packed column RRT
windows were wider in 88  percent of cases, with  only  8
percent 2-fold wider. In figure nos. 1  and 4, most of the  RT
prediction errors were on the same order as errors caused  by
the integrator's step size. This explains the noisy appearance
of these RRT and RT index plots.

Capillary Column Trends.  Under variable-temperature
conditions (figure nos.  2 and 3), the RRT and RT index  plots
exhibit pronounced spikes at RTs that correlate with a one- to
four-minute retention time region after the junction points
where the GC program is changed.  (At 7 minutes, the capillary
column ramp begins, and at 15.75 minutes, the ramping rate
shifts from 4 to 10°C per minute.)  The fact that these spikes
occurred at roughly the same RTs on the PID and ECD, despite
the elution  of the PID's middle IS nearly four minutes earlier
than the middle IS  used for the ECD,  suggests that this
 phenomenon is  not just a  consequence  of decreasing
 predictive ability (higher RT window width) as analytes  elute
 farther away from  the nearest  IS.  However, the adverse
 effects of a GC program's junction point may be partially
 offset if  an  IS elutes very close to the junction point.  For
 example, the spikes at  16 minutes in the ECD plots in figure
 nos. 2b and 3b are relatively smaller than corresponding spikes
 in the PID plots in figure nos. 2a and 3a.

 One  feature unique to figure no. 2 was the unusually high
 values for RRT window width at the beginning of the PID and
 ECD plots.  This  is  a consequence  of  high  ambient
 temperatures adversely affecting the GC's ability to maintain
 precise control over the desired 40°C initial temperature.  (A
 second unique feature of figure 2b was a bimodal distribution
 caused by inclusion of a separate analytical run containing
 only four analytes in the RT window plots.)

 Packed  Column Trends.   Packed  column  RT windows
 generated  under variable  and  controlled temperatures
 exhibited a  range of values very similar in magnitude to  those
 obtained on the  capillary column. It  is interesting that RT
 prediction  errors are  of similar magnitude because  other
 properties  of the packed column (namely resolution) are
 markedly  inferior  in  comparison with  capillary
 chromatography.  Packed column behavior under variable-
temperature conditions (figure no. 5)  also revealed  RRT
and/or RT index plots with spikes at certain RTs.  The RRT
window spike between 13 and 14 minutes in the ECD  plot
(figure no 5b) does not appear in the PID plot (figure no. 5a).
This spike may be attributable to a relatively greater distance
between these ECD analytes and the first IS. (Note that the RT
index method's windows were nearly optimum throughout
the region where this RRT spike occurred, resulting  in
windows that were narrower than the RRT method by up to a
factor of six.) The RRT and the RT index method exhibited
spikes of nearly equal height at elution times of 23 to 26
minutes in the PID plot in figure no. 5a.  These spikes may be
related to wide spacing between  ISs and also to the third IS
eluting in the final temperature hold zone that appears to be
associated with higher RT variance for all  compounds (as
suggested by wider external standard RT windows).  Because
the third ECD IS  eluted much earlier,  this trend was not
observed on the ECD.

                     CONCLUSIONS

The RT index method performed somewhat better than the
RRT method under controlled-temperature environments.
The variable-temperature environments afforded by the
mobile laboratory resulted  in wider windows for all three
methods, but the  RT index method  showed more dramatic
advantages over the RRT method under these conditions. The
 RT index method  showed the greatest degree of  superiority
over the RRT method at elution times that were affected by
GC program ramp shifting points and when analyte  RTs were
 most distant from  the nearest IS.

 Because all three techniques generate RT windows that are
 sensitive to temperature fluctuations over the course of a day,
 ambient temperature monitoring in the vicinity of the GC is
 useful during RT window development to define the range of
 conditions over which the derived windows will be applicable.
 Precise temperature control over the laboratory environment
 is desirable in  GC  analysis; however,  because mobile
 laboratory temperature control is rarely as good as that in a
 fixed laboratory,  the use of the RT index method will greatly
 assist in achieving the most specific GC analysis possible.  By
 using the RT index approach, mobile laboratory GC analysis
 can be performed in a manner that  minimizes the chances of
 erroneous (false positive) identification of target analytes.

                       REFERENCES

  1.  Sloboda, R. and Cohen, R., "FIT 3 SOPIII-1:  FASP Analysis
     for Volatile  Organic Compounds in Aqueous Samples
     Using Purge and Trap  Gas  Chromatography with
     Photoionization and  Electron Capture  Detectors in
     Series," TDD F3-8911-55, December 31,1990.

  2.  Brownlee, K.A.,  "Chapter 11:  Simple Linear Regression,"
     Statistical Theory and  Methodology in Science and
     Engineering. Second  Edition, John Wiley and Sons,
     Incorporated, New York, New York,  1965, pp. 334-361.

  3.  Zar, J.H.,  "Chapter  16:  Simple  Linear Regression,"
      Biostatistical Analysis. Prentis-Hall,  Incorporated,
      Englewood Cliffs, New Jersey, 1974,  pp. 205-215.
                   ACKNOWLEDGEMENTS

  This material has been funded wholly or in part by the
  Environmental Protection Agency under contract 68-01-7346
  to NUS Corporation.  It  has been subject to the Agency's
  review, and it has been approved for publication. Mention of
  trade names or commercial products does not constitute
  endorsement or recommendation for use.
                                                         779

-------
"                                         t
f   FI0.1U HT WINDOWS FOR CAP.CCT.JMD. AMBIENT TEMP M-24.5 C

N 0.15
                      11   13   15   17
                    RETENTION TIME (MINI
    FIQ.ll: RT WINDOWS FOR CAP.COL.PID.AMBIENT TEMP. 24-29.1 C
  0.05
 3   5   7   0   II   13   15   17
               IIEIEHIION IIHE (MINI

FIO.J1: RT WINDOWS FOR CAP.COL.. PID. AMBIENT TEMP 21.8-S5 C
   06



   0.5



   01



   0.3



   02



   U I
    o —i	K—i	1—•"—i	1— ^—H— i

     3    5    7    9    11   13    15    17   18   21
                    RETENTION TIME (MINI

    FIO 41: RT WINDOWS FOR PACKED COL..PID. AMBIENT TEMP 2o'c


  025T
  005
     6   8  10  12  14  16 18 20 22  24 26 28 30 32  34
                   RETENTION TIME (MIN)

   FIQ.Il: RT WINDOWS FOR PACKED COUPID, AMBIENT TEMP 20-84 C
    6  8  10  12  14  18 18 20 22  24  26  28 30 32 34
                   RETENTION IIME |MIH|
         EXTERNAL  STD


         MIN  ALLOWABLE
                                                                       FIO.tDi RT WINDOWS FOR CAP.COL, ECO, AMBIENT TEMP 23-84.8 C

                                                                    U. 15
                                                                                        11   13   15   17   19   21
                                                                                      RETENTION TIME IUINI

                                                                      FI0.2t>: RT WINDOWS FOR CAP.COUECD.AMBIENT TEMP 24-2l.s'c
                                                                                            II   13   15

                                                                                          DETENTION IIME (WIN)
                                                                           FI0.3b: RT WINDOW FOR CAP.COL, ECD. AMBIENT TEMP 21 S-2s'c
                                                                  N 0.5
                                                                  I
                                                                  i
                                                                  o
                                                                    0.4
                                                                          -4-
                                                                                               =F

                                                                       3    5    7   9   11   13    15   17   tg   31

                                                                                     ntHNIION TIME IMIN)

                                                                     FIG.4k RT WINDOWS FOR PACKED COUECO. AMBIENT TEMP 20*C
025
l) I,S
0 1
w 0.05
M n
f
*-^w»
^-vVvk A ~tr^°

8 10 12 14 16 18 2U 22 24
                                                                                     AEIEHtlONllUE|MIH|

                                                                    FIO.Bb, RT WINDOWS FOR PACKED COL4CD. AMBIENT TEMP 20-84'C
                                                                      I  °3
                                                                      N 0.25

                                                                      n
                                                                      »  CI.2
                                                                      T

                                                                      " 015
                                                                      w

                                                                      8  01
                                                                      0
                                                                      w

                                                                      « 0.05


                                                                      11   o
                                         RELATIVE  RT


                                         INTERNAL  STANDARDS
6    8    K)   12   14   16   16   20   22   24
              REIENTION TIME (MINI

-e-   RETENTION  INDEX
                                                          780

-------
      FIG.6a: RTW RATIOS ON CAP.COL.,PID,24-29.5 C vs. 23-24.5 C
                                                           FIQ.6b: RTW RATIOS ON CAP.COL,ECD,24-29.5 C vs. 23-24.5 C
n
E

E
N
T

0
N

T
I
M
E

W
I
N
                                              .—€^
     5    7    9    11    13    15   17
                MEAN RETENTION TIME (MIN)

 FIQ.7a: RTW RATIOS ON CAP. COL.PID, 21.5-25°C vs. 23-24.5°C
                    9    11    13    15    17
                     MEAN RETENTION TIME (MINI
                                                  21
3    5    7


FIQ.8a: RTW RATIOS ON PACKED COLUMN, PID, 20-24 C vs. 20 C
w
i
N
8
   o-
           10  12  14  16  18  20  22  24  26  28  30  32
                     MEAN RETENTION TIME (MIN)
                                                   	     o
                                                               3   5    7    9   11    13   15   17   19   21   23
                                                                               MEAN RETENTION TIME (MIN)

                                                                FIG.7b: RTW RATIOS ON CAP.COL, ECD, 21.5-25°C vs 23-24.5°C
                                                             15--
3   5    7   9   11    13   15    17   19   21  23
                MEAN RETENTION TIME (MIN)

FIQ.Sb: RTW RATIO ON PACKED COLUMN, ECD, 20-24°C vs. 20°C
                                                            6-r
                                                                   10
              EXTERNAL  STD
                                         RELATIVE RT
              12    14    16    18    20    22
               MEAN RETENTION TIME |MIN)

              RETENTION  INDEX
24
              F  VALUE  (P-0,1)      ~~-  INTERNAL  STANDARDS
                                                      781

-------
                              DETECTION OF AIRBORNE MICROORGANISMS
                           USING A HAND-HELD ION MOBILITY SPECTROMETER
A. Peter Snyder
U.S. Army Chemical Research,
Development and Engineering
Center
Aberdeen Proving Ground, MD
   21010-5423
                                  David  A.  Blyth,  John A.  Parsons
                                  CEO-CENTERS,  INC.
                                  10903  Indian  Head  Highway
                                  Suite  502
                                  Ft.  Washington,  MD  20744
              Gary A.  Eiceman
              Department of Chemistry
              New Mexico State University
              Las Cruces, NM  88003-0003
Microorganism detection  and  identification
challenge  even the  best  microbiological,
clinical and  analytical  Instrumentation  tech-
niques.  Analytical  techniques  such as gas
chromatography of inherent volatiles, ^*C02
radiotnetry, electrochemical  measurement  by
molecular  hydrogen  and mtcrocalorimetry  have
typical sensitivities and  response times  in
the  10° cells/ml  and 1.5 hr  ranges, respec-
tively.  Colorlmetric and  fluorometric
microbiological procedures fare much better
and  can be found  in  the  103-105 bacteria
cells/ml and  0.25-4  hr time  ranges.

The  detection of  fecal colI form bacteria  (£.
coll, Klebslella, Cltrobacter and Entero-
bacter) is of prime  concern  In water, waste-
water and  soil analysis  and  management.
Since J5. coll Is  found in  100-fold greater
concentrations than  the  other bacteria,  its
detection  can indicate the presence of
pathogenic organisms.  A Graseby-Ionics
hand-held  Ion mobility spectrometry device
was  investigated  for its potential in the
detection  and determination  of microblal
presence in both  liquid  suspension and aero-
sol  form.   A  standard microbiological test
that  is used  to detect the presence of total
fecal collforms is  the extracellular enzyme
reaction with that  of ortho-nitrophenyl-
galactopyranoslde (ONPG).  If these bacteria
are  present,  the  constitutive enzyme beta-
D-galactosidase cleaves  the  colorless ONPC
substrate  to  produce the galactosldase sugar
and  the yellow ortho-nitrophenol (ONP)
products.   Because of Its  relatively low
melting point (45-46°C), ONP has a consider-
able vapor pressure  (0.54  torr).  This concept
was exploited by  the use of  the ion
mobility spectrometry analytical technique.
Strips of  filter  paper were  inoculated with
microliter amounts of fecal coll form cells
and the ONPG substrate,  and the strip was
then placed in a vial and stoppered.  After
15 min, a clear ion mobility peak that
matched that of pure ONP was registered when
the hand-held unit was placed near the open-
ing of the vial.  With pure bacterial sus-
pensions, approximately  300 cells of £. col1
were detected in 15 minutes with ONPG and
1,000 cells of Bacillus  subtilts were
detected in 15 minutes with ONPG acetate.
Indeed, the accumulated  headspace vapor of
the liberated ONP from the bacterial enzyme
reaction was the source  of the signal.

Under controlled (0.45 ra^ container) con-
ditions, bacterial aerosols were collected
by a four stage impactor and the filter
paper strips from each stage were subse-
quently analyzed by IMS  monitoring of the
ONP vapor product.  Positive responses were
observed in the second and third stage of
the impactor (corresponding to 2-4 microns)
while the fourth stage (1 micron size)
produced no signal.

These observations indicate the potential
of ion mobility spectrometry in the
detection of extremely complex analytes
as that of living microorganisms under both
suspension and aerosol conditions.
                                                  783

-------
               FIELD  ANALYSIS  FOR  HEXAVALENT  CHROME  IN   SOIL
                 Robert L. Stamnes
              Environmental  Engineer
        U.S.  Environmental  Protection Agency
                Seattle, Washington

                 Gregory D. DeYong
                 Research Chemist
                  HACH Company
                    Ames, Iowa

ABSTRACT

Hexavalent chromium [Cr(VI)] has a high aqueous
solubility and  has entered the groundwater at many
hazardous waste sites. Chrome  is present at over
one hundred  sites of  concern nationally (1).
Hexavalent chromium is environmentally
significant. It moves rapidly from the soil through
groundwater to receptors. Soils contaminated with
hexavalent chromium are targeted frequently for
removal and  treatment.

Electroplating sites containing chromium are an
important  class of sites in  Region 10 of the EPA.
Region 10 has determined the extent of soil removal
at sites by first determining  the  extent of leaching
that corresponds to various levels of chrome in the
soil.  The cleanup criteria (concentration of Cr(VI)
allowable  in the soil) is then dictated  from the
desired groundwater  quality.  A  field test procedure
for the detection of chrome can then be applied to
direct the soil  removal effort.

The test method for defining the removal effort
should be simple, field portable, quick and accurate
in the parts  per billion (ppb) range.  X-ray
fluorescence (XRF) is a proven  and useful
analytical tool for investigation and  management  of
metals in  soils. However, chromium is one of the
 heavy metals that has more  than one oxidation state.
 Hexavalent chromium is the oxidation state of
environmental concern.  Hexavalent chromium can
 not be differentiated by the XRF. Another procedure
 must be used in conjunction with the XRF to
 establish  hexavalent chromium  levels in the soil.
 The HACH Company developed a field procedure for
 the analysis  of hexavalent chromium in soils. The
 procedure reportedly permits the measurement  of
 Cr(VI) in  soil down  to 250 ppb  in less than 30
 minutes.
            Clark D.Carlson
           Inorganic Chemist
       The  Bionetics Corporation
            ESAT Region 10
       Port  Orchard, Washington
HACH PROCEDURE (OUTLINE)

1) Preparation of sample:
       A)  Dry sample
       B)  Homogenize sample

2) Measure  appropriate sample into Whirl-pak™
bag. The appropriate Sample Size is based on the
Estimated Concentration in the soil.
         Estimated
    Cr6+  Concentration         Sample Size
       250-5000  ppb
       0.50-10  ppm
       2.50-50  ppm
          50-1000 ppm
         500-10000  ppm
20 g
20 g
20 g
 1 9
 19
 3) Add extractant solution to Whirl-pak™ bag.
 Prepare the solution by adding  one extractant
 pillow to 40 mis of deionized water.

 4) Extract  the sample:
        Shake mixture for 15 seconds at two
        minute intervals.  Continue for 15
        minutes.

 5)  Filter the  extraction  mixture.

 6)Transfer appropriate and equal aliquots of
 extraction  fluid to  two (2) 25 mL graduated
 cylinders and dilute to 25 ml.  The aliquote size  is
 based on the estimated Cr6+ concentration in the
 soil.
         Estimated
    Cr6+ Concentration         Sample Size
        250-5000ppb
        0.50-10  ppm
        2.50-50  ppm
           50-1000 ppm
          500-10000  ppm
 10 mL
  5 mL
  1 mL
  1 mL
  0.1 mL
                                                      785

-------
 7)Add one (1) Chromaver 3™ pillow to one of the
 graduated cylinders. Allow the solution to react for
 a minimum  of 10 minutes.

 8)Pour the contents of the graduated cylinder into
 viewing tubes.

 9)Place the viewing tubes into the viewing box and
 determine the concentration
 Cr6+ ppb
   disk reading (mg/mL) x 10*>

aliquot vol.(ml) x sample size (g)
 Quality Control:      Add standard solution to soil
 and analyze as described above.

 DISCUSSION

 Region 10 has applied this procedure to actual site
 soils from several sites.  The procedure could not
 be used at some sites due to interferences, but was
 successful detecting hexavalent chrome in soils
 from other sites.  The Cr(VI) data  are presented in
 Table  1 and Figures 1 through 4, where the
 procedure was successful.  Several laboratory
 procedures were used for comparison with the
 HACH kit results.  Method  1310 (Toxicity
 Characteristic Leaching  Procedure-TCLP). method
 1311  (EP Toxicity) and  method 3060 are the
 laboratory procedures used  to extract samples
 (2),(3),(4).  The  concentration  of hexavalent
 chrome in the extract was determined by Method
 7196  (5).

 Figure 1 depicts the relationship between the
 laboratory methods and HACH kit results. The best
 correlation is between Method 3060 and the HACH
 kit results.  Both methods use alkaline extractants.
 The TCLP and EP Toxicity methods correlate poorly
 with the HACH kit results.  Unlike the HACH kit,
 these procedures  require the adjustment of the pH
 to slightly acid.

 Temperature and extraction time dependance were
 also evaluated using actual site soils.  Figures 2, 3
 and 4  present this information for two different
site soils.  Concentrations acquired by Method 3060
can be achieved using the HACH kit with adequate
extraction  times and  extract temperatures.
Adequate extraction time is depicted  more clearly in
Figure  4 for a particular extract temperature.
                                               Figure 1   Laboratory Correlation with
                                                           Test Kit
                                                      300
                                                                    250
                                                                    200
                                                              Laboratory Correlation
                                                                      in PPM
                                                   ^  100

                                                   1   50
                                                   o
                                                   "*    0
                                                         0    50   100  150  200  250  300
                                                              Test Kit Concetration

                                                        ^.method 3060 ^.TCLP
                                                        -4-EPTox      .B. Theoretical
                                              Figure 2   Extraction Time  Dependence of
                                                          Test Kit Results
                                                     100
                                                          Extraction Time Dependence
                                                  ex
                                                  D.

                                                  .
                                                     90
80
                                                  1  7°

                                                 5  60
                                                     50
                                                                     10   20   30   40   50   60   70
                                                                         Extraction Time (minutes)
                                                                 86.1F     ^71.2 F     ^62.5 F
                                                                 48.8 F     ^28.5 F     _Method30
                                                        786

-------
Figure 3   Temperature  Dependence  of
           Test Kit Results,
           Site #1, Sample #1
       Temperature Dependance of Results
       100
                                                Table 1  Results from use o> the test kit
OH
CX
       90

       80
    O

    I™
    
-------
CONCLUSIONS

The HACH kit is a valid field screening procedure
for hazardous waste site investigation and
remediation.  However, attention must be given to
the possibility of interferences and adequate
extraction time  for the particular temperature
conditions.

While the HACH procedure correlates well with
Method 3060 results, it over estimates TCLP and
EP Toxicity results. No false positives were
detected with the limited number of analysis and
limited number of wastes used in this study.

False negative results can  be detected by spiking  the
extract with Cr(VI)  after an  analysis indicates no
Cr(VI)  present.  This verification procedure could
be run on each  soil matrix to reveal interferences.
REFERENCES

1.     U.S. EPA, "Alternative Treatment
       Technology Information Center" (ATTIC),
       Version 2.0, U.S. EPA Office of
       Environmental Engineering and Technology
       Demonstration, October 1990, (ROD
       Database)

2.     "Hazardous Waste Management System;
       Identification and Listing of Hazardous
       Waste ;Toxicity Characteristic Revision;
       Final  Rule", Federal  Register 55:126 (29
       June  1990), P. 26986.

3.     Identification and Listing of Hazardous
       Waste", Code of Federal Regulations Title
       40, Pt.  261/1990 ed.  (Appendix  II  - EP
       Toxicity Test Procedures)

4.     Radian  Corporation, "Evaluation of Soil
       Remedial  Action Levels for Frontier  Hard
       Chrome, October 1990, p. A-3.

5.     U.S. EPA, "SW-846 Test  Methods for
       Evaluating Solid Waste, U.S. EPA,  Office of
       Solid Waste, Third Edition, November 1986
NOTE: No official support or endorsement by the
Environmental Protection Agency, federal employees,
or any other agency of the Federal Government of the
product,  the procedure, or its manufacturer,  is
intended or should be inferred by this paper or
presentation.
                                                          788

-------
                    TRANSPORTABLE TUNABLE DYE LASER FOR FIELD ANALYSIS
                        OF AROMATIC HYDROCARBONS IN GROUNDWATER
                           Randy W. St. Germain and Gregory D. Gillispie
                           Department of Chemistry
                           North Dakota State University
                           Fargo, ND  58105
INTRODUCTION
     We have developed a transportable and fully
wavelength tunable laser system  for remote fiber
optic  fluorescence analysis  of  aromatic  hydro-
carbons.  System components include a pulsed Nd:
VAG  pump  laser, a two stage dye  laser,  fiber
optic probe,  monochromator/photomultiplier tube
/digital oscilloscope detection  system, and  386
portable control computer.  The  system can  eas-
ily  be moved in a van to remote  locations  for
field  operation with power supplied by a  5  kW
generator.  The data we present  here  accurately
represent  the  system's  current   capabilities
since the data were taken with the field version
of  the  unit; the only difference from  actual
field  operation  is that system power  was  not
derived  from the generator.

     Our approach improves on other field  laser
fluorescence  schemes because it employs a  com-
pletely wavelength tunable laser and because  we
emphasize time resolved detection.

SYSTEM DESCRIPTION
     The system components are rigidly  attached
to a wheeled 2'Wx4'Lx3'H shock-mounted  unistrut
cart. The Nd:YAG pump laser,  harmonic generator,
and 0.32 m emission monochromator  are located on
the base of the cart.  The dye laser,  frequency
doubling  crystal,  and  associated  optics  are
mounted  on a 2' x 4' optical breadboard  bolted
to the top of the cart; the digital oscilloscope
is supported over the breadboard.  Connected  to
the system by an umbilical, the  YAG laser  power
supply (approximately a cube  two feet on a side)
sits on another movable cart. Data  acquisition
boards,  power supplies, stepper motor  control-
lers, etc. are located in an  expansion box under
the optical breadboard.

     The pulsed light emerging horizontally from
the harmonic generator is directed up through  a
one-inch hole in the optical  breadboard.  As the
beam emerges from the hole, a Pellin-Broca prism
directs  the  light parallel  to  the  breadboard
surface  as it separates  the  desired  dye  laser
pump  wavelength (532 or  355  nm) from  the  1064
fundamental.  After  this separation  the  pump
light  is  split  into separate  beams  for  the
oscillator  and amplifier cells of  the  Littman
grazing incidence dye laser.   The  monochromatic
and wavelength-selectable visible light from the
amplifier  cell  is frequency-doubled  into  the
ultraviolet  with a KDP crystal.  The  following
filter  rejects residual  visible light from  the
KDP crystal and passes about  40% of the UV light
onto  a lens for focussing onto the  600  micron
delivery fiber.  Figure 1 illustrates the  pulse
energies at various points in the optical  train
for 355 nm pumping of Coumarin 500 dye.
                                       125 pj U.V
            20 pJ U.V.
                                      5m Probe
 Figure 1.  Pulse energy conversion
                                                 789

-------
RESULTS AND DISCUSSION
   Although  typical of current  day-to-day  per-
formance,  the UV pulse energies  available at the
probe  distal end can be markedly  improved  with
minor modifications. For example,  increasing the
355 nm pump energy to 50 mJ and  changing the os-
cillator/amplifier split to ca.  10 mJ/40 mJ will
bring  the  amplifier output to  ca.   8  mJ.  The
efficiency with which the UV is  freed from  vis-
ible light after the doubling crystal can  prob-
ably  be  boosted by a factor of two.   Some  im-
provement in doubling and launch efficiencies is
likely  also.  Ultimately we expect   to achieve
150-250 uJ pulse energies at the distal end of a
5 m probe.

     The BTX components require  excitation wave-
lengths  shorter than those available from  fre-
quency doubling the output of a  dye  laser pumped
at 532 nm; they require pumping  at 355  nm.  How-
ever, nearly any other aromatic  hydrocarbon  can
be probed at sufficiently long excitation  wave-
length to permit 532 nm pumping  of highly effic-
ient dyes such as Rhodamine 6G.    Owing to  such
factors as greater available pump  energy at  532
nm (up to 150 mJ with our DCR-11 Nd:YAG laser),
conversion efficiencies up to 35%  with  R6G,  and
lower fiber attenuation,  we can  confidently pre-
dict that pulse energies of 2 mJ or  more can  be
delivered at 280 nm, for example.  In the lab we
have routinely produced 3 mJ ultraviolet  (prior
to launch into the fiber) with only  60  mJ of 532
nm pump energy.

     The anthracene calibration  curve in  Figure
2 illustrates the high sensitivity (better  than
10 parts-per-trillion detection  limit)  and  wide
dynamic  range available for a PAH that absorbs
reasonably  strongly at a wavelength achievable
with 532 nm pumping and not strongly attenuated
by the fiber.
 CD
 W
 O  -1
 o
 _J
    -2
    -3
                        Slope
0.96
      -3      -2      -1       0
                LOG CONC. (ppb)
    Figure  2. Anthracene calibration
                curve
                           For reasons already described (need to pump
                      the   dye laser at 355 nm, higher light  attenua-
                      tion  by the  fiber) and molecular factors  (lower
                      molar absorptivity, generally lower fluorescence
                      quantum yield), the detection limits for the BTX
                      components   are higher than for anthracene.   For
                      example, the limit of detection for p-xylene  is
                      currently  about 1 ppb (Figure 3).  Benzene  and
                      toluene  are even weaker emitters and their  de-
                      tection limits are correspondingly higher (5 ppb
                      for toluene,  20 ppb for benzene).  Nevertheless,
                      the   linear  calibration plots found over a  wide
                      concentration range  for  all  three  BTX  com-
                      ponents  are encouraging.  Future  work  should
                      yield detection limits below the MCL's.
                      CO
                      o
                      o
                         -1
                                       Slopo-1.00
                                                    L.O.D.
                            01234
                                      LOG CONC. (ppb)
                           Figure 3. p-Xylene calibration
                                       curve
                           As mentioned  in the introduction we plan to
                      exploit the pulsed nature of our laser source as
                      an   aid to quantitate  multi-component  samples.
                      Figure 4 demonstrates the accuracy with which we
                      can  measure fluorescence lifetimes.
                                                       0.40
                                                       0.32-
                                                    o
                     m
                         0.24-
                         0.16-
                                                       0.08-
                         0.00
                                            TIME (ns)
                        Figure 4.  Anthracene lifetime test
                                                 790

-------
     We  chose anthracene in cyclohexane  as  our
test case because accurate literature  values are
available.  Fluorescence decay profiles in  both
air-saturated and degassed cyclohexane are  shown
in  Figure 4 along with the laser  time-profile.
Owing lifetime being short compared to the  laser
pulse  duration,  deconvolution  is   necessary.
Moreover, a separate correction must be   applied
for  the  differential  transit  time  of  light
through  the  fiber at  different  wave-lengths.
Software to accomplish this in nearly  automated
fashion  yielded the indicated lifetimes,  which
are  in excellent agreement with the  literature
values.

     Another   way   to  prove   that    we   are
satisfactorily correcting for the fiber   transit
time  effect ia to compare lifetimes  taken  for
the identical sample with and without  the  fiber
optic  probe.  Results are shown for toluene  in
water in  Figure 5.  The  same  lifetime is derived
after deconvolution in each case.  Note that the
time gap  between the maximum  laser intensity and
the  maximum  fluorescence  intensity is less  for
   0.12
   0.00
   0.12
         Laser
 2 0.09-
 o
 >
Fiber  Probe

     = 4.9  nsec
   0.06-
 w
 UJ
   0.03-
   0.00
         — —Fluor, (obs.)
              Fluor, calc.
                               the  probe example;  the fluorescence,  at   longer
                               wavelength than the scattered laser light,  takes
                               less time relative to the scattered laser  light
                               to travel via the fiber to the PMT.  It is note-
                               worthy  that the fluorescence lifetimes  of  the
                               individual  BTX  components in  water  are  sig-
                               nificantly different, whereas in aliphatic  sol-
                               vents  they  are all about the same.    Figure  6
                               summarizes  the values we find for degassed  and
                               air-saturated solutions.  The roughly factor  of
                               two variation between benzene and toluene and  a
                               similar factor for toluene relative to  p-xylene
                               may  prove  helpful for  separating  their  con-
                               tributions to the total fluorescence signal.
                                  16.0
          Figure 5.  Toluene lifetime
          Benzene       Toluene       P-Xylene

       HI Oxygen Saturated  |   I Oxygen Removed


          Figure 6.  BTX  lifetimes

     Two popular multi-dimensional   fluorescence
techniques for chemical analysis are excitation-
emission  matrices  (EEM's) and  phase   resolved
fluorescence. A third possibility is illustrated
in Figure 7 (next page); the analyte is JP-4 jet
fuel in water.  We refer to these plots aa wave-
length-time  matrices, by analogy to EEM's.    To
generate  a wavelength-time matrix,  either  the
excitation  or emission wavelength  is   stepwise
varied and a fluorescence time profile  collected
at  each  setting.  For figure  7  the   emission
wavelength  that has been varied in 3 nm  incn
ments  for excitation  at 262 nm, such  that  the
signal is primarily due to the BTX components.

     The  wavelength-time matrix at the top  of
the  figure is  for water equilibrated with JP-4.
This   sample  therefore contains a  high  concen-
tration   of  BTX.  The  bottom of Figure  7  shows
the  corresponding  wavelength-time  matrix   for
pure water.   The  narrow feature represents Raman
scattering of  the water solvent.   In the  middle
picture   the  water Raman  and   BTX fluorescence
make  comparable  contributions to the  overall
 intensity.    Over the concentration  range  that
the  water Raman  band can  be distinguished  from
the  fluorescence,  it  can be  used as an  internal
 standard.
                                                   791

-------
                               SATURATED
                               SOLUTION
                               1/20 DILUTION
                               OF SATURATED
                                 SOLUTION
      We will  therefore in the future reduce  the
laser  pulse duration to under 1 ns to gain   the
advantages  shown   in Figure 8.   The  indicated
fluorescence   decay profiles at  two  different
laser pulse durations are for lifetimes close to
the  actual values for  benzene,   toluene,   p-
xylene,  and naphthalene.  The shorter the  laser
pulse relative to the fluorescence lifetime,  the
easier it is to select a time gate for detection
such that the  laser intensity is nearly zero  but
the  fluorescence   intensity is still  near   its
maximum.   We  believe that the resulting  better
rejection   of    background    scatter   will
significantly   improve our detection limits   for
benzene.  This technique will also aid rejection
of  physical   scattering as, for  example,  from
soil particles in a monitoring well environment.

ACKNOWLEDGMENT
   This  work has been supported by the Air Force
Engineering Services Center.
                               Pure Water
                                                                         Laser, 5 ns duration
 Figure 7. JP-4 wavelength-time matrices
FUTURE WORK
    Because the  BTX component fluorescence  life-
times  are not long compared to the laser  pulse
duration,   their fluorescence profiles are   only
slightly altered in time relative to that of the
laser.   Consequently,  the Raman  signal  (rig-
orously coincident in time with the  excitation)
is heavily overlapped temporally with the fluor-
escence, making  them hard to separate.
                                                                         Laser, 0.5 ns duration
                 5 nsec/division

    Figure  8. Fluorescence  decays
                                                  792

-------
                        Real Time Detection of Biological  Aerosols
         Peter J.  Stopa,  Michael  T.  Goode,  Alan W.  Zullch, David W. Slckenberger,
                           E. William Sarver, Raymond A. Mackay

                               Detection  Technology Division
                       U.S. Army Chemical Research* Development and
                                    Engineering Center
                          Aberdeen Proving  Ground,  MD  21010-5423
    Interest In the environmental  Impact of
biological aerosols has Increased  due to the
Implications of aerosolized bio-materials In
Indoor building pollution; the release of
genetically-engineered organisms Into the
environment; and the release of potentially
pathogenic organisms downwind from sewage
treatment plants.  Efforts to date for the
real time detection of biological  aerosols
have proven unsuccessful  due to the lack of
the technology to discriminate between
potentially hazardous materials and
background materials.

   The Integration of rapid Immunoassay
technology with a real-time air sampling
capability Is presently under Investigation.
A two-stage air sampler with Impingement
capability has been developed and  Integrated
with an 1mmunolog1cally-based biosensor to
effect a real-time aerosol detection
capability.  The sampler concentrates and
Impinges 100 liters of air Into 100 ul of
fluid.  The Impinged sample 1s then mixed
with the Immunoreagents, and the resulting
Immune complex Is trapped onto a
nitrocellulose filter while unreacted
materials are washed away.  Urease, an enzyme
which effects a pH change 1n the substrate
buffer, 1s used as the enzymatic tag.  The
slope of the resultant pH change Is then
determined through the use of the Light-
Addressable PotentlometHc Sensor.

     Mass spectrometry offers an alternative
means of detection of aerosols.  A small*
field portable mass spectrometer,  based on
Quadrapole Ion Storage (QUISTOR) technology.
Is also being developed and Integrated with
an air-sampling capability.  The
concentrated sample will  then be pyrolyzed,
with the resulting pyrolysates being
analyzed by MS or MS/MS.   The resulting
spectrum will be compared against an onboard
library.  An artificial Intelligence
capability will allow unknown materials to
be analyzed and retained for future
reference.

   Efforts to date have centered on
materials which are of military Interest.
Detection levels as low as 1 ng have been
achieved In a one minute assay time for the
Immunoassay system.  Non-milItary uses of
this technology can be developed, depending
on the applications of the customer and the
development of appropriate antibody reagents
or mass spectral libraries.
              INTRODUCTION

   The contributions of biological aerosols
to both Indoor and outdoor pollution
problems have often been neglected.  This
has been largely due to the difficulty one
has 1n the detection and characterization of
the aerosol.  Past attempts at detection
have relied upon measuring changes In bulk
properties, such as heme content or the
presence of specific reactions.  These
attempts failed because they did not possess
the sensitivity and/or specificity to
distinguish between hazardous and non-
hazardous materials.  Recent advances In
both biosensor and mass spectroscoplc
technologies are leading to the development
                                                  793

-------
of small, lightweight, and rugged Instruments
which can be Introduced Into field
applications for detection of biological
materials.
           THE  BIOCHEMICAL DETECTOR

   The field assay of biological materials Is
difficult due to the need for detection of
often minute quantities of specific material
1n the presence of a large background.
Immunologlcal-based sensors are being
developed which have the capability to
differentiate between analytes and
background.  Antibodies are bound to either
optical or electrochemical  transducers, and
the binding event Is measured due to the
generation of an optical or an
electrochemical event.  Typically,
fluorescent dyes or chromogenlc enzyme
substrates are used for the transductlon of
an optical signal while enzyme-substrate
combinations which yield electrochemlcally
active species are used with the
electrochemical sensor.  Both types of
sensors were Initially evaluated 1n this
project.  The electrochemical one was chosen
for further development.

     This sensor 1s referred to as the
"Light-Addressable Potent1ometr1c Sensor"
(LAPS), and 1s marketed by Molecular Devices
Corporation, Menlo Park, CA, as the
Thresholds system.  It  1s simple 1n design,
consisting of a silicon wafer, on which a
layer of silicon oxlde/sllIcon nitride has
been grown.  This silicon oxide/silicon
nitride serves as an electrical Insulator and
makes the  surface Impervious to 1on
migrations from solutions In contact with the
surface,  Imparting a neutral pH sensitivity
to the sensor.  The transducer  Is used to
monitor the activity of enzyme-labelled
antibodies which are used 1n the reaction.
Presently  urease, which catalyzes the
hydrolysis of urea to carbon dioxide and
ammonia.  Is used.

     The  1mmunolog1cal  reaction takes place
on a nitrocellulose filter which Is later
placed on the sensor surface.  A controlling
electrode  and a reference electrode back-
bias the  Insulator/silicon Junction, creating
a depletion layer at the Junction (absence of
charge carriers).  A light emitting diode
(LED), driven at 10 kHz, Illuminates a small
area of the silicon chip, creating charge
carriers  In the depletion layer at that
point, which results In an alternating
current between the controlling electrode
and the bulk silicon.  The magnitude of this
current Is dependent on the surface
potential  of the silicon oxide/silicon
nitride.  This surface potential responds 1n
a Nernstlan manner to changes 1n the surface
pH.  Several LED's can be used on the chip
so that multiple sites can be addressed 1n
succession.  When coupled with the
appropriate Immunologlcal reagents, a sensor
can be obtained which has a detection
capability for several materials 1n a small
area.

     The Bio-Chemical Detector, currently
under development by the Army, utilizes this
sensor technology In conjunction with an
aerosol sampler.  The sampler Impinges
aerosols Into a liquid medium.  This liquid
1s then transferred to a reaction manifold
where the reagents are added.  The resulting
Immune complexes are then filtered through
an active membrane which captures the
complex.  Unreacted reagents are then washed
away, and the filter Is transferred to the
reading module where the pH change 1s
obtained.

     This detection system has as a design
goal the detection of six different classes
of biological materials- bacteria,
rlckettsla, viruses, and small, medium, and
large molecular-weight toxin materials.
Other goals Include a total sample
acquisition time of two minutes with
capability of repetitive analysis over a 24
hour period.  Initial results have been
encouraging with detection limits of
nanograms per mllUllter of toxin materials
being realized with the sensor.  Detection
limits of 10*5-10*6 organisms per ml have
been realized with two of the mlcroblal
materials.  These detection limits are
realized 1n the 3-4 minute time frame.  The
next phases of development Include better
Integration of modules and Improvements to
the  Immunoassay format.

    Although this system  Is being developed
for materials which are of military
Interest, the use of this system can be
extended to other materials through the
development of appropriate antibody
reagents.   It  1s conceivable that this type
of detection system could be utilized where
real-time detection of hazardous biological
materials  Is required.
                                                    794

-------
               THE CBMS SYSTEM

     There Is much Interest 1n the
development of small,  portable mass
spectrometer units for field analysis of
hazardous materials.   Although these units
are significantly smaller than their
laboratory counterparts* compromises are made
with respect to capability and resolution.
In addition, the system must be capable of
Identifying trace quantities of hazardous
materials often 1n the presence of other
Interfering compounds.  This can often be
accomplished through  the use of a GC/MS
system; however, the  complexity and
logistical burden of  this type of system may
make It too cumbersome for routine use 1n the
military environment.

     The CBMS 1s an attempt to develop a
small, portable mass  spectrometer which has
the capability to detect trace quantities of
materials 1n the midst of significant amounts
of Interferents.  It  utilizes Quadrapole Ion
Storage technology (QUISTOR) to accomplish
this goal.   This technology has allowed for
the development of a  small, sensitive mass
spectrometer which has an MS/MS capability
and can be fitted with a variety of  probes  to
enable sampling of ground contaminants or the
detection of biological aerosols. The ground
sampling probe 1s commercially available and
will not be described In detail here.  The
biological aerosol sampling capability was
the result of an In-house effort at  CRDEC.
The aerosol material 1s impacted Into a
quartz tube with subsequent pyrolysls by IR
radiation.  The resultant vapor 1s then
Introduced Into the Instrument and analyzed.
In the case of bacteria where similar
primary spectra are obtained, the unit Is
then switched Into an MS/MS mode where
daughter Ion spectra are obtained.  This has
allowed bacterial  Identification to the
Genus level; species level  should be
possible with the development of the
appropriate spectral libraries.  In
addition, an artificial intelligence
capability Is being built Into this unit so
that it will be capable of analyzing spectra
of unknown materials and trying to "best
guess" what they are.
                  SUMMARY

   Both these systems offer viable
approaches to the real time detection of
biological aerosols.  They also demonstrate
two of the principles which can be used to
achieve this:  one being the specificity of
antibody molecules, while the other being
the chemical signature; a material would
give in the mass spectrometer.  The bio-
specificity approach requires a lot of up
front work in the development of the
biological reagents (although the hardware
development is not trivial) while the other-
requires a significant amount of work in the
development of appropriate spectral
libraries.  Both systems are In the early
stages of development and show great promise.
                                                  795

-------
                  LASER FLUORESCENCE EEM INSTRUMENT FOR In-Situ
                              GROUNDWATER SCREENING
                  Todd A. Taylor, Hong Xu, and Jonathan E. Kenny
                             Department of Chemistry
                                Tufts University
                               Medford, MA 02155
ABSTRACT

We have constructed and laboratory
tested a field transportable
fluorescence instrument for aqueous
pollutant screening.  This instrument
acquires 3-dimensional laser-excited
excitation-emission matrices (EEMs) of
environmental solutions.   Computer
analysis of these EEMs by the least
squares method allows determination of
the chemical composition of the
solutions.  Our instrument generates
more than 30 laser beams of different
wavelengths throughout the ultraviolet
region of the spectrum using a laser-
pumped Raman shifter, and these beams
act as a source for fluorescence
excitation.  Laser light and fluores-
cence emission is transported between
the environmental sample and the
instrument by optical fibers.  Our
instrumental response is linear with
concentration over 3-4 orders of
magnitude and detection limits are in
the ppb range for many pesticides and
pollutants.  Three and four component
mixtures of groundwater pollutants have
been directly analyzed at sub-ppm
concentrations in methanol solutions
with less than 20 % error using least-
squares EEM analysis.
INTRODUCTION

Our group has previously constructed and
field tested a Nd:YAG (neodymium:
yttrium aluminum garnet) laser-based
instrument for in-situ monitoring of
groundwater pollution by fluorescence
analysis (1-4).  This first generation
instrument employed optical fibers to
carry laser light and fluorescence
between the instrument and the ground-
water sample.  Optical filters were
placed before the photomultiplier tube
detector to select the emission spectral
detection range.  Concentrations of
fluorescent pollutants that absorbed 266
nm radiation could be determined in the
parts per million (ppm) to parts per
trillion (ppt) concentration range.
However, this instrument was not
sensitive to compounds having a low
absorptivity at 266 nm and had limited
capability to distinguish different
fluorescent compounds from each other.

Our second generation instrument,
described in this paper, has a much
improved ability to analyze the chemical
composition of groundwater.  This
instrument generates an array of laser
beams by pumping a Raman shifter with a
Nd:YAG laser (5).  Addition of a Raman
shifter to our initial instrument was
relatively simple and inexpensive and
allowed the generation of laser beams of
different colors throughout the ultra-
violet and visible regions of the
spectrum (5).  This second generation
system also employs a small spectrograph
and an intensified diode array detector
to allow full spectral analysis of
laser-induced fluorescence and scattered
light.  The high-intensity of laser ex-
citation allows rapid emission spectral
acquisition, thus, an excitation-
emission matrix can be acquired rapidly
by sequentially launching each laser
beam into the excitation optical fiber
and recording the emission spectrum.
                                          797

-------
Spectral acquisition, processing and
analysis is performed using compiled
Pascal programs on a 80386 personal
computer.  This computer arrangement
allows rapid formatting and analysis of
the  large EEM  data matrices produced by
this  instrument (e.g. 25 excitation
wavelengths X  700 emission wavelengths
=  17,500 data  points/EEM).
EXPERIMENTAL

Figure  1  shows a block diagram of our
second  generation  instrument.  The third
or  fourth harmonic of a pulsed Nd:YAG
laser  (Quanta-Ray model DCR-11)  is
focussed  into a Raman shifter where much
of  the  beam is converted  into laser
beams of  different colors.  The  YAG
laser was operated at 10  Hz, and pulse
energies  of 35 mJ/pulse at  355 nm and 14
mJ/pulse  at 266 nm were employed. Twenty
different laser beams between 240 and
330 nm  were suitable for  the fluores-
cence excitation of the compounds used
in  this study.  Each Raman-shifted laser
beam was  selected  sequentially by the
dispersion unit and launched into one
end of  a  seven-meter optical fiber.  The
laser light is transported  by the
excitation fiber to the solutions where
it  is absorbed by  the analyte.   An
optical fiber similar to  the excitation
fiber collects and transports analyte
fluorescence and scattered  light back to
the detection system.

Spectra are acquired by dispersing col-
lected  light with a 0.27  m  spectrograph
onto a  cooled diode array detector.  The
emission  spectral bandpass  of this ar-
rangement is 4 nm.  Individual fluores-
cence spectra are  collected, processed,
formatted into excitation-emission
matrices, and analyzed by least-squares
using Turbo Pascal programs.  Least-
squares analysis involves minimizing the
differences between the EEM spectrum and
a linear  combination of the EEM  spectra
of  each of the individual analytes (6).
All  programs for data analysis were
developed in-house and run  on a  personal
computer.   EEM plots shown  in this paper
were obtained using Surfer  software
(Golden Software Inc.).
RESULTS

Figure 2 shows the excitation-emission
matrices of four groundwater pollutants:
the cresol isomers p-cresol and m-
cresol, and the carbamate pesticides
carbaryl and carbofuran.  Figure 3a
shows the fluorescence EEM of a solution
containing a mixture of all four com-
pounds, and Figure 3b shows the EEM of a
mixture containing the compounds m-
cresol, carbofuran, and carbaryl.  All
spectra were obtained with our laser-
based instrument using methanol solu-
tions of 1 ppm or less in concentration.
The total spectral integration time to
acquire these EEMs was less than 5
minutes.  Exposure times for individual
spectra ranged from 0.5-40 seconds.

The results of the least-squares
analysis of the above mixtures are
compared with the actual prepared
concentration in Table 1.  The
concentration of all components were
determined quantitatively within 20 %
for these sub-ppm solutions.  We have
previously analyzed the composition of a
3-component mixture to within 1.5 %
under similar experimental conditions
when the components of the mixture
exhibit less spectral overlap (7).

Table 2 shows our current detection
limits and linear dynamic ranges for the
above pesticides and pollutants.  These
results were obtained from spectra
acquired for 5 seconds upon exciting
solutions at 266 nm and by measuring the
fluorescence intensity at the intensity
maximum.
DISCUSSION

The EEMs in Figure 1 show that three of
the compounds - m-cresol, p-cresol and
carbofuran - have very similar fluores-
cence EEM spectral profiles.  The
absorption and emission spectra of these
three compounds have intensity maxima
within 10 nm of each other.  Also, the
spectral profiles of these compounds are
similar and exhibit little vibronic
structure.  This type of strong simil-
arity in spectral characteristics makes
it relatively difficult to accurately
analyze these compounds by EEM analysis.

Our capability to analyze these pol-
lutants as 3- and 4-component mixtures
is demonstrated in Table 1.  Least-
squares EEM analysis provided a
reasonably accurate determination of
sub-ppm pollutants in simple mixtures
without any prior preconcentration,
extraction, or chromatographic treatment
of the analyte solution.  These results
                                           798

-------
show that we can analyze mixtures of
chemical isomers (m- and p-cresol) in
addition to mixtures of unrelated
pollutants with our instrument.

Table 2 shows that the detection limits
of these pollutants are in the ppb
concentration range and the linear
dynamic extends over 3-4 orders of
magnitude.  Similar results have been
obtained with our instrument for phenol,
dibenzofuran and carbazole (7).

More complex mixtures of fluorescent
chemicals could be analyzed with our
laser EEM instrument by adapting it for
use as a liquid chromatography detector.
All laser beams from the Raman shifter
are produced simultaneously, so each
could serve to excite fluorescence at
the end of a chromatographic column
simultaneously.  By using an array of
excitation and emission optical fibers
and by employing a two-dimensional array
(CCD) detector, all spectra of the EEM
could be acquired simultaneously.  At
least fifteen Raman-shifted laser beams
in UV region of the spectrum are of
sufficient intensity that they could be
employed to obtain fluorescence EEM
spectra in 1-2 seconds.  This instru-
mental design would provide a dramatic
improvement in the ability of our
instrument to analyze complex solutions.
A recent study by the National Institute
of Standards and Technology has found
liquid chromatography with fluorescence
detection to compare favorably with gas
chromatography/mass spectrometry for the
analysis of polycyclic aromatic hydro-
carbons in environmental samples (8).

Our fluorescence EEM instrument has
performed well in laboratory tests.  The
Nd:YAG laser is a solid state laser and
requires little maintenance.  The Raman
shifter is the only other non-linear
optical component in our system, and the
operation parameters of this device have
been characterized previously (4).

An important aspect of the field
compatibility of our instrument is near
absence of moving parts.  The only
moving parts in our instrument are the
YAG harmonic generation crystal mounts,
the prism rotation stage (to select the
color of the laser beam), and the fiber
positioning elements (to allow efficient
launching of the laser beam into the
excitation optical fiber).  Relatively
large optical fibers are employed in
this instrument (0.60 mm core diameter)
to simplify the alignment of the laser
beam with the excitation optical fiber.

The power requirements of our current
instrument are 3000 W, and 9 ft3/hour of
dry nitrogen gas is used to purge
moisture sensitive optical components.
Approximately half of the weight of our
instrument (450 Ibs) is found in the
Nd:YAG laser (220 Ibs).

The components of our current fluor-
escence EEM instrument cost about
$50,000, which is well below the total
cost of a transportable gas chromat-
ography-mass spectrometer ($300,000).
Also, unlike GC-MS, our instrument has
the capability to directly analyze non-
volatile samples in aqueous solutions.
CONCLUSIONS

The ability of our laser-excited
fluorescence EEM instrument to analyze
mixtures of cresol isomers and pesti-
cides directly at sub-ppm concentrations
in methanol was demonstrated.  Detection
limits were in the ppb concentration
range and the linear dynamic range was
3-4 orders of magnitude.  This
instrument is a promising device for
direct in-situ screening of groundwater
pollutants or for interfacing with a
liquid chromatograph for EEM spectral
analysis of chromatographic eluents.
ACKNOWLEDGEMENT

The authors greatfully acknowledge the
assistance of Anthony Bevilacqua, George
Jarvis, and Mark Regina in setting up
much of the equipment involved in these
experiments.
REFERENCES

1. Chudyk, W.A., Carrabba, M.M., Jarvis,
   G.B., Kenny, J.E.  Anal. Chem., 57,
   1985, 1237.

2. Kenny, J.E., Jarvis, G.B., Chudyk,
   W.A., Pohlig, K.O.  Anal. Instrum.,
   16.  1987, 423.

3. Chudyk, W.A., Kenny, J.E., Jarvis,
   J.B., Pohlig, K.O.  InTech.  34. 1987,
   53.

4. Jarvis, G.B., Ph.D. dissertation,
   Tufts University, Medford, MA.  1989.
                                          799

-------
5.  Kenny, J.E.,  Jarvis,  J.B., Xu, H.
   Proceedings of the First Internation-
   al Symposium of Field Screening
   Methods for Hazardous Waste Site
   Investigations, U.S.  EPA, 1988, 133.

6.  Warner, I.M., Davidson, E.R.,
   Christian, G.D.  Anal. Chem.,  49,
   1977, 2155.

7.  Taylor, T.A., Xu, H., Bevilacqua,
   A.C., Kenny,  J.E.  In preparation for
   submission to Analytical Chemistry.

8.  Wise, S.A., Hilpert,  L.R., Byrd,
   G.D., May, W.E.  Polycyc. Arom.
   Comp.. 1,  1990, 81.
          TABLE 1.  Least-Squares EEM Analysis Results for Two Mixtures of
                   Pollutants.
          Mixture   Component

                    m-cresol
             1      p-cresol
                    carbofuran
                    carbaryl

                    m-cresol
             2      carbofuran
                    carbaryl
          Prepared
        Concentration
       X 10'6 M  (ppm)

         3.21 (0.44)
         2.58 (0.35)
         2.96 (0.84)
         0.795(0.20)

         3.21 (0.44)
         2.96 (0.84)"
         0.795(0.20)
       Least-Squares
        Calculated
       Concentration
         X 10'6 M

           3.20
           2.27
           3.53
           0.933

           3.75
           2.88
           0.838
              Percent
               Error

                0.03
                12.0
                19.3
                17.3

                16.8
                2.70
                5.41
          TABLE 2.  Detection Limits and Linear Dynamic Ranges of
                   Pollutants.
           Compound

           m-cresol
           p-cresol
           carbofuran
           carbaryl
 Detection
   Limit,
X 10'7 M  (ppm)

2.0  (0.030)
2.0  (0.030)
3.0  (0.085)
2.0  (0.050)
Linear Dynamic
 Range, ppm
 0.030
 0.030
 0.085
 0.050
10
10
 5
 2
                                           800

-------
Figure 1: BLOCK DIAGRAM OF SECOND GENERATION INSTRUMENT
»•••••••••••••••••••••••»•«
:
Nd:YAG Laser ^ ]
i
I
I
^^^^^^^^^^^^
Power
Supply
«^^«>ji^_«
L

CONNECTIONS:
	 electrical
'>SN* optical



Raman ^^ Dispersion
Shifter Unit
*
Optical
Fibers ^^
4_l 	




Ng Gas


Detection
System
^^^^^^^
\

Computer



                            Sample

-------
Figure 2:  EEM SPECTRA  OF POLLUTANTS
                                  m-cresol
   p—cresol

     CH3
                                   0 H
                                   II I
                                  0-C-N-CH,
   0 H
   II I
 0-C-N-CH,
                  802

-------
Figure 3: EEM SPECTRA OF MIXTURES
 V
*-
         ' n
              803

-------
                                  Analysis of Total Polyaromatic Hydrocarbon Using
                                       Ultraviolet-Fluorescence Spectrometiy
                          T.L. Theis, A.G. Collins, P.J. Monsour, S.G. Pavlostathis, C.D. Theis
                                               Rowley Laboratories
                                                Clarkson University
                                                Potsdam, NY 13699
      The first step in the remediation of Manufactured Gas
Plant (MGP) sites is an estimate of the extent and degree
of contamination.  This research has been concerned with
the development of a procedure for the field determination
of total concentration of polyaromatic hydrocarbons (PAH)
in contaminated soils at MGP sites.  It is based upon the
principle  of  ultraviolet  fluorescence  whereby  certain
chemical substances, PAH among them, emit a portion of
incident  UV  radiation  at   longer  wavelengths.    The
experimental program which has been undertaken contains
three  elements:  optimization  of  the  PAH  extraction
technique, miminization of interferences, and quantitation
of  PAH  using  UV-flourescence  at  several  excitation
wavelengths in order to control nonuniform effects.-

      Soils  were supplied from  three MGP  sites,  and
consisted  of one or two  samples considered to vary in
their degree of contamination, as well as uncontaminated
samples.   Soxhlet extraction (benzene) followed by gas
chromatographic measurement of 16 PAH compounds was
performed according to EPA method 610. These values were
used as baseline information against which comparisons of
solvent extraction efficiencies were made.   Five solvents
were evaluated  for  their  extraction efficiency: benzene,
hexane, acetonitrile,  2-propanol, and  acetone.   Results
indicated  that  both  benzene  and  hexane  extracted the
maximum amount of PAH, with hexane showing a small
advantage for lower-ring  and benzene an  advantage for
higher-ring PAH. The differences, however,  were small and
hexane was chosen as the solvent of choice based on safety
considerations.

      Three  factors  affecting dispersal  and extraction of
PAH into hexane were evaluated: sample  size (actually
solvent/sample  ratio),  mixing  time  (with and without
sonication),  and effects of  added moisture followed  by
blending.    The primary consideration   for  maximum
extraction efficiency was  the dispersal of  waste  particles
in the medium.  A further constraint was the fixed size of
extraction tubes, 50 ml, which was  considered  to  be a
convenient size for  manipulations in  the  field.  Under
these conditions, a sample size  of 2-4 grams (wet weight)
in 50 ml of hexane gave maximum PAH removal from the
 soil.

      It was anticipated that the dispersal  of particles,
 and  hence extraction efficiency, might  be enhanced  by
 sonication, however,  this  proved  not to  be the  case.
 Figure   1  shows  results  for  three  separate  extraction
 experiments, no sonication, and sonication  for 10 to  20
 minutes.  Total mixing time was the same, 90 minutes. The
 data reveal  that sonication  does result in a  more rapid
 initial release  of PAH,  however, this  effect is  negated
 and  in  fact reversed  after the  sonication period.    As
 there was  no apparent  advantage to the use of sonication
 in  improving extraction efficiency or  decreasing the total
 time, its use was discontinued.  Figure 1 also shows the
 minimum  mixing time  (without  sonication) was about  70
 minutes.   This  was further  verified in other experiments
 resulting  in  the  adoption  of this mixing  time  as  the
 standard value.

      Somewhat surprisingly, it was found that the addition
 of  distilled water to samples, followed by blending, gave
 excellent  dispersal  of  particles resulting  in acceptable
 PAH extraction.  This  is  illustrated  in  Figure 2  which
 shows total  PAH extracted as a  function of  the  water
 added/sample ratio.   Blending also resulted in a greater
 sample  uniformity, thus replicate variability (also  shown
 in Figure 2)  is improved.  In effect the addition of water
 acts to yield a more uniform moisture content, regardless
 of the initial (field) moisture content of the sample.

      There are three primary sources of error in measuring
 total PAH by UV fluorescence: background fluorescence  of
 non-PAH  substances,  different  distribution   of   PAH
 compounds  from  sample  to  sample, and  fluorescence
 quenching  effects (either direct or "concentration").   The
 approach used to minimize quenching  was dilution  of the
 hexane extract (using more  hexane)  until the interference
 was eliminated  and  spike recoveries gave a  clear linear
response  against the fluorometric reading.  Figure 3  shows
the   fluorescence  of   a  hexane   extract  for  several
dilutions.   As  the sample  is  diluted the response first
increases,  as  the effects of quenching are  lessened, and
then declines, as  the  fluorescing substances themselves
                                                        805

-------
are diluted.  Quenching effects can be quite  subtle;  the
important factor  is to make  fluorescence readings at  the
dilution level which  gives a linear recovery curve.   For
example, Figure  4 shows the results of a spike recovery
study for a mixture of PAH added to 1:10,000 dilution of
the  hexane   extract.    The  x-axis   intercept  of  the
regression  curve represents  the  concentration of PAH in
the sample.  In  general it has been found  that it is best
to make measurements  at the highest dilution  for which
recoveries can be made accurately, usually 10  or 10 .

      The  accompanying table presents  a summary of the
results of  the research.   Gas chromatographic  analysis of
Soxhlet  and  hexane extracts  are  generally comparable
indicating  acceptable recoveries  using the hexane method.
As state3 previously,  fluorescence measurements were made
at several  excitation and emission wavelengths.  It was
found that  results  were  relatively  insensitive  to  the
emission wavelength, 410 nm giving acceptable responses
throughout. The excitation wavelength, however, was very
important  in determining the response  of the instrument
and appeared to  depend on the distribution of PAH in the
sample extract.   Unfortunately, no  one  wavelength was
satisfactory.  As can be seen from results for sample 1C
the  fluorescence method  can  be quite  accurate  when
interferences are  lacking.   Results for  samples 1A,  IB,
2A, 3A, and 3B are considered to be acceptable with errors
of  12,  3.7,   5.6,   1.5,  and 5.3  percent,  respectively,
relative to the  gas chromatographic analysis of the hexane
extract.

      The  accuracy  of  the  fluorescence  method  at each
excitation   wavelength   was  assessed  through   linear
regression  of the UV-fluorescence values against the  gas
chromatographic analysis  for total PAH.  The results of
these regressions were used  to compute the relative error
of  fluorescence  at  any   given  level  of  total  PAH
concentration.   The  results  are  shown in Figure 5.  One
obvious feature  is  the radically different character  of
the  error  at  250 nm  in  comparison  with other error
curves.  The error analysis at other wavelengths indicate
more  well-balanced  curves,  the error  decreasing  as  the
concentration  of  PAH  increases.     It   appears   that
measurements  made  at  280  nm excitation  give the most
consistent  accuracy.  Averaging the readings for selected
wavelengths,  as  shown,  does   not substantially improve
accuracy.   It  is  interesting to note that the  errors for
most of the wavelengths at 100  mg/kg, the  target goal for
the   detection   limit,   are   nearly   the  same,   about
twenty-five percent.   Given  the intended purpose of the
method, this is considered acceptable.

Acknowledgements

      This research  was supported by  the  Gas Research
Institute, David G. Linz, Manager, Land and Water Quality
Research.
               Sununaiy of UV-Fluoreicence Reiulu*

Sample #
1C (Clan)
M
IB
2A
3A
3B

Soxhlet
Extraction
7.32
3539
4064
2936
675
45.54

Heune
Extraction
4.03
3398
3353
2916
794
39.18
UV-Fluoretcence
250/410*
6.2610.15
5436±147
4725±178
6375*226
96fi±61
41.25±0
280/410
6.910.15
3817198
2449198
3273±106
782±14
67.513
315/410
7.5110.24
48511192
2650148
3082158
890132
9515
360/410
7.3210.29
44131268
34761108
25951117
1204152
150±5
'All values are mg/kg dry weight for total PAH.
bFirst number ii excitation wavelength.
 Second number U emission wavelength.
                                                           806

-------
     o»
                 o no  sonication

                 A 10  minutes sonication

                 o 20  minutes sonication
                                          A
                                          a
c
•J3
|
V
o
c
o
o

3C
<£
a.

1500-
.
1000-
«
"
•

500-
f


0-
C






A


o


) 20
A 0

a
A

a


o °
0

D

40 60 80 1C
                    Time of  Extraction  in  Hexane (minutes)



Figure  1:  Comparison of Alternative Hexane Extraction Procedures
      4000-
0»3000-





*-, 2900-




^2000-

Q



-r 1500-
      1000-
    O
    •+*
    O
       500-5,
         0.0
                                  Soxhlet Value
                                     A


                                     A
                                     Site  1   Sample A

                                     Initial Moisture  Content:   60.255
              0.5      1.0      1.5      2.0      2.5      3.0

                  Mass  of Water  Added  / Mass  of Sample
3.5
Figure 2:  Dependence of PAH Extraction Efficieny on Sample Moisture Content
                                    807

-------
        3-3 -i
                                               Site  1   Sample B
                                               Hexane Extract
                                               Excitation:  315 nm
                                               Emission:   410  nm
        0.0
                                    I' i
                                   2           3
                                   Log  Dilution
 Figure 3:   Impact  of  Sample Dilution on PAH Fluorescence
       230-,
       200-
       1 SO-
 OT
•f
'c

 u
•c
•4J
 0)
 E 100
 8
 o

C
   50-
       -0.
      02  -
                                     Site 1   Sample B
                                     Excitation:  315 nm
                                     Emission:  410 nm
                                     1:10000 Dilution of Hexone Extract
_    ii 11 ii 11 n inn ii| in i in iiiniii 11 ii| ii inn 11|
0.00    0.01    0.02    0.03     0.04     0.05
   Total PAH Added  (ug/mL)
Figure 4:   PAH Determination by Method of Standard Additions
                                    808

-------
                   100-.
s
co
                         '»' I I I I  I I I I < I I I 1 I  I I I I I I I I I I I I  I I I I I I 1 I I
                                                                     250 nm
                                                                     Ave. (250+280+315)
                                                                     315 nm
                                                                     360 nm
                                                                     Ave. (280+315)
                                                                     280 nm
                         log  PAH  concentration   (mg/kg)
                            Figure 5. Error Analysis of the UV-Fluorescence Method

-------
                   ON-SITE ANALYSIS OF CHLORINATED SOLVENTS IN GROUNDWATER
                                           BY PURGE AND TRAP GC
                 Stephen A. Turner, Daniel Twomey, Jr., Thomas L. Francoeur and Brian K. Butler
                                ABB Environmental Services, Inc., Portland, Maine
INTRODUCTION

Historically, analysis of volatile organic compounds via purge
and trap gas chromatography (GC) has been conducted in
laboratory settings where controls  were strictly monitored.
Recently, however, increased reliability of GC instrumentation
combined  with the adaptation of quality control procedures
make this a suitable  analytical  technique  for  successful
incorporation into field sampling activities.  This  screening
technique  produces  an  accurate  "real  time"  profile  of
groundwater contamination which can be subsequently used in
deciding the  placement of  monitoring  well screens.   This
approach not only aids in detecting the most contaminated zone
within a given aquifer, but it also serves to reduce field costs
associated with traditional "cluster" type well  installations.
Costs are further reduced as only essential samples are selected
for Contract Laboratory Program (CLP) analysis.

This study involved an investigation at a National Priorities
List  site which  produced approximately 200  groundwater
samples from screened auger borings.  These samples were
analyzed   on-site   for   selected   chlorinated    solvents
(1,1-dichloroethane, 1,2-dichloroethene, trichloroethene, and
tetrachloroethene) by purge and  trap GC.  Field  screening
results  were used  to   determine  the  distribution  of
contamination; these results were  confirmed by submitting
selected samples  to a laboratory to  be analyzed according to
CLP protocols.

The following sections describe how the field screening data
were  collected and subsequently used  in determining  the
placement  of groundwater monitoring  well  screens.   In
addition, the  quality of these data were evaluated  as to their
precision and accuracy through rigorous statistical modeling.
Finally, a general cost comparison  of field screening versus
CLP  analysis establishes the economic practicality  of this
technique.
TECHNICAL APPROACH

This field screening procedure was based on USEPA Method
601.  A GC equipped with a purge and trap device and an
electrolytic conductivity detector (ELCD) was set up  in a field
trailer in close proximity  to the study area.   A wide-bore
capillary column was used for compound separation, and the
total cycle time was approximately 22 minutes.  Under these
conditions approximately 20 samples could be screened in a 12
hour shift  (not  including QC  samples).   Quality control
included initial calibration runs, continuing calibration (at the
start and end of each  day), method blanks, and  surrogate
spikes (bromofluorobenzene).  A description of these quality
control measures is provided in the following section. Analyte
detection limits  as determined  by statistical analysts were
approximately 1 pg/L.
Temperature Programs and System Operating Conditions:

              Purge and Trap:
       o Purge:
       o Purge Flow:
       o Desorb Preheat:
       o Desorb:
       o Bake:
6 minutes at 35 °C (max)
28 - 33 mL/minute
175°C
3 minutes at 180°C
6 minutes at 225°C
              Gas Chromatograoh/ELCD:
          Column Flow (He):
          Makeup Gas (He):
          Hydrogen:
          Initial Temperature:
          Initial Time:
          Rate:
       o  Final Temperature:
       7-10 mL/minute
       30 mL/minute
       80 mL/minute
       35°C
       3 minutes
       8eC/minute
       135°C
                                                                       o  Approx. GC Run Time:    15.5 minutes
                                                           811

-------
With respect to drilling activities, groundwater samples were
collected using 4.25-inch inside diameter hollow stem augers.
The lead auger was plugged and modified with the installation
of  0.010-inch stainless  steel well  screen  sections (i.e.  a
screened lead  auger).   Monitoring wells  were subsequently
installed in borings  where  the highest concentrations  of
chlorinated  solvent were identified as established by  field
purge and trap GC results.
QUALITY ASSURANCE/QUALITY CONTROL

QA/QC  was  performed to a  degree sufficient to evaluate
general  data  quality  and  system performance while  not
inhibiting  the ability  to provide  real-time data and  high
throughput.  In general, under normal operating conditions,
five analytical runs per day were devoted to QA/QC.

o  Initial Calibration:   A  minimum of  one 3-point  initial
calibration  was  performed  and  percent relative standard
deviations (%RSD) were calculated  for  each  analyte.  An
average  response factor was used to calculate sample analyte
concentrations  if   %RSD   <  30%.   Otherwise  analyte
concentrations were derived from the corresponding calibration
curve.  The calibration range was 5 to 50 /xg/L

 °  Continuing Calibrations: Continuing  calibrations (run  at
approximately mid-level concentration) were performed  at a
minimum of every 12 hours and at the beginning and end of
each analytical day. Percent deviations (%D) of less than 35%
were  considered  acceptable for  verification of  the  initial
calibration curve.   Failure of this  critera necessitated the
reconstruction of the 3-point initial calibration curve.

 o  Blanks:  Method blanks were performed daily before the
 analysis of any groundwater samples.   In addition, system
 cleaning blanks were run after  any  groundwater  sample
containing any single analyte at a concentration exceeding five
times the highest calibration level.

 o  Surrogate Spikes: Bromofluorobenzene was spiked in  each
analytical run to evaluate sample recoveries and matrix effects.
However, no actions  were implemented  as a result of poor
surrogate recoveries.

o  Independent Check Standards:  To verify the quality of the
calibration  standards  and to evaluate  standard preparation
procedures, two complete sets of  standards were purchased
from two independent chemical suppliers.  After completing
the calibration curve with one set of standards, a blank spike
with a known concentration (at or near  the mid calibration
level) of the second set of standards was  analyzed to confirm
analytical accuracy.
GROUNDWATER INVESTIGATION

This  approach  was  first  incorporated  into  a  remedial
investigation conducted at a military industrial complex where
previous   studies    showed   groundwater   to   contain
1,2-dichloroethene,  trichloroethene, and  tetrachloroethene.
The subsurface geology at this site was characterized as glacial
outwash consisting of fine to medium sand, and the water table
was located approximately 85 feet below the ground surface.
The purpose of this remedial investigation was to determine
the extent of groundwater contamination migrating from the
site.    In  order  to  properly  delineate  the   extent  of
contamination,   monitoring  well  fences  were  installed
perpendicular to the plume axis.  Three of these monitoring
well fences were required to  sufficiently characterize  the
contaminant plume as to its horizontal and vertical boundaries.
In addition to isolating the limits of the plume, field screening
data aided in defining the  intervals of highest concentration
widiin the aquifer. Where possible within a given monitoring
well,  screens were  installed to intercept the groundwater at
depths where the highest solvent concentration was identified.
At  the lateral extent  of the plume, where screening results
detected no target analytes at any interval, well screens were
installed   at  depths   similar  to  those  where  the  highest
concentration of contaminants  were  identified  in adjacent
wells.  The plume cross section is shown if Figure 1.

Since the local geology  was comprised almost exclusively of
fine outwash  sands  and  provided for relatively fast and
efficient  drilling, the investigation  was  designed to  fully
envelop the contaminant plume by collecting a large number
of samples for field screening. By doing so, a profile defining
both the minimun concentrations around the edge of the plume
edge  and  the maximum concentration along its axis was
generated.  This method of plume isolation results in very few
data gaps.  This volume of data provides tremendous benefit
to interpretation of groundwater contamination as compared to
the traditional approach to well installation which is generally
directed by historical  data and produces a limited number of
data points.

The groundwater exploration program consisted of 29 screened
auger borings  from  which 190 groundwater samples were
collected at five to ten foot intervals and analyzed by field purge
and trap GC. Sample depths ranged from the water table to 204
feet below the ground surface,  and all screened auger borings
were completed with the installation of monitoring wells with
5-foot 0.010-inch slot PVC screens.  Field GC analyses of the
screened  auger  samples  determined  the geometry  of the
groundwater plume.  These results indicated the solvent plume
exceeded  3,000 feet in  width  and 7,000 feet in  length.
Subsequent CLP laboratory analysis of groundwater samples for
target compound list (TCL) volatile organic compounds (VOCs)
confirmed the field GC results as determined through statistical
analysis (to be discussed in the following section).
                                                              812

-------
        Plume Cross Section B-B'
     West
     B
East
  B1
       i      in      nil
                       Figure 1

           Contaminant Plume Cross Section

CLP  VOC  analysis   indicated  the  samples   contained
1,2-dichloroethene concentrations from  less than  1 /tg/L to
54 /tg/L. The predominant site contaminants, trichloroethene
and tetrachloroethene, were also identified at concentrations as
high as  670 ng/L and 430 /tg/L, respectively.  Overall, the
field GC program determined that the plume boundaries were
sharp,  and  the  plume  consisted  of  two  parallel  high
concentration (>  100 M5/L) lobes.
COMPARISON   OF   FIELD  AND   LABORATORY
RESULTS

Field GC screening data are used in real-time to facilitate field
decisions.  Additionally, these data, in conjunction with CLP
results, provide information for site characterization. In order
to assure that field results  are of sufficient quality for use in
interpretation of groundwater  contamination, the field  GC
screening procedure was calibrated against laboratory CLP
methods. A statistical evaluation was therefore performed to
determine the usability of field  data.

A set of 35 replicate samples was collected and submitted for
both CLP  and field GC analyses.  The statistical evaluation
was  based  on  five   replicates  of  seven  samples  (five
groundwater samples from three screened auger borings and
two blind  spikes) analyzed both by field GC and by three
 independent laboratories.  Each independent laboratory result
was considered to be a "true" value, and the field GC results
were compared to each laboratory result individually to test the
accuracy of the field results.  Target compound concentrations
in these samples ranged from non-detect to 125 /tg/L in all
samples analyzed.

The field GC results, laboratory results, and spike concentrations
represent paired data points.  To determine if field GC results
adequately represent the true concentration in the sample, a
statistical test (paired t-Test) was performed under the hypothesis
that field  GC  results  equal  laboratory   results  or  spike
concentrations. Testing this hypothesis is equivalent to testing
whether  the differences between field results and laboratory
results are significantly different from zero.  The paired t-Test
results   indicate that field GC  results are not significantly
different  than  either  laboratory  results   or actual  spike
concentrations.  At the 95% confidence  level the original
hypothesisis not  rejected.    In other  words,  there was no
statistically significant evidence  to indicate that  there  is a
difference between field and laboratory measurements.

In addition to the tests for zero differences between field GC and
laboratory  results  and spike concentrations,  a polynomial
regression analysis  was performed to evaluate the correlation
between  field  and laboratory data.  This procedure  involved
defining  field  GC results as the dependent variable and
laboratory results and spike concentrations as the independent
variable. The regression model used was
                                            c + ax  +
                                                      bx*
                          where:  y = field GC results,
                                  x = laboratory results / spike concentrations,
                                  a,b,c = constant coefficients.
                   The results of this analysis indicate a strong correlation between
                   field GC results and laboratory results/spike concentration with
                   an adjusted multiple rs of 0.965.  As can be seen in Figure 2, a
                   graph of the data supports this assertion. The middle curved line
                   in the graph is the least squares quadratic best fit to the data.
                   This best fit line appears to level off at around 100 /tg/L; this
                   effect results from  the narrow  dynamic range of the ELCD as
                   compared to CLP GC/MS techniques (i.e. the linear range of the
                   ELCD was exceeded). The upper and lower curved lines are the
                   approximate 99% confidence bounds about the best fit line.

                   The leveling off characteristic of the best fit line indicates that
                   concentrations  above  this  range  may  be underestimated.
                   However, the calibration was designed to accurately quantitate
                   analyte concentrations over a relatively low range  of  1 to
                   50 /tg/L thereby producing an accurate description of the edge
                   (lowest concentrations) of the contaminant plume.  Although
                   beneficial,  accurate  quantitation   of  analytes   at   high
                   concentrations was not essential to the success of this  field
                   program.  When anticipated,  this problem  was remedied by
                   diluting highly contaminated samples by an appropriate factor,
                   thereby achieving the linear range of the calibration.
                                                             813

-------
    150
    100  -
 -J
 W
 E
                                       100
                                                      150
                              LAB
                         Figure 2

                Quadratic Regression Fit of
             Field v. Lab Data (all units jig/L)
The paired t-Test  and the polynomial  regression analyses
demonstrate a strong correlation between field and laboratory
results. Therefore, this screening procedure,  combined with
the confirmation of selected samples using CLP protocols,
provides data of sufficient quality for use in plume delineation.

COMPARISON OF FIELD AND LAB COSTS

In addition to the obvious field operations advantages field
screening  presents  by producing  accurate "real-time"  data,
substantial savings  are realized  by conducting field analyses
and  shortening  the  duration  of the  investigation.    By
performing analyses  of  groundwater samples in the field,
analytical  costs are minimized. The expense of purge and trap
volatile analysis performed by a CLP laboratory ranges from
$200 to $400 per sample.  In contrast, field screening analysis
can be performed at between $50 to $150 per sample.  The
final per  sample cost  of field screening analysis is  largely
dependent on the volume of samples requiring analysis. Since
the  cost  of  mobilization and  demobilization  of a  field
laboratory is constant  regardless of the  volume  of samples,
larger field programs tend to have the lowest per sample field
screening  analytical costs.  If all 200 samples generated from
this field program were analyzed by a CLP laboratory the total
analytical  costs (assuming $300/sample)  would run $60,000.
Field analysis at $75/sample cost $15,000,  and the additional
confirmatory laboratory  analyses  performed  in  conjunction
with the field program (29 samples), brings  total analytical
cost to $23,700.  This represents a savings of $36,300 over the
traditional approach.
Although substantial,  the overall  program  savings  are not
restricted to reduced analytical costs. The traditional approach
to site investigation inevitably requires multiple site visits many
months apart. These additional visits double and triple program
costs relative to  a one  visit  investigation.   The costs of
mobilization,  sampling, shipping ($100 to $300/cooler), and
peripheral expenses are also significantly reduced in conjunction
with a one visit investigation that field screening provides.
CONCLUSIONS

While the cost savings are an important aspect of field screening
analysis, the real-time data acquisition is fundamental to the
success of the field program.  Typical turn-around time for
laboratory data is one month, which relegates the placement of
monitoring   well  screens  and   surface   and   subsurface
contamination delineation to educated guesses based on historical
data. The successes of these field investigations are not known
until the field events have ended; multiple iterations may be
required to complete remedial investigations. In fact, the success
of traditional field  investigations may never be  fully realized
since the presence of many data gaps is inherent to the approach.
At this military installation, it is quite possible that the dual lobe
character of the contaminant plume would have been overlooked
during a multiple phase program dependent on the evaluation of
CLP data for well placement information.   The use of field
analysis allows many critical difficulties to be overcome. Wells
were placed at depths of the highest contaminant concentration
based on  real time information which provided  an accurate
determination of the contaminant plume.  Additional wells can
be installed if field screening data indicate the need to fill in data
gaps; the real-time information allows these installations to occur
while drilling crews are still on-site, thus eliminating additional
•mobilization charges.

It should be  stressed that using  field GC screening as  an
analytical tool is not designed to as a substitute for CLP analysis.
One of the principle reasons the technique is so useful and time
efficient in the field is much of the QA/QC  associated with its
CLP analog has either been scaled back or eliminated. Field GC
screening should be regarded as merely a  single factor in a
holistic approach to conducting site and remedial  investigations
rather than a stand alone analytical method.

The cost effectiveness, time savings, and quality of field analysis
combine to and demonstrate the utility of purge and trap GC
analyses in the field.  Analytical and overall program costs are
significantly reduced while the sample database is increased,
providing  information  critical  for  completing  remedial
investigations in a timely fashion.
                                                              814

-------
                 U.S. EPA EVALUATION OF TWO PENTACHLOROPHENOL IMMUNOASSAY SYSTEMS
                  J. M. Van Emon
         U.S. Environmental Protection Agency
     Environmental Monitoring Systems Laboratory
                 Las Vegas, Nevada
     R. W. Gerlach, Ft. J. White, and M. E. Silverstein
       Lockheed Engineering & Sciences Company
                  Las Vegas, Nevada
INTRODUCTION

The Superfund Amendments and Reauthorization Act of
1986 (SARA)  charged the  U.S. Environmental Protection
Agency (EPA) with effecting more timely and cost-effective
remedies at the nation's Superfund sites. The development
of improved field screening methods that yield immediate or
short-turnaround environmental data can result in major cost
savings per monitoring site. The EPA Superfund Innovative
Technology Evaluation (SITE) Program was established in
response to legislation within SARA.  The EPA mandate to
research,  evaluate,  test,   develop,  and   demonstrate
alternative   or  innovative  treatment   technologies  is
accomplished within the SITE program. One aspect of the
SITE program focuses on monitoring and  measurement
technologies for contaminants occurring at hazardous waste
sites.

This report presents the results of a demonstration of
individual field and laboratory-based  immunoassays for the
detection and measurement of pentachlorophenol (POP) at
hazardous  waste sites.   POP  has  beneficial uses in
agriculture and as a wood preservative; however, there are
risks associated with its use.  Numerous sites on the EPA
National Priorities List contain hazardous levels of POP.
Rapid  and  inexpensive  monitoring  and  measurement
technology  is  useful for  monitoring   the  extent  of
contamination and the  effectiveness  of  remediation.
Immunoassays are gaining recognition as one cost-effective
alternative to chromatographic and spectroscopic analytical
procedures in large-scale environmental monitoring studies.
Immunoassays can be used in the field, have the capacity
for an increased sample throughput, and can be used to
screen and rank samples for analysis by more traditional
analytical methods. Although specific immunoassays have
been  developed for  hazardous  compounds of Agency
interest, many of these systems have not been properly
evaluated for environmental matrices.  Agency restraint in
utilizing immunoassay technology is partly due to the scarcity
of immunoassay methods that have been fully evaluated for
environmental applications.
MATERIALS AND METHODS

Westinghouse  Bio-Analytic  Systems  (WBAS), Rockville,
Maryland,  developed  two  immunoassay  technologies
appropriate for screening drinking water, surface water, and
ground water samples to detect and measure the presence of
POP.    One  technology  is a  96-well  microtiter  plate
immunoassay designed to accommodate the high sample
capacity that might be encountered in a laboratory setting.
The plate immunoassay is based upon a rat  monoclonal
antibody selective for POP. The method has a stated sensitivity
of 30 ppb and a linear dynamic range from 30 to 400 ppb.
Although the procedure involves an overnight incubation step,
analysis requires less than 0.5 hour per sample.

The second technology is an 8-well immunoassay kit designed
to provide rapid, semi-quantitative analysis for PCP in the field,
for example, at hazardous waste sites. The kit immunoassay
is based upon rabbit polyclonal antisera and requires only
about 30 minutes per run. The kit immunoassay has a linear
dynamic range of 3 to 40 ppb  and a stated detection limit of 3
to 5 ppb for water.

These technologies were submitted to the EPA Environmental
Monitoring Systems Laboratory-Las Vegas (EMSL-LV) for
evaluation. The study was conducted in two phases. The first
phase  evaluated the plate immunoassay under laboratory
conditions; the second phase focused on the kit immunoassay
under field conditions.
                                                       815

-------
EXPERIMENTAL PROCEDURES
In the first phase of the study the plate immunoassay was
evaluated  using spiked environmental water samples (i.e.,
drinking water, surface water,  and  ground water).   A
methods comparison was conducted between the plate
immunoassay and a gas chromatography (GC) detection
protocol, as described in EPA Method 604.  Extracts were
prepared following EPA Method 604 and quantified by both
the plate immunoassay  and GC.  Extracts from a simple
solid-phase extraction technique, developed by WBAS, were
also analyzed by both the plate immunoassay and GC.  For
relatively clean water samples, the plate immunoassay can
be  run without an extraction.  Thus, unextracted samples
were also analyzed directly by the plate immunoassay. The
direct plate immunoassay data were compared to the GC
results obtained using the solid-phase and EPA Method 604
extracts.

The second phase of the study consisted of evaluating the
kit  immunoassay in a field demonstration under the SITE
program. The field demonstration occurred at the MacGillis
and Gibbs Superfund Site in New Brighton, Minnesota. The
kit immunoassay demonstration was conducted  in tandem
with another  SITE demonstration of  a technology  to
biodegrade  PCP  (BioTrol  Aqueous Treatment Systems,
Chaska, Minnesota). Though the majority of the SITE study
was  directed towards the  demonstration of  the  kit
immunoassay, part of the  study also evaluated the plate
immunoassay, because it can be performed under the same
field- or mobile-laboratory conditions as the semiquantitath/e
field analysis kit.

Samples consisted of (1) raw ground water known to contain
PCP,  (2) ground water treated with nutrients prior to the
application of a bioremediation technology, and (3) effluent
from the bioremediation process. The kit immunoassay was
performed   by  personnel  from  Science  Applications
International Corporation at the field demonstration site, and
by  personnel from  the EMSL-LV and  WBAS  in  their
respective laboratories.  Splits of these samples were also
analyzed with the plate immunoassay at the EMSL-LV and
WBAS.
RESULTS

The evaluation results for the first phase showed no practical
difference among:  (1) the plate immunoassay and GC
detection of Method 604 extracts, (2) the plate immunoassay
and  GC detection  of  solid-phase extracts,  (3) analysis
laboratories for the WBAS solid-phase and EPA Method 604
extraction protocols followed by  immunoassay detection,
and  (4) the  precision  of  the  direct  plate immunoassay
obtained by the two laboratories.  Figure 1 illustrates the
comparability between the immunoassay and GC results.
This first-phase evaluation generated a  9 percent  false
   3.2
   2.0

   1.6-

   1.2-

   0.6-

   0.4-

    0
SOLID PHASE EXTRACT
  A turtle* Wiur
  O Drinking WMor
  O around *•'•'
METHOD 604 EXTRACT
  AturlMO WMW
  Q Drinking W*t«r
  O Ormiritf W.l.r
          0.4
               0.8    1.2    1.6   2.0    2.4
                  OC Amlyslt RoiulU (ppm)
                                          2.1
                                               3.2
 Figure 1.  EMSL-LV Plate Immunoassay vs GC Method 604
          Results

positive rate (n = 115) and a 0 percent false negative rate
(n = 192). The performance of the direct plate immunoassay
was not clearly demonstrated by the results of this study.
However, with additional testing, it is expected that this
particular mode of operation can provide acceptable results in
the field as well as in a laboratory to provide a quantitative,
high-sample throughput analytical methodology. The minimum
detectable  level of  PCP  in the plate  immunoassay,
approximately 30 parts per billion (ppb) for clean environmental
water samples, is appropriate for regulatory use.

The phase-two method comparison was conducted between
the two immunoassay methods and EPA Method 8270 (a gas
chromatography/mass spectrometry (GC/MS) procedure). For
the kit immunoassay, 87.5 percent of the  results for the
samples taken prior to the bioremediation process were within
a factor of two of the GC/MS results. When these same
samples were analyzed  by the  plate immunoassay,  94.4
percent were within the f actor-of-two window. Figure 2 shows
the results for on-site kit immunoassay and EMSL-LV plate
immunoassay  compared  to the GC/MS   results.  Both
immunoassay analyses of the low-level (effluent) samples gave
a comparable range of results, but they were biased high by
up to a factor of 2.5 in comparison to GC/MS analysis results.
                                                                    KIT IA vs GC/MS      Plate IA vs GC/MS
                                                                                       501
       I ' 10 ' 20 ' 3'0 ' 40 ' 50 ' 60   0   10 ' 20 ' 30 ' 40 ' SO
                M*.n PCP Cone, (ppm) by GC/MS

          InMu.nl Srnnpl.  OEI.hiwH Simplo  AlntlmfH S.mpl* [
  Figure 2. On-site Kit and EMSL-LV Plate Immunoassay vs
          GC/MS Results
                                                         816

-------
Ranges in ppm for the low-level samples were 0.008 to 0.91
(GC/MS), 0.20 to 2.27 (kit immunoassay), and 0.31 to 1.82
(plate immunoassay).  This bias, in view of the rank order of
relative responses, does not significantly impact the utility of
the  technology  as  a  screening tool.    Thus,  both
immunoassays were comparable to the GC/MS results. The
kit immunoassay false positive rate was 19 percent (n = 98)
and the plate immunoassay false positive rate was 0 percent
(n = 21).  However, the majority of the kit false positive
results were between 3 and 7 ppb, which is near the  lower
detection limit of the  method.  If a protocol specifying a
minimum value of 7 ppb were used in this study,  there
would have been a substantially lower false positive rate
(5 percent) for the  kit immunoassay.  No  false  negatives
were observed  in  this study  with either immunoassay
technique. This is  a critical criterion in determining  if the
immunoassays can  be used as effective sample screening
tools for expensive analytical methods.

A  variety of  QA/QC  samples  were  included  in the
experimental design  for the demonstration to provide
comparison  performance data  for  samples  of known
concentration and matrix.  The immunoassay plate results
for QA audit samples (nominal 25 ppm PCP concentration)
are shown in Figure 3.
             EMSL-LV
WBAS
ao-
L:
i .
1 40-
I '
& 20'
a.
0-
1
o !
j
i






Nominal Concentration 1
25 ppm PCP |
g_ L --£-
° • ; ° : |
1
o 0 6
• B B a

_*_.
A

                      56     1
                       Ai»y Numbtr
                       S*mpl« Dilution:
                   01:30  A 1:60  01:«0 O1:120
Figure 3. Plate Immunoassay QA Sample Chart for EMSL-
         LV and WBAS Laboratories
CONCLUSIONS AND RECOMMENDATIONS

Although the immunoassay methods evaluated are not as
precise  as  traditional  methods,  they  have  several
outstanding features.  Principal among these are the rapid
turnaround times for total analysis  and the low cost per
sample.  The immunoassay methods are also field portable
and require minimal training to perform. The immunoassays
have detection limits and linear dynamic ranges comparable
to those of traditional methods. A reduced level of accuracy
is a limitation usually  encountered when a system  is
configured for simple  field-portable  use.   Though  an
immunoassay can be developed for a specific target analyte,
                        it may be subject to certain interference effects from non-target
                        compounds as well as from matrix effects unless a more
                        involved extraction is used. However, most cross-reacting
                        compounds are structurally related to the compounds of
                        interest and, because of  their intrinsic toxicity, they are
                        frequently analytes of interest for regulatory monitoring as wett.
                        It is noteworthy that this study found no evidence of false
                        negative results, an important feature of a screening method.
                        The majority of false positives occurred at values near the
                        lower limit of detection for each method. This rate might be
                        significantly reduced by simple changes in the acceptance
                        criteria for the method.

                        The information provided by immunoassay analysis, while not
                        solely sufficient for initial site characterization, is frequently
                        useful for detailed characterization of previously identified
                        compounds of interest. Immunoassays are developed to be
                        sensitive to only a particular target compound or a class of
                        related compounds. The overall performance shown in these
                        studies  demonstrates that  immunoassays can provide
                        appropriate information for rapid on-site field decision making
                        (see Table 1). In addition,  the high sample capacity of the

                        TABLE 1. Method Comparison for PCP Analysis in Water
                                                                         WBAS      WBAS
                                                                           Kit       Plate
                                                            Performance Immuno-   Immuno-
                                                            Parameters   assay      assay
                                                            EPA      EPA
                                                           Method    Method
                                                            8270      604
                                                           GC/MS     GC
                        Detection
                        Limit (ppb)      3-5       30-40       30-50     1-15

                        Linear
                        Dynamic      3-40      30-400     30-200     1-200
                        Range (ppb)

                        Precision     10-15%     5-10%  concentration  10%
                                                         dependent
                        Accuracy    +25-40%   ±15-25%   +10-20%   +15%


                                                 No         Yes      Yes


                                                 Yes         No        No
Extraction
Required       No
Rapid
On-Site        Yes
Analysis
                        Total       1.5 hours/   5 hours/     5 hours/  4.5 hours/
                        Analysis    10 samples 40 samples   sample    sample
                        Time

                        Cost/Sample  $7.50      $2.50     $300-$750 $1004300
                                                       817

-------
 plate immunoassay provides a low-cost screening alternative
 to higher cost laboratory analysis, the results of which are
 frequently unavailable for several weeks after sampling.
 REFERENCES

 Silverstein, M. E., R. J. White, R. W. Gerlach, and J. M. Van
 Emon, 'Superfund Innovative Technology Evaluation of the
 Westinghouse Bio-Analytic  Systems Field  Immunoassay
 Methods for the Analysis of Pentachlorophenol in Water,
 EPA Environmental Monitoring Systems  Laboratory,  Las
 Vegas, NV, 1991, (in preparation).

 Van Emon, J.  M.,  'EPA Evaluations of  Immunoassay
 Methods' in Immunochemical Methods for Environmental
 Analysis, ACS Symposium Series 442, J. M. Van Emon and
 R.  0.   Mumma,   Eds.,  American  Chemical  Society,
 Washington, D.C., pp 58-64,1990.

 Van Emon, J. M. and  R. W. Gerlach, EPA Evaluation of the
 Westinghouse   Bio-Analytic  Systems  Pentachlorophenol
 Immunoassays,  EPA/600/X-90/146, EPA  Environmental
 Monitoring Systems Laboratory, Las Vegas, 1990, NV, 79 pp.

 Van Emon.  J.  M., J. N. Seiber,  and  B. D. Hammock,
 •Immunoassay  Techniques for  Pesticide  Analysis'  In
 Methods  for Pesticides and  Plant  Growth  Regulators,
 Academic Press, New York, Vol. XVII, 1989, pp 217-263.

 White,   R.  J.,  and   J.  M.  Van  Emon,  Report  on
 Immunochemical Techniques for Identifying and Quantifying
 Organic  Compounds in Biological  and  Environmental
 Samples, EPA/600/X89/288, EPA Environmental Monitoring
 Systems Laboratory, Las Vegas, NV, 1989, pp 1-A6.
NOTICE

Although the research described herein has been funded
wholly  or in part by the U.S. Environmental Protection
Agency under Contract Nos. 68-03-3249 and 68-CO-0049 to
Lockheed Engineering & Sciences Company, it has not been
subject to Agency review and therefore does not necessarily
reflect the views of the Agency, and no official endorsement
should  be inferred.  Mention of trade names or commercial
products   does   not  constitute   endorsement   or
recommendation for use.
                                                       818

-------
         RAPID SCREENING TECHNIQUE FOR POLYCHLORINATED BIPHENYLS (PCBs)
                       USING ROOM TEMPERATURE PHOSPHORESCENCE
                                 T. Vo-Dinh, G. H. Miller, A. Pal, W. Watts,
                                               and M. Uriel
                                 Advanced Monitoring Development Group
                                      Oak Ridge National Laboratory
                                     Oak Ridge, Tennessee  37831-6101

                                        D. Eastwood and R. Lidberg
                             Lockheed Engineering & Management Services Co.
                                         Las Vegas, Nevada  89122
ABSTRACT

The  analysis  of  Polychlorinated biphenyls (PCBs)
generally requires selectivity and sensitivity. Even after
cleanup, PCBs are usually at ultra-trace levels in field
samples,   mixed   in   with   other   halocarbons,
hydrocarbons, lipids, etc. The levels of PCBs typically
found  in water, soil,  tissue,  food, biota and  other
matrices of interest are in  the parts per billion (ppb)
range.  Most current measurement techniques for PCBs
require chromatographic  separations  and   are  not
practical for routine analysis.

There  is a  strong need  to have  rapid and simple
techniques to screen for PCBs  under  field conditions.
The  use  of field  screening  analysis  allows  rapid
decisions in remedial actions and  reduces the need for
sample preparations and time consuming laboratory
analyses.   Field screening techniques  also reduce the
cost of clean-up operations.

This paper  describes  a  screening  technique  room
temperature phosphorescence (RTP), and provides an
overview of both this  analytical procedure and the
instrumentation  to  detect  trace levels of  chemical
pollutants  and  related  biomarkers   in   complex
environmental samples.

INTRODUCTION

Polychlorinated biphenyls  are  a  class of chlorinated
aromatic compounds which have found widespread
applications  because of their general  stability  and
inertness as well as their excellent dielectric properties.
The PCBs have been used in electrical capacitors,
transformers, vacuum pumps, adhesives, plasticizers,
pesticides, etc. The discovery of PCBs in environmental
samples has spurred renewed concerns due to their acute
and chronic toxicity, and other long-term health effects.
The analysis of PCBs generally requires selectivity and
sensitivity. The levels of PCBs typically found in water,
sofl, tissue, food, biota, and other matrices of interest are
in the parts per billion (ppb) range.  It is therefore
important to develop simple,  sensitive and  rapid
screening procedures for PCBs.

Most of the analytical techniques used for PCBs are not
easily adapted to field measurements and generally
employ chromatographic separations coupled to a
specific detection scheme  (e.g., flame ionization
detection-FID;  electron  capture detection-BCD;
photoionization detection-PID; thermal conductivity-TQ
mass spectroscopy-MS; Fourier transform infrared-FTTR,
etc.). A review of analytical techniques for PCBs has
been described by Erickson (1). Packed column gas
chromatography (GC), thin-layer chromatography (TLC),
or high-performance liquid chromatography (HPLC) can
be used to provide data on "total PCB" contents in
samples. Packed column GC/ECD is the  common
method for quantification of PCBs as Aroclors in the
American National  Standards  Institute  (ANSI)
procedures. The PCBs are quantified against an Aroclor
standard using the largest peak, or a secondary peak.
The GC/ECD technique was used to determine PCBs in
sediments and   soils (2).   If  congener-specific
determination  is required,  high-resolution  gas
                                                   819

-------
 chromatography  (HRGC),  which  uses fused silica
 capillary columns, would be the technique of choice (3).
 High-resolution gas chromatography has been used for
 the analysis of PCBs in transformer fluids or waste oils.
 Various MS techniques (electron impact MS, chemical
 ionization MS, coupled MS/MS, etc.) have been used to
 analyze  complex PCB samples.   Methods involving
 perchlorination of the  biphenyl  ring  of the PCB
 congeners  have  been used in the determination of
 PCBs.  One of the limitations  of the perchlorination
 approach is due  to the fact  that biphenyl can also be
 perchlorinated, thus leading to erroneously high blank
 levels.

 Room Temperature Phosphorimetrv fRTP):

 The screening technique  involved in this study involve
 Room Temperature Phosphorimetry (4).  Conventional
 phosphorimetry requires the use of low-temperature
 matrices   to   reduce   the  collisional   quenching
 mechanisms and  radiationless deactivation  processes.
 Due to the requirement  of  cryogenic  equipment and
 refrigerant, conventional phosphorimetry has  limited
 usefulness   for   routine   applications   in  field
 measurements.

 Unlike conventional low-temperature phosphorimetry,
 RTP  is based on  detecting  the  phosphorescence
 emitted from  organic compounds adsorbed on solid
 substrates  at  ambient temperatures.   The  general
 approach  is to  obtain  a  solution  containing  the
 materials  to  be  analyzed   using  rapid   extraction
 procedures (1-3 min). A  few microliters of the sample
 solution are then spotted  on  a filter paper. The spot is
 dried for about three minutes with a heating lamp then
 transferred to  the  sample  compartment  of  the
 spectrometer.   Measurements can be  performed with
 any commercial  spectrofluorimeter equipped  with a
 phosphoroscope.

 The sensitivity and selectivity of RTP can be enhanced
 by  mixing the sample or pretreating the filter paper
 with a  heavy-atom salt solution.  Salts such as thallium
 acetate, in lead  acetate are efficient in enhancing
 phosphorescence  quantum yields for most PCBs.

 Figure 1 illustrates the  characterization of  Aroclor
 1254, a PCB mixture commonly found in environmental
samples, using the RTP technique.  This figure shows
the RTP spectra of Aroclor 1254 using thallium acetate
as the  heavy-atom perturber. The  efficacy and cost-
effectiveness  of  the  RTP  technique for  screening
complex   environmental   samples    have   been
 demonstrated in previous studies (4). Figure 2 shows the
 improved selectivity of the RTP technique by using the
 second-derivative method. The precision of the RTP
 measurements  is of the  order of 15%.

 The RTP technique approach offers several advantages:
 (a) rapid analysis, (b) simple set-up, (c) field applicable,
 (d) low per-analysis cost. These features of merit make
 RTP suitable for screening where a rapid estimation for
 specific PCBs is needed.  The use of field screening
 analysis allows rapid decisions in a cleanup operation
 and reduces the need for either return visits to a site by
 a cleanup crew, or extensive and costly laboratory
 analyses of samples that contain no detectable levels of
 PCBs.  Field screening techniques also reduce the cost of
 remedial actions  by preventing unnecessary excavation of
 uncontaminated soil.

 ACKNOWLEDGEMENT

 This research  is  jointly  sponsored by  the U.S.
 Environmental   Protection  Agency  (Interagency
 Agreement EPA No. DW899339001, DOE No. 1824-
 B124-A1) and the Office of Health and Environmental
 Research, U.S. Department of Energy, under contract
 DE-AC05-84OR21400 with Martin Marietta Energy
 Systems, Inc. Although research described in this article
 has been funded by the United States Environmental
 Protection Agency, it has not been subjected to Agency
 review and therefore does not necessarily reflect the
views of the Agency, and no official endorsement should
be inferred. Mention of trade names or commercial
products does not constitute Agency endorsement of the
product.

REFERENCES

 1.  Erickson, M. D., Analytical Chemistry of PCBs,
   Butterworth Publishers, Boston, 1986.
2.  Spittler,  T.  M.,  in   Environmental   Sampling
   for  Hazardous  Wastes,  G.  E.  Sweitzer and
   J. A. Santolucito, Eds., ACS Symposium Series 267,
   ACS, Washington, D.C., 1984.
3.  Safe, S., M. Muffin, L. Safe, C. Pochini, S. McCrindle
   and M. Romkes, in Physical Behaviour of PCBs
   in the Great Lakes,   D.  MacKay,  S. Paterson,
   S. J. Eisenreich and M. S. Simmons, Eds., Ann Arbor
   Publishers Inc., Ann Arbor, MI  1983.
4.  Vo-Dinh, T, Room Temperature Phosphorimetry for
   Chemical Analysis, Wiley, New York, 1984.
                                                     820

-------
QC
<
oc
H
00
QC
z
iu
        !     I     I    I     I     I     I     I     I
 400  420  440  460  480  500  520  540  560  580  600

                   WAVELENGTH (nm)


                      Figure 1


  Example of RTP Spectrum of Aroclor 1254
                        821

-------
z
IU
111
oc
UJ
Q

D
Z
o
o
111
        PCB  1221
                                    CONC. = 5 ng/sq cm

                                    BANDPASS = 4, 2 nm

                                    EX = 270 nm

                                    WHATMAN No 1

                                    FILTER PAPER TREATED

                                    WITH LEAD ACETATE
    400
425
450
475
500
525
550
                                 Figure  2
                    SECOND DERIVATIVE RTF EMISSION

                    OF POLYCHLORINATED BIPHENYLS

-------
           RAPID DETERMINATION OF DRUGS AND SEMIVOLATILE ORGANICS BY DIRECT
                      THERMAL DESORPTION ION TRAP MASS SPECTROMETRY*
                  Marcus B. Wise, Ralph H. Ilgner, Michelle V. Buchanan, and Michael R. Guerin

                                         Analytical Chemistry Division
                                        Oak Ridge National Laboratory
                                      Oak Ridge, Tennessee 37831-6120
ABSTRACT
   Direct thermal desorption  of analytes into an  ion
trap mass spectrometer (ITMS) is being investigated as a
technique  for the rapid  screening of a  wide variety of
samples for target semivolatile organic compounds. This
includes the direct detection of drugs  in physiological
fluids, semivolatile organic pollutants in water and waste
samples,   and  air  pollutants   collected   on  sorbent
cartridges.   In  order to minimize  the analysis time,
chromatographic separation  is  not  performed on  the
sample prior to introduction into the ITMS. Instead,
selective  chemical  ionization   and  tandem  mass
spectrometry (MS/MS) are used achieve the specificity
required for the target  analytes.  Detection limits  are
typically 10-50  ppb using a 1  uL aliquot  of a liquid
sample without  prcconcentration. Sample turn-around
time is 2 to 5 minutes and 3  to 5 target analytes can be
quantitatively determined simultaneously.

INTRODUCTION
   Recent advances  in ion trap technology have led to
the development  of  very sensitive and highly versatile
mass   spectrometers.    These  instruments   exhibit
mechanical simplicity, relatively small size,  electron or
chemical ionization options, and MS/MS capability  on
the most sophisticated systems.  Because  of their small
size, ruggedncss,  and ease of operation,  ion traps are
especially  attractive for  potential use in field screening
applications.  Much work is currently being performed
by several groups on  the development  of screening
methods for volatile organics in environmental matrices.
This  paper  describes  the  use of  ion  trap  mass
spectrometry for the direct analysis of target semivolatile
organics in  a  variety  of  samples  including organic
pollutants  in  water and drugs and  metabolites  in
physiological fluids.
EXPERIMENTAL
   All experiments were performed with a Finnigan MAT
ITMS ion trap mass spectrometer. This instrument was
equipped with an electropolished vacuum manifold and dual
300 L/sec turbomolecular pumps to minimize problems with
background contamination. Additionally, the manifold and
ion trap cell were heated continuously to a temperature of
200°C  using  infrared  lamps  to  further  minimize
contamination problems. The ITMS was configured with all
of the necessary electronics for ion ejection, MS/MS, and
axial modulation experiments.  Scan  functions  for the
detection of target compounds were written by the authors
using the editor program provided by the manufacturer.

   Samples were introduced into the ITMS by means of a
thermal desorption  device which was coupled with an
open/split interface  as shown  in Figure 1. The  thermal
desorber was  constructed in our laboratory and was
designed to be compatible with sorbent tubes which are 3
inches long and 0.25 inches in diameter.  The cap of the
thermal desorber has a septum which  enables the direct
injection of liquid samples onto a sorbent tube  without
removing it from the desorber body.

   The analysis of target semivolatile  compounds is
typically performed by injecting 1 uL of liquid sample into
the thermal desorber onto the head of a 1 cm long packing
of Tenax. The Tenax trap is heated for 10 seconds with 450
watts of power while simultaneously being purged with a
flow of helium  at 40 to 60 mL/min. Compounds which are
vaporized from the sample are carried by the helium into an
open/split interface  which is constructed from a 6 inch
length of 500 micron megabore capillary tubing and a 15
inch length of  150 micron uncoated fused silica capillary
tubing.  The  150 micron capillary acts as a restrictor
between the ITMS vacuum chamber and  atmospheric
                                                       823

-------
pressure.  The split ratio is approximately 95:1 with the
bulk of the sample diverted to the split vent.  In order
to  minimize  the  condensation of compounds  in the
transfer  lines,  they  STS  maintained  at  a  constant
temperature cf 20G°C  No chromatographic separation
is performed  on the  sample prior to introduction into
the ion trap.

    For target compound analysis, the ITMS is typically
operated  in chemical ionization MS/MS mode.  Iso-
butane is  normally  used  as  the chemical ionization
reagent gas due  to  its ability  to discriminate  against
common hydrocarbon interferences.  Quantification  is
performed by collecting a series of MS/MS spectra for
the  target  analyte over a  period of  approximately  3
minutes starting at the time the sample is  heated and
desorbed from the Tenax.  Reconstructed plots  of ions
which arc characteristic of the MS/MS fragments for the
target analyte are integrated  and compared with an
external calibration curve.

RESULTS AND DISCUSSION
    An important  application of this technology is for
the  rapid   detection  of target  semivolatile  organic
pollutants  in  ground  water  samples.  For example,  as
shown in  Figure  2,  acrylamide can be  quantitatively
determined at low part-per-billion levels  in  1  uL  of
ground water with a  sample turn around  time  of less
than 4 minutes.   Calibration curves are linear over at
least  3  orders of magnitude  and reproducibility   is
typically 10-20%  at  the 95%  confidence intervals  as
shown in Figure 3. This method has also been shown to
be capable of the detection of pesticides such as dieldrin
as shown in Figure 4.

    In addition to the detection of semivolatile organic
pollutants   in  ground  water  samples,   the   same
methodology may  be  used for the detection of target
drugs in physiological fluids including urine and saliva.
As  shown  in  Figure 5,  cocaine can  be  directly
determined in urine at less than 100 ppb using 1 uL of a
sample without prcconcentration or chromatographic
separations. Sample turnaround time is approximately 3
minutes and linearity is good  over 2 to 3 orders of
magnitude as shown in Figure 6.  Using this technique  a
wide  range of  drugs of  abuse have  been examined
including    cocaine,   phenobarbital,   amphetamine,
methamphetamine, THC,  and  codeine, among  others.
Further, extensive work has demonstrated that nicotine
and several of its major metabolites can be quantitatively
determined  in the urine of smokers and  potentially in
the urine of non-smokers as well.  This  should prove
especially useful as a technique  for helping to assess the
level  of human  exposure  to  environmental  tobacco
smoke.

   In  addition to  the application of this technology  to
the  determination   of  target  semivolatile   organic
compounds in liquid samples, it is also useful as a technique
for the determination of trace levels of semivolatile organics
in air samples.  For this particular application, an air sample
is collected off-line using a small air sampling pump and a
suitable resin cartridge.   Samples can be collected for a
preset period of time for TWA values or grab samples may
be collected for  an instantaneous measurement.  The
sorbent tubes  are then simply loaded into  the thermal
desorber and  heated for 10 seconds to vaporize the
adsorbed analytes.  The experimental parameters are
otherwise identical to what has previously been described.
Using this method, part-per-trillion level detection limits
have  been  demonstrated   for   several  different
organophosphonate compounds in  air.   In addition,
extensive work has demonstrated that the level of nicotine
in environmental tobacco smoke can be quickly determined
using this method.

   Although the examples which have been described have
only involved the detection of a single target analyte, it is
currently possible to determine up to 4 different compounds
in a  single injection by use of alternating or multiplexed
scan functions. This is a means of using the host computer
to rapidly switch the conditions within the ion trap so that a
spectrum of a different analyte can be obtained. Current
software requires approximately 0.5 second to switch from
one compound to another, however, thus limiting the total
number of compounds that can realistically be determined
during the  time scale of a direct thermal desorption
experiment. Improvements in the ion trap software should
help to alleviate this problem.

CONCLUSION
   Direct  thermal  desorption into  an  ion trap  mass
spectrometer is a sensitive tool for the rapid detection of
target semivolatile organic pollutants in air and ground
water and drugs and metabolites in physiological fluids.
Detection limits are generally 10-50 ppb in a 1 uL liquid
sample without preconcentration and linearity is typically 2
to 3 orders of magnitude. Sample turn around time is 3 to
4  minutes  and up  to 4  different  compounds can be
simultaneously determined in  a single injection.
ACKNOWLEDGEMENT

'Research sponsored jointly by the National Cancer
Institute, DOE Interagency Agreement 0485-0485-A1, the
U.S. Army Toxic and Hazardous Materials Agency, DOE
Interagency Agreement 1769-A073-A1, and the Office of
The Program Executive Officer-Program Manager for
Chemical Demilitarization, DOE Interagency Agreement
1769-1757-A2, under United States Department of Energy
Contract DE-AC05-84OR21400 with Martin Marietta
Energy Systems, Inc.
                                                         824

-------
         Direct Thermal Desorption  ITMS
            CAP
HELIUM  IN
 FLOW CONTROL
 HEATED DESORBER

     BLOCK
.SEPTUM




 O-RING
.<-"


—TENAX TRAP
                                           ION TRAP MS
HtSTRICTOR
U X
5FER LINE
1

CELL.
\ J(
-)L(
i
                   SPLIT VALVE
                               SPLIT VENT
  Figure 1  Apparatus for the direct thermal desorption of semivolatiles into the ITMS.
            Acrylamide in Water
                     3 Injections
                         MS/MS

                    Total Ion Response
                          A..        fv
            — .•—'-,/ '•M-.W.-.W.'J - ..'/ ** - ,-,SW,v.v.v_..' ***
            M 'i i I i i  i i I i i fi i i i i i I i i" i i
                    RESPONSE FOR M/Z 55
                    FRAGMENT ION
3785
11
, ) V/,,
5760
V

6230
I
' ii\i i &( 'tf '
                         TIME (min)
 Figure 2- Repeated 1 uL injections of acrylamide in water into thermal desorber.
                            825

-------
                        Acrylamide  in Water
                          ITMS Direct  Injection
                             MS/MS mass  72
100X
TOT-
         Figure 3  ITMS response curve for acrylamide in water.


              DIELDRIN IN HEIHANOL
DIRECT INJECTION AND FLASH VAPORIZATION INTO THE IMS

  NETHANOL
               TOTAL  ION PROFILE        HETHANOL
          DIELDRIN


            INJECTION ii
                                                      DIELDRIN
                                                        INJECTION If 2
      i' i' i
6.3X
345-
                       HASS 345 FRAGHENT ION PROFILE
                I   I  I  I
  SCAN Ii      450
  TIHE(HIN)   7:47
               see
              8:40
  I
559
9:31
10:23
 T
 650
11:16
             Figure 4  Repeated thermal desorptions of dieldrin into the ITMS.
                                 826

-------
 101



 0,8X

 153
   Cocaine-N-Methyl-D3 in  Urine
               1 uL injections


Blank  98 pg    98 pg   98 pg  245 pg  245 pg 245 pg
       ,»'"•-.>   -•"'•-••   ,,.v'-v-s,   A/*X.   AAVv   ft
.,/- *«W   -S^'    N-4,*   >V'  ^*vJ"   ^J
      62
      199
336
2i!9
100
311
298
200
673
369     437
175     150
624     574
                                   ww*
185
      83      155     '!&    297     369     437
     1 12    1462    1337    3662    3287   2637
     5 96    4945    4784    12157   11394   9258
           196
                             I I I I n'Ti'J I I I'i | I

                          300       400
       4:58      n:47       14:35      19:23

   Figure 5 Repeated injections of cocaine in urine into the ITMS.
            Cocaine-N-Methyl-D3 in Urine
                        1 uL injections
                           m/z 185
           100   200  300  400  600  600  700  800  900 1000

                          Picograms

            Figure 6 ITMS response curve for cocaine in urine.
                          827

-------
                         A NEW APPROACH  FOR  ON-SITE MONITORING OF
                                            ORGANIC VAPORS
                                          AT LOW  PPB LEVELS
                                       H. Wohltjen, N.L. Jarvis. and J.  Lint
                                           Microsensor Systems, Inc.
                                              6800 Versar Center
                                            Springfiled, VA   22151
ABSTRACT

A  new, very  compact, gas  chromatograph  has been
developed that uses  scrubbed ambient  air for its carrier
gas.  The instrument uses an internal pump to collect a
vapor sample,  concentrate it onto a small Tenax sorbent
tube,  and thermally  desorb  it  (via  backflush) into the
isothermal  (65°C)  chromatographic   column  for
subsequent  analysis.  Pump,  valve, and concentrator
sequencing  is controlled automatically by an on-board
microcomputer that also records the  detector  responses
and determines baseline corrected chromatographic peak
heights, positions, and concentrations.  The instrument  is
small, (ca. 1/3  cu.  ft.)  lightweight,  (12  Ibs) and
consumes little power  (e.g.,  10 Watts).  The selective
measurement   of  a variety  of  organic  vapors  at
concentrations  from a few ppb (by volume) to 100 ppm
has been demonstrated.
 INTRODUCTION

 Environmental monitoring  is a growing challenge, driven
 by ever increasing demands for  higher sensitivity, higher
 selectivity, greater portability,  and lower cost.  In the last
 decade, portable gas chromatographs have become very
 popular for this  application owing to  their high sensitivity
 and the broad acceptance  enjoyed by the chromatographic
 method within the  analytical community.   Many of the
 current instruments,  while  effective,  offer less than
 optimal solutions to workers  required to perform on-site
 environmental measurements.  The disadvantages exhibited
 by some  existing portable chromatographs include such
 things as  the need  for external supplies of carrier gas,
 poor or non-existent control of column temperature,  high
 cost, extensive  reliance  on operator adjustments and
 operator interpretation of chromatograms.
The main objective of this project was to  develop a very
small, low cost, portable gas chromatograph that was fully
automatic,  requiring  merely that the  user turn  on  the
power, provide a sample,  and  read the  results  from  a
digital display.   A further aim was to afford the user great
flexibility both  in  the power supply  requirements and in
the reporting  of  data  directly  to personal computers,
printers, and modems.   Finally,  it was required that the
instrument  exhibit  sensitivities   and   selectivities
significantly  higher  than that demanded  by most
environmental  monitoring  applications.   These objectives
were  successfully  met   by  the  portable instrument
described here.

MATERIALS AND METHODS

System Overview
A schematic of the vapor monitor instrument is shown in
figure 1. Air samples  can be drawn into the instrument
using a small on-board air sampling  pump.  The samples
pass through a tenax GC concentrator tube where many
organic vapors are effectively trapped.  Introduction of the
concentrated sample into the GC column is accomplished by
heating  the concentrator and backflushing  the vapors into
the column.  Sample injection,  operation of  the GC column,
and  data analysis are controlled by the microcomputer
following a  schedule  illustrated  in  figure 2.   Once the
chromatogram   is   completed,  the  microcomputer
determines retention time, baseline corrected height,  and
vapor concentration  for all peaks,  using the calibration
tables stored  in  memory  for up to three user-defined
retention time  windows.  On power-up  there  is a warm-up
cycle during which the  oven temperature  stabilizes at
65°C.      During  the  warm-up  period  the  system
microcomputer performs diagnostic  checks  on  the oven,
detector, carrier gas pump, and  internal circuitry.  Local
interaction  with the instrument is  via the front panel
keypad and display which  prompts  the user with menus
through the operation of  the  instrument.  It can  also be
controlled  remotely  via  connection to the 1200  baud
RS232C serial communication port of  a PC or a modem
connected to a telephone line. The instrument is contained
in a rugged 1/8  inch  thick aluminum case and can be
powered either  from  an  external  12 volt  battery  or
directly from 120 volt AC  power.
                                                      829

-------
                                                                           L CONSOLE PO
                                            SYSTEM MICROCOMPUTER
                                          Wlllt 64K BATTERY BACKED RAM
                                         AND BATTERY BACKED CLOCK/CALENDAR
      RS232C
      SERIAL
     COMM PORT
                                                                         trU
'";:

:,™;
GC INTERFACE
PC BOARD

"-1





r^~

H

•;r,»

                                                            r^t
                         Figure  1.    Portable gas chromatograph system  diagram
Chromatographic Column
The instrument uses an isothermal (65°C)  1/8" diameter,
35  inch, Chromatographic column packed with  10% TCEP
(tricyanoethoxypropane)  on  80/100  mesh  Supelcoport.
The  column  provides  satisfactory  resolution   for
monitoring  all but very complex  vapor mixtures.   The
column is wound  on  a 2.5 inch diameter aluminum block
that contains a heater.  The column and heater are housed in
a  styrofoam block to  assure  a constant temperature
environment.   Once  hot, the  column  oven  assembly
consumes less than 6 watts of power to maintain the 65 °C
temperature to within ±  0.5°C.

Carrier Gas
The carrier gas  is generated  from ambient air using a
small compressor  pump and an activated carbon scrubber
to  remove  impurities.  The pressure in the scrubber  tank
is held constant by means of a silicon chip pressure sensor
connected  in a feedback control loop to the  carrier gas
pump.  The carrier gas pump pulses on and off to add air to
the scrubber tank as required.  The duty cycle of this pump
is low, typically less than 10%.  Power consumption of the
carrier gas  generator is typically less  than 100 mW when
generating  a  12  seem  flow  rate  through  the  35  inch
column.   Flow rate is maintained constant  to  better  than
±0.5  seem  of  the   setpoint  regardless  of  ambient
temperature  or pressure  variations.   Scrubber life  is
dependendent on ambient conditions but typically  is 500  to
1000  hours.   Scrubber replacement  takes less than  5
minutes.
                         -QC DURATION-
SAMPLE DURATION «•"
PREHEAT DURATION
INJECT DURATION
PURGE DURATION


CONC HEAT -

FAN


VALVE#1

VALVE#2


VALVE#3 -
CALCULATION

DETECTOR
SIGNAL













— i
«•


n





-4» 1
**•
'


I

i i i



1 1
i — ,




i































































r
WINDOW #i #2 #3
1 1 ,
^JiA^




ON
ON


ON

ON

ON

FF
ON

ION



i i 1 1 1 1 1 'T~
0 120 240 360
Figure 2.
            TIME (sec)

Typical run schedule for chromatograph
                                                       830

-------
Sample Enrichment
The high  sensitivity of this instrument results from  the
use of a concentrator  to  enrich the  sample  vapor
concentration prior to injection  into the column.  Sample
enrichment is achieved by compressing the organic  vapors
present in a large volume of air  into a  much smaller
volume for injection into the GC column.  This is done by
sampling  the ambient  air at a high collection rate (e.g.,
600 seem) and passing the sample through ca. 50 mg of
tenax GC  in an adsorbent tube that traps most organic
vapors  except for the very low boiling compounds (and
water vapor). The adsorbent tube is then heated to 140°C
to vaporize the trapped organics  which are then injected
into the GC column using a much lower rate of flow (e.g.,
12  seem).  After injection, the concentrator is  cooled by a
fan to permit efficient sample collection on the next trial.
The comparatively low  140 °C desorption temperature was
selected to extend the operating lifetime of the  adsorbent to
1000  hours or more.  Sample collection can be performed
concurrently  with  the  chromatographic run  to minimize
the analysis  time.  The sample  collection and  injection
times are  user programmable to optimize the  performance
of the gas chromatograph.  The enrichment factor of the
sample concentration  procedure Is thus normally adjusted
by  varying the sampling time.   The  practical limits for
sample concentration are a function  of such parameters as
the adsorbent  used,  the bed  depth,  the  vapors to be
concentrated, and desorption  temperature.   At a  typical
sampling  time of 30 seconds  the effective concentration
factor will be approximately SOX.   The instrument  was
designed  to  collect vapor samples either automatically
from ambient air present at a 1/8 inch  Swagelok fining on
the front panel or from a sampling line to permit access to
more localized areas  or  in restricted locations.  It  is also
possible to add  a septum to the gas  inlet fitting so  that
vapor samples can be  introducted directly onto the sample
concentrator  by syringe  injection.
                                                      System Purging
                                                      The  instrument  offers  the  capability of  purging the
                                                      internal sample injection system with clean air.  When the
                                                      user selects a purge duration from the system menu, the
                                                      concentrator and  associated valving are flushed with  clean
                                                      air immediately  after the sample is  injected  onto the
                                                      column.   This process removes residual vapors from the
                                                      concentrator sorbent and  assures  that  low volatility
                                                      contaminants are not injected into the column.

                                                      Solid State Sensor
                                                      The  proprietary  solid-state  detector  used   in  the
                                                      instrument  is  a  very  robust device whose operating
                                                      lifetime is measured in years. Vapors  eluting from  the GC
                                                      column adsorb onto the chemically sensitive coating of the
                                                      detector.  This vapor/coating interaction results In a signal
                                                      whose frequency  is related to the vapor concentration. The
                                                      detector exhibits some preferential sensitivity to aromatic
                                                      hydrocarbons  but  it also  responds well  to  aliphatic
                                                      hydrocarbons, alcohols,  esters, and halocarbons.  By
                                                      changing the detector coating composition the detector can
                                                      be tailored to respond in a selective fashion to many other
                                                      organic vapors such as organophosphorus pesticides.

                                                      System Calibration
                                                      The  instrument  is  normally  calibrated  using five
                                                      concentrations of each vapor (e.g., benzene, toluene and
                                                      xylene). These data define  the  shape of the calibration
                                                      "curve",   which is then stored in memory for data analysis.
                                                      The instrument can be readily re-calibrated to compensate
                                                      for changes in sensitivity over time.   A single point span
                                                      calibration  is sufficient.   If the experimentally measured
                                                      concentration of  a vapor  standard differs from the known
                                                      concentration, a system  Response  Factor, R,  can  be
                                                      adjusted  so that the measured concentration equals the
                                                      known  value.  Changing the R  factor adjusts the entire
                                                      calibration  curve.
     25000-
 S
 I
 Q
 N
 A
 L
(Hz)
     20000-
15000-
10000-
      5000-
                  I     i      i      i     i
                  60   120   180   240   300   360
 S
 I
 G
 N
 A
 L
(Hz)
                                                                25000
                                                                 20000-
15000-
                                                           10000-
                                                                 5000-
                     RETENTION TIME (sec)
                                                                        60   120   180   240   300   360
                                                                           RETENTION TIME (sec)
 Figure 3.  Typical chromatogram for benzene (1), toluene (2)     Figure 4.
             ethyl benzene (3), and o-xylene (4)
                                                                 Chromatogram for four consecutive runs of
                                                                  unleaded gasoline vapors in air showing BTEX
                                                        831

-------
Data Analysis and Reporting
Once  the  chromatogram  is complete,  the on-board
microcomputer  determines  retention  times and baseline
corrected heights of the peaks occurring  in up to three
user selected retention time windows.  Window location and
width are specified independently for  the analytes of
interest. After the chromatogram data have been acquired,
the data are  smoothed and analyzed to  check for peaks.
When  a peak  is detected  in a  window  then baseline
correction is  performed prior to determination of the  peak
height.  This is done by drawing  a straight line from the
valley points  that are detected before and after the  peak
location. Thus, tangent skimming is achieved.   It  then
automatically  determines the concentrations of the selected
vapors  using the calibration tables  stored in memory.
Each calibration "curve" consists of up to  five, piecewise
linear segments  that help  correct for  non-linearities in
the  overall  system response.    Results  of the latest
measurement are displayed on the Liquid Crystal Display
(LCD) on the  instrument panel. Up to 8 hours of the  most
recent  data are stored in non-volatile memory. The  user
can  report  results of the  latest run  or  report all stored
data in a variety of  formats selected from the menu.  The
instrument can  be interfaced directly  to  a  PC terminal to
display the  data visually, or to a printer to  provide a  hard
copy. The user can choose  to display or  print the latest
chromatogram in real time,  print out a listing of all peaks
in  the chromatogram, or print a listing of  all data stored in
memory without header  information to make it compatible
with many  popular  spreadsheet  programs available for
personal computers.    The user  also has the option of
reporting data remotely via a telephone modem.
                                                       40000
                                                    i
                                                        30000 -
                                                    LU
                                                    *   20000 H
                                                    g   10000 •
                                                             0.0    0.2   0.4    0.6    0.8   1.0
                                                                    BENZENE CONC. (ppm)

                                                              Figure 6. Benzene calibration curve
                                                The calibration curve for benzene is shown in figure 6.  It
                                                should be noted that on this curve the benzene sample
                                                having a concentration  of 0.011  ppm  produced a peak
                                                height of approximately 2000 Hertz.  The typical detector
                                                noise observed on this instrument  was approximately ±2
                                                Hz. Thus, the 11 ppb benzene sample produced a signal to
                                                noise ratio of approximately 1000 to 1 thereby suggesting
                                                that   highly  accurate   quantitative  work  at  these
                                                concentration levels is possible.  Other compounds besides
                                                BTEX can also be monitored with this simple instrument.
                                                Figures  7 & 8 show chromatograms for trichloroethylene
                                                and  tetrachloroethylene, respectively, using the same
                                                column conditions as used in the BTEX analysis.
 RESULTS AND DISCUSSION

 A sample chromatogram for benzene, toluene, ethylbenzene
 and o-xylene in air is  presented  in figure  3.  Figure 4
 illustrates the  chromatographic   repeatability  of  the
 instrument.   Here four consecutive chromatograms  of
 unleaded gasoline vapors in room air are superimposed.
 The repeatability of the peak  height and  peak position
 exhibited  by the  instrument  is  noteworthy.    Further
 evidence of the excellent repeatability was  obtained from
 numerous automatic samples taken by the instrument.  For
 example, when exposed to repeated samples of a benzene
 gas standard  at 0.93 ppm the  results shown in figure 5
 were obtained.
Run No.
    1
    2
   3
   4
   5
   6
   7

Figure 5.
   Signal
   (Hertz)
   26,010
   26,041
   25,995
   26,010
   26,064
   26,022
   26,084
 Measured
Cone, (ppm)
    0.930
    0.931
    0.930
    0.930
    0.932
    0.931
    0.933
                                                      25000
                                                  S
                                                  I
                                                  Q
                                                  N
                                                  A
                                                  L
                                                 (Hz)
                                                      20000-
                                       15000-
                                       10000-
                                                                  5000-
                                                                     0  -
          0      60   120   180   240   300   360
                    RETENTION TIME (sec)


Figure 7.   Chromatogram for trichloroethylene
Reproducibility of Portable GC Peak Response
       To Benzene at 0.93 ppm
                                                        832

-------
CONCLUSION

This portable gas chromatograph was designed  to take
advantage of the most recent solid state technology. The use
of modern solid state chemical and physical sensors affords
a dramatic  reduction in the instrument complexity, size,
and cost  by  allowing the use of scrubbed  ambient air as a
carrier  gas.   This  very  important   feature  greatly
simplifies the  logistics  of operating and  maintaining a
portable gas chromatograph.  The extensive use of digital
microcomputer  technology makes  it extremely easy to
connect  into  multi-instrument   networks  or  data
processing  computers.   The  instrument  has been
demonstrated  to be  highly  effective  for  monitoring
hazardous  organic  vapors in  the  low parts per billion
range in  "real-world"  matrices.  It can be programmed to
respond to a wide range of organic vapors  and can be used
for  single   sample   analysis  as  well  as  continuous,
unattended  monitoring.
     25000'
 S
 I
 G
 N
 A
 L
(Hz)
     20000-
15000-
10000-
      5000-
         0  -
           0     60    120   180   240   300   360
                     RETENTION TIME (sec)


 Figure 8.   Chromatogram for  tetrachloroethylene
                                                       833

-------
                 A RAPID SCREENING PROCEDURE FOR DETERMINING TRITIUM IN SOIL
                                           Kai M. Wong, Tina M. Carlsen
                                      Lawrence Livermore National Laboratory
                                                  Livermore, CA
Tritium, as tritiated water, is occasionally found in soil
samples collected for investigations being conducted to
identify the sources of soil and ground water contamination
at the Lawrence Livermore National Laboratory (the major
constituents of concern are volatile organic compounds and
gasoline). It is important to quickly identify samples that
contain tritium, and to estimate tritium activities or concen-
trations before submitting samples to contract analytical
laboratories for conventional chemical analyses as most
laboratories are not equipped to handle radioisotopes.
Traditional methods for determining tritium concentrations
in soil require distillation or freeze-drying to collect soil
water, but both methods are time consuming and costly.
Therefore, we needed to develop a rapid screening method
for determining the tritium concentration in soil. Initially,
we attempted to determine the soil tritium content by direct
liquid scintillation counting of small aliquots of the soil
samples.  However, direct counting limited the size of the
sample, raising the detection limit of the analysis. Also,
many soil samples contained materials that were extracted
by the liquid scintillation cocktail solvent, causing drastic
reduction of counting efficiency and further decreasing
analytical sensitivity.
Upon further work, we found we could obtain reliable,
reproducible results by direct extraction with water. This
method is simple, eliminates interfering materials from
most soil matrices, and is very cost effective compared to
traditional methods. A weighed amount of soil is vigorously
agitated in a measured volume of water. After centrifuging,
the supernatant is decanted and tritium activity is deter-
mined by liquid scintillation counting for 10 minutes. The
measured activity is adjusted for the amounts of soil and
water used in the extraction. Depending on the amounts of
soil and water used, the method can measure tritium con-
centrations down to 1 pCi/g of soil (approximately 10,000
pCi/L at 10% soil moisture). The entire extraction proce-
dure for 12 samples requires less than 30 minutes. Results
can then be used to choose samples to be analyzed for
tritium, using more rigorous distillation methods.
Work performed under the auspices of the U.S. Department
of Energy by Lawrence Livermore National Laboratory
under Contract W-7405-Eng-48.
                                                        835

-------
                           ™ SW^1*" of Volatae °«*-fc Constitutents of
                           °^Ulie ST and TraP ^ EIi^beth Woofenden, Perkin-
                        Seer Green, Buckinghamshire, England and James Ryan  The
        Perkin-EImer Corporation, 761 Main Avenue, Norwalk, CT 06859-0219
  INTRODUCTION
  Purge-and-tra
  year histo
  m water.                           , ensve, a
  able to be automated. Purge-and-trap is used in all
  SSrvb? "A monitoring programs; RCRA, CERCLA,
  NPDES industrial wastewater, and drinking water.

  A conventional system is illustrated in Figure 1  It is
  an integrated system, i.e. the purge vesseland sorb-
  ent trap are connected directly to the gas chromato-
  grapn in a laboratory environment.  Water samples
  must be collected in the field, chemically stabilized
  atmospherically-sealed, and shipped to a laboratory
  SieH  * llr.    -,n "^tod at the lab, they must be
  stored at 4° C until analyzed, and at least for CER-
  CLA, these samples must be analyzed within 10 days of
          Open-loop purging system with on-line trap
          and thermal desorption facility
                     Figure 1

While this existing analytical system works well, this
paper will demonstrate that it is not necessary for the
anaf'sistO   d°ne '" proximityto the chromatographic

Purge-and-trap systems which incorporate an integral
(on-line) thermal desorption device (shown in Fiimre 1)
have been found to suffer from several limitations
pese limitations can adversely affect the practical per-
formance of such on-line systems. Among the limita-
tions are:
   -  RISK OF CARRYOVER BETWEEN SAM-
      PLES. This can occur when a particularly high
      concentration sample is analyzed.

   -  RESTRICTED COMPATIBILITY WITH
      HIGH RESOLUTION CAPILLARY GC + MS
      DETECTION.

   -  RESTRICTED STORAGE TIME FOR WA-
      TER SAMPLES.

   -  INCREASED CHANCE OF SAMPLE CON-
      TAMINATION.  This can happen because of
      stabilizers added to the sample to prevent halo-
      form formation, or from atmospheric contami-
      nation of the aqueous sample from improper
      sample seals.

One solution to overcoming these potential problems is
to separate the purge-and-trap volatile chemical collec-
tion and concentration from tne desorption-chromato-
graphic analysis. In other words, perform the chroma-
tography off-line from the sample concentration.

The performance of one such off-line, fully automatic
thermal desorption system (Figure 2) has been evaluat-
ed for this work. System details and results from these
investigations are presented below.
                                                              Open-loop purging apparatus with removable trap
                                                              suitable for off-line, two-stage thermal desorption
                                                                     Open-loop purging apparatus
                                                                        with removable trap

                                                                           Adsoibenl tnpft)
          Optional
          line heater
                     Figure 2
                                                837

-------
ADVANTAGES OF SEPARATING PURGE-AND-
TRAP FROM CHROMATOGRAPHIC ANALYSIS
Using portable traps in combination with an automatic
off-line purge unit enables water to be sampled using
conventional EPA purging methodology at field sam-
pling stations. Once sampling is completed, the tubes
may be capped and transferred for thermal desorption
GC analysis at a central laboratory facility. This ap-
proach immediately overcomes  two of the major draw-
backs of conventional on-line methodology:

    -   NO RISK OF CARRYOVER BETWEEN
       SAMPLES.

    -   GREATLY EXTENDED MAXIMUM SAM-
       PLE STORAGE TIMES.

    -   DISTRIBUTED FIELD SAMPLING COM-
       BINED WITH CENTRALIZED LABORA-
       TORY ANALYSES.

An automatic thermal desorption instrument allows
multiple sample tubes to be analyzed without operator
attendance. With the Perkin-Elmer Model ATD-400
up to 50 sorbent traps can be can be analyzed in one
carousel loading. These traps, in the form of sampling
tubes, are compatible with the method detection Omits
specified by EPA 500 and 600 series methods, as well as
those purge-and-trap methods in the RCRA SW-846
analytical method manual.

For long term storage, the sorbent tubes can be capped
with brass Swagelok <™> caps and one-piece PTFE fer-
rules. Such tubes, spiked with benzene, toluene and m-
xylene, are available as certified  standards (Ref.l), and
have been shown to be stable for up to two years of
storage time.

In addition, data reported by the Netherlands Organi-
zation for Applied Scientific Research shows that chlo-
rinated hydrocarbons on Tenax(TM) are stable for over
2 years. Multiple analyses for  trichloroethylene and
tetrachloroethylene (Figure 3) carried out over a two
year period had a reproducibility with less than 10%
RSD at storage temperatures ranging from 4°C to  40 °C
(Ref. 4).
               STABILITY OF VOLATILE CHLOROALKANES ON TENAX
      Storage
     Temperature
      •C
               Component
             24 Month
Initial Mean Charge   Mean Recovery
ng RSO% »Rep   ng %Hec.
               Trichloroethylene
               Telrachtofoelhylen*
               Trichkxoelhylene
               Telrachkxoelhylene
               Trfchlwoelhylene
               Tetrachloroethylene
840
eoe
2.0% 15
1.9% IS
640 2.0%  15
eoe 1.9*  is
840 2.0%  IS
806 1.9%  IS
856 102%
781 97%
         811  97%
         756  94%
         6X2 100%
         765  95%
      Rel: TNO Division ol Technology lor Society: Netherlands Organization lor Applied Sclenlinc
        Research, Report No. R90/266.
                     Figure 3
                               The ATD-400 also overcomes a third limitation of con-
                               ventional procedures, i.e. incompatibility with high
                               resolution capillary GC and mass spectrometric detec-
                               tion, by using an optimized two-stage thermal desorp-
                               tion process (Figure 4).
                                    Two-stage thermal desorption using a packed cold trap
                                            Stage 1 - Primary (lube) desorption
                                     'Inlet split'
                                        Hot sample tube
                                        f
                                               -t-
                                                         'Desorb' flow
                                     Carrier
                                     inlet
                                                     Cold trap
                                                        i
                                                                        GC
                                                                      detector
                                                          Carrier inlet >

                                                              GC analytical column
                                            Stage 2 - Secondary (trap) desorption
                                      Sample t
                                        ube I


                                           t
                                                              -^-'Outlet split1

                                                                       GC
                                                                     detector
                                                 Hot trap
                                           Carrier inlet
                                                          GC analytical column
                      Figure 4

In this system, the primary trap (or tube) is heated and
purged with carrier gas for several minutes to ensure
complete elution of all retained components. These
components are swept by the gas stream into a second-
ary cold trap held at subambient temperatures and
comprising l/8th-inch quartz tubing packed with ap-
proximately 20 mg of a selected adsorbent. Following
this primary tube desorption stage, the secondary cold
trap is heated rapidly at approximately 2400° C, thus
transferring the pollutants into the GC analytical col-
umn in a very narrow band of vapor. This extremely
rapid  heating of the secondary trap produces peak
widths of approximately 1 second, resulting in:

    -  UNCOMPROMISED HIGH RESOLUTION
      CAPILLARY CHROMATOGRAPHY and

    -  EXCELLENT COMPATIBILITY WITH
      MASS SPECTROMETRIC DETECTION
      (Figure 6).

The quantitative performance of the system for trace
levels (nanograms) of components in the presence of
relatively large masses (-20 mg) of water was investi-
 §ated. These results are tabulated in Figure 8. These
 ata demonstrate that the excellent quantitative per-
formance of the ATD-400 was unaffected by a large
mass of water present in the sample.
                                                    838

-------
         Recovery from sample tubes showing the

         effect of water
                       Calculated Amount In ng
                      Lorotorm ••nil
             Knq **•!*«««
                       M.»


                       22 It
                       24 3f



                       237*
4M  *«  3»


3M  l-Iff  114


n t.   i» t   ft


tfl.M  1* 10  »1 M


1fl.40  ItM  JtJ«


10.33  1«M  »ST


        if n


        at os


        ooo


    H.OT  nn


     1.30  2.4*
21-fi  H.3S


n«o  t«»r


 040  (1,1.
                     Figure 5


A big advantage of the packed type of secondary cold
trap is that there is negligible risk of trap blockage by
ice formation.  The formation of ice blockages is a
major drawback of conventional capillary cryofocusing
systems. By using a packed cold trap, it is also possible
to retain extremely volatile components (i.e. C3 and
even C, hydrocarbons) using electrical cooling rather
than a liquid cryogen.  [N.B. The inability of simple liq-
uid nitrogen cooled cyrofocusing systems to retain even
medium volatility components has been reported by
Grob and co-workers (Ref. 2). It is also advantageous
to any form of automatic chromatographic analysis if
liquid coolant is NOT required.]

By using only a low mass of adsorbent (-20 mg) in the
secondary trap of the ATD-400 and by enabling high
maximum temperatures (up to 400°C) to be selected
when required, the system eliminates any possibility of
sample carryover on the secondary trap.

Detection limits as low as 5 ppt have been reported
using the  off-line  thermal desorption-GC analysis tech-
nique for the determination of VOCs in water (Ref.3).

 CONCLUSION By using standard EPA purge-and-
 trap methodology in combination with enhanced ufL
 line, two-stage automatic thermal desorption-GC analy-
 sis, confidence in analytical methodology is retained
 and the inherent disadvantages of conventional on-line
 instrumentation are overcome. This approach offers:
    -  VASTLY  EXTENDED SAMPLE STORAGE
       CAPABILITY
    -  NO RISK OF CARRYOVER BETWEEN
       SAMPLES
    -  EXCELLENT COMPATIBILITY WITH
       HIGH RESOLUTION GC AND MS DETEC-
       TION (Figure 6)
                            In addition, by using a fully automatic two-stage de-
                            sorption instrument, this approach lends itself to
                                -   FIELD SAMPLING AND CENTRALIZED
                                   ANALYSIS.


                                      Sample tube containing 4 ng each of chloroform,
                                      benzene, toluene and p-xylene and 20 nig
                                      of water analyzed using a Model AID 400/
                                      Model 8700 CC/Mass Spectrometer System

Toluene

Chlorc

form

Benzene





p-Xylene



I/KU 	
                                                                        600
                                                                        10:01
                                                      1000
                                                      16:41
1200
20:01
                                                                             13:21

                                                                        Scanned between 78 • 91 Mass nos
1400 Scan nos
23:21 Mins
                                                                                 Figure 6
                                REFERENCES
                                1.  Certified Standard Material Reference No.
                                   CRM 112. Available from the European Com-
                                   munity Bureau of Reference:
                                      Community Bureau of Reference (BCR)
                                      Rue de la Loi 200
                                      B-1049 Brussels
                                      Belgium

                                2.  How Efficient are Capilliary Cold Trans?, J. W.
                                   Graydon and K. Grab, Chrom. 15,327,1983, pp.
                                   265-269.

                                3.  Modified Analytical Technique for Determination
                                   of Trace Organics in Water Using Dynamic Head-
                                   space and Gas Chromatography-Afass Spectrome-
                                   try.  Bianchi, Varney, and Phillips, J. of Chrom.
                                   467 (1989), pp  111-128.

                                4.  Stability of Chlorinated Hydrocarbons on Tenax,
                                   F. Lindqvist and H. Bakkeren, Netherlands Or-
                                   ganization for Applied Scientific Research, TNO
                                   Division of Technology for Society, Report No.
                                   90/268.
                                                    839

-------
            A FIELD-PORTABLE SUPERCRITICAL FLUE) EXTRACTOR FOR CHARACTERIZING
                 SEMIVOLATILE ORGANIC COMPOUNDS IN WASTE AND SOIL SAMPLES
                         Bob W. Wright, Cherylyn W. Wright, and Jonathan S. Fruchter
                                   Battelle, Pacific Northwest Laboratories
                                           Richland, Washington
INTRODUCTION

Rapid,  field-portable  methods  for  measuring  the
concentration of semivolatile  organic compounds are
desirable for on-site characterization of contaminated soils
and sediments. Supercritical fluid extraction (SFE) provides
a viable alternative to current liquid extraction methods,
which include Soxhlet and sonication methods.  Compared to
current  liquid extraction  methods, SFE is  rapid, large
quantities of glassware are not  needed, large volumes of
solvent are not required to be used or concentrated, and fewer
sample handling and sample preparation steps  are involved.
Because of these characteristics, SFE lends itself to in-the-
field extraction of solid samples of environmental concern.

In SFE, a supercritical fluid is used as a mobile phase passing
through  the solid  matrix.  The semivolatile organic
compounds of interest are  partitioned  into the supercritical
fluid, after which they are collected and analyzed. The liquid-
like solvating power and rapid mass-transfer properties of a
supercritical fluid provide the  potential for more rapid
extraction rates and more efficient extraction  due to better
penetration of the matrix than is feasible with liquids.

A prototype field-portable supercritical extractor  was
developed and tested  at  four  different field locations,
including two coal-tar-contaminated sites, a petroleum-tar-
contaminated site, and a pplychlorinated biphenyl (PCB)-
contaminated site.  In addition, the results obtained from
replicate SFE extractions of a coal-tar contaminated soil were
compared to the  results obtained from  replicate Soxhlet
extractions of the same soil; the results obtained from the SFE
extraction of several  coal-tar-contaminated  soils  were
compared to  the  results obtained  from  an on-site
micrpextracdon of replicates of the same soils; and the results
obtained from the SFE extraction of PCB-contaminated soils
were compared  to the results  obtained from analyses
performed by a CLP laboratory using replicate soil samples.
EXPERIMENTAL

Apparatus

A schematic diagram of the field-portable SFE apparatus is
shown in Figure  1.   Although not  apparent  from the
schematic diagram, the extraction cell heating mantles and
restrictor heaters are mounted on top of the apparatus and the
collection vessels are mounted vertically on its right side; this
configuration maintains the device's compact design and
allows easy manipulation of the extraction cells, restrictors,
and  collection vessels.   Overall,  the device  measures
approximately 14 in. wide by 14 in. high by 13 in. deep and
weighs approximately 23 kg. It was designed specifically for
field applications where portability, extraction speed, ease of
operation, minimal requirements for  ancillary supplies, and
sample analysis flexibility are more significant factors than in
laboratory applications. The apparatus was designed for use
with carbon dioxide, but other pressurized liquids or ambient
pressure liquids could also be used.

A reciprocating high-pressure  liquid  chromatography pump
supplies pressurized carbon dioxide  to the extraction cell,
where the sample to be extracted is housed. To prevent the
pump from vapor-locking, it is necessary to  cool the
pumphead assembly and the incoming flow of liquid carbon
dioxide;  lightweight cooling  is obtained  by single-stage
thermoelectric devices.  The pressurized carbon dioxide and
the extraction cell are heated in a cylindrical heating mantle
oven. The carbon dioxide pressure is reduced to atmospheric
pressure  through a flow restrictor  made of fused silica
capillary tubing (50 cm x 100-jun I.D.); the restrictor passes
through a heated ceramic tube furnace into the collection
vessel that contains collection solvent.  A close-up of the
collection vessel assembly is shown  in  Figure 2.   The
restrictor is  passed through a septum seal into  the glass
restrictor support tube of the glass collection flask.  Solvent is
added to the flask (usually to at  least one-half the height of the
finger, 10 to 15 mL).  A thermoelectric-cooled copper block
with a tortuous flow path is connected to the exit of the glass
collection flask to  serve as a condenser to  minimize losses
                                                       841

-------
                   1/4" S.S.
                  TUBING
1/16" S.S.
TUBING
                  FLUID SUPPLY COOLING
                  ASSEMBLY
FLUID SUPPLY
                     PRESSURE TRANSDUCER
                                                                         COLLECTION
                                                                         VESSEL
                                                                         ASSEMBLIES
                                                                                            EXTRACTION
                                                                                            CELL
                  Figure 1.  Schematic diagram of the portable, analytical-scale SFE apparatus.
    CONDENSOR
    ASSEMBLY -
                           CLASS COLLECTIOH
                           FLASK ASSEMBLY
                           RESTRICTOR
                           SUPPORT TUBE
    Figure 2.  Extract Collection Vessel Design
       due to solute volatility or entrainment in the escaping carbon
       dioxide.

       The apparatus is designed to extract a single sample at a time,
       but it has tandem sample processing capabilities with two
       extraction cell ovens, two restrictor heaters, and space for
       two  collection  vessel  assemblies.   This  allows near-
       continuous extraction of samples since one sample can be
       connected or removed from the apparatus while another one
       is being extracted.

       Field Extractions

       The equipment necessary for the extraction and analysis of
       coal-tar-contaminated soil (including the SFE apparatus and a
       gas chromatograph  for extract analysis) were  shipped via
       overnight air express to a utility site in  the Midwest. It  was
       set up in about 2 h  on the day before  samples  were taken.
       The utility provided some soil samples taken  from different
       locations on the site; several of the samples (1 to 2 g) were
       extracted using the  SFE apparatus with carbon dioxide at
       100°C  and about 290 bar for approximately 20 min.  The
       flow rate was very fast because the  soil was sandy.  2-
       Chloroanthracene (approximately 50 (ig)  was added to the
       methylene chloride collection solvent to serve as an internal
       standard.  The extract from one of the more contaminated
       samples  (IL-B-3) was analyzed in the  field  for polycyclic
       aromatic hydrocarbons (PAH) by gas chromatography (GC)
       using a fused silica capillary column  coated with DB-5 (J &
       W Scientific, Folsom, CA)  and a flame ionization detector
       (FID);  calibration  was performed with standard PAH
       mixtures at three different concentration levels.  Results are
       given in parts per million (ppm; (ig/g)  of the  soil sample as
       taken.

       Additional coal-tar-contaminated soil  samples  were extracted
       and analyzed   at a second  field  location.   Nine coal-tar-
                                                    842

-------
contaminated soil samples obtained from drill cores at the site
of a waste dump from an abandoned manufactured-gas site in
the Northeast were extracted in the field using carbon dioxide
at 100°C and between 300 and 400 bar for IS min.  The
extracts were analyzed for PAH using GC, as described
above.

Ten petroleum-based oil-tar-contaminated solid samples were
collected from  various places around a gas plant in die
Northeast, the third field location.  Some of mem required the
addition of clean soil, silica gel, or glass beads to facilitate
extraction because of their "sticky" consistency. The samples
were extracted in the field using carbon dioxide at 100°C and
about 350 bar for approximately 20 min. The extracts were
analyzed for PAH using GC, as described above.

The applicability of the SFE method for the analysis of PCB-
contaminated soils was demonstrated at a contaminated site in
the Northwest, where 17 soils were extracted and analyzed in
the field.  Fourteen  of these samples were extracted with
carbon dioxide at 100°C and pressures of 350 to 400 bar for
IS  to 20 min; three of these samples were extracted at
pressures of 250 bar for 20 to 25 min. The extracts were
analyzed in the field for Aroclor 1260 PCB compounds using
GC, as described above, except calibration was performed
using a standard Aroclor 1260 PCB mixture. Eight of the
extracts were further analyzed in the laboratory using GC
with a chlorine-specific electron capture detector (ECD).

SFE Comparisons

SFE was compared to  Soxhlet extraction using a coal-tar-
contaminated soil sample from the  second field location
described above. Five 2-g replicates were extracted using
SFE with carbon dioxide at 100°C and approximately 350 bar
for 30 min. Each of the five replicate SFE extractions were
analyzed by GC in triplicate,  and  IS individual PAH
compounds (ranging in size from two to five rings) were
quantified. Five 2-g replicates of the same soil were Soxhlet-
extracted overnight using 2SO niL each of methylene chloride.
The extracts were then concentrated using a rotary evaporator
operated at 40°C. The Soxhlet extracts were then analyzed by
GC in the same way as were the  SFE  extracts.  The
quantitative results obtained from the two extraction methods
were compared using an F-statistic at 95% confidence limits.

The same nine coal-tar-contaminated soil samples that were
extracted using SFE at the second field location described
above  were concurrently extracted in  the field by an
independent laboratory using a microextraction method.
Two-g samples of each soil were extracted in 15-mL culture
tubes with 10 mL each of 1:1 methylene chloride: acetone.
The soil-solvent mixture was agitated for 30 min, after which
the solvent was decanted and concentrated to 0.5 mL. The
extracts were analyzed for PAH by GC.  The relative
percentage differences of the quantitative results between the
SFE and microextraction methods were calculated for each
compound, these differences were then averaged for each of

results from the two field methods.

Subsamples  of  the  17 PCB-contaminated soil samples
described above  were extracted and  analyzed by  an
independent CLP laboratory using Soxhlet extraction and
analysis by GC using an ECD. The results were compared to
the SFE results.

RESULTS

The two-ringed PAH compound, naphthalene, was found to
be the PAH of highest concentration in the IL-B-3 soil sample
at 7SO ppm.   It was followed by the three-ringed PAH
compound, phenanthrene, at a concentration of 400 ppm.
The highest concentration of a heterocyclic PAH compound
was  determined to be dibenzothiophene (a three-ringed
compound containing one sulfur heteroatom) at 20 ppm. The
highest concentration of an alkylated PAH was determined to
be 2-methylnaphthalene at 320 ppm. Pyrene (a four-ringed
pericondensed PAH compound) and chrysene (a four-ringed
catacondensed PAH compound) were detected at 170 and 48
ppm, respectively.  Benzo[a]pyrene (a five-ringed PAH
compound) was detected at 36 ppm. The concentrations of
the parent PAH compounds decreased with increasing
molecular weight

Of the nine coal-tar-contaminated soil samples extracted and
analyzed at the second field site, only one contained PAH
compounds at  concentrations  greater  than  1   ppm.
Naphthalene  was  the  PAH  compound  of  greatest
concentration in this sample with a concentration of 180 ppm.
Phenanthrene was detected at 29 ppm, and pyrene at S.S
ppm.  The highest concentration of a PAH compound
detected in five of the other soil samples ranged from 0.1 to
0.9 ppm. No PAH compounds were detected in two of the
soil  samples  above the  minimum detectable  limit of
approximately 0.01 ppm.

The  levels of PAH detected in the ten  petroleum-tar-
contaminated samples ranged from low ppm to low parts per
thousand (mg/g).  The samples that contained the highest
levels of PAH were described as tar, oily tar, or sluff tar
samples.  The highest overall levels of PAH were detected in
the sample described as tar. The next highest overall levels of
PAH were detected in the two field pile samples (described to
be a mixture of soil and sluff tar) and an oily tar sample from
a manhole pit; these samples contained about five times less
PAH than did  the aforementioned tar sample. Five of the
samples were described as tar-contaminated dirt and their
PAH levels were about one order of magnitude less than
those detected in the oily tar or sluff tar samples. The lowest
levels  of PAH were detected from  a sample from a
composting tub in the curing stage.

There was wide variability in the levels of PCBs detected in
the soils from the fourth field site.The levels of Aroclor 1260
detected in the 17 PCB-contaminated soil samples ranged
from <10 to 19,000 ppm. These results show how the areas
of most contamination can be located with little turnaround
time during a  site characterization using SFE.  The ECD
results obtained later in the  laboratory were all either less than
or the same as the FID results obtained in the field; these
results were not surprising since many of the soil samples
were contaminated with diesel fuel and hydraulic  oil, which
may have caused the FID results to be inaccurate because of
coeluting hydrocarbon compounds.
                                                     843

-------
When the quantitative results from five SFE extractions and
five Soxhlet extractions were compared for  15 PAH
compounds ranging from two to five rings in size, it was
found that statistically significant differences could only be
detected for two of the IS  individual  PAH compounds.
Thirteen of the compounds were detected at the same level for
both extraction methods.  The  two compounds that gave
differing  results  were  both high-molecular-weight
benzopyrenes; for these compounds slightly lower amounts
(approximately 20%) were detected in the SFE extracts than
were detected in the Soxhlet extracts. These results indicate
good agreement between the SFE and the traditional Soxhlet
extraction methods.

The same two coal-tar-contaminated  soil samples that
contained  no detectable limits of PAH compounds when
extracted using SFE also contained no detectable limits of
PAH when extracted using the microextraction method.  For
the  seven other samples that were extracted by the SFE and
microextraction methods, the relative percentage differences
between the results of the two extraction  methods ranged
from 16 to 44%.  The data indicated that SFE generally gave
higher concentrations  of the  lower-molecular-weight
compounds (two rings in size) than did the microextraction
method. Overall results, however, indicated that the two field
methods were comparable for these particular soil samples,
especially considering the extremely low levels of PAH
contamination that were present (mainly in the parts  per
billion, ng/g, range), and that each sample was only analyzed
one time by each method

The comparison of the results from the field SFE extraction
and analysis of PCB-contaminatcd soils to those obtained by
an independent CLP laboratory indicated the two methods
gave results of the same order of magnitude.

CONCLUSIONS

SFE was shown to be a rapid, field-portable method for the
analysis  of PAH and PCBs in  soil  and other solid
environmental samples. The SFE method gave results that
were comparable to the results obtained by traditional
extraction methods.

SFE should prove useful as an efficient means  for rapid
characterization during site  assessments and at sites
undergoing remediation treatments. The method should also
be applicable for measuring  the concentration of other
semivolatile organic compounds of environmental concern.

ACKNOWLEDGMENT

This work was supported by the Electric Power Research
Institute,  Land  and Water Quality Studies Program,
Environment Division, under Contract RP-2879-04.
                                                      844

-------
                              Detection of mercuric ions in water with
                                      a mercury-specific antibody
                                        Dwane E. Wylie
                                        Larry D. Carlson
                                        Randy Carlson
                                        Fred W. Wagner
                                        Sheldon M. Schuster
               INTRODUCTION
Exposure to toxic amounts of mercury can lead to
serious health problems, with long-term
consequences for the affected population. Thus,
simple, sensitive, and convenient procedures are
needed for detection of mercury in the environment
to prevent these problems from arising.

In this paper, an ELISA is described that detects
mercury at concentrations of 0.5 ppb or greater in
water. Between 0.5 and 10 ppb mercury, the
absorbance is proportional to the log of the mercury
concentration. The assay is specific for mercury, in
that no other metal tested interferes with the
quantitation of mercury. In addition, the assay
requires little preliminary processing of the sample
and can be done with only one hundred microliters of
sample.

         MATERIALS AND METHODS
Materials
An EP Extract Metals Quality Control Sample was
obtained from the Environmental Protection Agency,
Quality Assurance Branch, Environmental Monitoring
and Support Laboratory, Cincinnati, OH, 45268.  The
sample contained 0.2 mg/L Hg"^,  100 mg/L Ba4"1", 1
mg/L Cd44", 5 mg/L Cr44+, 5 mg/L Pb44, and 5 mg/L
Ag+ in distilled water adjusted to pH 5.0 with acetic
acid.

Standard Reference Materials 1641 and 3133 were
obtained from the National Institute of Standards and
Technology, Office of Reference Materials,
Gaithersburg, MD, 20899. SRM  1641 consisted of
mercury at a concentration of 1.52 ug/ml in 2% nitric
acid, and SRM 3133 contained 10 mg/ml mercury (as
16.2 mg/ml mercuric nitrate) in 10% nitric acid.
Detection of mercury in wqfor by enzvme-linked
immunosorbent assay
Ninety-six-well microtiter plates (EIA/RIA grade.
Costar Corp.. Cambridge, MA) were treated with
BSA-glutathione, blocked, and used for ELISA. One
hundred microliter aliquots of water containing
known amounts of mercuric chloride, ranging from
0.2-200 parts per billion, were added to the wells of
the microtiter plate for 30 minutes. The plates were
washed three times, then ascites fluid containing a
mercury-specific monoclonal antibody was added for
30 minutes at room temperature, followed by goat
anti-mouse u chain conjugated to horseradish
peroxidase (Kirkegaard and Perry Laboratories, Inc.,
Gaithersburg, MD). After incubation for 30 minutes at
room temperature, the plates were washed, and 100
ul of ABTS peroxidase substrate (Kirkegaard  and
Perry Laboratories, Inc., Gaithersburg,  MD)were
added to each well. After 15 minutes incubation, the
absorbance of each well at 405 nm was measured
with  a Trtertek Multiscan MC multichannel
spectrophotometer (Flow Laboratories,  Rockville,
MD).

Cold-vapor atomic absorption spectrometry
Mercury concentrations of some samples were
determined by cold-vapor atomic absorption in the
Diagnostic Laboratory, Department of Veterinary
Sciences,  University of Nebraska-Lincoln with a
Mercury Monitor flameless atomic absorption
spectrophotometer (Model 1255, Milton Roy Inc.,
LDC Division, Riviera Beach, FL).  Before analysis,
the samples were treated with SnCl2 in 10%  HCI to
reduce mercuric ions to elemental mercury, and the
mercury concentration was determined by
comparison with a mercury standard (Mercury
Reference Standard Solution, Fisher Scientific)
treated in the same manner.
                                                 845

-------
For direct comparison of mercury quantitation by
atomic absorption and ELISA, the mercury standard
was diluted to a nominal concentration of 100 ppb in
0.1 M HEPES, pH 6.8. Two aliquots were then
removed. One was diluted in HEPES buffer to the
appropriate concentrations for analysis by
immunoassay, while the other was diluted in 10%
nitric acid ('Baker Analyzed' 70-71%, Trace Mineral
Analysis, Baker Chemical Co.) for atomic absorption
measurement. Mercury was then measured as
described above for each method.

Mercury quantitation in EPA and NIST samples
Each of the samples from the Environmental
Protection Agency and the National Institute of
Standards and Technology was diluted in water to
mercury concentrations of 1-200 ppb, then used in
the ELISA as described above.  The results were
compared with a standard curve constructed from
ELISA analysis of water containing known
concentrations of mercury.  The mercury
concentrations of the samples used for construction
of the standard curve were also measured by atomic
absorption.

Interference with mercury detection bv other metals
in the ELISA
A 2 mM solution of each metal salt in water was
diluted to concentrations of 20 uM, 200 nM, 20 nM,
and 2 nM. Fifty microliters of each concentration
were added to individual microtiter wells treated with
BSA-glutathione. Fifty microliters of  SRM 3133
containing mercury at concentrations ranging from 1-
200 ppb were added to the appropriate wells. The
plates were incubated at room temperature for 30
minutes,  after which time the plates were washed
and assayed by the ELISA described above.

         RESULTS AND DISCUSSION
An immunoassay capable of detecting small
amounts of mercury in water was developed with the
use of an antibody that reacts with immobilized
mercuric ions. Table 1 shows the results of seven
replicate  analyses for each mercury  concentration,
along with the means, standard deviations, and
coefficients of variation. Absorbance approximately
twice that of background was consistently noted for
mercuric ion concentrations as low as 0.5 ppb when
compared to water with  no added mercury, and
concentrations of 0.2 ppb were 50% above
background.  Frequently, concentrations of mercuric
ions at 0.1 ppb demonstrated absorbance in this
same range (data not shown). A linear relationship
between A4Q5 and the log of the mercury
concentration was obtained in the range of 0.5-10
ppb, as indicated by a correlation coefficient of 0.998
within this interval.
In addition to its linearity, the assay was highly
reproducible.  Standard deviations were less than
11% of the mean at 0.2 ppb and generally decreased
as the concentration of Hg++ increased to 10 ppb. In
all cases, except for the sample containing no Kg**,
the coefficient of variation was 10% or less. These
results also indicated that the ELISA was as sensitive
for mercuric ion detection as the atomic absorption
procedure recommended by the EPA, which is
capable of mercury detection down to 0.2 ppb, but
requires a 100 ml sample to do so (1).
Since cold-vapor atomic absorption is currently the
method of choice for mercury determination, it was
important to determine how well ELISA results
correlated with atomic absorption analyses. To do
so, an atomic absorption mercury reference standard
was diluted in 0.1 M HEPES, pH 6.8, to a mercury
concentration of 100 ppb.  At this point, two aliquots
were removed and diluted to the appropriate
concentrations for immunoassay or atomic
absorption as described in Materials and Methods.
Samples containing 0, 2, 4, 6,10, and 15 ppb
mercury were then analyzed by both methods. As
shown in Figure 1, the results obtained from the two
methods were in dose agreement, as indicated by a
correlation coefficient of >0.99. In addition, the
standard deviation of the immunoassay at most
mercury concentrations was the same or less than
that obtained by atomic absorption. These results
demonstrated that, under the conditions of this assay,
quantitation of mercury by ELISA was as precise as
cold-vapor atomic absorption.

Since most samples for mercury analysis by cold-
vapor atomic absorption are stabilized in strong acid,
it was of interest to determine whether the
immunoassay could detect mercury in samples
treated similarly. Two samples obtained from the
National Institutes of Standards and Technology,
SRM 3133, which consisted of mercuric acetate in
10% nitric acid, and SRM  1641, which contained
metallic mercury in 2% nitric acid, were assayed by
the Hg++-specific ELISA.  Each sample was diluted
in water to mercury concentrations from 1 ppb to  100
ppb before analysis. As shown in Figure 6, mercury
could be detected in each sample at concentrations
of 1 ppb, although the absorbance at that
concentration was approximately half that obtained
with water containing the same amount of mercury.

The specificity of the assay for mercury was
investigated with the use of an EPA quality control
sample containing 0.2 mg/L Hg44. 100 mg/L 63++, 1
mg/L Cd44, 5 mg/L Cr444, 5 mg/L Pb44, and 5 mg/L
Ag+ in distilled water adjusted to pH 5.0 with acetic
acid. The sample was diluted to known Hg44
concentrations, which were assayed by ELISA and
                                                  846

-------
compared to results obtained with standards
consisting of known concentrations of mercuric
chloride in water (Figure 3).  Reactivity was obtained
with both the EPA sample and the water standard at
2 ppb mercury, and the absorbance for both samples
was linear up to 20 ppb mercury. Reactivity was due
to the presence of mercury and not to recognition of
one of the other metals, since a sample containing all
of the metals except mercury in the same
concentrations as in the EPA sample gave the same
absorbance as water containing no mercury.

The results in Figure 3 did not reveal whether higher
concentrations of these or other metals would
interfere with the assay.  Therefore, concentrations of
individual metal ions from 1  mM to 10 nM were
examined for interference with detection of various
concentrations of mercury in SRM 3133. Several
metal salts, including ferrous sulfate, lead acetate.
selenium dioxide, and silver nitrate, did not interfere
with mercury detection, even when they were present
at a concentration of 1 mM and mercuric ion was only
2 ppb. Other metal salts, however, including barium
chloride, cadmium chloride, chromic chloride, cupric
chloride, gold chloride, nickel chloride, and zinc
chloride, did interfere, but usually only at the highest
concentration (1 mM), although gold chloride also
demonstrated interference at 10 uM. Figure 4
represents results obtained with barium chloride,
which is typical of all metal chloride salts tested.
except gold chloride, which also demonstrated some
interference at a concentration of 10 uM.

ELISA assays have been adapted to a variety of
analytical procedures in recent years because of the
exquisite specificity of monoclonal antibodies. The
assay described here  involves recognition of
immobilized mercuric ions by a specific monoclonal
antibody. The use of an ELISA for detection of metal
ions circumvents many problems associated with
atomic absorption.  For instance, samples can be
analyzed in parallel, enabling large numbers of
samples to be processed at one time. In addition,
quantitative analysis can be performed with a simple
spectrophotometer or microtiter plate reader.
Automation of the photometer thus makes practical
the processing of a large number of samples,
allowing for the implementation of large-scale
monitoring programs.  Since the assay yields a
visible color change, semi-quantitative procedures
can be developed which require no electronic
instrumentation for evaluation. Thus, the assay has
the potential for field use. Finally, the procedure
requires only 0.1 ml of sample, up to 1000-fold less
than required by atomic absorption for maximum
sensitivity, and it can. therefore, be used to analyze
samples available in volumes insufficient for cold-
vapor atomic absorption.
                   References
 1.  Methods 7470 and 7471.  (1986) Test Methods
for Evaluation of Solid Waste, Physical/Chemical
Methods.  2nd Edition. Office of Solid and
Emergency Response, U.S. Environmental
Protection Agency, Washington, O.C.

 2.  Determination of Mercury  by the Cold Vapor
Technique. (1985) p. 171-173.  Standard Methods
for the Examination of Water and Wastewater.
Sixteenth Edition. American Public Health
Association. Washington, D.C.

           Table 1. Statistical analysis of ELISA data from
                  mercury detection in water.
                 Mercury concentration (ppb)
          00
               0.2
                     0.5
                          1.0
                                2.0
                                     5.0
                                           10.0
1
2
3
4
5
6
7
mean
stddev.
coeff. var.
0.196>
0.1S3
0.140
0.123
0.108
0.123
0.113
0.137
0.030
22.277
0.302
0.272
0.272
0.278
0.237
0.303
0.280
0.278
0.022
7.994
0.401
0.469
0.413
0.496
0.445
0.398
0.520
0.449
0.048
10.710
0.759
0.765
0.749
0.787
0.711
0.716
0.588
0.725
0.066
9.118
1.123
1.180
1.338
1.323
1.195
1.093
1.044
1.185
0.112
9.419
1.592
1.750
1.665
1.817
1.751
1.610
1.717
1.700
0.082
4.806
2.064
2.000
1.988
2.053
1.963
2.059
1.968
2.014
0.044
2.187
 avalues represent the absorbance at 405 run of ELISA analyses done

 as described in Materials and Methods. The data shown are the same

 as used to derive the graph in Figure 1. The correlation coefficient (r)

 between A4Q5 and the log of the mercury concentration between 0.2

 and 10 ppb is 0.998.
                                                   847

-------
M
(A
E
     ts
     10-
o
u

e
u
Ol
              r > 0.994
                                                                        2-
                                                                   405
                                                                         I -
                                      10
                                                     IS
                                                                                                 10
                                                                                                            100
              Hg Concentration by A A.  (ppb)

    Figure 1. Comparison of mercury detection by ELISA
    and atomic absorption. An atomic absorption
    mercury reference standard was diluted to mercury
    concentrations of 2,4, 6,10, and 15 ppb in either 0.1
    M HEPES, pH 6.8. for analysis by ELISA or in 10%
    nitric acid for cold-vapor atomic absorption. Each
    sample was then analyzed as described in Materials
    and Methods. The values shown represent the mean
    and one standard deviation of quadruplicate
    analyses by imrnunoassay and triplicate analyses  by
    atomic absorption.
405
               Hg  Concentration (ppb)
Figure 3. Detection of mercury in the EPA Quality
Control sample by ELISA.  The QC sample (») and
HgClg (E) were diluted in water to mercury
concentrations ranging from 0.5-200 ppb, then
analyzed by ELISA as described  in Materials and
Methods. A sample was included that contained the
same concentration of all other metals as the QC
sample but without mercury (a). The absorbance
obtained in  analysis of both water without added
mercury and the EPA sample without mercury was
0.263. Each point represents the average
absorbance obtained from quadruplicate analyses of
each sample.
                                                                             Bad
                                                                         2-
                                                                   405
                               10
                                          100
                                                                                                  10
                                                                                                             100
                   Hg  Concentration  (ppb)

   Figure 2.  Detection of mercury in National Institute of
   Standards and Technology Standard Reference
   Materials. SRM 1641 (0) and SRM 3133 (*) were
   diluted in water to mercury concentrations of 1.10,
   and 100 ppb, then analyzed by ELISA as described
   in Materials and Methods. A control consisting of
   known concentrations of mercury in water was
   included for comparison (ffi).  SRM 1641 consisted of
   metallic mercury at a concentration of 1.52 ug/ml in
   2% nitric acid, and SRM 3133 contained 16.2 mg/ml
   mercuric nitrate in 10% nitric acid. Each point
   represents the average absorbance obtained from
   quadruplicate analyses of each sample.
               Hg Concentration  (ppb)
 Figure 4. Effect of barium chloride on mercuric ion
 detection by ELISA. Barium chloride at 0 nM (a). 1
 nM (•). 10 nM (*), 100 nM (a). 10 uM (»). and 1 mM
 (0) concentrations were added to mercuric ion
 standards to determine their effects on the
 quantitation of mercury at concentrations ranging
 from 0.5-200 ppb.  For each concentration of metal
 salt, a control containing the same concentration of
 metal salt but with no added mercury was included.
 These values were below 0.2. Each point represents
 the average absorbance obtained from
 quadruplicate analyses of each sample.
                                                             848

-------
       THE EFFECTS OF PRESERVATIVES ON RECOVERY AND ANALYSIS OF VOLATILE
                                     ORGANIC COMPOUNDS.
 Kaveh Zarrabi, Steven Ward, Thomas Starks and
              Charles Fitzsimmons.
         Environmental Research Center
         University of Nevada-las Vegas
OBJECTIVE:  To evaluate the effects of selected
preservatives (stabilizers) on the recovery of volatile
organic compounds (VOCs) from contaminated soils.
The  study  provides practical improvement to  the
current   sampling,  preservation,  and   analytical
methods for VOC measurement in soil.

APPROACH:  Prior to evaluation of preservatives,
two concerns should have been addressed.

a)    The establishment  of  an appropriate delay
      time beyond spiking to achieve equilibrium
      between spiked VOCs and the soil matrix.
      Second, allowing for  significance losses of
      spiked VOCs in soil to be able to evaluate the
      effectiveness of preservatives.

b)    To  choose an effective method  to  retard
      biodegradation of VOCs in soil from time of
       spiking till completion  of analysis.

Upon establishment of the  above objectives, we
proceeded   with  evaluation   of  the  following
preservatives:

1)     The effects of methanol/water mixture  (1%
       and 10%), as the  extracting solvent  in the
       purge and trap analysis of VOCs in soil.

2)     The effects of two types of anhydrous salts in
       preservation of VOC contaminated soil at low
       moisture content.

3)     The effects of two types of anhydrous salts in
       preservation of VOC contaminated soil at
       high moisture content.
4)     The  effects of  two  solid  adsorbents  in
       preservation of voc contaminated soil at low
       moisture content.

               EXPERIMENTAL

Soils: Three real soils each with different TOC. The
soils were collected from same geological area but
different horizons.

VOCs:      Chloroform,   1,1,1-Trichloroethane,
Trichloroethene,   Tetrachloroethane,   1,1,2,2-
Tetrachloroethane,  Benzene,   Ethylbenzene  and
Toluene at the level of 150 ug/Kg.

Time Delay:   The delay time experiments  were
conducted for 8 days beyond spiking with sampling
intervals at 0, 3, and 8 days.  Statistical analysis of
data revealed after 3 days and from 3 to 8 days
storage at 4°  C  significant  losses of  VOCs are
occurring.   The  U.S.  EPA's  contract laboratory
program specifies analytical methods which include
holding time requirements for all  soil and water
samples collected  through Superfund and  RCRA.
These programs require analyses of VOCs in soil and
water to be  completed within 10 days of sample
receipt  by the laboratory. The result indicated the
importance  of  stabilizers and preservatives for
accurate measurement of VOCs in contaminated soil.

Biodegradation:  A literature  review revealed  no
preservative have been used to stop biodegradation of
VOCs in soils. However, mercuric chloride was used
to retard biological activities in soil  and water. We
used mercuric chloride at a rate of 2.5 mg/ 5.0 g soil.

Aqueous Methanol: Mixture of methanol/water (1%
and 10%)were prepared to reduce surface tension of
extracting solvent  (water) used in purge and trap
method of analysis for  VOCs.  Soils with organic
contents  higher   than   1.5%  did   show   any
improvement.  Soils with very low organic  content,
however, resulted in improved recovery.

Anhydrous  Salt   (Low  and  High  Moist  Soil):
Anhydrous salts were added to soils  at two moisture
                                                 849

-------
levels to pick up water from soil matrices and release
the adsorption sites.  In addition,  the added salts
could have produced a saline solution by adding water
prior to purge and trap method of analysis.  The
result  indicated no  improvement  on recovery of
VOC's were noticed even at higher moisture content.

Solid Adsorbents: Solid adsorbents (desiccants) were
added  to different soil types.  The statistical analysis
of results revealed solid adsorbents  are found to be
significantly more likely to give higher recoveries of
spiked  VOCs (P=0.05) in 42 out  of 60 situation
tested.  Solid  adsorbents always significantly gave
better  recoveries for soil  C  (  TOC <0.1%).  In
addition the effects of sample prep using current EPA
methods and utilization of solid adsorbents were
evaluated.

Detection Limits: The detection limits of an SRI gas
chromatograph in series with a Tekmar LSC-2000
sample concentrator and a Dynatech  PTA-30S
autosampler were evaluated  using water and soil
control samples.

                CONCLUSIONS

Further work is needed to develop more preservatives
especially for VOCs and field test current findings of
this study.
                                                    850

-------
                            Second  International Symposium
                              Field Screening Methods  for
                           Hazardous Wastes  and Toxic  Chemicals
                                 February 12-14, 1991
                            Sahara  Hotel - Las Vegas,  Nevada
                                Final  Participants'  List
Kim Aaron
Manager
Jet Propulitlon Laboratory
4800 Oak Grove Drive
Pasadena, CA  91109
U.S.A.
818-354-2816

David I. Actor
Manager of Sampling Programs
Chemical Waste Management, Inc.
ISO W. 137th Street
Riverdale, IL  60627
U.S.A.
708-841-8360

Robert E. Adams
Head, Instrumental Analysis
Southern Research Institute
2000 Ninth Avenue South
Birmingham, AL  3S20S
U.S.A.
205-581-220'.

James H. Adams, Jr.
Chief, Inorganic Section
U.S. EPA
Region V, CRL
536 South Clark Street
Chicago, IL  60605
U.S.A.
312-353-9604
Jane Anderson
Environmental Protection Spec.
U.S. EPA
Region I Superfund
556 E. Seventh Street
South Boston, MA  02127
U.S.A.
617-464-0918

Thomas D. Anderson
Chief, EDT&E Branch
U.S. Department of Energy
12800 Middlebrook Road
Germantown, MD  20874
U.S.A.
301-353-7295

Isamu [Sam) Aoki
Environmental Scientist
U.S. Department of Energy
725 DOE Place
Idaho Falls. ID  83404
U.S.A.
208-526-0583

Alutu E. Arave
Senior Engineering Specialist
EG&G Idaho, Inc.
Box 1625
Idaho Falls, ID  83415-1203
U.S.A.
208-526-2481
Jason Ai
Staff Chemist
Dames & Moore
2025 First Ave.
Suite 500
Seattle, WA  98121
U.S.A.
206-728-0744

Linda Alexander
Supelco, Inc.
Supelco Park
Beliefonte, PA  16823
U.S.A.
814-359-3441

Richard Alt
President
Bionomics Laboratory
4310 E. Anderson Road
Orlando, FL  32812
U.S.A.
407-851-2560

Paris Althouse
Technician
Lawrence Livermore National Laboratory
7000 East Avenue
Livermore, CA  94550
U.S.A.
Joseph Arlauskas
Mngr. Environ. Chemistry Lab
Science Applications Int'l Corp.
4224 Campus Point Court
San Diego, CA  92121
U.S.A.
619-535-7693

Janice Armour
Senior Scientist
Lockheed
1050 East Flamingo
Las Vegas, NV  891119
U.S.A.
702-361-0220 Ext. 253

Ted Amdorff
Optimal Technology
6430 Via Real
Suite 6
Carpinteria, CA  93013
U.S.A.
605*664-6226

Neil S. Arnold
Femtoscan Corporation
1834 West
4700 South
Salt Lake City, Utah  84119
U.S.A.
801-581-8431
                                          851

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February  12-U, 1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final Participants' List
 Kenneth  Asbury
 Program  Manager
 Lockheed ESC
 1050 E.  Flamingo Road
 Suite  120
 Las Vegas, NV  89119
 U.S.A.
Doris Ash
Research Chemist
Tennessee Valley Authority
Nat'l Fertilizer & Environ-
  mental Research Center
Muscle Shoals, AL  35661
U.S.A.
205-386-2458

John B. Ashe
President
Ashe Analytics
1705 Capital of Texas Hwy. S.
Suite 202
Austin, TX  78746
U.S.A.
512-327-6945

L. Rudolph Askew
V.P. Marketing & Sales
Viking Instruments Corporation
12007 Sunrise Valley Drive
Reston, VA  22091
U.S.A.
703-758-9339

Peggy Augustin
Ruska Instrument Corporation
P.O. Box 630009
Houston, TX  77263-0009
U.S.A.
713-975-0547

Gerry Auth
Midac Corporation
1599 Superior Ave.
Suit* B-3
Costa Mesa, CA  92627
U.S.A.
714-645-4096

Linda Baetz
Chemist
Army Environ. Hygiene Agency
Attn:  HSHB-ML-A
Aberdeen Proving Ground, MD  21014-5422
U.S.A.
301-671-3269

Carl Bailey
Environmental Scientist
U.S. EPA
25 Funston Road
Kansas City, KS  66115
U.S.A.
913-236-3881
Ronald J. Baker
Environmental Engineer
U.S. Geological Survey
810 Bear Tavern Road
Trenton. NJ  08628
U.S.A.
609-989-2016

John Barich
Environmental Engineer
U.S. EPA
Region X
1200 Sixth Avenue
Seattle, WA  98101
U.S.A.
206-553-8562

Helen J. Barker
Sales/Marketing Manager
Kevex Instruments, Inc.
355 Shoreway Road
San Carlos, CA  94070
U.S.A.
415-591-3600

Bob Barry
Senior Research Scientist
Millipore Corporation
80 Ashby Road
Bedford, MA  01730
U.S.A.
800-225-3384  Ext. 8768

Delbert Barth
Director, ERG
University of Nevada
Environmental Research Center
4505 Maryland Parkway
Las Vegas, NV  89154-4009
U.S.A.
702-739-3382

David L. Bartley
Research Physicist
NIOSH
4676 Columbia Parkway, R-8
Cincinnati, OH  45226
U.S.A.
513-841-4277

Marllew Bartling
Project Manager
Martin Marietta Energy Systems
k-25 Site, Hwy. 58,  Drop A-20
Oak Ridge, TN  37830
U.S.A.
615-574-5270

Raymond J. Bath
Toxicologist
KUS Corporation
1090 King Georges Post Road
Edison, NJ  08837
U.S.A.
201-225-6160
                                           852

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February 12-1*, 1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final Participants'  List
Wolfgang Bather
Manager
Draegerwirk
Moislinger Allee 53-55
Lubeck,   D-2400
Germany
0451-882-3138

Gorman Baykut
Bruker-Franzen Analytik
GMBH, Bremen
Fahrenheit Str. 4
D-2800 Bremen 33,
Germany
49-421-2205164

Werner F. Beckert
Research Chemist
U.S. EPA
EMSL-LV
P.O.Box 93478
Las Vegas, NV  B9193-3478
U.S.A.
702-798-2137

Willard Becraft
Manager, Business Development
Thermo Instrument Systems, Inc.
114 Crestview Lane
Oak Ridge, TN  37830
U.S.A.
615-482-7227

Bernie Beemster
Environmental Mktg.  Consultant
Biotronics Technologies, Inc.
12020 W. Ripley Avenue
Wauwatosa, MI  53226
U.S.A.
414-475-7653

Joseph V, Behar
Environmental Scientist
U.S. EPA
EMSL-LV
P.O.Box 93478
Las Vegas, NV  89193-3478
U.S.A.
702-798-2216

Michael G. Bell
Analytical Chemist
Los Alamos National Laboratory
P.O. Box 1663
E 518
Los Alamos,  KM  87545
U.S.A.
505-667-9996

Suzanne Bell
Los Alamos National Laboratory
MS K484
Los Alamos,  NM  87544
U.S.A.
505-667-0175
Thomas H. Bellus
Scientist
Hestlnghouse Hanford
P.O. Box 1970
Mail Stop 50-61
Richland, WA  99352
U.S.A.
509-373-4871

Jim Bentley
Division Director
Enseco - CRL
7440 Lincoln Way
Garden Grove, CA  92641
U.S.A.
714-898-6370

Donald R. Berdahl
Staff Chemist
GE - CRD
1 River Road
P.O. Box 8
Schenectady, NY  12301
U.S.A.
518-387-6083

Richard E.  Berkley
Research Chemist
U.S. EPA
AREAL/RTP
MD-44
Research Triangle Park, NC  27711
U.S.A.
919-541-2439

Bernie B. Bernard
Vice President/Dir of Tech
O.I.Analytical
P.O. Box 2980
Graham at Wellborn Rds.
College Station, TX  77841
U.S.A.
409-690-1711

Mark Bernick
Chemist
Roy F. Weston
GSA Raritan Depot
2890 Woodbridge Ave. 209 Annex
Edison, NJ  08837
U.S.A.
908-906-3489

John Bernstchy
Chemist
Science Applications Infl Corp.
4224 Campus Point Court
San Diego,  CA  92121
U.S.A.
619-535-7432
                                          853

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Hastes and Toxic Chemicals
                                 February 12-14. 1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final Participants'  List
Peter F. Berry
Chief Scientist
TN Technologies, Inc.
2555 North IH-35
Round Rock, TX  78664
U.S.A.
512-388-9211

Don Betowski
Research Chemist
U.S. EPA
EMSL-LV
P.O. Box 93*78
Las Vegas, NV  89193-3478
U.S.A.
702-798-2116

Philip J. Bierbaum
Director
NIOSH
4676 Columbia Parkway
Cincinnati, OH  45251
U.S.A.
513-841-4321

Stephen Billets
Chairperson
U.S. EPA
EMSL-LV
P.O. Box 93478
Las Vegas, NV  69193
U.S.A.
702-798-2232

Richard C. Birkmeyer
Research Director
Strategic Diagnostics, Inc.
128 Sandy Drive
Sandy Bear Industrial Park
Newark, DE  19713
U.S.A.
302-456-6789

David Blair
Orr Safety Corporation
2360 Mi liens Lane
Louisville, KY  40216
U.S.A.
502-774-5791

Ken Blake
Product Manager
IN Technologies, Inc.
2555 N I.H. 35
Round Rock, IX  78680
U.S.A.
512-308-9183

William Blanton
ManTech Environmental Technology
10625 Fallstone Road
Houston, TX  77099
U.S.A.
713-983-2115
Wayne Bliss
Director
Office of Radiation Programs
Las Vegas
P.O. Box 98517
Laa Vegas, NV  89193-8517
U.S.A.
702-798-2476

David A. Blyth
Senior Scientist
Geo-Centers, Inc.
10903 Indian Head Highway
Ft. Washington, HD  20744
U.S.A.
301-671-2416

Itamar Bodek
Arthur D. Little
15 Acorn Park
Cambridge, MA  02140
U.S.A.
617-664-5770

Mark Boedigheimer
Technical Group Director
CH2M Hill
2300 Northwest Walnut Blvd.
Corvallis, OR  97339
U.S.A.
503-752-4271

Kevin Bolger
Chemist
U.S. EPA
536 S Clark
Chicago, IL  60605
U.S.A.
312-353-7712

Patricia M. Boone
FASF Coordinator [Region II]
U.S. EPA
2890 Woodbridge Avenue
MS-220
Edison, NJ  06837
U.S.A.
201-906-6998

Robert L. Booth
Senior Science Advisor
AScI Corporation
26 W. Martin Luther King Drive
Cincinnati, OH  45268
U.S.A.
513-569-7364

David Bottrell
Chemist
U.S. Department of Energy
OTD/PSD/LMB
18901 Tributary Lane
Gaithersbuzg, KD  20879
U.S.A.
301-353-7251
                                            854

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Wastes  and Toxic Chemicals
                                 February  12-14, 1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final  Participants' List
William W. Botts
President/CEO
O.I. Analytical
P.O. Box 2980
Graham at Wellborn Rds.
College Station, TX  77841
U.S.A.
409-690-1711

Robert A. Bove
VF Domestic Sales & Marketing
Gilian Instrument Corporation
35 Fairfield Place
West Caldwell, NJ  07006
U.S.A.
201-808-3355

Matthew A. Bower
Senior Staff Scientist
ATEC Environmental Consultants
8989 Herrmann Drive
Columbia, MD  21045
U.S.A.
301-381-0232
                                                   J.L. Brokenshire
                                                   Graseby  Ionics Ltd.
                                                   Analytical Division
                                                   Park Avenue, Bushey
                                                   Watford, Herts,   WD2 2BW
                                                   United Kingdom
                                                   0923-38483

                                                   Buck Brooks
                                                   Ecologist
                                                   Ecology  and Environment, Inc.
                                                   2504 HE  61st Street
                                                   Apt. #14
                                                   Gladstone, MO  64118
                                                   U.S.A.
                                                   816-459-8993

                                                   Kathy Brown
                                                   Oak Ridge National Laboratory
                                                   Bldg. 7S03
                                                   MS  6382
                                                   Oak Ridge, IN  37831-6382
                                                   U.S.A.
                                                   615-574-7808
William D. Bowers
Femtometrics
1721 Whittier Avenue
Suit* A
Costa Mesa, CA  92627
U.S.A.
714-722-6239

Theresa Brandabur
Geologist
ICF Kaiser Engineers
160 Spear Street
Suit* 1380
San Franciaco. CA  94105
U.S.A.
415-882-3044

Brian Brass
Environmental Scientist
Roy F. Weston
2890 Woodbridge Avenue
Edison, NJ  08837
U.S.A.
201-632-9772
                                                    Ken W.  Brown
                                                    Technical Support Ctr.  Manager
                                                    U.S.  EPA
                                                    944 E.  Harmon
                                                    Las Vegas,  NV  89119
                                                    U.S.A.
                                                    702-798-2270

                                                    William L.  Brown
                                                    Analytical Chemist
                                                    Pace, Inc.
                                                    1710 Douglas Dr. N
                                                    Minneapolis, MN  55422
                                                    U.S.A.
                                                    612-525-5476

                                                    William C.  Brumley
                                                    Research Chemist
                                                    U.S.  EPA
                                                    EMSL-LV
                                                    P.O.Box 93748
                                                    Las Vegas,  NV  69193-3478
                                                    U.S.A.
                                                    702-798-2684
Michael G. Bray
Senior Chemist
Ecology and Environment
101 Yesler Hay
Suite 600
Seattle, WA  98104
U.S.A.
206-624-9537

Jon Broadway
Engineer
U.S. EPA
1504 Avenue A
Montgomery, AL  36115-2601
U.S.A.
205-270-3434
                                                    W.L.  Brutscher
                                                    Technical Associate
                                                    Shell Development
                                                    P.O.  Box 1380
                                                    Houston, TX  77001
                                                    U.S.A.
                                                    713-493-7260

                                                    Rex Clair Bryan
                                                    Geostatician
                                                    Viar Company
                                                    2221 East Street
                                                    Golden.  CO  80401
                                                    U.S.A.
                                                    303-277-0070
                                          855

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Hastes and Toxic Chemicals
                                 February 12-14, 1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final Participants  List
Ralph J. Bulger
President
Andersen Instruments, Inc.
4180 Fulton Industrial Blvd.
Atlanta, GA  30336
U.S.A.
404-691-1910

Deborah A. Bunlskt
Clean Technologies
DCS Annex - Academy Street
Newark, DE  19715
U.S.A.
302-368-7961

Lloyd W. Burgess
Research Scientist
University of Washington
CFAC
BG-10
Seattle, WA  98101
U.S.A.
206-543-0579

Ed Burroughs
Industrial Hygienist
HIOSH
4676 Columbia Parkway
Cincinnati, OH  45226
U.S.A.
513-841-4275

Neill  D. Butcher
Technical Assistance Team
Ecology and Environment
160 Spear Street, #930
San Francisco, CA   94105
U.S.A.
415-777-2811

William Buttner
Transducer Research, Inc.
999 Chicago Avenue
Naperville, IL  60540
U.S.A.
708-357-0004

Lupe  Buys
Lab Licensing Consultant
Arizona Dept.  of Health Services
1520  W. Adams
Phoenix,  A£  85007
U.S.A.
602-542-6100

Robert H.  Caldwell
Marketing Executive
 Graseby Ionics
 10640 Main Street
 Suite 200
 Fairfax, VA  22030
 U.S.A.
 703-352-3400
Jim Cannon
Millipore Corporation
80 Ashby Road
Bedford, MA  01730
U.S.A.
617-275-9200  Ext. 2337

Robert Carley
Laboratory Manager
University of Connecticut
191 Auditorium Road
Storr, CT  06269
U.S.A.
203-486-5488

Clark Carlson
Senior Inorganic Chemist
The Bionetics Corporation
7411 Beach Drive East
Port Orchard, MA  98366
U.S.A.
206-871-0748

Randy Carlson
Research Scientist
BioNebraska
319 Manter Hall
City Campus
Lincoln, HE  6858B
U.S.A.
402-472-2767

Kenneth R, Carney
Research Associate
Institute for Environmental  Studies
Louisana State University
Baton Rouge, LA   70803
U.S.A.
504-388-4281

Michael M. Carrabba
Group Leader
EIC  Laboratories,  Inc.
 Ill  Downey St.
Norwood, MA  02062
U.S.A.
 617-769-9450

Bob  Carter
Manufacturer's Representative
Photovac  International
 25-B Jefryn Blvd.  West
Deer Park, Mf   11729
 U.S.A.
 516-254-4199

 Ray E.  Carter.  Jr.
 Research Assistant
 University of Kansas
 4002 Learned Hall
 Lawrence,  KS  66045
 U.S.A.
 913-864-3731
                                             856

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February 12-14, 1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final Participants'  List
Bill Caushie
Technical Sales
SKC, Inc.
334 Valley View Road
Eighty Four, PA  15330
U.S.A.
412-941-9701

Anne Cavanaugh
Strategic Diagnostics, Inc.
128 Sandy Drive
Sandy Bear Industrial Park
Newark, DE  19713
U.S.A.
302-456-6789

Debra Chaloud
Senior Scientist
Lockheed-ESC
1090 East Flamingo Road
Las Vegas, NV  89119
U.S.A.
702-734-3227

Thompson Chambers
Technical Assistance Team
Ecology & Environment, Inc.
160 Spear Street
San Francisco,  CA  94105
415-777-2811

Cal Chapman
President
Environmental Fuel Systems, Inc.
P.O. Box 1899
Bandera, TX  78003
U.S.A.
512-796-7353

6. Hunt Chapman
Chemist
Ecology and Environment, Inc.
1700 North Moore Street
Suite 1105
Arlington, VA  22209
U.S.A.
703-522-6065

Rob Cherney
Field Engineer
Hewlett-Packard Company
1421 South Manhattan Avenue
Fullerton, CA  92631
U.S.A.
714-758-5520

Tom Chiang
Staff Scientist
Lockheed Corporation
1050 E. Flamingo Road
Las Vegas, NV  89119
U.S.A.
702-734-3341
Mei-Chen Chuang
Research Scientist
6E Corporate R&D
KWD-279
P.O.Box 8
Schenectady. NY  12301
U.S.A.
518-387-5654

Wayne Chudyk
Professor
Tufts University
Civil Engineering Department
Medford, MA  02155
U.S.A.
617-381-3211

Mary E. Cisper
Technician
Los Alamos National Laboratory
P.O. Box 1663
MS - J585
Los Alamos, NM  87545
U.S.A.
505-665-0836

Glen A. Clark
Laboratory Section Chief
REECO
PO Box 98521
Las Vegas, NV  89193
U.S.A.
702-295-6559

Scott Clifford
Environmental Engineer
U.S. EPA
Region I
60 Westview Street
Lexington, MA  02173
U.S.A.
617-860-4631

David Clift
Chemist III
ADHS, State Lab Services
Haz Mat
1520 H. Adams Street
Phoenix, AZ  85007
U.S.A.
602-542-6121

Athol L. Cline
President
American Analytical Labs
3441 E. Hilber Street
Tucson, AZ  85714
U.S.A.
602-889-5787
Thomas J.  Cody,  Jr.
Vice President
Arlano Technologies,
12 Clematis Avenue
Haltham, MA  02154
U.S.A.
617-891-0778
                                                                         Inc.
                                         857

-------
                            Second International Symposium
                              Field Screening Methods  for
                           Hazardous Wastes and Toxic  Chemicals
                                 February 1Z-U, 1991
                            Sahara Botel - Las Vegas,  Nevada
                                Final Participants'  List
Larry Coe
Vic* President
SC&A
1311 Dolley Madison Boulevard
McLean, VA  22101
U.S.A.
703-893-6592

Michael J. Coggiola
Group Leader
SRI International
333 Ravensmod Avenue
Menlo Park, CA  94025
U.S.A.
415-859-3045

Mary Cogliano
Environmental Scientist
ICF Technology
9300 Lee Highway
Fairfax. VA  22031-1207
U.S.A.
703-934-3298

Mort Cohen
Director, Business Development
Teledyne CME
P.O. Bo* 58133
Santa Clara, CA  95052
U.S.A.
408-982-3989

Roy J. Cohen
Chemist
NUS Corporation
999 West Valley Road
Wayne, FA  19087
U.S.A.
215-687-9510

Steven Cohen
Commercial Development Dir.
Union Carbide
Chemicals & Plastics Company
39 Old Ridgebury Road
Danbury, CT  06817
U.S.A.
203-794-3216

Dan Colemen
Director of Marketing
CMS Research Corporation
200 Chase Park South
Suite  100
Birmingham, AL  35244
U.S.A.
205-733-6900

Doug Combs
Senior Geologist
Science Applications Int'l Corp.
800 Oak Ridge Turnpike
Oak Ridge, TN  37831
U.S.A.
615-481-2361
Charles F. Costa
Dir. Nuclear Radiation Assess.
U.S. EPA
EMSL-LV
P.O. Box 93478
Las Vegas, NV  89193-3478
U.S.A.
Larry Cottran
MS/IR Systems Product Line Mgr
Hewlett & Packard
1601 California Avenue
Palo Alto, CA  94304
U.S.A.
415-857-6118

Stuart Cram
Hewlett-Packard
Route 41
P. 0. Box 1100
Avondale, PA  19311
U.S.A.
215-268-5453

Bruce Crane
Marketing Manager
EM Sciences
480 Democrats Road
Gibbstown. HJ  08027
U.S.A.
800-222-0342

Steve Creech
The Foxboro Company
P.O. Box 500
600 N. Bedford St.
East Brldgewater, MA  02333
U.S.A.
508-378-5666

John Creed
Research Chemist
U.S. EPA
26 West Martin Luther King Dr.
Cincinnati, OH  45268
U.S.A.
513-569-7307

Alan B. Crockett
Technical Staff Consultant
Idaho National Engineering Laboratory
EG&G Idaho, Inc.
P.O. Box  1625
Idaho Tails,, ID  83415-2213
U.S.A.
208-526-1574

Tom A. Cronk
Research Associate
Oak Ridge National Laboratory
Grand Junction
P. 0. Box 2567
Grand Junction, CO  81502
U.S.A.
303-248-6265
                                           858

-------
 Stcond International Symposium
   Fiild Screening Methods for
Hazardous Wastes and Toxic Chemicals
      February 12-14, 1991
 Sahara Hotel - Las Vegas, Nevada
     Final Participants'  List
Amy J. Croas-Sniecinski
QA Officer
University of Nevada-Las Vegas
Environmental Research Center
4505 S. Maryland Parkway
Las Vegas, NV  89123
U.S.A.
702-739-3382

Colin Gumming
Graseby Ionics Ltd.
Odhams Trading Estate
St. Albans Road
Watford, Rerta,   WD2 SJX
United Kingdom
0923 38483

Timothy J. Curry
Environmental Engineer
U.S. EPA
25 Funston Road
Kansaa City, KS  66115
U.S.A.
913-236-3881 Ext. 215

Allan Curtis
Senior Chemist
U.S. EPA
Region VIII
Box 2S366 DFC
Denver, CO  80225
U.S.A.
303-294-1154

Arthur P. D'Silva
Senior Chemist
Iowa State Univeraity
Ames Laboratory - U.S. DOE
9 Spedding Ball
Ames. IA  50011
U.S.A.
515-294-9317

Clifford Dahra
Associate Professor
University of New Mexico
Biology Department
Albuquerque, NM  87131
U.S.A.
505-277-2850

Don E. Dale, Jr.
Technician
New Mexico State University
1628 Stull
Las Cruces,  NM  88001
U.S.A.
505-522-1200

Bill Daniels
Regional Representative
PHS
1961 Stout Street
Denver, CO  80294
U.S.A.
303-844-6166
                         Dennis  M.  Davis
                         Research Chemist
                         U.S.  Army  CRDEC
                         SMCCR-RSL/D.M.  Davis
                         Aberdeen Proving Ground,
                         U.S.A.
                         301-671-2437

                         Geoffrey C.  Davis,  Jr.
                         Senior  Program Manager
                         Teledyne CME
                         P.O.Box 58133
                         Santa Clara, CA 95052
                         U.S.A.
                         408-982-1980

                         Richard De Filippi
                         President
                         Ariano  Technologies,  Inc.
                         12 Clemati* Avenue
                         Waltham, MA 02154
                         U.S.A.
                         617-891-0778
                         M.  Scott  DeSha
                         Physicist
                         U.S.  Army CRDEC
                         SMCCR-DDT
                         Aberdeen  Proving  Ground, MD
                         U.S.A.
                         301-671-5841

                         Greg  DeYong
                         Research  Chemist
                         HACK  Company
                         P.O.  Box  907
                         100 Dayton  Avenue
                         Ames,  IA  50010
                         U.S.A.
                         515-232-2533

                         Michael Cellared
                         Program Manager
                         OWSQA
                         U.S.  EPA, RD-680
                         401 M Street
                         Washington, DC  20460
                         U.S.A.
                         202-382-5794

                         Jack  Demirgian
                         Chemist
                         Argonne National  Laboratory
                         9700  S. Cass Avenue
                         Building  211
                         Argonne,  IL 60439
                         U.S.A.
                         708-972-6807

                         Gregg  D.  Dempsey
                         Branch Chief
                         U.S.  EPA
                         PO  Box 98517
                         Las Vegas,  KV   89193
                         U.S.A.
                         702-998-2476
MD  21010-5423
    21208
              859

-------
                            Second International  Symposium
                              Field Screening Methods  lor
                           Hazardous Wastes  and Toxic  Chemicals
                                 February 12-14,  1991
                            Sahara Hotel - Las  Vegas,  Nevada
                                Final Participants'  List
Gurpal Deol
Quality Assurance Director
IT Corporation
1355 Vander Way
San Jose, CA  95112
U.S.A.
408-283-2261

Sefcsan Dheandhano
Extrel Corporation
575 Epsilon Drive
Pittsburgh, PA  15238
U.S.A.
412-967-5717

Gordon L. Dippo
Program Manager, RAPP, HazWrap
Martin Marietta Energy Systems
P.O. Box 2003
Oak Ridge, IN  37831-7606
U.S.A.
615-435-3211

John 1. Ditillo
Chemist
U.S. Array
CRDEC
ATTN: SMCCR-DDT
Aberdeen Proving Ground, MD  21010-5423
U.S.A.
301-671-3021

James Doesburg
Associate
Colder Associates, Inc.
4104 146th Avenue HE
Redmond, HA  98052
U.S.A.
206-883-0777

David Dogruel
Student
Los Alamos National Laboratory
NMIMT
Box 3586 c/s
Socorro, MM  87801
U.S.A.
505-835-4639

Linda Dohrman
Marketing Manager
Milllpore Corporation
80 Ashby Road
Bedford, MA  01730
U.S.A.
617-275-9200

Robert Donohoe
Staff Member
Los Alamos National Laboratory
INC-4, MS C345
Los Alamos, KM  87545
U.S.A.
505-667-7603
Russell C. Drew
President
Viking Instruments Corporation
12007 Sunrise Valley Drive
Reston, VA  22091-3406
U.S.A.
703-758-9339

Jo Ann Duchene
Program Manager
Life Systems, Inc.
ICAIR
24755 Highpoint Road
Cleveland, OH  44122
U.S.A.
(216)464-3291

Michael Duffy
HHU Systems
160 Charlemont Street
Newton Highlands, MA  02161
U.S.A.
617-964-6690

Joseph P. Dugan, Jr.
Senior Scientist
Hestinghousa Idaho Nuclear Co.
P.O. Box 4000
[IRC/MS 2202]
Idaho Falls, ID  83403
U.S.A.
208-526-3975

Peter H. Duquette
Research Chemist
Bio-Metric Systems, Inc.
9924 W. 74th Street
Eden Prairie, MN  55344
U.S.A.
612-829-2714

Nicholas Durand
Environmental Chemist
Sandia National Laboratories
P.O. Box 5800
Albuquerque, NM  87185
U.S.A.
505-644-0676

Damien Durbin
President
Gulf States Analytical
5450 NU Central
Suite  110
Houston, TX  77092
U.S.A.
713-690-4444

Philip Durgin
Senior Research Scientist
Veeder - Roof, Inc.
125 Powder Forest Drive
Simsburg, CT  06070-2003
U.S.A.
203-651-2785
                                            860

-------
                            Second International  Symposium
                              Field Screening Methods  for
                           Hazardous Wastes  and Toxic  Chemicals
                                 February 12-1*,  1991
                            Sahara Hotel - Las Vegas,  Nevada
                                Final Participants'  List
Chuck Out el
Civil Environmental Engineer
Ecology and Environment,  Inc.
12251 Universal
Taylor, MI  48180
U.S.A.
313-946-0900

Robert Dvorin
Senior Chemist Engineering
Science Applications Int'l Corp.
1 Sears Drive
Paramus, NJ  07652
U.S.A.
201-599-0100
Gary A. Eiceman
Professor
Hew Mexico State University
Chemistry Department
Las Cruces, KM  88003
U.S.A.
505-646-2146

John A. Elton
Research Chemist
Eastman Kodak Company
Federal Systems Division
Rochester, NY  14650-2156
U.S.A.
716-477-5642
Dennis T. Ealey
Senior Staff Scientist
Cham-Nuclear Geotech. Inc.
2597 B3/4 Road
Grand Junction, CO  81503
U.S.A.
303-248-6173

David G. Easterly
Quality Assurance Manager
U.S. EPA
EMSL. LV
P.O. Box 93478
Las Vegas, NV  89193-347S
U.S.A.
702-798-2556

D. Eastwood
Senior Staff Scientist
Lockheed Engineering & Sciences Co.
(LESC)
1050 East Flamingo Rd.,Ste.242
Las Vegas, NV  89119
U.S.A.
702-734-3287

Susan Eberlein
Technical Group Leader
Jet Propulsion Laboratory
4800 Oak Grove Drive
168-522
Pasadena, CA  91109
U.S.A.
818-354-6467

Brian Eckenrode
Project Scientist
Viking Instruments
12007 Sunrise Valley Drive
Reston, VA  22091
U.S.A.
703-758-9339

Vicki Ecker
Principal Scientist
Lockheed Engineering & Science* Co.
1050 East Flamingo
Las Vegas, NV  89119
U.S.A.
702-734-3223
William H. Engelmann
Geochemiat
U.S. EPA
EMSL - LV
P.O. BOX 93478
Las Vegas, NV  89193-3478
U.S.A.
702-798-2664

Robert E. Enwall
Staff Scientist
Lockheed ESC
1050 East Flamingo Rd.
Suite 242
Las Vegas, NV  89119
U.S.A.
702-734-3243

Barbara Erickson
Laboratory Supervisor
Pima County Mastewater Mngt.
Ina Road
7101 North Casa Grande Highway
Tucson, AZ  85743-9577
U.S.A.
602-744-4236  ext. 226

Mitchell D. Erickson
Associate Dir. R&O Programs
Argonne National Laboratory
Research and Development
9700 S. Cass Avenue
Argonne, IL  60439
U.S.A.
708-972-7772

Ed Eschner
Scientist
Lockheed Engineering & Sciences
1050 East Flamingo Road
Las Vegas, NV  89119
U.S.A.
702-734-3258

Eugene Esplain
Biologist
Navajo Superfund Program
PO Box 2946
Window Rock, AZ  86S1S
U.S.A.
602-871-6859
                                          861

-------
 Second International Symposium
   Field Screening Methods for
Hazardous Wastes and Toxic Chemicals
      February 12-14. 1991
 Sahara Hotel - Las Vegas, Nevada
     Final Participants' List
Carolyn Esposito
Project Manager/Chemist
U.S. EPA
RREL, STDD, RCB  MS-104
2890 Woodbridge Ava.
Edison, NJ  08837-3679
U.S.A.
201-906-6895

John C. Evans
Staff Scientist
Battelle Northwest
K6-81
P.O.Box 999
Richland, WA  99352
U.S.A.
509-376-0934

Joseph D. Evans
Environmental Chemist
Science Applications Int'l Corp.
10240 Sorrento Valley Road
#204
San Diego, CA  92121
U.S.A.
619-587-9071

David Everitt
Environmental Scientist
McLaren/Hart Environmental Mngt.
919 Porter Street
Easton, PA  18042
U.S.A.
215-258-5792

Scott Faller
Chemist
U.S. EPA
NRD/NRA
P.O. Box 93478
Las Vegas, NV  89193-3478
U.S.A.
702-798-2333

Tseng-Ying Fan
ManTech Techology
10625 Fallstone Road
Houston, IX  77099
U.S.A.
713-983-2115

James Fargo
Senior Scientist
Science Applications Int'l Corp.
8500 Cinder Bed Road
P.O. Box 1148
Newington, VA  22122-9998
U.S.A.
703-550-0516

Werner Faubel
Staff Member
Nuclear Research Center, Karlsruhe
P.O.Box 3640, D-7500 Karlsruhe
Karlsruhe,
West Germany
0721-07247-82-3357
                         Martin Favero
                         Vice President
                         Tracer Research Corporation
                         3855 North Business Center Dr.
                         Tucson, AZ  85705
                         U.S.A.
                         602-688-9400

                         Maury Fee
                         Systems Manager
                         Aerojet
                         P.O.Box 296
                         1100 West Bollyvale
                         Azus>, Ca  91702
                         U.S.A.
                         818-812-2564

                         Thomas J. Farmer
                         Geophysiclst/Applications Engr
                         Geophyaical Survey Systems, Inc.
                         P.O. Box 97
                         13 Klein Drive
                         N. Salem, NH  03073-0097
                         U.S.A.
                         603-843-1109

                         Peter Ferron
                         Environmental Supervisor
                         Ohio EPA
                         1800 Watermark Drive
                         Columbus, OH  43266
                         U.S.A.
                         614-644-2094

                         Stephen R. Finch
                         Research and Development
                         Dexsil Corporation
                         1 Ramden Park Drive
                         Hamden, CT  06517
                         U.S.A.
                         203-288-3509

                         Lynn Ann Fischer
                         Chemist
                         Ecology and Environment, Inc.
                         1057 West Fireweed Lane
                         Suite 102
                         Anchorage, AK  99503
                         U.S.A.
                         907-257-5000

                         J. Berton Fisher
                         Staff Research Scientist
                         Amoco Production Company
                         4502 E. 41st Street
                         P.O. Box 3385
                         Tulsa, OK  74102
                         U.S.A.
                         918-660-4078

                         Philip C. Fisher
                         Sensor Engineering Manager
                         Gas Tech, Inc.
                         8445 Central Ave.
                         Newark, CA  94560-3431
                         U.S.A.
                         415-794-6200
                 662

-------
                            Second International Symposium
                              Field Screening  Methods  for
                           Hazardous Mastes and Toxic  Chemicals
                                 February 12-14, 1991
                            Sahara Hotel - Las Vegas,  Nevada
                                Final Participants'  List
Richard Fitzpatrick
Research Chemist
Occidental Chemical Corp
2801 Long Road
Grand Island, NY  14072
U.S.A.
716-773-8100

Gerald Flanagan
President
Emissions Detection & Control.  Inc.
90S Hillside Dr.
P.O. Box 533
Southampton, FA  16966
U.S.A.
215-364-4255

William Fleming
Chief Operating Officer
Gilian Instrument Corporation
35 Fairfleld Place
West Caldwell, NJ  07006
U.S.A.
201-808-3355

Thomas H. Flor
Supervisory Hydrologist
NEESA
Port Rueneme, CA  93043-5014
U.S.A.
805-982-3512

C.R. Flynn
President
Chemrad Tennessee Corporation
701 Scarboro Rd. Ste. 2030
Oak Ridge, IN  37830
U.S.A.
615-481-2511

Chuck Forbes
Regional Sales Manager, West
U.S. Analytical Instruments
1511 Industrial Road
San Carlos, CA  94070
U.S.A.
415-595-8200

Michael S. Ford
Chemist
Science & Technology Corporation
118 East Vine Street
Touele, UT  84074
U.S.A.
801-882-1*36

John W. Fordham
Senior Research Engineer
Desert Research Institute
7010 Dandini Boulevard
Reno, NV  89512
U.S.A.
702-673-7364
Ken Forsberg
Technical Director
Summit Interests
P.O. Box 1128
Lyons, CO  80540
U.S.A.
303-772-3073

Jeff Fox
Regional Sales Manager
MDA Scientific
P.O. Box 1353
Rancho Santa Fe, CA  92067
U.S.A.
619-756-9540

Thomas Francoeur
Project Chemist
ABB Environmental Services, Inc.
261 Commercial Street
Portland, ME  04112
U.S.A.
207-775-5409

Clyde Frank
Assoc. Dir. Off. Technol. Dev.
U.S. Department of Energy
1000 Independence Avenue SE
Washington, DC  20585
U.S.A.
202-586-6382

Jochen Franzen
General Manager
Bruker-Franzen Analytik GmbH
Fahrenheitstr, 4
D-2800 Brennen 33,
Germany D-2800
49-421-2205-150

Philip R. Fresquez
Environmental Soil Scientist
Los Alamos National Laboratory
K490
Los Alamos, NM  87545
U.S.A.
505-667-0815

Steve Fraudenbarger
Student
Science & Technology Corporation
118 East Vine Street
Salt Lake, UT  84072
U.S.A.
801-882-1436

Howard Fribush
Project Officer
U. S. EPA
AOB, OS-230
401 M Street, SH
Washington, DC  20460
U.S.A.
202-382-2239
                                         863

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February 12-14, 1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final Participants'  List
David Friedman
U.S. EPA
Off. Research & Development
Washington, DC  20460
U.S.
202-245-3535

Stephen Friedman
V.F.  Research & Development
EnSys, Inc.
4222 Emperor Blvd.
Royal Center
Morrisville, NC  27560
U.S.A.
919-941-5509

Christopher Fromme
Manager of Engineering
RedZone Robotics
401 Bingham Street
Pittsburgh, PA  15203
U.S.A.
412-481-3477

Kishor Fruitwala
Manager
Ecology and Environment, Inc.
1509 Main Street
Dallas, TX  75201
U.S.A.
214-742-6601

Marvin H. Frye
Supervisor, Environ. Science
U.S. EPA
Region VIII
Box 25366 DFC
Denver. CO  80225
U.S.A.
303-236-5073

Jon C. Gabry
Environmental Chemist
Ebasco Environmental
160 Chubb Avenue
Lyndhurat, NJ  07071
U.S.A.
201-460-6468

Susan Gagner
Chemist
Science Applications Int'l Corp.
4900 Hopyard Road
Suite 310
Pleasanton, CA  94588
U.S.A.
415-463-8111

Kishor Gala
Consulting Chemist
CH2M Hill
6060 S. Willow Drive
Greenwood Village, CO  80111
U.S.A.
303-771-0900
Richard B. Gammage
Group Leader
Oak Ridge National Laboratory
P.O. Box 2008
Bldg. 7509    MS 6383
Oak Ridge, TN  37831-6383
U.S.A.
615-574-6256

Lizbeth Garcia-Gonzalez
Graduate Student
New Mexico State University
Analytical Chemist
P.O.Box 331
Mesilla, MM  88046
U.S.A.
505-526-8725

Jack D. Generaux
Chief, Environmental Support
U.S. Army
Engineer District [CEMRKJ
601 E. 12th Street
Kansas City, MO  64106-2896
U.S.A.
913-426-7885

Joseph E. Gibb
Senior Environmental Scientist
Aluminum Company of America
ALCOA Technical Center
Rt. 780, Bldg. C
Alcoa Center, PA  15069
U.S.A.
412-337-2597

G. Gibson
Scientist
Lockheed Engineering & Sciences Co.
   (LESC)
1050 East Flamingo Road
Las Vegas, NV  89119
U.S.A.
702-734-3256

Gregory Gillispie
Professor
North Dakota State University
Chemistry Department
Ladd Hall
Fargo, NO  58105
U.S.A.
701-237-8244

Dick Glass
E-N-G Mobile Systems, Inc.
2950 Cloverdale Ave.
Concord, CA  94518
U.S.A.
415-798-4060

Para Claude
Gas Tech. Inc.
8445 Central Avenue
Newark, CA  94560
U.S.A.
415-745-8700
                                           864

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February 12-14, 1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final Participants' List
Gary Goken
Lab/Hfg. Operations Manager
3M
Bids. 2:8-35-0«
St. Paul, UN  55144
U.S.A.
612-733-6030

Larry Golden
V.F. Equipment Services
Clean Air Engineering
500 W. Hood St.
Palatine, IL  60067
U.S.A.
800-627-0033

Randy Golding
Tracer Research Corporation
3855 North Business Ctr Drive
Tucson, AZ  85705
U.S.A.
602-888-9400

Hugh Goldsmith
President
SRI Instruments
3870 Del Amo Blvd.
Suite 506
Torranee, CA  90503
U.S.A.
213-214-5092

John Graves
President
Environmental Chemistry Services
96 Inverness Drive East
Suite H
Englewood, CO  80112
U.S.A.
303-792-5920

David W. Gray
On-Scene Coordinator
U.S. EPA
Region VI
1445 Ross Avenue
Dallas, IX  75202-2733
U.S.A.
214-655-2275

Phillip Greenbaum
Chemist/Senior Analyst
US Army Foreign Science & Tech Center
220 7th Street NE
Charlottesville, VA  22901
U.S.A.
804-980-7857

Rudolf H. Greulich
Senior Engineer
I. Kruger Ltd.
Gladsaxevej 363
2860 Soborg,
Denmark
45-31690222
Robert Gross
Electronic Engineer
U.S. Army CRDEC
SMCCR-DDT
Aberdeen Proving Ground, MD  21010
U.S.A.
301-671-3021

John M. Gurchiek
Senior Technician
ENSR Consulting & Engineering
740 Pasquinelli
Westmont, IL  605S9
U.S.A.
708-887-1700

Donald Gurka
Chairperson
U.S. EPA
EMSL-LV
P.O. Box 93478
Las Vegas, NV  89193-3478
U.S.A.
702-798-2312

John H. Haas
Research Associate
Oak Ridge National Laboratory
P.O. Box 2008
Oak Ridge, TN  37831-6113
U.S.A.
615-574-5042

John T. Hadley
Environmental Geologist
ICF Kaiser Engineers
165 S. Union Blvd.
Suite 802
Lakewood, CO  80228
U.S.A.
303-236-7412

Andrew J. Hafferty
Field Investigation Team Mngr
Ecology & Environment, Inc.
101 Yesler Way
Seattle, HA  96104
U.S.A.
206-624-9537

Rolf Hahn
Landesansta.lt f.  Uraweitschutz
B.-W.
Griesbachstr. 3,
D-7500 Karlsruhe 221,
Germany
0721 8406-394

John Hall
Chief
U.S. Department of Energy
Technology Development Branch
P.O. Box 98518
Las Vegas,  NV  89193
U.S.A.
702-295-5855
                                          865

-------
                            Second International  Symposium
                              Field Screening Methods  for
                           Hazardous Hastes  and Toxic  Chemicals
                                 February 12-14,  1991
                            Sahara Hotel - Las Vegas,  Nevada
                                Final Participants'  List
John D. Hanby
President
Hanby Analytical Labs, Inc.
4400 S. Wayside #107
Houston, IX  77087
U.S.A.
713-649-4500

Steven Hanst
Sales Manager
Infrared Analysis, Inc.
1424 H. Central Park Avenue
Anaheim, CA  9280Z
U.S.A.
714-535-3057

Lisa Hanusiak
Environmental Chemist
ICF Kaiser Engineers
160 Spear Street
Suite  1380
San Francisco, CA  94105
U.S.A.
415-882-3063

Kim G. Hanzelka
Chemist
Martin Marietta Energy Systems, Inc.
Bldg.  9115, Y-12 Plant
P.O. Box 2009  MS - 8219
Oak Ridge, TN  37831
U.S.A.
615-574-1599

Charles S. Harden
Chairperson
U.S. Army CRDEC
Attn:  SMCCR-RSL
Aberdeen Proving Ground, MD  21010
U.S.A.
301-671-3129

Anthony Harding
Spectrace Instruments, Inc .
2401 Research Blvd. #206
Fort Collins, CO  80526
U.S.A.
303-493-2219

John Harju
Geologist
University  of North Dakota
Energy & Environ. Research Ctr
 IS North 23rd Street
Grand Forks,  ND   58202
U.S.A.
 701-777-5208

Robert 0. Harrison
Manager of  R&D
 Immunosystems,  Inc.
 4 Washington  Ave.
Scarborough,  ME   04074
U.S.A.
207-883-9900
Marty Harshbarger-Kelly
Editor
EnviroMatrix
505 Hater Street
Chardon, OH  44024
U.S.A.
216-286-5818

Linda Hashlamoun
Registration Coordinator
Life Systems, Inc.
ICAIR
24755 Highpoint Road
Cleveland, OH  44122
U.S.A.
(216)464-3291

Philip Hemberger
Los Alamos National Laboratory
Analytical Chemistry Group
MS  6740
Los Alamos, NM  87545
U.S.A.
505-665-4896

Susan Hennigan
Product Manager
The Foxboro Company
P.O. Box  500
600 N. Bedford St.
East Bridgewater. MA   02333
U.S.A.
508-378-5556

Art Hentschel
Territory Manager
U.S. Analytical Instruments
1511  Industrial Road
San Carlos,  CA  94070
U.S.A.
415-595-8200

Steven  Hern
Chief,  Exposure Monitoring
U.S.  EPA
EMSL-LV
P.O.  Box  93478
Las Vegas, NV  89193-3478
U.S.A.
702-798-2594

Jack  Herndon
Laboratory Manager
Hart  Crowser.  Inc.
 1910  Fairview Ave.  East
Seattle,  WA  98102-3699
U.S.A.
206-324-9530

Roy C.  Herndon
Director, Ctr.  BTR&WM
Florida State  University
 2035  East Paul  Dirac  Drive
Morgan  Building,  Suite 226
 Tallahassee,  FL   32310
 U.S.A.
 904-644-5524
                                            866

-------
                            Second International Symposium
                              Field Screening  Methods  for
                           Hazardous Hastes  and Toxic  Chemicals
                                 February 12-14,  1991
                            Sahara Hotel - Las Vegas,  Nevada
                                Final Participants'  List
Laura Hernon-Kenny
Researcher
Battelie
SOS King Avenue
Rm. 7238
Columbus, OR  43201
U.S.A.
614-424-5491

Rebecca Herrmann
Technician
Lawrence Livermore National Laboratory
7000 East Avenue
Livermore, CA  94550
U.S.A.
Alan D. Hewitt
Physical Scientist (Research]
U.S. Army CRREL
72 Lyme Road
Hanover, NH  03755-1290
U.S.A.
603-646-4512

Monica Heyl
U.S.Army CRDEC
SMCCR-RSL
Aberdeen Proving Ground, MD  21010
U.S.A.
301-671-3129

Ann C. Heywood
Assistant Vice President
Science Applications Int'l Corp.
370 L'Enfant Promenade SW
Suite 902
Washington. DC  20024-2518
U.S.A.
202-586-8922

Cornelius J. Higgins
Principal
Applied Research Associates, Inc.
4300 San Mateo Blvd.  NE
Albuquerque, NM  87110
U.S.A.
505-881-8074

Herbert Hill
Professor of Chemistry
Washington State University
Department of Chemistry
Pullman, WA  99164-4630
U.S.A.
509-335-5648

John L. Rill
Manager of Marketing
UTD,  Inc.
P.O. Box 8560
8560 Cinderbed Road,Suite 1300
Newington, VA  22122
U.S.A.
703-339-0600
Gary D. Hippie
Environmental Chemist
3M/Environmental Engineering
Bldg. 2-3E-07
935 Bush Avenue
St. Paul, UN  55106
U.S.A.
612-778-6439

Mari Beth Hilton
Chemistry Instructor
Tulsa Junior College
Science & Math Division
909 South Boston
Tulsa, OK  74135
U.S.A.
918-587-6561

John 1. Hogue
Geologist
Ecology and Environment, Inc.
6440 Hillcroft Suite 402
Houston, TX  77081
U.S.A.
713-771-9460

Humid Hojoji
Catholic University of America
400 Hannan Hall
Washington, DC  20064
U.S.A.
202-319-6705

Paul  M. Holland
Senior Scientist
General Research Corporation
5383 Hollister Avenue
Santa Barbara, CA  93111
U.S.A.
805-964-7724

Bruce Hornaday
Environmental Protection Spec.
David Taylor Research Center
Code 0231
Bethesda, MD  20084-5000
U.S.A.
301-227-1510

John Hosenfeld
Section Head
Midwest Research Institute
425 Volker Boulevard
Kansas City. MO  64110
U.S.A.
816-753-7600  Ext. 336

Masaaki Hosomi
Senior Research Scientist
National Institute for
  Environmental Studies
16-2 Onogawa
Tsukuba,
Japan 305
81-298-516111
                                          867

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February 12-U, 1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final Participants'  List
James Houle
Project Manager
Eastman Kodak Company
3/82/RL
Rochester, Sf  14650-2156
U.S.A.
716-477-3710

Robert J. Howard
Product Marketing Executive
Graseby Ionics, Ltd.
9515 Deereco Road
Suite 500
Timonium, MD  21093
U.S.A.
301-560-0385

William Hughes
Sales Manager
GMD Systems, Inc.
Old Route 519
Hendersonville, PA  15339
U.S.A.
412-746-3600

Jim Hunter
Thermo Environmental Instruments
6 West Forge Parkway
Franklin, MA  02038
U.S.A.
508-520-0430

Emile Hyman
Principal Investigator - R&D
Public Svc. Electric & Gas Co.
P.O. Box 570-16G
Newark, NJ  07101
U.S.A.
201-430-6654

Peter Isaacson
Project Manager
Viar and Company
300 North tee Street
Alexandria, VA  22314
U.S.A.
703-684-5678

Stephen C. James
Chief
Risk Reduction Engineering Laboratory
Site Demonstration & Eval.Br.
26 West Martin Luther King Dr.
Cincinnati, OH  4S268
U.S.A.
513-569-7301

N. Lynn Jarvis
Vice President
Microsensor Systems, Inc.
6800 Versar Center, Suite 118
Springfield, VA  22151
U.S.A.
703-642-6919
Roger A. Jenkins
Group Leader
Oak Ridge National Laboratory
Bethel Valley Road
Bldg. 4500-S, P.O. Box 2008
Oak Ridge, TN  37831-6120
U.S.A.
615-576-8594

Thomas F.  Jenkins
Research Chemist
U.S. Army CRREL
Geochemical Sciences Branch
72 Lyme Road
Hanover, NH  03755
U.S.A.
603-646-4385

Janine Jessup-Arvizu
Chairperson
Idaho National Engineering Lab
EG&G Idaho, Inc.
P.O. Box 1625 CFA 633
Idaho Falls, ID  83415
U.S.A.
208-526-0470

Jeff Jeter
Senior Scientist
Lockheed Engineering and Sciences
   Company (LESC)
1050 E. Flamingo Rd.
Las Vegas, NV  89119
U.S.A.
702-734-3286

Dave Johnson
Business Development
Standard Manufacturing Company
4012 W. Illinois
Dallas, TX  75211
U.S.A.
214-337-8911

James L. Johnson
Chemist/Team Leader
U.S. EPA
ESB/P-3-1-1
944 E. Harmon Avenue
Las Vegas, NV  89119
U.S.A.
702-798-2118

P. Elizabeth Jones
Senior Environmental Engineer
Midwest Research Institute
5109 Leesburg Pike
Suite 414
Falls Church, VA  22041
U.S.A.
703-671-0400
                                           868

-------
                            Second  International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February 12-14, 1991
                            Sahara  Hotel - Las Vegas, Nevada
                                Final Participants' List
Robert K. Jones
Physical Scientist
National Oceanic & Atmospheric Admin.
Hazardous Materials Division
7600  Sand Point Way, Northeast
Seattle, HA  98115
U.S.A.
206-526-6317

Roy R. Jones
Environmental Scientist
U.S.  EPA
1200  Sixth Avenue
Seattle, WA  98101
U.S.A.
206-553-7373

Thomas Jordan
SKC - West
P.O.  Box 4133
Fullerton, CA  92634-4133
U.S.A.
714-992-2780

Howard L. Joseph
Representative
Jontec Systems, Inc.
P.O.  Box 2163
Laguna Hills, CA  92634
U.S.A.
714-581-5972

Bill  Jow
Mobil* Lab Supervisor
GTEL Environmental Labs, Inc.
20000 Mariner Ave.
Suite 300
Torrance, CA  90503
U.S.A.
213-371-1044

Thomas Junk
Research Associate
Institute for Environmental Studies
Louisiana State University
Room 42   Atkinson Hall
Baton Rouge, LA  70803
U.S.A.
504-388-4289

Larry Kaelin
Chemist
Roy F. Weston
REAC
2890 Hoodbridge Avenue
Edison,  NJ  08837-3679
U.S.A.
201-906-3492

Robert C. Kaercher
Mngr., Market Research & Anal.
Aerojet
Electronic Systems Division
1100 Hollyvale,  M/S Bldg.1-3
Azusa, CA  91702
U.S.A.
818-812-1034
Robert Kamensky
NRT Corporation
MS 14-100
3SSO General Atomics Ct.
San Diego, CA  92121
U.S.A.
619-455-3281

Mel Kaminsky
Analytical Services Mngr./FASP
Ecology & Environment,  Inc.
Ill W. Jackson Blvd.
Chicago, IL  60604
U.S.A.
312-663-9415

Stephen Kane
Senior Applications Chemist
Photovac International, Inc.
25-B Jefryn Boulevard West
Deer Park, NY  11729
U.S.A.
516-254-4199

Weng F. Kang
Assistant Professor
Vanderbilt University
Station B
Box 1661
Nashville, IN  37235
U.S.A.
615-322-0952

Steven Karr
Mngr. Sensor Signal Processing
General Electric Company
P.O.  Box 8
Shenectady, NY  12301
U.S.A.
518-387-5450

Jay Kasner
Research Scientist
Department of Defense
Washington, DC  20505
U.S.A.
703-281-8064

Roy M. Kay
Publisher
EnviroHatrix
505 Water Street
Chardon,  OH  44024
U.S.A.
216-286-5818

Garrett Keating
Lawrence Livermore National Labs
P.O.  Box  5507
L-453
Livermore,  CA  94550
U.S.A.
415-422-0921
                                          869

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February 12-14, 1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final Participants'  List
Rebecca Keen
Technical Assistant
Science Applications lnt'1 Corp.
12800 Middlebrook Road
Germantown, MD  20874
U.S.A.
301-353-7949

Carl Keller
Division Leader
Science & Engineering Associates, Inc.
612 Old Santa Fe Trail
Santa Fe, NM  87501
U.S.A.
505-983-6698

Lori J. Keller
Scientist
Bechtel National, Inc.
800 Oak Ridge Turnpike
Oak Ridge, TN  37830
U.S.A.
615-482-0596

Marty R. Keller
Scientist
Bechtel National, Inc.
P.O. Box 350
Oak Ridge, TN  37831-0350
U.S.A.
615-576-5915

Paul Kelley
Consultant
Teledyne CME
3034 Sunny Meadow Lane
San Jose, CA  95135
U.S.A.
408-223-8939

Scott Kellogg
University of Idaho
Dept. of Bacteriology and
 Biochemistry
Moscow, ID  83843
U.S.A.
208-885-6966

Peter Kesners
TecimLsche Universitat Hamburg
Harburg
Harburger Schlobstr. 20
Hamburg 90,   2100
Germany
040-7718-2378

Koumudi A. Ketkar
Senior Environmental Scientist
CH2M Hill
625 Herndon Parkway
Herndon, VA  22070
U.S.A.
703-471-6405  Ext. 4238
Suhas N. Ketkar
Staff Physicist
Extrel Corporation
575 Epsilon Dr.
Pittsburgh, PA  15238
U.S.A.
412-963-7530

Eric Kett
The Foxboro Company
P.O. Box 500
600 N. Bedford St.
East Bridgewater, Ma  02333
U.S.A.
508-378-5556

Bhupi Khona
Remedial Project Manager
U.S. EPA
Mail Code 3HW23
641 Chestnut Street
Philadelphia, PA  19107
U.S.A.
215-597-0439

Yasmine Khonsary
Environmental Scientist
U.S. EPA
ORP
4220 S Maryland Pkwy. Bid. C
Las Vegas, NV  89119
U.S.A.
702-798-3159

Man-Goo Kim
University of Utah
MARC
214 EMRL
Salt Lake City, UT  84112
U.S.A.
801-581-8431

S.M. Klainer
Scientist
Lockheed engineering and Sciences
   Company (LESC)
1050 E.Flamingo Rd.
Las Vegas. NV  89119
U.S.A.
702-734-3308

Mark Klusty
Product Manager
Microsensor Systems, Inc.
725 South St Asaph Street
Apt Bill
Alexandria, VA  22314
U.S.A.
703-683-4267
                                           870

-------
                            Second International Symposium
                              Field Screening Method!  for
                           Hazardous Hastes and Toxic  Chemicals
                                 February 12-1*, 1991
                            Sahara Hotel - Las Vegas,  Nevada
                                Final Participants'  List
Steven C. Knight
Manager
Terra Tek
Environmental Products Divn.
400 Wakara Hay
Salt Lake City, UT  84108
U.S.A.
801-584-2456

Stephen Xnollmeyer
Mobile Laboratory Supervisor
Tetra K Testing
53 Southampton Road
Westfield, MA  01085
U.S.A.
413-562-9193
Eric Koglin
Matrix Manager,
U.S. EPA
EMSL - LV
P.O. Box 93478
Las Vegas, NV
U.S.A.
702-798-2432
 AFMMP
89193-3478
Larry Kopjak
Microsensor Technology, Inc.
41762 Christy Street
Fremont, CA  94538
U.S.A.
415-490-0900

Thomas Korb
Marketing Manager
National Draeger, Inc.
101 Technology Drive
Pittsburgh. PA  15275
U.S.A.
412-788-5608

John Koutsandreas
U.S. EPA
RD 680
401 M Street SW
Washington, DC  20460
U.S.A.
202-382-5789

Walter H. Kovalick
Dir., Technology Innovation
U.S. EPA
Emergency & Remedial Response
401 M Street SH
Washington. DC  20460
U.S.A.
Ronald J. Kovein
Electronics Technician
NIOSH
DHHS, CDC
4676 Columbia Parkway, MS R-8
Cincinnati, OB  45226
U.S.A.
513-841-4284
                                     Paul  Kowalski
                                     Brucker  Instruments
                                     19  Fortune  Drive
                                     Manning  Park
                                     Billerica,  MA   01821
                                     U.S.A.
                                     508-667-9580
                                     Mark  L.  Kram
                                     Hydrologist
                                     NEESA
                                     Code  112E2
                                     Port  Hueneme,
                                     U.S.A.
                                     805-982-3512
              CA  93043-5014
R. T. Kroutil
Research Chemist
U.S. Army CRDEC
SMCCR-RSL/R.T. Kroutil
Aberdeen Proving Ground, MD  21010-5423
U.S.A.
301-671-3021

Lisa Krowitz
Chemist
NUS Corporation
19 Crosby Drive
Bedford, MA  01730
U.S.A.
617-275-2970

Paul Kueser
Director, Programs & Planning
Air & Waste Management Assn.
3 Gateway Center
4 West
Pittsburgh, PA  1S222
U.S.A.
412-232-3444

Cynthia A. Ladouceur
Commander
U.S. Army CRDEC
Attn:  SMCCR-RSB/
 Dr. Cynthia A. Ladouceur
Aberdeen Proving Ground, MD  21010-5423
U.S.A.
301-671-4284

Johannes Lagois
Dragerwerk AG
Moislinger Allee 53/55
D 2400 Lubeck,
Germany
0451-882-2616

Emily Landis
Geochemistry
Ecology and Environment, Inc.
6777 Engle
Cleveland, OH  44130
U.S.A.
216-243-3330
                                         871

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February 12-14, 1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final Participants' List
Ken Lang
Chief, Chemistry Division
U.S. ATHMA
Technical Support Division
Attn:  CETHA-T5-C
Aberdeen Proving Ground, MD  21020-5401
U.S.A.
301-676-7569

Frank Laukien
Vice President
Broker Instruments, Inc.
Manning Park
Billerica, MA  01821
U.S.A.
508-667-9580

Donald S. Lavery
General Analysis Corporation
140 Water Street
Norwalk, CT  06854
U.S.A.
203-852-8999

John R. Leetnan
President
Leeman Labs, Inc.
55 Technology Drive
Lowell, MA  01851
U.S.A.
S08-454-4442

Chris Leibman
Section Leader
Los Alamos National Laboratory
Organic Analysis
MS K484
Los Alamos, NM  87545
U.S.A.
505-667-5889

David Leland
Spectrace Instruments
2401 Research Blvd. #206
Ft. Collins, CO  80526
U.S.A.
303-493-2219

Douglas K. Lemon
Manager, Measurement Sciences
Battelle Pacific Northwest Labs
902 Battelle Blvd.
P.O. Box 999
Richland, NA  99352
U.S.A.
509-375-2306

Joseph Leonelli
Associate Director
SRI International
333 Ravenswood Avenue
Menlo Park, CA  94025
U.S.A.
415-859-23Z6
Steven Levine
Professor
University of Michigan
School of Public Health
Ann Arbor. MI  48109-2029
U.S.A.
313-936-0759

Timothy £. Lewis
Principal Scientist
Lockheed
1050 East Flamingo Road
Las Vegas, KV  89119
U.S.A.
702-734-3AOO

R. Lidberg
Lockheed Engineering & Sciences Co.
(LESC)
1050 E.Flamingo Road
Las Vegas. NV  89119
U.S.A.
702-734-3256

Stephen H. Lieberman
Naval Oceans Systems Center
Code 522
San Diego, CA  92152
U.S.A.
619-553-2778

Shirley Liebman
Geo-Centers, Inc.
c/o U. S. Army CRDEC
SMCCR-RSL/S. Liebman
Aberdeen Proving Ground, MD  21010-S423
U.S.A.
301-671-3764

Thomas Limero
Toxicology Lab Supervisor
KRUG Life Science!
1290 Hercules Drive #120
Houston, IX  77089
U.S.A.
713-483-7251

Greg Linder
Toxicologist
MAHTECH
200 Southwest 35th Street
Corvallis, OR  97330
U.S.A.
503-754-8335

Eric R. Lindgren
Sr. Member of Technical Staff
Sandia National Laboratories
Waste Mangwnent Sys. Div. 6416
Albuquerque, NM  87185
U.S.A.
505-844-3820
                                           872

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February 12-14, 1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final Participants'  List
Kazumasa Lindley
Imnunoaseay Chemist
University of Nevada-Las Vegas
4505 Maryland Parkway
Las Vegas, NV  89154
U.S.A.
702-790-2379

Diane R. Lindsay
Civil Engineer/Graduate Stud.
New Mexico State University
Box 3010 UPB
Las Cruces, KM  B8003
U.S.A.
505-646-1499

A. Linenberg
President
Sentex Sensing Technology, Inc.
553 Broad Ave.
Ridgefield, NJ  07657
U.S.A.
201-945-3694

Ben Litteral
Sales
Outokumpu Electronics
10614 Countryside Drive
Rancho Cucamonga, CA  91730
U.S.A.
800-229-9209

Scott Little
Manager, Analytical Frod. Dev.
TN Technologies, Inc.
2555 North IH-35
Round Rock, TX  78664
U.S.A.
512-388-9235

Viorica Lopez-Avila
R & D Manager
Midwest Pacific Environmental Lab
625B Clyde Avenue
Mountain View, CA  94043
U.S.A.
A. Lord
Graseby Ionics, Ltd.
Analytical Division
Park Avenue, Bushey
Watford, Herts.   WD2 2BW
United Kingdom
0923-38483

Roger Loughrey
Project Chemist
NUS Corporation
Lab Services Group,
6605 Landview Rd.
Pittsburgh, PA  15217
U.S.A.
412-747-2527
William Lowry
Program Manager
Science & Engineering Associates, Inc.
612 Old Santa Fe Trail
Santa Fe, NM  87501
U.S.A.
505-983-6698

Rebecca Kun-Er Lu
Research Manager
Corning Incorporated
Sullivan Park
Corning, NY  14831
U.S.A.
607-974-3133

Daniel Lucero
Project Engineer
III Research Institute
4600 Forbes Blvd.
Lanham, MD  20607
U.S.A.
301-459-3711

Steven L. Ludmerer
Business Director
Union Carbide Chemicals & Plastics Co.
39 Old Ridgebury Road
Danbury, CT  06817
U.S.A.
203-794-6131

Nile A. Luedtke
Program Manager
Martin Marietta Energy Systems
P.O. Box 2003
Oak Ridge, TN  37831-7261
U.S.A.
615-574-8752

Robert M. Lugar
Senior Scientist
Idaho National Engineering Lab
EG&G Idaho, Inc.
P.O. Box 1625,Mail Stop 1406
Idaho Falls, ID  83415
U.S.A.
208-525-5649

Timothy 0. Lynn
Dexsil Corporation
One Hamden Park Drive
Hamden. CT  06517
U.S.A.
203-288-3504

Larry Lynott
President
SCITEC Corporation
1029 N. Kellogg
Kennewick, WA  99336
U.S.A.
509-783-9850
                                          873

-------
                            Second International Symposium
                              Fiald Screening Method! for
                           Hazardous Hastes and Toxic Chemicals
                                 February 12-14. 1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final Participants'  List
Harvey Lyon
Safety/Environmental Manager
David Taylor Research Center
Code 0231
Bethesda, MD  20084-5000
U.S.A.
301-227-1510

Raymond Hack ay
Detection Directorate
U.S. Army CRDEC
Detection Technology Div.
Aberdeen Proving Ground, MD  21010-3423
U.S.A.
301-671-5532

David Madsen
Reynolds Electrical & Engineering Co.
P.O. 98521
Las Vegas, NV  89193-8S21
U.S.A.
702-295-1913

Jack Mahon
Sales Manager
Dexsil Corporation
1 Hamden Park Drive
Hamden, CT  06517
U.S.A.
203-288-3509

Aaron M. Mainga
Graduate Student
Institute for Environmental Studies
Louisiana State University
Baton Rouge, LA  70803
U.S.A.
504-388-4283

Dennis A. Malick
Environmental Inspector
USMC
HQ Bn. Co. "A"  MCAGCC
29 Palms, CA  92278
U.S.A.
719-368-5200

Deborah Malone
Environ Health Specialist II
Az. Dept. Environmental Quality
2655 E. Magnolia St. Suite 2
Phoenix, AZ  85034
U.S.A.
602-392-4030

Laura L. Manley
Hydrologist
AZ. Dept. of Environmental Quality
2005 N Central
Phoenix, AZ  85004
U.S.A.
602-257-2123
Chung-Rei Mao
Chemist
USACE
Missouri River Division
12565 West Center Road
Omaha, NE  68144
U.S.A.
402-221-7494

Mark Marcus
Sr. Dir. Analytical Programs
Chemical Waste Management, Inc.
150 W. 137th Street
Riverdale, IL  60627
U.S.A.
708-841-8360

Glen A. Marotz
Professor
University of Kansas
Dept. of Civil Engineering
2006 Learned Hall
Lawrence, KS  66045
U.S.A.
9M-864-3731

Paul Marsden
Project Manager
Science Applications Int'l Corp.
4224 Campus Point Court
MS *210
San Diego, CA  92121
U.S.A.
619-535-7302
Edwin Marti
Scientist
Triangle Laboratories,
801-10 Capitola Dr.
Durham, NC  27713
U.S.A.
919-544-5729

John Martin
Product Manager
Rosemount Analytical
Dohrmann Division
3240 Scott Blvd.
Santa Clara. CA  95054
U.S.A.
408- 727-6000

Michael L. Mastracci
Senior Engineer
US EPA
401 M Street SW
MAshington, DC  20460
U.S.A.
202-475-8933
Inc.
                                            874

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February 12-14, 1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final Participants' List
John M. Mateo
QA Officer
Roy F. Hasten
REAC
2890 Hoodbridge Avenue
Edison, NJ  06837
U.S.A.
201-632-9200

Rosemary Mattuck
Environmental Chemist
ENSR Consulting & Engineering
33 Nagog Park
Acton, MA  01720
U.S.A.
508-635-9500

Gerhard Matz
Tachnische Universitat Hamburg
Harburg
Harburger Schlobstr. 20
Hamburg 90,   2100
Germany
040-7718-3113

Carl Mazzuca
President
Gas Tech, Inc.
8445 Central Avenue
Newark, CA  94560-3431
U.S.A.
415-794-6200

Richard G. McCain
Principal Engineer
Westinghouse Hanford Company
P.O. Box 1970
H4-55
Richland. Wa  99336
U.S.A.
509-378-0777

Charles McCammon
Regional Consultant
NIOSH
1961 Stout Street
Denver, CO  80294
U.S.A.
303-844-6166

William H. McClennen
Analytical Chemist
University of Utah
Ctr. for Microanalysis
214 EMRL
Salt Lake City,  UT  84118
U.S.A.
801-581-8*31

William A. McClenny
Supervisory Research Physicist
U.S. EPA
MD 44
Research Triangle Park, NC  27711
U.S.A.
919-541-3158
Linda McConnell
Senior Environmental Engineer
Midwest Research Institute
5109 Leesburg Pike
Suite 414
Falls Church, VA  22041
U.S.A.
703-671-0400

J.J. McCown
Advisory Scientist
Weatinghouse Hanford Company
T6-07
P.O. Box 1970
Richland, WA  99352
U.S.A.
509-373-3762

Chuck McDonald
Western Region
Microbics Corporation
6810 West 13th Avenue
Kennewick, WA  99337
U.S.A.
509-735-6402

James McElroy
Research Env Scientist
U.S.EPA
tO Box 93478
Las Vegas, NV  89193-3478
U.S.A.
702-798-2361

Elizabeth "Betsy" McGrath
Environmental Coordinator
University of Washington
CPAC
Seattle. WA  98115
U.S.A.
206-543-3430

Michael H. Mclaughlin
Mgr, Laser Technology Program
GE Corporation Research & Development
P.O. Box 8
RWD 272
Schenectady, NY  12301
U.S.A.
518-387-6113

Eugene Meier
Oir. Adv. Monitoring Sys. Dlv.
U.S. EPA
EMSL-LV
P.O. Box 93478
Las Vegas, NV  89193
U.S.A.
                                          875

-------
                            Second  International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February  12-14, 1991
                            Sahara  Hotel - Las Vegas, Nevada
                                Final Participants' List
 Richard G. Melcher
 Research Associate
 Dow Chemical Company
 Analytical Sciences Department
 1602 Building
 Midland,  MI  48667
 U.S.A.
 517-636-2126

 Kevin Meldrum
 Manager,  Business Development
 Teknekron Company
 1080 Marsh Road
 Menlo Park, CA   94025
 U.S.A.
 415-322-6200

 Raymond G. Merrill
 Program Manager
 Radian  Corporation
 Nelson  Highway
 3200 East Chapel Hill Road
 Research Triangle Park, NC  27709
 U.S.A.
 919-481-0212

 Henk L.  C. Meuzelaar
 University of Utah
 Ctr.  for Microanalysis &
 Reaction Chemistry, 214 EMRL
 Salt Lake City,  UT  84112
 U.S.A.
 801-581-8431

 Theodore J. Meyer
 Senior  Scientist
 EG&G Idaho, Inc.
 P.O.  1625
 Idaho Falls, ID  83415-4153
 U.S.A.
 208-526-4397

 Fred P.  Milanovich
 Physicist
 Lawrence  Livermore Nat'l Labs
 P.O.  Box 808
 Livermore, CA  94550
 U.S.A.
 415-422-6838

 Dennis Miller
 Lockheed  Engineering & Sciences Co.
 1050  Flamingo Road
 Las Vegas, NV  89119
 U.S.A.
 702-361-1626  Ext. 257

Gary A.  Miller
Senior Hydrogeologist
PRC Environmental Management,  Inc.
 1099  18th Street, Suite 1960
Denver,  CO  80202
U.S.A.
303-295-1101
 Herbert C. Miller
 V.P./Research Director
 Southern Research  Institute
 2000 Ninth Avenue  South
 Birmingham, AL   35205
 U.S.A.
 205-581-2513

 Jack Miller
 Sales & Marketing  Manager
 The  Foxboro Company
 600  North Bedford  Street
 P.O. Box 500
 East Bridgewater,  MA  02333
 U.S.A.
 508-378-5556

 Michael Miller
 Research Chemist
 ITT  Research Institute
 10 W. 35th Street
 Chicago, IL  60616
 U.S.A.
 312-567-4234

 Timothy R. Mlnnich
 Manager, Air Services Division
 Blasland, Bouck  &  Lee
 Raritan Plaza III
 Fieldcrest Avenue
 Edison, NJ  08837
 U.S.A.
 201-225-8484

 Pete Mitchell
 Chemist
 ICF Technology
 P.O. Box 280041
 Lakewood, CO  80228
 U.S.A.
 303-236-7412

 Tuijauna Mitchell-Hall
 Senior Scientist
 Lockheed ESC
 6585 S. Paradise Road
 Las Vegas,  NV  89119
 U.S.A.
 702-361-0220. ext.212

 John E. Moerlins
 Assoc. Director, Ctr. BTR&WM
 Florida State University
 2035 East Paul Dirac Drive
 Morgan Building, Suite 226
 Tallahassee,  FL  32310
 U.S.A.
 904-644-5524

 Paul Monsour
Graduate Research Assistant
Clarkson University
Rowley Lab
Rowley/CEE
Potsdam,  NY  13699
U.S.A.
 315-268-3776
                                           876

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February 12-14. 1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final Participants' List
Victor Montelaro
Enforcement Program Manager
Louisiana Dept. Environ. Quality
Office of Legal Affairs Enf.
333 Laurel Street
Baton Rouge. LA  70804
U.S.A.
504-342-9171

Mark Montgomery
Environmental Engineer
NEESA
Port Huenema, CA  93043-5014
U.S.A.
805-982-3512

T.E. Moody
Senior Scientist
Mestinghouse Hanford Company
450 Bills Street
RichUnd, MA  99352
U.S.A.
509-376-0396

Douglas Moore
Research Associate
University of New Mexico
Biology Department
Albuquerque, NM  87131
U.S.A.
505-277-2715

Gerald Moore
President
GMD Systems, Inc.
Old Route 519
Hendersonville, PA  15339
U.S.A.
412-746-3600

Jim Moore
Director of Marketing
Spectrace Instruments, Inc.
345 E. MiddleCield Rd.
Mountain View,  CA  94043
U.S.A.
415-967-0350

Thomas W. Moran
Manager
Galson Technical Services, Inc.
Environmental Investigation
6601 Kirkville
Syracuse. NY  13057
U.S.A.
315-432-0506

Todd R. Morgan
Environmental Scientist
McLaren/Hart Environmental Mngt.
26 Independence Blvd.
Warren, NJ  07060
U.S.A.
201-647-8111
Frank A. Morris
Managing Scientist
BC Analytical
1255 Powell Street
Emeryville, CA  94608
U.S.A.
415-428-2300

John V. Morris
Metals Team Leader
U.S. EPA
Region V, CM.
536 South Clark Street
Chicago, IL  60605
U.S.A.
312-353-3594

Linda Morrison
Chemist
C.C. Johnson & Malhotra P.C.
215 Union Blvd.
Suite 215
Lakewood, CO  80228
U.S.A.
303-987-2928

John Morrow
Chemist
NUS Corporation
1927 Lakeside Parkway
Suite 614
Tucker, GA  30084
U.S.A.
404-938-7710

Robert Mustacich
General research Corporation
5383 Hollister Avenue
Santa Barbara, CA  93111
U.S.A.
805-964-7724

Gerald Muth
Chemist
U.S. EPA
Region X
7411 Beach Drive E
Port Orchard, WA  98366
U.S.A.
206-871-6565

Steve Nacht
Reynolds Electrical & Engineering Co.
PC. Box 98521
Las Vegas, NV  89193-8521
U.S.A.
702-295-1913

Charles H. Nauman
Matrix Manager for Biomarkers
U.S.EPA
EMSL-LV, EAD
P.O. Box 93478
Las Vegas, NV  89193-3478
U.S.A.
702-798-2258
                                          877

-------
                             Second International Symposium
                               Field Screening Methods for
                            Hazardous Wastes and Toxic Chemicals
                                  February 12-1*, 1991
                             Sahara Hotel - Las Vegas, Nevada
                                 Final Participants' List
 Gary Nebel
 Research Engineer
 John Crane, Inc.
 International Sealing Systems
 6400 West Oakton Street
 Morton Grove, IL  60053
 U.S.A.
 708-967-2414

 Ed Heel
 Thermo Environmental Instruments
 8 West Forge Parkway
 Franklin, MA  02038
 U.S.A.
 508-520-0430

 David C.  Nelson
 Vice President
 Perkin-Elmer Nelson Systems
 10040 Bubb Road
 Cupertino,  CA  95014
 U.S.A.
 408-725-1107

 Michael T.  Nemergut
 Thermo Environmental Instruments
 8  West Forge Parkway
 Franklin,  MA  02038
 U.S.A.
 508-520-0430

 Bruce H.  Neuman
 Senior Advisor
 Ecology and Environment,  Inc.
 7  Scenery Drive
 Greensburg,  PA  15601
 U.S.A.
 412-837-5488

 Bruce Nielson
 Environ. Research Engineer
 U.S.  Air Force
 HQ Engineering £ Services Ctr.
 Building  1117
 Tyndall Air  Force Base, FL   32403-6001
 U.S.A.
 904-283-6011

 Bruce  Nielson
 Program Manager
 Air Force Engineering & Services Cntr.
 Engineering  and  Services Lab.
 HQ AFESC/RDV
 Tyndall Air  Force Base, FL  32403-6001
 U.S.A.
 904-283-6011

 Janis Nilsson-Runyon
 Partech Environmental
 192 Technology
Suite S
Irvine, CA  92718
U.S.A.
 John Nocerino
 Chemist
 U.S. EPA
 EMSL - LV
 P.O. Box 93478
 Las Vegas, NV
 U.S.A.
 702-798-2110
89193-3478
 Stig Nybo
 Turner Designs
 920 W. Maude Ave.
 Sunnyvale,  CA  94086
 U.S.A.
 408-749-0994

 Maureen O'Mara
 On-Scene Coordinator
 U.S.  EPA
 5HS-12
 230 S. Dearborn
 Chicago,  IL  £0604
 U.S.A.
 312-886-1960

 Robert O'Neil
 Senior Consultant
 Arthur D. Little
 25  Acorn Park
 Cambridge,  MA  02140
 U.S.A.
 617-864-5770

 Christopher M.  Ohland
 Environmental Chemist
 CH2M  Hill
 2203  Lefeber Avenue
 Wauwatosa,  WI  53213
 U.S.A.
 414-774-2081

 James  Osborn
 Project Supervisor
 Field  Robotics  Center
 The Robotics Institute
 Carnegie Mellon University
 Pittsburgh,  PA  15213
 U.S.A.
 412-268-6553

 Joe Osborne
 Litigation/Haz Waste Unit
 Ontario Ministry of the Environment
 P.O.Box 213
 Rexdale, Ontario,
Canada M9W 5LC
 416-235-5759

Edward B. Overton
Director
 Institute for Environmental Studies
Louisiana State University
Baton Rouge, LA  70803
U.S.A.
504-388-8521
                                           878

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February 12-14, 1991
                            Sahara Hotel * Las Vegas, Nevada
                                Final Participants'  List
Ashley Page
Marketing Specialist
CMS Research. Corporation
200 Chase Park S.
Suite 100
Birmingham, AL  35244
U.S.A.
205-733-6900

Winston Pao
Mngr, Corporate Business Dev.
Corning Incorporated
HP-AB-2-9
Corning. NY  1*831
U.S.A.
607-974-8328

Kevin Parrett
Environmental Chemist
Dames & Moore
1750 SW Harbor Way
Suite 400
Portland, OR  97201
U.S.A.
503-228-7688

Brad Parsons
Sr. Hazardous Materials Spec.
Calif. Dept. of Health Service
10151 Croydon Hay *3
Sacramento. CA  9S827-2106
U.S.A.
916-855-7816

Michael Parsons
Project Engineer
Exxon Research and Engineering Company
Analyzer Applications
P.O.Box 101, Park Avenue
Florham Park,  NJ  07932
U.S.A.
201-765-2933

James Pasmore
Outokurapu Electronics
P.O. Box L 1069
Langhorne, PA  19047
U.S.A.
215-741-4482

Kyle Peterson
Soil Scientist
P&W Land Consultants, Inc.
671 Lamberton Street
Trenton, NJ  08611
U.S.A..
609-695-2249

J. Gareth Pearson
Dir. Exposure Assessment Res,
U.S. EPA
EMSL-LV
P.O. Box 93478
Las Vegas, NV  89193
U.S.A.
Doug Peery
Field Analytical Specialist
IT Corporation
312 Directors Drive
Knoxville, TN  37923
U.S.A.
615-690-3211

Bob Pellissier
Manager
Gas Tech
Environmental Monitors [GEM]
8445 Central Avenue
Newark, CA  94560-3431
U.S.A.
415-794-6200

Barry V. Pepich
Section Leader
The Bionetics Corporation
7411 Beach Drive East
Port Orchard, MA  98366
U.S.A.
206-871-0748

Juan J. Perez
Chemistry Lab Technician
Science & Technology Corporation
425 E. 3rd Avenue tl
Salt Lake City, UT  84103
U.S.A.
801-355-7579

Randy W. Perils
Chemist
Ecology & Environment
1776 S. Jackson St.
Suite #200
Denver, CO  80210
U.S.A.
Symala Perry
ICF Technology, Inc.
10 university City Plaza
Suite 2400
Universal City. CA  91608
U.S.A.
818-509-3192

Gary Ferryman
Chemist
U.S. EPA
Region VIII
Box 25366 DFC
Denver, CO  80225
U.S.A.
(303)293-1541

Joe Peters
Millipore Corporation
SO Ashby Road
Bedford, MA  01730
U.S.A.
617-275-9200  Ext. 2337
                                         879

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February 12-14, 1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final Participants'  List
Mark A. Peters
EG&G Rocky Flats, Inc.
Building T130C
P.O.Box 464
Golden, CO  60402
U.S.A.
303-965-2322

Sabrina Peterson
Manager
IT Corporation
17461 Derian Avenue
Suite 190
Irvine, CA  92714
U.S.A.
714-261-6441

John M. A. Fetinarides
Graseby Ionics Ltd.
6 Millfield Hse, Hoodshots Mdw
Croxley Ctr., Watford, Herts,   WD1 8YX
United Kingdom
44 923-38483

Christian R. Petrich
Student
University of Idaho
Moscow, ID  83843
U.S.A.
208-885-5956

Colleen F. Petullo
Health Physicist
U.S. EPA-ORP
4220 South Maryland Pkwy.
Las Vegas, NV  89154
U.S.A.
702-798-2446

William G. Phillips
Program Manager
U.S. EPA
Off-Site Radiation Safety
P.O. Box 93478
Las Vegas, NV  89193-3478
U.S.A.
702-798-2305

Eugene Pier
Manager, Business Development
Rosemount Analytical
Dohrmann Division
3240 Scott Blvd.
Santa Clara, CA  95054
U.S.A.
408-727-6000

Brian M. Fierce
Senior Project Engineer
Hughes Aircraft Company
A1.2C851
P.O. Box 9399
Long Beach, CA  90810-0399
U.S.A.
213-513-4556
Harley V. Piltingsrud
Senior Research Physicist
NIOSH
MS R-8
4676 Columbia Parkway
Cincinnati, OH  45226
U.S.A.
513-841-4280

Stanislaw Fiorek
Director
Outokunpu Electronics, Inc.
Technology Development
860 Town Center Drive
Langhorne, FA  19047
U.S.A.
215-741-4482

Ann Pitchford
Environmental Scientist
U.S. EPA
EMSL-LV
P.O. Box 93748
Las Vegas, NV  89193-3478
U.S.A.
702-798-2366

Ron Polhill
Presentation Coordinator
Life Systems, Inc.
ICAIR
24755 Highpoint Road
Cleveland, OH  44122
U.S.A.
216-464-3291

Marc A. Fortnoff
Manager, Sensor/Laser Lab
Carnegie Mellon Research Inst.
4400 Fifth Avenue
Box 131
Pittsburgh, PA  15213
U.S.A.
412-268-3495

Judd Posner
Senior Research Chemist
NIOSH
4676 Columbia Parkway
MS R-8
Cincinnati, OH  45226
U.S.A.
513-841-4279

Edward J.  Poziomek
Senior Scientist
University of Nevada-Las Vegas
Environmental Research Center
Las Vegas, NV  89154-4009
U.S.A.
702-739-3382
                                            880

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February 12-14, 1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final Participants' List
Thomas H. Fiitchett
Chemist
U.S. EPA
Environmental Response Team
2890 Woodbridge Avenue
Edison, NJ  08837
U.S.A.
201-321-6738

John C. Querido
President
Jontec Systems, Inc.
P.O. Box 2163
Laguna Hills, CA  92654
U.S.A.
714-581-3972

Walfred R. Raisanen
V.P. Research & Development
Arizona Instrument
1100 E. University
Tenpe, AZ  85280
U.S.A.
602-731-3400

Gaurav Rajen
Hydrologist
Navajo Superfund Program
PO Box 2946
Window Rock. AZ  86515
U.S.A.
602-871-6859

Margaret M. Ravasz
Staff Assistant
Argonne National Laboratory
R&D Program Coordination
9700 S. Cass Avenue
Argonne, IL  60439
U.S.A.
708-972-3982

Larry G. Reed
Director
U.S. EPA
Hazardous Site Evaluation Div.
401 M Street SW
Washington, DC  20460
U.S.A.
202-475-8602

Keith Reichenbach
Senior Project Chemist
OHM Corporation
16406 US Route 224 East
Findlay, OH  45839-0551
U.S.A.
800-537-9540

Hank Reider
Manufacturer's Representative
Photovac International
25-B Jefryn Blvd. West
Deer Park, NY  11729
U.S.A.
516-254-4199
Dennis G. Revell
Chemist
U.S. EPA
Region IV
College Station Road
Athens, GA  30605
U.S.A.
404-546-3387

Peter I. Richter
Technical University Budapest
Institute of Physics
Budapest, XI. Budafoki
ut 8,   1111
Hungary
361-1853-230

Andrew Riddell
Chemist
Ecology and Environment, Inc.
101 Yesler May
Suite 600
Seattle, WA  98104
U.S.A.
206-624-9537

Mai Riddell
Managing Director
BioNebraska
29 Robin Rd.
Moorestown. NJ  08057
U.S.A.
609-235-8920

Bruce Riggle
Program Manager
Acurex Corporation
4915 Prospectus Drive
Durham, DC  27713
U.S.A.
919-544-4535

Cleve Rix
Technical Representative
Tekmar Company
7143 E. Kemper Road
Cincinnati, OH  45249
U.S.A.
800-543-4461

Michael D. Roach
President
Bison Engineering
P.O. Box 1703
30 S. Ewing
Helena, MT  59624
U.S.A.
406-442-5768

Albert Robbat
Tufts University
62 Talbot Ave.
Medford, MA  02155
U.S.A.
617-381-3474
                                         881

-------
                             Second International Symposium
                               Field Screening Methods for
                            Hazardous Wastes and Toxic Chemicals
                                  February 12-14. 1991
                             Sahara Hotel - Las Vegas, Nevada
                                 Final Participants'  List
 Adele J. Roberts
 Staff Engineer
 General Atomics 63-107
 P.O.  Box 85608
 3550  General Atomics Court
 San Diego,  CA  92186-9784
 U.S.A.
 619-455-2517
 Paul Runyon
 President
 Fartech Environmental
 192 Technology
 Suite S
 Irvine,  CA  92718
 U.S.A.
 Gary Robertson
 Chemist
 U.S.  EPA
 P.O.  Box 93478
 Las Vegas,  NV  89193-3478
 U.S.A.
 702-798-2215

 Wilson  Rodriquez
 VP International  Sales
 Gilian  Instrument Corporation
 35 Fairfield Place
 West Caldwell,  HJ  07006
 U.S.A.
 201-808-3355

 Jeffrey C.  Rogers
 Environmental Scientist
 Stata of Delaware
 Division of Air & Waste Mngmt.
 715 Granthen Lane
 New Castle,  DE 19720
 U.S.A.
 302-323-4542

 Jeffrey Rosenfeld
 Science Supervisor
 Lockheed Engineering & Sciences Co.
 1050  East Flamingo-Suite  120
 Las Vegas,  NV  89119
 U.S.A.
 702-734-3211

 Margaret A.  Rostker
 Acting  Deputy Director
 U.S.  EPA
 EMSL-LV
 P.O.  Box 93478
 Las Vegas,  NV  89193-3478
 U.S.A.
 702-798-2522

 Peter Rubens
 Project Manager
 Nachant Environmental
 5345  Timken  Street
 La  Mena,  CA   92042
 U.S.A.
 619-589-9000

Amy Rubert
EG&G  Idaho
P.O. Box  162S
Mailstop  3416
 Idaho Falls.  ID  83415
U.S.A.
208-526-9761
 James  Ryan
 Manager,  Environmental Mktg,
 The  Perkin-ELmer  Corporation
 761  Main  Avenue
 Norwalk,  CT  06859
 U.S.A.
 203-834-6856

 Mary Ryan
 Chemist
 Clean  Air Engineering
 500  West  Wood Street
 Palatine, IL  60067
 U.S.A.
 708-991-3300

 Sean Ryan
 Marketing Manager
 IDEXX  Corporation
 100  Fore  Street
 Portland, ME  04101
 U.S.A.
 207-774-4334

 Annette R.  Sackman
 Ecology and Environment,  Inc.
 1967 North  Gateway Boulevard
 Fresno, CA  93727
 U.S.A.
 209-453-9093

 M. A.  Saleh
 Chemist
 University  of Nevada-Las  Vegas
 Environmental Research Center
 Las Vegas,  NV  89154
 U.S.A.
 713-739-3382

 Harry Salem
 Chief, Toxicology Division
 U.S.  Army CRDEC
 SMCCR-RST
 Aberdeen  Proving Ground,  MD  21010-5423
 U.S.A.
 301-671-3034

 Stanley Salisbury
Regional Program Consultant
 NIOSH
Atlanta Regional Office
 101 Marietta Tower,  Suit* 1106
Atlanta,  GA  30323
U.S.A.
 404-331-2396
                                           882

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February 12-14, 1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final Participants1 List
 Howard  Salman
 Chemist
 Bureau  of Reclamation
 Denver  Federal Center
 Denver, CO  80225
 U.S.A.
 303-236-4290

 John Santolucito
 University of Nevada - LV
 Environmental Research Center
 4504 Maryland Parkway
 Las Vegas, NV  89134
 U.S.A.
 702-739-3382

 Richard Sasaki
 Branch  Chief
 Hawaii  Department of Health
 Air Surveillance & Analysis
 1270 Queen Emma St., Suite 900
 Honolulu, HI  96813
 U.S.A.
 808-548-3676

 Drew Sauter
 Consultant
 A. D. Sauter Consulting
 2356 Aqua Vista Avenue
 Henderson, NV  89014
 U.S.A.
 702-454-7884

 Sonia Savage
 Sales Representative
 Enseco, Inc.
 8335 Columbine Avenue
 California City, CA  93505
 U.S.A.
 619-373-8852

 Wayne N. Sawka
 Environ. Sciences Supervisor
 Aerojet Propulsion Division
 P.O. Box 13222
 Bldg. 2001.  Dept. 5730
 Sacramento, CA  95613
 U.S.A.
 916-355-5763

 Laura Scalise
 U.S. EPA
 Region  II
 2890 Hoodbridge Avenue
 Edison, NJ  06837
 U.S.A.
 201-906-6171

 Douglas T.  Scarborough
Chemist
U.S. Army T&HMA
Attn:  CETHA - TS - C
Building E4460
Aberdeen Proving Ground,  MD  21010-5401
U.S.A.
 301-676-7569
 John F. Schabron
 Manager
 Western Research Institute
 Analytical Research Division
 P.O.Box 339S
 Laramie, MY  82071
 U.S.A.
 307-721-2445

 Ronald Schalla
 Senior Research Engineer
 Battelle, Pacific Northwest Labs
 P.O. Box 999
 Richland, HA  99352
 U.S.A.
 509-376-5064

 Bill Scheidler
 ESAT Team Manager
 The Bionetics Corporation
 7411 Beach Drive
 Fort Orchard, HA  98366
 U.S.A.
 206-871-0748

 Ray Scheinfeld
 Senior Technical Manager
 Roy f. Heston
 One Weston Way
 West Chester, PA  19380
 U.S.A.
 215-430-7330

 Ray Schlosser
 Marketing Manager
 MDA Scientific
 405 Barclay Blvd.
 Lincolnshire, IL  60069
 U.S.A.
 708-634-2800

 Lawrence S.  Schmid
 President & CEO
 Strategic Directions International
 6242 Hestchester Fkwy, Ste 100
 Los Angeles,  CA  90045
 U.S.A.
 213-641-4982

 Steven P.  Schuetz
 Project Scientist
Roy F.  Heston/REAC
GSA Raritan Depot
Edison,  NJ  06837
U.S.A.
 208-632-9200

Mike Scott
Field Engineer
Hewlett-Packard Company
9606 Aero Drive
San Diego,  CA  92123
U.S.A.
619-541-7251
                                         883

-------
                            Second International  Symposium
                              Field Screening Methods for
                           Hazardous Wastes  and Toxic Chemicals
                                 February 12-1*.  1991
                            Sahara Hotel - Las  Vegas, Nevada
                                Final Participants'  List
Joseph A. Scroppo
President
Bladon International, Inc.
880 Lee Street
Oes Flaines, IL  60016
U.S.A.
708-803-2396

David P. Seely
Environmental Engineer
U.S. EPA
Region V, SHS-11
230 S. Dearborn
Chicago, IL  60604
U.S.A.
312-886-70S8

Craig Sellman
The Perkin-Elmer Corporation
M/S 12
761 Main Avenue
NorwaUc, CT  068S9
U.S.A.
203-762-6050

Rashmi Shah
Laboratory Director
Enseco - CRL
7440 Lincoln Hay
Garden Grove, CA  92641
U.S.A.
714-898-6370

Mahmoud R. Shahriari
Research Professor
Rutgers University
Fiber Optic Center
P.O. Box 909
Piscataway, NJ  088SS
U.S.A.
201-932-5033

David Sheealey
Manager
Western Research Institute
Environmental Technology
1624 Custer
Custer, WY  82070
U.S.A.
307-721-2355

Jeffrey B. Sherard
Director, Western Opereations
Tracer Research Corporation
3855 N. Business Center Drive
Tucson, AZ  85705
U.S.A.
602-888-9400

Liu Shili
University of Connecticut
Environmental Research Inst.
191 Auditorium Road
Storr, CT  06269
U.S.A.
203-486-5482
Donald Shoff
U.S. Army CRDEC
SMCCR-RSL
Aberdeen Proving Ground, MD  21010
U.S.A.
301-671-3129

Mark E. Silverstein
Senior Scientist
Lockheed Engineering & Sciences Co.
   (LESC)
1050 E.Flamingo Rd.,St«.1050
Las Vegas. NV  89119
U.S.A.
702-734-3291

David M. Simanic
Manager - Field Operations
NUS Corporation
Laboratory Services Group
5350 Campbells Run Road
Pittsburgh. PA  15205
U.S.A.
412-747-2506

David Simmons
Environ. Quality Coordinator
Louisiana Dept. Environ. Quality
Office of Legal Affairs Enf.
333 Laurel Street
Baton Rouge, LA  70804
U.S.A.
504-342-9171

Eileen M. Simmons
Project Manager
ICF Technology
P.O. Box 280041
Denver, CO  80228
U.S.A.
303-236-7081

Mila Simmons
Associate Professor
Dept. of Env & Ind Health
The University of Michigan
2530 SPHI
Ann Arbor, MI  48109
U.S.A.
313-763-9269

Stephen J. Simon
Project Manager
Lockheed
900 Grier Drive
Suite  B
Las Vegas, NV  89119
U.S.A.
702-361-5789
                                            884

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February 12-14, 1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final Participants' List
Orman Simpson
Director, Remote Sensing Tech
MDA Scientific
3000 Northwoods Parkway
Suite 185
Horcross, GA  30071
U.S.A.
404-242-0977

Robert R. Sims
Environmental Chemist
Science Applications Int'l Corp.
4224 Campus Point Court
San Diego, CA  92121
U.S.A.
619-535-7507

Mahadeva P. Sinha
Technical Staff
Jet Propulsion Laboratory
4800 Oak Grove Drive
Pasadena, CA  91109
U.S.A.
818-354-6358

Barbara Skiffington
Engineer Analyst
General Research Corporation
1655 N.  Fort Meyer Drive
Arlington, VA  22209
U.S.A.
703-2*3-4822

Henry Skinner
Organic Chemist
ICF Technology
P.O. Box 280041
Lakewood, CO  80228
U.S.A.
303-236-7265

Russell J. Sloboda
Senior Chemist
NUS Corporation
999 Wast Valley Road
Wayne. PA  19067
U.S.A.
215-687-9510

Gary Small
Asst.  Professor
University of Iowa
Department of Chemistry
Iowa City, IA  52242
U.S.A.
319-335-1370

Richard Smardzewski
Supervisory Research Chemist
U.S. Army CRDEC
SMCCR-RSL/R.  Smardzewski
Aberdeen Proving Ground, MD  21010-5423
U.S.A.
301-671-2560
Donald Smith
Chemist
U.S. EPA
NEIC
Box 25227, Bldg. 53, DFC
Denver, CO  80225
U.S.A.
303-236-5132

J. Michael Smith
Engineer
U.S. EPA
1S04 Avenue A
Montgomery. AL  36115-2601
U.S.A.
205-270-3422

John M. Smith
Director, Haz Mtls. Res. Fac.
Battelle
505 King Avenue
Columbus, OH  43201-2693
U.S.A.
614-424-5392

Randy L. Snipes
Project Manager
Martin Marrietta Energy Systems
P.O. Box 2003
Oak Ridge, IN  37831-7606
U.S.A.
615-435-3128

John Snyder
Senior Chemist
Lancaster Laboratories Inc
2425 New Holland Pike
Lancaster, PA  17601
U.S.A.
717-656-2301

James W. Soroers
Engineering Group Manager
General Research Corporation
1655 N. Fort Meyer Drive
Arlington, VA  22209
U.S.A.
703-243-4822

Chris Sooner
Senior Research Chemist
Occidental Chemical Corp
2801 Long Road
Grand Island,  NY  14072
U.S.A.
716-773-8100

Herking Song
Electrochemist
S.R.I. International
333 Ravenswood Avenue 4AG100
Menlo Park,  CA  94025
U.S.A.
415-859-6135
                                          885

-------
                            Svcond International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February 12-14, 1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final Participants'  List
Carl Soong
Reynolds Electrical & Engineering Co.
P. O. Box 98521
Las Vegas, NV  89193-8521
U.S.A.
702-295-1913

G. Wayne Sovocool
Chemist
U.S. EPA
EMSL-LV
P.O. Box 93476
Las Vegas, NV  69193-3478
U.S.A.
702-798-2212

Glenn E. Spangler
Environmental Technologies Grp.
1400 Tylor Ave.
P.O. Box 9640
Baltimore, MO  21204
U.S.A.
301-321-5261

Thomas Spargo
Chemical Engineer
Ecology and Environment, Inc.
One Techview Drive
Cincinnati, OH  45215
U.S.A.
513-733-3107

S.J. Spijk
Technologic TNO
7334 DT Apeldoorn
Hoofdgroep Maatschappelijke
LN V Wetenenk 501,
Thomas M. Spittler
Lab Director
U.S. EPA
Region I
60 West View Street
Lexington, MA  02173-3165
U.S.A.
617-860-4334

Stan Spurlin
Head, Special Programs
Midwest Research Institute
425 Volker Blvd.
Kansas City, MO  64110
U.S.A.
816-753-7600. Ext. 312

Randy St. Germain
Graduate Student
North Dakota State University
Chemistry Department
Ladd Hall
Fargo, ND  56105
U.S.A.
701-237-8588
Robert L. Stamnes
Civil Engineer
U.S. EPA
Region X
1200 Sixth Avenue
Seattle, WA  96101
U.S.A.
206-553-1512

Allan Staple
President
EnSys
Emperor Boulevard,Royal Center
Research Triangle Park, NC  27709
U.S.A.
919-941-5509

Steve Stasko
Manager, Technical Programs
Air & Waste Management Assn.
3 Gateway Center
4 West
Pittsburgh, PA  15222
U.S.A.
412-232-3444

Ken Stehr
Kevex Instruments
35S Shoreway Road
San Carlos, CA  94070-1308
U.S.A.
415-591-3600

Kathie Stephens
Assistant Registration Coord.
U.S. EPA
EMSL-LV
P.O. Box 93478
Las Vegas, NV  B9193-347S
U.S.A.
702-798-2432

Richard L. Stephenson
CIH Corp. Industrial Hygienist
Thomson Consumer Electronics, Inc.
3301 South Adams Street
Marion, IN  46953
U.S.A.
317-662-5439

Matt Stinchfield
Vice President
Zenitech Corporation
1951 Grant Road
Suite 110
Tucson, AZ  85745
U.S.A.
602-798-1466

Peter J. Stopa
Chemist
U.S. Army CRDEC
SMCCR-DDT
Aberdeen Proving Ground, MD  21010-5423
U.S.A.
(301)671-5578
                                            886

-------
                            Second International Sympoilum
                              Field Screening Methods for
                           Hazardous Hastes and Toxic Chemicals
                                 February 12-14,  1991
                            Sahara Hotel - Las Vegas, Nevada
                                Final Participants'  List
Michael Story
Corporate Technology
Finnigan Corporation
355 River Oaks
San Jose, CA  95134
U.S.A.
408-433-4800

Reginald Stroupe
President
Nutech Corporation
2806 Cheek Road
Durham, NC  27704
U.S.A.
919-682-0402

Janes D. Stuart
Assoc. Prof, of Chemistry
University of Connecticut
Department of Chemistry
215 Glenbrook Rd.
Storrs, CT  06269-3060
U.S.A.
203-486-2125

Daniel Stubblefield
Manager, Business Development
Corning, Inc.
35 W. Market, MP-RO-2
Corning, NY  14331
U.S.A
607-974-4516

Ralph J. Sullivan
Senior Chemist
ICF Kaiser Engineers
2700 Chandler Avenue,Bldg. C.
Las Vegas, HV  89120
U.S.A.
702-795-0515

Chris Sundeen
Ecology & Environment
1776 South Jackson *200
Denver, CO  80210
U.S.A.
303-757-4984

Chris Button
Vice President of Research
Ruska Laboratories
3601 Dunvale
Houston. TX  77063
U.S.A.
313-975-0540

Basil Swanson
Los Alamos National Laboratory
INC-Y, MSC344
Los Alamos, NM   97544
U.S.A.
505-667-5814
Rick Swatzell
Project Manager
Martin Harrietts Energy Systems
P.O. Box 2003
Oak Ridge, TH  37831-7606
U.S.A.
615-435-3126

Mary Beth Tobacco
Geo-Centers
7 Wells Ave.
Newton Centre, MA  021S9
U.S.A.
617-964-7070

Jason Talbot
Environ. Quality Specialist
Louisiana Dept. Environ. Quality
Office of Legal Affairs Enf.
333 Laurel Street
Baton Rouge, LA  70804
U.S.A.
504-342-9171

Charles Tanner
Exhibit Coordinator
Life Systems, Inc.
ICAIR
24755 Highpoint Road
Cleveland. OH  44122
U.S.A.
(216)464-3291

E. Jennings Taylor
Mngr. Electrochemical Technol.
Physical Sciences, Inc.
20 New England Business Center
Andover, MA  01810
U.S.A.
508-689-0003

John H. Taylor
President & General Manager
Analytical Technologies
5550 Morehouse Drive
San Diego, CA  92121
U.S.A.
619-458-9141

Robert L. Taylor
Inorganic Analytical Chemist
ICF Technology
255 Zang Street  #2531
Lakewood, CO  60228-1012
U.S.A.
303-986-7059

Todd A. Taylor
Research Associate
Tufts University
62 Talbot Avenue
Medford, MA   02155
U.S.A.
617-381-3095
                                          887

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February 12-14. 1991
                            Sahara Hotel - Las Vegas,  Nevada
                                Final Participants' List
Victoria Taylor
Senior Associate
ICF Kaiser Engineers
160 Spear Street
San Francisco, CA  94105
U.S.A.
415-882-3029

Tim Theisen
Optimal Technology
6430 Via Real
Suite 6
Carpinteria. CA  93013
U.S.A.
805-684-6226

Tony Theisen
Optimal Technology
6430 Via Real
Suite 6
Carpinteria, CA  93013
U.S.A.
805-684-6226

Jay Thomas
New Business Development
Sippican
Seven Barnabas Road
Marion, MA  02738
U.S.A.
508-748-1160

John R. Thompson
Senior Scientist
Science Applications Int'l Corp.
2109 Ail Park Road SE
Albuquerque, NM  87106
U.S.A.
505-247-8787

Darwin Throne
Managing Director
OQ Alliances
1290 Ridder Park Drive
San Jose, CA  95131-2398
U.S.A.
408-437-8300

Richard M. Tinlin
Vice President
Geraghty & Miller
9831 South 51st Street
Phoenix, AZ  85044
U.S.A.
602-496-0025

Harold W Tomlinson
Manager-Optical Processing
GE-Corporate Research & Development
1 River Rd, KWC-1317
Schenectady, NY  12309
U.S.A.
518-387-5687
Karl A. Touet
Project Manager/Chemist
NUS Corporation
19 Crosby Drive
Bedford, MA  01730
U.S.A.
617-275-2970

Lila Accra Transue
Chemist
Ecology and Environment, Inc.
101 Yesler Hay, Suite 600
Seattle, WA  98104
U.S.A.
206-624-9537

Jean-Luc Truche
R&D Section Manager
Hewlett-Packard
1601 California Avenue
Palo Alto, CA  94304
U.S.A.
415-857-6079

Pdtricio E. Trujillo
Staff Member
Los Alamos National Laboratory
P.O. Box 1663
EES-1, Mail Stop D469
Los Alamos, NM  87545
U.S.A.
505-667-1224

Ernesto C. Tuazon
Research Chemist
University of California
Statewide Air Pollution
Research Center
Riverside, CA  92521
U.S.A.
714-787-5140

Bruce R. Tucker
Director Marketing & Sales
Quadrel Services, Inc.
10075 Tyler Place #9
Ijamsville, MD  21754
U.S.A.
301-874-5510

Alta Turner
Statistician
CH2M Hill
777 108th Avenue, Northeast
Bellevue, HA  98009-2050
U.S.A.
206-453-5005, ext.5657

Stephen Turner
Environmental Chemist
ABB Environmental Services
261 Commercial St.
Portland, ME  04112
U.S.A.
207-775-5400
                                           888

-------
                            Second International Symposium
                              Field Screening Methods for
                           Hazardous Wastes and Toxic Chemicals
                                 February 12-1«, 1991
                            Sahara Hotel - Las Vegas,  Nevada
                                Final Participants'  List
Bruce G. Tuttle
Industrial Hygienist
Westinghouse Hanford Company
P.O. Box 1970 N3-06
Richland, MA  99352
U.S.A.
509-376-2648

Joe Vagaggini
Senior Chemist
ICF Kaiser Engineers
9300 Lee Highway
Fairfax. VA  22031
U.S.A.
703-934-3163

Cornelius A. Valkenburg
Senior Chemist
ICF Kaiser Engineers
2700 Chandler Avenue
Las Vegas, NV  B9120
U.S.A.
702-795-0515

Melinda Van
Vice President
FemtoScan Corporation
1834 West
4700 South
Salt Lake City, UT  84118
U.S.A.
801-964-2317

Jeff Van Ee
Electronics Engineer
U.S. EPA
EMSL-LV
P.O.Box 93478
Las Vegas, NV  89193-3478
U.S.A.
702-798-2367

Jeanette Van Emon
Research Molecular Biologist
U. S. EPA
EMSL-LV
944 E. Harmon
Las Vegas, NV  89119
U.S.A.
702-798-2154

Derk Van Ree
Drs.
Delft Geotechnics
Stieltjesweg 2
Delft
Z600 AB,
The Netherlands
31  15 693500
Katrina E. Varner
Environmental Scientist/Chem.
U.S. EPA
EMSL-LV
P.O. Box 93478
Las Vegas, HV  89193-3478
U.S.A.
702-798-2645

Judy Vaughn
Industrial Hygienist
Westinghouse Hanford Company
P.O. Box 1970 N3-06
Richland, WA  99352
U.S.A.
509-376-2709

Nina Vendelboe
Chemical Engineer
Danish Geotechnical Institute
Maglebjergvej 1
2800 Lyngby,
Denmark
+45 4288 4444

David Vener
Program Manager
Xontech. Inc.
6669 Hayvenhurst Avenue
Van Nuys, CA  91406
U.S.A.
818-787-7380

John W. Verbicky
Mngr. Environ. Technol. Lab
General Electric
Corporate R&D
Bldg. K-l Room 5A58, River Rd.
Schenectady. NY  12301
U.S.A.
518-387-6177

Harold A. Vincent
Chemist
U.S. EPA, EMSL-LV
QAD/QAB
P.O. Box 93478
Las Vegas, NV  89193-3478
U.S.A.
702-798-2129

Rick Vollweiler
President
Art's Manufacturing & Supply Company
105 Harrison
American Falls, ID  83211
U.S.A.
800-635-7330

Christoph Von Hoist
Instistut Biocontrol Germany
D-6507 Ingelheim
Hamburgerstr.7,
Germany
06732-787-200
                                          889

-------
                             Second International Symposium
                               Field Screening Methods for
                            Hazardous Wastes and Toxic Chemicals
                                  February 12-14,  1991
                             Sahara Hotel - Las Vegas,  Nevada
                                 Final Participants'  List
 Brian Wagner
 Field Chemist III
 GTEL Environmental Labs
 Meadowbrook Industrial Park
 Milford,  NH  03055
 U.S.A.
 603-672-4835

 Sandy Wagner
 Project Leader
 Los Alamos National Laboratory
 P.O.  Box 1663 MS K481
 Los Alamos,  NM  87545
 U.S.A.
 505-665-2126

 Robert I.  Wallace
 Hydrogeologist
 Target Environmental Services
 9180  Rumsey Road
 Columbia,  MD  20145
 U.S.A.
 301-992-6622

 Barb  Walsh
 Technologist
 Esso  Petroleum Canada
 4S3 Christina St S
 Sarnia ON,    N7T 7M1
 Canada
 519-339-7023

 Paul  R.  Walsh
 Research Technologist
 Esso  Petroleum Canada
 Research  Dept
 PO Box 3022
 Sarnia ON,    N7T 7M1
 Canada
 519-339-4815

 James L. Walsh.  Jr.
 Senior Research Engineer
 Georgia Tech Research Institute
 O'Keefe Building,Rm.039
 Altanta, GA   30332
 U.S.A.
 404-894-8054

 Claudia Walters
 Chemist/Q.A.  Officer
 U.S.  EPA
 Chesapeake Bay Liaison Office
 410 Severn Avenue
 Annapolis, MD 21403
 U.S.A.
 301-267-0061

 Hui Wang
 Environmental Chemist
Roy F. Weston
Landmark One
On* Van do Graff Drive
Burlington, MA  01803
U.S.A.
617-860-4394
 Steven Ward
 Senior Research Chemist
 University of Nevada  - LV
 Environmental Research Center
 4505 South Maryland Parkway
 Las  Vegas,  NV  89154
 U.S.A.
 702-739-1042

 Susan J. Ward
 Environmental Mgmt Specialist
 Clark County Health District
 625  Shadow Lane
 Las  Vegas,  NV  89127
 U.S.A.
 702-383-1274

 William A.  Warner
 Environmental Scientist
 U.S.  EPA
 P.O.  Box 25366
 Denver Federal Center
 Denver, CO  80225
 U^S.A.
 303-236-5064

 Robert Watson
 Regional Coordinator
 Superior Analytical Labs,  Inc.
 835  Arnold  Drive, Suite  106
 Martinez, CA  94533
 U.S.A.
 415-229-1590

 Steven  E. Way
 On-Scene Coordinator
 U.S.  EPA
 Region  8
 999  18th Street
 Denver, CO   80202
 U.S.A.
 303-293-1723

 Charles A. Weisberg
 Chemist
 U.S.  EPA Central Regional Lab
 839 Bestgate  Road
 Annapolis, MD  21401
 U.S.A.
 301-266-9180

 Jennifer Wendel
 Chemist
 Ecology and  Environment,  Inc.
 208 South LaSalle,Suite 1300
Chicago, IL   60604
 U.S.A.
 312-201-3809

Clarence S. Wentzel
Director of Research
Arnel, Inc.
 3145 Bordentown Avenue
Parlin, NJ  08859
U.S.A.
201-721-4300
                                           890

-------
                            Second International Symposium
                              Field Screening Methods  for
                           Hazardous Wastes and Toxic  Chemicals
                                 February 12-14,  1991
                            Sahara Hotel - Las Vegas,  Nevada
                                Final Participants'  List
Ralph Was twig
Senior Research Associate
Corning Incorporated
SP-FR-1-7 Sullivan Park
Corning, NY  14831
U.S.A.
607-974-3141

Kevin Whitehead
Hewlett-Packard
Box 1100
Route 41
Avondale, PA  19311
U.S.A.
215-268-5453

Kevin Whitehead
Marketing Specialist
Hewlett-Packard Company
5161 Lankershire Blvd
North Hollywood, CA  91601
U.S.A.
818-505-5785

Red [William] Whittaker
Director, Field Robotics Ctr
Carnegie Mellon University
5000 Forbes Avenue
Pittsburgh. PA  15213
U.S.A.
412-268-6559

Terry Wilks
Technical Director
Dakatech Precision Sampling
P.O. Box 15886
Baton Rouge, LA  70895
U.S.A.
504-927-1128

Llewellyn Williams
Director, QAMD Division
U.S. EPA. EMSL-LV
QAD
P.O. Box 93478
Las Vegas,  XV  89193-3478
U.S.A.
702-798-2138

Rod Williams
Hollingsworth Dames and Moore
24 Goskar Avenue Alderley
Queensland,   4051
Australia
07356-5278

Tony Williams
V.P. Inorganic Division
VG Instruments
32 Commerce Center
Danvers, MA  01923
U.S.A.
508-777-8034
Doug Winters
Manager
Mantech Environmental
Athens Operations
545 Research Drive
Athens, GA  30605
U.S.A.
404-546-7611

Marcus B. Wise
Research Staff Member
Oak Ridge National Laboratory
Analytical Chemistry Division
P.O. Box 2008, MS 6120
Oak Ridge, TN  37831-6120
U.S.A.
615-574-4867

Hank WohLtjen
Microsensor Systems
6800 Versar Center
Springfield, VA  22151
U.S.A.
703-642-6919

Clayton Wood
R&D Manager
HNU Systems, Inc.
160 Charlemont Street
Newton, MA  02161-9987
U.S.A.
617-964-6690

W. S. "Bud" Hood
MTS
ANCAL
3305 Spring Mountain Road
Suite 60
Las Vegas, NV  89102-8624
U.S.A.
702-434-1501

Stan Woods
Development Engineer
Hewlett-Packard
3500 Deer Creek Road
Palo Alto, CA  94304
U.S.A.
415-857-6496

Ray Worden
President
RUSKA Laboratories
3601 Dunvale
Houston, TX  77063
U.S.A.
713-975-0547

Bob W. Wright
Senior Research Scientist
Battelle Pacific Northwest Labs
P.O. Box 999
Richland, WA  99352
U.S.A.
509-376-1661
                                          891

-------
                             Second International Symposium
                               Field Screening Methods for
                            Hazardous Wastes and Toxic Chemicals
                                  February 12-14, 1991
                             Sahara Hotel - Las Vegas, Nevada
                                 Final Participants'  List
 John Wronka
 Brucker Instruments
 19 Fortune Drive
 Manning Park
 Billerica, MA  01821
 U.S.A.
 508-667-9580

 Duane S.  Wydoski
 Chemist
 United States Geological Survey
 National  Water Quality Lab.
 5293 Ward Road
 Arvada, CO  80002
 U.S.A.
 303-236-5345

 Dennis  J.  Wynne
 Chief,  Technical Support Div.
 U.S.  Army TSHMA
 812 Yvette Drive
 Forest  Hill,  MD  21050
 U.S.A.
 301-671-2466

 Nabil Yacoub
 Operations Manager
 State of  California
 Hazardous  Materials  Laboratory
 2151  Berkeley Hay
 Berkeley,  CA  94704
 U.S.A.
 415-540-3315

 Tracy Yerian
 Senior Chemist
 Ecology and Environment,  Inc.
 101 Yesler Way Suite 600
 Seattle, WA  98104
 U.S.A.
 206-624-9537

 Matthias Yoong
 Mngr. Environ,  Systems Group
 Xontech
 6862  Havenhurst  Ave.
 Van Nuys,  CA   91406
 U.S.A.
 818-787-7380

 Barbara Young
 Senior Research  Scientist
Millipore  Corporation
 80  Ashby Road
Bedford, MA 01730
U.S.A.
617-275-9200   Ext. 2303

David R. Youngman
Staff Scientist
Lockheed Engineering & Sciences Co.
6585 Paradise Road
Las Vegas, NV  89120
U.S.A.
702-361-1626
Edward Zahnow
Research Associate
DuPont Co.
1103 Greenway Road
Wilmington, DE  19803
U.S.A.
302-695-1433

Andrew T. Zander
Director,Measurements Lab.
Varian Research Center
3075 Hansen Way
Palo Alto, CA  94303
U.S.A.
415-424-6187

Kaveh Zarrabi
Principal Investigator
University of Nevada-Las Vegas
Environmental Research Ctr.
4505 Maryland Parkway
Las Vegas, NV  89154
U.S.A.
702-739-1042
Yasoob Zia
Louisiana Dept. Environ.
333 Laurel Street
Baton Rouge, LA  70804
U.S.A.
504-342-9171
Quality
                                           892

-------