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 P