SEVENTH ANNUAL
   WASTE TESTING
        AND
QUALITY ASSURANCE
    SYMPOSIUM
      JULY 8-12,1991
  GRAND HYATT WASHINGTON
    WASHINGTON, D.C.
    PROCEEDINGS
       Volume II

-------
              VOLUME
                    II
THE SYMPOSIUM IS MANAGED BY THE AMERICAN CHEMICAL SOCIETY
                  pi inted on recycled paper

-------
                    TABLE  OF CONTENTS
                                        Volume  II

Paper                                                                             Page
Number                                                                           Number


 ORGANICS
 44.      Methods for the Determination of Volatile Organic Compounds in Soil Samples. T. A. Bettor,       U -   1
         J. W. Eichelberger

 45.      Concentration of Water Soluble Volatile Organic Compounds From Aqueous Samples by          n -   2
         Azeotropic Microdistillation. M. L. Bruce, R. P. Lee, M. W. Stephens

 46.      Pollution Reduction in the Laboratory Through the Use of Smaller Initial Sample Size.            n -  17
         D. J. Bencivengo, B. N. Colby, P. W. Ryan

 47.      Extraction of Phenolic Compounds From Water Samples Using Styrene-Divinylbenzene SPE       n -  27
         Disks. C. G. Markell, D. F. Hagen

 48.      Comparison of Alternative Methods for Analysis of Volatile Organic Contaminants.              n -  38
         J. E Ryan, T. C. Voice

 49.     Evaluation of Sample Preparation Methods for Solid Matrices. V. Lopez-Avila, J. Milanes,         n -  40
         N. Dodhiwala, J. Benedicto, W. F. Beckert

 50.      Analysis for Selected Appendix DC Compounds in Environmental Matrices by High              II -  53
         Performance Liquid Chromatrography/Particle Beam Mass Spectrometry. J. L. Cornell,
         J. C. Lowry, M. D. Tilbury

 51.      The Implementation of HPLC/Post-Column Techniques for Rugged Carbamate and              n -  66
         Glyphosate Analysis. M. W. Dong,  M. V. Pickering

 52.      Determination of Low-Level Explosive Residues in Water by HPLC: Solid Solid-Phase           n -  68
         Extraction vs. Salting-Out Solvent Extraction. M. G. Winslow, B. A. Weichert, R. D. Baker

 53.      Reduction of Azo Dyes to Aromatic Amines for Environmental Monitoring. R. D. Voyksner,        E -  82
         J. T. Keever, H. S. Freeman, W. N. Hsu, L. D. Betowski

 54.      Hazardous Waste Component Identification Using Automated Combined GC/FTIR/MS.           n -  84
         R. J. Leibrand

 55.      Environmental Applications of Multispectral Analysis. J. M, McGuire                        H -  96

 56.      Sample Preparation Using Supercritical Fluid Extraction Methodology. W. F. Beckert,             n -  97
         V. Lopez-Avila

 57.      The Research Status of Supercritical Fluid Extraction for the Analysis of PCBs In Incinerator       n - 106
         Ash. R A. Pospisil, M. A. Kobus, C. R. Hecht

 58.      Supercritical Fluid Extraction (SFE) of Total Petroleum Hydrocarbons (TPHs) With Analysis       H - 117
         By Infrared Spectroscopy. Af. L. Bruce, R. P. Lee, M. W. Stephens

-------
59.       Application of Supercritical Fluid Extraction of Dioxins/Furans From Soil and "PUF".               n - 125
         J-PHsu, J. C. Pan, K. Villalobos, G. P. Miller

60.       The Application of Supercritical Fluid Chromatography to the Analysis of Herbicides and            n - 126
         Pesticides in TCLP Extracts. C. R. Hecht, P. A. Pospisil, M. A. Kobus, M. F. Marcus

61.       Problem Solving in the Organic Extractions Laboratory: Herbicides. />. Smith, J, Doeffinger,         n - 139
         T. Wittwer, J. Giannella, C. Lott, L. Stanton, E. Alverson, S. Fitzgerald, K. Klinger

62.       Infrared Microsampling for the Qualitative Analysis of Organics Extracted from Soil Samples.        n - 152
         M. R Putter, F.  J. Weesner

63.       A Performance  Comparison Study of Different Types of Devices for Solid Phase Extraction.         n- 153
         K /. Lee, E. N.  Amick, J. A. Berges, G. L. Robertson

64.       Standard Reference Spectra for MS/MS Quality Assurance, Performance Evaluation, and            n- 154
         Proficiency Testing: X[rf]Q Tandem Mass Spectrometers. R. I. Martinez

65.       Improved Techniques for Formaldehyde Analysis by HPLC Using Automated Sample               n- 165
         Preparation and Diode Array Detection. B. Goodby, S. Vasavada, J. Carter, L. Schaleger

66.       An Interlaboratory Comparison Study of Supercritical Fluid Extraction for Environmental            n- 179
         Samples. T. L. Jones, T. C. H. Chiang

67.       An Analytical Manual for Petroleum Products in the Environment. M, W, Miller,                   H - 181
         M. M. Ferko, F. Genicola, H. T. Hoffman, A. J. Kopera

68.       Evaluation of Liquid/Solid Extraction for the Analysis of Organochlorine Pesticides and PCBs        n- 182
         in Typical Ground and Surface Water Matrices. A. D. O'DonneU, D. R. Anderson,
         J. T. Bychowski, C. G. Markell, D. F. Hagen

69.       Improving the Analysis of Semi-Volatile Pollutants. C.  Vargo, N. Mosesman, G. Barone             n - 195

70.       Electrospray Combined with Ion Trap Mass Spectrometry for Environmental Monitoring.            n - 203
         R. D. Voyksner, H-Y Lin

71.*     Recent Advances in the Use of Supercritical Fluid Extraction for Environmental Applications.        n - 205
         /. M. Levy, A. C. Rosselli, D. S. Boyer, M. Ashraf-Khorassani

72.       Using Supercritical Fluid Extraction to Separate Diesel From Soil Matrices. C. A. Craig,            U - 206
         S. Prashar, J. Cunningham, B. E. Richter, A. Rynaski

 73.      Creative Review of 'Tentatively Indentified Compound" Data Using the Retention Index.            n - 217
         W. R Eckel

74.       High Efficiency GPC Cleanup of Environmental Samples—Column Optimization.                 n - 232
         G. J. FalUck, R. Cotter, R. Foster, R. L. Wellman

75.       Efficient Aqueous Extraction Using an Emulsion Phase Contactor. K. R Kelly, L. C. Schrier,         n - 242
         K. C. Kuo

76.       SFE Practical Applications for Environmental and Industrial Samples. L. J. D. Myer,                n - 249
         J. Tehrani, P. K. Mignon
INORGANICS
77.      Microwave Sample Preparation Methods for Environmental Analysis. H. M. Kingston,              n - 253
         F. A. Settle, M. A. Pleva, L. lassie, P. Walter, J. Petersen. B. Buote

-------
78.      Comparison of Procedures for TCLP Extract Digestion; Conventional vs. Microwave.              EL - 255
        V. L. Verma, T. M. McKee

79.      Sample Decomposition in Closed Vessels with a Pressure Controlled Microwave Oven.             n - 265
        E Panholzer, G. Knapp, P. Kettisch, A. Schalk

80.      State-of-the-Artof Microwave Digestion Methods for Environmental Analysis. M.E.Tatro        n-266

81.      The Application of X-Ray Fluorescence Spectroscopy for Rapid Hazardous Waste                n - 271
        Classification and Screening. P. A. Pospistt, H. Van Kley, M. F. Marcus, C. Taylor,
        N. K. Shah, E. S. Tucker

82.      Semi-Quantitative Determination of the Inorganic Constituents in Specific and Non-specific        n - 282
        Categorical Solid Waste Using EDXRF. K. Fennell, R. M. Olbrot, T. G. Howe

83.      Identifying Sources of Environmental Contamination through Laser Sampling ICP-Mass           n - 284
        Spectrometry. K. J. Fredeen, M. Broadhead

84.      ICP/MS Analysis of Toxic Characteristic Leaching Procedure (TCLP) Extract: Advantages         H - 298
        and Disadvantages. M. G. Goergen, V. F. Murshak, P. Roettger, I. Murshak, D. Edelman

85.      Chromium VI:  An Overview of Its Relevant Environmental Occurrence, Analytical Methods       n - 312
        of Quantitation, and Report on Recent Ion Chromatography Methods Development and
        Validation Activities. L. B. Lobring

86.      Rapid High Performance Microwave Digestion. R. Rubin,  M. Moses                           n - 314

87.      The Performance of a Low Cost ICP-MS for the Routine Analysis of Environmental Samples.       n - 315
        R. C. Seeley, T. M. Rettberg, P. D. Blair

88.      Robotics for Automated Digestion of Environmental Samples. A. C. Grillo, C. Balas              n - 317

89.      Application of Laser Sampling ICP-Mass Spectrometry to  Environmental Analysis.               n - 318
        E. R. Denoyer,  K. J. Fredeen, R. J. Thomas
REGULATORY COMPLIANCE
90.      Status of Developing Land Disposal Restrictions for Superfund Soils. R. Troast, C. K. Ofrutt,       n - 321
         J. O. Knapp

91.      Certification Protocol for Meeting Organic Treatment Standards for Incineration Ash.              n - 337
         W. R. Schofield, J. W. Kolopanis, T. S. Johnson

92.      Factors Affecting the Admissibility and Weight of Environmental Data as Evidence.               n - 349
         J. C. Worthington, K. G. Luka

93.      Review of Groundwater Monitoring Requirements at RCRA Sites. W. G. Stek                   n - 350

94.      The Paperless Environmental Laboratory: A Plan for Realization. /. C. Worthington,              n - 361
         G. A. Duba

95.      Data Management Issues in the Hazardous Waste Industry. G. A. Austiff, J. Krecisz                n - 362

Affi/GROUNDWATER

96.      New Directions in RCRA Ground-Water Monitoring Regulations. /. R. Brown, V. B. Myers,        H - 379
         A. E. Johnson

-------
97.     De-Mystifying the Problem of Filtered Vs. Unfiltered Samples. It D. Brown                    H - 394

98.     Determination of Target Organics In Air Using Long Path Spectroscopy. R. D. Spear,
        P. D. Greenlaw, R. J. Bath                                                            H - 409

99.     Measurement of Toxic Organic Compounds in Landfill Gas Samples Using Cryogenic             n - 410
        Trapping and Full Scan GC/MS. S. D. Hoyt

100.     The Determination of the Heat of Combustion and Water Content of Incinerator Feeds Using        n - 411
        Near Infrared Spectroscopy. N. K. Shah, P. A. Pospisil, R. A. Atwood,
        D. L. Wetzel, A. Eilert

101.     Source Sampling and Analysis Guidance: A Methods Directory.  M. D. Jackson,                  n - 422
        L. D. Johnson, K. W. Baughman, R. H. James, R. B. Spafford


AIR &  GROUNDWATER POSTERS


102.     A Field Investigation of Groundwater Monitoring Well Purging Techniques. V Maliby,             n - 430
        J. P. Unwin

103.     Analysis of Polychlorinated Biphenyls in Water and Stack Emissions by High Resolution Gas       n - 4SO
        Chromatography / High Resolution Mass Spectrometry. E. A. Marti, J. Amin, H. S. Karam,
        T. J. Yagley, A. F. Weston

104.     Continuous Analysis of VOCs in Air Using a New, Phenyl-Methyl Silicone Stationary Phase        n - 465
        for High-Resolution Capillary GC. R. P. M. Dooper, N. Vonk, H. J. Th. Bloemen
GENERAL
105.     Developing a Uniform Approach for Complying with EPA Methods. J. Parr, P. Sleevi,             n - 479
        D. Loring, N. Rothman

106.     Performing TCLP Analyses to Get Meaningful Data. K Dolbow, J. Price                       n - 490

107.     Total Cyanide By Photolysis. J. Gutierrez                                                H - 492

108.     Ammonia and Total Kjeldahl Nitrogen Determinations Using Flow Injection Analysis with         n - 493
        Gas Diffusion. J. P. Calvi, B. P. Bubnis, J-A. Persson

109.     An Objective Criterion for Terminating Permeability Tests. M. S. Meyers                       n - 508

110.     Sampling and Analysis Plans to Evaluate the Performance of Lead-Based Paint Abatement         n - 521
        B. S. Lint, J. Schwemberger, R. Cramer, B. Buxton, S. Rust, G. Dewalt, B. Lordo, J. McHugh

111.     Further Evaluation of the Cage Modification to the TCLP. P. White                            n - 532

112.     Comparative Study of EPA TCLP and California W.E.T. for Metals in Different Matrices.           H - 533
        G. S. Sivia, M. S. Iskander,  J. T. Coons

AUTHOR INDEX

-------
ORGANICS

-------
44      Methods for the Determination of Volatile Organic Compounds in Soil Samples
         Submitted by:
         Thomas A. Bellar, James W. Eichelberger
         Organic Chemistry Branch, Chemistry Research Division
         Environmental Monitoring Systems Laboratory
         Office of Research and Development
         U.S. Environmental Protection Agency
         26 West Martin Luther King Drive
         Cincinnati, Ohio 45268-1564
         513-569-7512, FTS 684-7512
         Grace Plemmons-Ruesink
         Technology Applications, Incorporated
              Methodology for the determination of volatile organic compounds (VOCs) in
         soil and some other solid matrices has traditionally been fraught with
         problems.  Sample integrity is jeopardized when samples are manipulated to
         introduce internal standards or surrogates, or when the sample is exposed to
         the atmosphere while being transferred to the extraction device.  There has
         also been a problem with the incomplete extraction of the VOCs from the solid
         matrices.  Recently, instrument manufacturers have developed analytical
         equipment specifically designed to efficiently extract VOCs from a variety of
         solid matrices, while preserving the integrity of the original sample.
         Evaluations of two such units, the Dynatech PTA-30 W/S and the Tekmar Model
         7000 Equilibrium Headspace Analyzer, are described for the determination of a
         broad spectrum of organic compounds contained in several soil  types.  For
         comparison purposes, similar analyses were performed with both systems
         according to Method 8260.  Problems such as excessive amounts  of water vapor
         interfering with the reproducibility of the gas chromatographic retention
         times are addressed.  For both evaluations, several types of matrices were
         fortified and analyzed.  The same gas chromatograph equipped with a wide-bore
         capillary column, and the same ion trap detector were used for separation and
         measurement in both studies.  The features of each instrument, accuracies,
         precisions, and method detection limits are discussed for representative VOCs.

                                           M-1

-------
45        CONCENTRATION OF WATER SOLUBLE VOLATILE ORGANIC COMPOUNDS FROM
                     AQUEOUS SAMPLES BY AZEOTROPIC MICRODISTILLATION

                        Mark L. Bruce, Richard P. Lee and Marvin W. Stephens

                               Wadsworth/ALERT Laboratories, Inc.
                                      4101 Shuffel Dr. N.W.
                                    North Canton, Ohio 44720

         ABSTRACT

         Methanol and other similar volatile organic compounds in zero headspace extracts and
         other aqueous matrices can be analyzed by azeotropic microdistillation, followed by gas
         chromatographic separation  and detection.  The method detection limits for methanol,
         1-butanol and 2-methyl-l-propanol are at least an order of magnitude below the current
         Land Disposal treatment standards using the Toxicity Characteristic Leaching Procedure
         (TCLP).

         A microdistillation system was developed to address the limitations of direct sample
         injection, purge-and-trap  and other azeotropic distillation systems.  Sample volume
         requirements range from 10 to 40 ml.  The concentration factors range from 90 to 250
         (depending  on  the analyte) with a 40 ml sample.  The total distillation time is
         approximately five minutes.  Typical detection limits are between 5 and 15 ng/1 when the
         distillate is analyzed by gas chromatography with flame ionization detection.

         Aliquots of zero headspace extraction fluid and ground water were spiked with methanol,
         1-propanol,  2-methyl-l-propanol,  1-butanol, 1,4-dioxane, acetonitrile, propionitrile,
         acrolein, acrylonitrile and ethyl acetate at 0.10 mg/1 and 0.75 mg/1.  Each aliquot was
         distilled and analyzed in duplicate during a 10-day period. Accuracy and precision were
         determined.   System bias for most compounds was less than 15% (i.e., the average
         percent recovery was between  85-115%).  The relative standard deviation for percent
         recovery for most compounds was also less than  15%. The microdistillation was most
         effective for the alcohols.

         INTRODUCTION

         The Hazardous  and Solid  Waste Amendments of 1984 amended RCRA by banning all
         land disposal of untreated hazardous waste  within 5*/2 years  after passage on
         May 8, 1990.  The basic purpose of the  land disposal restrictions  is to discourage
         activities that involve placing untreated wastes in or on the land when a better treatment
         or destruction alternative exists. Under the land disposal restrictions  (40  CFR part
         268.41) for spent  solvents,  methanol has a treatment standard  of 0.25 mg/1 for
         wastewaters  containing spent solvents and 0.75 mg/1 for all other spent solvent wastes in
         the waste extract using zero headspace extraction (ZHE). To date there are no EPA-
         approved methods for methanol that have detection  limits  below these treatment
         standards. The effect of this situation is that residues from the treatment of solvent
         wastes and multi-source leachate wastewaters cannot presently be certified to meet the
         corresponding treatment standards and thus cannot be landfilled.

         This paper presents the development of an aqueous sample concentration, cleanup and
         analysis method with a detection limit lower than the spent solvent treatment standards
                                              1-2

-------
for methanol.  The total sample handling  time  from the start  of distillation to the
completion of analysis is less than one-half  hour. The initial experimental parameters
were derived from a method for the azeotropic distillation of water soluble volatile
organic compounds  (1,2).   This method is based on the fractional distillation of
compounds which form azeotropes with water.* When distilling a 40 ml aqueous sample,
or ZHE extract, total distillation time, including warm-up, is five minutes. GC run time is
approximately 17 minutes. The distillate is free from nonvolatile  organic and inorganic
interferences. These nonvolatile components may degrade gas  chromatographic
performance and shorten the life of the GC column.

^INSTRUMENTATION, EQUIPMENT AND SUPPLIES

Gas Chromatograph/Data System
    Hewlett Packard  5890 equipped with a  flame ionization detector,  Macintosh Ilci
     (Apple) with Lab View (National Instruments) and  GC Integrator & Workmate
     (WillStein) software.
Gas Chromatography Columns
    Quantitation: DB-Wax, 30 m X 0.53 mm I.D., 1.0 micron film thickness
    Confirmation: DB-1, 30 m X 0.53 mm I.D., 1.5 micron film thickness
Hardware
  Wadsworth MicroVOC^ Systemฎ, Shamrock Glass (see Figure 5.)
    Round bottom flask,  100 ml, 14/20 joint
    Fractionation column, 14/20 joint, 1.6 cm O.D., 1.3 cm I.D., 60 cm in length,
         Shamrock Glass (see Figure 3.)
    Pipe insulation, polyurethane foam, lV2"  O.D., 5/s" I.D., 55 cm in length
    Glass beads, 5 mm O.D.
    Keck clamps, for  14/20 ground glass joint, Shamrock Glass
    Glass reducing union, 14/20 ground glass joint to 6 mm O.D. tube, Shamrock Glass
         (see Figure 4.)
    Stainless steel reducing union, Vi6" to l/4"
    Air condenser, Teflonฎ tubing, Vie" O.D., 1/32" I-D. (40 cm in length)
    GC autosampler vials
    Autosampler vial inserts,  100 |jl, calibrated
  Graduated cylinder, 50 ml
  Support stand with rod, 1 meter
  Three-finger clamp
  Heating mantle, Glas-Col, 115 volts, 230 watts, STM 400
  Temperature controller, Glas-Col PL115-Cordtrol,  115 volts, 600 watts
  Porous carbon boiling chips, VWR cat # 26397-409
Reagents and Standards
  Ethanol, Everpure, 200 proof
  Methanol, B&J Brand, purity 99.9%
   1-Propanol, Baxter, purity 99%
  2-Methyl-l-propanol, Aldrich, purity 99.9%
   1-Butanol, Aldrich, purity 99.8%
   1,4 Dioxane, Aldrich
  Acetonitrile, Aldrich, purity 99.9%

* Note:  Methanol does not form an azeotrope with water; nevertheless it can be
effectively distilled with this method.
                                       1-3

-------
  Propionitrile, Aldrich, purity 99%
  Acrolein, Aldrich, purity 97%
  Acrylonitrile, JT Baker, purity 99%
  Ethyl Acetate, Aldrich, purity 99%
  Reagent water, deionized
  Zero headspace extraction fluid: Refer to Method 131 1 of the Federal Register Vol 55
      No. 126, Friday, June 29, 1990, pg 26986-26998
The goal was to develop a sample preparation/introduction system which when combined
with GC-FID analysis would provide methanol method detection limits (MDL) below
0.1 mg/1, use less than 100 ml of sample and require less than 30 minutes of sample
preparation.   Reaching  the   MDL goal would require a  concentration factor of
approximately 30.  Concentration factor is the ratio of the analyte concentration in the
collected distillate fraction to that in the original sample.

Many physical parameters were investigated, such as the sample volume, boil/reflux rate,
total  distillation  time and volume of distillate  collected.   The  physical design
characteristics of the  distillation  system  itself were  investigated.    Several
distillation/condenser designs were used: a commercial modified Nielson-Kryger and two
miniaturized Nielson-Kryger (Peters) systems.  Several alternate overflow systems were
studied: the straight tube, notched, flared, wick and hoop systems (3).  In addition, a
completely redesigned capillary condenser was developed.  The capillary condenser
system was later refined into a more rugged form, the Wadsworth Micro VOC3. Two
chemical parameters were also studied: analyte concentration and matrix.  Table 1 lists
the parameters that were studied.

                  Table 1.   Distillation Parameters Investigated
       Physical
          sample volume
          boil/reflux rate
          distillation time
          distillate volume collected
10 to 1000 ml
2 to 7 ml/min
5 to 120 minutes
2 ill to 20 ml
      Physical design
         Fractionation column

         Distillation system design
          collection chamber volume
          condenser height/cooling surfaces
          overflow design

          overflow tube inside diameter
          overflow tube height
          capillary condenser
          Wadsworth MicroVOC3
Vigreux, glass bead, sand, glass wool,
Rashig ring, spinning band

1 to 20 ml
15 to 60 cm, cooling coil, baffles
Peters/Dow, straight, notched, flared,
side drain, wick, hoop
2 to 10 mm
2 to 35 mm
      Chemical
          analyte concentration
          matrix
0.025 to 10 mg/1
ground water, ZHE extract
                                      11-4

-------
Modified Nielson - Kryger condenser

Initial experiments employed a commercially available modified Nielson - Kryger (N-K)
condenser from Ace Glass (3).  Its design was similar to that described  by Peters (2)
except with larger dimensions and sample removal through a stopcock was used rather
than a syringe (Figure 1). The N-K collection chamber volume was larger; 20 ml vs 1 ml.
Factorial design experiments* indicated that 70% recovery and an estimated detection
limit in the mid ppb range could be obtained for methanol with a distillation time of one
hour using a one  liter sample aliquot. Azeotropic distillation appeared to be the right
process.   However, this large  scale system  was not practical because of the long
distillation time and large  sample volume requirement.  Many miniaturized condenser
overflow systems  were investigated (3).  Most miniaturized systems were more practical
than the modified Nielson-Kryger system, but none produced a  concentration factor
greater than 20.   Examination of fundamental distillation principles led to a radical
change in condenser design.
         t
        E
        E
        o
        c
         t
                                     -Cooling Water

                                     Condensed Steam'
                                         m Chamber
                                         (20ml)
                                     Overflow Tube
                                       (4 mm ID)
                                     Overflowing Water
                  \
Vigreux Column
 2 L Flask
  Heating Mantle
          Figure 1.   Modified Nielson-Kryger Condenser Distillation System

ANALYSIS

EPA SW-846 Method 8015 (modified) was used for analyzing the concentrated aqueous
samples. The analytical conditions are summarized in Table 2.

The instrument detection limit (IDL) was calculated to be 0.15 ng using 10 2 |il injections
of a 0.10 mg/1 standard.  IDL = (tn-i,99%)(Std Dev) = (2.821)(0.0528) = 0.15 ng. A
special note of caution regarding the GC temperature program is in order.  Even though
methanol and most other analytes elute relatively early, the GC column temperature must
be ramped high enough and held  long enough to remove all water from the capillary
column. Retention time shifts may result if the water is not eluted from the  column.


* Note: Factorial design is a statistical procedure which facilitates optimization of several
parameters at the same time.  Precision estimates can also be obtained.
                                       1-5

-------
Since methanol is a common laboratory solvent it is difficult to obtain methanol-free
water.  One deionized water system contaminated reagent water with methanol.  Also,
airborne methanol can be absorbed by water in open containers.

                         Table 2.   Analysis Parameters
quantitation column
confirmation column
instrument calibration range
response factor %RSD
continuing calibration
   response factor %D
injection volume
injection type
injection port temperature
temperature program
                           DB-Wax
                           DB-1
                           0.2 to 2000 ng
                           2 |il
                           splitless
                           180ฐC
external & internal standardization
earner gas
carrier gas flow
detector
detector temperature
hydrogen flow
airflow
make-up gas
make-up gas flow
helium
2.5 ml/min.
FID
230ฐC
37 ml/min.
426 ml/min.
nitrogen
30 ml/min.
                           30ฐC for 5 min., 5ฐC/min. to 70ฐC, 20ฐC/min to 150ฐC
PROTOTYPE VOC3

Previous N-K distillation systems had not
met the goals described above.  A  new
condenser design improved  both the
concentration factor and the simplicity of
the distillation  system.  The capillary
condenser, an early prototype of the VOC^,
is  shown  in Figure 2.   The fractionation
column and condenser were very simple
and inexpensive to  make.  The system
consisted  of a sample flask, fractionation
column packed with glass beads (35 cm
length),  capillary column (0.53 mm I.D.
and 35 cm length) and microcollection vial.

The capillary  tube was normally water-
cooled. The first 10 to 100 ^1 of distillate
were collected in  the micro vial.  When
100 p.1  of distillate  were  collected  a
concentration factor of 80 was achieved in a
7-8 minute distillation.  The methanol
absolute recovery was 20%.  The method
detection limit of methanol in reagent water
was 0.018 mg/1.
                                                              t
                                                        Air or Water Cooled
                                                        Capillary Column
                                                                   Micro vial
                                                     Glass Bead Fractionation Column
                                                          Flask
                                                 Figure 2. Capillary Condenser
Various types of fractionation  columns
were studied.  Glass  and Teflonฎ  tubes
were packed with sand, glass wool,  Rashig rings and glass beads.   A spinning band
fractionation column was also studied. Small increases in distillation efficiency (relative
to glass beads) were found with some fractionation column types, but the columns were
either more difficult to clean or mechanically complex. Thus, the glass bead fractionation
                                      11-6

-------
column was  chosen as the best compromise between ease of use and distillation
efficiency.

WADSWORTH MTCROVOC3

The Wadsworth Micro VOC^ is a rugged version of the capillary condenser system
constructed from standard glass, stainless steel and Teflonฎ components. VOC^ is an
acronym for Volatile Organic Compound Concentration and Cleanup. The glass bead
fractionation column is constructed from glass tubing with standard 14/20 ground glass
joints (Figure 3).  The air condenser consists of three parts (Figure 4): a custom glass
reducing union which converts from the ground glass joint to 1/4" glass tube, a stainless
steel reducing union which joins the V4" glass tube to a Teflonฎ tube and a Teflonฎ tube
(Vie" O.D., Vsz" I-D.) which was substituted for the 0.53 mm capillary column used in
the prototype. The complete system is shown in Figure 5. The total system cost is about
$300 with glassware comprising less than $70.

This microdistillation system more effectively concentrates methanol (and other alcohols)
than the previous prototypes. The concentration factors range from 100 to 250 depending
on  analyte.  The absolute analyte recoveries range from 20% to 60% (Table 3).  The
microdistillation  system is more effective than  purge-and-trap or other azeotropic
distillation systems even though the absolute percent recovery is significantly less than
100%.  Relative  recoveries,  calculated by using standards  which are  also distilled,
average 99%.

                    Table 3.   Analyte Concentration and Recovery
Analyte
Methanol
1-Propanol
2-Methyl- 1 -propanol
1-Butanol
1,4 Dioxane
Acetonitrile
Propionitrile
Acrolein
Acrylonitrile
Ethyl acetate
CAS number
67-56-1
71-23-8
78-83-1
104-51-8
123-91-1
75-05-8
107-12-0
107-02-8
107-13-1
141-78-6
Typical
Concentration
Factor*
140
240
250
250
150
200
200
100
100
100
Typical
Absolute
Recovery*
35%
60%
63%
63%
38%
50%
50%
20%
20%
20%
Average
Relative
Recovery
100%
92%
86%
89%
100%
101%
96%
99%
116%
114%
    * When a 40 ml sample aliquot is used and the first 100 ul of distillate are collected.

Method Summary

The azeotropic microdistillation method is summarized in Figure 6.  A 40 ml aliquot of
sample is transferred to a round bottom flask. Boiling chips and internal standard(s) are
added to the  sample.   Matrix spike compounds are  added when appropriate.  The
distillation apparatus is  assembled using Keck clamps at both ground glass joints after
insuring that the fractionation column and air condenser are completely dry. The sample
                                        1-7

-------
                                      14/20 Ground Glass Joint
                                      Glass beads
                                      Indentations
                                       14/20 Ground Glass Joint
                   Figure 3. Fractionation column
      Stainless Steel
      Reducing Union
                                        OD Teflonฎ Tube

1



1
T'oe
Glass Reducing Union
                                       6 mm OD Tube
                                         14/20 Ground Glass Joint
                      Figure 4.   Air condenser
                             I-8

-------
                            Collection Vial
\
         Reducing Unions
Keck Clamp
     Fractionation Column
     Insulation
       Keck Clamp
          100 ml flask
              Heating Mantle
       Figure 5.   Micro distillation system
                  I-9

-------
                       Measure sample aliquot
                             I
                       Transfer sample to flask
                             I
                    Add I.S., spikes and boiling chips
                             I
                      Assemble distillation system
is heated to the boiling point (2-3 minute warm-up) and held at
a boil for 2 minutes. The first 100 |il of distillate are collected
in a microvial for analysis by GC-FID.   All  calibration
standards are distilled in the same manner as  samples to
compensate for system bias since the absolute recoveries of
analytes are typically 50%.  This calibration procedure is
analogous to purge-and-trap calibration procedures.

TECHNIQUE COMPARISONS

Four sample introduction/preparation techniques were
compared to this microdistillation for low molecular weight
alcohols such as methanol (Table 4).  Direct sample injection
did not provide adequate analyte detection limits because there
was no concentration step.  In addition, direct sample injection
deposited  nonvolatile  sample   constituents   in  the
chromatographic system, which degraded performance. This
was particularly true for zero headspace extracts.  Purge-and-
trap  sample introduction  did not meet the detection limit
requirements because the analytes were very water soluble and
thus difficult to purge.  Absolute analyte recovery  was very
low, (typically <1%) and highly variable.

Two modified Nielson-Kryger (N-K) azeotropic  distillation        p.    (-
systems have been used.  A large scale N-K system (5) did    M A*?^
provide adequate analyte detection limits but required one liter    Metnoa summary
of sample and a one hour distillation.  The one liter sample requirement was problematic
since ZHE extraction produced only a few hundred milliliters.  Also, the precision of
methanol recovery was poor (40% RSD). A small scale N-K  system (3) did not meet the
detection limit (concentration factor) requirement.  However, only a small sample aliquot
was required and the distillation time was short relative to the large scale N-K system.

The microdistillation system presented in this paper has the highest actual concentration
factor and lowest analyte detection limits of these five sample introduction/preparation
techniques.  Sample volume requirements and equipment cost are low  and preparation
time is short.

METHOD VALIDATION

A method validation study following the guidelines specified in the EPA Test Method
Equivalency Petitions guidance manual (4) was performed.   A data summary of the
aqueous matrix study for samples spiked at two concentration levels is presented below.

Two sample matrices were studied: ground water and ZHE extraction fluid.  The ground
water was taken from a  residential drinking water well.   It was  high in  calcium,
magnesium and iron content. The ZHE extraction  fluid was prepared from reagents with
low methanol content.  Appropriate amounts of each matrix were spiked with each of the
compounds  listed in Table 5.  The spiking concentrations were 0.10 mg/1  for the low
concentration spike and 0.75 mg/1  for the high  concentration spike.   Both matrices
contained low concentrations of methanol.   Unspiked aliquots of each matrix  were
11-10

-------
processed and analyzed to allow the percent recoveries to be corrected for the "native"
analyte concentrations.

Each spiked matrix was  subsequently shaken  briefly (with minimal headspace) to
homogenize it.  Each spiked and un spiked matrix was divided into sample aliquots and
stored in glass 40 ml VOA bottles with Teflonฎ  lined caps at 4ฐC with zero headspace.
Each day for 10 consecutive working days each matrix was distilled six times:   two
unspiked samples, two low concentration spikes and two high concentration spikes. A
total of 12 "samples" were distilled each day. All calibration standards were distilled in
the same manner as the samples to automatically compensate for system bias.  An internal
standard (ethanol) was used to improve precision. Analysis of Variance (ANOVA) was
used to estimate method accuracy (bias) and precision.

                          Table 4.   Technique Comparison
Method* Theoretical
Concentration
Factor
Direct sample
injection
Purge-and-trap
Nielson-
Kryger(5)
Nielson-
Kryger(3)
Wadsworth
MicroVOC3
1

2500**
350

8

400

Absolute
%
Recovery
100

1
40

100

50

. . Method Sample „ ,
Actual Detection Preparation ^lQ
Concentration LimU Time Volume
Factor mg/1 minutes ml
1

25
150

8

200

2

0.1
0.05t

0.3

0.01

0

10
60

10

5

0.002

5
1000

40

40

Notes and Equations:

*   Assume a 2 ul injection into the GC-FID for comparison purposes.
**  The purge-and-trap TCP assumes a 2 ul final sample volume to be consistent with the injection volumes
    used by the other techniques.
t   The methanol recovery precision is low so the method detection limit is not improved as much as expected
    based on the actual concentration factor.

„,    ,. .      .	,.   f    /-V-,T*\      original sample volume
Theoretical concentration factor (TCP) = ———&	—i—-	-	
                                 final prepared sample volume

Absolute % Recovery = amount of analyte in prepared "sample"  .
                     amount of analyte in original sample

Actual Concentration Factor = TCF • (Absolute %Recovery / 100)
                                   Direct Inject DL
Method Detection Limit (estimated) =
                              Actual Concentration Factor
' Sample prep precision
The method detection limits  (MDL) for both matrices  are shown  in Table 5.  The
detection limit was calculated from two different data sets. The one-day detection limit
was derived from seven replicate analyses performed on the  same day.  The 10-day
                                        11-11

-------
detection limit was derived from the equivalency study data and consisted of 20 replicates
spread over 10 days.  The one-day MDL is often much lower than the 10-day MDL. This
is expected since day-to-day  reproducibility is usually not as good as same-day
reproducibility. In addition, the distillates from the one-day ground water detection limit
study were analyzed on a less sensitive GC. The detection limits remained essentially
unchanged. This indicates that in this study the precision of the distillation is the limiting
factor for method detection limits.  Thus, using a less sensitive detector will not
necessarily raise the method detection limit. Regardless of the GC used for analysis, the
methanol, 2-methyl-l-propanol and  1-butanol method detection  limits are well below
current land disposal treatment standards.

                            Table 5.  Target Analytes
Analyte
Methanol
1-Propanol
2-Methyl- 1 -propanol
1-Butanol
1,4 Dioxane
Acetonitrile
Propionitrile
Acrolein
Acrylonitrile
Ethyl acetate
Method Detection Limit* (mg/1)
Ground Water THE Fluid
Iday1 Iday2 10 day1 1 day1 10 day1
0.008
0.007
0.005
0.002
0.007
0.004
0.002
0.012
0.010
0.011
0.014
0.005
0.007
0.009
0.012
0.005
0.005
0.021
0.020
0.021
0.017
0.029
0.018
0.026
0.022
0.037
0.082
0.10
0.11
0.008
0.018
0.004
0.004
0.018
0.030
0.011
0.019
0.014
0.015
0.028
0.024
0.029
0.027
0.042
0.027
0.029
0.080
0.092
0.089
       * Microdistillation with modified 8015 analysis.
       1 GC number 1, nominal instrument detection limit 0.
       2 GC number 2, nominal instrument detection limit 0.
lmg/1
5tol.0mg/l
The method may be  extended to 2-butanone,  2-propanol and  acetone,  but  these
compounds were not included in this study.

The results of the equivalency study are summarized in Tables 6 and 7. No outlying data
points were found in any of the data sets.  The day effect was significant for  some data
subsets.  Day effect is statistically significant when the precision within days is much
better than the precision between days.  This is a normal situation for  analytical
procedures.  The bias column in Tables 6 and 7 shows the 95% confidence interval of
analyte fraction recovered.  A value of 1 corresponds  to 100% recovery. The lower
bound for precision is the lower limit of the 95% confidence interval of the true variance
(of analyte recovery). The EPA has used 0.25 as an example maximum (4).

Figures 7 and 8 graphically present the bias data of Tables 6 and 7. The 95% confidence
intervals (CI) of analyte percent recovered are plotted for both low and high spike levels.
Most 95% CIs are small and near 100% recovery.  This indicates that both accuracy
(bias) and precision are good.

Overall the method is very effective for concentration and cleanup of the two aqueous
matrices studied. The alcohols exhibited excellent accuracy (bias) and precision. The
nitrile results were also quite good.  The method is not as effective  for acrolein,
                                       1-12

-------
   acrylonitrile and ethyl acetate although it may be adequate for some uses.  Ethanol was
   not a good internal standard for these three compounds.  A more appropriate internal
   standard may solve most of the precision problems associated with these compounds.

   Table 6. EPA Equivalency Study-Analysis of Variance (ANOVA) Results for THE Fluid
Analyte
Methanol
1-Propanol
2-Methyl- 1 -propanol
1-Butanol
1,4 Dioxane
Acetonitrile
Propionitrile
Acrolein
Acrylonitrile
Ethyl acetate
Low Concentration
Outliers Day Bias
effect
No
No
No
No
No
No
No
No
No
No
No
Yes
No
Yes
No
Yes
No
Yes
No
Yes
1.02-1.12
0.92-1.04
0.81-0.92
0.82-0.95
0.95-1.10
1.04-1.17
0.94-1.05
0.89-1.28
1.06-1.40
0.97-1.41
Lower
bound for
Precision
0.007
0.006
0.008
0.007
0.017
0.007
0.008
0.059
0.082
0.072
High Concentration
Outliers Day Bias
effect
No
No
No
No
No
No
No
No
No
No
Yes
No
No
No
Yes
No
No
No
No
No
0.97-1.06
0.87-0.92
0.77-0.87
0.83-0.90
0.97-1.06
0.90-1.00
0.83-0.97
0.70-1.03
0.84-1.21
0.87-1.28
Lower
bound for
Precision
0.003
0.002
0.007
0.003
0.003
0.007
0.015
0.080
0.096
0.123
Table 7.  EPA Equivalency Study-Analysis of Variance (ANOVA) Results for Ground Water
Analyte
Methanol
1 -Propanol
2-Methyl- 1 -propanol
1-Butanol
1,4 Dioxane
Acetonitrile
Propionitrile
Acrolein
Acrylonitrile
Ethyl acetate
Low Concentration
Outliers Day Bias
effect
No
No
No
No
No
No
No
No
No
No
No
Yes
Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
0.90-0.98
0.87-0.95
0.80-0.94
0.82-0.91
0.92-1.01
0.99-1.10
0.93-1.12
1.00-1.39
1.18-1.68
1.07-1.63
Lower
bound for
Precision
0.004
0.003
0.008
0.003
0.006
0.005
0.013
0.063
0.098
0.117
High Concentration
Outliers Day Bias
effect
No
No
No
No
No
No
No
No
No
No
No
No
Yes
Yes
Yes
Yes
Yes
No
Yes
Yes
0.91-0.98
0.88-0.94
0.82-0.96
0.85-0.95
0.94-1.04
0.87-0.97
0.81-0.98
0.71-0.89
0.82-1.06
0.77-1.08
Lower
bound for
Precision
0.003
0.003
0.008
0.004
0.004
0.004
0.010
0.015
0.024
0.032
   LIMITING FACTORS

   The method is most effective for water soluble compounds having a boiling point low
   enough that the distillate is enriched in the target compounds relative to the original
   sample.  The precision with which the distillate is collected significantly affects overall
   method precision.  The distillation rate also affects method performance.  System and
   reagent  contamination must be kept  to  a minimum.  The type and condition of
   fractionation  column affect the recovery of the target analytes.  Although the
   microdistillation removes many nonvolatile and semivolatile interferences, it does not
   remove interferences from nontarget water soluble volatile organic compounds. Absolute
   analyte recovery ranges from 20 to 65%.  Although the recovery is significantly less than
   100%, the bias is consistent and the results can be corrected to account for this limitation
                                        11-13

-------
using  internal standards and  calibration procedures similar to  the  purge-and-trap
technique.
   140  --
   130  --
T3
2 120
CD

I 110
CD
tr
+ฑ 100
CD
y
    90 --
    80 --
    70
                           0  0.10 mg/1 spike concentration
                           ฐ  0.75 mg/l spike concentration
                                            I
                                                               ^	o—^
         Methaflol   1-Propanol  2-Methyl-1-   1-Butand  1,4Dbxane   Acetonitnle  Propionitrile    Acrolan    Acrytonitrile   Ethyl acetate
                         propand
  Figure 7  EPA Equivalency Study-Bias Results for Zero Headspace Extraction Fluid
    170  '
    150  -
 CD

 I  13ฐ
 CD
 EC
    100
     90  -
     70
                          o  0.10 mg/l spike concentration
                          D  0.75 mg/l spike concentration
                                                                                    o1?
          Methanol    1-Propanol  2-Methyl-1-   1-Butanol  1,4Dioxane   Acetonitrile  Propionitrile    Aaolein   Acrylonitrile   Ethyl acetate
                          propanol

           Figure 8  EPA Equivalency Study-Bias Results for Ground Water

The target analyte must be sufficiently volatile to be distilled from the aqueous sample.
Significant enrichment of the analyte in the distillate (relative to the original sample) only
occurs when the vapors released from the boiling water have a higher analyte to water
ratio than the original sample. This happens when the analyte forms an azeotrope with
water which is > 50% analyte.  If the azeotrope is <  50% analyte no enrichment of the
vapors will take place in the  fractionation column. Some low boiling analytes such as
                                         11-14

-------
methanol do not form an azeotrope with water but still are effectively concentrated by
this system. In general this method is most effective for compounds that have boiling
points below that of water.  However, some butanols have boiling points higher than
water but form azeotropes that boil at less than 100ฐC.  This method appears to work for
such compounds. It does not  work for compounds such as 2-ethoxyethanol which form
an azeotrope that is predominantly water.

The precision with which the distillate is collected significantly affects method precision.
We recommend collecting the FIRST  100 jj.1 of distillate. This seems to be a reasonable
compromise between maximum concentration factor and ease of handling.  The first few
\i\ will contain the highest concentration of analyte; however, it is very difficult to
manually collect this small fraction in a reproducible manner. Larger volumes such as
1 ml can be collected. However, this  significantly reduces the method concentration
factor. The volume collected should be  100 ฑ 20 fj.1. An internal standard (added prior to
distillation) should be used to help compensate for these small variations in volume in the
same manner that an internal standard compensates for purge-and-trap, chromatographic
and detection variations.

The  distillation rate  also affects method performance.   If the distillation rate is
significantly higher than 2  ml/minute the fractionation column may not function
efficiently. The analyte enrichment in  the distillate may be reduced.  If the distillation
rate is too slow the distillate will not reach the air condenser or the distillation may take
too much time.

System and reagent contamination must be kept to a minimum. The specific maximum
contaminant concentration varies according to the quantitation limit required. Methanol
and acetone are common contaminants in deionized water, reagents and laboratory air. If
either of these compounds are target analytes, special laboratory practices  may be
necessary.  Some water deionizers actually increase the amount of methanol and other
potential target compounds in  the laboratory water system. High purity reagents may also
be necessary, particularly in the preparation of ZHE extraction fluid.

The type and condition of fractionation  column affect the recovery of the target analytes.
For best reproducibility and efficiency, the fractionation column, reducing unions and air
condenser must be completely dry before use.  Only 50 p.1 of water in the condenser can
seriously  reduce the analyte concentration in the distillate.  Therefore, the entire
distillation apparatus should be oven-dried before use.

Although the microdistillation removes many nonvolatile and semivolatile interferences,
it does not  remove  interferences  from nontarget  water soluble volatile  organic
compounds.  Nonvolatile sample components will not be distilled and thus will not be
introduced into the GC. Most semivolatile components will also be eliminated or greatly
reduced.  This greatly reduces contamination of the injection port and GC column. Many
water soluble volatile organic compounds may be collected in the distillate.  Some of
them may be difficult to resolve chromatographically. For example methanol, 2-butanone
and 2-methyl-2-propanol elute very closely on a polyethylene glycol stationary phase
(J&W DB-Wax). Such interferences may require different GC columns and/or detectors
to resolve.

The method bias due to  low  analyte recoveries can be corrected  by using an  internal
standard and distilling all calibration standards.   This is  similar to purge-and-trap
                                       11-15

-------
procedures except that the microdistillation system is not directly interfaced to the GC at
present.

CONCLUSION

Methanol and other water soluble volatile organic compounds in zero headspace extracts
and other aqueous matrices can be analyzed by azeotropic microdistillation, followed by
gas chromatographic separation and detection.  The method detection limits for methanol,
1-butanol and 2-methyl-l-propanol are much less than the current land disposal treatment
standards.

This microdistillation system (Wadsworth MicroVOC^) addresses the shortcomings of
direct sample injection, purge-and-trap and other azeotropic distillation systems. Small
sample aliquots are required (40 ml).  Analyte concentration factors are about two orders
of magnitude when a 40 ml sample aliquot is used.  The total distillation time is five
minutes.  Typical detection limits are  between 5 and 15 (ig/1  when the distillate is
analyzed by gas chromatography with flame ionization detection.  The cost of the
complete system is less than $300 with glassware comprising less than $70 of the total
cost.
REFERENCES

1) Report to EPA: Measurement of Polar. Water - soluble. Nonpurgeable VQCs in
Aqueous matrices by Azeotropic Distillation - Gas Chromatography / Mass Spectrometry,
Midwest Research Institute, September 30,1989.

2) Peters,  Steam Distillation Apparatus for Concentration of Trace Water Soluble
Organics, Anal. Chem.. 52 (1), pp. 211 - 213,1980.

3) Bruce, M.L., Lee, R.P., Stephens, M.W., A Method for the Concentration and Analysis
of Trace Methanol in Water  by Distillation and Gas Chromatography, Sixth Annual
Waste Testing and Quality Assurance Symposium Proceedings Vol n, p. 93,1990.

4) Test  Method  Equivalency  Petitions. A Guidance Manual.  EPA/530-SW-87-008,
OSWER Policy Directive, No.  9433.00-2, February 1987

5) Cramer, P.H., Wilner, J., Eichelberger, J.W., Azeotropic Distillation-A Continuing
Evaluation for the Determination of Polar, Water - Soluble Organics, Sixth Annual Waste
Testing and Quality Assurance Symposium, July, 1990.
                                      11-16

-------
46              POLLUTION REDUCTION IN THE LABORATORY THROUGH
                     THE  USE  OF SMALLER INITIAL SAMPLE SIZE
                                       by

                           Dante J.  Bencivengo,  Ph.D.
                             Bruce N. Colby, Ph.D.
                             Philip W. Ryan, Ph.D.

                            Pacific  Analytical,  Inc.
                           6349 Paseo Del Lago, #102
                          'Carlsbad,  California  92009
                                  Presented  at

                       SEVENTH ANNUAL WASTE TESTING AND
                          QUALITY ASSURANCE SYMPOSIUM
                                July 8-12, 1991

                                Washington,  DC
                                    1-17

-------
                      1.0   INTRODUCTION

Since  the first of EPA's  wastewater  analysis methods were
put  forth in 1976,  the environmentally sensitive chemical,
methylene chloride,  has  been  the  solvent  specified for
extracting   semivolatile   organic  compounds   (BNA's)  from
aqueous  media.  Present  methods  for  determining  BNA's  in
aqueous  samples  use  about  500 mL  of methylene  chloride
(MeCl2)  for  each 1  L  sample  exclusive of that required for
GPC.  Of  this,  only one mL  is  retained  for  analysis; the
rest  is  either  lost  to  the atmosphere during handling or
disposed  of  by  a waste removal firm.

Because  MeCl2  is  on  most of EPA's  lists of undesirable
chemicals,   reducing  the  amount  required  by  EPA's  own
analytical   methods  seems  highly  desirable.  The   first
attempt  to  do this was  noted  in Method  525,  a  method for
determining  BNA's  in  drinking water  where  analyte removal
from  the sample is accomplished  by adsorption  in  a  "SEP"
cartridge or disk.

Although  SEP  technology  seems   unlikely to  be  directly
applicable to complex samples such as those associated with
industrial discharges or those from  test  wells,  one aspect
of Method 525 is important. This is the fact that the GC/MS
calibration  curve  is  pushed  downward   from  the  typical
10 nG/uL  low point  to 0.1 nG/uL. The  significance  of this
is  that   it  should  be  possible  to use  a  smaller  initial
sample  size,  on the order  of  100  mL,  yet retain  the
existing  1  mL  final  volume  and  still  be well  within the
calibration  range of  the GC/MS equipment.  This should make
it possible  to  reduce the  amount of MeCl2 required by about
a factor  of  ten.

Further,   by  calibrating  to the existing  method's  high
point, the effective dynamic range is increased by an order
of magnitude.  This  should  result in  fewer  sample  extracts
requiring dilution  and reanalysis.  Clearly  this  would save
on  analytical  costs.  Perhaps less  clear  is the  savings
which should result from being able  to predict  analytical
effort more  accurately  due to reducing the uncertainty  in
the time and effort associated with reruns.

There  are  several  minor  benefits associated with  using
100 mL water samples. These  include  reduced  sampling and
shipping costs, smaller, i.e.,  less expensive glassware and
general   improvements  associated   with   improved   space
utilization.  Such  improvements,  while  significant   in  a
conventional  laboratory,   are  much  more  important  in  a
mobile/field laboratory environment.
                             1-18

-------
The purpose  of the activity reported here  was to evaluate
the  possibility  of  using  smaller  initial   sample  size,
100 mL  versus  1000 mL,  for the  analysis  of  semivolatile
organic compounds in aqueous waste samples. For the purpose
of  these  experiments,  a  100  mL continuous   liquid-liquid
extractor was  designed  and fabricated.  It was  then used to
prepare  seven   spiked  clean   water  samples   for  method
detection  limit assessment  and  five spiked field samples
(provided  by ICF Technology Inc.,  Las Vegas)   for analyte
recovery  assessment.  The  quantities  of  solvent used  in
preparing the samples were recorded for comparison to those
used with the standard 1000 mL extractor design.
                     2.0  EXPERIMENTAL

CONTINUOUS LIQUID-LIQUID EXTRACTOR DESIGNS

A set of  100  mL continuous  liquid-liquid extractors (CLLE)
were  fabricated according  to  the design  shown  in Figure
2.1.  To  use them,  25 mL  of methylene chloride  (MeCl2)  is
added  to the  CLLE  and  Another  25  mL,  to  a 50 mL  round
bottom  flask  (RB)  which  is  attached  and  used  as  the
collector. A  100 mL aliquot of water sample  is  then added
to the CLLE.  Spikes are added to the water  at  this point.
The water is then acidified with 2 mL  of six  N sulfuric
acid,  the condenser is placed on top of  the extractor and
the solvent  in the  RB  is heated to  boiling for 18  to 24
hours.
  Figure 2.1 - 100 mL Continuous Liquid-Liquid Extractor.
                              11-19

-------
GC/MS  CONDITIONS

The  instrument used  to  collect the data  in this  study was  a
VG  Trio-1.  This  is  a  current generation instrument which
provides   better  sensitivity   than   older  instruments.
Acquisition  parameters  were equivalent  to  those specified
in   the   CLP  10/89   Low  Concentration   Water   Method   for
Semivolatiles.  The   instrument  was  calibrated for  each
target  analyte  using  a  5-point   linear  (regression)
calibration   curve.   Concentrations  of  the   calibration
solutions ranged from  one  to  100 ng/uL;  1 uL  injections
were used in  all  cases.

A method  detection limit  (MDL)  study was carried out using
seven  clean  water   samples  spiked  to  30  ug/L  with  each
target analyte  with  CLLE design A  and  10  ug/L with CLLE
design B. MDLs  were  calculated from  data acquired  on both
instruments.  Percent recoveries were also  calculated from
the  calculated  concentration  values for the five spiked
field  samples.
                3.0  RESULTS AND DISCUSSION

SOLVENT USAGE

Solvent   usage,   identical  for  both   CLLE   designs,   is
summarized  in  Table  3.1.  The  quantities  of  methylene
chloride  (MeCl2)  were recorded  for both  the CLLE  and RB
charge  volumes  and  the  glassware washing  volumes.  The
washing   volumes   were   included   because  they   are   of
significant magnitude  and because  the washing cycle  is a
true  part of  the analysis.  For comparison  purposes,  the
solvent volumes use  with 1000  mL CLLEs is also included in
Table  3.1.  Also  included in  the  table  is  the  volume of
solvent used to clean up the K-D apparatus.

           TABLE  3.1 - METHYLENE CHLORIDE  USAGE

  Usaqe        	100 mL CLLE	1000 mL CLLE
CLLE Charge
RB Charge
CLLE Cleanup
RB Cleanup
K-D Cleanup

25 mL
25
75
30
100
Total 255 mL
250
250
225
150
100
975 mL
As can be  seen,  the charge volumes used win  the  CLLEs and
the RB collector are directly proportional to  the volumes
                              1-20

-------
of sample extracted. Thus, by going from 1000 mL to 100 mL,
there is a savings of 90 percent in MeCl2 usage.

The  wash volumes  for  the  CLLE  and  RB  however,   are  not
proportional to sample  volume.  This is  because the washing
process must  clean the surface  area  of the glass  and the
ratio between  the  two devices in terms of  surface  area is
on the  order  of  0.3-to-l.  This  correlates well with the
MeCl2 usage of 0.28-to-l.

It is important to note that both CLLE sizes  result  in an
extract which  must be concentrated via  Kuderna-Danish (K-
D) . The wash  volume for the K-D  is remains unchanged with
sample size  because the  K-D  apparatus,  in  particular the
Snyder column, is the same size in both cases.  In principle
it should  be  possible to decrease  the  size of  the Snyder
column  but  in practice  this  may  be  difficult  due  to
fabrication difficulties.

The overall reduction in  MeCl2  achieved by  going to 100 mL
initial  sample  size  and  100  mL CLLEs  is  82.2  percent
excluding  the  K-D. It drops  to a savings  of  73.8  percent
when the K-D is included.

CONTINUOUS LIQUID-LIQUID EXTRACTOR DESIGN

Two significant behavior  characteristics  were  noted during
the extraction process. First there was  a tendency for the
RB  to  go  dry.   This  results   in  target  analytes  being
volatilized and driven up into the solvent vapor return arm
where they condense.  When dry RBs were  noted,  the extracts
were discarded and an additional sample  aliquot prepared,
this  time with  less heat  applied  to  the  RB.  The second
problem resulted when  insufficient heat was applied to the
RB to maintain solvent  condensation in  the  condenser. When
this happened, solvent  condensed  in the  solvent return arm
of the CLLE  and  the sample was  not extracted.  This second
problem was   harder to  monitor than  the first  because it
was  dependant on  laboratory temperature.  This tended to
change due to the  day/night  settings  on  the  thermostat
which result in the lab becoming rather cool on cold winter
nights. When  the  lab temperature drops, the volume of hot
solvent needed to keep solvent condensation taking place in
the condenser and not in the transfer  arm becomes larger.

Both of the above problems could be overcome by using a 100
mL RB  for the collector  and charging  it with  50,  rather
than  25  roL  of MeCl2.  This  however,  would result  in  a
significant increase in solvent usage.
                              11-21

-------
INSTRUMENT CALIBRATION

The  VG Trio-1  was readily  calibrated across  the  one  to
100 ng/uL range. Most of  the polar compounds were detected
in  the 1 ng/uL  injection.   Exceptions  were  Benzoic acid,
Hexachlorocyclopentadiene,  2,4-Dinitrophenol,  4,6-Dinitro-
2-methylphenol and Pentachlorophenol.

METHOD DETECTION LIMITS AND SPIKE RECOVERIES

Method  detection  limits  (MDLs)  were  calculated  for each
analyte based on the  seven  spiked  clean water samples. The
MDLs  were  not  as   low  as  had   been  anticipated  when
calculated  from data acquired from  all  seven  runs. They
ranged from about 20 to 30 ug/L for most analytes. Problems
were most  significant with  the  highly polar  analytes and
the  reactive   analytes.   These   included   Benzoic  acid,
Hexachlorocyclopentadiene,    2,4-Dinitrophenol    and   4-
Chloroanaline.

        TABLE  3.2  - METHOD  DETECTION LIMITS (ug/L)

                              7 Sample Data   4 Sample Data
Phenol
bis ( -2-Chloroethyl ) Ether
2-Chlorophenol
Benzyl Alcohol
2-Methylphenol
bis (2-Chloroisopropyl ) Ether
4 -Methylphenol
N-Nitroso-Di-n-Propylamine
Hexachloroethane
Nitrobenzene
Isophorone
2-Nitrophenol
2 , 4-Dimethylphenol
Benzoic Acid
bis (2-Chloroethoxy) Methane
2 , 4-Dichlorophenol
1,2, 4-Trichlorobenzene
Naphthalene
4-Chloroaniline
Hexachlorobutadiene
4-Chloro-3 -Methylphenol
2-Methylnaphthalene
Hexachlorocyclopentadiene
2,4, 6-Trichlorophenol
2,4, 5-Trichlorophenol
2-Chloronaphthalene
2-Nitroaniline
39.1
29.6
29.9
31.5
29.3
23.0
27.4
16.3
29.3
26.3
23.5
21.8
10.8
nd
35.4
23.1
26.7
26.4
nd
27.8
24.8
23.9
nd
18.6
20.0
22.9
17.6
2.8
2.2
3.0
3.6
4.8
7.2
5.6
17.4
4.6
2.6
1.4
2.0
8.7
nd
3.7
5.4
6.6
2.9
nd
10.0
3.4
3.5
nd
8.7
2.4
2.1
2.7
                              1-22

-------
 Dimethyl  Phthalate                  29.3             5.6
 Acenaphthylene                      29.6             3.3
 2,6-Dinitrotoluene                  16.9             4.8
 3-Nitroaniline                      16.1            10.8
 Acenaphthene                        30.0             2.6
 2,4-Dinitrophenol                    nd             nd
 4-Nitrophenol                       16.8             9.4
 Dibenzofuran                        22.8             4.6
 2,4-Dinitrotoluene                  15.2             6.0
 Diethylphthalate                    28.3             6.4
 4-Chlorophenyl-phenylether          29.0            14.3
 Fluorene                            29.6             4.3
 4-Nitroaniline                      18.7            17.8
 4,6-Dinitro-2-Methylphenol          8.5            11.4
 N-Nitrosodiphenylamine              32.8            25.4
 4-Bromophenyl-phenylether           27.3             5.0
 Hexachlorobenzene                   15.0             6.6
 Pentachlorophenol                   8.9             5.6
 Phenanthrene                        27.8             4.5
 Anthracene                          25.1             4.7
 Di-n-Butylphthalate                 32.9            29.0
 Fluoranthene                        31.6            13.8
 Pyrene                              33.0            12.0
 Butylbenzylphthalate                29.8            18.1
 3,3'-Dichlorobenzidine              14.4            14.3
 Benzo(a)anthracene                  34.0            10.8
 Chrysene                            32.1             7.5
 bis(2-Ethylhexyl)phthalate          37.2            31.4
 Di-n-Octylphthalate                 36.9            31.3
 Benzo(b)fluoranthene                37.6            31.8
 Benzo(k)fluoranthene                46.0            36.3
 Benzo(a)pyrene                      24.0            10.5
 Indeno(l,2,3-cd)pyrene              28.8            15.1
 Dibenz(a,h)anthracene               29.2            18.2
 Benzo(g,h,i)perylene                25.1             8.1
 1,3-Dichlorobenzene                 25.9             5.9
 1,4-Dichlorobenzene                 26.7             4.1
 1,2-Dichlorobenzene           	27 .1	4. 0
                    avg             26.0             9.4

The  higher  than  anticipated  MDLs  are  believed  to result
 from  the operational  problems  with the  extractors  (see
above).    The  reason  for   this   conclusion   is  that  the
recoveries for  the  target  analytes in  three of  the  seven
samples was quite low  (less than  40 percent).  Further,  the
recoveries  were  particularly  low for   the   more  polar
compounds. This suggests that there may have  been problems
with incomplete extraction caused by the low night time lab
temperature.
                             11-23

-------
If it  is  assumed that the three  low  recovery extracts are
the  result  of  a  circumventible  problem  and  MDLs  are
recalculated based on  four,  rather  than seven samples, the
MDL values  are  much  closer to those  anticipated (4 Sample
data in Table 3.2) with an average of 9.4 ug/L.

Results from the  five field  samples spiked  in duplicate
were used to calculate mean percent recoveries (Table 3.3).
The percent  recoveries averaged between 65  and  80 percent
which seems reasonable.

              TABLE  3.3  -  PERCENT RECOVERIES

             Analvte	Ava  %R

             Phenol                      47.8
             bis(-2-Chloroethyl)Ether   71.2
             2-Chlorophenol              54.0
             Benzyl  Alcohol              83.5
             2-Methylphenol              52.5
             bis(2-Chloroisopropyl)Ether 73.1
             4-Methylphenol              51.1
             N-Nitroso-Di-n-Propylamine 73.3
             Hexachloroethane            72.0
             Nitrobenzene                69.8
             Isophorone                  81.3
             2-Nitrophenol              55.2
             2,4-Dimethylphenol          45.8
             Benzoic Acid                54.6
             bis(2-Chloroethoxy)Methane 63.5
             2,4-Dichlorophenol          56.1
             1,2,4-Trichlorobenzene     78.0
             Naphthalene                73.7
             4-Chloroaniline               nd
             Hexachlorobutadiene         73.9
             4-Chloro-3-Methylphenol     62.7
             2-Methylnaphthalene         74.1
             Hexachlorocyclopentadiene      nd
             2,4,6-Trichlorophenol       60.1
             2,4,5-Trichlorophenol       59.8
             2-Chloronaphthalene         76.1
             2-Nitroaniline              92.4
             Dimethyl  Phthalate          83.6
             Acenaphthy1ene              77.7
             2,6-Dinitrotoluene          85.7
             3-Nitroaniline              51.7
             Acenaphthene                73.6
             2,4-Dinitrophenol           79.7
             4-Nitrophenol              70.3
             Dibenzofuran                79.7
             2,4-Dinitrotoluene          86.4
                             11-24

-------
              Diethylphthalate
              4-Chlorophenyl-phenylether
              Fluorene
              4-Nitroaniline
              4,6-Dinitro-2-Methylphenol
              N-Nitrosodiphenylamine
              4-Bromophenyl-phenylether
              Hexachlorobenzene
              Pentachlorophenol
              Phenanthrene
              Anthracene
              Di-n-Butylphthalate
              Fluoranthene
              Pyrene
              Butylbenzylphthalate
              3,3'-Dichlorobenz idine
              Benzo(a)anthracene
              Chrysene
              bis(2-Ethylhexyl)phthalate
              Di-n-Octylphthalate
              Benzo(b)fluoranthene
              Benzo(k)fluoranthene
              Benzo(a)pyrene
              Indeno(1,2,3-cd)pyrene
              Dibenz(a,h)anthracene
              Benzo(g,h,i)perylene
              1,3-Dichlorobenzene
              1,4-Dichlorobenzene
              1,2-Dichlorobenzene
              Nitrobenzene-d5
              2-Fluorobiphenyl
              Terphenyl-dl4
              Phenol-d5
              2-Fluorophenol
              2,4,6-Tribromophenol
              2-Chlorophenol-d4
              1,2-Dichlorobenzene-d4
                                     Avg
                      4.0  CONCLUSIONS

The  experiments   described   are  promising  in  terms  of
reducing  pollution  associated  with  environmental  sample
preparation.  The  technique  of  using  a   smaller  initial
sample size  in  conjunction with calibrating the GC/MS to a
lower concentration  provided a  reduction  of approximately
75 percent in the  amount of  Methylene chloride required to
perform  an  extraction.  The  quality  of  the  analytical
results was  approximately equivalent  to that achieved with
                             11-25

-------
the traditional  1  L sample and extractor but  a new design
is recommended to increase the ruggedness of the ruggedness
of the experimental design.
                             1-26

-------
47        EXTRACTION OF PHENOLIC COMPOUNDS FROM WATER SAMPLES
                    USING STYRENE-DIVINYLBENZENE SPE DISKS

        Craig  G. Markell.  Research  Specialist,  New  Products Department,
        Donald F.  Hagen,  Corporate  Scientist, Corporate Research  Analytical,
        3M,  3M Center  Bldg.  201-1S-26, St. Paul,  Minnesota  55144

        ABSTRACT

        Phenolic compounds, especially the more polar  ones,  can be difficult
        to extract from  water samples  using  solid phase extraction  with CIS
        functional silica  as the  paniculate.   The  cause  of low  recoveries is
        almost  certainly   unfavorable  partitioning  between  the   CIS   and
        water,  resulting in  rather low  breakthrough volumes and  recoveries.
        Our  research  has  shown  that  the  use  of pH adjustment  and heavy
        salting, along with low  sample volumes,  can  help  the situation  by
        altering  the  partitioning,  but  another  solution  is  a  different solid
        phase  particulate.

        Styrene-divinlybenzene  particles   were   incorporated   into  47   mm
        solid phase  extraction disks  and used to extract  a  variety of phenolic
        compounds  from  water  samples.   To  preserve  the high  flow rates
        that  make solid phase  extraction  disks  so  attractive,  small particles
        (3-10  um)  were  used  to  preserve the  fast kinetics  seen with  the
        usual 8 um CIS silica.

        This presentation  will briefly discuss the basics  of extracting  phenols
        from water using  SPE disks, followed by the details and results of our
        research.  The  preliminary conclusion is that the resin  disks do  have
        some  advantages  over CIS  disks  for the extraction of phenols,  and
        perhaps  other polar  compounds, from water samples.

        INTRODUCTION

        One  of  the  more  significant  trends  in  environmental   sample
        preparation  is  the replacement  of  liquid/liquid  extraction  (LLE)  with
        liquid/solid extraction (LSE),  also  called  solid  phase  extraction  (SPE),
        for  concentrating  semi-  and  non-volatiles from  aqueous  samples.
        Although LSE  works  very  well  for   extracting  most analytes  of
        environmental significance, low recoveries  are  expected for the more
                                        11-27

-------
polar,  water-soluble  analytes,  such  as certain  phenols.    This  is
expected,  since  LSE and  LLE use very  similar mechanisms  and low
recoveries of  water-soluble compounds are well known in LLE.

Partition  ratios  between  the  organic phase  and  the aqueous phase
govern  the  percentage  of analyte  extracted  in  both  LLE  and  LSE,
where  it is  convenient to think of the organic portion of the particle
as being  analogous  to the solvent  in LLE.   For  hydrophobic,  water-
insoluble  compounds,  such   as  PAH's,  PCB's,  and  many  other
pollutants,  the  partition  ratio  is  overwhelmingly  in favor  of  the
organic phase,  resulting  in good recoveries  from large volumes  of
water.   On the other hand, polar and water-soluble  compounds have
less  favorable partition  ratios,  resulting  in  relatively  low recoveries.

Often,  the addition  of additives to the  aqueous phase is effective  in
changing  the  partition  ratio,  thus increasing low recoveries.   This
practice  is  well  known  in  LLE  in the  form  of  sodium  chloride
addition, or "salting out."   Another familiar  example is pH  adjustment
to  convert  ionic  analytes  to the   corresponding  neutral  species.
Salting  out  or buffering  can be equally  effective  in  LSE  and is one
approach to increasing LSE recoveries of difficult  compounds.

Other methods of increasing  low recoveries  in  LLE are  to  use larger
volumes  of  extracting  solvents,  different  extracting  solvents,   or
extractions  of  the  same water  sample  with  several  portions   of
organic solvent.   Approximate  LSE  analogs of these techniques  are
respectively  a   higher  mass  of  sorbent,   a  sorbent  with  more
selectivity  for   the  analytes,  and   multiple  sorbent  beds.    An
alternative  to  a  higher  mass  of  sorbent  is  a  smaller  volume  of
sample, which also increases  the  sorbent/sample  ratio.

This paper/presentation  will  explore  the use  of  experimental  solid
phase extraction disks which are similar  to the Empore™  disks used
in Method  525,  but containing  polystyrene/divinylbenzene  (SDVB)  in
place of the CIS  silica.   The premise of this work is that  the  SDVB
disks offer both  a  higher mass  of sorbent  (in terms  of  organic
content)  and  perhaps  more selectivity for aromatic  compounds than
CIS.  These features are expected  to  result in a disk  which will offer
significantly  higher  recoveries  for polar compounds,   such   as
                                 H-28

-------
phenolics, from  water samples.   The analytes  used for  this work  are  a
series  of  phenols  ranging from  relatively  hydrophilic to hydrophobic.

EXPERIMENTAL

The  experimental portion  of  this  work  consisted  of  two  parts:
scouting  to  determine the efficiency  of several  sorbents  at  extracting
phenolics  from  water, and more thorough  triplicate  extraction studies
of the most efficient sorbents identified  by  scouting.   The scouting
was  done by  spiking several phenolic  compounds into  100  ml  of
reagent water and passing the water  through experimental 47  mm  x
0.5 mm  SPE  disks  containing a  variety of sorbents, using a standard
47  mm  filtration apparatus.   The  use of  the  disks  has been  well
documented elsewhere and  won't  be detailed here (1,2).   The  disks
were  then eluted using  acetonitrile  or  acetonitrile followed  by  ethyl
acetate, depending on how  tightly  the analytes were  sorbed to  the
disk.   The  final  determination was done with HPLC,  using  a reverse
phase  system  with UV  detection.   The compounds were:   phenol,  o-
cresol,  2-nitrophenol,  4,6-dinitro-o-cresol, 2,4-dichlorophenol,  2,4,5-
trichlorophenol,  and  2,4,6-trichlorophenol  at  concentrations in  the
water  ranging from 0.5  ppm  to  20  ppm,  depending on  the  extinction
coefficient.   The effect  of  salt addition and pH adjustment  was also
briefly studied  during this  phase.

Recovery  data  were  determined by  spiking the  11  phenols  shown  in
Table  II  into  100,  200,  300,  and 500 ml  of  water and  passing  the
water  through  47 mm   disks,  as in  the  scouting studies.   Samples
were  processed  using a  vacuum of  about 25  inches  Hg,  generated
with an  aspirator.   Sample  flow times ranged  from 0.5  minutes  for
100  ml through  the  SDVB  disk  to 7  minutes for 500 ml  through  the
CIS disk.  The 500  ml  sample took  2.5  minutes to pass  through  the
SDVB disk, which would correspond to 5 min/L.   Each spike was done
in triplicate.   The  approximate  concentration of  each phenol  in  the
water  was 200  ug/L.  Since the  11  phenols weren't well  resolved by
the HPLC, the phenols were tested in two mixtures,  one  with five and
one  with  six  of  the  compounds.   Elution was  done  with 2 x  2 ml
aliquots  of  tetrahydrofuran (THF),  followed by  2 x 2  ml aliquots  of
methanol.   These aliquots  were then  combined  and made  to  10 ml
with methanol for HPLC analysis.   Again, CIS reverse  phase  HPLC
was used with  270 nm  detection  and  a  water:methanol  gradient.
                                  1-29

-------
Each of  the mobile phases  contained  0.1  percent  acetic  acid to
suppress ionization  of the phenols  and improve  peak  shape.   Three
types of disks were  used  -  CIS and cyclohexyl (CH) bonded silica, and
SDVB.

In  both  phases  of  the  study,  calibration  was  single  point,  the
standard  being  at  the  concentration  expected  from  a  100 percent
recovery.   Except for the standard 47  mm CIS  and CH disks,  which
are  commercially  available  (Varian  Sample  Prep  Systems,  Harbor
City, CA),  the  Empore™  disks were  experimental,  each  being
prepared  at  3M  from  the  specific  sorbent  particles  mentioned.
Except  for the SDVB particles used  in  the scouting  phase,  which were
in the 50-100 um size range,  all of the particles  were  5-15  um.  All
disks were 47 mm  x 0.5  mm with  particle loadings of 75-90  percent
by  weight.

RESULTS AND DISCUSSION

The  intent of the scouting work was  to  quickly  test several  sorbents
for their ability to extract  the probe phenolics from  100 'ml  of  water,
then use the  more promising phases for further study.  Standard  CIS
disks were used as  a control, since these  disks are beginning  to  find
wide use in  environmental  laboratories.   CH  bonded  silica  was  also
incorporated  into  scouting  and  the  subsequent  recovery  studies,
since CH has  gained  a  reputation  of  being  effective  for  phenol
extractions.   Besides the  CIS bonded silica and CH  bonded  silica,  two
proprietary bonded silicas were  tried,  plus  a  cyano  bonded silica  and
the SDVB.  The phases showing  the best recoveries were CIS, CH, and
SDVB.   Although CH is  often mentioned  as an effective phase  for the
SPE  of phenolics  from  water,  the  results  failed  to  show  a clear
advantage  over CIS.   These results   are  shown  in  Table  I.    The
compounds  that  presented  problems  with the  extraction  were  the
more polar, water soluble  compounds,  while  the  hydrophobic phenols
were  easily extracted from  100  ml  samples by  most  of  the  sorbents
tried.

As  an extension of  this  work, 25% NaCl  was added to the water  and
the pH  was  lowered to 2 with  HC1.   These  modifications  of the
sample,  done  before extraction,  were   successful  in  raising  the
recoveries  of  several analytes,  also   shown  in Table I.   Sample
                                11-30

-------
modification  steps,  which added little  time or  cost  to  the analysis,
make  it  possible  to use  standard  CIS  disks  for  the  quantitative
extraction  of many  phenolics  of environmental  interest.   The only
test  compound  showing  a  low  recovery  was  phenol, which  has  a
water solubility  of almost 10 g/100 ml  in  water.

In the scouting  work, the SDVB resin  was  clearly  the most effective
for quantitative extractions  of the test compounds  from  100  ml of
water, even  without  modification  of  the  sample.    Because  of the
strong  interaction  between  some of  the  analytes   and   the  resin,
acetonitrile elution  alone  wasn't  strong  enough,  as  evidenced  by
generally  low  recoveries  (not  shown)  of  even  the   hyrophobic
phenols.   To  overcome  this  problem,  the  usual  acetonitrile  elution
was  followed by 2 x  1 ml ethyl  acetate elutions, which were added to
the  acetonitrile.   There  are  undoubtedly  a  number  of   alternative
elution  solvents which would  have been equally effective.

Once the scouting work  had identified  SDVB  as an  effective  sorbent
for phenolic compound  extractions,  a  more rigorous recovery  study
was   undertaken to  confirm  the  scouting  results  and  progressively
increase sample volumes  to define the  limits of this  technique.   CIS
and  CH disks  were again included for  comparison.   The SDVB disks
used  for these results contained sorbent  particles approximately  10
um in size.   For this study, the analyte  list was modified to  contain
the traditional   priority pollutant phenols,  at approximately 200 ug/L
each.  The reagent  water  used  was unmodified in terms of  pH  or salt
content.

Recovery  results  are  shown  in Tables  II, III, and IV,   at  several
sample  volumes,  for CIS,  CH,  and  SDVB  disks,  respectively.
Generally, the  results contain  no  surprises.  In  order  of effectiveness
for  phenol  extractions  from  water,  SDVBปC18>CH,  which  also
parallels  the organic  content of  each sorbent  particle.   As expected,
increasing the  volume of the  samples decreased  the recoveries of
marginally  recovered compounds.     There are  a  few   anomalous
results,  e.g. the 2,4-dinitrophenol  results on the  SDVB  disk   at 300 ml.
This  may be a  reflection  of compounds with pKa's  near the pH of the
matrix, where   a  slight change  in pH  would  result  in  a  substantial
change in  the  percentage of  the ionic   form   of  those  compounds.
                                11-31

-------
Dropping  the  pH  of  the  samples to  2 would  overcome  this effect and
may increase  some of the low recoveries.

While  resin   sorbents   have  occasionally  been   reported   in   the
literature  for  phenolic  extractions,  including one  of  the  pioneering
publications in SPE,  heroic  efforts are often  needed  to  clean  up these
resins  before  use  (3).   Although our work was  conducted well above
method detection  limits, we  saw no evidence  of  interferences from
the SDVB, which would be  expected  to contain UV  chromophores.
The only  cleanup step  used for these disks  was  the  initial wash  step
with a  few ml  of the  eluting  solvents  (1).   Given  the  small particle
diameter and  short  distances  needed for  contaminants to diffuse  into
the  wash  solvent,  the  initial  wash  step  plus   the  methanol
conditioning step  may be sufficient  to remove any  contaminants.   An
independent  researcher,  using  these  disks  prior  to   LC/MS,  also
noticed no interferences  (4).

Conclusions

This work demonstrates the utility  of experimental  SDVB,  SPE disks
as  a  technique  for isolating phenolics from  water.   Even at  500  ml,
quantitative recoveries  were  seen  for all but  a  few  phenols  with
extraction  times  corresponding  to  about  5   min/L.    With   pH
adjustment  and  salting out,  the   low  recoveries  may  have  been
improved.  Using 10 um SDVB particles in  the disks,  no interference
problems   were  encountered,  in contrast  to   literature  reports  of
extensive   soxhlet extractions needed  for  the  much   larger  SDVB
particles used  in  previous  research.
                                 1-32

-------
References

[1]    Hagen, D. R, Markell,  C.  G., Schmitt, G. A., and Blevins, D.  D.,
      Analytica  Chimica Acta. 236, 157-164,  (1990).

[2]    Markell, C. G., Hagen, D. R, and Bunnelle, V. A., LC-GC. 9, 332-
      337,  (1990).

[3]    Junk,  G.  A. et al., Journal  of  Chromatography.  99,  745-762,
      (1974).

[4]    Behymer,  T. D., personal  communication (1991).
                                 1-33

-------
                    TABLE I.  Scouting  Results  for Selected Phenols
                                                    Sorbent
Phenol




o-Cresol




2-Nitrophenol




2-Methyl-4,6-Dinitrophenol




2,4-Dichlorophenol




2,4,5-Trichlorophenol




2,4,6-Trichlorophenol
CIS
4
20
35
14
106
108
110
CH
5
9
23
19
102
109
1 11
SDVB*
90
120
108
96
107
95
96
C18**
23
94
90
94
92
92
92
CH**
14
99
69
94
88
87
89
*  eluted with acetonitrile  followed by ethyl acetate




u* pH = 2, 25% NaCl

-------
CO
01
               TABLE II.  Results Using CIS Disks  - % Recovery (RSD, n=3)




                                                    Volume  (ml)


                                 100             200             3ฃQ             500


Phenol                         12.8  (9.5)         7.8 (4.1)         4.5  (6.7)         2.8  (7.3)


2-Nitrophenol                  63.9  (3.2)        41.2 (3.2)       25.0  (5.9)        15.4  (5.2)


4-Nitrophenol                  29.9  (6.1)        19.1 (6.2)       11.3  (3.8)         6.8  (5.8)


2-Chlorophenol                50.6  (3.9)        30.3 (4.3)       18.2  (5.0)        10.7  (5.2)


2,4-Dinitrophenol               5.9  (17.3)        4.2 (15.2)        8.8  (1.6)         1.6  (5.3)


2,4-Dichlorophenol             67.4  (5.1)        82.9 (10.9)      63.8  (1.1)        47.5  (4.8)


2,4-Dimethylphenol            99.2  (3.0)        89.5 (4.7)       61.8  (7.2)        37.1  (6.7)


4-Chloro-3-Methylphenol      93.0  (7.6).      104.9 (3.9)       89.7  (7.4)        60.5  (7.9)


2,4,6-Trichlorophenol          92.7  (2.5)       111.3 (1.7)       99.7  (2.7)        97.0  (1.4)


2-Methyl-4,6-Dinitrophenol    31.0  (6.6)        31.4 (3.0)       18.6  (4.0)        11.7  (4.3)


Pentachlorophenol             95.1  (2.0)        98.8 (3.8)       102.9  (1.5)        98.2  (1.8)

-------
               TABLE  III.  Results Using CH Disks - % Recovery (RSD, n=3)






                                                     Volume (ml)




                                 100             200              300              500




Phenol                         12.2 (8.3)         6.9  (14.9)        2.9  (16.0)       2.0 (16.6)




2-Nitrophenol                  41.1 (2.5)        20.2  (6.1)        15.0  (10.5)       8.9 (8.2)




4-Nitrophenol                  28.8 (1.4)        16.5  (5.7)         7.9  (7.6)        5.2 (6.1)




2-Chlorophenol                41.6 (3.2)        21.4  (4.6)        15.2  (8.2)        8.9 (6.9)




2,4-Dinitrophenol               2.7 (15.5)        1.9  (8.7)         3.6  (6.5)        0.8 (19.7)




2,4-Dichlorophenol             64.9 (12.9)       59.2  (10.1)       53.4  (6.9)       34.4 (12.6)




2,4-Dimethylphenol            91.6 (0.8)        70.9  (14.4)       33.6  (10.0)      21.6 (18.4)




4-Chloro-3-Methylphenol      89.9 (6.0)        96.5  (9.3)        55.9  (9.3)       35.7 (20.5)




2,4,6-Trichlorophenol          92.4 (4.9)       109.5  (3.1)        79.4  (11.2)      86.7 (9.3)




2-Methyl-4,6-Dinitrophenol    19.8 (18.8)       14.2  (6.6)         9.8  (9.1)        6.3 (11.0)




Pentachlorophenol             99.9 (2.4)        93.5  (2.2)      105.0  (2.0)       98.7 (1.1)

-------
CO
-vl
             TABLE IV.  Results  Using SDVB Disks - % Recovery (RSD, n=3)




                                                    Volume  (ml)


                                 100             200             300             500


Phenol                         50.6  (12.4)      26.8 (10.0)       16.7  (20,9)        9.2  (13.6)


2-Nitrophenol                  90.8  (3.6)      101.6 (0.8)        95.9  (2.0)       93.0  (2.7)


4-Nitrophenol                  88.2  (10.3)      84.3 (6.1)        53.7  (13.4)      33.1  (12.9)


2-Chlorophenol                 89.7  (3.6)      107.7 (0.9)        91.8  (3.4)       71.6  (5.9)


2,4-Dinitrophenol              43.7  (28.2)      54.1 (15.8)      105.1  (12.3)      22.2  (11.6)


2,4-Dichlorophenol             91.3  (6.3)      102.4 (1.2)        94.1  (3.7)       91.8  (0.6)


2,4-Dimethylphenol            98.5  (4.7)      108.6 (3.1)        93.6  (8.4)       95.9  (1.3)


4-Chloro-3-Methylphenol      91.3  (12.3)     111.4 (3.3)        97.3  (14.0)     100.3  (2.9)


2,4,6-Trichlorophenol           90.9  (5.5)      115.7 (3.0)       100.5  (16.3)      97.8  (1.6)


2-Methyl-4,6-Dinitrophenol    64.0  (7.6)       96.9 (4.9)       100.0  (14.4)      85.1  (1.5)


Pentachlorophenol             95.3  (4.6)      100.2 (3.2)       105.0  (0.9)       98.2  (0.1)

-------
AQ              Comparison of Alternative Methods for Analysis of Volatile Organic Contaminants

                                                      by

                       Thomas C. Voice, Associate Professor of Environmental Engineering,
                                       Michigan State University, East Lansing, Ml 48824, and
                       James F. Ryan,   Environmental Marketing Manager, The Perkin-Elmer Corporation,
                                       Norwalk, CT 06859

        The U.S. Environmental Protection Agency (EPA) has published methods for the quantitative analysis of
        volatile organic contaminants (VOCs) in a variety of environmental sample matrices such as ground water,
        industrial effluents, drinking water, sludge, soil, and so forth. These methods are largely based upon a
        "purge-and-trap" methodology in which the volatile constituents are purged from the sample, collected on
        an adsorbent trap directly connected to a gas chromatograph, and then thermally desorbed onto the GC
        column for separation and quantitative analysis. These methods vary somewhat in the type of GC column
        and detectors specified, but the purge-and-trap technique for collecting the organic constituents and in-
        troducing them to the GC is essentially the same in all methods.  While the purge-and-trap technique of-
        fers the advantage of sub parts-per-million sensitivities for many compounds with specific GC detectors, it
        suffers from several significant disadvantages, including lack of a universally applicable trap adsorbent
        material, high sample-to-sample carryover, introduction of large quantities of water to the GC and detec-
        tors, poor compatibility with capillary columns, and limitations to automating the overall technique.

        This study describes an investigation of two alternatives to the common implementation of the EPA purge-
        and-trap procedures: (1) automated static headspace analysis and (2) what might be termed "off-line"
        purge-and-trap. Static headspace analysis involves equilibrating a sample with a fixed gas volume in a
        closed vessel and subsequently introducing an aliquot of this gas directly into the GC. The entire process
        can be automated using equipment available from a variety of manufacturers.  Off-line purge-and-trap
        involves purging samples using separate adsorbent traps for each sample independent of the GC system.
        The traps are then thermally desorbed into the GC using automated equipment.  Analyses of a series of
        VOCs have been performed using these two alternative techniques, along with the traditional purge-and-
        trap approach. Samples analyzed include both water and soil matrices.

        In the figures presented below, we show that off-line purge-and-trap methodology compares very
        favorably with the chromatographic and reproducibility data generated by an on-line methods.  However,
        the carryover is reduced by a factor of 10 with the off-line method because of the use of multiple traps.
                        On-line purge and trap
Off-line purge and trap
        5 ml of sample containing 10 ug/L of
        benzene, trichloroethane, toluene,
        tetrachloroethylene, ethylenebenzene,
        p-xylene, and o-xylene
0.53 mm x 50 m DB-624 column, PID detector
4 min @ 50ฐC, 8ฐ/min to 190 ฐ, 4 min hold
250 mg Tenax trap
4mindesorb@ 180ฐ
                                                      I-38

-------
                    Comparison of Sample Reproducibiltty and Sample Carryover.

                        %RSD Reproducibility                     % Carryover
Compound            On-line         Off-line                On-line       Off-line
benzene              2.42           2.24                   0.10          0.01
trichloroethylene       5.29           1.41                   0.16          0.01
toluene               1.10           2.61                   0.22          0.04
tetrachloroethylene     4.59           1.37                   0.28          0.02
ethylenebenzene              2.49           0.94                  0.34           0.03
p-xylene              3.39           1.72                   0.36          0.03
o-xylene              1,80           3.40                   0.43          0.04

Data has also been gathered on automated headspace analysis of these compounds in water and in soil.
                                               I-39

-------
49      EVALUATION OF SAMPLE PREPARATION METHODS FOR SOLID MATRICES

          Viorica Lopez-Avila. J. Milanes, N. Dodhiwala, and J. Benedicto, Mid-Pacific Environmental
          Laboratory, Mountain View, California 94043, and W. F. Beckert, U.S. Environmental
          Protection Agency, EMSL-LV,  Las Vegas, Nevada 89109.

          ABSTRACT

          Four sample preparation methods: Soxhlet extraction  (Method 3540), Soxtec extraction
          (Method 3541), sonication extraction (Method 3550), and supercritical fluid extraction (SFE)
          with carbon  dioxide (Method 3560) have been  evaluated.   Thirty  target compounds
          representing organochlorine  pesticides,   nitroaromatic  compounds,  haloethers,  and
          chlorinated hydrocarbons were spiked on wet and dry clay, topsoil, sand, and sand mixed
          with organic compost and were extracted by Soxhlet, Soxtec, and sonication techniques using
          hexane-acetone (1:1) and methylene chloride-acetone (1:1) and by SFE with carbon dioxide.
          Data are also presented  for 43 base/neutral/acidic compounds spiked on sand or clay and
          extracted by SFE with carbon dioxide and by Soxtec extraction with hexane-acetone (1:1),
          and for three standard reference materials extracted by Soxtec and SFE with carbon dioxide.

          INTRODUCTION

          There are currently two extraction methods listed in SW-846 (1) for the extraction of solid
          matrices: Method  3540  (Soxhlet  extraction) and  Method 3550 (sonication extraction).
          Method 3540 is generally applicable, and a  large number of samples can be extracted side
          by side with limited manpower requirements. However, Soxhlet extractions usually take
          between 8 and 26 hours, require relatively  large amounts  of solvents, and involve extract
          cleanup and concentration. Sonication extractions require much shorter extraction times,
          but they are labor-intensive, use large amounts of solvent, and require extract cleanup and
          concentration.

          Two new techniques that have become available recently are Soxtec extraction  (Method
          3541) and  supercritical  fluid extraction (SFE) (Method  3560).  Soxtec extraction is a
          modified Soxhlet extraction: the thimble with the sample is first immersed in hot solvent,
          then, after  a boiling period of usually up to 1 hour, is raised physically and extracted a la
          Soxhlet for another hour.  The very limited results reported so far by others indicate that
          Soxtec extraction is at least as exhaustive as Soxhlet extraction (2), but the extraction time
          is reduced to about 2 hours, less solvent is needed, and the solvent is evaporated and
          NOTICE:  Although the research described in this paper has been supported by the U.S.
          Environmental Protection Agency, it has not been subjected to Agency review and therefore
          does not necessarily reflect the views of the Agency, and no official endorsement should be
          inferred.   Mention of trade  names  and  commercial products does not  constitute
          endorsement or recommendation for use.
                                                 11-40

-------
condensed without requiring extract transfer.  SFE uses a supercritical fluid as extraction
solvent in a special extraction system that is operated at pressures and temperatures higher
than the  critical pressure and  critical temperature of the particular fluid.  The most
commonly used fluid  is carbon dioxide; others  that are  being used, or  have been
investigated, include nitrous oxide, sulfur hexafluoride, Freon-13, ammonia, xenon, and
several hydrocarbons. Carbon dioxide is so popular because of its low critical temperature
(31.3ฐC) and pressure (72.9 atm) and because it is non-toxic, non-flammable, relatively non-
reactive and inexpensive, and its use does not result in a waste disposal problem.  It is a
rather non-polar solvent, similar to  hexane or benzene, but both solvent  strength and
selectivity can be improved by the addition of small amounts of modifiers such  as acetone,
methanol, or toluene.

The objective of this study was to evaluate the applicability of Soxtec extraction to samples
of interest to the EPA and to generate performance data for these four extraction methods
for  solids.  We focused on  30 analytes covering the following groups of compounds
environmentally significant to EPA: organochlorine pesticides, chlorinated hydrocarbons,
nitroaromatics, and haloethers.  To a limited extent, we also generated data  for 43
base/neutral/acidic compounds currently  on the Hazardous Substances List.  The matrices
evaluated included sand, clay,  topsoil, sand mixed with organic compost, and  standard
reference materials certified for a limited number of organic compounds (mostly polynuclear
aromatic hydrocarbons).

EXPERIMENTAL

Apparatus

    •   Soxhlet extractor - 40  mm  ID with 500-mL round bottom flask, condenser and
        heating mantle

    •   Sonication system - Horn-type sonicator equipped with titanium tip (Heat Systems
        Ultrasonics Inc., Farmingdale, New York,  Model W-375)

    •   Soxtec  HT-6 extraction system with controlled heated  oil bath (Tecator, Inc.,
        Herndon, Virginia)

    •   Kuderna-Danish apparatus with 10-mL concentrator tube, 500-mL evaporation flask,
        three-ball macro Snyder column

    •   Supercritical fluid extractor - Suprex Model SE-50 including a 4-port and  a 12-port
        valve configured with electronic actuators for automated operation. The system was
        set up either with two or four extraction vessels for parallel extractions. The 3-mL
        extraction vessels (1 cm ID x 4 cm length) were obtained from Suprex Corporation
        (Pittsburgh, Pennsylvania), the 2-mL extraction vessels (0.9 cm ID x 3 cm length)
        from Alltech  Associates (Deerfield, Illinois).    Supercritical  pressures were
        maintained inside the extraction  vessels by using 60 cm  of uncoated  fused-silica
        tubing (50 urn ID x 375 urn OD) from J&W Scientific  (Folsom, California) as
        restrictor.   Collection of the extracted material was performed by inserting the
                                       11-41

-------
       outlet  restrictor into a  15-mm  x 60-mm  glass vial (Supelco Inc.,  Bellefonte,
       Pennsylvania) containing 5 mL hexane.

       Gas chromatograph ซ A Varian 6000 equipped with two constant-current/pulsed-
       frequency electron capture detectors and two megabore fused-silica open-tubular
       columns (30-m x 0.53-mm ID x 0.83-/^n film thickness DB-5 column and 30-m x
       0.53-mm ID x 1.0-f/m film thickness DB-1701 column), connected to a press-fit Y-
       shaped glass splitter (J&W Scientific Inc., Folsom, California) was used to analyze
       for the 30 target analytes. The columns were temperature-programmed from 100ฐ C
       (2-min hold) to 275ฐ C (6-min hold) at 5ฐC/min; injector temperature 250ฐ C;
       detector temperature 320CC;  helium carrier gas 6 mL/min; nitrogen makeup gas
       20 mL/min.

       Gas chromatograph/mass spectrometer — A Finnigan 4510B (Finnigan MAT, San
       Jose, California) interfaced with a data system for data acquisition and processing
       and equipped with a 30-m x 0.32-mm ID DB-5 fused-silica open-tubular column (1-
       |im film thickness) was used for all PAH and base/neutral/acidic compound
       analyses.  The column was temperature-programmed from 40ฐ C (4-min hold)  to
       300ฐ C (6-min hold) at 8ฐC/min; injector temperature 270ฐ C; interface temperature
       270ฐ C.
Materials
       Standards — Analytical reference standards of the organochlorine pesticides,
       chlorinated hydrocarbons, nitroaromatics, haloethers, PAHs and base/neutral/acidic
       compounds  were obtained from the U. S. Environmental Protection Agency,
       Pesticides and Industrial Chemicals Repository (Research Triangle Park, North
       Carolina), Aldrich Chemical (Milwaukee, Wisconsin), Ultrascientific Inc. (Hope,
       Rhode Island), and Chem Service (West Chester, Pennsylvania). All compounds,
       except the PAHs and the  base/neutral/acidic compounds, were obtained as neat
       materials. Their purities were stated to be greater than 98 percent. Stock solutions
       of each  test compound  were prepared in pesticide-grade hexane at 1 mg/L.
       Working calibration standards were prepared by serial dilution of a composite stock
       solution  prepared from  the  individual stock solutions.  The PAHs and the
       base/neutral/acidic compounds were obtained as composite mixtures in methylene
       chloride or methylene chloride/toluene.

       SFC-grade carbon dioxide (Scott  Specialty Gases, Plumsteadville, Pennsylvania)

       Hexane, acetone, methylene chloride ~ nanograde or pesticide-grade

       Sample matrices: sand, clay, topsoil, sand mixed with 10 percent organic compost,
       marine sediments HS-3 and HS-4 (National  Research Council of Canada, Halifax,
       Nova Scotia, Canada), PAH-contaminated  soil SRS 103-100  (Fisher  Scientific,
       Pittsburgh, Pennsylvania).  The sand and the standard reference materials were dry.
       The  clay, topsoil, and the  sand/compost  matrices contained  10.6, 2.6,  and
       4.2 percent moisture.
                                        1-42

-------
Procedures

Spiked samples (10 g each) of sand, sand with 10 percent organic compost, clay, and topsoil
were extracted with hexane-acetone (1:1) or methylene chloride-acetone (1:1) following the
procedures given in Methods 3540 and 3550. Spiking of the samples (that were extracted
by SFE, Soxhlet, and Soxtec) with the 30 target compounds or the base/neutral/acidic
compounds was performed as follows: the sample was weighed out in an aluminum cup and
a concentrated stock solution (100 to 1000 jiL) containing the target compounds in hexane
or methylene chloride and methylene  chloride/toluene was added to the  sample with a
syringe while making sure that the solution did not contact the aluminum cup.  Mixing was
performed with the tip of a disposable pipette. After the solvent had evaporated completely
(approximately 15 min),  the spiked sample was transferred to the extraction vessel. Spiking
of samples that were extracted by sonication was performed directly into the amber bottle
used for extraction.

Soxtec extractions were performed with 10-g samples and 50 mL solvent using an immersion
time of 45 or 60 min and an extraction time of 45 or 60 min as indicated in the tables.

SFEs were performed as specified in the Results Section.  All SFEs were carried out using
the Suprex SE-50 system.

RESULTS AND DISCUSSION

Table 1 presents the average recoveries of the 30 target compounds spiked on sand with
10 percent organic compost and on the clay matrix and extracted by sonication and Soxhlet
extraction with hexane-acetone (1:1). The results from the Soxtec extraction are presented
in Table 2 and the SFE  data in Table 3.

The following conclusions can be drawn from these data:

    •  The repeatability of the sonication  extraction with hexane-acetone (1:1) is much
       better than that of Soxhlet extraction.  The percent RSDs for the 30 target
       compounds for sonication ranged from 2.3 to 3.9 percent (except for one value at
        14.7 percent) for the sand/compost matrix and 0.2 to 6.5 percent for the clay matrix.
       The percent RSDs for the recoveries from the Soxhlet extraction ranged form  3.9
       to 86.9 percent for the clay matrix, with most of the values above 20 percent.

    •  The repeatability  of the Soxtec technique is significantly better than that of the
       Soxhlet technique.  Only the more volatile compounds  such as nitrobenzene,
       benzotrichloride, 4-chloro-2-nitrotoluene, and the dichloronitrobenzenes exhibited
        RSD values above 10 percent when the extraction was performed with either
       hexane-acetone  or methylene chloride-acetone. The percent RSDs for the other
       compounds were  below 10 percent.  The average recoveries using the Soxtec
       technique were significantly higher than those obtained by Soxhlet or sonication, and
        similar or slightly higher than the SFE recoveries, for both the hexane-acetone and
        the methylene chloride-acetone solvent combinations.
                                        1-43

-------
    •  SFE recoveries were comparable to those obtained by sonication (except for the clay
       matrix where SFE recoveries were significantly higher than the recoveries via
       sonication) and Soxhlet techniques, but the RSDs for the SFE values were quite
       high.  The data reported in Table 3 were obtained with our four-vessel setup;
       therefore,the RSD values for the SFE data are actually those from the combined
       results from four extractions carried out simultaneously. Work is in progress in our
       laboratory to investigate the vessel-to-vessel variability.

The data for the base/neutral/acidic compounds are presented in Table 4 and 5 for the SFE
and Soxtec extraction, respectively.  The recoveries by SFE and Soxtec extraction are
comparable (except for the very low SFE recoveries for compounds  4, 5, 7,  12, probably
because of their volatilities and for benzoic acid and 4-nitrophenol, probably because of their
low solubilities in supercritical carbon dioxide) and the percent  RSDs  follow the same
pattern as discussed above for the group of 30 compounds.

In the case of the three standard reference materials, we noticed significant differences in
recoveries obtained by Soxtec and by SFE. For the SRS 103-100 standard reference soil
(Table 6), the SFE naphthalene and acenaphthylene recoveries were only about 50 to 60
percent of those measured in the Soxtec extracts.  This could be explained by the high
volatilities of the two compounds.  However, the recoveries of the higher-molecular-weight
PAHs benzofluoranthenes and  benzo(a)pyrene were 53 and 32 percent by SFE versus 118
and 80 percent by Soxtec. Additional extractions were performed with supercritical carbon
dioxide modified with 10 percent hexane, 1 percent toluene,  or 15 percent propylene
carbonate to improve the  extractabilities of the higher-molecular-weight PAHs.  Only
propylene carbonate showed increased extractabilities for the compounds cited above. For
the HS-3 and HS-4 (Table 7), SFE recoveries were approximately around 20 percent when
the extraction was performed with carbon dioxide.  Addition of modifiers increased the
recoveries somewhat.  Furthermore, presence of elemental sulfur in these marine sediments
created restrictor plugging problems on four commercial extractors evaluated by us as part
of another study.

In conclusion, sonication and  Soxtec extraction of environmental samples with hexane-
acetone (1:1) give comparable  results in terms of method precision and accuracy and are
fast.  However, they both require  large amounts of solvents, and the extracts need to be
subjected to gel permeation chromatography or some type of column chromatography (e.g.,
alumina, silica) especially if an electron capture detector will be used for analysis. SFE, on
the other hand appears to be much faster and more  selective.  However, the technique is
matrix-dependent, and although we have shown that  many compounds of interest to EPA
can be extracted from spiked sand, more developmental work is required  before SFE can
be used routinely with environmental matrices.

REFERENCES

1.  Test Methods for Evaluating Solid Waste (1986), 3rd Ed., SW-846, U.S. Environmental
    Protection Agency, Washington, DC.
                                       1-44

-------
2.  J.H. Stewart,  Jr.,  C.K. Bayne, R.L. Holmes,  W.F. Rogers and  M.P. Maskarinek.
   "Evaluation of  a  Rapid  Quantitative  Extraction  System  for  Determining the
   Concentration of PCBs in Soils." Proceedings of U.S. Environmental Protection Agency
   Symposium on Waste Testing and Quality Assurance, July 11-15,1988, Washington, DC.
                                       11-45

-------
TABLE 1.    AVERAGE PERCENT RECOVERIES AND PERCENT RSDs FOR 30 TARGET COMPOUNDS EXTRACTED FROM
            SPIKED SAND/COMPOST AND CLAY SAMPLES BY SONICATION AND SOXHLET EXTRACTION WITH HEXANE-
            ACETONE (1:1)
                                                         Sonication8
Soxhletb
Sand/compost with

Compound
no.
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
26
27
28
29
30


Compound name
13-Dichlorobenzene
1,2-Dichlorobenzene
Nitrobenzene
Benzal chloride
Benzotrichloride
4-Chloro-2-nitrotoluene
Hexachlorocyclopentadiene
2,4-Dichloronitrobenzene
3,4-Dichloronitrobenzene
Pentachlorobenzene
23>4,5-Tetrachloronitrobenzene
Benefin
alpha-BHC
Hexachlorobenzene
delta-BHC
Heptachlor
Aldrin
Isopropalin
Heptachlor epoxide
trans-Chlordane
Endosulfan I
Dieldrin
2,5-Dichlorophenyl-4'-nitrophenyl ether
Endrin
Endosulfan II
2,4,6-Trichlorophenyl-4'-nitrophenyl ether
P|P:-DDT
23,6-Trichlorophenyl-4'-nitrophenyl ether
23,4-Trichlorophenyl-4'-nitrophenyl ether
Mirex
20 percent
Average
recovery
0
0
0
63.3
0
933
0
923
0
82.4
84.8
99.6
90.0
883
88.6
90.9
84.6
95.0
90.5
89.9
91.6
91.8
86.2
95.3
86.7
47.3
84.2
82.7
79.5
84.2
moisture
Percent
RSD

_
_
14.7
ซ
3.5
_
2.9

2.7
3.4
2.6
23
2.9
3.1
2.8
2.6
3.7
3.7
3.0
3.4
3.7
3.0
3.6
3.7
3.1
3.0
3.1
3.4
3.9
Clay with
20 percent
Average
recovery
0
2
0
0
0
34.1
0
37.0
34.0
30.1
35.5
35.4
45.0
34.4
47.6
40.7
42.1
38.0
46.1
44.7
453
48.9
44.7
44.9
47.4
23.5
44.7
47.1
44.1
513
moisture
Percent
RSD

_
_
_
_
6.5
—
2.0
5.2
7.1
2.5
5.8
3.4
5.5
0.2
3.7
43
4.8
1.5
1.4
1.0
1.0
3.2
1.1
03
3.1
1.8
3.1
4.6
3.0
Sand/compost with
20 percent
Rep. 1

0
0
0
30.7
29.6
46.0
0
40.4
34.4
55.0
32.0
493
67.5
58.8
78.1
65.4
72.1
61.1
77.7
75.4
73.6
79.4
51.8
86.0
74.6
NSC
69.4
35.6
46.2
74.7
moisture
Rep. 2

0
0
0
30.7
29.2
443
_
43.8
40.2
52.5
42.8
60.7
78.8
69.7
84.8
76.7
79.4
79.4
83.1
82.4
81.0
82.0
75.5
88.5
80.9
NS
82.2
71.1
67.6
79.8
Clay with
20 percent
Average
recovery
0
0
0
0
9.8
34.0
0
32.2
24.8
53.2
27.1
47.8
57.4
55.4
65.0
59.6
69.8
64.2
72.0
75.6
76.4
74.4
65.9
81.0
78.5
NS
73.6
64.4
62.5
75.5
moisture
Percent
RSD

_
_
_
86.9
44.4
_
57.8
44.0
16.6
38.4
11.5
47.5
24.5
27.1
34.1
8.8
20.8
20.8
12.5
5.5
20.0
26.9
3.9
6.7
NS
38.5
34.4
29.2
15.0
a Number of determinations was three. Spiking level was 500 ng/g, except compounds 23, 28, and 29 at 1500 ng/g, compound 26 at 3000 ng/g, compound 3 at 2000
 ng/g, and compounds 1 and 2 at 5000 ng/g.
" Number of determinations was three except for sand/compost matrix where only two determinations were performed.  Spiking level was the same as for the
 sonication experiments. Extraction time was 16 hours.
c NS - not spiked.

-------
  TABLE 2. AVERAGE PERCENT RECOVERIES AND PERCENT RSDs FOR THE 30
            TARGET COMPOUNDS FROM SPIKED CLAY SAMPLES BY SOXTEC
            EXTRACTION WITH HEXANE-ACETONE (1:1) AND METHYLENE
            CHLORIDE-ACETONE (!:!)•
                                            Hexane-acetone
Methylene chloride-acetone
Compound
no.
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
26

27
28

29

30
Compound
name
13-Dichlorobenzene
1,2-Dichlorobenzene
Nitrobenzene
Benzal chloride
Benzotrichloride
4-Chloro-2-nitrotoluene
Hexachlorocyclopentadiene
2,4-Dichloronitrobenzene
3,4-Dichloronitrobenzene
Pentachlorobenzene
23,4,5-Tetrachloronitrobenzene
Benefin
alpha-BHC
Hexachlorobenzene
delta-BHC
Heptachlor
Aldrin
Isopropalin
Heptachlor epoxide
trans-Chlordane
Endosulfan I
Dieldrin
2,5-Dichlorophenyl-4'-
nitrophenylether
Endrin
Endosulfan II
2,4,6-Trichlorophenyl-4'-
nitrophenylether
p,p'-DDT
23,6-Trichlorophenyl-4'-
nitrophenylether
23,4-Trichlorophenyl-4'-
nitrophenylether
Mirex
Average
recovery
0
0
77.1
383
33.4
92.8
46.0
115
783
48.6
122
82.0
94.9
81.7
104
87.1
78.2
97.5
92.4
85.8
90.5
68.8

99.7
112
903

127
61.4

97.2

91.6
84.0
Percent
RSD
_
—
18
7.8
17
17
21
8.0
83
12
4.6
3.7
5.5
7.1
9.7
5.4
5.7
6,9
0.6
2.2
2.0
2.6

2.0
4.4
10

5.0
6.5

2.0

13
5.1
Average
recovery
0
b
0
0
32.5
41.6
0
39.9
543
58.7
89.8
84.8
91.8
85.6
103
89.4
70.7
95.2
91.0
95.8
92.8
73.4

106
119
89.5

70.7
41.1

96.9

943
106
Percent
RSD
_
—
_
_
41
27
_
18
16
8.9
23
3.4
6.3
1.8
5.7
3.0
33
8.8
4.2
4.2
43
8.1

53
4.6
6.1

8.8
16

53

53
7.4
a The operating conditions for Soxtec apparatus were as follows:  immersion time - 60 min; extraction time - 60 min;
 the sample size was 10 g clay, the spiking level was 50 ng/g, except compounds 23, 28, and 29 at 150 ng/g,
 compound 26 at 300 ng/g, compound 3 at 200 ng/g, and compounds 1 and 2 at 500 ng/g. The number of
 determinations was four. The moisture content of the matrix was not altered.
"Not able to determine because of interference.
                                        11-47

-------
TABLE 3.  AVERAGE PERCENT RECOVERIES AND PERCENT RSDs FOR 30 TARGET COMPOUNDS EXTRACTED
           FROM VARIOUS SPIKED MATRICES WITH SUPERCRITICAL CARBON DIOXIDE8
Sand with 10 percent
Sand
Compound
no.
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
26
27
28
29
30

Compound
1,3-Dichlorobenzene
1,2-Dichlorobenzene
Nitrobenzene
Benzal chloride
Benzotrichloride
4-Chloro-2-nitrotoluene
Hexachlorocyclopentadicne
2,4-Dichloronitrobenzene
3,4-Dichloronitrobenzene
Pentachlorobenzene
2,3,45-TetrachIoronitrobenzene
Benefit!
alpha-BHC
Hexachlorobenzene
delta-BHC
Heptachlor
Aldrin
Isopropalin
Heptachlor epoxide
trans-Chlordane
Endosulfan I
Dieldrin
25-Dichlorophenyl-4'-nitrophenyl ether
Endrin
Endosulfan II
2,4,6-Trichlorophenyl-4'-nitrophenyl ether
p,p'-DDT
2,3,6-Trichiorophenyl-4'-nitrophenyl ether
2,3,4-Trichlorophenyl-4'-nitrophenyl ether
Mirex
Average
recovery
0
0
0
0
0
56.9
16.9
57.8
68.7
60.9
75.3
70.4
72.5
60.6
68.7
82.9
76.9
112
79.6
71.0
76.7
82.9
66.9
68.5
62.8
81.6
71.1
71.3
56.7
81.1
Percent
USD

—
_
	
_
25.9
11.8
16.1
30.1
16.8
15.1
9.5
13.6
10.2
2.8
125
19.2
11.8
11.6
4.3
32.2
29.5
2.5
27.8
26.2
9.9
15.6
3.5
22.6
30.3
Clay
Average
recovery
0
0
0
0
0
57.8
15.1
62.3
62.9
50.9
65.8
65.9
66.1
56.7
73.8
63.4
62.0
70.8
70.7
71.1
68.9
114
74.3
76.4
76.7
76.6
86.9
72.9
68.7
67.3
Percent
RSD
_
-
—
—
-
28.9
60.1
27.9
25.6
21.8
26.3
27.8
31.3
28.1
25.8
28.6
37.8
26.0
27.1
28.2
24.3
21.1
25.2
24.7
26.5
23.3
19.7
225
27.3
27.6
Topsoii
Average
recovery
0
0
53.6
30.1
0
683
47.4
68.9
69.9
68.3
74.2
76.0
75.3
735
81.3
74.9
75.3
76.9
79.4
80.4
79.2
845
793
79.7
76.9
78.9
82.6
79.1
68.9
79.0
Percent
RSD
_
—
32.9
36.3
-
24.4
33.1
24.2
23.4
20.1
19.1
18.7
18.7
18.2
14.7
17.6
19.3
16.8
18.7
17.3
15.7
15.8
16.0
15.8
13.8
14.9
15.6
18.2
18.6
16.4
organic compost
Average
recovery
93.6
795
76.6
81.6
80.3
86.4
81.8
73.4
76.0
84.0
73.9
73.9
75.3
74.1
81.7
87.7
775
81.1
81.9
80.4
78.3
82.3
80.6
83.9
82.8
88.3
84.7
79.1
75.7
78.2
Percent
RSD
29.2
24.3
22.7
24.6
27.1
245
23.6
22.9
18.9
24.0
205
21.9
22.7
21.9
21.7
20.9
195
17.9
19.6
21.3
16.2
19.6
19.7
18.8
22.8
15.4
12.7
17.9
19.1
20.2
aThe number of samples extracted in parallel for each matrix was four. The experiments were performed with supercritical carbon dioxide at 300 atm/70ฐC/60 min dynamic. The
 sample size was 2 g. The spiking level was 25 ng/g, except compounds 23, 28, and 29 at 75 ng/g, compound 26 at 150 ng/g, compound 3 at 100 ng/g, and compounds 1 and 2 at
250 ng/g. The moisture content of the matrix was not altered.

-------
TABLE 4. AVERAGE  PERCENT  RECOVERIES  AND  PERCENT   RSDs  FOR
          BASE/NEUTRAL/ACIDIC  COMPOUNDS  EXTRACTED FROM  SPIKED
          SAND WITH SUPERCRITICAL CARBON DIOXIDE3
Compound
no.
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
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
Compound name
Phenol
Bis(2-chloroethyi)ether
2-Chlorophenol
1 ,3-Dichlorobenzene
1,4-Dichlorobenzene
Benzyl alcohol
1,2-Dichlorobenzene
2-Methylphenol
Bis(2-chloroisopropyl)ether
4-Methylphenol
N-nitroso-di-n-propylamine
Hexachloroethane
Nitrobenzene
Isophorone
2-Nitrophenol
2,4-Dimethylphenol
Benzoic acid
Bis(2-chloroethoxy)methane
2,4-Dichlorophenol
1 ,2,4-Trichlorobenzene
Hexachlorobutadiene
4-Chloro-3-methylphenol
2-Methylnaphthalene
Hexachlorocylopentadiene
2,4,6-Trichlorophenol
2,4,5-Trichlorophenol
2-Chloronaphthalene
Dimethylphthalate
2,4-Dinitrophenol
4-Nitrophenol
Dibenzofuran
2,4-Dinitrotoluene
2,6-Dinitrotoluene
Diethyl phthalate
4-Qilorophenyl-phenylether
4,6-Dinitro-2-methylphenol
4-BromophenyI-phenyiether
Hexachlorobenzene
Pentachlorophenol
Di-n-butylphthalate
Butylbenzylphthalate
Bis(2-ethylhexyl)phthalate
Di-n-octylph thalate
Terphenyl-dj4
Spike level
(ng/g)
300
150
300
150
150
150
150
300
150
300
150
150
150
150
300
300
300
150
300
150
150
300
150
150
300
300
150
150
300
300
150
150
150
150
150
300
150
150
300
150
150
150
150
20.0b
Average
recovery
50.9
23.6
25.9
4.2
4.7
54.4
8.2
54.4
27.1
64.5
58.9
5.4
41.9
60.4
50.2
65.5
7.3
61.6
63.6
32.6
25.0
71.8
62.2
46.6
71.5
80.2
69.5
59.2
37.2
10.0
78.0
71.0
78.6
66.7
79.9
53.9
773
78.2
65.2
73.0
545
71.9
58.0
92.1
Percent
USD
26.3
75.0
64.6
160
156
135
119
25.1
69.1
155
17.9
151
38.7
12.1
31.8
16.4
24.6
23.1
16.6
51.7
66.4
10.1
19.4
33.8
9.4
7.3
14.4
15.5
19.7
27.4
6.9
7.3
7.3
135
10.6
16.7
11.0
7.7
12.4
12.0
19.8
14.0
16.1
2.2
 a The number of samples extracted in parallel was four. The experiments were performed at 150 atm/50ฐC/10 min static
  followed by 200 atm/60ฐC/10 min dynamic and 250 atm/70ฐC/10 min dynamic.  The sample size was 3 g dry sand.
 b Spiked at 20 ng//iL in the collection vial.
                                     11-49

-------
TABLE 5. AVERAGE   PERCENT  RECOVERIES  AND  PERCENT   RSDs  FOR
          BASE/NEUTRAL/ACIDIC COMPOUNDS EXTRACTED FROM SPIKED CLAY
          BY SOXTEC EXTRACTION WITH HEXANE-ACETONE (1:1)
Coaccntration (ng/jiL)
CoMpound
•a.
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
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
Compound Mine
Phenol
Bis(2-chloroethyl)ether
2-Chlorophenol
13-Dichlorobenzene
1,4-Dichlorobenzene
Benzyl alcohol
1,2-Dichlorobenzenc
2-Mcthylphenoi
Bis(2-ch!oroisopropy1)ether
4-Methylphenol
N-nitroso-di-n-propylamine
Hexachloroethane
Nitrobenzene
Isophorone
2-Nitrophcnol
2,4-Dimethylphenol
Benzoic acid
Bis(2-chloroethoxy)mcthane
2,4-Dichlorophenol
1,2,4-Trichlorobenzene
Hexachlorobutadiene
4-Chloro-3-mcthylphenol
2-Methylnaphthalene
Hexachlorocylopentadiene
2,4,6-Trichlorophenol
2,45-Trichlorophenol
2-ChloronaphthaIene
Dimethyl phthalate
2,4-Dinitrophcnol
4-Nitrophcnol
Dibenzofuran
2,4-Dinitrotolucne
2,6-Dinitrotol uene
Diethyl phthalate
4-Chiorophenyl-phenylether
4,6-Dinitro-2-methylphenol
4-Bromophenyi-phenylether
Hexachlorobenzene
Pentachlorophenol
Di-n-butyl phthalate
Butyibenzyl phthalate
Bis(2-ethylhexyl)phthalate
Di-n-octyl phthalate
Clayl
14.4
65
12.3
ND
ND
16.9
ND
5.3
3.8
7.2
12.6
ND
7.8
16.7
10.0
14.7
12.5
13.0
16.5
4.4
ND
20.0
12.7
45
20.8
7.9
17.2
22.2
28.1
20.8
23.9
25.1
20.8
22.9
19.7
20.1
18.7
21.4
19.7
33.2
20.4
23.6
23.8
Clayl
15.1
8.2
13.4
ND
ND
17.9
ND
5.6
4.6
7.4
13.1
ND
8.5
17.6
11.1
16.0
12.9
17.5
17.5
S3
ND
20.2
14.7
6.4
22.5
83
19.4
23.6
29.7
20.3
26.3
26.7
21.5
23.4
20.9
19.4
193
22.7
19.2
23.0
20.6
24.0
26.2
Clay 3
13.5
8.2
12.7
ND
ND
155
ND
4.9
5.1
65
11.6
ND
9.1
16.2
11.3
14.4
11.1
13.0
16.0
6.6
ND
18.4
14.9
65
19.9
7.9
185
21.3
24.9
24.9
23.7
24.0
19.2
21.1
19.9
17.6
18.2
212
175
14.3
18.7
21.9
24.8
Average
reeoveiy
(percent)
47.8
25.4
42.7
0
0
55.9
0
17.6
15.0
23.4
41.4
0
28.2
56.1
36.0
50.1
40.6
44.1
55.6
18.1
0
65.1
47.0
193
70.2
26.8
61.2
74.6
91.9
62.9
82.1
84.2
683
74.9
67.2
63.4
62.4
72.6
62.7
78.3
663
77.2
83.1
Percent
USD
5.6
13
43
-
—
7.2
-
6.6
14.6
6.7
6.2
—
7.7
4.2
65
5.7
7.7
3.0
4.6
31
-
5.1
8.6
19
6.3
2.9
6.0
5.2
8.9
16
5.9
5.4
5.8
5.4
3.2
6.8
3.0
3.7
6.1
40
5.2
4.8
4.8
•Soxtec samples included additional 21 compounds not listed here. The operating conditions for the Soxtec apparatus were as
follows: immersion time - 45 min; extraction time • 45 min; the sample size was 10 g clay, the spiking level was 6 ft-gjg- The
moisture content of the matrix was not altered.
                                      11-50

-------
TABLE 6.   PERCENT RECOVERIES OF COMPOUNDS EXTRACTED FROM THE
           SRS  103-100  STANDARD REFERENCE  MATERIAL BY  SOXTEC
           EXTRACTION  (HEXANE-ACETONE  1:1)  AND  BY   SFE  WITH
           SUPERCRITICAL CARBON DIOXIDE
Compound name
Naphthalene
2-Methylnaphthalene
Acenaphthylene
Acenaphthene
Dibenzofuran
Fluorene
Phenanthrene
Anthracene
Fluoranthene
Pyrene
Benzo(a)anthracene
Chrysene
Benzo(k + b)fluoranthene
Benzo(a)pyrene
Pentachlorophenol
Certified
value
(rag/kg)
32.4 ฑ 8.2
62.1 ฑ 11.5
19.1 ฑ 4.4
632 ฑ 105
307 ฑ 49
492 ฑ 78
1618 ฑ 348
422 ฑ 49
1280 ฑ 220
1033 ฑ 289
252 ฑ 38
297 ฑ 26
152 ฑ 22
97.2 ฑ 17.1
965 ฑ 374
Soxtec3
percent recovery
127
127
110
108
123
92.7
81.3
131
81.3
69.1
95.2
91.6
118
80.2
111
SFEb
percent recovery
63.8
82.6
64.6
98.2
92.9
80.4
124
78.4
92.3
78.2
67.6
68.4
53.3
32.2
141
a Single determination. The operating conditions for the Soxtec apparatus were as follows:
  immersion time - 45 min; extraction time - 45 min; the sample size was 10 g.
" The values given represent the average recoveries for three replicate samples extracted
  sequentially.  The sample size was 2.5 g. The extraction was performed with carbon
  dioxide at 300 atm and 70ฐ C for 60 min; 10 percent moisture was added to each sample
  prior to extraction.
                                    1-51

-------
01
ro
         TABLE 7.  PERCENT RECOVERIES OF COMPOUNDS EXTRACTED FROM THE HS-3 AND HS-4 MARINE SEDIMENTS
                   BY SOXTEC EXTRACTION (HEXANE-ACETONE 1:1) AND BY SFE WITH SUPERCRITICAL CARBON DIOXIDE
Compound name
Naphthalene
Acenaphthylene
Acenaphthene
Fluorene
Phenanthrene
Anthracene
Fluoranthene
Pyrene
Benzo(a)anthracene
Chrysene
Benzo(a)pyrene
Benzo(b)fluoranthene
Benzo(k)fluroanthene
Benzo(ghi)perylene
Dibenzo(ah)anthracene
Indeno( l,2,3-cd)pyrene
Certified I
value
(rag/kg)
9.0
0.3
4.5
13.6
85
13.4
60
39
14.6
14.1
7.4
7.7
2.8
5.0
1.3
5.4
ฑ 0.7
ฑ 0.1
ฑ 1.5
ฑ 3.1
ฑ20
ฑ 0.5
ฑ 9
ฑ 9
ฑ 2.0
ฑ 2.0
ฑ 3.6
ฑ 1.2
ฑ 2.0
ฑ 2.0
ฑ 0.5
ฑ 1.3
IS-3 percent recovery
Soxtec8
47.8
167
129
53.7
44.9
75.4
51.3
43.1
56.2
57.4
48.6
71.4
175
56.0
92.3
51.9
SFEb
11.1
18.9
15.1
26.1
12.3
28.7
27.7
28.1
30.5
9.5
22.1
85.7
—
—

Certified
value
(mg/kg)
0.15
0.15
0.15
0.15
0.68
0.14 ฑ0.07
1.25 ฑ0.10
0.94 ฑ0.12
0.53 ฑ0.05
0.65 ฑ0.08
0.65 ฑ0.08
0.70 ฑ0.15
0.36 ฑ0.05
0.58 ฑ0.22
0.12 ฑ0.05
0.51 ฑ0.15
HS-4 percent recovery
Soxteca
~
85.3
129
88.0
95.7
71.7
76.9
58.5
71.4
133
—
~

SFEb
—
26.5
ซ
21.6
18.1
—
-
~
—
~
—
—

         a Single determinations. The operating conditions for the Soxtec apparatus were as follows: immersion time - 45 min; extraction
           time - 45 min; the sample size was 10 g.
         b The HS-3 sample (2 g) was extracted at 350 atm and 60ฐC for 20 min (single determination). The HS-4 sample (four 1.5-g
           samples extracted in parallel, the extracts were then combined) was extracted at 350 atm and 70ฐ C for 30 min.

-------
CH     ANALYSIS  FOR SELECTED APPENDIX  IX COMPOUNDS IN ENVIRONMENTAL MATRICES BY
        HIGH  PERFORMANCE  LIQUID CHROMATOGRAPHY/PARTICLE BEAM MASS SPECTROMETRY

        Jeffery L. Cornell, Senior Scientist, Jeffrey C. Lowry, Director of
        Organics, Marshall D. Tilbury,  Scientist, Enseco-Rocky Mountain Analytical
        Laboratory, 4955  Yarrow Street, Arvada, Colorado 80002

        ABSTRACT

        A method  is described for the analysis of selected Appendix IX analytes in
        environmental matrices, using high performance liquid chromatography-mass
        spectrometry (LC/MS) employing  a particle beam interface.  The method is
        targeted  at compounds that are  not easily extracted and/or analyzed by the
        current methodologies.  The five selected compounds are chromatographed on
        an octadecylsilane  (C-18) reverse phase column, using methanol and ammonium
        acetate in water.  A commercial particle beam interface is used in
        conjunction with  a quadrupole mass spectrometer for detection and
        quantitation of the analytes.   One internal standard and two surrogate
        standards have been included.   Multipoint calibration curves indicate that
        concentration versus response data fit a second order polynomial model.
        This  second order equation can  then  be used for quantitation of subsequent
        check standards and samples.

        Sample preparation methods are  described to extract these compounds from
        water and soil samples.   For  controlled matrices, the average recovery of
        the analytes from water  samples is 73% and 84% for soils.

        INTRODUCTION

        In  1987,  the Environmental Protection Agency  (EPA) promulgated regulations
        which required owners  and operators  of hazardous waste  treatment,  storage
        and disposal facilities  to analyze their ground water for a list of 232
        constituents listed in Appendix IX of 40 CFR, Part 264  (52 Federal Register
        25942).   The Appendix  IX list consists of metal, anions, and a wide variety
        of organic compounds,  including nitrosamines, phenols,  polynuclear aromatic
        hydrocarbons, volatile organics,  pesticides,  herbicides and chlorinated
        dioxins.

        Since promulgation of  this regulation, commercial  laboratories have
        struggled to develop analytical protocols to  address this extensive list.
        In particular, a  number  of polar compounds which are not amenable  to
        conventional methodologies have presented challenges to the laboratory
        community.  A review of  the public docket to  the Appendix IX rule  making,
        funded research by EPA,  presentations at analytical methods caucuses and
        SW-846, indicate  that  these exotic compounds  still pose a challenge for
        routine analysis  (1).  Although several laboratories have evaluated
        conventional HPLC as an  analytical tool, this approach has not been widely
        accepted, due to  detection limit and identification reliability concerns,
        especially in contaminated matrices  (2).
                                            1-53

-------
Enseco's approach for Appendix IX analyses has been described elsewhere
(3).  The current approach provides reliable data for all Appendix IX
compounds except for five polar compounds.  These compounds are:
p-phenylenediamine, dimethoate, 4-nitroquinoline-n-oxide, famphur and
hexachlorophene.  Although it is possible to incorporate dimethoate and
famphur into method 8140, method 8270 was the recommended method for these
compounds.  Our recovery studies using method 8270 have consistently
indicated that these compounds are not measurable using method 8270 with
conventional sample preparation techniques (4).  Because of this, we have
developed the following LC/MS technique, combined with modified 8270-1 ike
sample preparation methods to provide reliable data for these compounds in
environmental matrices.

EXPERIMENTAL

Reagents and Chemicals

All analytes were obtained at 98% purity or higher from Aldrich, Cambridge
Isotope Laboratories and MSD Isotopes.  Standards at working concentrations
were verified using USEPA certified check standards, where available.
Labeled compounds were verified against their native counterparts.  All
reagents used were HPLC grade or equivalent. All standards stocks and
working concentrations are made in acetonitrile (p-phenylenediamine
degrades in methanol).  The compounds of interest are listed below and
their structures shown in Figure 1.

CAS Number     Target Compound

106-50-3       p-Phenylenediamine
60-51-5        Dimethoate
56-57-5        4-Nitroquinoline-n-oxide
52-85-7        Famphur
70-30-4        Hexachlorophene

Internal Standard:  Caffeine-13-C3

Surrogate Standards:  p-Phenylenediamine-d4
                      Malathion-dlO

The deuterated phenylenediamine (PDA) was chosen as a surrogate as the
native PDA has been the most difficult compound to extract, and can be
quite reactive.  Labeled malathion was included due to its similarity to
dimethoate and famphur.

Instrumental Conditions

The HPLC instrumentation consisted of a Hewlett Packard 1090L liquid
chromatograph with a ternary pumping system and a filter photometric
detector.  The UV detector was useful for off-line method development and
as a diagnostic tool. The LC was equipped with a variable volume injector
(2 uL was the nominal injection volume) and an autosampler.  The column
used was an Ultracarb ODS(30)  2x250mm manufactured by Phenomenex.  The
mobile phase consisted of water (modified with 0.01M ammonium acetate) and
methanol used in the following gradient at a flow rate of 0.20 mL/min.


                                   11-54

-------
                             Figure 1
CRS Number =  166563
CAS Number =  56575
CRS Number = 68515

            CH3
             I
             0
   CH3-0	P	S —CH2—C—NH-CH3
             ii              ii
             S              0
(taut oner)


CRS Number = 78384


      Cl       OH         Ci
                  CAS  Number =  52857


                       CH,    0

                          N—S=0
                                                        0
                                               CH3-0	P—0-CH3
                          OH       Cl
                                11-55

-------
       Water*   Methanol   Minutes

        50%       50%         0
         0%      100%         9

* with 0.01M arnmomium acetate

The particle beam interface used was a Hewlett Packard model 59980A.
Typical helium pressures were from 35 to 45 psi, desolvation chamber
temperatures from 45-55 degrees and nebulizer position was determined
experimentally based on flow injections of caffeine. The interface was
connected (via the standard transfer line) to a Hewlett Packard model 5988A
mass spectrometer operated in the electron impact ionization mode at 70 eV
and 300 uA emission current.  The ion source was operated at 250 to 300
degrees and the scan range 62 to 450 amu at sufficient speed to allow for
at least 10 scans per chromatographic peak.  The electron multiplier was
operated at between 2100 and 2300 volts.

As a starting point an autotune routine using perfluorotributyl amine
(PFTBA) can be used for mass spectrometer tuning.  It was often useful to
then maximize the tune on m/z 219 to provide good mid-mass sensitivity.
The instrument was tuned using the following guidelines for PFTBA.

       m/z     abundance

       69      100%
       131     25-75%
       219     25-75%
       502     >0.5%

These abundances will allow reasonable correlation with NIST or other El MS
libraries.  Mass peak width and axis calibrations are performed as needed.

The particle beam interface is optimized using manufacturer guidelines.
This performance is verified on a daily basis following tuning but prior to
the injection of calibration standard(s).  Several flow injections (column
bypassed) of 20 ng (2uL injections of a 10 ng/uL solution) of caffeine are
performed at 50:50 methanolrwater with 0.01M ammonium acetate.  Data is
acquired in the SIM mode monitoring m/z 194.   The peak areas are
integrated and evaluated for sensitivity and precision.  It was typical to
expect approximately 500,000 area counts with a precision of approximately
5%, injection to injection. This step establishes that the system is
functioning properly before doing any chromatography.

Calibration and Quantisation

There has been a great deal of discussion regarding quantisation in the
area of particle beam LC/MS (5,6,7).  We have found that the relationship
between concentration and response is not linear in the traditional sense
(e.g. GC/MS and the use of average response factors in environmental
analyses).  Instead,  this relationship is best described using a second
order polynomial expression.  Although this has not historically been the
approach for environmental analysis, data has recently been shown that the
                                   11-56

-------
accuracy of an existing method (based on linear calibration) can be
improved by utilizing a second order calibration (8). Once established,
quantitation can be performed using the second order equation, rather than
an average response factor.

The initial or multipoint calibration consists of a minimum of five points
covering one order of magnitude for each analyte and surrogate.  Plots of
concentration versus response (using extracted ion areas) are generated
following analysis and data processing of the points.  Given a reasonable
fit (r=0.95 or better), the data system is updated with the second degree
equations and the points requantitated against the curve.  The percent
difference between actual and theoretical concentration is calculated to
determine the quality of the calibration curve for each analyte.

The continuing calibrations consist of a midpoint level standard of all
analytes and surrogates, to be performed after the nebulizer performance
verification, but before sample analysis.  Again, concentration values are
calculated by the data system using the second order equations determined
in the initial calibration.  The percent difference between these
concentrations and the theoretical values are calculated for each compound
to determine if samples can be analyzed.  Other continuing calibrations (at
differing concentration levels) will be analyzed every five samples and
percent differences checked as before.

Background Subtraction

Due to the constant presence of background spectra, characteristic of the
interface, a background subtraction procedure was used to make low level
spectra identification easier.  First, a copy of the original data file is
made and archived so that an unaltered version will always be available.
Next,  spectra for ten scans (+/- 5 scans at 1 minute into the run) is
averaged and the resulting spectra is subtracted from each scan in the data
file.   (This spectra is also archived on the data system).  One minute was
chosen because this is before the void volume of the column has eluted as
was found to be representative of the background ions present.  A report is
generated showing the total ion chromatogram (TIC) before subtraction, the
spectra used for subtraction and the TIC after subtraction.

Sample Preparation

The preparation of water samples consisted of extraction with methylene
chloride by continuous liquid-liquid extractor and Kuderna-Danish (K-D)
evaporative concentration.  For the controlled matrix experiments, 1 liter
of de-ionized, carbon filtered water was spiked with the target compounds
and surrogates.  The pH was measured, and buffered at pH 7 with a potassium
dihydrogen phosphate/sodium hydroxide buffer.  This is added because pH
control is critical to the extraction of phenylenediamine.  Additionally,
sodium chloride (35 g) is added to the water to facilitate the extraction
of dimethoate.  Methylene chloride was then added and the extraction run
for 18 hours, as in method 3520, SW-846.  Following this, the extract was
concentrated to approximately 5 ml in a K-D.  The concentration was
continued to about ImL under a stream of nitrogen and then exchanged to
acetonitrile.  The extract was then evaporated to a final volume of 0.5 ml.
                                   11-57

-------
The sample preparation method for soils uses a 30 gram extraction using 1:1
methanoltmethylene chloride by sonication (as in method 3550), and
concentration by Kuderna-Danish (K-D).  For the controlled matrix
experiments, ottowa sand was spiked with the target compounds and
surrogates.  A 100ml aliquot of 1:1 methanol:methylene chloride was added
and the samples sonicated for 3 minutes at an output setting of 10 and a
duty cycle of 50%.  The extract was decanted and two more 100ml sonications
were performed.  The methanol/methylene chloride extract was then filtered
and concentrated to approximately 5 ml in a K-D.  The concentration was
continued under a stream of nitrogen and simultaneously exchanged to
acetonitrile.  The extract was then evaporated to a final volume of 0.5ml_.

RESULTS AND DISCUSSION

Several authors have discussed that mobile phase modifiers such as ammonium
acetate can enhance the MS response of some compounds when using the
particle beam interface (9,10).  Although we have also confirmed this
effect, ammonium acetate was also used for chromatographic reasons.  It was
found that without this modifier, p-phenylenediamine and hexachlorophene
showed very poor peak shape.   Two possible explanations are that the
ammonium acetate is acting as either an ion pairing agent, or simply
deactivating silanol sites on the stationary phase.  New columns must be
conditioned for several hours with the gradient described before acceptable
chromatography can be achieved.  Once this is done, however, all compounds
show good peak shape and separation, as shown in the total ion chromatogram
in Figure 2.

The numbers labeling the peaks in figures 2 and 3, as well as the numbers
labeling spectra in figure 4 all correspond to the following definitions:

        #   Compound

        1 - p-Phenylenediamine and p-phenylenediamine-d4
        2 - Dimethoate
        3 - 4-Nitroquinoline-n-oxide
        4 - Famphur
        5 - Malathion-dlO
        6 - Hexachlorophene

It is interesting to note that in the MS total ion chromatogram,
hexachlorophene exhibits significantly more peak tailing than in the UV
chromatogram shown in Figure 3.  This is likely due to memory effects in
the ion source itself.  This effect becomes more pronounced at lower source
temperatures (200-250 degrees) and diminishes as source temperature
increases (250-300 degrees).  Since a 300 degree source temperature had no
negative affect on other compounds response, much of the work was done at
this temperature.

All of the spectra obtained for these analytes show good correlation with
NIST spectra with one exception.  4-Nitroquinoline-n-oxide shows a base m/z
of 190 (the molecular ion) in the reference spectra, but this was not
obtained experimentally.  The base mass we obtained was m/z 144 and a small
(10% relative abundance) peak at m/z 190.  The difference of 46 is
                                   1-58

-------
                               Figure 2
File >PB89462.8-456.0 ami. 2UL-RP9  STD.5/9-180  fSB'SB MปOH/H20+.8ln HH
                            TIC

 1608001
 148880-
 120008-
 100800-
 80000-
 60000-
 48888-
 28880-
                   \
 188


-90


ซe


•70



•60


•50



r40


r30


-20


-18
         1.8 2.8 3.0 4.0 S.0 6.8 7.8 8.8  9.818.011.012.013.014.015.0
          UV  =  254 nm
-100



•99



ซe



•70



-68



-50



•40



-30



•20



10



0
      010 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.010.011.012.013.014.016.0
                               Figure 3

                                  11-59

-------
Figure 4
File >PB894 2UL-RP9 STD. Scan 69
Bpk Rb 22888. ENH 3.45 Bin.
25888} I08 I?2 r

28888:

16888-

leeoo-
6888-
a.
^^-
j
1

88
(
7 .1

{ 1 9l
-188

•80
•60

•48
-28
jป
88 188 128
Fil* >PB894 2UL-RP9 STD. Scan 149
Bpk Rb 8378. ENH 7.26 Bin.
-j 87 125 r
8888-


600B-

4888-
2008-
_
gj






!

1 i
•V







u
J




143 ,97
/ { 229
' f
LLk. 1.
-188

-68

-68

-40
-28

_o
188 150 208 258
Fil* >PB894 2UL-RP9 STD. Scan 115
Bpk Rb 15486.

16888-
-
1288*
-
8888-
_
4888-]



ENH 5.64 Bin.
197 r



111
68
/
[JJ
ie'0
*

168
\
/





223
•iee

ee
ซe

48
28
s
158 280 250
File >PB894 2UL-RP9 STD. Scan 178
Bpk Rb 12335.

12000-

•

8800-
-
4000-
-
B.





ENH 8.65 Bin.
144
*




117



88
s
1
jj \_ Jy^j

4




X
1 88
, /
-188

-60

-68

-48
-28

ja
128 168 288
File >PB894 2UL-BP9 STD. Scan 213
Bpk Rb 11328. ENH 18.32 Bin.
218 .
12888-


8888-
•
™
4888-
•
a_
^
5



93 125
\ f
1
1 C7 9 A 2
1 PB894 2UL-RP9 STD. Scan 245
Bpk Rb 8787.


8808-
-
6008-

4080-

2888-
:

6





72
/
L
'se'
ENH 11.84 Bin.
132 -
t



188
r


,
JU






158 183


•iee
ee

68

48
28
a
128 168 288
Fil* >PB894 2UL-RP9 STD. 5/9-188 |58/50 H*OHxH20+.81H NH40RC-> Scan 261
Bpk Rb 11482. ENH
12888-


8888-

4888-


1 198
7


89 '/
\ / 132
1 1 { lซ
; L
JLAL^JL^L^.




299
231 2f5 \
dLjL JL .Jl
88 128 168 288 248 288
12.61 Bin.






327 /5 3?e 486
^Ik. . ,< \
328 368 ' 480
-iee

•88

•68
-48
•20
•8
   11-60

-------
accounted for  by the loss  of the nitro group,  caused by  decomposition in
the  ion source.   The relative abundance of m/z 190 increases( with  a
decrease in m/z  144) at  lower source temperatures (225 or less).  Given the
reduction in overall response of all  compounds at lower  temperatures (and
the  increased  memory effect), source temperatures of 250 to 300  degrees
were best for  this analysis.  Spectra obtained from a midpoint standard are
shown in Figure  4.

Calibration and  Quantisation

Several  calibration curves have been run for  the analytes and surrogates
using from five  to ten points, covering nearly one order of magnitude.  The
concentration  versus response plots  indicate  that the relationships are
best described by second order polynomial equations.  An example of a 5
point calibration for 4-nitroquinoline-n-oxide is shown  in Figure 5.  The
correlation of these curves is usually 0.99 or better.   Once the data
system is updated with this information, the  curve can be requantitated
against itself and percent differences calculated. The data for  four
different calibrations is  shown in  Table 1.
                                    Figure 5
             Cซlib File I  CREBP9IIQT  Coปp

             Cซlib Datvi  910611 88|42
             Coapi 4-Nitroquinolinป-l-oxidป

                   328-
             Cane.
             ug/BL
248-


zee-


168-


128-



 88-


 48-
                               fivปrปgซ RF
                               lปt Dปgrปป
                               2nd Dปgrป*
                     8.8
                            .2
                                                            1.4
                                                                 1.6
                                     Response Ratio

            Compound I 4 Callb File: CA5AP9::QT

            Conpound: 4-N1troqu1nol1ne-l-ox1de
               Istd: !3-C3-Caffe1ne

            File: >PB895 >PB896 >PB894 >PB897 >PB898
            Cone:  20.00  50.00 100.00 150.00 200.00
              Rf: .22431 .44609 .54931 .58918 .70075

               Average of 5 Rfs: .50193  (35.86 X Rsd)   Rx: .0000000  Ry: .0000000
             1st Degree Equation: y - .2089807 + 1.339279(x)

            'gd^'g.tfi! ;"361066935 + 1.872203(x) +
            2nd Degree Corr Coef: .9996580

              In the above equations:

                 Cone Std        Area Std
              y -	      X •	
                 Cone Istd        Area Istd

              Istd Cone for all calibration points Is: 100.00

                                       11-61

-------
                                  Table 1
INITIAL CALIBRATION:

       Percent Difference Data
                                Curve "A"
                                Curve "B1
Compound

p-Phenylenediamine
Dimethoate
4-Nitroquinoline-n-oxide
Famphur
Hexachlorophene
Ave. %Diff.

   19
   5.9
   16
   4.7
   22
S.D.    Ave. %Diff
12
6.8
4.8
4.3
8.5
11
5.2
4.1
4.1
3.8
S.D.

5.3
3.4
3.4
3.4
3.2
                                Curve "C"
                                Curve "D"
Compound

p-Phenylenediamine-d4
p-Pheny1enedi ami ne
Diemthoate
4-Nitroquinoline-n-oxide
Famphur
Malathion-dlO
Hexachlorophene
Ave. %Diff.

   n/a
   2.7
   2.7
   0.72
   1.8
   n/a
   1.0
S.D.    Ave. %Diff.
n/a
2.2
2.5
0.93
1.8
n/a
0.89
3.1
6.1
8.6
1.3
1.9
4.8
1.5
S.D.

2.6
3.8
12
1.5
1.9
5.8
1.1
* For curves "A" and "B", the concentration range covered is as follows:
p-phenylenediamine; dimethoate and 4-nitroquinoline-n-oxide ranged from 40
to 180 ug/mL in 20 ug/mL increments; famphur covered 20 to 90 ug/mL in 10
ug/mL increments; and hexachlorophene ranged from 200 to 900 ug/mL in 100
ug/mL steps.

I For curves "C" and "D", the concentration range covered in five points is
as follows; p-phenylenediamine, p-phenylenediamine-d4 ("D" only) and
4-nitroquinoline-n-oxide ranged from 25 to 200 ug/mL; dimethoate and
malathion-dlO ("D" only) ranged from 50 to 400 ug/mL; famphur ranged from
12.5 to 100 ug/mL; and hexachlorophene ranged from 125 to 1000 ug/mL.
                                        11-62

-------
Based on the data in Table 1, the points fit the second order calibrations
very well.  However, one has to look at how well the continuing
calibrations over time, compare with the curves to know if this response
remains predictable.  Table 2 summarizes the percent differences obtained
for 16 standards run over a two week period after the analysis of curve
"A".  Concentrations were calculated by the data system using the second
degree equations and then percent differences were calculated comparing
these concentrations to true values.  The data shows that all percent
differences were less than 30% until the 13th day after the initial
calibration, indicating good stability of the initial calibration.  Based
on this data, one would have likely decided to establish a new initial
calibration on the 13th day.
                                  Table 2
CONTINUING CALIBRATION:

        Percent Difference Data
        2-M   2-L
            Day and Standard Level

            2-H   3-M   3-M   6-M   6-L
                                    6-H   6-M
PDA
DMT
NQO
FMR
HXN
0.2
22
3.6
12
12
0.6
1.6
26
4.6
3.2
6.3
8.6
6.4
0.4
1.7
5.8
14
13
16
8.4
3.6
18
9.5
20
6.5
11
12
15
7.0
20
6.4
5.0
7.1
3.0
10
12
4.6
19
3.8
1.6
14
15
9.9
11
16
        7-M   8-M   9-M   9-M   13-M  13-M  14-M    avq.%d1ff  S.D.
PDA
DMT
NQO
FMR
HXN
13
1.4
5.7
2.8
3.9
13
4.1
13
11
5.2
11
19
23
11
25
23
4.4
19
3.3
2.3
28
0.8
37
9.9
13
19
25
39
5.7
45
0.9
4.4
28
3.9
9.0
10
10
17
7.8
11
8.0
7.9
11
5.4
11
Notes:

The initial multipoint calibration (curve "A") was run on Day 1.

L = Low standard, usually half the concentration of the midpoint,
M = Midpoint standard, the middle of the calibration range.
H = High standard, usually 1.5-2 time the level of the midpoint.
                                   11-63

-------
Sample Extraction Results

Seven 1L replicates of carbon filtered, deionized water were spiked at the
levels shown below.  The average percent recovery and standard deviation
are also shown.

                                  ug/L
Compound                        Cone.Spiked     Avg.% Rec.     S.D.

p-Phenylenediamine-d4              50              59          16
p-Phenylenediamine                 50              59          15
Dimethoate                        100              86          5.8
4-Nitroquinoline-n-oxide           50              74          6.2
Famphur                            25              81          6.8
Malathion-dlO                     100              87          9.3
Hexachlorophene                   250              63          7.9


Similarly, seven controlled soils (Ottowa Sand) were spiked at the levels
shown below and the results expressed as average percent recovery and
corresponding standard deviations.

                                  ug/kg
Compound                        Cone.Spiked     Avg.% Rec.     S.D.

p-Phenylenediamine-d4             1670             64          23
p-Phenylenediamine                1670             65          21
Dimethoate                        3330             90          9.0
4-Nitroquinoline-n-oxide          1670             77          12
Famphur                            833             97          12
Malathion-dlO                     3330             97          14
Hexachlorophene                   8330             96          10


The levels at which the waters and soils were spiked, falls at the mid
point in terms of instrument calibrations and final extract concentration.
Because of this, the actual method detection limits  will likely be 2 to 4
times lower than the spiking level shown above.

SUMMARY

A method using particle beam LC/MS has been developed for the analysis of
some intractable Appendix IX compounds.  Together, with previously
established methods, it will be possible to measure the entire Appendix IX
list.  Calibration and subsequent quantisation is performed by taking
advantage of the second order behavior that appears to be characteristic
of the particle beam interface for these compounds.  The data demonstrates
that this approach provides a reliable method of initial calibration.
Furthermore, the analytical stability of the curves over several days has
been demonstrated.
                                   1-64

-------
It is noteworthy that quantisation by UV detection was not practical given
the poor UV response of two compounds and one surrogate.  Additionally,
the labeled PDA could not have been used as a recovery surrogate with UV
as the quantitation method.  Lastly, when "real" environmental samples
with various contaminants are analyzed, the interferences associated with
UV detection could make quantitation difficult.

ACKNOWLEDGEMENTS

Special thanks go to Todd Burgesser, Robert E. Moul, Bill Zdinak, Susan
Davis, and members of the Enseco-RMAL organic sample preparation group.
Thanks also to Jerry Parr for his valuable input.

Additionally, Hewlett Packard is to be acknowledged for their outstanding
level of technical assistance and support.

REFERENCES

1.)  "Summary of Analytical Methods for Appendix IX",RCRA Docket number
     F-87-AX9F-FFFFF.

2.)  Porter, J.W., "Technical Guidance on Ground Water Analysis for
     Appendix VIII", USEPA, February 1986.

3.)  Parr, J.L., et.al., "Establishing an Analytical Protocol for the
     Measurement of EPA's Appendix  IX List",  in Ground Water Quality and
     Analysis of Hazardous Waste Sites, Marcel Dekker Inc., In Press.

4.)  Capability of EPA Methods 624  and 625 to Measure Appendix IX
     Compounds, American Petroleum  Institute, Publication Number 4454,
     January 1987.

5.)  Budde, W.L., Bellar, T.A., Behymer, T.D., Analytical Chemistry, 1990,
     62, 1686-1690.

6.)  Winkler, P., et.al., "The Effect of Sample Concentration on PBI
     Response", Finnigan Technical  Report.

7.)  Brown, M.A., et.al., Analytical Chemistry 1991, 63, 819-823.

8.)  Parr, J.L., et.al., "Improving The Pesticide Method: A Laboratory
     Perspective", USEPA Analytical Methods Caucus, March, 1991

9.)  Bellar, T.A., Behymer, T.D., Budde, W.L., "Investigation of Enhanced
     Ion Abundances from a Carrier  Process in High Performance Liquid
     Chromatography Particle Beam Mass Spectrometry", J. Am. Soc. Mass
     Spectrometry, 1990,1,92-98.

10.)  Perry, L., "Effect of Mobile Phase Additives on Liearity in  Particle
     Beam LC/MS", Pittsburgh Conference, March 1991.
                                   11-65

-------
        THE IMPLEMENTATION OF HPLC/POST-COLUMN TECHNIQUES
          FOR RUGGED CARBAMATE AND GLYPHOSATE ANALYSIS
Michael V. Pickering, Ph. D. Pickering Laboratories, 1951 Colony
Street, Mountain View, California 94043; Michael W. Dong, Ph. D.,
The Perkin-Elmer Corporation, 761 Main Avenue, Norwalk,
Connecticut 06859-0250
ABSTRACT

This  paper  provides an overview of the fundamental concepts  of
post-column  derivatization techniques used in EPA Methods  531.1
(carbamates)  and  Method 547  (glyphosate).   Problem  areas  for
their practical implementation are described.  Specific solutions
leading to a more reliable analysis are discussed.


INTRODUCTION
Carbamates  are  broad spectrum pesticides which  exhibit  strong
cholinergic effects on insects.   Their low soil persistence  and
phytotoxicity,  make  them a favorite for food crop applications.
The  recent discovery of aldicarb (Temik  ) in the ground  waters
of  agricultural  regions has  prompted  the  U.S.  Environmental
Agency  (U.  S.  EPA) and other agencies to regulate pesticide use
and  require routine monitoring of drinking water and raw  source
water.  The recommended HPLC analytical method   (EPA method 531.1
for  drinking water and method 8318 for solid wastes) is based on
a   2-stage   post-column  reaction  followed   by   fluorescence
detection.  Carbamates are hydrolyzed at elevated temperatures by
sodium  hydroxide  to  provide  methylamine,  which  subsequently
reacts with o-phthalaldehyde (OPA) and 2-mercaptoethnaol (MCE) at
a  high  pH to produce a  highly  fluorescent  isoindole.    This
technique  has  excellent  sensitivity and selectivity  to  allow
direct  injection  of  drinking  water  samples  without   sample
enrichment or cleanup.

                                                       TM
Glyphosate  (N-(phosphono-methyl)-glycine)  or  Roundup  )   is  a
nonselective herbicide commonly used in post-harvest application.
Maximum   residue   tolerance  limits  for  glyphosate  and   its
metabolite aminomethylphosponic acid (AMPA) in various food crops
vary  widely  from 0.1 to 15 mg/kg.  Glyphosate  is  a  trivalent
negative  anion under neutral pHs (pK. =2.3) ,  though it can be
analyzed  by  cation exchange chromatography  under  acidic  pHs.
Analysis  of glyphosate according to EPA method 547 utilizes  the
same  HPLC  post-column  equipment used  in  carbamate  analysis.
                              1-66

-------
Hypochlorite  is used as the first post-column reagent to oxidize
glyphosate into glycine which is subsequently reacted with OPA to
form a fluorophore.


IMPROVED HPLC/POST-COLUMN TECHNIQUE

The  practical implementation of several improvements to  enhance
method  performance and ruggedness is discussed.   For  carbamate
analysis,  which  utilizes a 0.05 N sodium  hydroxide  hydrolysis
reagent,  the  prevention  of backflow of this reagent  into  the
silica-based  analytical column is critical (1).   For glyphosate
analysis,  the  replacement of the calcium  hypochlorite  oxidant
with  sodium  hypochlorite eliminates reactor  blockage  problems
(due  to the formation of calcium phosphate from the reaction  of
calcium ions with phosphate ions of the mobile phase).  Also, the
regeneration  of  the cation exchange column after each  analysis
with  5mM potassium hydroxide is necessary to maintain  retention
time reproducibility.


For both carbamate and glyphosate assays,   the incorporation  of
several  post-column  pressure  monitoring  points  and  pressure
relief   valves  in  the  system  significantly  enhances  system
reliability by aiding problem diagnostics and preventing  rupture
of  the  heated fluorocarbon reaction coil.    The use  of  guard
columns  is  mandatory  to prolong  analytical  column  lifetime.
Additional  sample  cleanup  (i.e.,  filtration  and  solid-phase
extraction)  are  required for some -water samples  and  vegetable
extracts.   The substitution of volatile 2-mercaptoethanol in the
ortho-phthalaldehyde   (OPA)  reagent  with the  nonvolatile  N,N-
dimethyl-2-mercaptoethylamine hydrochloride (ThioFluor  ) reduces
odor problems in the laboratory.   The use of borate salts  which
contain  high  levels of insoluble matter,  should be avoided  in
preparing  the OPA buffer.   Boric acid,  available  in very  pure
form,  should  be used instead and adjusted to pH 10 with  sodium
hydroxide.    The teflon tubing in the OPA reagent line should be
replaced by Saran tubing which prevents oxygen permeation  (causes
OPA  degradation).   The  proper operating  sequence  for  system
start-up  and  shutdown  is also important  to  prevent  possible
reagent precipitation or system damage  (2).


REFERENCES

1.  M.  V.  Pickering,"Assembling  an  HPLC  post-column  system:
practical considerations," LC.GC, 6, 1988, 994-997.

2. M. W. Dong, F.L. Vandemark, W. M  . Reuter and M. V. Pickering,
"Analysis of carbamate pesticides by LC," Amer. Environ. Lab.,
2(3), 1990, 14-27,
                               11-67

-------
52
^"~   DETERMINATION OF LOW-LEVEL  EXPLOSIVE  RESIDUES IN  WATER BY HPLC:
          SOLID-PHASE EXTRACTION VS. SALTING-OUT  SOLVENT EXTRACTION
      Michael  G. Winslow.  Manager,  Organic  Analytical  Division,  Bradley A. Weichert,
      Manager, GC/HPLC Department, and  Robert  D.  Baker,  Senior  Associate Scientist,
      GC/HPLC Department,  Analytical Services, Environmental Science & Engineering, Inc.,
      P.O. Box 1703,  Gainesville,  Florida  32602.

      ABSTRACT

      The December 1990 SW846 draft protocol for the  determination of low concentrations
      (1-50 u,g/L) of nitroaromatic and nitramine compounds in ground and surface water by
      high pressure liquid chromatography (HPLC) (Method 8330) proposes the use of a
      salting-out solvent extraction technique using sodium chloride (NaCI)  and acetonitrile
      (ACN), followed  by a Kuderna-Danish  extract concentration.  This sample preparation
      procedure was developed and validated by the  U.S. Army  Cold Regions Research and
      Engineering Laboratory (CRREL) for 8  selected analytes - RDX,  1,3,5-trinitrobenzene
      (1,3,5-TNB),  1,3-dinitrobenzene  (1,3-DNB),  2,4,6-trinitrotoluene  (2,4,6-TNT),
      4-amino-2,6-dinitrotoluene   (4-Am-2,6-DNT),  2-amino-4,6-dinitrotoluene  (2-
      Am4,6-DNT),  2,6-dinitrotoluene  (2,6-DNT),  and  2,4-dinitrotoluene  (2,4-DNT).
      The adoption  of this procedure  in Draft Method 8330, which includes six additional
      analytes - HMX, nitrobenzene (NB), tetryl, and the 2,3,4-isomers of  nitrotoluene (2
      -NT, 3-NT,  4-NT)- should be  assumed to be applicable only to the  eight analytes
      validated by CRREL, because its applicability to the six additional  analytes  is not
      supported with experimental data.   A  discussion of this salting-out solvent procedure
      and the  results  of laboratory analyses  applying it  to the determination of all fourteen
      target analytes are presented.

      An alternative sample preparation  procedure  for  the  determination  in water  of low
      concentrations of  all fourteen  nitroaromatic and nitramine compounds listed in  Draft
      Method 8330 is proposed.  This procedure, which has been routinely used to determine
      explosive residues in water samples for the  Army, uses  Porapak R solid sorbent for the
      extraction of explosive residues from water samples. A discussion of this solid-phase
      procedure and the results of laboratory analyses are presented.

      Broad scope applicabilty, ease of use,  and cost effectiveness are three factors which
      should be considered when adopting an analytical method or procedure  for inclusion in
      SW-846.   In proposing the  salting-out extraction procedure in Draft Method 8330 for
      the determination  of low concentration explosive  residues in water samples, these
      factors seem  to have  been  neglected.  For comparison,  the solid-phase extraction
      procedure, which does successfully test itself against these three factors, is presented.

      INTRODUCTION

      Nitroaromatic  and nitramine compounds are some the most widely  used munitions
      components.  They have been and continue to be produced in large quantities and are
      therefore, along  with certain of their degradation  products and production impurites,
      subject to environmental regulation.  The primary concern  has been the contamination

                                              II-68

-------
of ground and surface waters near ballistic test ranges and munitions processing and
storage facilities.  In  recent years, the EPA has issued health advisories on several of
these compounds in drinking water (2,6-DNT 2,4-DNT, TNT, RDX).  The result has
been the need for an analytical  method that can routinely  acheive detection limits of 1
     and less for the  majority of  the compounds of interest.
Gas chromatographic (GC)  methods  have been used effectively to  detect nitrated
munitions components  with  excellent  sensitivity and  selectivity, especially  when
employing the electon capture detector.  However, these methods have been applicable to
only a limited number of target analytes, for various reasons:   partial or complete
degradation of thermally  labile species; loss due to volatilization of some species during
extract concentration; the difficulty in selecting a single organic extraction solvent.

The use of HPLC with UV detection has become the preferred method for the analysis of
wide range of munitions compounds.  In order to achieve detection limits less than 1
|ig/L in water samples,   sample concentration  prior to  HPLC analysis is required.
SW846 Draft Method 8330  has proposed a  salting-out  solvent extraction procedure
using NaCI and ACN.  It is the intent of this paper to  suggest that  a solid sorbent
extraction  procedure using  the hydrophilic resin Porapak  R  warrants strong
consideration  as the preferred  extraction  procedure for incorporation into Method
8330.  Tests  employing both  procedures are described below, and  the results  are
presented for comparison and discussion.

EXPERIMENTAL

Analytical standards  were prepared  from Standard Analytical  Reference Materials
(SARMs)  obtained from  the  U.S. Army Toxic and Hazardous Materials Agency
(USATHAMA), except for the three nitrotoluenes and the surrogate 3,4-DNT which were
obtained from Aid rich Chemical Co.,  and the two amino-dinitrotoluenes which were
obtained from the Naval Surface Weapons  Center  (NSWC).  For each of the  two
extraction procedures, ten 500-mL samples were  prepared at ten  concentration  levels
on four consecutive days  - a total of  40 samples per procedure were analyzed. The
laboratory samples were prepared in ASTM Type  II/HPLC grade water.  Target analytes
were spiked at the levels indicated in Table  1.  Concentration level x represents the
target limit of detection.  In addition, each of  the laboratory samples was spiked  with a
surrogate compound,  3,4-dinitrotoluene,  at approximately 5  u.g/L.  Calibration
standards were prepared in 30% ACN/70% H20 so that the UV responses of each  of the
target analytes would  bracket the predicted responses of the target analytes in the final
extracts of the spiked laboratory samples.

Salting-Out Solvent Extraction  Procedure

A 400-mL aliquot of water sample was placed into a 500-mL separatory funnel and
shaken vigorously with 130 g of NaCI until the NaCI was completely dissolved. 100 ml
of acetonitrile (ACN) was added to the separatory funnel and the contents were shaken
for 5 min. The phases were then allowed to separate for 30 min. The lower water layer
was then drained off and discarded.  The upper layer (~23 ml) was collected in a 25-mL
Kuderna-Danish  (K-D) receiver.  The separatory  funnel was rinsed with 5 mL of ACN
and the rinsate was added to the extract in the receiver. (If the ACN extract is turbid, it
                                      I-69

-------
should be  transferred  to a 40-mL centrifuge tube with teflon-lined screw cap and
centrifuged at 4000 rpm for 5 min.  The ACN  layer is  then removed with a  pasteur
pipette to the 25 mL K-D receiver.)  The receiver was then fitted with a a micro (40
mL)  K-D flask and modified two-ball micro snyder column.  The ACN extract was
reduced to less than 1.0 mL and brought to the 1-mL mark with ACN.  The extract was
then diluted to a final  volume of  4 mL with  ASTM  Type II/HPLC water and filtered
through a 0.45 uM Teflon filter.  The first 0.5 mL was discarded.  The remaining filtrate
was  then ready for HPLC  analysis.  The salting-out solvent extracture  procedure  is
summarized in Table 2.

Solid-Phase Extraction  Procedure

An empty 6-mL Baker Disposable Extraction Column with a 20-p.M frit at the bottom
was  packed with 0.5 g  of cleaned  80-100 mesh  Porapak R.  Another frit was placed  at
the top of the sorbent bed to assist  packing and help prevent channeling. The column was
first conditioned with 15 mL  of ACN followed by 30 mL of ASTM Type II/HPLC water.  A
500-mL aliquot of water sample was passed through the column at 10 mL/min. utilizing
a  Visiprep Solid-Phase Extraction  Vacuum Manifold (Supelco).

Figure 1  plots the results of an experiment to determine the optimum sample flow rate
through the extraction system. Five representative target analytes were spiked at 40-
50 ug/L into 500 mL of  ASTM Type II/HPLC water.  Duplicate samples were extracted  at
five different flow rates ranging from 2 to 50 mL/min.

This system allows 12  samples to be processed simultaneously in  about 50 min.  The
sorbent column was then eluted with 3 mL of ACN at <. 3 mL/min. into a graduated
centrifuge tube.  The ACN  eluent  was concentrated to 2 mL under a gentle stream  of
nitrogen.  The eluent was diluted  to a final volume of 6 mL with  ASTM Type  II/HPLC
water  prior  to  HPLC  analysis.   Table 2  summarizes the solid-phase extraction
procedure.

HPLC Analysis

A Shimadzu model LC-6A high-pressure liquid chromatograph (HPLC) equipped with a
Shimadzu  SPD-6A autosampler  and a  Kratos 757  variable  wavelength ultraviolet
absorbance (UV) detector set at 250 nanometers was used for analysis of the ACN/H20
extracts.  The samples were eluted from a 25 cm x 4.6 mm I.D. Phenomenex ODS  (5-u.M
particle size) reverse-phase column.  Analyses were  performed isocratically with a 55
% methanol/45 % H2O (V/V) mobile phase at a 0.8 mL/min. flow rate.  Analytical runs
lasted 30 min., the last target compound eluting  at about 25 min.  Refer to Figure 3,  4,
and 5 for calibration standard and spike sample chromatograms.  The injection volume
was 500  uL. The instrument operating conditions are summarized in Table 2.  Data was
collected and quantitated using a Nelson 2700 Turbochrom data system.

Calculations

The percent recovery for each target analyte in the spiked water samples was calculated
by comparison of the  calculated concentrations  to  the target concentrations.   The
calculated concentrations were obtained from the initial calibration quadratic regression
                                      II-70

-------
equations.   Calibration standards were analyzed at a minimum of five concentration
levels  with responses that bracketed the responses of the samples. The lower limits of
detection for each procedure were determined by the Certified Reporting  Limit (CRL)
test used by USATHAMA (1990) rather than by the  Method Detection Limit (MDL) test
outlined by EPA (Federal Register 1984).  For an excellent comparison of the two tests,
see Reference 2.

RESULTS AND DISCUSSION

The accuracy and precision data for both sample preparation procedures are presented
in Table 3.  Percent recovery outliers were eliminated from the calculation of the mean
percent recovery of each target analyte.  Figure 2 graphs a comparison of the mean
percent recoveries for each procedure.  Table 4 lists the CRLs for each procedure.

For both procedures, the mean  percent recoveries exceeded 70% for all  analytes.
However, the overall accuracy of the solid-phase procedure  was significantly greater
than that of the salting-out solvent procedure. The average percent recovery of the solid-
phase  procedure (94.3) exceeded that of the salting-out solvent procedure  (84.6) by
nearly 10%.   The overall precision of both  procedures were very similar, although the
average standard deviation of the solid-phase procedure  (10.3) was nearly  one percent
point lower than that of the  salting-out solvent procedure (11.2).  As a result of these
differences in accuracy and  precision, for all analytes except 2,4,6-TNT, the calculated
CRLs were significantly lower for the solid-phase procedure.

From  a comparison of  the test results, it is clear that the solid-phase procedure, when
applied to  laboratory water samples  spiked with the 14 nitroaromatic and  nitramine
munitions  compounds listed  in SW846 Draft Method 8330, performs  better  than the
salting-out  solvent  procedure proposed  in  the  method.  But this is  not surprising,
considering  the  complexity and labor  intensive nature of the  salting-out procedure. It
required 16  hours using the  salting-out procedure to prepare a batch of 20 samples for
HPLC  analysis.  It required  only 6  hours using the  solid-phase procedure. Further, it
was quite surprising that the test results for the salting-out procedure were as good as
they were.  This would not be predicted  for a procedure  that applies considerable
amounts of  heat to a group of target analytes containing species that  are either very
thermally labile, such as tetryl, or quite volatile, such as the nitrobenzenes.

In light of  the above,  it  is strongly  recommended  that the  solid-phase  sample
preparation procedure using Porapak R be  given  consideration for adoption in SW846
Method  8330   as  the  sample  preparation procedure for determination  of  low
concentrations of nitroaromatic and nitramine compounds in water.

SUMMARY

For the determination  of low concentrations of fourteen nitroaromatic and  nitramine
compounds in  water,  SW846  Draft  Method 8330 (Revision  1,  December 1990)
proposes  a  salting-out solvent extraction  procedure  using sodium chloride  and
acetonitrile.  An alternative procedure, which employs the solid sorbent Porapak R for
the extraction  process, is presented.  Both extraction procedures are tested on spiked
water samples.  The sample extracts are  analyzed  by HPLC with UV  detection.  The
                                       1-71

-------
analytical results from both extraction procedures are presented for comparison.  The
advantages for  inclusion of the solid-phase extraction procedure into SW846 Method
8330 are discussed.

ACKNOWLEDGEMENTS

We thank P. Dumas, D. Dabney, and S. McMillen for their laboratory assistance.

REFERENCES

1. Environmental Science & Engineering, Inc., "Development of Standard Analytical
   Methods for the Analysis of Explosives in Water by High Pressure Liquid
   Chromatography  Precertification/Certification Report and  Method Writeup", March
   1991, Contract No. DAAA15-90-D005, U.S.  Army Toxic and Hazardous Materials
   Agency, Aberdeen Proving  Ground, Maryland.

2. C.L Grant, A.D. Hewitt, and T.F. Jenkins, "Experimental Comparison of EPA and
   USATHAMA  Detection and Quantitation Capability Estimators". American Laboratory.
   15-33,  February,  1991.

3. B. Lesnik, "The HPLC Methods Development Program:  An Overview", Environmental
   Lab.  18-41, April/May,  1990.

4. M.P. Maskarinec, D.L. Manning, R.W. Harvey,  W.H. Griest and B.A. Tomkins,
   "Determination of Munitions Components in  Water by Resin Adsorption and High-
   Performance liquid Chromatography-Electrochemical Detection," Journal of
   Chromatography. 302, 51-63, 1984.

5. P.H. Miyares and T.F. Jenkins, "Salting-Out Solvent Extraction Method for
   Determining  Low Levels of Nitroaromatics and Nitramines in Water", Special Report
   90-30,  U.S. Army Corps of Engineers, Cold Region Research & Engineering
   Laboratory,  1990.

6. SW846 Draft Method 8330, "Nitroaromatics  and  Nitramines by  High Pressure
   Liquid  Chromatography (HPLC)",  Revision 1, December 1990, U.S Environmental
   Protection Agency, Office of Solid Waste and Emergency Response, Washington, D.C.

7. USATHAMA, U.S. Army Toxic and Hazardous Materials Agency, Quality Assurance
   Program, January  1990.
                                     II-72

-------
Table 1    TARGET ANALYTE CONCENTRATIONS fag/L)

J
H0C
•~fv '•
ซK '- , •
1,3^TNฃ
1f^f>N8
Teiryj
"•;•••••• •.
iป *;",\~ ^
ฃ,4>TNT
% f •. •.
• t : :
4ปAJR*&&8NT
•sj
ฃปAffi *2*443i NT
^DNT .> .
: ^
2J4HDHT ^
t^T , \ '
4-Mf
3-NT
OX
0
0
0
0
0
0
0
0
0
0
0
0
0
0
JX ,
0.302
0.292
0.148
0.138
1.08
0.645
0.28
0.052
0.055
0.061
0.053
0.307
0.301
0.292
X. • ;
0.604
0.584
0.296
0.275
2.15
1.29
0.56
0.104
0.11
0.122
0.106
0.613
0.602
0.584
2X
1.21
1.17
0.592
0.55
4.3
2.58
1.12
0.208
0.22
0.244
0.212
1.23
1.2
1.17
6X
3.02
2.92
1.48
1.38
10.8
6.45
2.8
0.52
0.55
0.61
0.53
3.07
3.01
2.92
10K
6.04
5.84
2.96
2.75
21.5
12.9
5.6
1.04
1.1
1.22
1.06
6.13
6.02
5.84
aox
12.1
11.7
5.92
5.5
43
25.8
11.2
2.08
2.2
2.44
2.12
12.3
12
11.7
$OX i
30.2
29.2
14.8
13.8
108
64.5
28
5.2
5.5
6.1
5.3
30.7
30.1
29.2
t&OX
60.4
58.4
29.6
27.5
215
129
56
10.4
11
12.2
10.6
61.3
60.2
58.4
200X
121
117
59.2
55
430
258
112
20.8
22
24.4
21.2
123
120
117
                                II-73

-------
Table  2 SUMMARY OF EXTRACTION PROCEDURES AND CHROMATOGRAPHIC
         CONDITIONS
                    SALTING-OUT SOLVENT EXTRACTION
         400 ml water sample in a 500 mL separately funnel.
         Add and dissolve 130 g of NaCI.
         Add 100 mL ACN and shake for 5 min.
         Let phases separate for 30 min.
         Discard lower water layer; recover ACN; (-23 mL) in a 40 mL vial.
         Rinse separatory funnel with 5 mL ACN; recover in the 40 mL vial.
         If ACN extract is turbid, centrifuge at 4000 rpm for 5  min.
         Remove ACN layer with a pasteur pipette to a K-D  evaporator.
         Reduce to < 1.0 mL  and bring to 1.0 mL with ACN.
         Dilute with 3.0 mL reagent water to 4.0 mL final volume.
         Filter through 0.45  u.M Teflon filter; discard  first  0.5  mL.
         Analyze by RP-HPLC/UV.
                         SOLID-PHASE EXTRACTION
          6 mL Disposable Extraction Column packed with 0.5 g of cleaned 80-100mesh
          Porapak R.
          Precondition column first with 15 mL ACN and then with 30 mL ASTM Type
          II/HPLC water.
          Measure 500 mL of water sample and pass through column at 10ml_/min.
          Elute column with 3 mL ACN at < 3 mL/min. into a graduated centrifuge tube.
          Concentrate eluent to 2 mL under a gentle stream of nitrogen.
          Dilute to 6 mL final volume with ASTM Type II/HPLC water.
          Analyze by RP-HPLC/UV.
                      CHROMATOGRAPHIC CONDITIONS
          Column: Phenomenex ODS reverse phase HPLC column, 25-cm x 4.6-mm, 5-
          Mobile Phase:  Isocratic, 55% methanol/45% water (V/V).
          Flow Rate: 0.8 mL/min.
          Injection Volume:  500 uL.
          UV Detector:  250 nm.
                                      I-74

-------
Tables  PRECISION   AND  ACCURACY  DATA
Salting-Out
Solvent Extraction
IftK
mx
t^TNB
1,3-DNB
tEraYL . .
MB ....
3,4-DNT (SUR)
2t4h6"TMT
4'Attf-2,e*DNT
2'AMซ4>6ปDNT
2,$*ONT
2,4*ONT
2*NT
4-MT
3-NT
Solid-Phase
Extraction
HVIX
BOX
tf%5"TNB
1,3-DNB
TEmYL
N8
3,4-DNT (SUR)
2^eปTNT
4*AM'2,e'DNT
2*Ay*4,e*D*JT
2,$>DNT
2,4'ONf
2-NT
4-NT
3ปNt
MEW
$RBฃ
91.8
86.7
70.8
83.9
80.1
81.5
aus
89.4
126
78.4
83.6
76.7
76.6
75.1
77.3
STDDEV
5.5
6.2
18.6
8.6
14.5
9.1
12.5
14.1
19.1
11.2
12.8
9.9
7.3
7.1
11.1
RANGE
84.2 - 106
67.2 - 98.6
44.3 - 98.3
64.5 - 103
46.5 - 103
67.8 - 109
70.0 - 121
64.5 - 123
101 - 167
54.9 - 102
58.6 - 117
56.4 - 99.2
63.4 - 103
63.2 - 95.7
62.6 - 96.2
K
36
36
36
36
36
34
21
35
19
35
36
35
34
34
35

97.8
95.5
84.7
97.1
91.1
92.7
102
96.8
123
92.2
90.3
85.2
91.9
84.1
90.8
4.1
6.6
16
6.5
11.6
7
8.3
12.6
18.3
12.4
11
9.1
8
10
12.6
87.9 - 106
81.3 - 109
54.5 - 105
78.8 - 112
73.6 - 120
77.4 - 108
69,4 .- 125
75.4 - 123
96.8 - 164
68.2 - 117
64.4 - 108
66.2 - 101
67.7 - 106
63.6 - 96.3
69.4 - 129
36
36
35
36
36
36
40
35
35
35
35
36
34
36
32
                           11-75

-------
Table 4    CERTIFIED   REPORTING   LIMITS  (UG/L)

HMX
FOt
1^5-TNB
1>DMB
TEm^L
NITROBENZENE
2t4^7#T
4*AM-2re>I>NT
2-AM-4>e'DNT,
2,6-DMT
2,4*DNT
2ปNT ^ ^
4*NT .
3*NT
$OOD^PHA$E
0.3
0.29
0.45
0.15
2.49
0.65
0.64
1.57
0.16
0.074
0.064
0.41
0.62
1.4
SALTiNO-OUT
BOLVENFF
0.45
0.64
0.75
0.38
4.07
4.4
0.57
3.98
0.86
0.123
0.088
2.11
2.07
2.03
                    II-76

-------
     100
      95
      90
%Rec  85
      80
      75
      70
                 Figure 1   PERCENT RECOVERY vs FLOW RATE
                               4          1 0
                                  mL/min
20
                                                                     RDX
               -*- NB
               •O TNT
               -*• 2,4 DNT
50

-------
                            Figure 2   PERCENT RECOVERY COMPARISON
•si
CXI
                         50
                                               SOLID PHASE


                                               SALTING OUT
60
70
80
 90

%Rec
100
110
120
130

-------
       FileNane   :  D:\2700\DATA\HTRC012.raw
       Start Tiie  :  0.00 lin        End Tile  : 30.00 lin
       Scale Factor:   0            Plot Offset: 0 งV
Date :  5/16/9!  2:58 PM
Low Point :  -0.00 nV
Plot Scale:  100 iV
Hioli Point
 Page 1 of 1
:  100.00 iV
                                                    CNซO
>
 E
 D
 U)
 C
 o
 Q.
 V)
 0>
                                                                                                         ntf/mL
                            1,3,5-TNB
                            1,3-DNB
                            TETRYL
                            NB
                            3,4-DNT
                            2,4,6-TNT
                            4-Am-2,6-
                            2-Am-4,6-
                            2,6-DNT
                            2,4-DNT
                            2-NT
                            4-NT
                            3-NT
1
X
I
I 1 I 1 1 !
1 5
i
X
o
a:
1 1
I
CD
t—
in
n
i i i
UJ
ISJ
1-,!
T-~}^ 2
~TTm7~r
Retention
1 1 1 1 1 1 i
CN i" ^
H- 1 1 P
ZT Z Z Z Z ฃฃ
O 1 23 OO f-:
rOCN? -^ CMCNCNI CM
1 1 1 1 1 1 1 1
15 20
Time [min]
i
ir
z
I
1
I
0
cr
i
' 2J5 ' ' ' '
                Figure 3    CHROMATOGRAM OF CALIBRATION STANDARD
                                                      1-79

-------
C h
                                                      Hb <:• g -r ฃ
       FileNaie    : D:\2700\DATA\SALT021.rav
       Start Tiie  : 0.00 tin        End Tiie  ; 30.00 tin
       Scale Factor:   0            Plot Offset: 0 ซV
                  Date : 5/16/91  3:26 PH
                  lav Point : -0.01 iV
                  Plot Scale: 50  iV
High Point
 Page 1 of 1
: 49.99 iV
      45-
>
 CD
 CO
 C
 O
 a
 t/i
 (D
QL
                          r- CM
1
X
2
X
1 1 1 I 1 1
b
I I
m
•z.
\
x in-
o K!
ce •ซ-
t
m
•
^—
TEJRYL
NITROBENZE -
1 1 1 1 1 1 1 1
Retention
i i
1:
KTCM
iiit
CM •<*-"
1 1
O O t— t—
^ =500
< < ' 1
1 I <ฐ*.
•M- CM CMCM
I ' '
Time
1 I
2
[mi

I
n

l
]
2-NITROTOL -
1
i
o
5
a:
5E
\
I
o
s
a:
7Z.
1
to
1 I 1 1 1 1
25
             Figure 4    CHROMATOGRAM OF SALTING-OUT EXTRACTION SAMPLE
                                                     1-80

-------
      FileNaw   I Ds\2700\DATA\HTRC039.ran
      Start Tite : 0.00 tin       End Tiie  : 30.00 nin
      Scale Factor:   0           Plot Offset: 0 iV
                                            Date : 5/16/91 3:06 PH
                                            Low Point :  -0.00 iV
                                            Plot Scale:  50 tV
High Point
       Page 1 of 1
      : 50.00 iV
     45—
     4O-
     35-
     30-
 QJ
 CO
 C
 O
 Q.
 V)
 0)
o:
2O-
      15-
      10-
                       oooo
                        I  I 1 1
                        lปป
                         I
                                 co CM
                                 10 r^_
                                 oC) O'
                                  I   I
IO-— U"> U")CN CM CO CO ID
oq^r ro ojcq CD in *ป;ซ-;

I'" "~ "~"~ ~~ "~ ~~~
                                               I  ill  I  II
                                                                        O
7
      7
                                                                                      a
                                                                                      CT
                                                         Compound

                                                         HMX
                                                         RDX
                                                         1,3,5-TNB
                                                         1,3-DNB
                                                         TETRYL
                                                         NB
                                                         3,4-DNT
                                                         2,4,6-TNT
                                                         4-Am-2,6-DNT
                                                         2-Am-4,6-DNT
                                                         2,6-DNT
                                                         2,4-DNT
                                                         2-NT
                                                         4-NT
                                                         3-NT
         UQ/L

         3.02
         2.92
         1 .48
         1 .38
         10.8
         6.45
         4.94
         2.80
         0.52
         0.55
         0.61
         0.53
         3.07
         3.01
         2.92
1




X
^
X
1 1 1 1 1 1
ct
1




X
O
tt
1 1
1
m
z
H~

**!
*^
T-
' 1
1 1 1
UJ
INJ
LU , ,
"^^ 111
ง— ' Q-1

1 ttf CK
^i fTj i
'— t- Z
III!
I I I
ป
cst
1- 1
P—Z o
d^.' . Z
o ^ ^
1 ™ <;
"^"^ |
r^CM •*
1 1
1 1 1
*
1
OH- i-
"Z. "Z. "Z.
^K ฃ$ C)
< 1 1
1 _ .
CM CM CM
1 1 1
1
O
c5
cr

z
i
CM
1 ' '
1
O
5
cr
t:
z
i
•^
1
i
o
5
o:
t=
z
i
•o

                                   Retention  Time  [min]
           Figure 5   CHROMATOGRAM OF SOLID-PHASE  EXTRACTION SAMPLE
                                             11-81

-------
53       REDUCTION OF AZO DYES TO AROMATIC AMINES FOR ENVIRONMENTAL MONITORING

       Robert  D.  Voyksner  and Jeffrey  T.  Keever,  Analytical  and Chemical  Science,
              Research Triangle Institute,  P.  0.  Box 12194,  RTP, NC  27709
           Harold S.  Freeman  and W.  N.  Hsu,  North Carolina State Uuniversity,
                              Box 7003,  Raleigh,  NC  27695
                Leon D.  Betowski,  EMSL US-EPA,  Las Vegas,  NV  89193-3478

          Azo  dyes are of  great environmental  concern due to their potential  to form

      carcinogenic aromatic amines under  reducing conditions.   As a result,  it is

      necessary to evaluate both the intact molecule and its potential  reductive

      cleavage products to adequately  assess the  potential risk of a dye  stuff to  the

      health of man and the environment.   With over 100 million pounds  of azo dyes

      produced annually, the development  of a  method that effects the reductive

      cleavage products in vitro and permits their characterization would aid in

      determining modern complex and structurally unknown dyes and their  genotoxicity.

          A logical approach would involve the evaluation of procedures for the

      reductive cleavage of azo dyes followed by  mass spectroscopic (MS)  analysis  of

      their products.  To best determine the approach in achieving this goal, the

      reduction of representative samples from several of the major azo dye classes

      (e.g. disperse and solvent dyes), was accomplished using chemical means.  Two

      chemical procedures  were evaluated for the  reduction of azo dyes.  The first

      reduction agent, SnCU, is especially important in the reductive  cleavage of azo

      linkages in the presence of other easily reduced groups such as a nitro group.

      The second method involved sodium dithionate, Na2S20., which has  been used to

      effect the reductive cleavage of water soluble azo dyes for decolorizing

      purposes and for Salmonella bacterial assays for mutagenicity.

          Initial screening of the various reduction products from each procedure was

      accomplished using thin layer chromatography (TLC).  Confirmation of the

      postulated reduction products involved a combination of particle beam and

      thermospray LC/MS and GC/MS.  Standards of  the proposed reduction products and


                                            11-82

-------
aromatic amines of the products tentatively identified, when available,  were
employed to confirm identities.  It was observed that the chemical reduction
methods resulted in nearly 100% reduction of the azo bond to form the
characteristic amines for the 16 dye standards evaluated.  Overall the SnCl2
method was a more powerful reducing agent yielding a greater number of products.
In addition to the reduction of the azo bond,  dyes containing acetate groups
exhibited both acid and base catalyzed hydrolysis of the ester groups to form
the respective alcohols.  The presence of electron withdrawing halogen groups on
the aromatic ring appear to make the nitro groups more susceptible to reduction.
Also small yields of N-dealkylation products were observed for SnClp reduction
of some dyes.
    Current efforts are evaluating the use of chemical reductions for
determining aromatic amine content of wastewater, sludge and sediment
contaminated with azo dyes.  Reduction conditions using SnCl2 or Na2S204 needed
to be modified to  insure complete reduction of azo dyes  in wastewater and
sludge.  Variables, including  the presence of sediment in a sample, temperature,
reaction time and  amount of reductant,  influence  the yield of the aromatic
amines.  The analysis of the reduced azo dye samples provided an estimate of
total dye  (amine)  content, but identity of specific azo dye could not be
determined.  Most  mono and diamino reduction products  could be analyzed by
GC/MS.  More polar reduction products containing  three amine groups or multiple
function groups  (e.g. SO,, OH,  NO,,) were best analyzed by LC/MS.
                         O        ฃ
    Although the information described  in this article has been funded wholly or
in part by the Environmental Protection Agency under contract 68-02-4544 to
Research Triangle  Institute, it does not necessarily reflect the views of the
Agency and no official endorsement should be inferred.
                                      1-83

-------
54             HAZARDOUS WASTE COMPONENT IDENTIFICATION
                  USING AUTOMATED COMBINED GC/FTIR/MS

       Roger   J.    Leibrand,   Scientific  Instruments  Division,
       Hewlett-Packard Co, 1601 California Avenue,  Palo  Alto,  CA
       94304

       INTRODUCTION

       Our modern industrial civilization contains many items which
       have the potential  to  produce  hazardous  waste  in  their
       production or use.  In general, plastics, paints, petroleum,
       petrochemicals, leather, textiles, pesticides, and medicines
       can generate volatile solvents, reaction residues, oils, and
       non-volatile heavy metals, dyes, pigments, salts, acids, and
       caustics.   To properly dispose of these hazardous wastes is
       a complex problem.   The  first  step  in  this  process  is
       determining  what  is  there, i.e., chemical analysis of the
       waste   sample.    The   scheme   for   identification   and
       quantitation  of  certain  specific toxic compounds is quite
       well delineated in the U.S.  Environmental Protection Agency
       methodology.    However,  tentatively  identified  compounds
       (T.I.C.s) and other hazardous components that are non-target
       compounds   in   the  E.P.A.   methods  are  not  accurately
       determined by those methods.  These contaminants need to  be
       identified  before  proper  disposal  of hazardous waste can
       occur or a waste site can be remediated.

       Qualitative  analysis of organic pollutants currently relies
       heavily on the mass spectrometer,  both  in  its  GC/MS  and
       LC/MS   forms.   The  mass  spectrometer  produces  powerful
       structural information  based  on  molecular  fragmentation,
       often  including  molecular  weight data.  It is weak in the
       areas of aromatic substitution, isomer differentiation, ring
       junctions,  alcohol  identification,  and  functional  group
       classification.  Fortunately, the infrared  spectrometer  is
       strong  in  these areas.  The E.P.A. has recognized this and
       has developed Method  8410  for  the  GC/FT-IR  analysis  of
       semivolatile  organics  (1).   Combining  GC/MS and GC/FT-IR
       provides a higher confidence result than  either  one  alone
       (2).   Combining  the  two into one doubly hyphenated system
       with powerful automated computing  capability  provides  the
       analyst  with  an  efficient  system for analyzing hazardous
       waste.  An example of that is shown in this paper, utilizing
       a sample from an actual hazardous waste drum.


       HARDWARE

       The  gas  chromatograph,  an  HP 5890A, was set up using the
       common,  most  routinely  used  column   for   environmental
                                     1-84

-------
screening, a 25 meter 5% phenyl methyl silicone  (HP-5) under
standard  analytical  operating  parameters.     The   column
effluent  was  split  at  the end of the column  at a 10 to 1
ratio with the bulk of the flow going to the  HP 5965B  IRD
and  the  lesser amount to the HP 5970B MSD.  The details of
this parallel flow  configuration  are  described  elsewhere
(3).
SOFTWARE

One  of  the  characteristics  of  environmental  samples in
general and hazardous waste samples in  particular  is  that
they contain many components.  The sample used in this paper
contains many  dozens  of  compounds,  the  specific  number
analyzed   for   is  determined  only  by  the  instruments'
sensitivities and operational parameters.  Figure 1  is  the
Total  Ion  Chromatogram  (TIC)  from  the MSD and the Total
Response Chromatogram (TRC) from the IRD.  Easily well  over
100  components are detectable in each Chromatogram, but for
illustration, the integration threshold  was  set  to  allow
only  about 50 to be analyzed.  Those peaks are shown in the
integrated chromatograms of Figure 2.

In    order   to   aid   the   chemist   in   characterizing
multicomponent, complex chromatograms, the standard software
of  the HP 5965B includes the Macro Program 'Qualrpt'.  This
automated software routine performs a qualitative  analysis,
i.e.,    library    searchs    on    previously   integrated
chromatograms.  This Qualrpt macro works on  the  integrated
Total  Response  Chromatogram  (TRC) from the IRD and/or the
Total Ion Chromatogram (TIC)  from  the  MSD.   The  Qualrpt
output  for  individual  IRD  and  MSD  data  consists  of a
tabulation of the GC peaks found (times ,  areas,  etc.),  a
Chromatogram,  and  library  search  results.  When the data
from the IRD and  MSD  are  combined  in  Data  Editor,  the
Qualrpt  macro  produces  a tabulation of the peaks found, a
combined chroraatogram, and combined list of  library  search
results  for  each peak.  The combined search list is merged
by common CAS Registry numbers into three categories.  Class
1  contains  those  entries  which  are on both lists.  When
comparing IRD and MSD library searches, these  entries,  (or
more often, this entry) have a high probability of correctly
identifying unknowns.  Class 2 contains those entries  which
are  only  on one hit list because they exist in only one of
the two libraries  searched.   Class  3  consists  of  those
entries  which  are in both libraries but nonetheless showed
up in only one of the two hit lists.   Isomers,  because  of
their  nearly identical spectra often appear only on the MSD
hit list while homologous series because of  their  spectral
similarities  only  appear  on  the IRD hit list.   If the IR
spectrum of a  specific  unknown  compound  is  not  in  the
library,  typically the near misses are of the same chemical
                             1-85

-------
class.   This  is  a  very  powerful  feature  of   infrared
spectroscopy which is favorably exploited in the IRD Qualrpt
software.

Once the appropriate libraries are selected,  in  this  case
the  49000  entry  NIST/NBS  library of mass spectra and the
3000 entry EPA Vapor Phase Infrared library, Qualrpt can  be
initiated.   Several  examples  have  been selected from the
Qualrpt combined search results for discussion.
RESULTS

The  sample  was from a waste drum presumed to contain paint
and perhaps other hazardous  solvent  materials.   50  peaks
were  integrated  in the TRC and the TIC.  The chromatograms
are not totally identical.  There are some differences,  air
and  water peaks in the TRC, and a few more small components
in the TIC.  There were 41 common peaks found and 23 Class 1
hits.  Some peaks were found in the TIC only and some in the
TRC only.  Differing detector response factors  account  for
this  fact.   Carbon  dioxide  (from  air),  water (from the
sample and/or air), and the solvent methylene  chloride  are
early components found by the IRD.  In addition, acetone and
methanol were found and elute before the  solvent  raethylene
chloride.

A  typical  example  of  a Class 1 hit is shown in Figure 3.
The PBM mass spectrometry search and  the  IRD  search  both
indicate  that the compound is ethyl acetate.  Note that the
highest quality hits are in Class  1.   All  of  the  lesser
quality  IRD matches are acetates.  Infrared spectroscopy is
very    good    at    functional    group/compound     class
differentiation.   IR and MS confirm each other in this case
for a very high confidence result.

Figure 4 shows an example where the top quality PBM  hit  is
not  confirmed  by the IRD.  In this case the top IRD hit is
ethyl benzene and the top MSD hit is meta xylene.  As can be
seen  from the Class 3 listing, all three xylenes are in the
EPA IR library but their spectra are so different they don't
appear  on  the hit list while the spectrally similar ethyl,
propyl,  and  butyl  benzenes  do.   Clearly  the   infrared
information  confirms  that  ethyl  benzene  is  the correct
assignment.

Figure 5 is an example where the top IR hit is not confirmed
by  PBM.   Here  the top quality IR hit is butyl benzene and
the top MSD hit is propyl benzene.  The likely molecular ion
at m/z 120 is very important information pointing toward the
assignment of propyl benzene.  Note that all the  listed  IR
hits are alkyl benzenes.
                             11-86

-------
Since  the  NIST/NBS  MS  library is more than sixteen times
larger the than EPA IR library it is obvious  many  MS  hits
cannot be confirmed by IR.  Most of the time in this case of
Class 2 hits,  however,  the  chemical  class  and  isomeric
configuration  is confirmed.  Figure 6 is an example of this
situation where the spectra of the ethyl methyl benzenes are
not  in  the  EPA Library but the closest IR hits are mostly
all 1,4-disubstituted benzenes.   This  indicates  that  the
best qualitative assignment is l-ethyl-4-methyl benzene, not
the 1,2- or 1,3- isomers.

Table 1 is a condensation of  the  Qualrpt  combined  search
results.   It can be seen there are 56 total peaks reported,
41  common  ones,  and  23  Class  1  hits.   The   compound
assignments when no Class 1 hit was found is the most likely
based on further examination of the spectra.  In some  cases
the  molecular  ion  was  of  some  use.  Clearly use of the
largest MS and IR  libraries  available  would  improve  the
searching and is the subject of future work.
CONCLUSION

The combined technique of GC/FT-IR/MS using the HP 5890A, HP
5965B, and HP 5970B  with  the  Qualrpt  automated  spectral
output  and  combined library searching has been shown to be
very  useful  in  the  rapid  high  confidence   qualitative
analysis of hazardous waste components.
                             11-87

-------
REFERENCES

1. Method 8410, U.S.E.P.A., Rev. 1, 1990

2.    Gurka,   D.   F.   and  Pyle,  S.M.,  QUALITATIVE  AND
QUANTITATIVE     ANALYSIS      BY      CAPILLARY      COLUMN
CHROMATOGRAPHY/LIGHTPIPE   FOURIER  TRANSFORM  SPECTROMETRY,
Environ. Sci. Technol. 1988, 22, 963-967

3.  Leibrand, R.  J.  and Duncan, W.  P.,  INVESTIGATION  OF
THE  CHROMATOGRAPHIC  OPTIMIZATION  OF COMBINED GC/FT-IR/MS,
Int. Lab., 1989, 46-52
CONDITIONS

Gas  Chromatograph  Column:   25  m  x  0.32  mm id HP-5 (5%
phenylmethyl silicone), 0.52 micrometer  film  Carrier  Gas:
Helium  at  10 psi, 2.0 mL/min Oven: 40 C (1.0 min) to 240 C
at 4 C/min  Injection  Port:   250  C  Sample  Injection:  2
microliters split 10:1

IRD  Parameters  Light  Pipe:  250  C Transfer Lines:  260 C
Sweep Gas:  Nitrogen,  15  psi  inlet,  5  psi  outlet  Scan
Parameters:  8  cm-1  resolution,  2 co-adds, 3 scans/second
stored Detector: Wide band (550 to 4000 cm-1) MCT

MSD  Parameters  Mass  Range:   10  to  310   daltons   Scan
Parameters:  2 A/D samples, 1.4 scans/second stored Transfer
Line: 280 C
                             11-88

-------
CD
CO
                                                                                 TRC of DHTHiMRSTCIRZ.D
                                                                                 ukuwu
                                                                                 T:C of  DRTRiWHSTEMSZ.D
                                                                                                                       JLJL
JL
                                        Figure 1. Total Response Chromatogram (TRC) from IRD and Total Ion Chromatogram (TIC) from MSD of

                                                hazardous waste sample

-------
                                                                                     ป  TIC  of DF1TR; WRSTEMS2. D
                                                                                        TRC of DfiTfl:WRSTEIR2.D
CD
O
                                                                   ป
                                                                   n



                                                                   I
                                                                                              M     IB
                                                                                                                  20     52
                                                                                                                                      26      23
                                           Figure 2. Total Response Chromatogram (TRC) from IRD and Total Ion Chromatogram (TIC) from MSD of
                                                    hazardous waste sample showing integration of peaks used in this brief

-------
Peak 7 of 56.
(3 10:313 -
7.0E>5:
o
o 5.0L>5-
ซ 4.0E+5'
1 3.0ฃ>5-
1 2-0H ,-
\.BfH-5- <

(704:709 -
3 12-
i IB-
o 8-
8 B-
.0 4-
L.
ฐ 2~
K 0-1-+
4000

	 Both MS and IR found
...) flvg 3.571:4.088 ml n . from DfiTR : WBSTEMSE
43
ซ5
1 . .' / t2t 27t
, 1 / H2 \ ป' 175 187 H7 \
Hl^|li / \ / /// \
50 100 150 200 250
Mass/Charge
...) RSf 3.976:4.004 min. from DRTR: WRSTEIR2
- 5
at II m 1 1 "~
en 1 — H •
A A JIL. .
3500 3000 2500 2000 1500 1000
Trequency (cm-1)
COMPARISON OF RESULTS FROM
PBM Search of Library file: DATA:NBS43K.L
AVQ 3.971:4.008 nin. fron DATA:UASTEnSZ.O
AND
IR Search of Library file: DATA:EPA_REVA.L
ASP 3.376:4.004 nin. fron DATA:UASTEIR2.D
Class 1 (on both lists)

.D
. D

PBM IR
CK Nunber Dual Oual MUt Fornula Nane
1. OO014I-78-6
64 379 88 C4H802 Acetic acid, ethyl eater
Class 2 (in only one library)

PBM IR
Cfl^ Nunber Oual Oual NUt Fornula Nane
Z. OOOS95-46-0 25 	 I3Z CSH804 Propanedloic acid, dimthyl-
3. OO0078-98-8 12 — 72 C3H402 Propanal . 2-oxo-
4. 002203-36-3 — 9EO ISO C6HI1C10Z ACETIC ACID. 3- CHLOROBUT YL ESTER
5. 006363-44-6 — 953 188 C9HI604 1 ,5-PENTANEDIOL . DIACETATE
Class 3 (in both libraries, but on only one list)
PBM IR
CAS Nunber Oual Oual MUt Fornula Nane
6. 000628-63-7 — 952 130 C7HI402 ACETIC ACID. PENTYL ESTER
7. O0443S-53-4 — 951 146 C7HI403 ACETIC ACID. 3-METHOXYBUTYL
ESTE
Figure 3.  Qualrpt output of peak 7, Class 1 result indicating ethyl acetate
                                   11-91

-------
 Peak Z3 of S6.
                       —  Both HS and IR found
(BG4:874 - ...) Rvg 10.821:10.945 min. from DflTfi: WRSTEMS2 . D

a
0
C
n
T3
c
3
jo
a:



c
g
cr
E
"
a
o
c
a
JO
t.
o
cc

3.
2.
2.

1 .
1.
5.
0.


0E+6:
5E+6:
0E+6-

5E*S-
0E+S-
91




SI
V \. K
\ ,NJ J .1 .




IBC
it? IBI 214 nt 245 as ai
J / Vs ',"/ / / / / \

50 100 150 800 250 300
Mass/Charge
1925:1938 - ... RSP 10.868:10.942 m!n. from DflTfl: WHSTEIR2 . D











B.0:


B.0:

4.0-

2.0:



01
2

CO 1
01 f

01 H "" - * m /
raoi m u JL a ru g I
^jJ^L^J L
4000 3500 3000 2500 2000 1500 1000
Frequency (cm— 1)
                          COnPARlSON OF RESULTS FROM
                  PBH Search of Library file: DflTfl:NBS49K.L
                  Avg 10.821:10.945 ซln. fron DrtTfl:UftSTEMSZ.D
                                     AND
                  IR Search of Library file: DflTfl:EPa_REVfl.L
                  RSP la.BEB: 18.942 din. froH DflTfl:UftSTEIR2.D
                           Class  I   
-------
  Peak  34 of 56.
                        -—  Both US and  IR  found
(


C
3
_Q
cc



1IGB:1176 - ... flvg 14.579:14.677 m 1 n . from DRTR: WRSTEMS2 . D
1.0E+6-
8.0E+5-

6.0E>5-

4.0E+5-

2.0E+5-







cs
39 \ 7ซ
(5 / 1 i
' i. li. iJ jl . i ,
91




121
/
ISI 22C 27<
IซI / (77 212 \ /
. .1 . . ' / / / \ /
50 100 150 203 250
Mass/Charge
(2586:2600 - ... RSP 14.600:14.679 mm. from DRTR: WRSTE IR2 . D
„
oz
E

ffi
O
C
a
c
O
-O
a:

3. 0-
2.5-

2.0-
1 .5"

1 .0-
0.5-
0.0-i
4
$

Jli
s ™
n 11 <"
i IR „, 2

/I "ft" - M
•^~" — -^ \ 	 A J VA. _ f^. ^ J Ly
4000 3500 3000 2500 2000 1500 1000
Frequency (cm— 1)
                          COMPARISON OF RESULTS FROM
                  PBM  Search of Library file: DATA:NBS49K.L
                  Avg  14.578:14.677 din. fron DATA:UASTEMSZ.D
                                     AND
                  IR Search of Library file: DATA:EPA_REVA.L
                  ASP  14.600:14.679 nin. fron OATA:UASTEIRZ.O
                          Class
                                   (on both lists)
     CAS Nunber
                 PBM
                 Oual
                IR
               Oual
                            MUt  Formula
                                                Nane
  I.  800103-65-1    80   941   IZ8  C9H1Z
                                              Benzene, propyl-
        CF)S Nunber
                    PBtl
                    Qual
                   Class Z   (in  only one  library)

                   IR
                  Oual  nut  Formula        Nane
     Z.
     3.
     4.
     5.
     6.
     7.
     8.
8B4I5Z-89-4   64   	  150  C9H14NZ       1,Z-Ethanedianine. N-(phenylneth
084545-85-I   43   —  I9Z  C7H7C1ZP      Phosphonous dichloride. (phenyln
017G34-51-4   33   	  1Z0  C9H1Z         I,3.5-Cycloheptatriene. 7-ethyl-
000620-05-3   32   	  218  C7H7I         Benzene, (iodonethyl)-
084464-74-8   23   —  184  C8H803S       I.Z-Ethanediol. phenyl-. cyclic
00O148-Z8-3   1O   ---  240  C16HZ0NZ      1.Z-Ethanedianine. N.N'-bis(phen
000588-67-0    8   ---  164  C1IH160       Benzene, (butoxynethyl)-
                Class 3  (in both libraries, but on only one list)
                    PBM   IR
        CAS  Nunber  Oual  Oual  MUt  Formula
                                                   Nane
     9.  000104-63-Z   64   -—  151  C9HI3NO      Ethanol.  Z-[ (phenylnethyl >aninol
    10.  008IZZ-78-1   37   —  IZ0  C8H80        Benzeneacetaldehyde
    II.  000184-51-8  —-   953  134  C10HI4       BENZENE.  BUTYL
    12.  800538-68-1  —   949  148  CI1H16       BENZENE.  PENTYL
    13.  881877-16-3  —-   S4Z  I6Z  CIZH18       HEXANE.  I-PHENYL
    14.  8BZ189-60-B  —   933  190  C14HZZ       OCTANE.  1-PHENYL
Figure 5. Qualrpt output of peak 34, Class 1 result indicating propyl
           benzene
                                         1-93

-------
     Peak 36 of 56.
                           	   Both MS and IR found
<

a
u
n
T.
c
D
I


1205:1208 - ... Rvg 1 5 . 05 8 : 1 5 . BSE mm. from DRTfi: WflST EH92 . D
a.5E>6-
2.0E + 6:

l.SEi-G-

1.0E + 6:
5.0E+S-






ป C 77 SI
1 T t , \, i ' ,
15




(II
us n?
i .. 7' T \/
B.BE-t-B"1-1 — ' — ' — • — i — • — ' — ' — • — i — • — • — • — • — i •"•" 	
50 100 150 200
Mass/Charge
(2666:2674 - ... RSP 15.051:15.096 mm. from DflTR: WF1STEIR3 . D
Zi
rr
E
—

U
C
n
X!
t
O
(A
jD
rr

5.0:

4.0:

3.0:

2.0;

1 .0-

I
mซ
ป F -
Is r -


I 1 | -v 0- ft
1 1 ^ ^ II <** nu M
I <* S 1 S is /
/I 2" K- — \\
J v^ y..^--' v-^ — vx; s^v
4000 3500 3000 3500 2000 1500 1000
Frequency ( cm- 1 )
                              COMPARISON OF RESULTS FROM
                      PBn Search of  Library file: DflTfl:NBS43K.L
                      Avg IS.eSB: 15.836 nin. fron DflTfl:UflSTErtSZ.0
                                         AND
                      IR Search of Library file: WTA:EPn_REVA.L
                      ASP IS.eSI:IS.83E nin. fron DflTfl:UftSTEIRZ.0
                               Class  1   (on both lists)
                     NO COMMON COMPOUNDS  FOUND  IN SEPARATE REPORTS ••••
                                                    1 ibrory)
                          pen   IR
             CAS Nunber  Dual  Oual   HUI   Fornula
                                                        None
          z. eeeies-67-8
          3. 0ซ3l4!-ซZ-4
          4. 895814-85-7
          S. 061142-17-4
          6. BZ9634-3S-6
          7. eซ7Zt4-BI-1
          8. (M1B38-B3-9
          9. eei484-se-e
         la. 0MI3Z-77-B
9,
83
S6
53
58
43
37
zs
t;
                               937
                        ,3(i
                        ,7<
                        Z67
                        1S,
                        ,zซ
                        27/i
                        I4H
 C9H1Z        Benzene, l-ethyl-4-nethyl-
 C9HIZ        Benzene. I,3.S-trinethyl-
 C9H12        1.3-Cyclopentadtene.  S-(l-nethyl
 CISHI6       Benzene, I.1'--
 C9HIZ        2.3-Heptadien-S-yne.  2.4-dinethy
 CUHItBrO    Ethanone. 2-brono-1 ,2-diphenyl-
 CIIH16       BENZENE. 1 .Z-DItlETHYL-4-ISOPROPY
                     Class 3  (in both libraries, but on only one list)
             PBH    IR
CAS Number   Oual   Oual   HUI
                                         Fornula
         n. ceaeze-u-4
         12. eeeen-14-3
         13. eซeSZB-73-B
         is. mee95-63-6
         16. aeeese-es-z
         17. eooies-es-s
         IB. eeee93-87-E
         13. WI632-I6-2
         ze.
              91
              91
             47
             2S
                  950
                  949
                  935
                  934
          ,7ป
          ,;ซ
          ,7^
          ,70
          ,;ป
          |Zซl
          134
          13ซ
          ,,/
          I4B
C9HIZ        Benz  ne.  1-ethyl-3-nethyl-
C9HI2        Benz  ne.  l-eปhyl-2-nethyl-
C9HIZ        Benz  ne.  I.2,3-trinethyl-
C9HI2        Benz  ne.  (l-nethylethyl)-
C9HI2        Benz  ne.  1.Z.4-trinethyl-
C8H80        Ethanone, 1-phenyl-
CiaHU       BENZENE.  P-OIETHYL
CI8HI4       BENZENE.  l-ISOPROPYL-4-METHYL
C8HI6        1-HEXENE. Z-ETHYL
CIIHI6       TOLUENE.  P-TERT-BUTYL
Figure 6.  Qualrpt output of peak 36, Class 2 and 3 results indicating
            l-ethyl-4-methyI benzene
                                           1-94

-------
Peak
#
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
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
. 53
54
55
56
Retention
Time, TIC
_
2.364
2.964
3.033
3.150
3.690
3.983
4.271
4.799
5.174
-
5.368
5.604
5.665
5.943
6.517
-
7.299
7.583
7.663
-
8.264
10.913
11.362
11.879
12.184
-
12.495
12.996
-
13.423
14.053
14.360
14.624
14.999
15.075
15.313
15.514
15.689
15.899
16.339
17.105
17.449
17.882
18.090
18.332
18.602
18.710
18.865
18.992
19.772
19.838
20.112
21.231
24.526
27.531
PBM
Quality
_
-
72
-
-
59
64
49
45
47
-
64
47
95
91
95
-
53
91
83
-
64
64
91
43
91
-
59
74
-
87
90
90
80
91
91
91
72
70
45
91
96
80
59
86
81
91
38
90
94
93
94
83
94
43
91
Retention
Time, TRC
2.293
1372
2.971
3.038
3.157
3.693
3.989
4.278
4.801
5.179
5.229
5.372
5.608
5.670
5.948
6.523
6.828
7.303
7.588
7.667
7.887
8.268
10.919
11.369
11.887
12.192
12.378
12.502
13.004
13.130
13.431
14.061
14.368
14.633
15.006
15.081
15.319
15.521
15.701
15.905
16.350
17.113
17.456
-
18.097
18.339
18.606
-
-
-
19.790
-
20.119
21.239
-
27.540
IRQ Class
Quality 1 Hit
	 _
843
978 +
940
915
921 +
979 t
984 +
930
965
923
958
917
939
984 +
971
938
920
981 +
977
906
922
966 +
953 +
952
976 +
948
972 +
969 +
945
967 +
970 +
951
941 +
972 +
950
901 +
950
926 +
950
967 +
982
949 +
-
951
947 +
939 +
-
-
-
920
-
929 +
983
-
989 +
Identification,
Most likely if not Class 1
Carbon dioxide (air)
Water
2-Propanol
Acetone
Methylene chloride
2-Butanone
Acetic acid, ethyl ester
1-Propanol, 2-methyl
2-lsopropoxy ethanol
Branched paraffin
2,3-Dimethyl pentane
Branched paraffin
Olefin or cycloparaffin
1,2-Dimethyl cyclopentane
Heptane
Methyl cyclohexane
Ethyl cyclopentane
1-Methoxy butane
Toluene
iso Butyl acetate
2-Methyl Heptane
2,2-Diethoxy propane
Ethyl Benzene
Meta xylene
alcohol
Ortho xylene
1,4-Dimethyl cyclohexane
2-Butoxy ethanol
Isobutyl butyrate
1,2-Dimethyl cyclohexane
Cumene
Propyl cyclohexane
3-Methyl nonane
Propyl benzene
1-Ethyl-3-methyl benzene
1-Ethyl-4-methyl benzene
1,2,4-Trimethyl benzene
4-Methyl nonane
1-Ethyl-2-methyl benzene
3-Methyl nonane
1,2,4-Trimethyl benzene
Decane
1,2,3-Trimethyl benzene
1-Ethenyl-2-methyl benzene
4-Methyl decane
Isobutyl cyclohexane
1,3-Diethyl benzene
Diethyl benzene
U,8-P-Menthatriene
1-Ethyl-2,4-dimethylbenzene
2-Ethyl-1,4-dimethylbenzene
1 -Methyl-3-isopropylbenzene
1,2,4,5-tetramethylbenzene
Undecane
1,3-Dioxolane-2-methanol
Phthalic anhydride
Table 1. Summary of compounds found in hazardous waste sample
                                11-95

-------
             Environmental Applications of Multispectral Analysis
                                      by
                                John M. McGuire

                       Environmental Research Laboratory
                     U.S. Environmental Protection Agency


     Beginning in the early '70s, extensive application of gas

chromatography/mass spectrometry (GC/MS) for identification of organics in

water led to its now being the accepted method for positive identification of

target analytes.  GC/MS with automated spectra matching against a reference

collection of mass spectra is known to be excellent for specific

substantiation of target compounds, but its current success rate for tentative

identification of unknowns is poor.  In particular, it fails to detect and/or

identify compounds whose mass spectra are not in the spectral libraries.



     In order to improve the identifications of non—target compounds,  we have

applied other existing techniques to environmental sample extracts.  This

Multispectral Analysis approach uses high resolution mass spectrometry (HRMS)

to determine elemental compositions of ions, Fourier transform infrared (FTIR)

spectroscopy to recognize sub-molecular structures, and chemical ionization

(Cl) mass spectrometry to establish molecular weights of the unknowns.  The

spectral information is then melded together to postulate the structures of

the unknown compounds.  Results on application of this technique to

unidentified compounds in environmental samples have been excellent.  Upon re-

examination of samples from a survey conducted by the EPA Office of Water, we

identified two series of aldehydes as well as a variety of organo—phosphates

whose spectra were not included in the reference collection of mass spectra.

In the course of the work, the approach was also applied to correct a

misidentification made by routine spectra matching.
                                  1-96

-------
56
SAMPLE PREPARATION USING SUPERCRITICAL FLUID EXTRACTION METHODOLOGY
         Werner F. Beckert,  U.S.  Environmental Protection Agency,  EMSL-LV,  Las
         Vegas, Nevada 89109, and Viorica Lopez-Avila,  Mid-Pacific Environmental
         Laboratory,  Mountain View,  California 94043.

         ABSTRACT

         Although extraction of  analytical  samples with supercritical  fluids
         (SFs)  has  received  much attention  during the  last 10  or 20  years,
         applications of supercritical fluid extraction  (SFE)  techniques  to  the
         extraction   of  compounds  regulated   by  the  Environmental  Protection
         Agency  (EPA)  from  matrices  of  concern  to  the EPA  have been  rather
         limited. In late  1988, we started a project  to develop SFE methods  for
         samples of  interest to the EPA.   Based on our  results, which are  summa-
         rized  in this paper, we  developed a draft  protocol  for  SFE of  environ-
         mental samples  that has undergone  a limited multi-laboratory  evalua-
         tion.    Furthermore, an  EPA work group  for  SFE development has  been
         formed,  with participants from EPA,  other Government agencies,  industry
         (especially the SFE  equipment manufacturers)  and academia,  and  evalua-
         tion of commercially available  SFE  instrumentation  is continuing.   The
         results to  date demonstrate that SFE is a viable alternative to conven-
         tional  methods  for  the extraction  of organic  pollutants  from  solid
         samples.  However,  our results  also demonstrate that  the many  factors
         affecting SFE efficiency make it  difficult to optimize  the method,  and
         that more developmental work has to be done before SFE becomes  an easy-
         to-use,  off-the-shelf method.

         INTRODUCTION

         The Environmental  Protection Agency  (EPA) is   interested  in  new  and
         improved analytical  methods  which are faster, better and cheaper  than
         present methods, and which, at the same time,  are safe and environment-
         friendly (by minimizing  the  generation of waste).   Such methods,  when
         not developed  specifically for  the analysis  of  environmental  samples,
         must  be  adapted to EPA needs  with  respect  to matrices, analytes  or
         analyte groups, sample  sizes,  data  quality  objectives  (precision  and
         accuracy requirements),  etc.  The methods  should be generic, as  far as
         analytes and matrices  are  concerned,  and  they should not be restricted
         to any particular brand of instrumentation or  equipment.
         NOTICE:    Although   the  research  described  in  this  paper  has  been
         supported by the U.S. Environmental  Protection  Agency,  it  has not been
         subjected to Agency  review and therefore does  not  necessarily reflect
         the  views  of   the   Agency,   and  no official  endorsement  should  be
         inferred.   Mention  of   trade  names  or  commercial  products  does  not
         constitute endorsement or recommendation for use.
                                          11-97

-------
Sample  extraction techniques  should,  to the  extent  possible,  yield
quantitative  recoveries  of the  target  analytes from  the  matrices,  be
selective so that extraction of interferants is minimized, not generate
large  volumes of  waste solvents,  require  little  sample  and  extract
handling to minimize analyte  losses  and contamination, and be fast and
inexpensive.  The two  methods that are at present  included in the SW-
846 methods  manual1,  Soxhlet  extraction  (Method 3540)  and sonication
extraction  (Method  3550),  only  partially   fulfill   these  extraction
goals.  A  third  extraction method, Soxtec extraction (which is basic-
ally  a modification of  Soxhlet  extraction), comes somewhat  closer to
reaching these goals.  These three methods have recently been evaluated
for their relative merits2.

For a number of  years now,  supercritical  fluid extraction  (SFE) has
been publicized as a new and  promising technique for the extraction of
organic compounds  from  solid  matrices.   Some of the claimed advantages
of  SFE  over  conventional  extraction methods include  much  shorter
extraction  times  and close to quantitative  recoveries.   No toxic and
expensive solvents are  required  which results in reduced materials and
waste disposal costs, and in reduced  environmental pollution.  No sol-
vent  removal  is  required,   and no glassware  cleaning.   SFE conditions
can be optimized  by  varying pressure  and   temperature  and  by  using
modified supercritical  fluids (SFs), and extractions  can be performed
at relatively low temperatures,  if  desired.   Overall,  the  use of SFE
techniques in place of conventional methods could result in substantial
cost and labor savings.

In principle,  SFE is  similar to other solvent extraction techniques,
except  that  the  solvent is in its supercritical (SC)  state.   SFs have
some unique  properties that put  them  between liquid  and gases.  Their
viscosities  are  much  lower  than those  of  liquids and  their surface
tension  is zero,  that  means,  they  can  penetrate  into the  pores  of
solids  much more easily  than liquids.   Their densities  are  close to
those  of  liquids which  means their capacities  for carrying dissolved
materials are similar to those of liquids.

The most commonly used SF is  C02; others that  are  being used,  or have
been  investigated,  include nitrous oxide, sulfur  hexafluoride, Freon-
13, ammonia,  xenon  and several  hydrocarbons.   SF C02 is  so popular
because  of its  low  critical  temperature (31.3ฐC)  and  pressure   (72.9
atm),   and  because it  is  non-toxic,  non-flammable,   relatively  non-
reactive and  inexpensive,  and its use does  not  result in a waste dis-
posal  problem.  It is  a rather non-polar solvent,  similar to hexane or
benzene, but  both solvent  strength and selectivity can be improved by
the addition  of  small  amounts of modifiers,  such as acetone, methanol,
or toluene.

Application of SFE to  the  extraction of compounds  regulated by the EPA
from  solid  analytical  samples  has  been  limited.     Brady   et  al.3
extracted PCBs, DDT and  toxaphene from  spiked soil  samples with SC C02.
Schantz and  Chesler4  extracted urban  particulate matter and sediments
                                  11-98

-------
with  SC  C02  and   found  that  recoveries  of  PCBs  and  PAHs  were
approximately  equivalent  to  those  obtained  via Soxhlet  extraction.
Smith and coworkers5'6 used SC CO, and  SC  isobutane  to extract various
condensed aromatic and heterocyclic  compounds  from urban dust and from
XAD-2  Spherocarb.    Hawthorne et  al.7~10  used SC  CO  and  SC nitrous
oxide  to  extract PAHs from samples of  urban  dust,  fly  ash  and river
sediment.   Other authors  reported  extraction of  triazine herbicides
with SC  C02  from spiked  sediment  samples11,  PCDDs and  PCDFs from fly
ash samples with SC nitrous oxide12, and PCDDs from sediment samples13.
Additional  applications  of  SFE  techniques   to  environmental  sample
extractions  have  been  reported  at  recent   scientific meetings  and
symposia14"17.

SUMMARY OF EXPERIMENTAL RESULTS

We started  with our  experimental  studies  in  late 1988  with a Suprex
Model SE-50  extraction system using either a  single-extraction vessel
arrangement, or  a two-  or four-vessel  arrangement  where  two or four
extractions were performed  simultaneously.  In the multi-vessel experi-
ments we were  mainly interested in establishing the equivalency of the
results obtained from the parallel extractions.   All experimental work
was  performed at  the Mid-Pacific Environmental  Laboratory  (formally
Acurex  Corp.).   The   bulk  of  the results  is summarized  below (for
details see ref. 18):

Single-Vessel  Extractions

0   Seventeen  organochlorine  pesticides (OCPs) were  spiked on sand at
    500 and  2,500  ppb and  extracted with  SC C02  for  30 min at 150 atm
    and 50ฐC.  The mean recoveries were  almost quantitative for most of
    the compounds.   A combination  of static and dynamic  extraction, as
    well as variations of P and T, gave  similar results.

0   Forty-one  OCPs  were spiked  on  sand  and extracted with  SC C02
    modified  with 10% methanol.   The recoveries from  the  triplicate
    samples were 79%  or higher for 38  of the 41 compounds.

0   OCPs were  spiked at  two levels on soil containing 10% moisture and
    extracted  with SC CO,  using a  combination of static  and dynamic
    steps at various p and  T  settings.  The mean recoveries were 80 to
    90%  (but  only  about  50% for endrin aldehyde).  The  moisture (which
    can be regarded  as a  modifier) obviously did not drastically change
    the extraction efficiencies under  these conditions.

0   Aroclors  1232 and  1260  were  spiked at 5000 ppb on Florisil and
    extracted  for 40 min  with SC C02 at  various conditions  for p and T.
    The recoveries were quantitative at  relatively low temperatures and
    high pressures.
                                  11-99

-------
0   Fourteen phenols were spiked on sand at 3.6 to 18 ppm and extracted
    with  SC  C02.   The  recoveries  ranged  from  53 to 129%,  except for
    2,4-dinitrophenol with only 27% recovery.

0   Twenty-five organophosphorus pesticides  (OPPs)  were spiked on sand
    at 2.5 ug/g  and  extracted with SC  CO  or  SC  C02 modified with 10%
    methanol.  The recoveries  were significantly  higher when methanol-
    modified  CO   was used  (20 recoveries  > 80%,  compared  to  only  8
    recoveries >  80%  for   SC CO   alone);  however,   in  both  cases,
    several OPPs were not recovered at all.

0   Sixteen  polynuclear aromatic  hydrocarbons (PAHs)  were  spiked  on
    coal, coal  fly ash, sand  and urban dust.    Mean  recoveries after
    extraction with SC CO  at  150 atm/50ฐC/60 min were almost quantita-
    tive for samples of  the  coal and  coal  fly  ash but only 57% for the
    urban dust samples.   Mean recoveries  from  the  spiked  sand samples
    under two sets of conditions improved with added modifier (200 yL
    acetone, added to  the sample)  from 74 to 81% in one case and from
    58 to 89% in the other case.

0   Soil  samples (SRS  103-100,  Fisher  Scientific), certified  for  13
    PAHs, dibenzofuran  and   pentachlorophenol,  were extracted  with  SC
    C02 at 300 atm/70ฐC/60  min.   Ten percent  water was added to each
    sample prior  to  extraction.   All recoveries  were >60%,  except for
    benzo(b and k)fluoranthene (53%) and benzo(a)pyrene (32%).

0   Sand was spiked with 43 neutral/acidic compounds and extracted with
    SC C02,  with and without  modifier (200 pL acetone) added  to the
    sample.   Some 20  recoveries  were  lower when the modifier was used,
    and only 14 recoveries increased.

Two-Vessel Extractions

0   Sand was spiked with 36  nitroaromatic  compounds and extracted with
    SC C02  under two sets of experimental conditions.  The agreement
    between   the  duplicate   extractions  performed  in  parallel  was
    excellent (generally within 10%).  The more volatile nitroaromatics
    gave good recoveries at  lower (200 atm) but not at higher (300 atm)
    pressures.

0   19 haloethers spiked on  sand  and extracted with SC C02  gave mean
    recoveries of 73  to 99%  for all  but two compounds.  The agreement
    between the duplicate extractions performed in  parallel was within
    15% for most  of the compounds.

Four-Vessel Extractions

0   Sand was spiked with 19  haloethers and extracted with SC C02 at 250
    atm/60ฐC/60 min.  Of the  19 compounds,  15  were  recovered at >75%,
    and  the  other four  all  at above 45%.   The  agreement  between the
                                  1-100

-------
    parallel extractions, expressed as % RSD, ranged from 1.2 to 35.6%,
    with 12 values being below 10%.

0   Of  42  OCPs,  spiked on  sand and  extracted  with  SC C02,  35  gave
    recoveries  >50%  whereas  two  (chlorobenzilate  and  endosulfan
    sulfate) were  not  recovered at all.   Twenty-six of  the RSDs  were
    <10%,  and  the  others  were  between  10  and  23%,  except   for
    l,2-dibromo-3-chloropropane (30.8%).

The  only  limitations  we  have  experienced  with  the  four-vessel
arrangement was in the duration of the extraction.   When working  with
2-mL extraction vessels and using 50-ym restrictors, the 250-mL syringe
pump allows a maximum extraction time of 60 min.

DISCUSSION

As can  be seen  from  the above summary of our experimental results, the
extraction efficiencies we achieved with  a variety of  samples are in
general good  to reasonable, especially since  in most  cases  we had not
tried  to  optimize our  extraction  conditions.   Some  of  our recoveries
were much  lower than  those reported by others.   However,  one has to
realize that  at least  in some of  the  cases reported in the literature
the  extraction conditions  had  been  optimized  in  a   trial-and-error
approach.   In  addition,  such experiments  were  often  conducted  with
homemade equipment,  focused mostly on PAHs, and used small sample sizes
(as small as  a  few milligrams).   In order to develop a SFE method that
can successfully be  applied to samples of  interest to the EPA, we have
to  use commercially available equipment,   consider  a wide  variety of
sample  matrices and groups  of pollutants, and use  sample sizes large
enough  (1  to  10  g,  preferably  at   least 5  g)  for  the  inevitable
inhomogeneities of most real environmental  and hazardous waste  samples.

There  is  a   lack  of  standard  reference   materials  that  include  the
matrices and  pollutants of environmental  concern.   The materials  that
are available are either  spiked matrices  (soils,  etc.), or  they are
certified for only a very  limited number of compounds, e.g., PAHs.  It
is  therefore  difficult,   even  currently  impossible,   to  determine
absolute  extraction efficiencies  for  most analytes  because  in  most
cases  removal  of  a spike  from a  sample  matrix  is much  easier  than
removal of  "incorporated"  or  "native" compounds.   This, however,  is a
problem that hampers the evaluation of all  extraction methods,  not just
SFE,  and  one  is  usually  confined  to  comparing  relative  extraction
efficiencies.

Temperature and pressure changes affect the density and viscosity of a
SF  and  therefore  its  solubilizing ability.   However,  it  is little
understood what happens on  the  surfaces of the  solid matrices during
the extraction  process, and what  the  desorption,  solvation and trans-
port mechanisms are, and little  is known about how to optimize p and T
for specific  matrices  and  analyte groups.  Just  raising the  pressure
does not seem to be  the answer.
                                  11-101

-------
An understanding of  the  desorption  and transport mechanisms of solutes
under SC  conditions  would provide  clues  whether the use  of  static or
dynamic extraction conditions,  or  a  combination of the  two,  would be
advantageous; whether rapid pressure fluctuations would improve extrac-
tion rates  and,  maybe,  extraction  efficiencies  and selectivities; and
whether application  of  ultrasound  could  enhance extraction efficiency
and rate, as has been suggested6'19.

CURRENT ACTIVITIES

Summarized  below are the SFE  activities in which EPA is at present
involved.
EPA SFE Methods Development Group

In January 1990,  an SFE  Methods  Development Group  was formed.   The
overall goal  of  this group is to assist  and  advise in the development
of SFE  to make it  a viable,  attractive  and  affordable alternative to
conventional extraction methods.  This effort includes

0   development  of  a general,  standardized  SFE  method,   or  set  of
    methods, for a variety of analytes and matrices,

0   generating performance data for  the method(s) through  intra- and
    interlaboratory evaluation studies, and

0   improving and, to the extent practical, standardizing hardware.

The SFE  Methods  Development  Group  members come  from  EPA (OSWER, ORD,
and Regional laboratory personnel), instrument manufacturers, academia,
and other  interested  contractor laboratories.    Semi-annual meetings
provide a  forum  for candid discussions of results and problems, of new
approaches and of specific applications.

Protocol Development and Evaluation

Based on our results we developed a draft protocol in  the SW-846 format
"Extraction  Procedure Using  Supercritical  Fluids."   Our goal  was to
write a  generic  protocol  that  is  applicable to  as many different SFE
systems as  possible.   It is written  for  solid  matrices like soils and
sediments;  the  target  analytes  include  organochlorine  pesticides,
polychlorinated biphenyls,  polynuclear aromatic hydrocarbons, phenols,
phthalate  esters,  and  organophosphorus  pesticides.    The  protocol
addresses  interferences,  apparatus  and  materials, sample  preparation
(including  extraction),  and  quality  assurance.   An  updated protocol
version  was  recently evaluated by   10  laboratories  for   its  feasi-
bility21.   The analytical  data generated by the different laboratories
varied substantially, however, it was  confirmed that the protocol  could
be followed without  problems  by all operators involved, independent of
                                  1-102

-------
the particular SFE  system used,  and this has been  the  main purpose of
the evaluation study.

Instrument Evaluation

Most of  the  commercially  available SFE systems have  been evaluated at
the Mid-Pacific  Environmental Laboratory under  contract to  the  EPA.
The  generous  help  from   the  instrument  manufacturing  companies  is
acknowledged and greatly appreciated.  The main goal was to assure that
our protocol was  compatible  with  all instruments.   In general,  all
instruments performed adequately, although, as can be expected, none of
the instruments was without some problems, and feedback was provided to
the manufacturers as to problem areas and perceived weaknesses of their
instruments.   However, it must  be understood  that  evaluation of an
instrument does not constitute endorsement by the EPA.

Extraction and Optimization Studies

ฅork is  continuing  in  EPA laboratories and laboratories under contract
to the EPA on  different matrices (e.g.,  fly ash,  bottom ash, clay-type
soil,  soil high  in organics,  river and marine  sediments,  etc.)  and on
method optimization.   Currently,  a method for the extraction of  oil/
grease and  total petroleum hydrocarbons  from soil  is being developed.
The  current  EPA  methods  specify extraction  with  Freon  ,  however,
Bicking  et  al.21  and  others have shown  that  these  materials  can be
extracted with C02  under SC  conditions.   Our own  results confirmed
this,  and a  draft protocol  for  the  determination of  oil/grease  and
total petroleum hydrocarbons has been prepared.  Another method of cur-
rent  concern is  the  extraction of phenoxyacid  herbicides  and other
acidic compounds for soil.  Miller et all22 have shown that acidic com-
pounds can be derivatized by adding trimethyl phenyl ammonium hydroxide
in  methanol to  the material  in  the  extraction  vessel,  followed by
static and then  dynamic SFE.  Finally,  we are looking at the effect of
ultrasound application during SFE which, as discussed earlier, seems to
increase extraction rate and possibly extraction efficiency.

CONCLUSION

Supercritical   fluid  extraction  is  an  attractive  method  for  the
extraction of organic  contaminants  from matrices of concern to the EPA.
The most-used  extraction  medium, carbon dioxide,  is non-toxic and non-
polluting,   and  creates   no   waste-disposal  problems.     Potential
advantages of  the method  include  reduced material  and manpower needs,
speed, high  efficiencies,  selectivity  (in combination with modifiers),
and high versatility,  especially  in combination with  advanced  analy-
tical  techniques.    However,  more  developmental work  has  to be  done
before SFE becomes an  easy-to-use,  off-the-shelf method.
                                  11-103

-------
References

1.  Test Methods  for Evaluating Solid  Waste (1986), 3rd  Ed.,  SW-846,
    U.S. Environmental Protection Agency, Washington, D.C.

2.  V.  Lopez-Avila,  J. Milanes,  N.  Dodhiwala,  J.  Benedicto and  W.F.
    Beckert.    "Evaluation  of  Sample  Preparation  Methods  for  Solid
    Matrices."     Proc.   Seventh  Annual  Waste  Testing  and  Quality
    Assurance Symposium, Washington,  D.C., July 8-12, 1991.

3.  B.O. Brady, C.P.C. Kao, K.M. Dooley,  F.C.  Knopf and R.P. Gambrell.
    "Supercritical  Fluid  Extraction  of  Toxic  Organics   from  Soils."
    Ind. Eng. Chem. Res. 26: 261-268 (1987).

4.  M.M.  Schantz  and  S.N.  Chesler.    "Supercritical Fluid  Extraction
    Procedure  for  the Removal  of Trace  Organic  Species  from  Solid
    Samples."  J.  Chrom.  363: 397-401 (1986).

5.  B.W.  Wright and R.D.  Smith.  "Supercritical Fluid Extraction  of
    Particulate and Adsorbent Materials:   Part  II."  EPA Report 600/4-
    87/040 (1987).

6.  B.W.  Wright,  J.L.  Fulton,  A.   J.  Kopriva  and   R.D.   Smith.
    "Analytical Supercritical Fluid  Extraction  Methodologies."    In:
    Supercritical Fluid Extraction and  Chromatography:   Techniques and
    Applications.    B.A.  Charpentier  and  M.R.  Sevenants,  Eds.    ACS
    Symposium Series 366:   44-62 (1988).

7.  S.B. Hawthorne  and D.J.  Miller.    "Directly  Coupled  Supercritical
    Fluid  Extraction—Gas  Chromatographic  Analysis   of  Polycyclic
    Aromatic  Hydrocarbons   and  Polychlorinated  Biphenyls   from
    Environmental Solids."  J. Chrom.  403: 63-76 (1987).

8.  S.B. Hawthorne, M.S. Krieger and  D.J.  Miller.   "Analysis of Flavor
    and  Fragrance   Compounds  Using  Supercritical  Fluid  Extraction
    Coupled with Gas Chromatography."  Anal. Chem. 60:  472-477 (1988).

9.  S.B.  Hawthorne  and D.J.  Miller.    "Extraction and  Recovery  of
    Organic  Pollutants from  Environmental  Solids  and  Tenax-GC  Using
    Supercritical C02."  J.  Chrom.  Sci.  24: 258-264 (1986).

10. S.B.  Hawthorne  and D.J.  Miller.    "Extraction and  Recovery  of
    Polycyclic  Aromatic Hydrocarbons  from Environmental  Solids  Using
    Supercritical Fluid."  Anal.  Chem. 59: 1705-1708 (1987).

11. V.  Janda,  G.  Steenbeke   and  P.   Sandra.    "Supercritical  Fluid
    Extraction of s-Triazine Herbicides from Sediment."   J. Chrom.  479:
    200-205 (1989).

12. N.  Alexandrou and  J.  Pawliszyn.    "Supercritical Fluid  Extraction
    for  the Rapid  Determination of  Polychlorinated Dibenzo-p-dioxins
                                 11-104

-------
    and Dibenzofurans in  Municipal  Incinerator Fly Ash."  Anal.  Chem.
    61:  2770-2776 (1989).

13. F.I. Onuska and K.A.  Terry.   "Supercritical  Fluid  Extraction  of  2,
    3,7,8-Tetrachlorodibenzo-p-dioxin from Sediment Samples."  J.  High
    Resol. Chrom.  12:   357-361 (1989).

14. Sixth  Annual  Waste   Testing  and  Quality  Assurance  Symposium,
    Washington, D.C.,  July 16-20, 1990.

15. International Symposium  on Supercritical Fluid Chromatography and
    Extraction, Park City, Utah, January 14-17, 1991.

16. 201st American Chemical Society National Meeting,  Atlanta,  Georgia,
    April 14-19, 1991.

17. Thirteenth  International  Symposium on  Capillary  Chromatography,
    Riva del Garda,  Italy, May 13-16, 1991.

18. V. Lopez-Avila and  N.S.  Dodhiwala.  "Method  for  the Supercritical
    Fluid Extraction  of Soils/Sediments."   EPA/600/4-90/026,  September
    1990.

19. Carl  A.   Mabee,  Dearborn  Chemical  Company Limited,   personal
    communication, 1991.

20. T.L. Jones and T.C.H.  Chiang.  "An Interlaboratory Comparison Study
    of  Supercritical  Fluid  Extraction  for  Environmental  Samples."
    Proc. Seventh Annual Waste Testing and Quality Assurance Symposium,
    Washington, D.C.,  July 8-12, 1991.

21. M.K.L.  Bicking,   F.L.  DeRoos  and  T.G.  Hayes.    "An  Experimental
    Design Approach to  Optimization of  Supercritical  Fluid  Extraction
    Conditions."     Presented  at   the  International  Symposium  on
    Supercritical Fluid Chromatography and Extraction,  Park City,  Utah,
    January 14-17, 1991.

22. D.J.  Miller,  S.B.  Hawthorne  and  J.J.   Langenfeld.    "SFE  with
    Chemical  Derivatization  for  the  Recovery  of  Polar  and   Ionic
    Analytes."  Ibid.
                                  11-105

-------
57       THE RESEARCH STATUS OF SUPERCRITICAL FLUID EXTRACTION FOR
                       THE ANALYSIS OF PCBs IN INCINERATOR ASH

         Dr. Peter A. Pospisil. Manager Methods Development, Matthew A. Kobus, Chemist
         Methods Development, Charles R. Hecht, Senior Chemist Methods Development,
         Chemical Waste Management, Inc. 150 West 137th Street, Riverdale, IL 60627


         ABSTRACT

         The extraction time of PCBs from incinerator ash can be reduced to less than one
         hour using supercritical fluid extraction with carbon dioxide.

         SW-846 Method 8080 is currently the only EPA approved method for the extraction
         of PCBs in solid matrices. The sample is first extracted in a Soxhlet apparatus for 16
         hours with hexane/acetone.  The solvent volume is then reduced using a Kuderna-
         Danish apparatus, prior to GC-ECD analysis. The time intensive extraction step is
         the current limiting factor in reducing the turnaround time of the analysis.

         Supercritical fluids combine the mass transport properties of a gas with the solvation
         properties  of a liquids.  This research applied supercritical fluid technology to a
         specific combustion matrix, incinerator ash, both to  reduce the turnaround time of
         the method and to minimize solvent usage. The conditions required for extraction,
         including sample preparation, extraction temperature, extraction time, modifier type
         etc. were determined in a systematic manner to maximize the extraction speed and
         PCB recovery. The effect of the adsorptive properties of the matrix on the  analyte
         were also investigated.

         The application of supercritical fluid extraction technology for PCB extraction will
         enable  laboratories to provide same  day analytical service,  while  reducing
         laboratory costs.


         PURPOSE OF WORK

         The purpose of this work was to  apply supercritical fluid technology to develop a
         method for the extraction of PCBs from incinerator ash. The  technical  approach
         used to define method parameters was the  systematic evaluation of each element
         while holding all remaining variables constant. In this type  of study it is extremely
         important  to differentiate  between Comparative and Absolute extraction.  In the
         comparative situation there is a fixed goal based on data generated by an accepted
         method.   In the absolute case there  is  confirmation via  several alternative
         techniques that the extraction  is indeed complete. The  authors have  chosen a
         comparative study because of matrix considerations and analyte concentrations.
                                            11-106

-------
                                                                     Page 2

MATRIX OVERVIEW

Incinerator ash is defined as a combustion matrix.  It is  a by-product of  the
incineration process containing  a variety of newly formed active organic and
inorganic adsorption sites. The adsorptive strengths of the sites can vary as well as
their distribution throughout the ash particle.  Although the ash is primarily a glassy
product containing  a small amount of carbon polymer, it does have  some pore
structure. An ash particle is illustrated schematically in Figure 1.

The irregularly shaped polyhedron represents the basic ash particle contaminated
with PCBs. The light and dark dots represent weakly and strongly adsorbed PCBs.
The material on the surface is fairly easy to remove, its "removability" depends on
the strength of the adsorptive  site compared to that of the extractant.  The material
within the pore not only must be desorbed, but must diffuse to the pore mouth prior
to being swept into the flowing CCซ2.  This makes the removal of these materials
diffusion limited.   Any material occluded within the vitreous ash will never be
removed unless the ash is physically degraded  to expose the PCB to the extractant.


SUPERCRITICAL FLUID OVERVIEW

The molecules of a liquid are bonded by electrostatic forces, which are a function of
the molecule's polarity. The heat of vaporization represents the energy  required to
break these associative bonds as the liquid becomes a vapor. When  a liquid in
equilibrium with its vapor, is sealed in a tube and heated, the pressure of the closed
system rises and the liquid's heat of vaporization, and corresponding intennolecular
associative forces  decrease. When the associative forces reach zero, the liquid  and
gas phases become  one.  This temperature and  the corresponding pressure, which
are unique for each liquid, are known at the critical constants.

This  non-associated  supercritical  phase  (fluid)  has  unique  physio-chemical
properties. Its viscosity and diffusion constant approximate those of a gas, making it
an ideal material to permeate small pores. Its density and solvency approach those
of a liquid, enabling  it to dissolve a broad  range of organic compounds.   The
technique is not thermally driven, thus it is also  possible to extract thermally labile
and non-volatile materials.

The solvency of the mobile phase is a function of its density. Increasing the density
generally increases the solubility of larger molecular weight species. Carbon dioxide
is the most frequently used material  because of its low  critical  temperature,
inertness and safety.
                                   11-107

-------
                                                                     Page 3

INSTRUMENT DESCRIPTION

The supercritical fluid extraction instrumentation was purchased from the Lee
Scientific Co., and consists of the following components.

       Pump - A pump to supply pressurized liquid CO? to the extraction cell, that
             has the ability to deliver at least 20 mL ofliquid CC*2 at an operating
             pressure of up to 400 atmospheres, (6150 PSI).

       Oven - An oven capable of maintaining a temperature of 100 degrees C.

       Extraction Cell - consisting of a 10  cm, 4.6 mm ID HPLC column, 2 micron
             frits, and hardware to seal the end of the column.

       Restrictors - Fused silica capillary producing a carbon dioxide flow rate of 1.8
             mL/min of liquid, or 900 mL/min gas through the cell, usually about
             60 cm long, 50 micron ID.

       Liquid Carbon Dioxide - Supercritical fluid grade, in a tank that must contain
             a dip tube to deliver liquid product (Scott Specialty Gases).

Figure 2a shows a schematic of the apparatus used for the study. The components
occupy about 6 ftz of bench and floor space,  including space for  a single  gas
cylinder.


EXPERIMENTAL PROCEDURE

                              Sample Selection

The authors felt that the  best technical approach was to select a hazardous waste
incinerator ash sample containing  native, as opposed to spiked,  PCBs.  Typical
incinerator ash PCB concentrations approximated 3 ppm, about 15 times lower than
the regulatory level.

The initial work for this study was performed on this type of ash.  Reproducibility
difficulties, arising from sample size and GC detection limits constraints, required
that a sample of higher concentration be obtained. This, due to the nature of the
typical incinerator ash, dictated that the sample be spiked.  A 250  gram sample of
incinerator ash was then  spiked to a level of 50 ppm with Aroclor  1260,  by  the
technique of solvent evaporation in a rotary evaporator.

                      Sample Preparation and Analysis

A gallon of incinerator ash was crushed in a reciprocating jaw crusher.  All of the
ash was sifted through a  9.5mm screen to remove non-extractable items, such as
large metallic shards etc. Particles greater than 1.0mm were hand  ground in  a
mortar and pestle until they passed through the  16 mesh screen.  The ash sample
was homogenized by passing it through a riffler three times. The moisture content
was determined to be 3.14% by oven drying at HOC.  The PCB was determined to
be 3.4ppm of Aroclor 1260, using SW-846 Method 8080.


                                  11-108

-------
                                                                     Page 4

                        Preparation of Spiked Sample

Preparation of 250g of ash, spiked to a concentration of SOppm, was accomplished
by slurrying 30 gram portions of incinerator ash with acetone and spiking the slurry
with 1.5mL of  lOOOppm  Aroclor  1260 spiking solution.  After mixing for a few
minutes, the solvent was  driven off slowly in a rotoevaporator over the period of
approximately 45 min.  This spiking method was  repeated several times to create
enough ash for experimental  purposes.  The spiked ash was homogenized by
combining all fractions into a single container and riffling three times. The PCB
was determined to be 54.4ppm of Aroclor 1260, using SW-846 Method 8080.

                    Extraction Procedure and PCB Analysis

The cell was assembled as shown in Figure 2b.  The extraction cell was completely
filled with ash to eliminate dead space. The carbon dioxide was turned on slowly
and brought up to pressure within one to two minutes.  The collection fluid was five
mL of hexane in a ten mL graduated cylinder.  Additional hexane was added to the
graduated  cylinder at the completion  of  the  extraction  to compensate for
evaporative losses. All PCB analysis were performed  using SW-846 Method 8080,
GC-ECD (a capillary column technique), along with a  compliment of QA including
calibration standards, spikes and duplicate spikes.

                      Effect of Time on PCB Extraction

In order to  determine the effect of time on the extractability of PCBs from the ash,
weighed ash samples were extracted with unmodified CC>2 for varying time  periods
ranging from 5 to 60 minutes.  These data are presented in Figure 3 and show that
the extraction  curve  reaches  a plateau  in  about 45  minutes.   The  following
extraction conditions were used:

                  CC-2 flow                 = 1.8 mL/min
                 Temperature               = 100 ฐC
                   Pressure                  = 400 atm

                   Effect of Temperature on PCB Extraction

In order to  determine  the  effect  of temperature,  and  thus density,  on the
extractability  of PCBs from the ash, weighed ash samples were  extracted with
unmodified CC>2 for varying time periods and at the following temperatures:

             Temperature          Phase              Density
                  ฐC                                 g/mL
                  35               Liquid              .98
                  100                SCF                .76
                  200                SCF                .50
                  300                SCF                .36

These data are presented in Figure 4 and show that the maximum amount of PCB is
extracted at a temperature of 100ฐC.  This indicates that density is more important
than the increase in diffusion coefficient for removing PCBs from the incinerator
matrix.  Liquid CC>2 doesn't work as well as the corresponding supercritical fluid
because of the decreased transport properties of carbon dioxide in the liquid state.

                                   11-109

-------
                                                                     PageS

                    Effect of Particle Size PCB Extraction

In order to determine the effect of particle size on the extraction of PCBs from ash,
weighed ash samples of two different mesh sizes were extracted with unmodified
CC>2 for varying time periods ranging from 5 to 60 minutes. The data are presented
below and show that more PCBs are extracted from the sample when it is ground to
a mesh size of 60 or more.
                   Mesh Size               Amount Extracted
                    10 -20                  3.2
                    40 -60                  4.2

The following extraction conditions were used.

                  CC>2 flow                 = 1.8 mL/min
                 Temperature                = 100 ฐC
                   Pressure                  = 400 atm
                   Time                     = 40 min

                 Static-Dynamic Extraction using a Prc-Modificr

Based on information presented at the NIST conference by Mary McNally [1], liquid
methanol  was introduced directly  into the cell in an attempt to improve  the
extraction efficiency.  After sealing  the  cell  containing methanol, the cell was
allowed to equilibrate for five minutes and then the CC>2 pressure was brought up to
400 atmospheres and held in the static  mode  for ten minutes.  The run then
proceeded as described earlier.  Samples were extracted using methanol,  acetone
and toluene with only small improvements being noted.

                    Dynamic Extraction using Modified CCh

Based on the slight increase of PCB extracted using the different premodifiers, work
was initiated with modifiers directly added to the CC>2. This work and that  of Larry
Taylor [2]  at VPI, influenced the choice of 5% toluene as the modifier of choice for
CO2-

                         Extractive Reproducibility
                  Comparison of Soxhlet and SFE Extraction

Replicate  SFE runs using toluene modified CC>2 extractions and Soxhlet extraction
were made using the same sample. The results are compared in Figure 5.  These
data show recoveries, PCB by Soxhlet extraction, of 1009fe from the spiked ash with
relative standard deviation (RSD) of only 5%. The SCF extractions show 81% PCB
recovery,  with a RSD of 12%. This  brings the  SCF extraction  for PCBs  in
incinerator ash into the equivalency range of Soxhlet technology.
                                   11-110

-------
                                                                     Page 6

                      Research Overview and Direction

Although the extraction recovery data has increased over the time period of the
research, more work needs to be done, in the areas of improved PCB recoveries,
and analytical precision. The use of toluene as a modifier improves PCB recoveries,
but considering the superior extractive properties of supercritical fluid CO?, one
would expect the recoveries of the two techniques to be more equivalent. Altering
the toluene modifier content or experimenting with  different  modifiers are two
possible approaches.

Analytical  precision  data generated in earlier experiment, for both extraction
techniques  is the reverse of that presented in Figure 5. This suggests that there is a
additional  degree  of freedom that has not yet been addressed. Before an SCF
method can be  finalized comparable  precision  data  must  be  generated, to
demonstrate the absence of any additional variables.

The work clearly shows that SCF is a viable  technique for the extraction of PCBs
from incinerator ash. These problems may be overcome and the application of SCF
for PCB extraction from incinerator ash will soon become a reality.


CONCLUSIONS
1 -    PCBs can be extracted from incinerator ash using  supercritical fluid carbon
      dioxide modified with toluene.

2 -    The extraction time is reduced to 50 minutes from 18 hours and uses only 5
      mL  of collection solvent.

3 -    SFE is clearly a viable equivalent technique for PCB extraction to minimize
      solvent use in  the laboratory.

4-    Additional  work needs  to  focus  on  improved recoveries  and  analytical
      precision.


REFERENCES
1 -    Consortium on Automated Analytical Laboratory Systems, 1st Workshop on
      Supercritical Fluid  Extraction of Soild Environmental Samples, October 31,
       1990, National Institute  of Standards and Techonlogy,  Gaithersburg, MD
      20899

2-    L.T.Taylor; A.  J. Sequeira presentation  at Pittsburgh  Conference,  1991,
      paper No. 1003
                                   11-111

-------
                      FIGURE 1
             RGB's IN INCINERATOR ASH
IV)

-------
                 FIGURE 2A
            INSTRUMENT SCHEMATIC
A
 CO
         COM3KLLEK
                  FIGURE 2B
               CELL SCHEMATIC
                     11-113

-------
                  FIGURE 3
                SFE OF RGB'S
    EFFECT OF TIME ON PCB EXTRACTION
3.5
  PPM EXTRACTED
2.5

 2

1.5

 1

0.5

 0
        10
20     30     40
    TIME MINUTES
50
60
70
                  100 deg C (10-20)

-------
                            FIGURE
                          SFE OF RGB'S
             EXTRACTION TEMPERATURE SELECTION
Ol
3.5

 3

2.5

 2

1.5

 1

0.5

 0
           PPM EXTRACTED
                                               —X
           0     10



           — 30 deg C (liq)

           ~B~ 200 deg C
              20     30     40
                  TIME MINUTES

                 -+- 100 deg C
                 -*- 300 deg C
50
60
70
  150 deg C

-------
                                             FIGURE 5
                              Dynamic  Extraction  using  Modifier
a>
E
a
a
XX

m
o
D.

o
z
O
o
          60
          50 -
          40 -
          30 -
          20 -
          10 -
                             MEAN
               T   I    I


                 SPIKE
                             \  \  \
                          i   i    I   i    r

                              SOXHLET
                                                      MEAN
                                                                       \  \
I   I    I   I    I   I    I   I    I   !


         SFE TOLUENE MODIFIER

-------
58     SUPERCRITICAL FLUID EXTRACTION (SFE) OF TOTAL  PETROLEUM
               HYDROCARBONS (TPHs) WITH ANALYSIS  BY INFRARED
                                      SPECTROSCOPY

        Richard P. Lee, Methods Development Chemist, Mark L. Bruce. Director of Research and
        Development, and Marvin W. Stephens, Vice-President, Corporate Technical Director

                               Wadsworth/ALERT Laboratories Inc.
                                     4101 Shuffel Dr. N.W.
                                   North Canton, Ohio 44720

        ABSTRACT

        Infrared spectroscopy is an attractive analysis procedure for the screening of petroleum
        hydrocarbons in solid matrices because of its  low cost and rapid sample throughput.
        Coupled with off-line supercritical fluid extraction (SFE), this method provides a rapid
        monitoring procedure with an order of magnitude reduction in the amount of solvent used
        compared to the present Soxhlet method. This method uses supercritical carbon dioxide as
        the extraction solvent to remove the target components  from a solid sample and deposit
        them into a collection vial containing 5 mL of solvent. Freon-113ฎ has been replaced in
        this application by Fluorinertฎ FC-77 as the collection solvent. An extraction time of 25
        minutes at 400 atmospheres and an oven temperature of 60ฐC provides a rapid, effective
        means of extracting petroleum hydrocarbons from sand and Kaolin matrices.

        INTRODUCTION

        The methodologies for sample preparation have not kept pace with the developments in
        sample analysis. The method that Franz Ritter von  Soxhlet developed at the turn of the
        century has changed very little.  It is still the predominant method for the preparation of
        solid samples.  The need for an alternative sample preparation method is critical in the
        analysis of petroleum hydrocarbons.

        According to EPA estimates there are three to five million underground storage tanks in the
        United States (1).  Approximately 100,000 of these  tanks are believed to be leaking. In
        addition, as many as 300,000 more tanks are predicted to begin leaking  in the next five
        years (1).  At present, semivolatile petroleum  hydrocarbons are extracted by Soxhlet,
        sonication, or Soxtecฎ using an organic solvent followed by  gas chromatographic or
        infrared analysis.

        Freon-113ฎ  is used when the analysis is performed by infrared spectroscopy.  It is a
        known ozone depleter (2), making it unacceptable as a laboratory solvent. According to
        M.P. McCormick of Langley Center's Aerosol Research Branch, polar stratospheric clouds
        provide a surface on which chlorofluorocarbons (CFCs) can react to free the chlorine to
        react with ozone (2).  In accordance to the Montreal Protocol on Substances that Deplete the
        Ozone Layer (Montreal Protocol) and the Clean Air Act Amendments of 1990 (CAA),
        CFCs will be phased out by the year 2000.

        This paper presents a method in which supercritical fluid extraction (SFE) is used in the
        preparation of solid samples containing trace concentrations of petroleum hydrocarbons. A
        fluorocarbon is used as the collection solvent followed by infrared spectroscopic analysis.
                                             1-117

-------
Fluorocarbons contain no chlorine and should not be confused with CFCs.  There has been
no evidence of the cleavage of carbon-fluorine bonds, nor has there been a link between
fluorine and  the depletion of stratospheric ozone (3).  The benefits of SFE are well
documented  in several  informative reviews (4,5,6).   The use of the fluorocarbon
Fluorinertฎ  FC-77 as the collection solvent  in this method eliminates the  use of
Freon-113ฎ and compliments SFE's ability to provide an effective alternative to the present
TPH extraction methods.

INSTRUMENTAL. EQUIPMENT  and SUPPLIES

Supercritical Fluid Extractor
 Suprex, SFE/50
 5 mL extraction vessel
 600 mm fused silica restricter, 32 micron ID
Infrared Spectrograph
 Perkin-Elmer, 710 Infrared Spectrophotometer
 10 mm, 3 mL quartz cell
Hardware
 16 x 60 mm glass vial
 16 x 100 mm glass culture tube
 500 mL round bottom flask
 Modified Neilson-Kryger distillation apparatus
 Heating mantel, Glas-Col 115 volts, 270 watts TM106
 Temperature controller, Glas-Col 115 volts, 1500 watts, PL-312 Minitrol
 Glass beads, 5 mm OD
 Glass Pasteur pipets
 Filter paper ashless 41, Whatman
 Glass wool-silane treated, Supelco
Reagents and  Standards
 Freon-113ฎ, EM Science
 Fluorinertฎ FC-72,3M
 Fluorinert.ฎ  FC-77, 3M
 Isooctane, Mallinckrodt
 Xylenes, Mallinckrodt
 Hexadecane, EM Science
 Kaolin, Baker Analyzed
 Diesel fuel, retail Fuel outlet
 Sand, washed and dried, Mallinckrodt
 CO2, SFC grade with 1500 PSIA Helium headspace with dip tube, Scott Specialty Gases
 Hexafluorobenzene, Aldrich*
 Octafluorotoluene, Aldrich*
 Bromopentafluorobenzene, Aldrich*

* These compounds were only used in the initial collection solvent search.

RESULTS and  DISSCUSSION

The development for this method was conducted in two areas the search for a suitable
collection solvent and the optimization of supercritical fluid extraction parameters. The
approaches and representative results are discussed below.
                                    11-118

-------
Collection  solvent

The search for an appropriate collection solvent was broken down into four stages. In the
first stage, the catalogs of seven chemical vendors were examined for potential solvents.
This search resulted in sixty-six possible collection solvents.

In the second stage, vendors were contacted for additional product information regarding
the potential collection solvents as to their hydrocarbon solubility, IR spectra, and material
safety data. This information was used to eliminate all but six solvents. Small amounts of
these solvents were obtained for further testing.

The third  stage  was to confirm conclusions drawn from the information provided by the
vendors. All solvents were tested for hydrocarbon contamination by measuring the C-H
stretch  region  of the infrared  spectrum (2900-3100 cm'1).  Even with background
correction, the  concentration of hydrocarbon contamination of the possible collection
solvents was prohibitively high for all solvents except the Fluorinertsฎ FC-77 and FC-72.
The 3M Fluorinertsฎ FC-77 and FC-72 provided an acceptable baseline over the spectral
region of interest with background correction.

A 10,000  mg/L diesel fuel standard was prepared in 25 mL Freon-113ฎ for solubility
comparison studies with solutions of FC-77 and FC-72 that were saturated with diesel fuel.
Using Freon-113ฎ as a solubility reference, it was estimated that the solubility limit of
diesel fuel in FC-77 and FC-72 is approximately 5000 mg/L. The low molecular weight
fraction of the  diesel fuel appeared to be preferentially more  soluble than the higher
molecular weight components when compared to the Freon-113ฎ standard.

The same  three  standard solutions used in the hydrocarbon solubility studies were used as a
mock collection solvent.  This was to test the ability of the FC-77 and FC-72 to retain
hydrocarbons when  COi was bubbled through them.  These standard solutions were
purged for 10 minutes with a calculated liquid CO2 flow rate of 2.5 ml/min. Freon-113ฎ
retained 74% of the diesel fuel.  FC-77 and FC-72 retained 46%  and 35% respectively of
the diesel fuel.  When FC-77 was purged with a liquid CC>2 flow rate of 0.7 ml/min the
retention rate unproved to 85%.

In stage four, FC-77 was chosen as the collection solvent over FC-72 because of its
slightly higher retention of hydrocarbons.  A sand sample spiked with 100 mg/kg diesel
fuel was extracted off-line using 0.5 mL/min  supercritical CO2.  The absorbance  was
measured by comparing the  extract to FC-77.  When FC-77 was used as the collection
solvent, a negative absorbance was obtained in the 3050 cnr1 region of the infrared
spectrum  (the reference IR cell contained a higher concentration of a component than the
sample  cell  containing the extract). A volatile component of the solvent had apparently
been purged out by the CCซ2 during the extraction. A 3M technical representative indicated
that this volatile component could be contamination resulting from a methanol wash used in
the production process of the  Fluorinertฎ solvent.

A 100 mL volume FC-77 was distilled for thirty minutes. The absorbance of the distilled
FC-77 was measured relative to FC-77 that had not been distilled.  After distillation, a large
negative peak  was obtained that extended over the 2900 to 3100 cnr1 region of the
                                      11-119

-------
spectrum.  It was concluded from this result that the volatile component contamination and
a portion of the hydrocarbon background contamination were distilled out of the solvent.
Distilled FC-77 was used as the collection solvent in an off-line extraction using CO2 at
0.500 mL/min of sand and Kaolin blank samples  When the absorbance of the extract was
measured using distilled FC-77 for background correction, a flat baseline was obtained
without an interfering negative peak. Sand samples spiked with 1000 mg/kg diesel fuel
were extracted using the distilled FC-77.  The spiked concentration recovery was 85%. It
is concluded from this data, that FC-77 can be used as  the collection solvent for this
method.

Optimization of Extraction Parameters

The goal of this stage of the method development was to optimize the extraction parameters
(Tables 1&3) so that this method could be used with as broad a spectrum of environmental
matrices and hydrocarbon mixtures as possible.  Due to  the lack of standard reference
materials (SRMs)  with known "native" TPH concentrations, spiked samples were used in
the development of this method. Using the recovery of spiked analytes to prove quantitive
extraction of "native" analytes is an uncertain comparison method. There is no way of
determining how spiked compounds compare to native pollutants in their interactions or
absorptive qualities that result from long term association with a matrix.  This work
attempted to approximate  matrix  absorptive  interactions by tumbling at  a rate of 30
revolutions per minute diesel spiked Kaolin samples (porcelain clay) for approximately 24
hours.   Kaolin is  a highly  absorptive, fine particle matrix with a high surface area. By
tumbling  a spiked Kaolin sample these characteristic  qualities would enhance the
absorbance of the spiked compounds.  It was speculated that tumbling agitation over an
extended period of time would be a more realistic approximation  of a "real  world"
environmental  sample than a sample that has been spiked and immediately extracted.  Both
spiking techniques were used to investigate the ability of SFE to yield quantitative recovery
of petroleum hydrocarbons.  Representative results from each  approach are discussed
below.

Spiked samples (immediate extraction")

The initial evaluation of extraction parameters was conducted with the aid of statistical
experimental design software (Design-Easeฎ). Two Plackett-Burman designs were used to
screen the effects of the major extraction parameters (see Tables 1&3) using Freon-113ฎ as
the collection  solvent.  A  Plackett-Burman experimental  design is  a special class of
fractional factorial design.  This design was used to screen variables for further study by
isolating strong main effects. Interactions between variables were not considered.

Sand was the  matrix throughout  the first Plackett-Burman experimental  design.  All
samples were spiked with a 100 mg/kg TPH mixture consisting of 33% isooctane, 24%
xylenes, and 42%  hexadecane, by weight.  As Table 2 shows, TPH recoveries were high
throughout most of the range of conditions.  This demonstrates the ease with which
hydrocarbons can be extracted when matrix interactions are minimal.

In the second Plackett-Burman in which Kaolin and sand were the variables, the matrix
was indicated as a main effect (see Table 4). When Kaolin  was the matrix a mechanical
problem was discovered during the extraction . Due to the small partical size of this matrix,
periodic blockage of the extraction vessel frits resulted in fluctuation of the CC*2 flow rate.
                                     11-120

-------
This fluctuation resulted in  unacceptably low precision  with replicate extractions
(53%-120%).  A steady vigorous CO2 flow rate was obtained when a paper cartridge
containing the Kaolin was extracted. A paper cartridge was made out of folded filter paper
into which the Kaolin was directly weighed.  The paper was folded and rolled so that the
cartridge could fit into the extraction chamber. Samples of 2 g of Kaolin in paper cartridges
were spiked to a 1000 mg/kg concentration of diesel fuel and extracted immediately with
approximately 0.7 mL/min CO2 (400 atm, 60ฐC, 10 min. static, 15 min. dynamic) using
FC-77  as the collection solvent.  Recoveries of 80%-85% were obtained. A higher
pressure was needed in these extractions to improve the extraction effeciency of the higher
molecular weight components of diesel fuel.

Table  1.   Plackett-Burman #1*
Factors
Mass of sample
Pressure
Oven temperature.
Type of extraction
Equilibrium time
Extraction time
Orientation of
extraction cell
Positive
Version of Factor
5
360
60
Dynamic (Dyn)
10
20
Vertical (Vert)
Negative
Version of Factor
1
150
35
Static
5
10
Horizontal (Horz)
units
Grams
Atmospheres
Degree-C

Minutes
Minutes

  All samples were spiked to a 100 mg/kg concentration of the TPH mixture. Samples
were extracted using a 5 mL extraction vessel and a 600 mm length of 32 micron ID fused
silica restricter.  Initial equilibrium time was a static step prior to the dynamic extraction
period.

Table  2.   Plackett-Burman #1 Results

Factors
Mass of
sample
Pressure
Oven
temperature
Type of
extraction
Equilibrium
time
Extraction
time
Orientation of
extraction cell
Recoveries
run order
Standard
Order
Units
grams
Atmospheres
Degree-C

Minutes
Minutes

Percent
1
8
1
150
35
Static
5
10
Horiz
66.5
2
3
1
150
60
Dyn
10
10
Vert
66.6
3
5
1
360
35
Static
10
20
Vert
137
4
7
5
360
35
Dyn
5
10
Vert
76
5
1
5
360
60
Static
10
10
Horiz
84.9
6
4
5
150
35
Dyn
10
20
Horiz
84.8
7
6
5
150
60
Static
5
20
Vert
97.6
8
2
1
360
60
Dyn
5
20
Horiz
80.4
                                     1-121

-------
Table 3.   Plackett-Burman #2*
Factors
Pressure
Oven temperature
Equilibrium time
Extraction time
Mass of sample
Type of extraction
Matrix of sample
Orientation of
extraction cell
Positive Version
of Factor
360
60
5
20
1
Dynamic (Dyn)
Kaolin
Vertical (Vert)
Negative Version
of Factor
150
35
10
10
3
Static
Sand
Horizontal (Horz)
* All samples were spiked to a 100 mg/kg concentration of the '.
were extracted using a 5 mL extraction vessel and a 600 mm lengtl
silica restricter. Initial Equilibrium time was a static step prior to the
Units
Atmospheres
Degrees-C
Minutes
Minutes
Grams



fPH mixture. Samples
i of 32 micron ID fusee
: extraction period.
Spiked. Tumbled Samples (delayed extractions)

Thirty grams of Kaolin was spiked with 3 mL of 5000 mg/L diesel standard to give a 500
mg/kg concentration. This sample was tumbled at a rate of 30 revolutions per minute for
approximately 24 hours. Two gram aliquots of this sample were extracted off-line using
approximately 0.7  mL/min supercritical CO2 (400 atm, 60ฐC) using FC-77  as  the
collection solvent. The initial static step was 10 minutes followed by a dynamic step of 15
minutes. Using these  extraction conditions  the best recoveries  achieved were 62%.
Pressure was increased to 420 atm. and the temperature of the oven was decreased to 40ฐC
to achieve a greater supercritical fluid density.  The liquid CO2 flow rate was increased to
approximately 0.8 mL/min.  The recoveries did not change appreciably. Maintaining the
same pressure the extraction oven temperature was increased to 100ฐ C.  This also did not
significantly change the recoveries of the diesel fuel. It was thought that possibly water
could displace the spiked hydrocarbons from the Kaolin and allow them to be swept out by
the supercritical CO2 and into the collection solvent (7). Water was added to the extraction
chamber prior to an extraction in three different cases in 200 uL, 300 uL and 1 mL
volumes. Recoveries decreased as the volume of water increased.

Method validation is in progress and the data is not available at this time. Validation will
consist of detection limit studies, performance comparision with Soxlet, sonication, and
Soxtecฎ extractions and  precision studies involving several soil matrices.

Summary

Quantitative extraction  and analysis  of petroleum hydrocarbons was demonstrated by
immediate extraction of spiked samples. Because of the limitations that are inherent to
spiked samples in estimating recoveries of native components spiked  samples were tumbled
for 24 hours. This was to allow greater mixing and enable the matrix to absorb the spiked
compounds.  Some general comments can made from the development of this method:
                                    1-122

-------
 1) Fluorinertฎ FC-77 provides an adequate solubility limit for this application and with
pre-extraction distillation has an acceptably low hydrocarbon background contamination
level.

 2) Though the approximate limit of hydrocarbon solubility in FC-77 is 5000 mg/L, lower
molecular  weight hydrocarbons are more  soluble than higher molecular  weight
hydrocarbons.  This should be taken into consideration when extracting higher molecular
weight hydrocarbon mixtures such as motor or crude oil.

 3) When the matrix interaction with the analytes of interest is minimal, as in the case of
sand and immediatly extracted spiked samples, recoveries can be quantitatively high.
When the matrix interaction with analytes are increased, as with the tumbled spiked
samples, extraction of petroleum hydrocarbons is less efficient and might be improved by
using higher pressures than those explored in this study.

ACKNOWLEDGEMENTS
The authors gratefully acknowledge Suprex Corporation for their help and cooperation and
the 3M Corporation for the generous supply of Fluorinertฎ FC-77 and FC-72.

REFERENCES

1) Industrial Safety & Hygiene News, 01/88.

2) NASA's  Innovators, NASA Tech Briefs Vol.15, No.4, 1991, pp.114.

3) Danielson, R.D., Understanding Fluorocarbons, Electronic Packaging and Production,
July 1976,  pp. 78.

4) Hawthorne, S.B, .Analytical-Scale Supercritical Fluid Extraction, Analytical Chemistry,
Vol. 62, No. 11, pp. 633A-642A, 1990.

5) Anderson,  M.R.,  Swanson, J.T., Porter, N.L.,  Richter, B.E., Supercritical Fluid
Extraction  as  a Sample Introduction  Method for Chromatography, Journal of
Chromatographic Science, Vol.27, pp. 371-377, 1990.

6) King,  J.W.,  Fundamentals  and Applications of  Supercritical Extraction in
Chromatographic Science, Journal of Chromatographic Science, Vol.  27, pp. 355-364,
1989.

7) Hayes, P.C.Jr, Bruce, M.L., Stevens, M.W., Test Method to Extract TPHs from Soil,
American Environmental Laboratory, 12/90, pp. 25-28.
                                      i-123

-------
Table #4.  Plackett-Burman #2 Results*

Factors
Pressure
Oven
temperature
Equilibrium
time
Extraction
time
Mass of
sample
Type of
extraction
Type of
matrix
Orientation of
extraction cell
Recoveries
run
order
Standard
Order
Units
Atmosphere
Degree-C
Minute
Minute
Gram



Percent
1
9
360
60
10
10
1
Static
Sand
Horiz
152
2
1
360
60
5
20
3
Dyn
Kaolin
Horiz
40
3
6
150
35
5
20
1
Dyn
Sand
Horiz
154
*These recoveries are not background corrected
Kaolin has been found to have a hydrocarbon co
4
2
150
60
10
10
3
Dyn
Sand
Horiz
68
5
4
150
60
5
20
3
Static
Sand
Vert
86
6
7
360
35
5
10
3
Static
Sand
Vert
38
7
3
360
35
10
20
1
Dyn
Sand
Vert
164
8
11
360
35
10
20
3
Static
Kaolin
Horiz
37
9
12
150
35
5
10
1
Static
Kaolin
Horiz
76
10
5
150
35
10
10
3
Dyn
Kaolin
Vert
59
11
8
360
60
5
10
1
Dyn
Kaolin
Vert
102
12
10
150
60
10
20
1
Static
Kaolin
Vert
150
by subtraction of native contaminates from sand or Kaolin.
ntamination of up to 40 mg/kg by other extraction methods (7).

-------
59                   APPLICATION OF SUPERCRITICAL FLUID EXTRACTION OF
                                 DIOXINS/FURANS FROM SOIL AND PUF
         Jong-Pyng Hsu, Ph.D.t  Joseph C. Pan, Ph.D.
         Kevin Villalobos, Gregory P. Miller
         Southwest Research Institute
         6220 Culebra Road
         San Antonio, Texas 78228-0510
        The supercritical fluid extraction of dioxins/furans from soil and polyurethane foam plug (PUF) is always
        interesting in the environmental application.

        In this study, dioxins/fiirans spiked on soils or PUFs will be evaluated.  The soils or PUFs will be
        extracted with Suprex supercritical fluid extractor using carbon dioxide. The final extract will be analyzed
        by a GC/MS.

        The extract can also be directly transferred  from  the supercritical fluid extractor to a GC/MS  for the
        purpose of reaching lower detection limits.

        For both cases, five concentrations of target compounds will be evaluated to determine the detection limit
        and linearity of the entire system.
                                                1-125

-------
60          THE APPLICATION OF SUPERCRITICAL FLUID CHROMATORAPHY
                   TO THE ANALYSIS OF HERBICIDES AND PESTICIDES
                                    IN TCLP EXTRACTS
         Charles R. Hecht. Senior Chemist Methods Development, Dr. Peter A. Pospisil,
         Manager  Methods   Development,  Matthew  A.  Kobus,   Chemist   Methods
         Development, Dr. Mark F. Marcus, Director of Analytical Programs, Chemical
         Waste Management, Inc. 150 West 137th Street, Riverdale, IL 60627
         ABSTRACT

         Supercritical fluid chromatography with electron capture detection was used to
         consolidate two hazardous waste methods for pesticides and herbicides, in TCLP
         extracts, into a single cost effective protocol.

         Herbicide and pesticide analysis of TCLP  extracts of hazardous waste is currently
         performed using SW-846 Methods 8150 and 8080 respectively. Both methods utilize
         different  sample preparation techniques.  Method  8150 incorporates  ether
         extraction, caustic hydrolysis and diazomethane esterification, while a selection of
         different preparatory  methods can  be utilized  for Method  8080.  The overall
         methodology required for the analysis produces  a turnaround time of up to two
         days, for a group of 6 to 8 samples.

         A  single SFC-ECD analytical method, with a consolidated sample preparation
         procedure, can be applied to the analysis of both the herbicides and pesticides in
         TCLP extracts. The technology is rugged enough to handle the compound type
         distribution of both the analytes and the typical interferences found in hazardous
         waste samples. Chromatograms, response factors, and the SFC mass spectroscopic
         data used to confirm the identity of the peaks will be presented. The application of
         SCF-ECD technology to the consolidation of these methods clearly reduces the
         costs and sample turnaround times.


         INTRODUCTION

         Pesticides and herbicides are analytes of major concern to regulatory agencies. The
         analysis of the pesticides Lindane, Heptachlor, Chlordane,  Endrin, Methoxychlor,
         and Toxaphene, coupled with  the chlorophenoxy herbicides 2,4-D and  Silvex, are
         now required for TCLP extracts. Two different methods are used for these analytes:
         SW-846 Method  8080 for the pesticides and  Method 8150 for the herbicides. The
         sample preparation procedures for both methods  are significantly different, labor
         intensive, and  hazardous.  Both   methods  require multiple  separatory  funnel
         extractions using ethyl ether or methylene  chlonde. Increasingly health, safety and
         environmental concerns are being raised from the usage and disposal of these and
         other hazardous solvents. Also of prime concern is the usage of diazomethane as a
         methylating agent for the chlorophenoxy acid herbicides. The Merck Index, Edition
         10 lists  diazomethane as a  very toxic, insidious  poison, that may explode  upon
         heating or contact with rough glass surfaces.
                                            1-126

-------
                                                                      page 2
PURPOSE

The purpose of this work is to develop a single streamlined method utilizing SFC-
ECD, which can determine both pesticides and herbicides in TCLP extracts. The
simplification of the extraction, hydrolysis and derivatization steps will increase
analyst  safety,  reduce  hazardous solvent usage, and  significantly reduce  the
analytical costs and sample turnaround time.


SCF THEORY

The molecules of a liquid are bonded by electrostatic forces. The energy required to
break these associative bonds, as the liquid becomes a vapor, is known as the heat of
vaporization. When a liquid, in equilibrium with its vapor, is sealed in a tube and
heated, the pressure of the closed system rises and the liquid's density and heat of
vaporization decrease. When the associative forces reach zero, the liquid  and gas
phases become one. This temperature and the corresponding pressure are unique
for each liquid  and termed the critical constants, which for CU2 are 31ฐC and 73
atm., respectively.

A supercritical phase has the solvency of a  liquid. It can dissolve, and thus partition
the analyte(s) between the mobile and stationary phase. It has the low viscosity and
high diffusion coefficient of a gas, resulting in low column pressure  drops and rapid
mobile/liquid phase equilibration. The chromatographic efficiencies approach those
of GC. The technique is not thermally driven  thus it is also possible  to  analyze
thermally labile and non-volatile materials.  A supercritical fluid, therefore combines
me best qualities of a gas and liquid in a single process. Carbon dioxide is the most
popular material used  for  extraction because of its low  critical temperature,
inertness, safety and ease of purification.

The solvency of the mobile phase can vary and is  a function  of its  density. Density
programming has the same effect on an SFC separation as temperature and solvent
composition have on GC and LC. When utilizing density programming to improve a
separation, the system controller must vary  the pressure to linearize  the density.

INSTRUMENT DESCRIPTION

The supercritical fluid instrumentation used for this work was purchased from the
Lee Scientific Company, the mass  spectrometer from the Finmgan Company, and
the electron capture detector from the Hewlett Packard Company. Two instruments
were used for the study and were configured as follows:

Instrumental SFC-MS

             Lee Scientific Model 600 SFC pump, oven and controller interfaced
             to a Finnigan INCOS-50 Mass Spectrometer via a heated transfer line.
             Transfer line manufactured by Lee Scientific.
                                    11-127

-------
                                                                     page 3


      Restrictor - Fifty micron fused  silica frit restrictors producing a carbon
             dioxide linear velocity of 0.6 cm/sec at 75ฐC and 75 atm were used for
             all SFC-MS work. Restrictors were purchased from Lee Scientific.

Instrument #2  SFC-ECD

      Supercritical Fluid  Chromatograph - Lee Scientific Model 600 SFC pump
             and controller interfaced to a Hewlett Packard 5890 GC equipped
             with an BCD detector.

      Restrictor - Fifty micron fused  silica frit restrictors producing a carbon
             dioxide linear velocity of 1.8 cm/sec at 75ฐC and 75 atm were used for
             all SFC-ECD work.

Liquid Carbon  Dioxide - Supercritical grade. The tank must contain a dip tube to
             deliver liquid product (Scott Specialty Gases).


EXPERIMENTAL PROCEDURE AND DATA REVIEW

                     EQUIVALENCY SUPPORT DATA

                     Analyte Recovery from The Empore
                              Solid Phase Disk

In order to evaluate the performance of the method, initial studies were done using
blank TCLP extraction  fluid,  and extracts from a variety  of sample matrices.
Recovery data was generated for the pesticides, herbicide acids, and their respective
methyl esters. The Empore disk extraction, hydrolysis, and esterification efficiencies
were determined. Recovery data is presented in Figure 1. The analyte recovery
range of 59% to 117% shows that the Empore Solid Phase disk  is  an acceptable
means of analyte concentration.

Included in the study were sample types such as: filter press cake, oil dry +  gas,
hydrocarbon  contaminated soil,  dimethyl  disulfide  spill  debris, terminal plant
sludge, grease, spent carbon filter media,  and others.


                     Analyte Chromatographv and Mass
                         Spectrometric Verification

In order to determine that  no analyte alteration, reaction or modification occurred
during the  supercritical fluid chromatographic procedure, the chromatograph was
linked to a mass spectrometer for spectral confirmation. The column  selected for
this work  was a Phenyl-5, 10 meter, 50 micron ID, haying  a 0.25  micron  film
thickness. A Finnigan INCOS-50 was used for the confirming spectrometer. Figures
2 and 3 show the SFC-MS total ion chromatogram and library matched  spectra
produced by the run. The  NIST library  searches for all of the analytes produced
library match factors of 800+. The data show that no analyte alteration, reaction or
degradation occurred in the system.
                                   1-128

-------
                                                                     page 4


                     EQUIVALENCY METHOD DATA

           CWM Combined Herbicide - Pesticide Sample Preparation

The first part of the method involves the two part streamlined sample preparation
procedure. Both procedures are compared with the corresponding SW-846 methods
in Figures 4 and 5.

1 -    One hundred mL of TCLP extract, including the surrogate, is made alkaline
      using  a KOH solution. The extract is stirred at  70ฐC for one hour. The
      extract is reacidified to a pH of 2 and passed through an Empore solid phase
      extraction disk, which removes the herbicide acids. The acids are eluted from
      the disk with methanol. Methane sulfonic acid is added to the extract and the
      extract solution is heated for one hour to esterify the acids and reduce the
      solution volume.

2-    A second  250  mL volume  of TCLP extract,  including surrogates, is
      neutralized with KOH and passed through a second Empore solid phase
      extraction disk. The pesticides are eluted using acetone. The extract solution
      volume is reduced using nitrogen blowdown.

The two extract solutions are combined, Aldrin is added as an internal standard and
the analysis is performed on the combined extract.

                         SFC-ECD Chromatography

An SFC-ECD chromatogram of the herbicide methyl esters and pesticides is shown
in Figure 6. All of the peaks are sharp, baseline resolved and of approximate equal
intensity.

Initial studies were performed using a mass spectrometer but to simplify the method
even further, final studies were done  using the ECD detector. All work was done
under isothermal,  density programmed chromatographic  conditions. They are as
follows:

            Oven temperature 100ฐC isothermal
            Initial density  0.13 (equivalent to 74 atm)
            Hold        2 minutes
            Ramp at .006 to  0.42 (equivalent to 178 atm)
            Ramp at 0.2 to  0.75( equivalent to 390 atm)
            Final Hold Time   7 minutes

Response factor, relative retention time and method detection limits are listed in
Figure 7.
                                    1-129

-------
                                                                     page 5
                   EPA Equivalency Study Data Generation

To begin formal EPA Equivalency spike recovery studies have been done on a wide
variety of sample types. To this date six samples have been completed in duplicate.
The six sample types include incinerator ash, wastewater treatment sludge, plating
filter  cake,  soil contaminated  with  1,1,1-TCE, grease/oil debris,  and copper
reclamation tailings.

Recovery data from these six sample types is listed in Figure 8. The data quality is
high and meets formal EPA Equivalency requirements.


CONCLUSIONS

1 -     A single SCF method requiring less than 5 hours has been developed for the
       analysis of both herbicides and pesticides in TCLP  extracts, at  detection
       levels well within the regulatory range.

2-     The data generated shows that  it meets the EPA's formal Equivalency
       criteria.

3 -     The use of ether  and  methylene chloride  is eliminated.  No  explosive
       derivatization agents are used, and general solvent use is reduced by a factor
       of 5.

4-     The analyte recovery  and SFC-MS support  data  show that there  is  no
       chemical reactivity between the analytes and the supercritical phase.
                                   11-130

-------
                                                                          FIGURE  1




                                                                   EQUIVALENCY SUPPORT DATA




                                                                       %  RECOVERIES
U

LINDANE
HEPTACHLOR
g-CHLORDANE
a-CHLORDANE
ENDRIN
METHOXYCHLOR
TOXAPHENE
2,4-D
SILVEX
TCMX(SURR)
DBC(SURR)
2,4,5-T(SURR>
Extract 1 Extract 2
91 92
55 56
76 80
82 84
99 98
85 88
81 77
98 100
55 56
54 57
101 102
87 81
Extract 3 Extract 4
92 91
59 49
105 88
108 92
110 100
114 96
148 126
85 118
93 68
61 47
124 133
87 110
Extract 5 Extract 6
94 92
58 60
73 75
78 80
103 93
" 93 87
123 140
91 88
51 61
50 79
98 95
88 107
Extract 7 Extract 8
94 92
65 63
76 77
81 80
102 90
106 93
139 102
95 106
79 84
57 51
99 95
121
Extract 9 Extract 10
92 91
60 62
75 71
78 77
99 98
96 89
115
90 99
95 99
48 48
96 92

AVERAGE
92
59
80
84
99
95
117
97
74
55
103
97

-------
       •1C                          QATfti Slซt2 II
       12/tl/ป 13i37iซ               CM.Ii CM.TM 13
       SffTUEl E8UIU STB  12739-2^27  St-TWtf
                  SCMS UN T023M
 1N.*-
                                                 tatt
                                                                                         3276M.
                                                                                       FIGURE 2

                                                                                        SFC-MS

                                                                                     PESTICIDES
                                                                43il9
                                                                               47t39
                                                      SCON
                                                      urn
      RIC                          ปTte SSli files       SOWS  9M TO 14ซ

      SMVLEt WMH  MMOT  B1PW4VL  FULL FRIT
      CQKK.i l.WL  LOOP  1GMDW   CT/OIP  IH^.1* IHJ
      WMZl C   1,1X3  UซLl H  ป. ซ,ป QUNNi ซ•ซ, l.i J •  BOSEt U 21,
1M.I
                  SFC-MS

                TOXAPHENE
                                                                        131C
                    18M
   36il6
11W
44t2ซ
1298
48i22
13M
52i24
14W  S0*i
56.23 TIIC
                                               11-132

-------
    1*8
I 288*9
             HID LIBMRV SEMCH (UMNmซ>
             12/18/38 UiXiM * 39i2C
             Smil EOUIU STD 57-19WW 82739-2^27
             COWS, i 13NOW  PWซ.-Vtl ซTQป RET 0
              11818 TO 11025 9MB) - 113*1 TO 11814
DOTfli S1831 11821
CM.Ii CM.TMI •  3
                                                                               BOSC
                                                    own
V2    38
                      L1WMC
                       SMVLE MINUS LIBMtY
                    A  ,li i   11  iJ  ซ	MI    •    J   -    ,1.1
                                  191
                                                                                    FIGURE  3

                                                                                      SFC-MS

                                                                                   LIBRARY  MATCH
                                                                                       SPECTRA
                                                                                          399
             HID LIMWtt SEMCH (LIBWTM)                   MTfli S787 8112*
                     Ili26i88 *• 27i99                       CM.li COLTAi •   3
                    K 11739-17-9
             COKB.I 1788BW  PtOm.-9  18KTB RET GflP
SMflf
                                         .,  1     ..  ll    IL.
                                                                                   K/2i  199
                                                                               RICl   349184.
C9.H8.03.CL2
I  21788
FUR 722
                       2,4-0 nmm. Esmt
                                ll             II,        1.    II,     i,     ll
                       SM*U MINUS LIMRY
                          II    p
                                         IM  l.ll
                                                11-133

-------
                        FIGURE 4

              FLOWCHART COMPARISON BETWEEN
         EPA METHOD 8150 and CWM HERBICIDE METHOD
                    FOR TCLP EXTRACTS
    EPA METHOD 8150

1) ETHER EXTRACT AT pH=2
     1 time at 150 mL
     2 times at 50 mL

2) HYDROLYSIS
     2mLof37%KOH

3) ETHER EXTRACT AT pH = 11
     2 times at 20 mL

4) ETHER EXTRACT AT pH=2
     1 time at 20 mL
     2 times at 10 mL
     2mLof(l:3)H2SO4
  CWM METHOD

NONE REQUIRED
TCLP EXTRACT HYDROLYSIS
   2mL37%KOH

NONE REQUIRED
NONE REQUIRED
                               EMPORE EXTRACT AT pH=2
                                 21mLofMETHANOL
                                 2mLof(l:l)H2SO4

                               EMPORE ANALYTE ELUTION
                                 15mLofMETHANOL
5) EXTRACT TRANSFER TO K-D
     1 time at 30 mL

6) INITIAL K-D

7) MICRO K-D

8) DIAZOMETHANE ESTERIFY
NONE REQUIRED
MSA ESTERIFICATION
   1001L of MSA
9) GC-ECD ANALYSES
SFC-ECD ANALYSES
COMBINED EXTRACT
(See Figure 6)
                         11-134

-------
                       FIGURES
             FLOWCHART COMPARISON BETWEEN
         EPA METHOD 8080 and CWM PESTICIDE METHOD
                   FOR TCLP EXTRACTS
    EPA METHOD 8080
1) SEPARATORY FUNNEL EXTRACT
    3 times at 60 mL MeCb
    NEUTRALpH
2) K-D TO 5 mL
3) NITROGEN BLOWDOWN
4) SOLVENT EXCHANGE
    9mLHEXANE
5) GC-ECD ANALYSES
  CWM METHOD
EMPORE DISK EXTRACT
    20 mL ACETONE
    NEUTRAL pH
ANALYTE ELUTION
    15 mL ACETONE
NITROGEN BLOWDOWN
NONE REQUIRED
SFC-ECD ANALYSES
COMBINED EXTRACT
                         11-135

-------
  AN .ป! 132964820. BNC
 RUN ป   168      flAY  13.  1991   19I19I24
 START
                                                                  FIGURE 6
                                                                  SFC-ECD
                                                              CHROMATOGRAPHY
                             11.331
                                                    ~ 11.142
                          ฃ9.ias  Z.'t-O Methyl  Ester
                1.093
                  39.722
                      44.720
                                                           30.957  Tetra-Chloro-M-Xylene  (Surr)
                                                          —— 31.923  Silvex Methyl Ester
                                                          ———  33.013 2,4,5-T Methyl Ester  (Surr)
                                                                   34.778  Lindane
———  38.394   Heptachlor


 40.849  Aldrin  (interns: 31.)

™""~~  42.398   Ga 1MB-Chi".' !3fl^
                                                                                 43. 899
                                                                  43.372   Endrin
                                               48. a is   Methoxychlor
                                                             40.084   Dibutylchlorendate (Surr)
TIMETABLE  STOP
                                                11-136

-------
                        FIGURE?

    RESPONSE FACTORS , RELATIVE RETENTION TIMES
        AND METHOD DETECTION LIMITS(NG/ML)
TARGET COMPOUNDS

                                 RF        RRT       MDL

2,4-D Methyl Ester                 .267        .712          10
Silvex Methyl Ester                .949        .781          10
Lindane                         .981        .851         0.5
Heptachlor                       1.20        .939         0.2
g-Chlordane                      1.13        1.05         0.2
a-Chlordane                      1.24        1.07         0.2
Endrin                          .992        1.11         1.0
Methoxychlor                     .525        1.19         0.6
Toxaphene(4 Peaks)               .140        1.13          75

                                           l'.26
                                           1.31

SURROGATES

2,4,5-T Methyl Ester               .885        .808         NA
Tetra-chloro-m-xylene              1.06        .758         NA
Dibutylchlorendate                .866        1.17         NA
INTERNAL STANDARD 0.5 PPM ALDRIN
                           11-137

-------
                                                                                    FIGURE 8
                                                                           TCLP EXTRACT SPIKE RECOVERY DATA
                                                                                    DUPLICATE RESULTS
00

LINDANE
NEPTACHLOR
g-CHLOftDANE
a-CHLOROANE
EHDRIN
NETHOXYCHLOR
TOXAPHENE
2.4-D
SILVEX
TCHX(SURR)
DBC(SURR)
2,4,5-T(SURR)
Uastewater
Treatment Sludge
94 99
86 93
92 94
92 95
111 106
108 108
121 126
93 96
86 81
75 87
101 102
83 96
Soil Cont.
With 1,1,1-TCE
106 98
88 68
96 78
100 84
108 92
121 110
118 90
48 82
92 66
86 68
110 100
38 75
Plating Filter
Cake
94 94
76 79
89 88
90 88
100 98
110 110
59 104
94 84
85 88
59 70
100 100
90 93
Incinerator
Ash
99 97
75 84
86 92
90 95
98 98
104 104
108 102
97 84
86 94
76 78
104 102
95 94
Grease/Oi I
Debris
76 71
76 98
86 82
90 91
88 89
102 100
78 128
76 83
81 93
82 84
94 90
81 88
Copper Reclaimation
Tailings
80 90
77 84
80 84
78 78
78 86
92 100
84 99
94 83
84 86
76 88
88 94
90 91
AVERAGE
92
82
87
89
96
105
101
84
85
77
98
84

-------
61
  PROBLEM SOLVING  IN  THE ORGANIC EXTRACTIONS  LABORATORY;
                         HERBICIDES

John Doeffinger, Teresa Wittwer, Jim Giannella,  Chris Lott,
Larue  Stanton,   Erik  Alverson,  Sean  Fitzgerald,  Kristine
Klinger, Deborah Smith.  Ph.D..  IEA,  Inc. - New  Jersey,  628
Route 10, Whippany, New Jersey 07981

ABSTRACT

In the  commercial  laboratory,  it is often difficult to  set
aside time and resources  to improve and optimize execution of
acceptable methods without  a  dedicated  "special  projects"
group.   At  IEA, Inc.  -  New Jersey, method development  for
sample preparation is carried out by the Organic Extractions
Group  as a  whole, from experimental  design through  data
interpretation,   within  the  normal  flow   of   production
laboratory  work.     This   problem-solving   process,   with
supporting  data,  including matrix  spike   and  surrogate
recoveries of real  samples,  is illustrated for the extraction
of herbicides.

INTRODUCTION

The "routine" extraction  of samples for herbicide analysis by
SW846 methods has presented difficulties for the laboratory.
Problems including the use  of the  reagents  diazomethane and
ethyl ether  in a production environment, the inconsistency of
spike  and  surrogate  recoveries,  and  the elaborate sample
manipulations required, result  in a  procedure that is time-
consuming and frustrating.   IEA, Inc.  -  NJ  has optimized an
existing trial  USEPA  method1 which  addresses  the compounds
2,4-D, Silvex, and 2,4,5-T.  We present here results of this
optimized IEA method for water and leachate samples and show
evidence that it is adaptable to a variety of matrices.

METHOD DEVELOPMENT

Preliminary  trials using the above mentioned USEPA method as
written yielded  poor surrogate and spike recoveries.  However,
we were  interested in pursuing optimization  of  this method
because of the many advantages it offered. A general meeting
of the entire extractions staff,  QA, and laboratory management
was  called  to  organize  a  systematic  approach   for method
optimization.   Data  from extractions using the  trial USEPA
method were  discussed and all agreed upon the next course of
action, which was a limited experiment to be conducted along
with normal extraction batches.  The group reconvened  the next
                                  11-139

-------
week to discuss the results of the experiment.  This process
evolved into an on-going program of weekly meetings followed
by limited experiments, which resulted in method definition
and refinement.

EXTRACTION OF HERBICIDES FROM AQUEOUS SAMPLES

Poor performance  (shown  in  Table 1.0)  of the  trial USEPA
method forced critical examination of the variables displayed
in Table 2.0.
                          TABLE 1.0
TYPICAL SURROGATE AND SPIKE RECOVERIES
USING USEPA TRIAL METHOD1
SAMPLE
LEACHATE 1
LEACHATE 2
LEACHATE 3
LEACHATE 4
LEACHATE 5
LEACHATE 6
LEACHATE 7
2,4-DB
SURR.
59
73
78
97
100
30
54
2,4-D


60
52
36


SILVEX


66
54
59


2,4,5-T


49
45
91


                            11-140

-------
                         TABLE 2.0
          VARIABLES INVESTIGATED FOR OPTIMIZATION
              OF HERBICIDE EXTRACTION
  1.0  Acid-Washing of all glassware and materials
  2.0  pH (2, 3, <2)
  3.0  Use of microsnyders
  4.0  Concentration of extract to dryness prior to
       esterification
  5.0  Use of Na2SO4 column cleanup
  6.0  Presence of acetic acid in TCLP leachates
  7.0  Use of BC13 versus BF3
  8.0  Temperature for esterification
  9.0  Methylene Chloride volume versus sample volume
 10.0  Alkaline hydrolysis required?
We found that acid-washing of all equipment was most crucial
to the  success  of the extraction, and  that aqueous samples
should be  taken to pH 1.   Neither use of  microsnyders nor
concentration to dryness  improved recoveries.  Also, we found
Sodium  Sulfate  column clean-up  to be  unnecessary.   Since
leachates performed better than  "plain" aqueous samples, we
investigated the addition of acetic acid to the water samples,
but the recoveries were  unaffected.   Best recovery occurred
when esterif ication was carried out with BF3  at 60ฐ C.   We are
still optimizing sample and solvent volumes,  and investigating
the necessity of the alkaline hydrolysis.  The IEA method is
summarized in Table 3.0:
                             11-141

-------
                          TABLE 3.0
     METHOD FOR HERBICIDE EXTRACTION OF WATER SAMPLES
1
2
3

4
5
6
7
8
9
0
0
0

0
0
0
0
0
0
       Acid wash all glassware and materials
       Adjust pH of sample to < 2.0
       Extract 500 ml sample three times with 60/40/40 ml
       methylene chloride
       Concentrate extract to 4.0 ml
       Solvent exchange with hexane
       Concentrate extract to 1.0 ml
       Esterify with BF3 at 60ฐ C  for  10 minutes
       Dilute to 5.0 ml with hexane
       Add 10 ml 7% NaS0, vortex
    .                  ?4,
 10.0  Collect  1.0 ml hexane extract for analysis
We tested this  IEA  method  for extraction of the most recent
USEPA WS series proficiency samples;  the results are shown in
Table 4.0.

                         TABLE 4.0
PROFICIENCY RESULTS: WS027
ANALYTE
2,4-D
SILVEX
REPORTED
45.9
17.9
TRUE
46.3
18.1
ACCEPTANCE LIMITS
15.1 - 59.1
7.47 - 24.4
The majority of aqueous herbicide analyses requested are for
TCLP  leachates;    surrogate recovery  results  for  leachate
blanks and samples are presented in Figures 1.0 and 2.0; spike
and  surrogate recovery  data  are  presented  in Table  5.0.
Results of  similar analyses of  a  series  of  spiked reagent
blanks are  shown in Table  6.0.   Note that  leached samples
consistently performed better than water samples.
                            11-142

-------
TABLE  5.0
FREE ACID SURROGATE AND SPIKE RECOVERIES
TCLP LEACHATES
SAMPLE
LEACH BLANK
LEACH BLANK SPIKE
LEACHATE X3
LEACHATE X3 MS
LEACHATE BLANK
LEACH BLANK SPIKE
LEACHATE WP1
LEACHATE WP1 MS
LEACH BLANK
LEACH BLANK SPIKE
LEACHATE WP9
LEACHATE WP9 MS
LEACH BLANK
LEACH BLANK SPIKE
LEACHATE D3
LEACHATE D3 MS
LEACHATE D4
LEACHATE D4 MS
LEACH BLANK
LEACH BLANK SPIKE
LEACHATE 191
LEACHATE 191 MS
LEACHATE 380
LEACHATE 380 MS
2,4-DB
100
104
106
109
95
97
98
93
116
101
114
120
75
89
78
85
94
104
85
79
80
77
70
70
2,4-D

83

95

83

82

88

105

65

64

72

81

75

72
SILVEX

85

94

83

77

89

110

71

75

74

66

69

58
2,4,5-T

66

78

67

66

76

89

59

57

61

55

54

53
    1-143

-------
                          TABLE 6.0
FREE ACID SURROGATE AND SPIKE RECOVERIES
REPLICATE BLANK ANALYSIS
SAMPLE
BLANK
BLANK SPIKE 1
BLANK SPIKE 2
BLANK SPIKE 3
BLANK SPIKE 4
BLANK SPIKE 5
BLANK SPIKE 6
2,4-DB
99
104
79
108
112
104
114
2,4-D

61
44
67
53
60
68
SILVEX

83
58
82
75
80
91
2,4,5-T

49
29
51
38
42
53
Although  there were  incidents of  low spike  and surrogate
recoveries, the results were generally good. Lower recoveries
were consistent throughout a complete batch, indicating that
an  isolated  extraction  procedure,  not  the  method,  had
performed  poorly.    The  compound 2,4,5-T was  the  poorest
performer  throughout  the aqueous studies, with  the lowest
recoveries noted in spiked reagent blanks.

EXTRACTION OF HERBICIDES IN ESTER FORM

We suspected that herbicides in various ester forms would not
be  converted  efficiently  to  the methyl  ester, since  the
alkaline  hydrolysis step  was  omitted4.   An  extraction  of
reagent water  spiked  with the  propylene  glycol  butyl ether
ester  of  Silvex  (Silvex  PGBE)  was   carried  out  without
performing an alkaline hydrolysis step.  The results of this
preliminary investigation are presented in Table  7.0 with an
example chromatogram shown in Figure 3.0.
                            11-144

-------
                          TABLE  7.0
PERCENT RECOVERIES OF ESTERS OF HERBICIDES
(BLANK SPIKES)
SAMPLE
BLANK
BUTYL ESTER,
BUTYL ESTER,
SILVEX
SILVEX DUP
2,4-DB
(SURR.)
109


SILVEX

26
24
The recovery of Silvex  PGBE  as  a methyl ester was marginal,
but we feel that additional  development work will result in
improved recoveries.  Further experiments are in progress to
determine the conversion efficiency  of  a greater variety of
esters.

EXTRACTION OF HERBICIDES FROM ORGANIC MATRICES

IEA, Inc. - NJ is frequently called upon to perform the TCLP
on  organic  matrices,  but  we had been  unable to  carry out
herbicide analysis on this  matrix type.  We tried an approach
similar to  a  BNA  partition 3.   First,  the  sample is washed
with a basic aqueous solution to separate all acids into the
water layer.    The  water  fraction  is  then acidified and
extracted with methylene  chloride.    The  methylene chloride
extract is then subjected to the IEA herbicide procedure.  The
results  presented  in Table  8.0  indicate that  analysis of
herbicides in  organic matrices is meaningful.   This capability
is  important  to  clients because  it  permits  complete sample
characterization;  inability  to  test  the  herbicide fraction
allows potential classification of a sample as a hazard.
                            1-145

-------
                          TABLE  8.0
PERCENT RECOVERY OF HERBICIDES IN AN ORGANIC MATRIX
SAMPLE
BLANK
BLANK SPIKE
SAMPLE 1
SAMPLE 2
SAMPLE 3
2 , 4-DB
96
82
69
96
82
2,4-D

64
61
88
52
SILVEX

59
48
54
47
2,4,5-T

52
51
57
44
In this example:
Sample 1 -
Sample 2 -
Sample 3 -
One gram  of motor oil  was
with  methylene  chloride,
during dilution.
One gram  of motor oil  was
with  methylene  chloride,
during dilution.
One gram  of motor oil  was
with  methylene  chloride,
diluted to  10 ml
spike  was  added

diluted to  25 ml
spike  was  added

diluted to  25 ml
spike  was  added
               directly to motor oil, prior to dilution.
EXTRACTION OF HERBICIDES FROM SOIL/SEDIMENT MATRICES

Most of our non-TCLP requests for herbicide analysis are for
soils;  therefore  we  wanted  to  extend  the  IEA  herbicide
procedure to soil analysis.  Data from preliminary trials of
the IEA method  are shown in Table 9.0.   Note  that the soil
sample showed better spike recovery than a blank sand matrix.
Further development is in progress.
                             1-146

-------
                          TABLE  9.0
FREE ACID SURROGATE AND SPIKE RECOVERIES
SOILS
SAMPLE
BLANK (SAND)
BLANK SPIKE
SOIL Dl
SOIL D2
SOIL Ul
SOIL Ul MS
SOIL Ul MSD
SOIL U2
SOIL BK
SOIL Rl
SOIL R2
2,4-DB
37
56
49
82
73
75
81
77
66
98
79
2,4-D

52



98
94




SILVEX

60



115
113




2,4, 5-T

93



91
86




CONCLUSIONS

Because the demand for herbicide analysis fluctuates and the
scope of requested analytes is limited, a method that can be
performed without a period of  fine-tuning required to achieve
acceptable recovery is needed. The IEA method described here
has the advantages  of simplicity,  use of routine reagents, and
improved  reproducible analyte  recoveries.   Production  is
doubled.    Along  with  the  development  of  the  extraction
procedure, there were many intangible benefits to working on
the problem as a group.  However,  the  ultimate success will be
in the achievement of full method approval.
                            1-147

-------
SUMMARY

The concept of  using  regular group problem-solving sessions
has resulted in development of a simple, effective methylene
chloride  extraction  for  the  determination of  herbicides.
Generally  acceptable  spike  and  surrogate  recoveries  of
commonly requested herbicide  analytes  in aqueous, leachate,
soil, and  organic matrices were  achieved.   This method is
easily implemented and results in increased capacity.
REFERENCES
     Analysis of Pesticide Residues in Human and Environmental
     Samples, June  1980,  "Determination  of  Some  Free Acid
     Herbicides in Water",  USEPA.
     Analysis of Pesticide Residues in Human and Environmental
     Samples, June 1980, "Sampling and Analysis of Water for
     Pesticides", USEPA.
     Test Methods for Evaluating Solid Waste, Volume IB, SW846
     3rd Edition, Method 3650 "Acid-Base Partition Cleanup",
     USEPA
     Analysis of  Pesticides  in Water, Volumes I  - III,  CRC
     Press, 1986.
                             1-148

-------
o
u
K
U
hJ
C
130



140



130



120



110



100



 90



 80



 70



 60



 50



 40



 30
                            FIGURE 1.0



                 2,4—DB Surrogate  Recovery

                        Leach Blank/Leach Blank Spike
                 2,4-DB  Surrogate Recovery

                           Samola/SamDle Soike
                UHU I I I I
                           20
                                           40
                                                           60
                              Sample No.
                              11-149

-------
         l'\
                          FIGURE 3
                            SILVEX
                           BUTYL ESTER
                                 CN
s

                                 CN
       >Q  GO  O>     CNOQD       J  I       CD    O  O       II       f\    h^ป
       cH   •  sO     O ซH  |^>       I  \       XT    CD
        •  ?H   •      •  •   •
       rH     rH     CN CN  CN

-------
CN
r\
          FIGURE 4

         HERBICIDES


         ORGANIC MATRIX

-------
52         Infrared  Microsampling  for  the  Qualitative  Analysis  of
                      Organics  Extracted  from  Soil  Samples

                             M.P. Fuller and F.J. Weesner
            Nicolet Instrument Corporation, Spectroscopy  Research Center,
                       5225 Verona Road,  Madison, WI  53711

         Infrared  spectroscopy  provides  a unique "fingerprint" that can  often
         be used  to identify  the  structure of unknown  compounds.   Often
         spectral library searches are used to make this type of identification.
         The primary  difficulty  in determining  the  identity  of unknowns in
         this manner is usually spectral contamination caused  by the
         absorbances  of components  other than  the  target  compound.    It is
         possible  to separate many organic compounds  using thin  layer
         chromatography (TLC)  techniques.   The  relative  elution distance can
         then  be  related  to separations  of  standard mixtures and the structure
         of unknown compounds  elucidated.  Many  times, however,  it is
         impossible  to  absolutely identify  components  in  this  manner.

         Recently  we have investigated the  use of TLC separations  combined
         with  infrared  microspectroscopy  for the  identification of organic
         compounds  extracted from soil samples.  The soil  extract  is
         separated using standard TLC procedures and  the  resulting  "spots"
         are analyzed  both directly on  the  plates and after extraction with
         appropriate  solvents.   The results of these experiments will be
         described and  compared with  GC/FT-IR measurements obtained on
         the soil extracts.
                                        11-152

-------
63
A PERFORMANCE COMPARISON STUDY OF DIFFERENT TYPES OF DEVICES FOR SOLID PHASE
EXTRACTION
       Y. Joyce Lee.  E. Neal Amick,  Jack A.  Serges, Lockheed  Engineering &  Sciences
       Company, Las Vegas,  Nevada

       Gary L. Robertson, U.S.  Environmental Protection Agency,  Environmental
       Monitoring  Systems Laboratory,  Las  Vegas, Nevada
       ABSTRACT


       Solid Phase  Extraction (SPE)  is rapidly becoming an alternative to  separately

       funnel extraction and continuous liquid-liquid extraction for the isolation of

       organic compounds from environmental samples.   SPE can provide analytical data

       in a timely  manner for decision-making during site inspections, remediations,

       and emergency removal activities.   It is especially useful if there is

       knowledge  regarding potential matrix interference at the site.  SPE utilizes a

       compact manifold that can process multiple samples simultaneously.   Solvent

       usage is minimized and the sample preparation can be performed rapidly.   The

       extraction can be accomplished by using glass cartridges, plastic cartridges,

       or extraction disks.  Characteristics of these extraction devices,  including

       recovery,  capacity, interferences, and contamination for polycyclic aromatic

       hydrocarbons, phenols, pesticides, and Aroclors have been compared and will be

       discussed.  The performance results presented are data generated as part of

       the Superfund Contract Laboratory Program Quick Turnaround Method development

       and validation process.
       Notice:   Although the research described in this article has been funded
       wholly or in part by the United States Environmental Protection Agency through
       contract number 68-CO-0049 to Lockheed Engineering & Sciences Company, it has
       not been subjected to Agency review and therefore does not necessarily reflect
       the views of the Agency and no official endorsement should be inferred.
                                        11-153

-------
  STANDARD REFERENCE SPECTRA for MS/MS QUALITY  ASSURANCE,
      PERFORMANCE EVALUATION, and PROFICIENCY TESTING:
              XC rf1Q TANDEM MASS SPECTROMETERS
Richard I.  Martinez, Research Chemist, Chemical  Kinetics  and
Thermodynamics Division, National Institute  of Standards  and
Technology, Gaithersburg, Maryland  20899
ABSTRACT

The collisionally-activated dissociation (CAD)  of  the
acetone cation (m/2 58) can be used for quality assurance,
performance evaluation, and proficiency testing of CAD
measurements in tandem mass spectrometry ( MS/MS) instruments
which use rf-only multipole collision cells.   The  absolute
branching ratios (product distributions) of  the CAD fragment
ions,  when measured as a function of the center-of-mass
collision energy E  ,  can provide an objective  basis for
quality assurance wnenever MS/MS methods are  used  (viz.,  to
validate how well the target thickness,  ion  containment
efficiency,  and collision energy are being controlled in
various instruments).

INTRODUCTION

Tandem mass spectrometry (MS/MS) instruments  which use  rf-
only muljj^inole collision cells are complex ion-optical
devices.      Such MS/MS instruments are denoted hereinafter
by the generic symbol XIrf1Q, where Q denotes  a quadrupole
mass filter,  Irf1 denotes an rf-only multipole  collision
cell used for collisionally-activated dissociation (CAD),
and X can be either a Q or a sector analyzer  (denoted by  EB
or BE) .   There are several types of Xtrf]Q MS/MS instruments
(e.g.,  QqQ,  BEqQ, QoQ, QhQ, etc.;  here q,  h,  and o denote,
respectively,  rf-only collision cells which  use quadrupole,
hexapole,  and octopole rod assemblies).   There  are currently
more than 400 XCrf]Q instruments worldwide,  representing  a
capital investment of more than $200M.

In this note we discuss an objective basis for  quality.
assurance of CAD measurements in "dynamically-correct"
X[rf]Q instruments.   The practical tuning  criteria and
guidelines herein can be used routinely (e.g.,  on  a daily
basis)  to check instrument and/or operator performance  once
it has been certified ( cf.  section 4a of ref.  10)  that
                             1-154

-------
                   Q
dynamically-correct  product distributions can be measured
in any particular instrument.

Technical Background:

To study ion-neutral reaction mechanisms in XIrf]Q
instruments,  it is crucial that one measure dynamically-
correct  product distributions which are instrument
independent.   Otherwise one may measure a distorted
representation of the reaction dynamics, which consequently
can lead to incorrect conclusions about the pertinent
reaction mechanisms.   The dynamical prerequisites for
obtaining dynamically-correct  product distributions
(branching ratios) within Xtrf] Q instruments have been
detailed elsewhere.

It follows,  therefore,  that to develop an instrument-
independent database (or library) for MS/MS measurements
within X( rf]Q instruments one must obtain substantially the
same representation for any reaction [e.g.,  CAD]  occurring
within any such instrument (i8e.0 no discrimination effects;
see Appendix of reference 9) .  '

A measurement protocol   was developed at the National
Institute of Standards and Technology (NIST;  formerly
National Bureau of Standards)  to provide a basis for precise
and accurate (ฑ10%)  instrument-independent,  dynamically-
correct measurements within XqQ instruments.   The precepts
of the NIST protocol should also be applicable to other
types of Xtrf) Q tandem mass spectrometers which have strong
focusing properties (e.g.,  QhQ, QoQ,  etc.),  so long as the
collision energy range is the same as for XqQ instruments.

The NIST protocol10 was validated by the recent NIST-EPA
International Round Robin   which indicated that at least
50% of the QqQ instruments which have been sold and are
currently in the field can provide an instrument-
independent,  dynamically-correct representation of any ion-
neutral reaction mechanism when this kinetics-based
measurement protocol is used.   Hence,  the NIST protocol can
be used to develop an instrument-independent database of  CAD
spectra for dynamically-correct X[ rf] Q tandem mass
spectrometers  '    (and/or to study the kinetics and
mechanism of ion-neutral reactions).   The NIST protocol is
to be incorporated into EPA' s SH-846 Test Methods for
Evaluating Solid Haste as an 8000 series tuning procedure
for dynamically-correct XI rf 1 Q instruments.
                            11-155

-------
DISCUSSION

After one has proven that a given XIrf]Q instrument is
capable of measuring dynamically-correct product
distributions ( cf.  section 4a of ref.  10), it becomes
imperative that standardized'operating conditions be
maintained to,ensure precision and accuracy of the CAD
measurements,   especially each time CAD spectra are to be
taken for inclusion in a NISI standardized database.

Maintaining Standardized Operating Conditions:

The CAD of the acetone cation is especially well suited for
the requisite quality assurance, performance evaluation, and
proficiency testing applications because:

(a)  it provides a relatively simple test case (if one
     cannot generate instrument-independent CAD spectra for
     the acetone cation,  then it may not be possible to
     develop an instrument-independent database under any
     conditions),  and

( b)  there is a wealth of information about the unimolecular
     and collisionally-activated dissociation of the acetone
     cation. T9 37

and, most importantly,  because of the following unique
characteristies:

(c)  there are distinct differences in the energy
     dependences of the branching ratios obtained under
     single-collision (SO vs.  multiple-collision 
     conditions .   Therefore, one can readily determine
     whether or not the target thickness is within the
     single-collision regime.  Comparison of the {SO and
      data in Tables 1 and 2 of ref.  9 indicated that
     the control of the target thickness becomes extremely
     critical if one hopes to measure instrument-independent
     product distributions (i.e.,  CAD spectra).

(d)  the production of 1 5  is a significant decomposition
     channel.     This allows one to gaugeghow well the
     reaction-induced mass discrimination '    due to CAD is
     controlled in various Xtrf1Q instruments (i.e.,  how
     well one can compensate for differences in ion
     containment efficiencies,  especially for low-mass
                            11-156

-------
     daughter ions (here md9ปghter/!Tparent= 1 ^5?= ซ• ซ) 1 .
     If one cannot measure^aynamicaliy-correct  branching
     ratios vs.  E   for 15 ,  then one obtains an incorrect
     representatiSB of the low-energy CAD mechanism  for
     Me2CO •.

(e)   the energy dependence of the branching ratio foe.,
     production of 15  goes through a sharp maximum.    This
     allows one to gauge how well the collision energy  is
     controlled in various instruments.

It is proposed,  thereforeg7that the standard spectra for  the
CAD of the acetone cation   be used by the MS/MS community
to periodically recheck the performance of dynamically-
correct Xtrf]Q instruments (viz.,  how well the key MS/MS
parameters such as target thickness, ion containment
efficiency,  and collision energy are being controlled in  a
XC rf]Q instrument).   These standard CAD spectra can  also  be
used to test the proficiency of Xfrf1Q operators of  varying
skill levels,  thus providing an objective basis for  quality
assurance whenever one uses  MS/MS methods such as the  EPA' s
SH-846 method.

Reference Spectra for the CAD of the Acetone Cation:

Table 1 shows the absolute branching ratios (from ref.  37)
for the CAD of Me-CO • [generated by 70 eV electron
ionization (EIX or acetone!.   They were measured with the
NIST protocol   in NIST's dynamically-correct QqQ
instrument   under single-collision conditions ( Ar  target)
at the center-of-mass collision energies (E  ) indicated.
The E   of Table 1 were selected iteratively to optimize  the
information about competitive reaction channels (including
the absolute maximum branching ratio for Me  production at
E  = 32.6 eV) .   The reader is referred to the EXPERIMENTAL
s8
-------
The ketene cation (m/z 42; branching ratios of 0.02-0.06 for
E  = 1-60 eV) is a minor  CAD  fragment.
 cm

MS/MS Quality Assurance:

One's ability to reproduce the  dynamically-correct  branching
ratios shown in Table 1 for the CAD  of  the acetone  cation
should indicate that one's XCrfJQ  instrument is functioning
properly, and is ready to measure  standardized CAD  spectra.
This provides an objective basis for quality assurance of
CAD measurements in "dynamically-correct"   XCrf]Q
instruments.

Branching Ratios and Target Thickness:

One should be able to replicate the  values in Table 1  to
within the maximum uncertainty  indicated  by the bracketed
values in Table 1 for branching ratios  >0. 01.   This would
ensure that the Ar target thickness  is  within the single-
collision regime.

Collision Energy:

It was shown in ref.  37 that  the complementary energy
dependences for production of MeCO  and Me  are due to a
competition between three fast,  primary (direct)  reactions,
each of which opens sequentially at  its respective  threshold
energy [viz., (1), (2), and (3)1.

Me2CO*-—ป MeCO* + Me- (X  2A"2>            AH= 0. 82 eV     (1)

       —ป Me* + Me- + CO                  AH= 4.24 eV     (2)

       —* MeCO* + Me- ( B, 1 2A'1)          AH= 6.55 eV     (3)

That is,  the maximum in the branching ratio vs.  E   curve
for Me  production at E   = 32. 6 eV corresponds toฐS'he
opening of reaction (3) when  the collisionally-activated
Me2CO • has acquired an internal excitation E.  .- 6.55 eV.
This E.  . is corroborated by  the increased pro-auction of 42
for E  X32.6 eV, which was attributed (in  ref.  37)  to the
opening of a new direct reaction channel  [(5)  or (6)]1  for
production of H2C=C=0 .


Me2CO+- —> H2C = C = 0+ + CH4                 AH= 0.89 eV     (4)

        ~ป H2C=C=0* + H + Me-            AH= 5.43 eV     (5)
                             1-158

-------
        —ป• H2C=C = 0  +• CH2 + H2            AH=  5.69  eV      (6)

Hence,  E  = 32.6 eV corresponds to E-  . -  5.43-6.55 eV  [for
reactionim( 3) ,  (5),  (6)].  That is,  afi uncertainty in  E.  .
of ca.  1 eV (=6.55-5.43) corresponds  to an  uncertainty In
Ecm of ca.  5-6 eV at Ecm= 32/6 eV.

SUMMARY

     The absolute branching ratios (product distributions)
for the CAD of the acetone cation (measured as  a function of
E  )  provide an objective basis for  quality assurance,
performance evaluation,  and proficiency testing of CAD
measurements in dynamically-correct  tandem  mass
spectrometers which use  rf-only multipole collision cells.
That  is, by replicating  NIST' s standard reference  spectra,
an operator can determine that the key MS/MS  parameters
(e.g.,  the target thickness, ion  containment  efficiency,  and
collision energy) are under control,  and  that one's Xfrf]Q
instrument is functioning properly and is ready to measure
standardized CAD spectra .

ACKNOWLEDGMENTS

     R.I.M.  gratefully acknowledges  the funding of this
work,  in part,  by the US Environmental Protection Agency
[the  Atmospheric Research and Exposure Assessment  Laboratory
(AREAL)] under Interagency Agreement  IAG  #DH-13934363-1 ,  and
helpful discussions with Dr. L. D. Betowski  ( EMSL - Las
Vegas).
                             11-159

-------
REFERENCES

1.   P. H.  Dawson, (a)  Quadrupole Mass Spectrometry and  its
    Applications,  Elsevier,  Amsterdam,  1976; ( b) Adv.
    Electronics  Electron Phys.  53.  153 (1980).

2.   S. C.  Davis  and B.  Bright;  Rapid Commun. Mass Spectrom.
    4.,  186 (1990),  and references therein.

3.   P. H.  Dawson, Int.  J.  Mass Spectrom.  Ion Phys. 20.  237
    ( 1976) .

4.   P. H.  Dawson  and J. E.  Fulford,  Int.  J.   Mass  Spectrom.  Ion
    Phys. 42. 195  ( 1982) .

5.   P. H.  Dawson, J. B.  French,  J. A.  Buckley, D. J. Douglas,
    and  D. Simmons,  Org.  Mass Spectrom.  (a) 17.  205  (1982);
    ( b)  17.   212  ( 1982) .

6.   (a)  K. L.  Busch,  G. L.  Glish,  and S. A.   McLuckey, Mass
    Spectrometry/Mass Spectrometry:  Techniques  and
    Applications of Tandem Mass Spectrometry, VCH, New York,
    1988; ( b) P. H.  Dawson and D. J.  Douglas, in  Tandem  Mass
    Spectrometry ( ed.  by F. H.  McLaf f erty)  , p. 125, Hiley,  NY
    (1983);   ( c)  R. A.  Yost and D. D.  Fetterolf, Mass Spectrom.
    Revs. 2_,  1  (1983).

7.   B.  Shushan,  D. J.  Douglas,  H. R.  Davidson, and S.  Nacson,
    Int.  J.   Mass Spectrom.  Ion Phys.  46.   71 (1983).

8.   R.I.  Martinez,  Rapid Commun.  Mass Spectrom.  ( a)  2.,  8
    ( 1988) ;   ( b)  2_,  41  ( 1988) .

    The  term "dynamically correct" was coined (see Appendix
    of reference 9)  '    to indicate those  branching  ratios
    measured in  Xtrf]Q instruments which  correspond  to the
    distribution of reaction products which, in principle,
    would be observed at the scattering center  of an
    idealized crossed molecular beam machine (if one were
    able to  integrate over all angles the  ion intensities  of
    each reaction  product channel).   This  correspondence  is
    attributed  to  the strong focusing properties of  rf-only
    multipoles  which provide high ion-containment
    efficiencies for ions scattered through a broad  range  of
    angles.      Hence,  dynamically-correct  branching  ratios
    are  those which have been appropriately corrected  for
    discrimination effects,  and,  therefore, provide  an
                             11-160

-------
    instrument-independent representation of the primary
    ion-neutral  interaction of  A  +B.

9.   R.I.  Martinez  and B.  Ganguli,  Rapid Commun. Mass
    Spectrom.  3.,  427 (1989).

10.  R.I.  Martinez,  (a)  J. Res.  Natl.  Inst.  Stand.  Tech.  (US)
    94. 281  (1989);  ( b)  J.  Am.  Soc.  Mass Spectrom.   1_,  272
    (1990).

11.  R.I.  Martinez,  Rapid Commun.  Mass Spectrom. 3.,   127
    (1989).

12.  R. I.  Martinez,  "WORKSHOP:  Instrument-Independent  CAD
    Database", Proceedings of the 37th ASMS Conference  on
    Mass Spectrometry and Allied  Topics (p.  1560),   Miami
    Beach, Florida,  May 21-26,  1989.

13.  R.I.  Martinez  and B.  Ganguli,  Rapid Commun. Mass
    Spectrom.  1,  377 (1989).

14.  D. J.  McAdoo,  F. H.  McLafferty,  and J. S.  Smith,  J.  Am.
    Chem.  Soc. 92.  6343 (1970).

15.  J. H.  Beynon,  R. M. Caprioli,  and R. G.  Cooks,  Org.  Mass
    Spectrom.  9_,  1  ( 1974) .

16.  C. C.  van  de  Sande and F. H.  McLafferty,  J.   Am.  Chem.  Soc.
    97. 4617  (1975).

17.  D. J.  McAdoo  and D. N.  Hitiak,  J.  Chem.  Soc., Perkin  II,
    770 (1981).

18.  C.  Lifshitz,  (a)  J.  Phys.  Chem.  87.  2304 (1983);  ( b)
    Int.  J.  Mass  Spectrom.  Ion  Phys.  43,  179 (1982).

19.  F. H.  McLafferty,  D. J. McAdoo,  J. S.  Smith,   and R.
    Kornfeld,  J.  Am.  Chem.  Soc.  93.  3720 (1971).

20.  J.  Diekman,  J. K.  MacLeod,  C.  Djerassi,  and J. D.
    Baldeschwieler,  J.  Am.  Chem.  Soc.  9J_,  2069 (1969).

21.  I.  Powis  and  C. J.  Danby,  Int.  J.  Mass Spectrom.  Ion
    Phys.  32.  27  (1979).

22.  R.  Stockbauer,  Int.  J.  Mass Spectrom Ion Phys.   25.  89
    (1977).
                              1-161

-------
23.  D. J. McAdoo and C. E. Hudson,  Int. J.  Mass  Spectrom.  Ion
    Proc.  59.  77 ( 1984)  .

24.  G.  Bouchoux and Y.   Hoppilliard,   Can.  J.  Chem.  60.  2107
    (  1982).

25.  A. J. Stace and A. K.  Shukla,  Int. J.  Mass Spectrom.  Ion
    Phys.  37.  35 < 1981)  .

26.  H. J. van  der Zande,  D.  P.  deBrui jn,   J.  Loss,  P. G.
    Ristemaker,  and S. A. McLuckey,  Int.  J.  Mass  Spectrom Ion
    Proc.  67.  161  (1985).

27.  C.  Lifshitz and E.   Tzidony,  Int. J.  Mass Spectrom Ion
    Phys.  39.  181  (1981).

28.  P. J. Derrick and S.  Hammer urn,  Can.   J.  Chem.  64.  1957
    (1986).

29.  C. S. T.  Cant,  C. J.  Danby,  and J. H. D.  Eland,  J.  Chem.  Soc.
    Faraday Trans.  II,   71.   1015 (1975).

30.  D. M. Mintz and T.  Baer,  Int.  J.   Mass  Spectrom.  Ion Phys.
    25.  39  ( 1977) .

31.  R.  Bombach,  J. -P.  Stadelmann,  and J.  Vogt,  Chem.  Phys.
    72.  259 (1982).

32.  A. K. Shukla,  K.  Qian,   S. L.  Howard,   S. G.  Anderson,  K. H.
    Sohlberg,  and J. H.   Futrell,  Int. J.  Mass Spectrom.  Ion
    Processes 92.  147 (1989).

33.  K.  Qian,  A.  Shukla,  S.   Howard,  S. Anderson,  and J.
    Futrell,  J.  Phys.  Chem.  93.  3889 (1989).

34.  K.  Qian,  A.  Shukla,  and J.  Futrell,  J.  Chem.  Phys.  92.
    5988 (1990).

35.  A. K. Shukla,  K.  Qian,   S.  Anderson,   and J. H.  Futrell,  J.
    Am.  Soc.  Mass Spectrom.  1_,  6 (1990).

36.  N.  Heinrich,  F.  Louage,  C.  Lifshitz,  and H.  Schwarz,  J.
    Am.  Chem.  Soc.  1 10.  8183 (1988).

37.  R.I. Martinez and B. Ganguli,  J. Am.  Soc.  Mass Spectrom.
    2_,  xxx  (1991);  submitted for publication.
                              1-162

-------
                 37
    In this  paper   we describe the kinetics and mechanism
    of the CAD of the acetone cation which indicates that
    there  is a correspondence between the distribution of
    internal energies accessed by the photoionization of
    acetone  [viz.,  the PEPICO data of ref.  29 and 311 and
    the energy deposition function accessed by collisions!
    activation of acetone cations formed by 70 eV El [viz.,
    our CAD  data].   That is,  the low-energy CAD of the
    acetone  cation involves electronic transitions (rather
    than vibrational excitation),   '    and dissociation
    occurs primarily from the same electronic states in both
    the CAD  and PEPICO experiments.

                                   37
    The concordance of our findings   with those of
    PEPICO  '    and,molecular beam experiments
    indicates again   that the HIST kinetics-based protocol
    developed in this laboratory makes it possible for one
    to measure dynamically-correct product distributions
    which  have been appropriately corrected for
    discrimination effects.   That is,  one can obtain an
    undistorted (instrument-independent)  representation of
    ion-neutral interactions (e.g.,  CAD).   This is essential
    for the  development of a standardized,  instrument-
    independent MS/MS database for XIrfJQ instruments.   The
    data in  Table 1  constitute some of the first elements of
    such a database.

38.  R.I.  Martinez,  Rev.  Sci.  Instrum. ,  58.  1702 (1987).
                            10
                            11-163

-------
 Table 1.  Branching  Ratios1  vs.  E   for the CAD of  58*  from Acetone

Parent Ion: C-H..O*  (m/z  58)    Source Compound: 2-Propanone  (>99.7Jt)
lonizat
cm
( ซT_1
1. 2
4. 1
10. 6
32. 6
44. 9
61. 2
Bcn,
( eV)
1. 2
	 - j b 	
ion Mode: 70 eV electrons Target Gas: Ar (single
a . Branchi no Ratios for the CAD of 56
UiL
24
[101
31
(15)
32
[151
34
(251
35
(101
34
[201
(14*)
0. 0000
0.0000
0. 0000
0. 0128
1751
0. 0000
0. 0000
(15*)
0. 0153
(8]
0. 0151
(101
0. 1581
( 31
0. 2561
[101
0. 0444
( 10]
0. 0278
[ 10]
Branch! nor

(31*)
0. 0000
(39*)
0. 0000
collision)
fr26*) <27*) (28*) (29*)
0. 0000 0.
0. 0000 0.
0. 0005 0.
[50] [
0. 0054 0.
(30) I
0. 0059 0.
(1001 (
0. 0000 0.
[
Ratios for
(40*) (
0. 0000 0.
0000 0.
0000 0.
0046 0.
20] (
0578 0.
71 [
0882 0.
15J [
0195 0.
20]
the CAD
41*1 I
0000 0.
0000 0.
0000 0.
0033 0.
15] (
0062 0.
251 t
0059 0.
100] C
0000 0.
[
of 58*
42*) (
0195 0.
0000
0000
0054
15]
0308
151
0738
20]
0124
351

43*)
965
   4. 1        0. 0000   0. 0000  0. 0000  0. 0000  0. 0241   0. 961
                                                 (20]     (4]

  10.6        0.0008   0.0000  0.0000  0.0000  0.0177   0.810
               [351                              [251     [51

  32.6        0.0020   0.0029  0.0014  0.0047  0.0218   0.598
               (601     (501     [501     (301     (15)     (2]

  44.9        0.0059   0.0036  0.0012  0.0084  0.0323   0.730
               (1001    (35]     (1001    [301     (201     [31

  61.2        0.0000   0.0120  0.0084  0.0169  0.0580   0.845
                        (20]     (301     (201     (151     (51


   Thฃ CAD^of 58* produces  only the  fragment ions  indicated  (e.g.,
   14 ,  15 ,  etc.)  and a  complementary neutral  fragment (not shown).
   Numbers in square  brackets represent maximum possible uncertainty
   in the cross  section a and in the branching  ratios,  expressed as a
   percentage of each o and of each  branching ratio.   Hence,  at
   E  =44.9 eY,   the branching ratio  for 58 —ป43  is 0.730 (ฑ0.02 max),
   wfiFle that for 58  —ป31   is 0.0059 (ฑ0.0059 max).
                                  11-164

-------
*e           IMPROVED TECHNIQUES FOR FORMALDEHYDE ANALYSIS
              BY HPLC USING AUTOMATED SAMPLE PREPARATION
                          AND DIODE ARRAY DETECTION

       by Brian  Goodby.  Smita  Vasavada, Jim  Carter  and Larry  Schaleger
             B C  Analytical, 801  Western Avenue, Glendale,   California, 91201

       ABSTRACT

       Formaldehyde,  one of  the  more  widely  produced  intermediates  in
       the  U.S.  chemical  industry,  is formed by  combustion  and  biological
       processes,   making   its  presence  in  the  environment   ubiquitous.
       Because formaldehyde  is  a  probable carcinogen,  reliable  analytical
       methods for  identifying  trace  levels  of  this  analyte  must  be  found.
       In this  study,  we  examined some  of  the  difficulties  involved in  a
       common HPLC  method, exemplified  by  draft  EPA  Method  8315   and
       California   Air   Resources  Board  (CARB)   430,   and   considered
       approaches   for  reducing  systematic  error.     Using     spectral
       confirmation,   we   also   investigated   the  frequency   at   which
       interferences  or   false  positives  occur.

       The  common  method   relies  on   pre-column  defivatization  of  the
       aldehydes   with    2,4-dinitrophenylhydrazine   (DNPH)   followed   by
       reversed-phase HPLC analysis.   As others have  noted,  this  method
       of  determining   formaldehyde   at  trace  levels  presents  a  major
       problem:   contamination,  which  results  in elevated  blank  levels   and
       increased  detection  limits.    Causes of contamination  include solvents
       and  solid-phase  extraction  columns  as  well  as  the  exposure  of
       reagents and  samples to  ambient  air.  Introduction  of  contamination
       counteracts  the  advantages  of   concentrating  the  sample  through
       extended   preparation  procedures,   such   as   liquid/liquid  and  solid-
       phase  extraction.  Because  concentration  is   not   needed  to  meet
       detection  limits on  the  order  of  20-50  ppb,  which  are  satisfactory
       for most  regulatory purposes,  we have   investigated  the  application
       of  automated  pre-column  derivation   using   the   Hewlett-Packard
       1090  Series  II  HPLC  system.    The paper  presents details  of  this
       procedure.
                                        1-165

-------
The  application  of  the  diode  array  UV-visible  detector  to  this
analysis   provides  many  advantages  over  fixed-wavelength  analysis.
Shifts  in  retention  time during  the course  of  analyzing  batches  of
real-world samples  are  quite  common,  and  lead  to  misidentification
in  the  case  of  single-wavelength,  single-column   detection.    With
spectral   information  available  from  the  diode  array   detector,  false
positives  can  be  virtually  eliminated.    The paper  provides  examples
of   chemical   interferents   and   misidentifications   observed   in
environmental   analysis.

INTRODUCTION

Formaldehyde  analysis   of  environmental  samples  is becoming  more
common  in the  analytical  lab.   The detection  of  this  compund  plus
other  carbonyls  is important  due to the health  hazards  and  possible
role   they  play   in  environmental    reaction   pathways.     These
compounds are  formed  by  incomplete  combustion  and   atmospheric
photoxidation  of  hydrocarbons.    Formaldehyde  is  a  very  common
ingredient  in  cosmetics,  building  materials,   as   well   as  a  key
chemical   for  chemical  synthesis.    The   fact  that   formaldehyde  is
universally present  leads  to  positive  detection  by any   analytical
procedure.     The  blank   levels   that  are   observed   can   vary
dramatically   when  an  extensive  sample  preparation  procedure  is
used.    This   paper  will   discuss  the  application   of  two  slightly
different   approved  procedures,  EPA   8315  and   CARB   430,  plus
compare  these   methods  to  an  on-line  HPLC  sample preparation
procedure  that   controls  the  sample contamination  problem.

EPA   Method   8315  involves  the  analysis   of  aqueous  and   solid
samples  by  DNPH  derivatization  followed  by  HPLC  detection.  This
research   did   not   investigate  the  application  of  8315  to   solid
samples.    Aqueous  samples  are  mixed  with  a derivatizing  solution
of DNPH in  ethanol  plus  acetic  acid.   The  pH  for this  reaction  is
adjusted  to around  5  and derivatization is allowed  to  proceed  for  at
least   a  half  hour.     This  solution  is   then  extracted   by  either
liquid/liquid  or  solid  phase  extraction.  The  liquid/liquid   procedure
partitions  the  DNPH  hydrazone  product  into  methylene  chloride.
The  SPE  process  uses   a  CIS   column to   separate  the  derivatized
product.    In  both  preparations  the  final   extract  must  be exchanged
for one   that  is  suitable  for HPLC  analysis,  acetonitrile  (ACN)  or
                                  1-166

-------
methanol.    The  reader  is  referred  to the  actual  methods  for  more
details  on  this  preparation.

GARB 430  is  the  California  approved   air  sampling  method  for
airborne  formaldehyde.    It  involves the  use  of   DNPH   impinger
solution  that   is  composed  of  approximately  2N   hydrochloric  acid
(HCL).  This  acidic  solution  traps  and   derivatizes   immediatley  all
carbonyl species  that  bubble  through.   The sample   solution is  then
extracted  with  70/30  vol/vol  %  hexane/methylene  chloride.    The
extract  is  again  exchanged to a  suitable  HPLC  solvent.   Therefore,
the  final  extracts  that  are  obtained by  these  two   methods   are
identical  and  all HPLC  conditions  are  the same.

Both  approved  methods  and  many journal  publications  mention  the
high  blank  levels  obtained  by these  sample preparation   procedures.
CARB 430  outlines  a  very  detailed  purification  procedure  involving
multiple  recrystallizations   of   DNPH  followed  by  storage   in   a
nitrogen  purged  dessicator.    All  impinger  solution  must  be checked
for contamination prior  to  use with  48  hours  as  the  maximum  time
between  preparation  and use.   8315  also mentions that  blank  levels
are a  major  problem  but  the only  precaution  mentioned  is to  use
the highest quality reagents.    In  both  methods  it is  suggested  that
the blank level  is  subtracted  from  all  sample  data.    8315  mentions
this   blank   subtraction  procedure  only   in   the  context  of  the
establishing  of  the  method  detection  limit.   The  work  reported  here
has  centered   around  trying  to   clean up  this  blank   problem  and
therefore  not   do  blank  subtraction  when  reporting  data.

EXPERIMENTAL

Reagents  and  standards.
All  organic   solvents  were  of  HPLC  UV  spectral   grade.     Many
vendors  were  consulted  during  solvent  selection  and   no  guarantee
of low ( <50   ppb)  formaldehyde  levels   could  be  confirmed.    The
results  obtained  here  indicate  that   the  purity  of  the  DNPH  and
water is more critical  then that  of the organic solvents.   DNPH was
purchased  from  ChemService  (West  Chester, PA)  and  recrystallized
twice in pure ACN  according to  the  CARB 430 method.    Impinger
solution  was   prepared  from  90  ml   of  concentrated   HCL  (Baker
Analyzed)  to  which  .250  grams of  pure   DNPH  is added.    After  the
                                  11-167

-------
crystals  have  dissolved  organic  free water  is  added  to  produce 500
ml  of solution.   The  water used  throughout this  work   was  distilled
then  passed  through  a  Barnstead  Nanopure  II  system,  heated  to
near boiling and  purged  for a hour  with helium,  and  finally charcoal
filtered.    This impinger solution  was   used  directly   as  the  on-line
(autoprep)   derivatization  solution  in the Hewlett  Packard  HPLC.

Formaldehyde   for   calibration  and  quality   control   solutions  was
obtained  also  from  ChemService  as  a  70%  aqueous  solution.   Known
concentrations   (1000  ppm)   of  a  aqueous   stock  were   prepared
according  to  method  8315.    This procedure  uses   a  pH  titration
procedure   to  establish  the  concentration   of  formaldehyde   in  the
stock.   CARB  430  recommends the  production  of  pure    formaldehyde
hydrazone   crystals   for  calibration.     Our   experience   with  this
procedure    has   never    produced   quantitative   information   that
compares  favorably  with  8315.    Even  though  the  melting  point
observed  for  the product  appeared to  be   acceptable  these  crystals
always  produce   standards   that  gave  very  low  responses.    All
calibration   standards  were  prepared   exactly  like  as  a  analytical
sample  for  all  three methods.  The  autoprep  technique   involved  the
production  of  spiked  water  at   4  or  5  calibration  concentrations
which were placed  on  the instrument  and  derivatized  on-line.

Instrumentation.
The HPLC  instrumentation  consists   of  a   Hewlett   Packard   1090
series  II with  a  diode array detector.    Computer   control  is  provided
with a HP  ChemStation  with  a  Pascal   based  operating  system.    This
instrument   is  a   binary  gradient  low  pressure  mixing   system.    The
mobile  phase  consists   of  0.01   molar  phosphoric  acid  (channel A)
and  pure ACN (channel  B).  The  selection  of  a  weak   acidic  mobile
phase  provides two  benefits.   First  it  assures  that  the  DNPH  reaction
proceeds to completion  plus  it  stops  the  columns   from  becoming
clogged.    Without  a acidic  mobile  phase the  reverse phase columns
used  in  our  lab   have  stopped   functioning   after  as  few   as  30
samples.    With  the  introduction of  a   phosphoric  acid   mobile  phase
the current  column  has  completed  over  300 analytical   runs.   Due  to
the nature  of the  samples  we analyze  (high  organic   contamination),
inexpensive  C18   columns  are  purchased   from   Alltech   (Deerfield,
IL).    The   column used  for this  work   is a  Econosphere  C18  with 5
micron  particle  packing  (150mm  x   4.6mm).        The  gradient
                                   1-168

-------
conditions  are shown  in  Table  1     The  spectroscopy parameters  are
also   presented.

The  procedure  for on-line  derivatization  is   presented  in  Table  2.
The  first  step  (line 1)  prompts the  autosampler  to  draw  up  into  the
sample  loop   2.5  micro   liters   (ul)  of  DNPH  impinger  (vial   #1)
solution.   Next 10 ul  of  sample  is  drawn  up  followed  by  a  syringe
rinse  in  solvent  (ACN).   Another  plug of DNPH  is then  drawn  up  to
sandwich  the  sample   Line 5  is the  mixing  step  which  moves  the
sample  plus  reagent  back   and  forth   through  the   heated  reaction
furnace  which  is at  50C.    After  mixing  a  final  10 ul of solvent  is
drawn in  order  to   optimize  the  position  of  the   sample   in   the
reaction  furnace.    The  autosampler  then  waits  two  minutes   before
performing  the  analytical  injection.

RESULTS AND SUMMARY

Method   development   on   the    autoprep  derivatization   technique
involved  experimenting   with  different  reagent  combinations.  The
starting  point  used the  reagents   suggested  by  EPA  8315. Thus,   the
derivatizing  solution   was   a  combination  of  5M  acetic   acid  plus
saturated  DNPH/ethanol.    This combination  did  not  give  satisfactory
results.    The  blank  values  observed were very high (at least  100
ppb).    Also,  the  sensitivity   was   not   comparable  to  standards
prepared   by  Method  8315.   This lack  of  sensitivity  was  most likely
due  to  the use  of a   weak acid for  derivatization.     It  had   been
observed  previously  in our  lab   that   this reaction was not  complete
in a  few  minutes.   In  order for  the   autoprep  technique   to  be  time
effective   different  reagents  had  to be selected.   The  application  of  a
strong acid  such as HCL or sulfuric  was suggested  to us  by a  fellow
researcher.  Because  we  always   are   preparing  clean   HCL  impinger
solution for Method  CARB   430  this  was  the   easiest  reagent  to  use.
It  was  proposed that  the  injection  of  5  ul's  of  this  acid  solution
could be  tolerated  by  the HPLC.

Over  the   course   of  approximately  3   months   a    calibration
comparison  between  the   autoprep  procedure  and   the   approved
sample  preparation  procedures  was  performed.    The   results  of  this
study is  graphically  displayed  in  Figure 1.    Each  calibration curve
has  a least  squares  fit  equation  and  line  associated  with it.    The
                                  1-169

-------
equation  in the top  of  the  each  plot is associated  with  the  top  line
on  the  right  hand border.    The  linearity  (correlation  coefficient  RA2)
of  these  calibration  curves  are  all  >  0.990  with  only   one  of  the
autoprep coefficients  being  <  0.995.

There  is  no  criterion for  linearity  in EPA  Method 8315.   GARB 430
mentions  that  linearity  through  the  origin may  be  assumed  if  an
RA2 of 0.999 is  obtained.   As  this data indicates an RA2 of  0.999  does
not  guarantee that  the  calibration  curve  passes  through   the  origin.
In  fact,  in  our  experience  these calibration  curves  rarely  include the
origin. If one tries to  force the  orgin  as  a data  point  then  the RA2
goes  down.    The  fact  that  these  calibration  curves  behave  in  this
manner is  not  surprising.   One  must remember  that  contamination
by   formaldehyde   can   occur  during   any  stage  of the  standard
preparation.    It  is  this   random  contamination  that leads  to  the
spread in  these  calibration  plots.   By  comparing  the  autoprep   data
to  that  obtained by  the two  approved  methods  it  is seen   that  all
three  techniques   are  fairly  comparable.

One  point   that  is  not  obvious  from  the  calibration   data  is  the
comparison  of  blank  levels  obtained  by  each  technique.    In  the
cases of  8315 and  CARB  430, the  blank  response  is  frequently as
high  as  the  lowest  calibration  standard   (100  ppb).    The  blanks
obtained   on  clean  water   by  autoprep   show  substantially   less
response.     Assigning  an  absolute   quantifiable  number   to   the
autoprep   blank  is  difficult,  because  preparation  of   calibration
standards  at  these   low  (  <100  ppb)  levels  is  nearly  impossible.
Basically, the  peak  areas obtained  for  the    autoprep  blanks  are  4 to
5   times   less   than   those   obtained  through   normal    sample
preparations.   This  lack  of  low  level  quantification  information  also
makes it  difficult  to  establish   a  method  detection  limit  (MDL)  for
the  autoprep  technique.   The  approved  methods however both  use
blank  subtraction  to  establish  their  MDL's.    If  one  looks  at  the
absolute   instrumental   response   from   the  HP1090   it   can  be
estimated  that  a  detection   limit  of  around  5-10   ppb  should  be
obtainable.

Figure  2   depicts  the  chromatgraphic  response   of a  high  level
formaldehyde  standard.     The   peak  shape  for  the  formaldehyde
derivative   is  obviously   not   symmetrical.    However,  with   the
                                   1-170

-------
spectral  power  of the  diode array  it  can  be  demonstrated  that  this
peak is  pure.   On a  brand new column  the  standards  do  initially
appear  symmetrical.  But after   a  few  real  world  samples  the  peak
shapes   deteriorate. This  deterioration  is not  due to  column  overload
because  even  low  level standards  are asymmetrical.    In  order  to
check  the  performance  of the used columns a  pesticide  mixture  was
run  under  the  same   mobile  phase.    The  gradient   program  was
slightly different.   As  can  be seen in  Figure  2B  the  chromatography
of this mix displays better   peak  shape. From the  point  of  view  of  a
production  laboratory   a  new   column  for    formaldehyde  analysis
cannot   be  justified.

The  analytical  results   for  a  number  of  different  types   of   water
samples  are  presented  in Table  3.    In every  case  the   normal   8315
preparation  of  these   samples   produced  higher   results   than  the
autoprep technique.   Not only  did the normal  preparation  produce
contaminated  sample   extracts  this   preparation   requires  a   lot  of
labor.    The  normal   preparation  involves  liquid/liquid   separatory
extractions   in  triplicate   followed   by  concentration   and   solvent
exchange.    Reagent   consumption  for  the   normal  preparation  is
hundreds   of  milliliters  compared  to  ul  by   autoprep.   The  only
manipulation   done   for  the   autoprep  procedure  is   to  load   a
autosampler vial  with   sample and  place .it  on the HPLC.

Another  advantage of  the   autoprep technique is  due to  the fact that
the  HPLC  injects  the  sample  directly  rather  then  a   concentrated
extract.    The  solvent  extraction  processes  of 8315  and  CARB 430
extract  all the  organic  components  in  the  sample  and  these then  can
cause  many types  of  chromatographic  problems.    Not  only  does   it
appear  that the  extraction  process   contaminates  the  samples  but
chromatographic   interferences    can   be    promoted    by    the
concentration  procedure.    The  HPLC  chromatograms   generated   by
autoprep   are   less  likely  to   produce   complicated   peak  shapes,
retention   time  shifts,     or  UV/Vis  spectral  complications.     An
example  of  the type  of problems  observed for  a  sample that  has
been prepared  according  to  Method  8315 is  shown  in  Figure 3.  The
top  portion  of this  figure  shows the  chromatogram and the   bottom
portion contains  selected UV/Vis spectral scans.   It is   fairly  obvious
from   the   chromatogram  that    the    formaldehyde   derivative
(retention  time  =  7.0)  peak  is  not  a single  component.   If  one  is
                                  11-171

-------
performing  this  test with  a  single  beam  UV/Visible  instrument  no
further  insight  into  this coelution is  possible.

Because  the   diode  array  system  gathers  and  stores   complete
spectral  scans   during   each  peak  the  presence  of  an   interfering
species  can  be  confirmed.    Three  spectral  responses  are  shown  for
1- the first peak  max at 6.7  min., 2 - the peak  max at  6.9 min., and
3- the tailing edge  at 7.3 min..   Spectra  1's response  is  indicative of
pure  formaldehyde  derivative   which  quickly   becomes   convolved
with  a  coeluting species spectra  shown  in  spectra  2.    Even  though
the interfering  species   peak maximum  is  at  a  different  wavelength
(250   nm)   than  that  used   for  formaldehyde   (360   nm)   the
absorbance  is  strong  and probably  adds  to the total  peak  area.   The
three  spectra shown  in the  figure  are all  scaled  to  the  same  axis.
Nothing  about  absolute  intensity  is  displayed.   It was  observed  that
the  absorbance  from   the   interfering   species    was   approximately
three  times  as  strong  as that  seen  for  formaldehyde.    The presence
of  this  interfering  species  remains  throughout  the  remainder  of  the
chromatographic  peak.   Spectra  3  still  displays  the  low  wavelength
absorption   due  to  its  presence.    Because  of  the   diode   array
information  these  analytical  results could  be  reported   to  the  client
as  elevated  due  to this interferent.
                                   1-172

-------
  2000
It
:
  1000
                  8315 Calibration
                y ป  0.28190 X  +113.87
                   R*2 = 0.996
                     y =  0.23850 X  + 33.865
                         RA2 = 0.999
                                                  2000
                                                  1000-
I
0.
                                                                CARB 430 Calibration
               y = 0.31547 X + 30.224
                   fi*2 = 0.999
                        y  =  0.2483  X +67.438
                            R*2 = 0.997
          1000    2000   3000   4000   SOOO   SOOO

               Concentration (ppb)
            1000   2000    3000   4000   SOOO   6000

                  Concentration (ppb)
  1MO
  1000
t
   500
           Impinger Autoprep Calibration
            y  = 0.25544  X - 21.179
                 R*2 - 0.991
                     y = 0.20985 X  +  88.569
                  •      Hป2 = 0.999
                                                 2000
                                                 1000-
           Composite Calibration Data
                                                 ป  IMP prep
                                                 •  art* 430
                                                 •  e*to 430
                                                 •  8315
                                                 a  S31S
     0    1000   2000   3000   4000   5000   SOOO        0     1000   2000   3000   4000   SOOO    8000
                Concentration (ppb)                               Concentration  (ppb)


                  Figure  1  - Comparison  of Calibration Data
                                                 11-173

-------
ซBB-
3BB-
2BB-
tee-
              DNPH
FORMflLBEHTOE
DERIVATIVE
                          A
                     B     B     IB
                  Time  (min.1

20-

13-
E IB-


SPIKE
B





SIHRZINC




Jb

5URFLBN




a ซ e e te 12
T I me ( m 1 n . )
 Figure 2  - HPLC  Chromatograms
              A) 5000 ppb formaldehyde std.
              B) Pesticide test mix
                     11-174

-------
IT
E
      2-
       6.8
6. 5
7.0      7.5
Ti me  (m i n.  )
8. 0
                                                    8. 5
TJ
I)

0
0
M
I
E
                      1 - Front  edge  spectra
    / 3  • Tall  edge  spectra
        2 - Peak max  spectra
              300            -400
             	Have length  (ntn)
                            500
                                 600
      Figure  3  -  Example of  a  Chromatographlc/
                         Spectral Interferent.
                              1-175

-------
LIQUID   CHROMATOGRAPH
                                                         initial parameters
           Flow :    0.500 ml/nin
     Solvent  A :     40.0 %
              B :     60.0 %
    Max Pressure :

      Stop Time :
      Post Time :

Injection Volume :
                      400 bar

                    14.00 min
                     0.00 min

                     10.0 ul
Min Pressure :
off
LIQUID   CHROMATOGRAPH
 Tin
(Bin)
1.00
5.00
7.00
8.00
10.00
14.00

Solvent
Solvent
Solvent
Solvent
Solvent
Solvent

A:
A:
A:
A:
A:
A:

40
30
0
0
40
40

.0
.0
.0
.0
.0
.0

B:
B:
B:
B:
B:
B:

60.0
70.0
100.0
100.0
60.0
60.0
DIODE-ARRAY   DETECTOR

      SIGNALS      ABC
Sample    (nm)
    Wavelength :    360    340     380
     Bandwidth      80     80      80
Reference (nm)
    Wavelength     560    560     560
     Bandwidth      40     40      40
                                                         signals & spectra
Store Spectrum
Threshold
PeaKwidth
Stop Time
Post Time
Prerun Balance.
peak controlled
0.1 mAU
0.150 min
14.00 Bin
0.00 min
Yes
                                   about     896 Records acquired during Run


                                              Sampling Interval     960 OB
                                             Spectrum Range from :    220 nm
                                                            to :    600 nn
                                                          step :      4 MI
           Table 1  -  HPLC  Chromatography  and
                          Diode Array Parameters
                                  11-176

-------
INJECTOR   PROGRRM

 Slowdown  Draw & Eject :    2
                  Mix :    2
Hold after Draw Be Eject :    0 seconds
Linett
1
2
3
4
5
6
7
8
Function
ggHB 2-s
Draw
Draw
Draw
Mix
Drav
Wait
1D.D
D.D
2.5
1D.O
1D.D
2. DO
Inject

ul
ul
u)
u]
u]
u]

from :
from :
from :
from :

Vialtt :
Sample
Vial* :
Vial* :
cycles : ID
from : Vial* :

1
D
1
D
minutes




        25.D   ul accumulated in Syringe with Line*   5
Table  2 - Autoprep Injector  Program
                        1-177

-------
TABLE 3 - COMPARISON OF ANALYTICAL RESULTS
                (ALL RESULTS ARE PPB)
 SAMPLE DESCRIPTION
8315 RESULT
AUTOPREP RESULT
 WASTE WATER

 SEWER COMPOSITE #1

 SEWER COMPOSITE #2

 GROUND WATER #1
             #2
             #3
             #4
             #5
             #6
             #7
             #8

 AQUEOUS SAMPLE
    104

    226

    230

    110
     45
     62
     87
     94
     95
     82
    113

    389
    53

    <25

    <25

    <25
     11
     it
     it
    <25

-------
AN  INTERLABORATORY  COMPARISON   STUDY   OF  SUPERCRITICAL  FLUID
EXTRACTION  FOR  ENVIRONMENTAL  SAMPLES;  Tammy  L.  Jones.  U.S.
Environmental Protection Agency,  Environmental Monitoring Systems
Laboratory -  Las  Vegas,  Las Vegas,  NV   89193, Tom  C.H.  Chiang,
Lockheed Engineering and Sciences Company,  Las Vegas, NV  89119
     The U.S. Environmental  Protection  Agency recently conducted

a multilaboratory evaluation of  a  supercritical fluid extraction

(SFE) protocol.   SFE  is a relatively new technique  which can be

used to  extract compounds of  environmental  interest  from solid

matrices (soils, sediments, fly ash, etc.)  by using supercritical

C02.  Ten laboratories participated in this study that was designed

to  evaluate  the  feasibility  and  applicability  of a  protocol

developed  for  the  extraction  of  environmentally  significant

analytes from environmental matrices.



     The efficiency of analyte (polynuclear aromatic hydrocarbons

and phenols) recoveries, using  SFE,  from three solid matrices  (two

standard reference materials  and  one spiked sand) was  studied.  The

analyses of the resulting extracts from all the laboratories were

performed by a single  laboratory using gas chromatography/mass

spectrometry  (GC/MS).     The  data  were evaluated  in  terms  of

precision,  accuracy, and the  intra-laboratory  and inter-laboratory

variations  within  this  technique.    In  general  the  percent

recoveries  of  the analytes from the  various laboratories ranged

from poor (< 40%)  to very good  (> 90%).  There was  a trend noticed

that those laboratories who performed satisfactorily  on one sample

matrix also continued to do so on the other two.  NOTICE:  Although

the research described  in this article  has  been supported by the
                             1-179

-------
U. S. Environmental Protection Agency,  it  has not been subjected



to Agency  review and  therefore does not necessarily  reflect the



views of the Agency.  This document is intended for internal Agency



use only.  Mention of trade names or commercial products does not



constitute endorsement nor recommendation for use.
                             1-180

-------
67
 AN ANALYTICAL MANUAL FOR PETROLEUM PRODUCTS IN THE ENVIRONMENT


M.W. Miller,  M.M.  Ferko,  F. Genicola, H.T. Hoffman and A.J. Kopera
New Jersey Department of Environmental Protection
Office of Quality Assurance
CN 027,  Trenton,  New Jersey  08625

Abstract

    The New Jersey Department of Environmental Protection (NJDEP)
Analytical  Chemistry  Manual  for  Petroleum   Products  in  the
Environment was drafted to help project managers select appropriate
analytical methods.  Eight NJDEP  programs administer regulations
concerning petroleum products.  The analytical  methodologies for
these  programs have not  been codified  within  federal  or  state
regulations, and  several method variants exist.

    Preparatory  to drafting  the Manual,  we conducted an extensive
review of  the regulatory  programs.   The methods and standards
reviewed include  those of federal  and state agency departments, as
well as those of  the American Public Health Association, American
Society  for  Testing  and   Materials,   and  American  Petroleum
Institute.   Selected methods  and procedures  for free product,
aqueous matrices  and nonaqueous matrices were edited to establish
a  Department Manual.   Methods for  volatile petroleum products
(e.g., gasoline,   jet fuel,  kerosene,  solvents)  and  semivolatile
petroleum products  (e.g. diesel,  fuel oils #2-#6) are presented.

    The analytical laboratory methods contained in the Manual will
become part of the revised NJDEP Regulations Governing Laboratory
Certification and Standards of Performance, N.J.A.C.  7:18.

    The paper discusses a survey  method,  two quantitative methods
and one fingerprint  method.   These  methods are representative of
the fifteen methods in the first edition of the Department Manual.
A gas chromatography-photoionization-flame ionization detector (GC-
PID-FID)   survey   method  is  presented   for  volatile  petroleum
products.   A  quantitative  GC-PID-FID   method  is discussed for
volatile petroleum products.  Gas  chromatography-mass spectroscopy
is discussed for  the identification and  quantification  of specific
semivolatile  compounds   in  petroleum  contaminated   soil.    The
identification of specific petroleum products in contaminated water
and  soil   or  free  product   is  accomplished   by  GC-PID-FID
fingerprinting.

    Each  method contains calibration  procedures for petroleum
products,  and quality  control requirements.    The   manual  also
contains a users  guide for environmental professionals.
                                  1-181

-------
QQ             EVALUATION OF LIQUID/SOLID EXTRACTION  FOR THE ANALYSIS  OF
            ORGANOCHLORINE PESTICIDES AND PCB'S  IN TYPICAL  GROUND  AND SURFACE
                                      WATER MATRICES

         Anne D. O'Donnell. Group Leader, Denise R.  Anderson, Group Leader, John
         T.   Bychowskl,    Manager,   Semi-Volatile   Organics  Department,   WMI
         Environmental  Monitoring  Laboratories, Inc.,  2100 Cleanwater  Drive,
         Geneva, Illinois,  60134; Craig G. Markell, Research  Specialist, I&E Sector
         New Products Department, Donald F.  Hagen, Corporate  Scientist,  Corporate
         Research Analytical Laboratory,  3M  Corporate Research Labs, 3M Center,
         Bldg. 201-1S-26, St. Paul,  Minnesota, 55144

         ABSTRACT

         Method 608/8080 is used for the analysis of organochlorine pesticides and
         PCB's  in  water and wastewater.   The  main features of  the method  are
         liquid/liquid extraction (LLE)  with methylene chloride,  removal of  the
         methylene chloride  to  concentrate the analytes,  a  solvent exchange into
         hexane, and gas chromatographic analysis with electron capture detection.
         Because several of the  500 series drinking water methods are being updated
         with   the   inclusion   of  liquid/solid  extraction   (LSE),  a  similar
         modification was evaluated  for Method 608.  The  LLE steps were replaced
         with a solid-phase  disk  (47mm C18)  extraction, elution  of analytes with
         ethyl acetate, and direct GC analysis of this eluate.

         The LSE evaluation study was performed with reagent water and composites
         of typical  ground  and  surface waters,  including groundwater  composites
         with very high particulate  content.  The single organochlorine pesticides
         and  the  multicomponent mixtures were  all  spiked  at two  concentration
         levels, a "validation"  level and an "MDL" level.   Elution efficiency was
         determined for all sample types.

         The recovery efficiencies,  %RSD's,  and  method detection  limits obtained
         demonstrate that LSE is at least equivalent  to  LLE for the Method 608/8080
         analytes,  and, in most  cases, an improvement. The  LSE disk modification
         was successfully applied to all water matrices typically  encountered in
         our laboratory.  Disk LSE provides a clear advantage  in terms of time and
         cost  per analysis  and solvent  use and disposal.

         INTRODUCTION

         The use of  LSE instead of  LLE  for  the isolation  and concentration of
         organic components  in  environmental  water  samples is  becoming  more
         extensive  because  of the time and cost benefits it provides.  LSE is less
         labor-intensive,  uses  substantially less  glassware, and  significantly
         reduces the  volume of hazardous  and  costly  solvents required.   The solid
         phase used most frequently is octadecane  (C18) chemically bonded to porous
         silica   particles.    It  is  commonly   packed  into  disposable  plastic
         cartridges,  producing LC mini-columns.   SPE using these  cartridges is an
         alternative  sample  preparation  procedure  cited  in  the  Drinking  Water
         Methods 506, 525.1,  and 550.1.   Method  525.1 addresses many  of the same
                                           1-182

-------
analytes covered by Method 608/8080 (1).

Recently 3M introduced membrane  disks,  Empore™,  as an  LSE medium (2).
Instead of being packed in a  cartridge,  the  CIS-bonded silica particles
are enmeshed in  PTFE  fibrils.  The  large  diameter, thin  disks  (47mm x
0.5mm) provide  a large cross-sectional area with low back pressure.  The
primary advantage  of  the  disk technology  is  the  speed  of extraction
possible with equivalent extraction  efficiency.   Efficiency is achieved
with a  smaller  particle  size, a uniform,  high density  packing,  and an
effective low linear velocity  through the disk at high sample flow rates.
The 47mm diameter disks fit standard glass filtration assemblies, allowing
the extraction  to be carried  out in  an all glass/PTFE  environment.  One
of  the  principal  disadvantages of the LSE  cartridges  is  the  amount of
trace  contaminants contributed  by   plastic  housings.     Disk  LSE  is
designated as an approved  technique for Methods 506/8061, 513, 525.1, and
550.1.

The organochlorine  pesticides and PCB's  determined by  Method 608/8080
present good candidates  for LSE.  They are extracted at a neutral pH, are
insoluble   in   water   (large  capacity  factor,   k',   for  CIS/water
reversed-phase  conditions, soluble in organic  solvents (easily eluted),
and are relatively non-volatile.  In addition,  the  608/8080 Method is very
susceptible to  interferences from trace level  contamination because  of the
high  sensitivity  of the  electron  capture detector.   A  procedure that
significantly reduces  both the amount  of glassware  that  must  be kept
scrupulously clean and the volume  of solvent  concentrated for the final
extract will also significantly reduce contamination interferences.

The evaluation  of Empore™ disks  for LSE of the Method 608/8080 analytes
was performed  using reagent  water,  composites  of  "average" ground and
surface waters, and composites of  groundwater samples with a very high
total suspended solids (TSS)  content. The technique was  to  be challenged
with  all  the types of water  samples normally encountered.   The  single
organochlorine  pesticides and the multicomponent mixtures were all  spiked
at  two  concentration  levels,  a  "validation"  level and  an "MDL"  level.
Elution efficiency was determined  for all analytes in all sample types.

EXPERIMENTAL

Materials.  Empore™ Extraction Disks, C18, 47mm (Varian  Sample Preparation
Products, Harbor City, CA, Cat. #1214-5004).   Whatman Multigrade GMF 150
graded density glass microfibre  filters,  37mm (Cat. #1841-047, Clifton,
NJ).

Apparatus.  Glass filtration apparatus,  47mm, 300mL funnel,  lOOOmL flask
Nuclepore  Cat.  #410502   (Pleasanton, CA).    Millipore  (Bedford,  MA)
vacuum/pressure pump  (Cat.  #XX55 000 00).   A tee with a pinch clamp is
placed in the line between the filtration assembly and the  pump to allow
fine control of the vacuum for the preconditioning and eluting steps.
                                   1-183

-------
Procedure.  The filtration unit  is  assembled  with  the  Erapore™ disk.  The
funnel  and  disk  are  washed  with  lOmL  of  ethyl  acetate  (the  elution
solvent), then with lOmL of methanol (preconditioning  wetting agent), and
finally with  two  lOmL  rinses of reagent water. The  1L water sample, to
which surrogate standard and  0.5% methanol  wetting agent have been added,
is then passed through the disk at full vacuum (25" Hg, 85 kPa).  A  thin
layer of liquid is maintained on the disk  from the methanol conditioning
step until the entire sample has been extracted. The disk is subsequently
eluted with two 5mL portions of ethyl acetate; the first portion is  also
used  to rinse the  sample bottle.   During the elution step,  the ethyl
acetate  is  allowed to equilibrate on  the  disk for a few  minutes.   The
eluate  is collected in a lOmL  Kuderna-Danish (KD)  concentrator tube.
Internal  standard  is added to the extract and it is  made  up to volume.
Na2S04 is added to dry the sample.

Liquid-liquid  extraction analyses   were   performed  using  continuous
liquid-liquid  extractors  for an  18-hour period.

GC Analysis.  Samples  were analyzed on a Hewlett-Packard  Model 5890A GC
with  electron capture  detection,  using  a  Hewlett-Packard  Model 7673A
autosampler and Fisons/VG Multichrom Data System, Version 1.8.  The column
was a J&W Scientific  (Folsom, CA)  DB608,  30m x 0.53mm i.d., 0.83pm  film
thickness (Part No. 125-1730).  Helium was  the carrier gas  at y=A5cm/sec,
with  argon-5ฃmethane  make-up at 65mL/min.   The  injection port  was at
200ฐC,  and   the  detector  at  300ฐC.    The  temperature  program   was:
isothermal at  140ฐC for 0.5  min, 140ฐ-275ฐ @ 6ฐC/min, hold 15 min.  The
injection was  2 uL  splitless.

RESULTS AND DISCUSSION

Table I summarizes the  results obtained for Empore™ LSE extraction of the
608/8080  analytes  from  reagent water  at  a  "validation"  concentration
levels.  For all experiments,  the amount of analyte spiked into  the sample
water was also spiked  into lOmL K-D concentrator  tubes containing ethyl
acetate,  the  "spike  check"  sample.      The  sample  extracts and  the
triplicate spike check  samples were treated identically for  analysis.  The
mean  of the spike check samples  was the basis for  the  recovery efficiency
calculation.

To test the completeness of elution  with  the two 5mL volumes of ethyl
acetate, a second  set of 5mL washes was passed through the disk, after the
first elution, and collected in  a second  lOmL concentrator  tube.  Elution
efficiency was calculated as the fraction of  analyte concentration in the
first eluant compared  to the  total  analyte  concentration in both eluants.
Elution efficiency  tests  were run for representative Aroclors,  not the
entire  set.

Table I data show excellent recovery and elution efficiencies.  The  mean
JKRecovery for all analytes was 91.5%,  and  the mean elution  efficiency was
0.991.  The lower  recovery value for  Aldrin is a  function of its higher
                                  11-184

-------
volatility.    The same  effect is  seen in  the lover  recovery  for  the
surrogate standard,  which  is more volatile than the rest of the analytes.
The precision of the method is also excellent.  The mean %RSD for all the
analytes vas 3.0%.

Method validation data  for the single organochlorine pesticides using LLE
are presented in Table II for comparison. The  mean accuracy of the method
is 88.2%; the mean  precision  is 2.7 %RSD.   Aldrin is seen  to  have the
lowest recovery  by  LLE  also.

Results for  the  disk extraction of the 608/8080 analytes from an average
groundwater  composite at validation concentration levels are compiled in
Table III.  Representative Aroclors were included in this study, not the
entire list.  The data  indicate that very good accuracy and precision can
be expected  for  disk LSE applied to actual samples.  For these groundwater
composites,  the  mean ^Recovery was 92.6, the mean precision was 4.4 %RSD,
and the mean elution efficiency was 0.993.

Summary Table IV contains  the  results for the disk LSE of all the analytes
from reagent water at MDL concentration levels. The MDL's calculated from
the Empore™ data and the current laboratory MDL's for LLE are also listed.
Several of the Empore™ MDL's  would  have to  be rerun at a lower level to
meet the  requirements  of  40  CFR, Part  136,  Appendix B.   Accuracy (%R)
values at these  levels  are good;  the mean ^Recovery is 91.9%.  Except for
a contaminant interfering with  endrin aldehyde,  the precision  data are
also good, with  the mean at 5.3 %RSD.   Comparison  of  the MDL values for
LSE and LLE  shows lower  results  for LSE in all  cases  except the endrin
aldehyde.  The  lower level of contaminants  accounts  for  the lower MDL's
by LSE, a factor more evident in  the  MDL  differences  for multicomponent
analytes (Chlordane, Toxaphene, and the PCB's).

Table V shows analysis results for disk LSE of the analytes from actual
groundwater   composites  at  the  low  MDL  concentration  levels.   MDL's
calculated from  these data compare  favorably with the MDL's determined in
reagent water.   Mean accuracy was 81.3% Recovery,  and  precision was 7.1
%RSD.

A liter of reagent water could be processed through  the  disk  in an average
time  of  7-8 min.   The  processing time  for  the  "average"  groundwater
composites ranged from  8-18 min.  In production, several samples could be
extracted simultaneously.   Many of these  experiments  were  run using  a
manifold with four extraction stations.

High Particulate Samples

Composites of samples with very high paniculate content were prepared to
study the procedure  modifications that  might be necessary to handle these
sample types. The composites  had TSS contents in  the range of 1.8-18 g/L.
In contrast, the "average" groundwater composites had TSS contents of 1-
5  mg/L.   The  high  particulate  samples took  several  hours  to process
through  the disk,   even after allowing the  particulates to  settle and
                                  11-185

-------
decanting most of the sample volume.

The  problem  of  excessive  filter  time  was managed  with  a pre-filter
positioned on  top  of the Empore™ disk.   Five  different pre-filters of
varying pore  size were evaluated.  The best results were achieved with the
Whatman graded density  filter.   With this prefilter,  and decanting most
of the sample volume  before  transferring the bulk of the particulates, the
high particulate composites  could be extracted in approximately 20-40 min.

The  more critical  problem  presented  by  high   TSS  content  samples  is
effective recovery of analytes  sorbed  on the particulates.   It requires
efficient elution  of the filter cake  of  particulates  that  results from
sample  filtered through the  Empore  disk and  prefilter.    Consistent
recoveries were obtained  by adding ImL  of  methanol  to  the  disk  and
collected particulates with the first 5ml of ethyl acetate eluant, mixing
the  particulates so  that  they were well dispersed  in the  eluant mix, and
than allowing  some  time  for  equilibration  ("3  min).   A  larger  K-D
concentrator tube is used to allow for the larger water/methanol layer (1-
4mL)  in  the  total collected eluant.

Table VI shows Empore™ LSE results  on two of  the very high TSS content
composite groundwaters.  Data  for LLE,  using  continuous  liquid-liquid
extractors,  were  also  obtained  for  comparison.    Recovery data  are
consistently good for the LSE analyses.   Precision data obtained for LSE
and  LLE  on these high particulate samples  are very comparable.

The  slightly  higher  total average recovery values for  LSE  compared with
LLE  are  the  result  of  poorer recoveries  for selected analytes  by LLE:
Aldrin,  Heptachlor,  Methoxychlor,  and  the 4,4'-DDT,  -DDE,  and  -DDD.
Various  mechanisms may be at work contributing to the loss/degradation of
these analytes —  light  and  temperature  conditions  during the 18-hour
extraction, or particulate  surface reaction effects.

Results  on elution efficiency tests for four different high  particulate
groundwater  composites  are  compiled in Table VII.   These data indicate
that  the procedure  used adequately eluted the analytes  from  the collected
particulates, prefilter, and Empore disk.  More  exhaustive elution  is not
required.

CONCLUSIONS

The  experimental  data  clearly  validate  the  substitution  of LSE using
Empore™ CIS disks for LLE in the analysis of the  organochlorine pesticides
and PCB's tested.  This  study went  beyond the required  validation and MDL
determination in reagent water;  the method was validated  in the types of
water  sample   matrices  typically  encountered  in   an   environmental
laboratory.   LSE is  not only  equivalent to LLE, it is  preferred because
of its time and cost  benefits, and especially because of its  environmental
benefits.    It  represents  a  substantial  reduction   in  the  volume  of
hazardous solvents  required for sample preparation.
                                   1-186

-------
ACKNOWLEDGEMENT
The authors acknowledge, vith appreciation,  the  technical assistance of
Laura K.  Bartoszek in conducting these extraction studies.

REFERENCES

(1)  Determination of  Organic Compounds in Drinking Water by Liquid-Solid
     Extraction and Capillary Column Gas Chromatography/Mass Spectrometry,
     J. W.  Eichelberger, T. D. Behymer, W. L. Budde, Revision 2.1  (1988).

(2)  Hagen,  D.  F.;  Markell, C. G.;  Schmitt, G. A.; Blevins, D. D.  Analyt.
     Chim.  Acta 1990,  236.  157-164.(1) Method 525.
                                  11-187

-------
                                 TABLE  I

          EMPORE™ EXTRACTION OF ORGANOCHLORINE PESTICIDES AND
      PCB's FROM  REAGENT  WATER  AT  VALIDATION CONCENTRATION LEVELS
                            SPIKE
SAMPLE ANALYSIS
      Analyte
Aldrin
a-BHC
b-BHC
d-BHC
g-BHC (Lindane)
Chlordane
4,4'-DDD
4,4'-DDE
4,4'-DDT
Dieldrin
Endosulfan I
Endosulfan II
Endosulfan sulfate
Endrin
Endrin aldehyde
Endrin ketone
Heptachlor
Heptachlor epoxide
Methoxychlor
Toxaphene
PCB-1016
PCB-1221
PCB-1232
PCB-1242
PCB-1248
PCB-1254
PCB-1260

Ug/L
1.27
1.30
1.33
1.24
1.33
12.44
1.33
1.19
1.36
1.41
1.27
1.22
1.13
1.36
1.38
1.34
1.60
1.33
4.21
6.31
5.86
11.83
7.95
7.13
5.75
2.51
2.08
Mean8
%R
77
93
93
94
93
94
93
90
92
92
93
94
93
95
93
93
87
93
91
90
96
94
88
87
89
88
96

%RSD
3.9
3.3
2.9
3.3
3.0
4.0
2.7
3.5
2.8
2.7
3.0
2.5
2.3
3.4
3.6
2.3
2.4
2.8
3.1
6.0
3.9
1.8
1.5
2.0
3.1
3.5
2.6
Elution
Efficiency1*
0.990
0.995
0.992
0.996
0.995
0.991
0.994
0.993
0.992
0.993
0.991
0.989
0.992
0.991
0.987
0.992
0.992
0.992
0.992
0.995
0.988
N.A.
N.A.
N.A.
0.990
0.986
0.996
an=8, R=recovery (accuracy)
bA/(A+B) A=analyte concentration in first lOmL eluant
         B=analyte concentration in second lOmL eluant
                                 1-188

-------
                                TABLE II

         LIQUID-LIQUID EXTRACTION  OF  ORGANOCHLORINE PESTICIDES
         FROM REAGENT WATER AT VALIDATION  CONCENTRATION LEVELS
      Analyte
                                    SPIKE
UE/L
SAMPLE ANALYSIS
Mean3
 %R        %RSD
Aldrin
a-BHC
b-BHC
d-BHC
g-BHC (Lindane)
4,4'-ODD
4,4'-DDE
4,4'-DDT
Dieldrin
Endosulfan I
Endosulfan II
Endosulfan sulfate
Endrin
Endrin aldehyde
Endrin ketone
Heptachlor
Heptachlor epoxide
Methoxychlor

an=4, R=recovery (accuracy)
  241
  236
  251
  093
  246
  220
  214
  177
  248
  264
  240
  225
  196
  268
  308
  Oil
  263
4.000
 65
 84
 91
 93
 85
 88
 87
 91
 88
 86
 88
 90
 95
 80
 85
113
 86
 90
3.4
3.8
1.5
3.0
3.6
1.7
1.8
0.9
2.6
2.3
1.9
2.0
2.6
3.7
2.1
6.8
2.8
2.1
                                 11-189

-------
                               TABLE  III

          EMPORE™ EXTRACTION OF ORGANOCHLORINE PESTICIDES AND
                PCB's FROM COMPOSITE AVERAGE GROUNDWATER
                   AT VALIDATION CONCENTRATION LEVELS
                            SPIKE
SAMPLE ANALYSIS
      Analyte
Aldrin
a-BHC
b-BHC
d-BHC
g-BHC (Lindane)
Chlordane
4,4'-ODD
4,4'-DDE
4,4'-DDT
Dieldrin
Endosulfan I
Endosulfan II
Endosulfan sulfate
Endrin
Endrin aldehyde
Endrin ketone
Heptachlor
Heptachlor epoxide
Methoxychlor
Toxaphene
PCB-1016
PCB-1260

Ug/L
0.83
0.81
0.87
0.52
0.84
12.36
0.97
0.85
0.99
0.96
0.85
0.87
0.82
1.00
0.94
0.98
0.94
0.88
3.43
5.51
6.23
2.18
Mean*
_*IL
84
96
96
96
96
82
94
90
95
96
95
94
96
100
90
96
86
96
95
93
86
88

%RSD
6.2
4.7
4.4
4.6
4.6
6.0
4.5
3.9
4.3
4.6
4.4
4.1
4.2
4.6
3.6
4.3
5.4
4.5
4.4
5.0
2.4
1.7
Elution
Efficiency
0.991
0.997
0.994
0.997
0.997
0.978
0.995
0.993
0.993
0.994
0.994
0.993
0.994
0.993
0.988
0.995
0.993
0.994
0.993
0.985
0.996
0.995
an=8
                                11-190

-------
                                TABLE IV

          EMPORE™ EXTRACTION OF ORGANOCHLORINE PESTICIDES AND
          PCB'S FROM REAGENT WATER AT MDL CONCENTRATION LEVELS
                       SPIKE      SAMPLE ANALYSIS         MDLb. Ug/L
                                  Mean3
	Analvte          ug/L        %R         %RSD         LSE     LLE

Aldrin                 0.019       89          7.7        0.004   0.011
a-BHC                  0.010      100          5.1        0.001   0.009
b-BHC                  0.020       95          4.1        0.002   0.013
d-BHC                  0.010       95          6.3        0.002   0.006
g-BHC (Lindane)        0.012       96          4.1        0.002   0.004
Chlordane              0.313       96          3.8        0.035   0.081
4,4'-DDD               0.017       93          4.9        0.003   0.006
4,4'-DDE               0.015       93          3.9        0.002   0.007
4,4'-DDT               0.019       84          6.0        0.004   0.006
Dieldrin               0.019      104         11.0        0.006   0.007
Endosulfan I           0.019       93          4.1        0.002   0.005
Endosulfan II          0.021      102          5.1        0.004   0.006
Endosulfan sulfate     0.020       82          4.0        0.006   0.020
Endrin                 0.021       95          3.5        0.003   0.006
Endrin aldehyde        0.026       89         23.5        0.017   0.011
Endrin ketone          0.019       91          3.5        0.002   0.015
Heptachlor             0.024       90          4.1        0.004   0.005
Heptachlor epoxide     0.019       93          3.9        0.002   0.004
Methoxychlor           0.078       89          5.1        0.011   0.036
Toxaphene              0.378      100          3.0        0.034   0.093
PCB-1016               0.583       87          4.6        0.070   0.21
PCB-1221               0.613       84          2.4        0.048   0.37
PCB-1232               0.431       90          4.2        0.049   0.11
PCB-1242               0.246      101          5.7        0.042   0.096
PCB-1248               0.307       78          3.3        0.024   0.11
PCB-1254               0.303       86          2.4        0.019   0.098
PCB-1260               0.130       87          3.8        0.013   0.016

an=8
b40CFR, Part 136, Appendix B.  The minimum detection limit (MDL) is
 defined as the minimum concentration of a substance that can be
 measured and reported with 99% confidence that the analyte
 concentration is greater than zero.
                                 11-191

-------
                                TABLE V

          EMPORE™ EXTRACTION OF ORGANOCHLORINE PESTICIDES AND
                PCB's FROM COMPOSITE AVERAGE GROUNDWATER
                      AT  MDL CONCENTRATION  LEVELS
     Analyte
Aldrin
a-BHC
b-BHC
d-BHC
g-BHC (Lindane)
Chlordane
4,4'-DDD
4,4'-DDE
4,4'-DDT
Dieldrin
Endosulfan I
Endosulfan II
Endosulfan sulfate
Endrin
Endrin aldehyde
Endrin ketone
Heptachlor
Heptachlor epoxide
Methoxychlor
PCB-1016
PCB-1260
                              SPIKE
Ug/L
SAMPLE ANALYSIS
Heana
           %RSD
0.016
0.011
0.024
0.011
0.014
0.317
0.018
0.022
0.017
0.018
0.018
0.029
0.022
0.024
0.027
0.019
0.027
0.019
0.076
0.563
0.127
109
82
71
80
76
80
77
59
90
90
89
62
74
76
82
83
71
84
85
91
95
8.4
6.1
6.5
7.6
6.8
11.2
5.7
5.2
5.6
14.0
5.7
10.3
8.1
5.4
12.9
6.2
6.5
6.0
6.0
2.5
2.7
 MDL

Ug/L

0.004
0.002
0.003
0.002
0.002
0.035
0.002
0.002
0.003
0.007
0.016
0.006
0.017
0.003
0.008
0.016
0.004
0.003
0.012
0.038
0.010
ln=8
                                  1-192

-------
                                                           TABLE VI

                                  COMPARISON OF LIQUID-SOLID AND LIQUID-LIQUID EXTRACTION OF
                                  ORGANOCHLORINE  PESTICIDES  FROM COMPOSITE HIGH PARTICULATE
                                   GROUNDWATERS AT VALIDATION CONCENTRATION  LEVELS  (lug/L)
                  Analyte
to
CO
Aldrin
a-BHC
b-BHC
d-BHC
g-BHC (Lindane)
4,4'-ODD
4,4'-DDE
4,4'-DDT
Dieldrin
Endosulfan I
Endosulfan II
Endosulfan sulfate
Endrin
Endrin aldehyde
Endrin ketone
Heptachlor
Heptatchlor epoxide
Methoxychlor

AVERAGE:
GROUNDWATER #1: TSS = 18 e/L
5/9 EMPORE
Mean %R
n=5
72
78
80
77
80
79
78
72
81
78
76
81
82
66
83
72
81
79
77
LSE

%RSD
6.5
8.5
8.0
9.3
8.3
5.6
5.8
5.4
7.6
7.5
7.5
8.4
7.1
6.5
8.7
6.6
7.4
5.6
7.2
4/25
Mean ฃR
n=3
31
79
94
84
82
35
31
26
54
61
64
75
56
77
78
38
64
36
59
LLE

%RSD
6.9
4.2
3.3
3.6
4.1
4.4
5.7
3.6
6.2
5.4
5.4
4.2
4.8
3.8
5.5
4.5
4.8
0.7
4.5
GROUNDVATER #2
5/14 EMPORE
Mean %R
n-5 %.
59
73
77
75
76
62
60
65
73
74
72
77
77
77
77
61
75
68
71
LSE

RSD
8.8
4.9
3.6
5.6
4.7
6.1
6.8
5.7
3.8
4.2
4.4
4.5
3.4
6.0
4.3
5.0
3.6
2.2
4.9
: TSS = 15
5/14
Mean XR
n=4
45
89
101
94
93
43
35
32
67
74
72
84
71
89
85
37
74
48
68
s/L
LLE

2RSD
13.6
0.4
2.7
1.1
3.1
9.8
8.3
8.6
6.2
6.4
6.7
6.1
6.0
2.1
3.9
10.2
5.1
7.7
6.0

-------
                               TABLE VII

          EMPORE EXTRACTION OF ORGANOCHLORINE PESTICIDES FROM
               COMPOSITE HIGH  PARTICULATE GROUNDWATERS AT
                VALIDATION CONCENTRATION LEVELS (lug/L)
                  MEAN  ELUTION EFFICIENCIES:  A/(A+B)
TSS (g/L)
n =
Analvte
Aldrin
a-BHC
b-BHC
d-BHC
g-BHC (Lindane)
4,4'-DDD
4,4'-DDE
4,4'-DDT
Dieldrin
Endosulfan I
Endosulfan II
Endosulfan sulfate
Endrin
Endrin aldehyde
Endrin ketone
Heptachlor
Heptachlor epoxide
Methoxychlor
18
5
GW n
0.925
0.991
0.980
0.987
0.991
0.929
0.926
0.928
0.955
0.954
0.956
0.968
0.956
0.973
0.975
0.932
0.959
0.923
15
5
GW #2
0.864
0.977
0.964
0.972
0.976
0.861
0.856
0.873
0.923
0.930
0.925
0.937
0.926
0.953
0.948
0.898
0.935
0.873
2.4
6
GW #3
0.930
0.977
0.956
0.973
0.974
0.933
0.922
0.917
0.946
0.941
0.942
0.949
0.943
0.948
0.958
0.925
0.946
0.910
1.8
3
GW #4
0.959
0.988
0.976
0.986
0.982
0.968
0.962
0.954
0.957
0.967
0.970
0.976
0.969
0.965
0.980
0.952
0.968
0.954
                                                                   Mean

                                                                  0.920
                                                                  0.983
                                                                  0.969
                                                                  0.980
                                                                  0.981
                                                                  0.923
                                                                  0.917
                                                                  0.918
                                                                  0.945
                                                                  0.948
                                                                  0.948
                                                                  0.957
                                                                  0.949
                                                                  0.960
                                                                  0.965
                                                                  0.927
                                                                  0.952
                                                                  0.915

AVERAGE:                  0.956      0.922      0.944      0.968       0.948
                                  1-194

-------
69               IMPROVING THE ANALYSIS OF SEMI-VOLATILE POLLUTANTS

          Christine Vargo. Applications Chemist, Neil Mosesman, Technical Marketing Manager,
          and Gary Barone, Research Chemist, Restek Corporation, Bellefonte, Pennsylvania  16823

          ABSTRACT
          Complex resolution and monitoring requirements established in EPA Method 8270
          demand the use of capillary columns that have high inertness, efficiency, and thermal
          stability. Recent polymer technology has been developed that substantially improves
          capillary columns used in the analysis of semi-volatile pollutants. Columns produced with
          this technology exhibit increased response factors for active compounds such as 2,4-
          Dinitrophenol and 4-Nitrophenol and increased thermal stability, resulting in faster
          analysis times and lower column bleed.  Data and chromatograms will be shown
          comparing the analysis of semi-volatile pollutants on conventional capillary column and
          columns made with new technology.  Direct comparisons will be shown of response
          factors, analysis times, and column bleed between the columns.
          INTRODUCTION
          The complex resolution and monitoring requirements established under RCRA, SARA,
          and SDWA, demanded improvements in existing analytical methodology. The EPA
          responded to this need with the development of method 8270, a GC/MS method for the
          analysis of semi-volatile pollutants.  This method utilizes high resolution  capillary
          chromatography. Capillary columns have the required inertness to allow  acidic, basic,
          and neutral compounds to be analyzed simultaneously, the efficiency to separate highly
          complex mixtures, and the thermal stability to analyze high molecular weight compounds.

          With the widespread use of method 8270, it has become evident that not all capillary
          columns have the necessary inertness for trace analysis of active compounds. Others do
          not have the efficiency to resolve isomer pairs which cannot be distinguished by their
          spectra alone.  Still other columns do not have the thermal stability essential to analyze
          high molecular weight compounds and reduce analysis times.

          Recent polymer technology has been developed that yields a column with substantially
          improved inertness, efficiency, and thermal stability for the analysis of semi-volatile
          pollutants - the XTI-5.

          The response of phenols is an excellent indication of capillary column inertness. Figure 1
          shows a total ion chromatogram of fifteen phenols and six internal standards on a XTI-5
          capillary column. The phenols show excellent peak symmetry and response at 50ng/w/,
          indicating the inertness of the column.

                      Figure I - Phenols Look Exceptional With GC/MS Analysis on an XTI-5
                                                    Phenols
                                                    1. Phenol
                                                    2. 2-chlorophcnol
                                                    3. 2-methylphenol
                                                    4. 4-raelhylphcnoI
                                                    S. 2-nitrophenol
                                                    6. 2.4-diinethylphenol
                                                    7. Benzoie acid
                                                    8. 2,4-dichlorophenol
                                                    9. 4-chloro-3-melhy!phenol
                                                    10. 2,4,6-irichlorophcnol
                                                    11. 2.44-trichlorophcnol
12. 2,4-dinitrophenol
13. 4-nitropbenol
14. 4.6-dinitro-2-methylphcfiol
15. Penlachlorophcnol
IS
IS I. l,4-dichlorobenzene-d4
IS2. N.phthalene-d8
1S3. Accnaphthalcne-dlO
IS4. Phenanthracene-dlO
IS5. Cnrysene-dl2
IS6. Perylcne-dl2
                                                   30m, 0.25mm ID, 0.25fim XTI-5 (cat.# 12223).
                                                   l.Ojil splilless injection. 40ng of phenols and IS mix.
                                                   Oven temp.:   40ฐCio350ฐC 3 15ฐC/min. Hold ISmin.
                                                   Inj-temp.:    350ฐC    Del.:       MS (TIC)
                                                   Scan rale:    1.5scan/sec. Scan Range:  35-400AMU
                                                 11-195

-------
Many EPA and CLP methods require minimum response factors and linear calibration
curves over a concentration range of 20 to 160ng.  Linear response factors are another
indication of column inertness and critical for environmental analyses. CLP protocols list
nineteen semi-volatile compounds as having minimum Relative Response Factors (RRF)
criteria of 0.010. These low response factors are due to these compound's poor linearity
and sensitivity. Figure 2 shows calibration curves on the XTI-5 for two erractically
performing compounds, 2,4-dinitrophenol and 4-nitrophenol. The calibration curves of
these phenols on the XTI-5 is very linear, even over a concentration range of 20 to 160ng.

                                  Figure i- XTI-5 Calibration Curve
                   l"
                   S. U

                   I *•
                   ac
                     01
30m.Oi23mmlD.OJ5|un XTM
Table 1 shows the response factors and percentage of Relative Standard Deviations (RSD)
calculated for 2,4-dinitrophenol, 4-nitrpphenol, pentachlorophenol, and benzoic acid. The
response factors were calculated by using the internal standard that elutes closest to the
compound (ie., dlO-phenanthracene for Pentachlorophenol and dlO-acenaphthene for 2,4-
dinitrophenol). Five data points at concentrations of 20, 40, 80, 120, and 160ng/w/ were
plotted for the calibration curve. The linear plots of the phenols clearly indicate the highly
inert nature of the XTI-5 column. All response factors meet or exceed the
criteria and all RSD percentages are well below the maximum deviation criteria of 20.5%.

                     Table ( - Response Factors are Linear for XTI-5 Capillary Columns
^Compound.*.-
2.4-dhuucfilwol
mm. CLP RF- 01
•rn.CLPRF-.OI
mv.CLFRF-.05
•ILCLPRF-WA
CoU
1
2
3
4
1
2
3
1
2
3
4
1
2
3
4
20ttf
0.402
0.418
0.38S
0.372
0.186
0.1S8
0.06S
0.127
0.184
0.181
0.160
0.275
0.302
0.321
OJ14
0.359
Stag
0.503
0.502
0.475
0.480
0.194
0.196
0.110
0.138
0333
0323
0.182
0243
0.453
0.453
0.426
0.451
-Star
0.523
0.550
0.496
0.498
0.208
0-223
0.122
0.157
0253
0.242
0.203
0.260
0.466
0.494
0,465
0.539
120nt
0.553
0.494
0.524
0.445
0325
0308
0.144
0163
0.268
0.233
0319
0272
0.491
0.523
0.520
0.433
' 160ng- ttmetn-
OJ61
O560
O.SSO
0.444
0^32
0.167
0.147
0133
0376
0315
0336
0267
0.504
0.583
0.485
0.489
0.508
0.507
0303
0447
0309
0.190
0.122
0144
0343
0319
0300
0263
0.443
0.475
0.442
0.454
stider. MtSD
0.057
0X150
0X65
0043
0.017
0.024
0.023
0014
0.033
0.021
0.027
0011
(XOBg
0.071
0.060
113%
9.9%
11.0%
96%
8.1%
12.8%
18.7%
97%
13.6%
9.6%
13.4%
4 3%
16.4%
105%
18.0%
Thermal stability is of extreme importance when analyzing high molecular weight
compounds, such as PNA's found in semi-volatile pollutant analyses. Column bleed can
present several problems when analyzing environmental samples.  The rise in baseline
associated with column bleed can lead to inaccurate quantitative results, confuse spectral
interpretation and, in extreme cases, cause misidentification.  Figure 3 shows total ion
chromatograms bleed profiles of the XTI-5, the conventional RV^> ^4 a ccunpctitors,
environmental column. MSD test results clearly show the XTI-5 exhibits the lowest bleed.
                                      11-196

-------
               Figure 3 - XTI Shows Lowest Bleed of any Column
                                                       Competitors Env.
                                                       Analysis Column
                                                        Rt-5
                  Time -
                    Oven temp.:
                    Inj. temp.:
                    Scan rate:
                    MS temp:
40ฐCio350<'C 3 15ฐC/min.  Hold 15min.
350ฐC    Del.:  MS (TIC)
1 Jscan/sec. Scan Range:   35-400AMU
270ฐC
Figure 4 shows the analysis of the semi-volatile compounds monitored in EPA's Contract
Lab Program on a 30m, 0.25mm ID, 1 .Own XTI-5. Analysis times is complete in 45
minutes and bleed is minimal at 325ฐC.

               Figure 4 - Semi-Volatile Pollutant Analysis on 30m, 0.25mm ED, l.Oum XTI-5
The new XTI-5 capillary column can improve the consistency and reliability of your semi-
volatile pollutant data.  The technology used to produce these columns yields capillary
columns with improved inertness, increased efficiency, and higher thermal stability.
                                        11-197

-------
File:             A:\0701002.D
Operator:
Date Acquired:     3 Jan 91   5:38 pro
Method File Name: benzphen.M
Sample Name:                      40ng std
Misc Info:                        rtx5 xti 30,.25,.25  40(1)-3JO  @  10/min
Bottle Number:    7
Abundance TIC: 0701002. D
3000000-
•
2800000-
2600000-
•
2400000-
2200000-
2000000-
•
1800000-


1600000-

1400000-

1200000-
1000000-
800000-
600000-
400000-

200000-
0-
rime ->
IS5





46

3

IS1




2
1.





i
























5



7.



IS3






10
9













,















11









12







IS4








15



14
I13


1









IS
















i. . u
5




IS6











J — ~










-trfMtviirfftiU^.
^
5.00 10.00 15.00 20.00 25.00 30.00 35.00
                                    11-198

-------
                                              Figure  2 - XTI-5 Calibration Curve
CO
CO
                 fe
0.8


0.6


0.4
                                   50
   100
Nanograms
                                       150
                                 ~*~ 4-nitrophenol       ""*" 2,4-dinitrophenol

                                 Co1# 18946B 30m, 0.25mm ID, 0.25nm XTI-5
200

-------
N>
8
                    Table*! - Response Factors are Linear for XTI-5 Capillary Columns
• Compound
ColJ
2pซg
SOiig
8%&
^/JlOftgl;!;
160ng
mean ซ
jtd. dev.
f *fe iFfc C3tr*1i
/V m%,K?K,Jp

2,4-dinitrophcnol
min.CLPRF-.Ol
4-nitrophenol
min.CLPRF-.Ol
Pentachlorophenol
min. CLP RF-.05
Benzoic Acid
min. CLP RF-N/A
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
0.402
0.419
0.389
0.372
0.188
0.158
0.085
0.127
0.184
0.181
0.160
0.275
0.302
0.321
0.314
0.359
0.503
0.502
0.475
0.480
0.194
0.196
0.110
0.138
0.233
0.223
0.182
0.243
0.453
0.453
0.426
0.451
0.523
0.550
0.496
0.498
0.208
0.223
0.122
0.157
0.253
0.242
0.203
0.260
0.466
0.494
0.465
0.539
0.553
0.494
0.524
0.445
0.225
0.206
0.144
0.163
0.268
0.233
0.219
0.272
0.491
0.523
0.520
0.433
0.561
0.560
0.550
0.444
0.232
0.167
0.147
0.133
0.278
0.215
0.236
0.267
0.504
0.583
0.485
0.489
0.508
0.507
0.503
0.447
0.209
0.190
0.122
0.144
0.243
0.219
0.200
0.263
0.443
0.475
0.442
0.454
0.057
0.050
0.055
0.043
0.017
0.024
0.023
0.014
0.033
0.021
0.027
0.011
0.073
0.088
0.071
0.060
11.3%
9.9%
11.0%
9.6%
8.1%
12.8%
18.7%
9.7%
13.6%
9.6%
13.4%
4.3%
16.4%
18.5%
16.0%
13.2%

-------
Figure*^ - XTI Shows Lowest Bleed of any Column
  •3
   c
  JO
       700000-
       650000-
                                                Competitors Env.
                                                Analysis Column
                                                     RtK-5
                                              XTI-5
Time ->?.oo   10.00

  Oven temp.:
  Inj. temp.:
  Scan rate:
  MS temp:
                           i—•
                     15.00  20.00
                               25.00
                                    30.00
                                     35.00  40.00
                  40ฐC to 350ฐC ฎ 15ฐC/min. Hold 15min,
                  350ฐC     Del.:   MS (TIC)
                  1.5scan/sec. Scan Range:   35-400AMU
                  270ฐC

-------
    File:              D:\DATA\40NGPP.D
    Operator:
    Date Acquired:      1 Apr 91  12:39 pm
    Method File Name: PP.M
    Sample Name:
    Misc Info:
    Bottle Number:     1
Abundance

 1800000-j


 1700000-


 1600000:


 1500000-


 1400000:


 1300000-


 1200000:


 1100000-


 1000000-
                                   TIC: 40NGPP.D
rime ->    15.00
                      20.00
25.00
30.00
35.00
40.00
                                        11-202

-------
yn                Electrospray Combined with Ion Trap Mass Spectrometry
'v                            for Environmental  Monitoring
                           Robert 0.  Voyksner and Hung-Yu Lin
                               Research Triangle Institute
                                     P. 0.  Box 12194
                          Research Triangle Park,  NC  27709 USA
           Environmental  monitoring of many non-volatile or thermally unstable
       polar  organics relies on the development  of sensitive,  specific and cost
       effective LC/MS techniques.  Electrospray can meet these goals since it
       has the capability to generate molecular  ions from low pg quantities of
       most environmentally relevant compounds.   The coupling of electrospray
       with an ion trap mass spectrometer (ITMS) offers the potential to gain
       structural information through MS/MS without the additional cost of
       multiple mass analyzers, as well as  achieving better sensitivity than
       conventional quadrupole mass analyzers.  This paper reports on the
       coupling of a commercial electrospray interface to an ITMS.  The system
       was evaluated for  its use for environmental monitoring.
           The electrospray source was interfaced to a second analyzer mounted in
       the ITMS vacuum chamber with minimal changes to either commercial unit.
       The use of a second analyzer in the  ITMS  minimized switch time between El
       and electrospray operations.  Ions formed in the electrospray interface
       were gated into the ITMS analyzer, using  the 180 V gating circuit employed
       for El operations, with good efficiency and minimal losses from
       collisional activation.   The determination of numerous pesticides,
       herbicides, dyes,  and potential DMA  adducts proved that the electrospray
       ITMS combination could acquire high  fg to low pg full scan spectra.  These
       sensitivities were 10-30 times superior to those obtained by electrospray
                                         1-203

-------
on a quadrupole mass analyzer.  The electrospray spectra of these
compounds usually only consisted of [M+H3* and/or [M+Na]  ions and no
fragment ions.  No thermal decomposition products were detected for the
thermally labile compounds analyzed.  Occasionally other adduct ions were
detected, such as [M+NH41+ and [M+m-triethylamine]"1" when buffers such as
ammonium salts or triethylamine were used in the LC mobile phase.  The use
of collisional activation decomposition in the ITMS analyzer proved useful
in generating MS/MS spectra from the protonated molecular ion or adduct
ion for each compound, resulting in fragment ions for identification or
confirmation.
    Although the  information  described  in this article has been funded
wholly or in part by  the  Environmental  Protection Agency under contract
68-02-4544 to Research Triangle Institute, it does not necessarily reflect
the views of the Agency and no official endorsement should be  inferred.
                                  11-204

-------
Recent Advances In the Use of Supercritical Fluid Extraction for Environmental
Applications

J.M. Levy*, A.C. Rosselll, D.S. Boyer, M. Ashraf-Khorassani
Suprex Corporation, 125 William Pitt Way, Pittsburgh, PA 15238

One of the distinct advantages In using supercritical fluid extraction (SFE) is the
ability to achieve selective  extractions  based  upon  differences in  threshold
solubilities of different analytes. Different threshold solubilities can be attained by
varying extraction pressures and/or temperatures.  The addition of modifiers to
the  supercritical fluid  extracting phase has also  enhanced the  extraction
efficiency  of  specific  analytes.    Depending  on  the  sample  matrix,  the
enhancement of solubilities could be offset by diffusion enhancement or by the
displacement of analytes  from  the outer or  Inner surface of matrix particles.
There also Is the possibility of performing  chemical reactions, such  as acid
 hydrolysis and functional  group derivatization,  during the  SFE step  thereby
 achieving  distinct     ctivity enhancements  for  specific analytes in complex
 matrices. Further selectivity enhancements can be achieved by utilizing  different
 adsorbents which  are added to the extraction vessel with the  sample or are
 packed into secondary extraction vessels which  are placed down stream of the
 sample extraction vessel,   in this work, each of these enhancements will be
 investigated  and  demonstrated  using a  newly developed  directly  coupled
 SFE/GC field portath system with environmental matrices such as soils, marine
 sediments, drilling muds, sludges, and ashes.
                                   II-205

-------
-79          USING SUPERCRITICAL FLUID EXTRACTION TO SEPARATE
'^                      DIESEL FROM SOIL MATRICES


        Carl A.  Craig.  Ph.D.,  Suman Prashar,  and Jennifer Cunningham
        BC Analytical,  1255  Powell Street,  Emeryville,  CA   94608;
        Bruce E. Richter, Ph.D., and Amy Rynaski Lee Scientific,  4426
        South Century Drive,  Salt  Lake  City,  Utah 84123
        ABSTRACT
        The  initial  and  arguably the  most  crucial  step  in  most
        environmental  analytical  analyses  is  the  separation  of
        analytes from  the sample matrix.  Separating  or  extracting
        organic compounds  from  soil  is  currently accomplished  by
        several methods: heating, purging or  by  solvent extraction.
        Of these,  solvent  extraction  is the  principal  method  used.
        Because there is interest in  reducing dependance on solvent-
        use extraction technologies,  we have  examined Supercritical
        Fluid Extraction (SFE)  as a method to separate semi-volatile
        organic analytes from soil matrices. Our initial data indicate
        that SFE works very well to extract diesel from soil samples
        and reduces  the amount  of solvent required by a factor of ten
        compared to  Ultrasonic  Extraction (USE).  This paper  will
        present the  results of a direct comparison  study  of diesel
        levels in soil extracts;  where each soil sample was submitted
        for both USE and SFE extraction and the extracts analyzed by
        gas chromatographic methods.


        INTRODUCTION
        Supercritical fluid extraction of analytes from environmental
        sample  matrices is generating  interest  in the  analytical
        laboratory.  The interest stems from reports of initial success
        in separating organic analytes from samples.1"8  In addition to
        the technical feasibility, there is  a genuine concern among
        the environmental lab community  to  reduce the amount of toxic
        and  hazardous  solvents  in  the  work-place.   These  two
        influencing factors guided our efforts to investigate the use
        of SFE.  Our efforts have focused  upon comparing the extraction
        techniques   (USE and SFE)  directly.  We  accomplish  this  by
        submitting   soil  samples to  both  extraction  methods  and
        comparing  the results of the  GC-FID quantitation  for the
        extracts.  The soil samples that we investigated were actual
        soil  samples which  had  been  submitted  by  clients to  BC
        Analytical  for diesel  hydrocarbon analysis.
                                    11-206

-------
Diesel hydrocarbons are  herein defined as a  class of C12-C25
hydrocarbons. A GC chromatogram  of  a common diesel standard
is provided in Figure  1.  To quantitate the diesel in a sample
extract, integration  of peak  areas over the  entire diesel
spectrum range  was done.  In  addition  to the integration,
diesel must be qualitatively  identified by a characteristic
chromatographic fingerprint. This is necessary to insure that
diesel not gasoline or mineral spirits  has been quantitated
by the integration.
EXPERIMENTAL
Supercritical fluid  extractions  were all  performed  using a
Dionex Model  703 Extraction  System.  All  of  the extraction
cells  were 5  cm  X  9.4  mm  I.D.;  the  end  caps  contained
stainless steel  frits  (0.5  urn pore size). Eight extractions
were run simultaneously, the outlet of each cell was connected
to a separate temperature controlled restrictor. Collection
of  the analytes  involved a  unique  dual  chamber vial  and
double-sided  Teflon  coated  septa. Approximately  10  raL of
methylene chloride were placed into each vial. The collection
system was then  electronically cooled to 5ฐC. Extractions were
carried out with 100%  CO2 that contained  1500 psi  of helium
headspace  (Scott  Specialty  Gases).  All  extractions were run
at  75ฐC  and 300  atmospheres  for  a  total of  15 min.  The
restrictors  were heated  to  150ฐC  to eliminate restrictor
plugging. The flow rate was on average 250 mL/min as gas.

Three one gram  samples of each soil were weighed and placed
in separate vials. The  sample size was  selected for ease of
handling and extraction. No effort was taken to optimize the
weight of  the sample extracted by  SFE". Two  of  the three Ig
samples were extracted under the conditions detailed above.
The third Ig sample was used to determine the moisture content
of the soil.

Each SFE  extract (approximately  10 mL) was  reduced  to less
than 5  mL total  volume under a  flush  of high-purity grade
nitrogen  gas. The extracts were  then transferred to  10 mL
concentrator tubes via pasture pipets that were packed with
anhydrous sodium  sulfate. Each extract  vial was then rinsed
with 1-3 mL of  methylene  chloride, and  the rinsate added to
the concentrator  tubes. The extracts  were then reduced to 1
mL final volume under a flush of nitrogen. The 1 mL extracts
were quantitatively  transferred to 1.5  mL vials with Teflon
lined septa and screw caps. These extracts were stored at
4 ฐC prior to GC analysis.

Ultrasonic extraction was carried out in accordance with EPA
method 3550A. Flow diagrams of the USE  and  SFE methods are
shown in Figures  2 and 3.
                             11-207

-------
Gas chromatographic identification and quantitation of diesel
in the USE and  SFE  extracts  was accomplished with a HP 5890
gas chromatograph using a 30 meter DB5 capillary column. Prior
to each daily run, column and septa conditioning was conducted
at 300ฐC  for  90  minutes.  The oven temperature was brought to
40ฐC  and  allowed to stabilize.  The instrument was calibrated
using 50, 100,  250,  500,  and 1000 ppm diesel standard. Data
were analyzed with Nelson 2600 software.
RESULTS AND DISCUSSION
This investigation involved  analyzing extracts from 42 soil
samples. The  soil samples were  submitted by  clients  to BC
Analytical for analysis of diesel hydrocarbons.  Once the study
was under way, all of the samples submitted to  BCA for diesel
analysis were extracted by both USE and SFE methods. Many of
the  soil  samples investigated have less  than  the reporting
detection limit  of diesel  in the USE  extract.  Thirty two of
the  42  samples were  reported as "not  detected"  for diesel
hydrocarbons by ultrasonic extraction and GC-FID quantitation.
(See  Table  1.)   The  remaining   ten   samples  had  reported
quantities of diesel in the 3550A extract ranging between 1-
2800 ppm.

Similarly, 32  of the 42 SFE  extracts  were confirmed not to
contain  diesel  hydrocarbons. Supercritical   extracts  were
reported to  contain  diesel if duplicate  extracts contained
diesel. (See Table 1.) In several cases, a duplicate extract
was  unavailable;  the  data were  then  based  upon  a single
replicate.

Both extraction methods yielded ten soil extracts with diesel
hydrocarbons identified.  Seven of these ten extracts were for
the  same  soils independent  of the extraction method.  (See
Table 2.)  For three soils,  a diesel  quantity was reported
using extraction  method 3550A and no confirmation was reported
in the SFE extracts.  Alternatively, three SFE  extracts gave
reportable levels  of diesel  when  the 3550A extract  had no
diesel hydrocarbons reported. However,  there is  generally good
agreement between the  incidence  of  soil extracts containing
diesel hydrocarbons using these two extraction methods.

During the investigation, the accuracy of the SFE extraction
was not measured using surrogates or spiked soils. Therefore
the "true" value  of  analyte  in the soil  is taken  to be the
value reported for the  3550A method. Of the seven soil samples
for which both extraction  methods gave positive results for
diesel  analysis,  only  two  results  were   significantly
different. Sample 32  was reported to give 700 ppm diesel via
method 3550A while the same soil  gave  only 115  ppm in the SFE
                            11-208

-------
extract. However,  the converse was also observed.  For soil 18,
the reported quantity of  diesel  in the 3550A extract was 46
ppm and the SFE extract gave 535 ppm.

Most of  the  GC-FID chromatograms of the  SFE extracts had a
characteristic pattern which was not attributed to hydrocarbon
in the  soil.  (See Figure  4.)  However this  pattern  was not
observed in the sample blank. This interference was manually
subtracted from each chromatogram to allow for quantitation.
The source of the  contamination  is unknown and is currently
under investigation. As  a result  of  the  contamination, the
reporting detection limit (RDL) for the SFE extracts was set
at 5 ppm. The problem that one  encounters  with a  5 ppm RDL is
that  a   hydrocarbon  pattern  (such as  the  mineral  spirits
identified in  Figure 4)  may be  observed but not  reliably
quantitated because of the high level of interference.
SUMMARY
Supercritical  fluid  extraction will  find  wide  use  in  the
environmental  laboratory  because  of the  ability  to extract
organic analytes from soil matrices without the use of large
volumes  of  hazardous  solvents.  However,  before  this  will
happen, the utility of SFE techniques on actual field samples
must be demonstrated.  Our results indicate that SFE works very
well to  extract  diesel hydrocarbons from soil matrices.  In
fact these  results  indicate that  SFE is  as efficient  as
ultrasonic extraction in removing diesel from sample matrices.
Of the  42 samples that were analyzed, SFE gave  similar results
to those obtained using USE.  In addition the differences that
were observed were not one sided.  In three cases diesel was
identified in the ultrasonic  extract and was not confirmed in
the corresponding SFE  extracts.  Similarly,  there were three
sets of  SFE extracts  where reportable  levels  of diesel were
identified, yet no diesel was  found in the USE extracts.  In
these six instances,  the  reported  quantity of diesel in the
soil was  less  than 31  ppm. The differences observed between
these  extracts  might  very  well  be  attributed  to  non-
representative sample sizes,  or sample inhomogeneity. During
the next phase of the research  we plan  to  investigate  the
effect of sample size on extraction optimization. In addition,
a reduction in the interference observed by the FID detector
in the SFE extracts must be eliminated in order to lower the
detection limit to levels equivalent to current methods.

Supercritical fluid extraction does significantly reduce the
amount  of solvent  required  for  the  extraction  of  diesel
hydrocarbons from soil samples from approximately 400 mL for
EPA method 3550A to approximately 20 mL for the SFE. Because
of these initial  successes we plan to continue to investigate
the use of SFE to separate diesel from soil matrices.
                            11-209

-------
REFERENCES

1.) Hawthorne, S.B. Anal. Chem. 1990, 62, 633.

2.) Yu, X.; Wang,  X.;  Bartha,  R.;  Rosen, J.D. Environ. Sci.
    Technol.  1990, 24. 1732.

3.) King,  J.W. Journal  of  Chromatographic Science 1989, 27,
    355.

4.) Ndiomu, D.P.;  Simpson,  C.F.; Analytical Proceedings 1989,
    26, 393.

5.) Janda, V.; Steenbeke,  G.;  Sandra,  P. Journal of
    Chromatography, 1989, 479, 200.

6.) McNally, M.P.; Wheeler,  J.R. Journal of  Chromatography
    1988,  435, 63.

7.) Wright, B.W.; Frye, S.R.; McMinn, D.G.; Smith, R.D. Anal.
    Chem.  1987, 59, 640.

8.) Hawthorne, S.B.; Miller, D.J. Anal.  Chem. 1987, 59, 1705.
                            11-210

-------
                                                                          C\J
Figure 1 . Capillary gas chromatogram of a 250 ppm
         diesel hydrocarbon standard

-------
         Ultrasonic Extraction  (SW-846 Method 3550A)

         General Tasks                      Analysis Specific Tasks
CO
       Weigh 30g of sample
       into a 400 mL beaker
      'Add 150 mL of extraction/
          solvent to beaker  /
              I
         Sonicate sample
          for 5 minutes
              I
r       Decant solvent over NaฃO4/
           into evaporator
              1
       Add 100 mL of extraction
                to beaker
       Sonicate for 5 minutes.
              I
       Rinse beaker with 50 mL
        of extraction solvent
              I
      Decant solvent over
           into evaporator
              I
          Distil off solvent
      Evaporate remaining solution x
       to correct final volume
      (Transfer sample to vial)
    Pesticides. PCB. .. Hexane/Acetone
    Semivotetie organics.Diesel.. .
             Total volume of
             solvent required
                400 mL
      Solvent exchange to hexane
      for pesticide and PCS analysis
1 mL for ofesel analysis
2 mL for semivolatle organics
10 mL for pesticides and PCB
      Plods! cleanup... Pesticides
      Acid cleanup..  .PCB	
    Figure 2. Flow  diagram of ultrasonic extraction
                                  1-212

-------
                Supercritical Fluid Extraction

     General Tasks                       Analysis Specific Tasks
    Weigh 1g of sample
  into an extraction thimble
          T
   Load eel Wo extractor
          i
        Extract for
         15 minutes
          I
  Collect analyte h solvent trap
     Pesticide. PCS. ..Hexane
     Semivolatite organics, Diesel.
 Transfer solvent with pipet anc
   NagS(j into concentrator
     Rinse vial with
     5 mL of solvent
             Total volume of
             solvent required
                20 mL
  Evaporate remaining solution/
   to correct final volume
          I
  ("Transfer sample to vlaf)
1 mL for diesel analysis
2 mL for semivolatfle orgaracs
10 mL for pesticides and PCB
      Fkxisl cleanup. . . Pesticides
      Acid cleanup. ..PCB	
Figure 3.  Flow diagram of Supercritical fluid extraction
                                  11-213

-------
03
O
CO
CO
-t—'
o
10
C\J
              IB
              (•J
                                ID
                                n
                           r~
                           I
                Low level (3. 2 ppm) mineral spirits
   A	I	I	I	I	\	l_
                                      _1	I	L
                                                        —   Contaminants
                                                     U
                                                         CTi
                                                         tr
                                                         sc
                                                                      a
                                                                      •*•
                                                                       ซ
                                                                      CB
                                                                          J*^,
                                                       1    1    1   1
              Figure 4.  Capillary gas chromatogram of supercritical fluid extract

                        for sample number 31
                                                                                                    C\

-------
TABLE 1.   RESULTS OF DIESEL HYDROCARBON ANALYSES
SAMPLE
NUMBER
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
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
BY EPA 3550/8015 BY SFE/8015
(mg/kg) (mg/kg)
REPLICATE 1 REPLICATE 2
2800 2280
<1 17
<1 <5
<1 <5
<1 <5
<1 8.5
<1 <5
<1 7.0
<1 <5
<1 <5
<1 <5
<1 <5
<1 <5
<10 <10
<1
<1
<1 <5
46 510
560 635
5 <5
<1 <5
<1
6
<10 <10
<10 <10
<1 <5
<1 <5
1 <5
<1 <5
<1
<1 <5
700 118
<1 <5
140 65
<1
<1 <5
30 <5
<1 <5
30 55
15 6.6
<1 <5
<1 <5

<5
12
<5
<5
9.5
<5
7.8
10
<5
<5
<5
<5
<10
<5
<5
<5
560
581
<5
<5
17
5.2
<10
<10
<5
<5
14
<5
<5
<5
112
<5
167
<5
<5
<5
<5
15
<5
<5
<5
PERCENT
MOISTURE

19
20
23
19
12
19
4.7
21
14
6.0
14
22
14
12
10
11
17
6.6
14
7.2
11
19
13
11
15
15
21
23
25
20
22
22
19
20
25
26
18
16
15
18
19
                        11-215

-------
TABLE 2.   EXTRACTS  OF SOILS CONTAINING DIESEL HYDROCARBON
           LEVELS  GREATER THAN THE REPORTING DETECTION LIMIT
SAMPLE
NUMBER
1
6
8
18
19
20
22
23
32
34
37
39
40
BY EPA 3550/8015
(mg/kg)
AVERAGE
2800
<1
<1
46
560
5
<1
6
700
140
30
30
15
BY SFE/8015
(mg/kg)
OF 2 REPLICATES
2280a
9.0
7.4
535
608
<5
17a
5.2
115
116
<5
35
<5
a)  Single extract analysis  (no replicate available)
                             1-216

-------
70          CREATIVE REVIEW OF "TENTATIVELY IDENTIFIED COMPOUND"
'                       DATA USING THE RETENTION INDEX
                               WILLIAM P. ECKEL
                               VIAR AND COMPANY
                             300 NORTH LEE STREET
                           ALEXANDRIA, VIRGINIA 22314
                                      1-217

-------
INTRODUCTION

      Identifying non-target  compounds in gas  chromatography-mass  spectrometry (GC/MS)
data is a difficult and time-consuming task, even for those trained in interpretation of mass
spectra.  It is  rare that a compound can be "tentatively identified"  with confidence when  the
only information available is the mass spectrum and the computer library matches.

      Aside  from prior  knowledge of what types of compounds  to  expect in a  particular
sample, the only other  piece  of data available to  the data  reviewer when interpreting a
spectrum is the GC retention time.  Because absolute  retention  times are dependent on a large
number of experimental conditions, the retention index was developed to express retention
data relative to  a standard set of compounds.  The original system, called the Kovats  index,
dates from 1958 and uses the normal alkanes as the retention index standards.

      A retention index  system was developed  in 1979 by  Lee and co-authors (1)  for  use in
identifying polycyclic aromatic hydrocarbons (PAH).  In the "Lee" retention index system,  the
retention index standards and their retention indices are naphthalene (1=200.00), phenanthrene
(1=300.00), chrysene  (1=400.00) and picene  or  benzo(ghi)perylene (1=500.00).   Beeause  the
perdeuterated  analogs of the first three of these are  used as internal standards in the several
variations of EPA method 625 for extractable compounds, the Lee retention index can be used
NOW by  data reviewers in identifying unknown compounds.  The Lee retention  indices  of
several hundred compounds of environmental interest  are available in the literature (1-5).
                                        1-218

-------
CALCULATION OF THE LEE RETENTION INDEX

The Lee retention index, I,  is defined as:

I = 100 ( rtunk - rtz / rtz+1  - rtz ) + 100 (Z).

where rtunk is the retention time of the unknown compound, rtz and rtz+j are the retention
times of the bracketing retention index standards, and Z is the number of benzene rings in the
retention index standards.  To summarize:

Standard                   Retention Index                           Z

naphthalene                      200.00                             2
phenanthrene                     300.00                             3
chrysene                         400.00                             4

     For unknowns which elute before  naphthalene or after chrysene, the retention index is
"projected" using the two retention index standards closest to the unknown compound.-
                                      11-219

-------
APPLICATION OF THE LEE RETENTION INDEX TO DATA REVIEW

     This  paper  presents some  applications  of the  Lee  retention index  system to  the
identification of compounds in real environmental samples.

     The first  example (Figure 1) is the spectrum of a dimethylphenol whose experimental I
was  194.35.  The target compound 2,4-dimethylphenol was also found  in  this sample at
1=190.70.  Judging from the known retention indices of the isomeric dimethylphenols, and the
measured bias between the experimental index and the  known index  for the 2,4- isomer, the
most probable identification for the unknown is 3,5-dimethylphenol.

     Figure 2 shows  the spectrum of a polycyclic aromatic hydrocarbon of molecular weight
252. Three of the library matches are target compounds which were also found in the sample
at different scan numbers than this unknown.   Their scan numbers, and  experimental  and
known retention indices are shown.  Based on the known retention order of these compounds
(as  shown by the  retention indices) and considering the experimental bias, the  proximity of
the unknown to the scan number for benzo(a)pyrene identifies it as benzo(e)pyrene.

     Figure 3 shows the spectrum of an unknown with an apparent molecular weight of about
190. Comparison of the experimental spectrum with the library  spectra of compounds of MW
192 shows  much more intense ions at m/z 190 and m/z 189 in  the unknown spectrum.   This
suggests  that the  unknown  spectrum  may not be  that of  a pure  compound, but of  two
coeluting compounds.   In fact, the experimental retention index for the  unknown lies  just
between  the known retention indices of the top two library matches, one  of  which has MW
190  and  the  other   MW   192.     Thus  both  cyclopenta(def)phenanthrene   and   9-
raethylphenanthrene appear to be present in the unknown spectrum.

     Figure 4 is the spectrum of  another pair of coeluting compounds, this  time of dissimilar
chemical classes.   An isomer of the top library match is present, judging from  the  retention
index.  However, a hydrocarbon pattern is also present at m/z 43, 57, 71, 85 etc.  The library
search  results do not suggest the presence of an alkane, but the retention index of pentadecane
is close enough to the  experimental index that  it can also be tentatively  identified in  this
spectrum.

     Finally, Figures 5a through 5d show a series  of normal alkanes that were found in a
sample.  As is usual with alkanes, the top library  matches are of widely varying chain length.
Calculation of the  retention index quickly narrows the  possibilities to one or two compounds
for each  spectrum.  Note that the tentative identification for each spectrum is not among the
top library matches.
                                       1-220

-------
CONCLUSIONS

      The  Lee  retention  index is  an  easily calculated bit of  information which can  be
extremely valuable in the  identification of unknown compounds.   It is useful in identifying
specific members  of  homologous series with identical spectra; coeluting compounds of similar
and  dissimilar chemical classes; specific  positional isomers; and  structural isomers.   More
sophisticated applications of the existing Lee retention index data, using the  principles of gas
chromatography and structure-retention relationships, are also possible.
                                        11-221

-------
REFERENCES

1.   Lee,  M.L.,  Vassilaros,  D.L., White, C.M.,  and Novotny,  M., "Retention  Indices  for
     Programmed-Temperature  Capillary-Column   Gas  Chromatography   of   Polycyclic
     Aromatic  Hydrocarbons", Analytical Chemistry 51, 768-773,  1979.

2.   Vassilaros, D.L., Kong, R.C.,  Later, D.W., and Lee,  M.L., "Linear Retention Index
     System for Polycyclic Aromatic Compounds.  Critical Evaluation and Additional Indices",
     Journal of Chromatography, 252, 1-20,  1982.

3.   Willey,  C.,  Iwao,  M., Castle,  R.M.,  and  Lee,  M.L.,   "Determination  of  Sulfur
     Heterocycles in Coal Liquids and Shale Oils", Analytical  Chemistry, 53, 400-407, 1981.

4.   Rostad, C.E., and  Pereira, W.E., "Kovats and Lee Retention Indices Determined by Gas
     Chromatography/Mass Spectrometry for Organic Compounds of Environmental Interest",
     Journal of  High Resolution Chromatography and  Chromatography Communications,  9,
     328-334,  1986.

5.   U.S.  Environmental Protection Agency,  Office of Water  Regulations and Standards,
     Industrial Technology Division, "Analytical Methods for  the NationaLSewage  Sludge
     Survey. Method 1625, Revision C. Semivolatile Organic Compounds by Isotope Dilution
     GCMS", August 1988.

-------
                                                 Figure 1
     Identification  of  Specific Dimethylphenol Isomer
    1088

  SAMPLE 1
  C8.H10.0
    1088
  MWT122
  BPK 107
— RANK 1
JT IN 2643
K PUR 866
CO
  C8.H10.0
    1088
  M WT122
  BPK 107
  RANK 2
   IN 2634
  PUR 864
  C8.H10.0
    1088
  MWT122
  BPK 122-
  RANK 3
  IN 2645
  PUR 863'
             LIBRARY SEARCH
             08/04/8516:34:00 + 8:02


             SAMPLE:! ULCCJS7689 (8-1-85)
DATA:GH057689A16*534
ENHANCED (108 2NOT)


UNKNOWN
BASE M/E: 107
RIC:457727
                                                    lexp= 194.35
                   lref = 198.95
                     I	...l...l..,,,	,,


    PHENOL,3,4-DIMETHYL-   CAS*   95-65-8


                                           lref = 201.49







    PHENOL,2,6-DIMETHYL-   CAS*   576-26-1


                                           lref= 184.08






50         100          150         200         250         300
                   IJUMMM uiiiiiriiT-jiim,.,	...,,|i,,,.,,,,.	|. in i M,, , ii ,, , ,, , ,, | , i,	


              PHENOL,2,3-DIUETHYL-    CAS*  526-75-0
                                                                       Dimethylphenol
                                                                           Isomer
                                          2,6
                                                                            2,4
                                                                        (Target Compound)
                                          3,5
                                                                            2,3
                                          3,4
                                                     'ref
                                 184.08
                                                   192.69
                                 196.53
                                                   198.95
                                 201.49
                                                                                              exp
                                        190.70
194.35
                                               Bias
      +1.99
+2.18

-------
                       Figure 2
  Identification of PAH with Molecular Weight of 252
8

1237
SAMPLE '

C20.H12
MWT252J
3 PK 252 I
* 23964 j
C20.H12
M WT 252 j
3 PK 252 i
ป 23966 j
PUR 914 '
C20.H12
MWT252.
3 PK 252
* 23965
C20.H12
1237
MWT252 j
3 PK 252
RANK 4 j
ป 23961
C20.H12
1237
MWT252J
3 PK 252
RANK 5 J
9 23962
PUR 905 =
LIBRARY SEARCH DATA: 61 127645 ป 2758 BASE M/Z: 252
06/08/906:59:00*45:58 CALI: B1127645 1 2 RIC: 51263
SAMPLE: Lป1 12764
lexp = 447.12 |
;Ti m- i-tu -,- i •'' 'fl11 V
BENZO(J)FLUORANTHENE
lre,= 440.92 |
„ 	 	 f
... 	 	 	 	 ,ll ..ill „ i
8ENZO(K)FLUORANTHENE
|fe( „ 442.56 |

BENZO(8)FLUORANTHENE
lre) = 441.74 |
^ittrrnr -, •-; - •'' I1' V '
BENZO(A)PYRENE
lret- 453.44 |

BENZO(E)PYRENE
lrel- 450.73 |
,1 ,||
• •^.^Mt.lUMt . i^.4V>ป ,, Jtt,t,V,,~,-,t 	 ปlj-.ifi ff-r,; 4 t^XtTi 	 '.'TK-n'r-r T>I 	 t r r i •r'l-ri'A '







1










PAH
Benzo (j)
Fluoranthene
Benzo (b)
Fluoranthene
Benzo (k)
Fluoranthene
Benzo (e)
Pyrene
Benzo (a)
Pyrene
ire(
440.92
441.74
442.56
450.73
453.44
Scan

2698
2703
2758
2769
IซP

438.03
438.79
447.12
448.79
Bias

+3.71
+3.77
+3.61
+4.65
                          250

-------
                                          Figure 3
     File>90868HOU
     SpkAbl1258
            Identification  of Coeluting  PAHs
INST C 2/7/90 NEAT 340 PHC/DRS CLP 5  Scan 2272
SUBAODOVC             21.72min.

UNKNOWN(S)
50 62 70 * * 109 126 152 ^
'""I1" 	 '"'1 	 !•! ••••• • 	 {IIM..IMI 	 . .,.,1111 l|limH III
File >BIGDD Phenanthrene, 4-methyl-
dn\* Ak QQQQ
181
,,,,. ,,,,,,,,,,
Scan 46878
192
198
I /
)miii|imni|
                                              O.OOmin.
•
&
Ol
-;

50 63 75 82 95 ^ ^ 152 ^ 174
/ s^ / ' \ 1 / ' / \\
File>BIGDB 4H-Cyclopenta(def)phenanthrene Scan 46722
SpkAb9999 0.00 min.
94 1g3
50 63 74 87 ^ ^ ^ ^ ,50 - , ^
/ ^ / ' ,| / / / .1 \ „
I'M1 	 1 	 'i'" 	 .i.|..i 	 ,,•••[•,,.,.. 	
File>BIGDB Anthracene, 1-msthyl- Scan 47139
Spk Ab 9999 0.00 min.
44 63 74 87 K ^ ^ ^ ,52 ^ 176
X -^ / \\ / / X ''I '

192
194
190
196
192
194
I ..,ป.|. ,.!...,
60
                100
                        120
                                140
                                       160
                                               180
                                                       200
Sample: HQ14    INSTC    injected: 2/07/90 17:02
TIC # 26  Area = 390337.0 Tentative Cone = 510.00
              BEST MATCHES
  1. Phenanthrene, 9-methyl-        192 C15H12
  2. 4H-Cyclopenta(def)phenanthrene   190 C15H10
  3. Anthracene, 1-methyl-          192 C15H12
  4. 9H-Fluorene, 9-ethylidene-       192 C15H12
  5. Phenanthrene, 9-methyl-        192 C15H12
  6. Methyl-phenanthrene or
    Methyl-Anthracene            192 C15H12
  7. Anthracene, 2-methyl-          192 C15H12
PAH
2-Methyl-
anthracene
Cyclopenta(def)
phenanthrene
Unknown
9-Methyl-
phenanthrene
4-Methyl-
phenanthrene
1 -Methyl-
anthracene
'ref
321.57
322.08

323.06
323.17
323.33
'exp


322.99



MW
192
190

192
192
192

-------
                                        Figure 4
Identification of  Coeluting  PAH  and  Alkane
    Filg>90868H014
    BpkAb4430
    57
INST C 2/7/90 NEAT S40 PHC/DRS CLP 5  Scan 1 680
SUBADDDVC             16.66min.
                         UNKNOWN
                                          168
-
-.

J
•i
i
•

^
•










43
.1


51
\

' 42


51

|


6C

5C



/
iniin'ti
File>BIGD
BpkAb999

76
\ ^
Fits >BIGD
BpkAb999

Fite>BIGO
BpkAb999

65

lm|iiiW
3
9

^
9
9
69
*-"
JrniTn
"I1""
3
9

77
\
85 ^
... 153
of, 111 126 1.41 \
/ / / / \
, , , >>.i.iK 	 	 ,u ,1, , . , , . . , 1 1 ,\ , . , . M . . , . , 	 1 1.1, , ..iii
''"' 	 |IIIII11|IIMIIIIIII1| 	 | 	 l|l
1,1'-Biphenyl,2-Methyl

10? 141 153
\. 115 128 ^ \
,| ^~~-- / / \ JL Jil
s-Tetfazine,3,6,-bis(dimelhylamino)-
1
x
Benzene, 1,1'-methylenebis-

91 nQ
\ 102 115 128 i
\
188 21-~^_
| / ^
Scan 41631
0.00 min.
168
X
I
182 183
Scan 41439
0.00 min.
38 169
^
'" 	 " 	 I 	 i
Scan 41627
0.00 min.
168
170
Sample: H014   INSTC   Injected: 2/07/90 17:02
TIC # 11   Area = 173945.0 Tentative Cone = 200.00
            BEST MATCHES
  1. 1,1'-Biphenyl, 2-methyl-           168 C13H12
  2. s-Tetrazine, 3,6-bis(dimethylamino)-     168 C6H12N6
  3. Benzene, 1,1 '-methylenebis-        168 C13H12
  4. Benzene, 1,1'-methylenebis-        168 C13H12
  5. 2H-Pyran-2,4(3H)-dione,3-acetyl-6-methyl- 168 C8H804
  6. Benzene, 1,1'-methylenebis-        168 C13H12
  7. Benzene, 1,1'-methylenebis-        168 C13H12
                         120
                                140
                                       160
PAH
2-Methyl-
biphenyl
Diphenyl-
methane
3-Methyl-
biphenyl
4-Methyl-
biphenyl
Unknown
Pentadecane
(MW212)
'ref
239.84
243.35
254.33
256.12

256.75
exp




256.56


-------
                     Figure 5a

Identification  of  Normal Alkanes
         LIBRARY SEARCH
         08/04/85 14:46:00117:46


         SAMPLE: 1 UL CCK57686 (8-1-85)
DATA: GH057689A16* 1182
ENHANCED(! OB 2NOT)


UNKNOWN
BASEM/E:57
RIC:1505270
1081 q
SAMPLE j
!
C8.H10.0
1081
MWT296
BPK 43-i
RANK 1 :
IN 21792 :
PUR 798-
C8.H10.0
1081 3
M WT352 =
BPK 43-:
RANK 2
IN 25265 :
PUR 795:
C8.H10.0
1081 3
M WT 492 :
BPK 43-i
IN 25265
PUR 785 -.


J
"'I'

(


,


II


Jl





(|


h

lexp = 433.60 |

HENEICOSANE CAS# 629-94-7
lref = 347.42 j
If il (1 1 1 • •, r iimun • T r r I.HI ,
PENTACOSANE CAS# 629-99-2
lref = 400.45 |
It M n i , .
PENTATRIACONTANE CAS# 630-07-9
lref = 538.06 |
1
,,l ,,U...,I.1 	 1 	 1 	 	 M 	
                                             400

-------
N>
                                 Figure 5b

            Identification of Normal  Alkanes
                     LIBRARY SEARCH
                     08/04/8514:46:00* 18:24


                     SAMPLE: 1ULCCS57686 (8-1-85)
DATA: GH057689A16 #1224
ENHANCED (10B2NOT)


UNKNOWN
BASE M/E: 57
R 1C:1638390
SAMPLE j
C25.H52
1122 3
MWT352 :
BPK 43 -i
RANK 1 :
IN 25265 :
PUR 766 •
C35.H72
1122 a
MWT492 :
BPK 57 -i
RANK 2 i
IN 29645 :
PUR 760 -.
C21.H44
1122 3
MWT296 :
BPK 43 J
RANK 3 :
IN 21792 :
PUR 756-


,,.|l

J
T.i.ffl
50
lexp = 450.61 |
I - nil MI'* 	 iiiimn tw. '. i,....-.f !•-".•: ;!'
I
r JlL Jl L \\ w ..L ..... ....
PENTACOSANE CAS# 629-99-2
lref = 400.45 j
ll h ii ii i I . .
PENTATRIACONTANE CAS# 630-07-9
lref = 538.06 j
( | n ,i ...
HENEICOSANE CAS# 629-94-7
lref = 347.42 j
i II i
i i il il ,, . . ..
100 150 200 250 300 350 400

-------
                      Figure 5c

Identification of Normal  Alkanes
         LIBRARY SEARCH
         08/04/8514:46:00+ 19:08


         SAMPLE: 1ULCCW7686 (8-1-85)
DATA: GH057686A16* 1237
ENHANCED (108 2NOT)


UNKNOWN
   BASEM/E:57
   RIC:1138680
1052 q
SAMPLE J
!
C35.H72
1052
MWT352 i
BPK 43-:
RANK 2 i
IN 25265
PUR 790-
C25.H52
1052 a
MWT352
BPK 43-
RANK 2 •
IN 25265 =
PUR 790^
C22.H46
1052 3
M WT310
BPK 57-
RANK 3
INI 22753
PUR 784-



,,,|l
Miff

||


||


|,
1 1 1 | 1


Jl


||


||


.nil


lexp = 470.45 |
J „ 	 , 	
PENTATRIACONTANE CAS# 630-07-9
lref = 538.06 1
	 1

PENTACOSANE CAS* 629-99-2
lref = 400.45 |
|i ii ,i ...... r .
1'" ' 	 | 	 |lllllll IHIIJ 	 Ijllll.l 1 III II 	 	 |ll 11 	 ,.1111111.1,11111111111,111111111
DOCOSANE CASK 629-97-0
lref = 361.53 |
:-:•:•:• :-:.:-:.:':.:':.:':.:.:.:.:-:.:->:':-:.:.:.r.>:.:.:.:-:-:.:-:.:.:.:.:.;.:.:.
\ '' '' '1 I ' v i • -i • v r r

      50
           100
                150
                     200
                          250
      300
350
                                         400
                                              450

-------
ro
                                  Figure 5d

             Identification of Normal Alkanes
                      LIBRARY SEARCH
                      08/04/8514:46:00 + 20:01


                      SAMPLE: 1 ULCC*57686 (8-1-85)
DATA: GH057686A16* 1332
ENHANCED (10B2NOT)


UNKNOWN
BASEM/E.-57
RIC: 683007
SAMPLE j
C22.H4S
1146 a
UWT310
BPK 57-
RANK 1
IN 22753
PUR 751
C21.H44
1146 q
MWT296:
B PK 43-:
RANK 2 •
IN 21792
PUR 747:
C25.H52
1146 3
MWT352 :
BPK 43 H
RANK 3 i
IN 25265 :
PUR 742:
1, ,|l

L
ir...ff
II
Jjl
||
,,,ll
II
ooc
m-l
HEN
ml
PEN
1
lexp = 494.33 1
TilTi nltyiiifVI'ii itlii i VTJ t V n n 1 iVn i ifi ill Tun il il il nil i nil 1 1 il n 	 nil
OSANE CASI 629-97-0
lref = 361.53 j
*
i I fti niMi i i i i^rrrTit ' •< V i f |"| ii i I't i I I i*f I I I |T| I I 1 fl 1 1 1 1 1 1 I J 1 1 1 1 1 1 1 li 	 |] I I I 11 	
EICOSANE CAS# 629-94-7
lref = 347.42 1

FACOSANE CAS# 629-99-2
lref = 400.45 !|
. . .I.L.ll ... IL,,.. ,1. . , .!„...ซ 	 ซ„...ซ... H._..W. .. . > 	 4^ 	 ป 	
                   50
                        100
                             150
                                  200
                                       250
                                            300
                                                 350

-------
Alkane
n-C27
n-C28
n-C29
n-C30
n-C31
n-C32
n-C33
"ref
425.51
437.68
448.93
460.36
471 .96
484.94
499.88
'exp

433.60
450.61

470.45

494.33
Spectrum

5a
5b

5c

5d
Bias

+4.08
+1.68

+1.51

+5.55
11-231

-------
74        HIGH EFFICIENCY GPC CLEANUP OF ENVIRONMENTAL SAMPLES -
                                 COLUMN OPTIMIZATION

         Gary  J. Fallick. Richard  Cotter,  Waters  Chromatography  Division,  Millipore
         Corporation, 34 Maple  Street,  Milford, Massachusetts 01757; Russell  Foster,
         Richard L.  Wellman, Resource Analysts  Inc.,  P.O.Box  778,  Hampton,  New
         Hampshire 03842


         ABSTRACT

         Gel permeation chromatography (GPC) has been used for almost two decades to
         remove  unwanted  high  molecular weight  compounds from  agricultural  and
         environmental samples prior to final analysis.  Virtually  all  of this work has been
         done with  25mm ID x 40 to 100cm long columns packed with low efficiency 37-75
         micron particles.

         Using smaller diameter column packing particles produces major gains in column
         efficiencies, enabling the same separation to be done with much smaller columns.
         This enables the cleanup to be done in significantly less time using considerably less
         solvent.

         This study was done to determine the optimum grade of GPC packing material and
         preferred column dimensions for environmental sample  cleanup. High efficiency
         100A material packed in 19mm x 30cm and 19mm x 15cm columns, as well as the
         two in series, has performed the cleanup effectively while operating with less solvent
         and greater throughput than the traditional column.   Injecting 5ml  samples
         containing over 310mg of material did not overload the two column set.

         INTRODUCTION

         GPC  clean  up  of environmental  samples  is now mandatory for preparing
         semivolatile  and pesticide Superfund samples according to  the EPA Contract
         Laboratory Program Statement of Work.  Laboratories which participate in the
         EPA Contract Laboratory Program and those which follow CLP protocols, doing
         "CLP-like" work, must use GPC.

         Another aspect of environmental testing is concern about the quantities of solvents
         which are  routinely consumed in environmental cleanup and testing procedures.
         Recently a major refinement of the traditional GPC cleanup procedure produced
         over 70% reduction in solvent usage and 69% reduction in sample processing time1,
         Table 1.  This was  accomplished by substituting a  Waters Ultrastyragel high
         efficiency,  high  resolution GPC column for the low resolution column which has
         been used in the method for almost two decades.

         This high  resolution  column was chosen  based  on  general properties  and
         requirements of the method.   Since the work began, a  revised  set of calibration
         requirements were issued for the GPC cleanup method, prompting a formal column
         optimization study.  This study was concerned with two main variables - grade of
         column packing  and column dimensions.
                                           11-232

-------
REVISED CALIBRATION REQUIREMENTS

The  original  calibration mix contained  corn oil,  bis(2-ethylhexyl)phthalate and
pentachlorophenol (PCP). The basic requirement of the column was that it provide
85% or better resolution between the corn oil and the phthalate.  The com oil
represents the low volatility, high molecular weight material which is removed from
the sample before  analysis. The work  reported  by Bumgarner with the high
efficiency column met these requirements, Figure 1.

The revised calibration mix still contained corn oil and phthalate but Methoxychlor,
Perylene and Sulfur were added  in place of PCP.  Resolution of 85% or better
among each pair of compounds was also specified.

PACKING MATERIAL OPTIMIZATION

The  initial high resolution  columns studied  contained a high efficiency packing
material with an exclusion rating of 500A.   To determine whether the 500A or the
corresponding 100A material is better suited for GPC  cleanup work, the relative
retention profiles of each were compared to the Bio-BeadsK packing traditionally
used in this method.

Relative retention was defined as the ratio of the retention volume of a calibration
compound to the retention volume of sulfur. As indicated in Figure 2, the 100A
packing behaved essentially identically to the Bio-Beads.  This packing grade was
used during subsequent optimization of column dimensions.

COLUMN OPTIMIZATION - RESOLUTION

A 19mm ID  x 30cm  long column was used originally. It easily  met the initial
calibration requirements  and the revised requirements for all calibration pairs
except phthalate and Methoxychlor. To achieve the required level of resolution for
all pairs, the column length was increased by using a 19mm ID x 15cm long segment
in series with the original 19mm x 30cm column, Figure 3.

COLUMN OPTIMIZATION - MASS LOADING

As indicated in Figure 3, the configuration of a 15cm and 30cm column in series
more  than  satisfies  the  calibration  requirements,  with  the  phthalate  and
Methoxychlor  almost  baseline resolved.   The concentration  of  the calibration
markers has been increased by 2.5 times to provide the same total mass on column
as would have been loaded with a 5ml injection containing the concentrations listed
in EPA Method 3640A.

To further demonstrate the resolving power and loading capacity of this preferred
column set, a 2ml aliquot of the collected fraction was reinjected into the GPC
system, Figure  4. It is estimated  that this sample contains about 8-10%  of the
original mass of the collected peaks. The  outstanding capacity of the column is
shown again by the similarity in resolution of this diluted fraction and the original
2ml injection.
                                   11-233

-------
COLUMN OPTIMIZATION - VOLUME LOADING

To demonstrate volume loading capacity on the high efficiency column set, a 5ml
sample containing about 3l5mg was injected, Figure 5. The broadened peak shapes
are a consequence of the larger injection volume, resulting in more peak overlap.
Each peak was collected as a separate fraction with the valley between peaks taken
as the cut points.

About  10% of the volume of  each fraction was reinjected as a separate fraction,
Figure  6.  Although the detector sensed  what  appeared to be considerable peak
overlap in the initial injection, rerunning the individual collected fractions shows
excellent resolution, well above the calibration requirements. In all of the loading
considerations, mass or volume, the actual capacity of the high efficiency column set
surpasses the requirements of method.

SPEED. LOADING. RESOLUTION

The  availability of 15 and 30cm column sections which can be used alone or in series
provides maximum flexibility for cleanup of environmental samples. Used together
they provide maximum loading capacity and resolution. With lightly contaminated,
low  concentration samples the 30cm  length may be used alone for maximum
throughput and solvent  economy.  In either case,  the  high resolution of these
columns provides significant gains in operating effectiveness with corresponding
reductions in solvent usage versus the low resolution column traditionally used for
GPC cleanup.

SUMMARY

High resolution GPC columns have been shown to meet the resolution criteria of
EPA Method 3640A and the EPA Contract Laboratory Program Statement of Work
for Organics  Analysis. They provide major savings in solvent use and processing
time relative  to  the low  efficiency columns traditionally  used in  this work.
Maximum  resolution and  loading  capacity  for performing  GPC  cleanup  of
environmental samples with these high resolution columns is achieved with a 19mm
ID x 30cm column in series with a 19mm ID x 15 cm column.

NOTE

Bio-Beads is a registered trademark of Bio-Rad  Laboratories.

REFERENCE

1. Bumgarner Jr., J., International GPC Symposium Proceedings, Boston, MA, 1989,
Waters Chromatography Division, Millipore Corp., pp.787-793,1989.
                                   H-234

-------
                                               Table 1
                         GPC CLEANUP  OF  ENVIRONMENTAL  SAMPLES

                       COMPARISON OF Low RESOLUTION COLUMNS vs 500A ULTRASTYRAGEL*


                                      TYPICAL RUN TIMES  (MINUTES)

                             DUMP      COLLECT   WASH      TOTAL
ro          Low RESOLUTION      30        36       15         81
<ซ          ULTRASTYRAGEL       12.5      8.5      4         25
                                TYPICAL METHYLENE CHLORIDE VOLUMES  (ML)

                             DUMP     COLLECT   WASH      TOTAL
           Low RESOLUTION     150        180      75        405
           ULTRASTYRAGEL       56.25      38.25   18        112.5
                    ^BUMGARNER JR,  J.f OCTOBER 1989  GPC SYMPOSIUM

-------
                            Figure 1


     ULTRASTYRAGEL COLUMN RUN ON WATERS SYSTEM
        SOURCE: BUMGARNER GPC SYMPOSIUM PAPER, 10/89


GPC Calibration Standard
USEPA 2/88 SOW

22.5 minute run time
4.5 ml/minute flow rate
                           Pentachlorophenol
                    Bis (2-Ethylhexyl)
                       Phthalate
                 10   12.5
               Minutes
                            11-236

-------
                                          Figure 2

                HIGH EFFICIENCY GPC CLEANUP  OF ENVIRONMENTAL SAMPLES
                                    COLUMN OPTIMIZATION
                   COMPARISON  OF GPC PACKINGS  BY RELATIVE RETENTION
ro
00
              O 100 A MATERS
              • 500 A MATERS
              • BIO-BEADS™ S-X3
                                                 t

                                                 9.9
                                                                                •.*

                                                                                0.5

                                                                                9.4
                               COMPARISON OF GPC PACKING MATERIALS
                                USING RELATIVE RETENTION TIMES TO
                                SULFUR AS A TOTAL VOLUME MARKER
                  COBN OIL
BISO.ETHYLHEXYL)
  PH'l'HALATE
                                                   PERYLENE
                                     SULFUR

-------
                                  Figure 3
ro
u
CD
         HIGH EFFICIENCY  6PC CLEANUP OF ENVIRONMENTAL SAMPLES
                   COLUMN CALIBRATION AT METHOD LOAD
    COLUMN:  100A 19MM x  (30cM +  15cn)
    SAMPLE:  2000 UL

    SOLVENT: METHVLENE CHLORIDE
    FLOW RATE: SHL/MIN
    DETECTION: UV  9 254NM,  1.5 AUFS

    PEAK ID:
     1 CORN OIL,  62.5MG/HL
     2 BXS(2-ETHYLHEXYL)PHTHALATE, 2.5NG/HL
     3 HETHOXYCHLOR, O.SMG/ML
     4 PERYLENE,  O.OSNG/ML
     5 SULFUR,  0.2M6/HL
                                                   s
                                                   u
n
                                                                                 Ul
                       0.
                       o

                       to
                                                           COLLECT

-------
ro
to
co
          HIGH  EFFICIENCY  GPC CLEANUP OF ENVIRONMENTAL SAMPLES
                    COLUMN CALIBRATION AT METHOD LOAD

                   ALIQUOT  OF COLLECTED FRACTION REINJECTED
COLUMN:  1QOA WMM K (30cM + 15cw)
SAMPLE:  2000 UL

SOLVENT: METHYLENE CHLORIDE
FLOW RATE: SML/MIN
DETECTION: UV 9 254NH,  1.5 AUFS

PEAK ID:

  2 BzS(2-ETHYLHEXYL)PHTHALATEr  2.5HG/HL
  3 METHOXYCHLOR, O.BMG/ML
  4 PERYLENE, O.OSMG/HL
                                                     B
                                                                 lull

-------
                           Figure 5

HIGH EFFICIENCY GPC CLEANUP OF ENVIRONMENTAL SAMPLES
                  COLUMN OPTIMIZATION
                SML TEST Mix INJECTION
          COLUMN:  100A 19MM x  (30cM  + 15cM)
 1900.000
   .000
 •220.000
     10
4ง

-------
                             Figure 6

 HIGH EFFICIENCY GPC CLEANUP  OF ENVIRONMENTAL SAMPLES
                     COLUMN OPTIMIZATION
    REINJECTED FRACTIONS  FROM  SML TEST Mix  INJECTION
             COLUMN:  100A 19MM x  (30cM +  15cM>
Oft fiftfi          / \                             Corn Oil
28.000-j        / \                             FrQdion
   mV
 -7.
OR nnn               /1                        Phthalate
28-ฐฐ9l             / \                         Fraction
   mv
 -7.
28.000-
   mV
 -7.000-
                                              Rerylene
                                              Fraction
28.000-
   mV
 -7.000-
                                          4-Nitrophenol
                                              Fraction
on nnn J                                  / \      Sulfur
28-ฐฐ.ฐH                               A/ \    Fraction
   mv
 -7.000
      10    1*5    S    S    35    3ง    45    S
                        Minutes            Press Resume

-------
75
EFFICIENT AQUEOUS EXTRACTION USING AN EMULSION PHASE CONTACTOR
         Kevin  P. Kellv.  Ph.D., Manager, Applications; Loren C. Schrier,
         Applications  Chemist; Kenneth C. Kuo, Senior Research Chemist, ABC
         Laboratories  Inc.,  P. O. Box 1097, Columbia, Missouri  65205
        ABSTRACT

        The  Emulsion Phase Contactor (EPC) promotes the efficient liquid/liquid
        extraction  of aqueous samples with organic  solvents.   Use of transient
        high intensity  electric fields in  the  liquid/liquid extraction process
        produces  increased interfacial surface area, resulting  in more efficient
        mass transfer.   Transient fields also induce droplet motion  (distortion
        and  translation) and droplet coalescence, which promote phase transfer and
        phase separation,  respectively.

        In the EPC  technique, aqueous sample is introduced into a cell, where  a
        voltage is  applied to disperse the  aqueous phase into the bulk  (organic)
        phase.  Mass transfer is facilitated by formation of  small  (micron-sized)
        droplets.  Contact time of the droplets is increased through the  action of
        additional  charged plates,  which  also aid in the eventual aggregation of
        dispersed droplets.

        EPC  data  are presented which show excellent methylene chloride extraction
        recoveries  for  many priority pollutants, such  as would  be analyzed using
        EPA  SW-846,  Method 8270 (semivolatile organics by GC/MS), or  the EPA CLP
        (Contract Lab Program)  Statement  of Work (SOW).

        The  EPC extraction method  is suitable  for  the automation of analytical
        laboratory  sample  preparation,  and it replaces the more labor intensive
        extraction  methods, such as  separatory  funnel extraction,  or classical
        liquid/liquid extraction  with boiling methylene chloride.

        Instrumentation to accomplish the goal of an automated  EPC extraction is
        currently under development at ABC Laboratories.  Analytical applications
        of EPC derive from technology transfer under U.S.  Patent No. 4,767,515,
        "Surface  Area Generation  and Droplet Size  Control in Solvent Extraction
        Systems Utilizing High Intensity Electric Fields", issued August  30, 1988,
        which was developed by  Scott and  Wham at Oak Ridge National Laboratory.
         INTRODUCTION

         Traditional methods for the liquid/liquid extraction of most environmental
         samples are both labor and solvent intensive,  and have high potential for
         exposure of laboratory workers to solvents and other hazardous substances.
         Recent alternatives,  such as solid-phase extraction  (SPE) cartridges or
         discs, may not cope well with difficult aqueous matrices.  Improvement of
         ordinary liquid/liquid extraction techniques by automating fluid transfers
                                          11-242

-------
and using electrically enhanced mass transfer' provides opportunities for
superior analyte recoveries and more consistent analytical results, while
reducing solvent (and energy) consumption and decreasing the opportunities
for exposure of personnel to hazards.

Efficient mass transfer during liquid/liquid extraction depends directly
upon the availability of surface area for mass transport.  Since organic
solvents tend to be essentially nonconductive,  charged metal electrodes
mounted within a solvent-containing region can be used to polarize water
droplets, which leads  to droplet shape distortion, rotation, translation,
and breakage (Figure 1).  This greatly increases phase transfer kinetics
for extraction of organic analytes from the aqueous phase.

A device employing this principle for extraction of  aqueous samples is
diagrammed in Figure 2.  Aqueous sample is pumped into the bottom of the
solvent-filled chamber, where an electric field causes droplet disruption.
A second electric field,  situated above the first, induces further droplet
motion and also aids coagulation of daughter droplets formed by the action
of the first  field.  Accumulating volume of aggregate aqueous phase (above
the two electric fields)  overflows to another container.  Thus the organic
solvent is in equilibrium with only a small  portion  of the aqueous sample
at any time.   This contributes to excellent extraction efficiencies.

Efficiency  of  the  EPC  is  demonstrated  by   extraction  of  various
environmental pollutants with methylene chloride, followed by analysis of
the extract using GC/FID or LC/UV detection.
RESULTS AND DISCUSSION

One liter of simulated field water2 was spiked with 2.50 mL of a methanol
solution that contained  1000  pg/mL  each  of priority pollutant molecules
(analytes  were  divided  into  two groups to  facilitate  chromatographic
analysis).  The EPC  extraction  cell  was  charged with methylene chloride
(approximately 400 mL), and plate voltages were  set at 15 kV (lower pair)
and 10 kV  (upper pair).  Aqueous sample was introduced into the EPC at a
rate of 22 mL/min.  After all  the sample  had traversed the EPC extraction
chamber, the methylene chloride layer was drained, and one half was dried
with sodium sulfate,  then concentrated to 10 mL (to minimize evaporative
losses) using a steam bath with the Kuderna-Danish apparatus.

Chromatographic data  (FID) were obtained  using  an HP  5890  Series  II
temperature programmed gas chromatograph  (Hewlett-Packard)  plus 3396-A
integrator, with one of  the following two columns:  a 30  meter x 0.25 mm
capillary  (Supelco DB-5, 0.25 jim film thickness); or a 5 meter x 0.53 mm
     JScott,  Timothy C.; Wham, Robert M.;  Ind. Enq. Chem. Res.  1989 28, 94.

     Prepared by adding 24 mg KC1,  608 mg MgCl2 hexahydrate,  344 mg CaCl2
dihydrate, and 404 mg NaCl  to one gallon of reagent water.
                                  11-243

-------
        Figure 1.  Electric Fields in Extraction
DROP FORMATION
DROP OSCILLATIONS
DROP BREAKUP
                                       O o
DROP-DROP
INTERACTIONS
& COALESCENCE
o    o
 oo
                    11-244

-------
FIGURE 2. EPC EXTRACTION CELL

                 WATER OUT

                 METHYLENE CHLORIDE IN
                                    z. /WATER
                               PHASE BOUNDARY
                                10 KV
                               ELECTRIC FIELD
                               S.S. ELECTRODE
                                15 KV
                               ELECTRIC FIELD
                               TEFLON BODY
                               WATER  IN
                 METHYLENE CHLORIDE OUT
             11-245

-------
capillary  (HP-1, 2.65  pirn film thickness).  The recovery  data (Table 1)
were developed by comparison of peak heights to calibrations with external
standards.  Recoveries were good to excellent, with excellent precision.
One of  the two  analyte  sets  was also  processed  using a  lower  spiking
concentration (250 (iL of  solution, Table 2) to verify adequate recovery at
trace levels.  The  entire extract from the low level  runs was concentrated
to 5 mL for analysis.

Calculated recoveries  used the average of  duplicate  extract injections
from each of the four extractions.  Average recoveries for ten analytes,
which represented several classes of chemical pollutants, ranged from 82%
to 109%.   The largest  standard deviation measured  was 7.1%,  indicating
that in addition to good recoveries with faster turnaround time (less than
1 hour per extraction for a 1  Liter  sample), EPC methodology furnishes an
enhancement  in precision,  in comparison with  more  operator  dependent
techniques.

Influence of pH on  the recovery of acidic analytes was  assessed by spiking
water with 1.00 mL of a solution  containing  1000 ^g/mL each of phenol, 4-
nitrophenol (4-NP), and 2,4-dinitrophenol (2,4-DNP) dissolved in methanol.
The pH of the sample  was  adjusted using  12 M HC1.  The EPC extraction was
carried out at each pH, then HPLC analysis of the extracted water, with UV
detection of  analytes, was used  to  measure  the  concentration of phenols
recovered in the extract and remaining in the water.

Data from those extractions at different pHs (Table 3) indicate that the
technique  is  efficient enough to produce reasonably  good  recoveries of
acidic compounds (pK, values for phenol,  4-NP,  and 2,4-DNP are 9.89, 7.15,
and 3.96, respectively) with no pH adjustment of samples.   Thus 2,4-DNP,
although a stronger acid than acetic acid (pK, 4.75) was recovered in 53%
yield from an unadjusted (pH 5.9) sample.

EPC development efforts are now focusing on further reduction of solvent
usage, design of particulate-tolerant extraction  devices,  and continued
improvements in sample throughput and analyte recoveries.

SUMMARY

The Emulsion Phase Contactor  (EPC) is a very efficient extraction device
that electrically enhances mass transfer during liquid/liquid extraction.
It provides good recoveries for extraction of a wide  range of semivolatile
analytes from water samples using methylene  chloride,  plus greater degree
of precision  (reproducibility) than can be obtained using less automated
techniques.  Extraction efficiency is  so high that acidic analytes can be
recovered without pH adjustment of water samples.   Additional advantages
are savings in  labor and reduced exposure of lab  personnel to hazardous
substances.  Automated  EPC equipment is  under development for application
to SW-846, CLP, and other extraction methods.
                                  1-246

-------
Table 1.  Recovery Data  (%  by GC/FID)  for Analytes at 2500
Analyte
bis ( 2-Chloroethyl ) ether
Nitrobenzene
bis ( 2 -Chloroethoxy) methane

Naphthalene
Acenaphthene
2-Fluorophenol
Aniline
1 , 4-Dichlorobenzene
Hexachloroethane
2-Fluorobiphenyl
Replicates
1
91
98
101
97
98
91
106
85
82
83
2
86
94
94
90
89
91
110
86
80
83
3
94
100
96
93
86
91
110
87
90
86
1
94
106
101
96
101
90
109
88
88
85
Statistics
X
91
100
98
94
94
91
109
86
85
84
a
3.8
5.0
3.6
3.2
7.1
0.5
1.9
1.3
4.8
1.5
                             11-247

-------
         Table 2.   Recoveries for Analytes at 250
Analyte
bis ( 2 -Chloroethy 1 ) ether
Nitrobenzene
bis < 2-Chloroethoxy ) methane
Naphthalene
Acenaphthene
i
100
105
101
98
88
2
101
103
104
99
105
     Table 3.  Influence of pH on Extraction of Phenolics
Compound
Phenol
4-Nitrophenol
2 , 4-Dinitrophenol
Recovery (%) at Indicated pH
pH = 5.9
63
59
53
pH = 3.8
80
68
75
pH = 1.6
1813
58
76
*High result due to chromatographic interference
                             1-248

-------
SFE PRACTICAL APPLICATIONS FOR ENVIRONMENTAL AND
INDUSTRIAL SAMPLES
       .T.p. MHjR. JOE TEHRANI
Isco,  Inc.,  4700  Superior  Street,  Lincoln,  NE,
68504

EADLK. MEQKV
HWS Technologies, Inc.,  825  J Street,  Lincoln, NE,
68501

Supercritical fluid extraction (SFE)  is a technique
that  has  the  potential   to   revolutionize
conventional  methods  of  sample extraction  and
analysis.  When  003  is used,  SFE is  an exceedingly
quick, inexpensive, and environmentally safe method
of  sample preparation for GC,  HPLC,  UV-VIS, and
TLC.   SFE can also be used for percent extractable
determinations in foods and polymers.

Liquid  extractions   of  analytes  from  complex
matrices  often  require  labor  intensive and  time
consuming methods  such as Soxhlet extraction.  The
matrices  must be  extracted  with large  volumes  of
environmentally  hazardous solvents  which must  be
evaporated to the atmosphere or otherwise disposed
of.

This   presentation  will  discuss   practical   SFE
applications  for  a variety  of  environmental  and
industrial samples  such as  soils,  polymers,  and
foods.   SFE extraction  data will be presented and
comparisons  will   be  drawn  between  SFE   and
conventional methods of  extraction.
                                  1-249

-------
INORGANICS

-------
Microwave Sample Preparation Methods For Environmental
Analysis

H. M. Kingston, F. A. Settle*, M. A. Pleva**, Lois Jassie***, P. Walter, Jim
Petersen, and Bill Buote****, National Institute of Standards and Technology,
Center for Analytical Chemistry, Inorganic Analytical Research Division,
Gaithersburg, MD 20899.
*Dept. of Chemistry, Virginia Military Institute, Lexington, Virginia  24450
  Dept. of Chemistry, Washington and Lee University, Lexington, VA 24450
  Research Associate sponsored by CEM Corporation
   Zymark  Corporation
**•
***•
****
Microwave sample preparation is gaining a wide degree of acceptance and is
being applied in EPA methods for elemental analysis.  The nature of standard
procedures requires that they be readily transferable and reproducible between
laboratories^.  Because microwave sample preparation procedures are
quantifiable their reproducibility provides an opportunity to improve data
quality. The control and standardization of microwave methods is a matter
that must be examined.  Calibration of laboratory microwave equipment is
required to transfer the methods accurately and precisely. Calibration
methods have been evaluated and the error that can be expected in
transferring these procedures  has been determined.

Microwave sample preparation methods provide a platform to produce
procedures that can be used generally for sample preparation in many
environmental elemental analyses.  Currently, both the RCRA and CERCLA
programs share two microwave methods, developed cooperatively, applicable
for soils, sediments, sludges, oils, and waters.  These methods demonstrate
robust microwave procedures that improve precision and are more efficient
than many classical methods.  These attributes arise from the direct control of
energy transfer, and the mechanisms involved in that transfer.  Reaction
temperatures and their profile control the mineral acid reactions that are
necessary to release the elements for analysis.  The mechanisms of the energy
transfer and the relationships  that control these reactions  will be discussed
relevant to the two current EPA methods.

In addition to  manual methods, an automated microwave decomposition
system has been developed using a modular design^.  Three separate portions
of the automated microwave sample preparation system will be described  as
well  as a prototype quasi expert system that has been developed to assist in
standardizing procedures for microwave dissolution. A file structure has
been devised to transfer  these procedures from system to  system and provides
information adaptable to the  level of automation in different instrument
                                 11-253

-------
configurations.  Once a procedure has proven successful, it is stored for future
reference and can be sent to other systems where it will automatically
reproduce the method. Thus, the system has the ability to "characterize" and
"transfer" procedures.  A quality control sample log-in system was developed
that guides the analyst through data entry, weighing and barcoding of the
sample, and creates a sample data file. A computer-controlled microwave
unit has been developed for use with this system. The entire system is
robotidy integrated and is being tested as the first component of a fully
integrated inorganic analysis system.  Microwave sample preparation stations
following this basic design have been constructed for EPA EMSL-LV and EPA
Region 10 using commercial  versions of the research system.  EPA and NIST
are coordinating the testing and evaluation of the systems.

Reference
1. Binstock, David A., Grohse, Peter M., Gaskill, Alvia Jr., Kingston, H. M., and
Jassie, L. B., "Development and Validation of a Method for Determining Elements
in Solid Waste Utilizing Microwave Digestion", JOAC, 74,2,1991.

2. Walter, P., Kingston, H. M., Settle, R A., Pleva, M.  A., Buote, W., and
Chrosto, J., "Automated Intelligent Control of Microwave Sample
Preparation," in Advances in Laboratory Automation and Robotics 1990, eds.
Stramaitis and Hawk, vol. 7, Zymark Corporation, 1991.
                                  11-254

-------
70         COMPARISON OF PROCEDURES  FOR TCLP  EXTRACT DIGESTION;
/O                      CONVENTIONAL VS.  MICROWAVE

      V.  L. Verma,  PhD, Supervisor,  and T.  M.  McKee,  Director,
      Browning-Ferris Industries  Houston  Laboratory,  5630  Guhn
      Road,  Houston,  Texas  77040

      ABSTRACT

      This  paper  describes a  study that  compares and  evaluates
      metal data  resulting  from the  analyses  of replicate Toxicity
      Characteristic  Leaching Procedure  (TCLP) extracts  subjected
      to  both  the  metal  acid digestions  methods  specified  in
      SW-846 and  microwave digestion  procedures.   The  TCLP  ex-
      tracts analyzed by Inductively  Coupled Argon Plasma  (ICP)
      Spectroscopy  for   the  metals   arsenic,  barium,   cadmium,
      chromium,  lead, selenium and  silver  were  prepared by a CEM
      Model MDS  8 ID  closed vessel  microwave digestion and  EPA
      SW-846 Methods  3010 and 3020.

      INTRODUCTION

      Method 3010,  "Acid Digestion of Aqueous Samples and Extracts
      for   Total  Metals  for  Analysis  by  ICP  Spectroscopy"  and
      Method 3020,  "Acid Digestion of Aqueous Samples and Extracts
      for  Total Metals  for Analysis by Furnace Atomic Absorption
      Spectroscopy"  found  in Test  Methods  for  Evaluating  Solid
      Waste.   Physical/Chemical  Methods.  November   1986,   Third
      Edition,  USEPA, SW-846  are commonly  applied techniques  to
      digest metals from aqueous  samples in an open vessel.

      As evidenced by the numerous studies  reported in the litera-
      ture,  closed vessel  microwave  acid  digestion is  receiving
      considerable attention as a  state-of-the-art metal digestion
      technique.   Microwave heating was first  reported to  speed
      digestion of  samples by acids  fifteen years ago(!).   The
      technique has been used in a  variety of sample preparations
      since then  (!~12)  and is rapidly  gaining  recognition  as  a
      useful tool in analytical chemistry  (13).   Systems designed
      specifically for laboratory microwave  digestion  are commer-
      cially available.   These  systems  are  designed  to  overcome
      deficiencies identified  by  researchers  (4,5,9,11,12)   wno
      performed their initial work  with microwave  ovens manufac-
      tured for domestic use.  Recently, advanced  techniques  have
      become  available   for  doing  acid digestion of  metals  in
      closed TFE  vessels by microwave heating (14) .

      Browning-Ferris Industries  Laboratory in   Houston,   Texas
      analyzes 150 to 200 Toxicity  Characteristic  Leaching  Proce-
      dure  (TCLP)  metal  samples per month with a goal  of from two
      to  ten  days  turnaround time,   including  the  long  manual
                                  11-255

-------
digestion needed  for samples  by  the traditional  hot plate
techniques prescribed in SW-846.   In order to reduce costs,
turnaround time,  and to improve  the sample  handling tech-
nique while maintaining QA/QC  and laboratory  safety, it was
decided to investigate the microwave digestion technique.

To accomplish this task, three different TCLP waste extracts
of  both extraction  fluid  number  one and  extraction fluid
number  2  were  analyzed for metals.   The waste descriptions
were as follows:
     -  Soil contaminated with gasoline, diesel and heating
          oil,
        Soil contaminated with hydraulic and diesel oil,
     -  Filter press sludge,
     -  Digested domestic sewage sludge,
     -  Reactor rake-out residue  (magnesium chloride
          production),
        Sand/urea
All  of the  Toxicity Characteristic (TC)  metals,  arsenic,
barium, cadmium,  chromium,  lead,  selenium  and silver,  were
at  concentrations  below their  appropriate detection  levels.
These  six extracts were each divided into three samples and
spiked  with different concentrations each of  the TC metals,
except  for  silver.   Each  of  the  samples was  then  further
divided into two equal portions.  One portion was split into
four 50 milliliter (mL) aliquot and microwave digestion was
performed on each aliquot  in a closed  TFE vessel  with 3 mL
of  concentrated nitric  acid and 2 mL of concentrated hydro-
chloric acid for  40 minutes at 90%  power (515 Watts) .  The
other  portions  of the spiked  extracts  were also subdivided
into four aliquots, and each subjected to the hot plate open
vessel  acid digestion  procedure  as  prescribed  by method
3010.

The  same  procedure  was   repeated  for  the  TCLP  extracts
(extraction  fluid number  one   and  extraction  fluid number
two) spiked  with  silver at  three different concentrations.
Four  50  mL  aliquots  of   each were  microwave  digested in
closed  vessels with 5 mL of nitric acid.  Four portions each
of the  same spiked samples were also digested by the  conven-
tional  hot  plate acid  digestion  procedure,  Method 3020.
Method  3020, utilizing only nitric  acid, is used by our lab
for silver digestion to avoid  silver chloride precipitation.

All samples were analyzed by ICP spectroscopy.

Comparison of the data  reveals the accuracy,  applicability,
and performance efficiency of  each technique.
                            11-256

-------
SUMMARY
Results of the study suggest  that  the Methods 3010 and 3020
can effectively be substituted  by  the microwave acid diges-
tion procedures utilized.

DISCUSSION

Apparatus
     Microwave Digestion System, CEM Model MDS-81D
     Closed  TFE  digestion vessels  and turntable,  CEM Part
No.
          600050
     Capping Station, CEM Part No.  920030
     Hot plates
     Griffin beakers
     Watch glasses, ribbed
Reagents
     Hydrochloric acid, concentrated, spectrograde
     Nitric acid, concentrated, spectrograde
     Water, deionized

Methodology  - Six  different  Toxicity Characteristic (TC)
samples,  three utilizing  extraction  fluid  number  one and
three with extraction  fluid  number two,  were spike with the
following  three  concentrations   of  the  metals  arsenic,
barium, cadmium, chromium, lead, selenium, and silver:
     Concentration  level  1  -  Maximum  regulatory   limit,
except for barium which was spiked at 5.00 ppm.
     Concentration  level  2  -  Mid-range  of  the  maximum
regulatory limit, except  for  barium which was spike at 2.50
ppm.
     Concentration  level  3   -  Five  times   the  respective
detection levels of each metal.

The spiked  solutions were then divided  into two equal por-
tions.  To  determine accuracy, each portion was subdivided
into eight aliquots  and  four were  digested by the microwave
method, and  four by the conventional hot  plate methods.

All samples  were analyzed by  ICP  Spectrometry under  SW-846
Method 6010  at the following wave  lengths:
     arsenic   189       lead    220
     selenium  196       barium  233
     chromium  205       silver  328
     cadmium  214

Microwave digestion method [3010X] -
o  Transfer  a  50 mL  aliquot  of a well mixed sample to  a TFE
digestion  vessel.   Add  3 mL  of  concentrated  and  2  mL of
concentrated hydrochloric acid.   Place  the  safety  pressure
relief  valve  on  the  vessels  and then  cap to  12 ft.lbs.
torque using the capping station.
                             1-257

-------
o  Weigh the vessel and record its weight in grains. Place it
in the MDS-81D turntable and attach the venting tube.
o  Repeat  the above  step until  the turntable  contains  12
vessels.   It  is  critical  that  the  total volume  of  the
solution equals 660 mL during digestion and that each vessel
contains an equal  volume of  acid.   This  is  necessary  to
ensure uniform heating of all vessel solutions.
o  Turn  the microwave  unit  exhaust on  to the  maximum fan
speed.   Activate  the  turntable   so that  it  is  rotating
continuously.
o  Program  the  instrument time for 40 minutes and power  at
90% (515 watts).  Depress the  start  key  and heat the sample
mixture for the programmed time.
o  At the end of  the  digestion period,  remove  the turntable
from the microwave  unit  and  allow the  sample solutions  to
cool  to  room  temperature.   Shake  the  vessels  to  mix the
sample solutions.   Detach each venting  tube and remove the
vessels from the turntable.
o  Weigh and record the  weight of the  cooled vessel  after
digestion.   If there is a weight loss greater than 0.5 grams
from that recorded prior to digestion,  add DI water equal to
the weight  lost.   If there  is a  significant  weight  loss
(e.g., two  to  three grams),  one should  discard  the sample,
and repeat the digestion procedure.
o  Recap the vessel using the  capping  station  and shake the
vessel to mix the sample solution.
o  Open  the vessels  and filter the  samples to  remove any
insoluble materials  if necessary.   Do  not rinse  or dilute
the digested sample.

Microwave digestion method [3020X] -
This procedure is identical to the digestion method [3010X]
described  above,  except  that  instead  of  adding  3 mL  of
concentrated nitric acid and 2 mL of concentrated hydrochlo-
ric acid to the  sample to be digested,  add  5  mL of concen-
trated nitric acid.

The hot plate  digestion methods utilized in this  study are
those described  in SW-846,  method  3010 and  3020.   Method
3020  is  utilized  for samples  to be  analyzed  by  ICP for
silver.

RESULTS

Table I  contains the results  of  the study.   The  data  is
divided  into  two  parts  to  distinguish  between  the  two
different  TCLP extraction  fluids  examined.   The  type  of
digestion employed  is identified for each  extract media  at
the top of  the  table.  The left margin lists the metals and
their  spike  concentrations.    The  table  also  lists  the
concentrations of four replicates and percent recovery.
                            11-258

-------
Table II  illustrates  the average concentration  of  the four
replicates, and shows the  relative  percent deviation of the
two  values obtained  by microwave  and  hot  plate  digestion
methods.

CONCLUSION

The  intent  of  the study was to  evaluate  the data resulting
from the  metal analyses of  the two TCLP  extract solutions
digested  by both microwave  and the conventional hot plate
methods.

The data reveal that percent recoveries obtained from micro-
wave digested  samples  and  hot plate acid digested samples
are  approximately equivalent.   In  some instances,  the lead
and  barium  recoveries  are  low,   probably due  to  matrix
interferences, but overall, the recoveries  obtained by means
of the two methods are  quite comparable.

The  relative percent  deviation (Table II) shows  that hot
plate  3010,  3020  methods  can be easily substituted  by
microwave acid digestion method 3010X and  3020X.

The  advantages of microwave digestion  procedures over those
of conventional hot plated methods  are: i)  it is a rapid and
safe way  of  preparing  samples  for ICP  analysis,   ii)  the
acids  do  not  evaporate from  the   closed  container causing
elevated  concentrations of trace acid  impurities,  iii) the
digestion acids apparently do not  decompose under microwave
conditions,  iv)  there  are  no acid fumes,  v)  volatile ele-
ments  are retained in  the sample  solution, vi)  the method
requires  less  monitoring,  and finally,  vii) there is less
potential   for  external   sample  contamination.   The  only
limitation  of  this method  is  the  time-consuming assembling
and  cleaning of the digestion  vessels.

ACKNOWLEDGEMENTS

The  authors would like  to  express  their appreciation to Ms.
Linda DeLeon,  Ms. Nanette  Skal,  Ms. Jacqueline Palomino and
Mr.  Charles Jui   for their  interest and valuable assistance
in the study and  the preparation of this paper.
                            11-259

-------
                                TABLE  I
                    COMPARISON OF PERCENT  RECOVERY
                   MICROWAVE VS. HOT PLATE DIGESTION
     EXTRACTION FLUID # 1
Arsenic at 5.000 ppm
EXTRACTION FLUID # 2
MicroW % Rec. Hot Plate
4.584 92 4.313
4.598 92 4.466
4.653 93 4.284
4.640 93 4.413
Arsenic at 2.500 ppm
MicroW % Rec. Hot Plate
2.065 83 2.257
2.066 83 2.101
2.171 87 2.018
2.038 82 2.121
Arsenic at 0.500 ppm
MicroW % Rec. Hot Plate
0.417 83 0.409
0.405 81 0.420
0.420 84 0.415
0.444 89 0.457
Selenium at 1.000 ppm
MicroW % Rec. Hot Plate
0.948 95 0.891
1.107 702 0.856
0.906 91 0.904
0.991 99 0.897
Selenium at 0.500 ppm
MicroW % Rec. Hot Plate
0.521 104 0.509
0.475 95 0.477
0.509 102 0.477
0.490 98 0.497
Selenium at 0.500 ppm
MicroW % Rec. Hot Plate
0.492 98 0.404
0.450 90 0.424
0.444 89 0.430
0.459 92 0.466
Lead at 5.000 ppm
MicroW % Rec. Hot Plate
4.767 95 4.636
4.773 95 4.621
4.703 94 4.524
4.709 94 4.627
% Rec.
86
89
86
88

% Rec.
90
84
81
85

% Rec.
82
84
83
91

% Rec.
89
86
90
90

% Rec.
102
95
95
99

% Rec.
81
85
86
93

% Rec.
93
92
90
93
MicroW
4.484
4.301
4.319
4.390
MicroW
2.509
2.456
2.223
2.432
MicroW
0.417
0.405
0.420
0.509
MicroW
1.032
1.070
1.066
1.086
MicroW
0.551
0.541
0.541
0.517
MicroW
0.547
0.522
0.528
0.549
MicroW
3.958
4.022
4.080
4.126
% Rec.
90
86
86
88
% Rec.
100
98
89
97
% Rec.
83
81
84
102
% Rec.
103
107
107
1089
% Rec.
110
108
108
103
% Rec.
109
104
106
110
% Rec.
79
80
82
83
Hot Plate
4.214
4.345
4.309
4.413
Hot Plate
2.255
2.135
2.421
2.181
Hot Plate
0.409
0.420
0.504
0.501
Hot Plate
0.906
0.910
0.862
0.897
Hot Plate
0.453
0.467
0.467
0.493
Hot Plate
0.530
0.476
0.554
0.551
Hot Plate
4.000
3.915
3.937
3.902
% Rec
84
87
86
88
% Rec
90
85
97
87
% Rec
82
84
100
100
% Rec
91
91
86
90
% Rec
91
93
93
99
% Rec
106
95
111
110
% Rec
80
78
79
78
                                 11-260

-------
                          TABLE I  (CONTINUED)
                    COMPARISON OF  PERCENT RECOVERY
                   MICROWAVE VS. HOT PLATE  DIGESTION
EXTRACTION FLUID #
Lead at 2.500 ppm
EXTRACTION FLUID # 2
MicroW % Rec.
2.277 92
2.255 90
2.278 91
2.189 88
Lead at 0.500
MicroW % Rec.
0.487 97
0.508 102
0.502 100
0.515 103
Chromium at 5.
MicroW % Rec.
4.874 97
4.844 97
4.862 97
4.844 97
Chromium at 2.
MicroW % Rec.
2.401 96
2.394 96
2.417 97
2.354 94
Chromium at 0.
MicroW % Rec.
0.283 113
0.264 106
0.257 103
0.237 95
Hot Plate
2.273
2.222
2.212
2.205
ppm
Hot Plate
0.506
0.473
0.451
0.477
000 ppm
Hot Plate
4.438
4.438
4.476
4.325
500 ppm
Hot Plate
2.210
2.170
2.187
2.240
250 ppm
Hot Plate
0.280
0.261
0.277
0.275
% Rec.
91
89
88
88

% Rec.
101
95
90
95

% Rec.
89
89
90
87

% Rec.
88
87
87
90

% Rec.
112
104
111
110
Barium at 5.000 ppm
MicroW % Rec.
4.618 92
4.466 89
4.519 90
4.560 91
Hot Plate
4.573
4.642
4.488
4.627
% Rec.
91
92
90
92
Barium at 2.500 ppm
MicroW % Rec.
2.375 95
2.355 94
2.397 96
2.261 90
Hot Plate
2.346
2.391
2.379
2.345
% Rec.
94
96
95
94
MicroW
1.914
1.872
1.918
1.939
MicroW
0.487
0.508
0.501
0.510
MicroW
4.488
4.600
4.508
5.035
MicroW
2.262
2.463
2.490
2.476
MicroW
0.283
0.264
0.257
0.238
MicroW
4.412
4.480
4.473
4.464
MicroW
2.356
2.328
2.365
2.360
% Rec.
77
75
77
78
% Rec.
97
102
100
102
% Rec.
90
92
90
100
% Rec.
90
98
99
99
% Rec.
113
1.06
103
95
% Rec.
88
90
89
89
% Rec.
94
93
95
94
Hot Plate
1.964
2.067
1.846
1.946
Hot Plate
0.506
0.473
0.451
0.477
Hot Plate
4.924
5.085
4.427
4.215
Hot Plate
2.100
2.228
2.289
2.218
Hot Plate
0.261
0.277
0.275
0.280
Hot Plate
4.287
4.403
4.296
4.295
Hot Plate
2.246
2.253
2.371
2.186
% Rec.
79
83
74
78
% Rec.
101
95
90
95
% Rec.
98
102
89
84
% Rec.
84
89
92
89
% Rec.
104
111
110
112
% Rec.
86
88
86
86
% Rec.
90
90
95
87
                                 1-261

-------
                          TABLE  I  (CONTINUED)
                    COMPARISON OF  PERCENT  RECOVERY
                   MICROWAVE VS. HOT PLATE DIGESTION
EXTRACTION FLUID # 1
Barium at 0.100 ppm
Microw  % Rec.  Hot Plate
 0.094    94      0.095
 0.094    94      0.092
 0.095    95      0.087
 0.095    95      0.099

Cadmium at 1.000 ppm
Microw  % Rec.  Hot Plate
 0.946    95      0.913
 0.953    95      0.944
 0.943    94      0.894
 0.955    95      0.915

Cadmium at 0.500 ppm
MicroW  % Rec.  Hot Plate
 0.443    89      0.429
 0.431  0.86      0.419
 0.439    88      0.416
 0.425    85      0.415

Cadmium at 0.100 ppm
MicroW  % Rec.  Hot Plate
 0.110   110      0.104
 0.109   109      0.106
 0.107   107      0.105
 0.105   105      0.106

Silver at 5.000 ppm
MicroW  % Rec.  Hot Plate
                              EXTRACTION FLUID #  2
 5.022
 4.983
 5.065
 5.027
 100
99.6
 101
 101
4.230
4.275
4.040
4.325
Silver at 2.500 ppm
MicroW  % Rec.  Hot Plate
 2.431    97      2.200
 2.420    97      2.130
 2.430    97      2.340
 2.542   102      2.136

Silver at 0.250 ppm
MicroW  % Rec.  Hot Plate
 0.240    96      0.225
 0.239    96      0.227
 0.231    92      0.235
 0.240    96      0.223
                   % Rec.
                     91
                     94
                     89
                     91
                   % Rec,
                     86
                     84
                     83
                     83
                   % Rec.
                    104
                    106
                    105
                    106
% Rec.
  85
  86
  81
  87
                   % Rec.
                     88
                     85
                     94
                     85
                   % Rec.
                     90
                     90
                     94
                     89
MicroW
0.899
0.909
0.913
0.913
MicroW
0.829
0.853
0.845
0.848
MicroW
0.375
0.399
0.393
0.406
MicroW
0.106
0.104
0.103
0.098
MicroW
5.022
5.088
5.049
5.093
MicroW
2.407
2.591
2.604
2.453
MicroW
0.249
0.243
0.251
0.244
% Rec.
90
91
91
91
% Rec.
83
85
85
85
% Rec.
75
80
79
81
% Rec.
106
104
103
98
% Rec.
100
102
101
102
% Rec.
96
104
104
98
% Rec.
100
97
100
98
Hot Plate
0.891
0.874
0.877
0.889
Hot Plate
0.828
0.841
0.831
0.828
Hot Plate
0.380
0.386
0.399
0.380
Hot Plate
0.101
0.098
0.103
0.100
Hot Plate
4.440
4.305
4.165
4.230
Hot Plate
2.367
2.476
2.564
2.314
Hot Plate
0.223
0.206
0.223
0.215
% Rec
89
87
88
89
% Rec
83
84
83
83
% Rec
76
77
80
76
% Rec
101
98
103
100
% Rec
89
86
83
85
% Rec
95
99
103
93
% Rec
92
82
89
86
                                11-262

-------
                                      TABLE  II
                 COMPARISON  OF RELATIVE  PERCENT DEVIATION
                      MICROWAVE VS.  HOT PLATE  DIGESTION
    EXTRACTION FLUID #  1
Arsenic
  MW   H P      MW   H  P
 4.619 4.369    2.085 2.124
RPD  5.67,      1.90/.
Selenium
  MW   HP
 0.966 0.887
RPD  8.5%

Lead
  MW   HP
 4.738 4.602
RPD  2.9%

Chromium
  MW   HP
 4.856 4.419
RPD  9.4%

Barium
  MW   HP
 4.541 4.583
RPD  0.94%

Cadmium
  MW   HP
 0.949 0.917
RPD  3.4%
   MW   HP
0.499  0.488
2.2%
   MW   HP
2.250  2.228
0.98%
   MW   HP
2.391  2.202
8.2%
   MW   HP
2.347  2.365
0.76%
   MW   HP
0.435  0.420
3.5%
Silver
   MW   H P      MW   H P
 5.025 4.218  2.456  2.202
RPD 17.5%    10.9%
                             EXTRACTION FLUID # 2

                 MWHP      MWHP      MWHP       MWHP
                0.422 0.425   4.374 4.322   2.405 2.248    0.438  0.459
               0.71%         1.2%          6.7%          4.7%
  MWHP      MWHP      MWHP       MWHP
0.461 0.431   1.064 0.894   0.521  0.466    0.536  0.528
6.7%         17.4%         ll.l'/,          1.5%
  MWHP      MWHP      MWHP       MWHP
0.503 0.477   4.047 3.939   1.911  1.956   0.502  0.477
5.3%          2.7%          2.3%          5.1%
  MWHP      MWHP      MWHP       MWHP
0.260 0.273   4.659 4.663   2.423 2.210   0.261  0.273
4.88%         0.09%         9.1%          4.8%
  MWHP      MWHP      MWHP       MWHP
0.094 0.093   4.457 4.320   2.352 2.264   0.909 0.883
1.07%         3.1%          3.8%          2.9%
  MWHP      MWHP      MWHP      MWHP
0.108 0.105   0.844 0.830   0.393 0.386   0.103 0.101
2.8%          1.7%          1.8%          2.0%
                 MWHP      MWHP      MWHP      MWHP
               0.238 0.228   5.063 4.285   2.574 2.430   0.247 0.217
               4.3%         16.6%          3.4%         12.9%
                                      11-263

-------
REFERENCES;

  1.    Abu-Samra, A.; Morris,  J..S.;  Kortyohann,  S.  R.,  Anal.
       Chero., 1975, 47, pp 1475-1477.
  2.    Barrett,  P.;  Davidowski,  L.  J.,  Jr.;  Penaro, K.  W.;
       Copeland, T.R., Analy.  Chem. 1978, 50, pp 1021-1023.
  3.    Nadkarni, R. A., Anal.  Chem., 1984, 56, pp2233-2237.
  4.    White,  R. T.;  Douthit,  G.  E.,  J.  Assoc. Off.  Anal.
       Chem., 1985, 68, pp 766-769.
  5.    Matthes,  S.  A.; Parrell,  R.  F., Mackie,  A.  J.,  Tech.
       Prog. Rep., U.S. Bureau of Mines, 1983, p!20.
  6.    Fernando,  L.  A.;  Heavner,  W.   D.;  Cabrillei, C.  C. ,
       Anal. Chem., 1986, 58,  p  551.
  7.    Fisher, L. B., Anal. Chem. 18986, 58, pp261-265.
  8.    Lamonte,  P.  J.;  Fries,  T.  L.;  Consul,  J.  J.,  Anal.
       Chem., 1986, 58, pp 1881-1886.
  9.    Copeland,  T.  R.,  Work assignment for  the  Office of
       Solid Waste, U. S. EPA, June 1986.
  10.  Westbrook, W. T.; Jefferson, R.  J., J. Microwave Power,
       1986, 21, p25.
  11.  Jassie,   L.   B.;   Kingston,  H.   M.,   1985  Pittsburg
       Conference Abstracts, Paper 108  A.
  12.  Kingston, H.  M.;  Jssie,  L. E.,  Anal.  Chem.,  1986,  58,
       pp 2534-2541.
  13.  "Symposium   on   Microwave  Techniques",   Twenty-fith
       Eastern  Analytical  Symposium,  Oct.   1986,   New  York;
       "Symposium   on  Microwave   Techniques",   Twenty-sixth
       Eastern Analytical Symposium, Sept. 1987.
  14.  Reverz,   R.;  Hasty,   E.,  Pittsburgh  Conference  and
       Exposition   on   Analytical   Chemistry    and  Applied
       Spectroscopy, March 1987.
  16 epapa2
                             11-264

-------
7Q     Sample Decomposition in Closed Vessels with & Pressure Controlled
       Microwave Oven.

       F.flanholzer.  Q.  Knapp, P. Kettisch, A. Schalk
       Department for Analytical Chemistry, Micro- and Radiochesitry,
       Grass University of Tehcnology, Technikerstrale 4, Graz, Austria

       wet-chemical sample decomposition  in closed vessels is one of the
       most efficient methods for trace element analysis.  Temperatures of
       at least 300*C are required,  in order to guarantee the complete
       decomposition of organic matter with nitric acid.  For that purpose
       the decomposition must be done under high pressure of up to 80 bar.
       Thซ currently available high-pressure vessels  for microwave
       decomposition do not  permit control of the microwave energy and
       therefore unknown sample materials can cause the vessel to rupture at
       excessive internal  pressure.   The  high-pressure  microwave
       decomposition vessels we developed permit  the  control  of microwave
       energy by the internal pressure.   Thus also unknown sample materials
       can be decomposed quickly  and without problems.
                                   1-265

-------
80        STATE-OF-THE-ART OF MICROWAVE DIGESTION METHODS
                           FOR ENVIRONMENTAL ANALYSIS
          Mark E. Tatro. SPECTRA Spectroscopy & Chromatography Specialists, Inc., P.O.
          Box 352, Pompton Lakes, NJ 07442


          ABSTRACT
          The closed vessel microwave digestion methods approved by EPA for the preparation of
          waters and soils for trace metal analysis require a two stage microwave power program
          that is designed to achieve an initial target temperature of the digestion acid followed by
          a slow rise to a final temperature. There are inherent problems with this method that
          arise from the fact that the digestion acid temperatures are functions of vessel  type,
          microwave  power, line  voltage  and the  number  of vessels.  In-situ  temperature
          measurement of the acid  during digestion requires  the use of an expensive fiber-optic
          probe device that is beyond the budget of most environmental laboratories.  Therefore,
          users will be "flying blind" when attempting to reach target temperatures. This paper
          will present data that depicts in-situ temperature-time curves that demonstrate the effect
          of vessel design and increased microwave power on target temperatures.
          INTRODUCTION
          The approved EPA CLP and the proposed EPA SW-846 closed vessel microwave
          digestion  procedures  are based  on  very  rigid  formats regarding    power-time
          programming and the number and type of vessels used (1). The EPA methods for the
          digestion of water samples requires the use of 5 vessels all containing 45 mis of water
          sample and 5 mis cone, nitric acid with a two stage power program of 545 Watts for 10
          minutes followed by 344 Watts for an additional 10 minutes. This program is designed
          to allow the acidified samples to reach a target temperature of 160 ฑ 4 ฐC by the end of
          the first 10 minutes and to  allow for slow rise to 165 - 170  ฐC within the next  10
          minutes. The EPA methods for the digestion of soil samples requires the use of either 2
          vessels containing the samples and  10 mis cone, nitric acid with a one stage power
          program of 344 Watts for 10 minutes  or 6 vessels containing the samples and  10 mis
          cone, nitric acid with a one  stage power program of 574 Watts for 10 minutes.  This
          program is designed to allow the acidified samples to reach an inital target temperature of
          175 ฐC in less than 5.5 minutes and remain between 170 -180 ฐC for the  balance of the
          10 minute time period. The above microwave programs are based on the  use of 120 ml
          single walled teflon vessels.
                                             11-266

-------
If the user deviates from the above conditions, i.e. uses more than the prescribed number
of vessels,  uses double walled  instead of single walled vessels, uses a higher power
wattage than prescribed, then the methods as written will not work as they will not reach
the initial and final target temperatures in the prescribed time.  The proposed SW-846
Methods 3015 and 3051 allow users to use higher wattage ovens to digest more than the
prescribed number of samples at one time and allow the users to use alternative vessel
designs.  However, the methods allow these changes only if the user can document that
the temperature-time profiles remain unaltered.  This requires the use of an expensive
fiber-optic temperature probe that requires experience and can be dangerous if not used
properly with pressurized vessels containing acid.  Therefore, this author considers the
SW-846 allowed changes to be a moot point for most environmental laboratories.
The basic equation used to calculate microwave power absorbed by acid matrices (2) is
shown in Equation 1:
                        P(absorbed) =  [(K)(Cp)(m)(dT)]/t
This author receives feedback from users that are under the impression that Equation  1
can be used to predict the final temperature reached in a specific time if they know the
number of vessels to be used. This, in fact, is not the case since this equation does not
take into account cooling effects. In fact, the power-time programs developed for the
EPA CLP and  SW-846 methods had to be developed empirically using  an  expensive
fiber-optic temperature probe (3).  Again, without such a probe, users cannot develop
their own microwave digestion methods to stay within the EPA required temperature-
time confines.
As yet, manufacturers of microwave digestion systems have not provided the information
needed  to  alter conditions while  remaining  within the temperature-time  guidelines.
Therefore,  this author conducted the following experiments to  document the effect of
varying the types of vessels and  the power on reaching the  initial and final target
temperatures for the microwave digestion of water samples.  This study used a Luxtron
Model 750 fiber-optic temperature probe for in-situ temperature  monitoring and a Floyd
Model RMS-150 microwave digestion oven (4).
Figure l(a) depicts the temperature-time curve for the typical CLP and SW-846 type
water digestion microwave program.  It uses 5 single-walled vessels, each containing 45
                                      1-267

-------
mis deionized water and 5 mis cone, nitric acid, using the prescribed 2 stage program of
545 Watts for 10 minutes followed by 344 Watts for 10 minutes.  The temperature
profile, as expected met the required target values.
Figure l(b) depicts the temperature-time curve developed empirically using 5 double
walled vessels, each filled with 45 mis deionized water and 5 mis cone, nitric acid.  A 2
stage program of 480 Watts for 10 minutes followed by 234 Watts for 10 minutes met
the required target temperatures. The reduction in power to meet the temperature targets
was a result of the greater insulation afforded by the double walled vessels.  Again, it is
stressed that this alteration to the EPA prescribed microwave power-time program would
have been impossible without the use of the fiber-optic temperature probe.
Figure l(c) depicts the temperature-time curve when 5 double walled vessels containing
45 mis deionized water and 5 mis cone, nitric acid, instead of 5 single walled vessels,
were used with the EPA power-time program of 545 Watts for 10 minutes followed by
344 Watts for 10 minutes. As shown, by using more insulated vessels, the temperatures
reached far exceed the EPA required target temperatures.
As stated previously, predicting target temperatures must  be done empirically.  This
author wanted to predict how many more doubled walled vessels could be used for water
digestions if a higher wattage (745 W) microwave oven were used.  Figure 2(a) depicts
the temperature-time curve for 12 vessels all filled  with 45 mis deionized water and 5
mis cone, nitric acid were heated at full power. As  depicted,  it required 16 minutes for
the samples to reach the initial target temperaure of 160 ฐC.  From this curve, the number
of double walled vessels that can be used to reach 160 ฐC in 10 minutes using 745 Watts
of power can be predicted as follows:
Step 1:  Using Equation 1, determine  the actual power absorbed over the  16 minute
period. In this case, m = 624 gm [(12 x 45 ml water x 1 gm/ml) + (12 x 5 ml nitric acid x
1.4 gm/ml)]; Cp = 0.9297 (estimated from reference (2)); dT = 160 - 24.2 = 135.8 ฐC; t
= 16 minutes x 60 second/minute = 960  seconds; K = 4.184. The actual power absorbed
is therefore:
               P(absorbed) = f4.184X0.9297)(624)(135.8> = 343 Watts
                                         960
                                      11-268

-------
Step 2: Using the empirically determined power absorbed of 343 Watts, determine how
many vessels containing 45 mis deionized water and 5 mis cone, nitric acid (52 grams
total mass/vessel) can be heated from 24 ฐC to 160 ฐC in 10 minutes:
(a)  m = [(P)(t)]/[(K)(Cp)(dT)] = [(343)(600)]/[(4.184)(0.9297)(136)] = 389 grams

(b)  (389 grams)/(52 grams/vessel) = 7.5 = 8 vessels


Figure 2(b)  verifies  that  the  prediction  of 8  vessels is correct  since  the  actual
temperature-time curve for 8 vessels containing 45 mis deionized water and 5 mis cone.
nitric acid using 574 Watts reaches 160 ฐC in 10 minutes.



SUMMARY
The interest  by EPA in converting  from hot plates to microwave ovens for the
preparation of samples for trace metal analysis is to  be  commended.  However, the
confusion at the outset when SW-846 approves methods 3015 and 3051 is expected to be,
in  this author's  opinion,  overwhelming.   Without  expensive in-situ temperature
monitoring probes and without documentation from the microwave manufacturers on
how to deviate from the rather rigid temperature-time profiles as prescribed by EPA,
users are expected to be confused. The announcement of in-situ temperature monitoring
capabilities built into  the next generation of microwave ovens will go a long way to
reduce this confusion.
REFERENCES


1. Tatro, M.E., EPA approves closed vessel microwave digestion for CLP laboratories.
    Spectroscopy,  5 (6), 17 (1990).

2. Kingston, H.M. and L.B. Jassie. Introduction to Microwave Sample Preparation.
    Theory and Practice, Chapter 6. H.M. Kingston and L.B. Jassie, eds.  American
    Chemical Society, Washington, D.C. (1988).

3. H.M. Kingston, private communication (1990).

4. M.E. Tatro, From hot plates to microwaves. Environmental Lab, 3(1), 28 (1991).
                                      1-269

-------
Rgure 1
  210
  200
            Single- vs. Double-wall vessels
      01234567891DH121314151617181920
Rgure 2.
Predicting Target Temperatures
180 ~
1 en -
isn -


13Q
4 9A -
-.120

gXUU -
i on -






30 H
20?

	 o^4 	 ^ซ
S* ^^

•f ^m****^
/ ^-""""^
X -*"
^-**^
fs*^
4 S
2>
v
//
ฃ
/
r
0123456789101112131415 18
Minutes
• 2(a) + 2(b)
                          11-270

-------
g-j      THE APPLICATION OF X-RAY FLUORESCENCE SPECTROSCOPY FOR RAPID
                  HAZARDOUS WASTE CLASSIFICATION AND SCREENING
         Dr. Peter A. Pospisil. Manager Methods Development, Dr. Harold Van Kley,
         Chemist Methods Development, Dr. Mark F. Marcus, Senior Director Analytical
         Programs,  Chandler Taylor,  Chemist Analytical Lab, Dr. Nilesh Shah, Senior
         Analytical  Chemist Methods Development, Chemical Waste Management Inc.,
         Technical Center, 150 W. 137th Street, Riverdale, IL 60627;

         Dr. E. Scott Tucker, Director Chem-Nuclear Laboratory Services, 25 Woods Lake
         Road, Greenville, SC 29607

         ABSTRACT

         X-ray fluorescence spectroscopy (XRF) with pattern recognition data interpretation
         provides immediate elemental screening capabilities for the comparison of sales
         sample metal composition with compositional data from loads received at hazardous
         waste sites.

         The proper disposal of hazardous  waste through stabilization processes requires
         information about the elemental composition of the waste stream.  Based on the
         stream's elemental composition, which is both waste code and generator specific,
         processing  decisions are made to select  the most effective waste stabilization
         procedure.   Current procedures utilize SW-846 methods to generate elemental
         compositional data.  The analytical  method turnaround time creates significant
         process delays and  costs for materials received at CWM sites.  XRF can rapidly
         provide information regarding RCRA elemental analyte composition in hazardous
         waste streams, enabling processing decisions to be made on a comparatively realtime
         basis and provide assurance of effective stabilization.

         Advances in energy-dispersive (ED) XRF instrumentation with computer software
         have greatly increased interest in the technology.  Advantages for hazardous waste
         analysis include minimal sample preparation, applicability to a broad range of liquid,
         solid or semi-solid samples and simultaneous multi-element analysis over a broad
         concentration range with no elemental carryover. The study generated elemental
         pattern data for K061 and F006 wastes, produced by specific generators.  Fourteen
         elements were selected for the study:  calcium, chromium, iron, nickel, copper, zinc,
         arsenic,  selenium, silver,  cadmium,  barium, mercury, thallium and lead.  ICP
         analyzed K061 wastes were used as quantitation standards.
                                            1-271

-------
Univariate elemental analysis using a one standard deviation comparative criterion
demonstrates the effectiveness of XRF technology for distinguishing among waste
codes and generators.   Multivariate  analysis  using the Mahalanobis distance
technique improves the comparison by utilizing specific elements or the entire X-ray
spectrum for recognition of XRF patterns.  We conclude that ED-XRF is able to
provide data critical for the decision process at hazardous waste disposal sites, while
reducing the overall cost of operation.

INTRODUCTION

All wastes received at  Chemical Waste  Management  (CWM), both  for  waste
disposal decisions and as received loads at our disposal sites  are subjected to a
"fingerprint" analysis.  The purpose of this rapid  test series is to verify that the
material received for disposal matches the profile for that generator produced by the
sales samples.  Screening tests  are in place for  nine parameters,  and it would
improve the quality of the screening process and streamline disposal operations if
metals could be included in the fingerprint  screening process.  This  study was
undertaken to determine if x-ray fluorescence spectroscopy (XRF) could provide a
reasonable and rapid metals-based fingerprint analysis to augment the current series.

An important factor in any fingerprint screen is that it be rapid and sensitive enough
to screen the hazardous  parameter at the appropriate level of quantitation.  XRF
is an excellent screening choice for this research, since samples require little or no
sample preparation  and  data can be generated within minutes.

PROJECT PURPOSE

The purpose of  this project is to determine the applicability of  XRF analysis to
fingerprint screening of sales  and received samples.  The  project will proceed
through the generation of quantitative analytical data to determine the feasibility of
a univariate pattern recognition process in differentiating among  waste  codes and
generators.  The work  will continue  using a  multivariate  process based on a
Mahalanobis distance technique, which will eliminate the need for the generation
of quantitative analysis.

TECHNICAL OVERVIEW

In XRF,  electrons in the lowest  energy orbitals near the nucleus of the atom are
energized by external radiation and escape from the atom. Electrons from higher
energy orbitals fill the empty orbital and the energy lost in dropping to a lower
                                   11-272

-------
energy orbital is emitted as an x-ray.  Because the emitted x-rays are always at a
lower energy than the activating radiation, the process is called "x-ray fluorescence."
Each element has characteristic electron orbitals of specific energy and therefore a
characteristic x-ray fluorescence pattern. Because inner orbital rather than valence
electrons are involved in this method, the chemical form of the element has little
or no influence.  The method is applicable for qualitative and quantitative analysis
for chemical elements higher in the periodic chart than oxygen.

Although the technique of ED-XRF has been known for about 40 years,  recent
advances in instrumentation and especially computer software have greatly increased
interest in the technique. XRF advantages include minimal sample preparation,
applicability to liquid, semi-solid  or  solid samples, simultaneous  multi-element
analysis over a wide concentration range, no carryover to the succeeding sample,
rapid quantitation and potential to optimize the system for specific elements.

The  XRF spectrometer records counts received in individual channels of a multi-
channel analyzer. Each channel counts a small range of energies so that a spectrum
of counts at specific energies is obtained. On the basis of known energy values for
individual elements,  specific ranges are assigned to certain elements.

EXPERIMENTATION

                             Instrument Selection

The  Kevex 770 XRF spectrometer was chosen as  the most applicable instrument
based  on its sensitivity, flexibility  and speed of  analysis.   The instrument was
purchased with a DEC VAX 11-57 computer, TSX operating system and Toolbox
software.

                             Sample Preparation

Sample preparation for this technique is minimal.  Dry powdered materials may be
placed directly in an XRF cup. Liquid or semisolid samples may be run as received
or be dried in an  oven and  then analyzed as a  dried powder.  Claylike damp
samples can be packed into the cup and tamped gently to remove the air spaces.

                            Instrument Calibration

As a starting point for this work the instrument was calibrated  using a large set of
K061 wastes from a single generator.  The set of K061 wastes were first analyzed
                                    11-273

-------
for the following 14 elements using conventional ICP and AA methodology:

            Calcium           Chromium         Iron
            Nickel             Copper            Zinc
            Silver             Cadmium          Barium
            Arsenic            Selenium          Mercury
            Thallium          Lead

These data are presented in Table 1. The high values of thallium and arsenic were
found to arise from elemental interferences and were not considered in the pattern
recognition process. Additionally, values for calcium and  iron were not included,
since their consistently high values overshadowed the metals of interest.

A critical consideration for this work is that the instrument was not calibrated to
produce quantitative data but consistent data among the matrices encountered.

                             Analytical Precision

A single K061 sample was run repeatedly under the same activation conditions to
show the precision of the method.  The sample  was run under the activation
conditions used for elements calcium, chromium, iron, nickel, copper, and zinc.  The
sample was stirred every few runs to expose a different portion of the sample to the
analytical procedure.  Results are shown in Table  2.   Again close agreement for
each element is seen so that precision is acceptable for waste samples.

                               X-Ray Analysis

The following samples were analyzed by x-ray using the  calibration curves produced
by the K061 materials.

      Number of               Sample Type                    Generator
       Samples

         24                       K061                           A
          4                       K061                           B
          7                       F006                           C

It was recognized that changing the sample type changed the matrix, which in turn
reduced the data's  quantitative quality.   Again,  the project goals were not  data
quantitation but pattern recognition, using elemental analysis as a guide.   Varying
the calibration curves only adds an additional degree of freedom.
                                   M-274

-------
The analytical data for Generators A, B and C are presented in Figures 1-4 a and
b, in a bar graph format.  The "a" portion of the Figure presents the composition
in percent, while the "b" portion presents the ppm information. The brackets on
each bar in Figure 2 show the compositional variation of the 24 K061 wastes at one
standard deviation. In a univariate pattern recognition approach, if the elemental
concentrations for the sample  fall between the brackets defined in Figure 2, there
is a 65% probability that the waste is a K061 from Generator A.

A visual comparison of the bar graphs  in Figures 2 and 3 show that the overall
elemental pattern is reasonably similar for the K061 wastes from generators A and
B.  But, since the lead, cadmium and barium concentrations  fall  outside of the
defined brackets, there is a high  probability that the waste is not a K061 from
Generator A.

Comparing the elemental pattern data between Generators A and C, K061 and F006
wastes, the basic elemental pattern is extremely different.  No lead appears in the
samples from Generator C, but there are significant amounts of barium and nickel.
XRF can  differentiate among  waste codes and generators based on the univariate
elemental pattern.

CONCLUSIONS

1.     XRF coupled with univariate analysis can generate unique  elemental data
       patterns, within 15 minutes for specific waste codes and generators. Wastes
       from an individual generator are quite characteristic.

2.     The method is rapid enough to be applicable for the purpose of fingerprint
       screening.

3.     The data are also applicable to process related decisions which are a function
       of elemental distribution.
                                    11-275

-------
                                                        TABLE 1
                                   ANALYTICAL DATA FOR 24 K061 WASTES FROM GENERATOR A
                  LEAD
CHROMIUM   COPPER
CADMIUM
  PPM
BARIUM
 PPM
NICKEL
 PPM
SILVER
 PPM
N>
























AVERAGE
STD DEV
RANGE

RSD68%
RSD 95%
RSD99%
1.40
1.47
1.26
1.33
1.65
1.92
1.70
1.22
1.22
1.33
1.25
1.40
1.15
1.36
1.43
1.31
0.84
0.77
1.32
1.13
0.78
1.01
0.89
0.67
1.242
0.298
1.541
0.944
24.032
48.064
72.096
0.15
0.16
0.12
0.14
0.18
0.17
0.17
0.16
0.14
0.14
0.16
0.14
0.18
0.18
0.18
0.18
0.14
0.15
0.16
0.17
0.16
0.14
0.14
0.14
0.156
0.017
0.173
0.139
10.901
21.802
32.702
0.19
0.18
0.16
0.18
0.19
0.19
0.20
0.19
0.15
0.13
0.15
0.14
0.13
0.18
0.16
0.13
0.13
0.17
0.14
0.14
0.15
0.20
0.15
0.13
0.161
0.024
0.185
0.137
15.115
30.230
45.345
152
131
148
129
161
123
221
201
157
239
227
224
219
223
226
233
190
148
223
257
175
208
195
133
189.292
40.345
229.637
148.947
21.314
42.627
63.941
161
118
155
123
130
115
159
135
122
144
141
116
118
165
111
116
123
126
115
117
164
164
180
158
136.500
20.831
157.331
115.669
15.261
30.521
45.782
99.8
76.2
76.2
76.2
91.2
115.0
102.0
99.2
84.4
87.3
94.2
76.2
76.2
107.0
112.0
124.0
76.2
98.3
107.0
102.0
76.2
76.2
95.2
97.0
92.717
14.231
106.948
78.485
15.349
30.669
46.048
36.2
36.6
35.9
35.7
36.6
373
37.5
41.2
36.4
38.0
40.0
36.9
35.6
38.2
38.7
35.4
35.5
36.0
37.8
36.7
35.4
34.7
35.7
36.0
36.833
1.513
38.346
35.321
4.107
8.213
12.320
0.10
0.10
0.10
0.10
0.72
0.10
0.10
0.10
0.10
0.75
0.10
0.10
0.10
0.10
0.10
0.10
0.10
0.70
0.10
0.10
0.52
0.10
1.01
0.10
0.229
0.272
0.502
-0.043
118.870
237.740
356.610
0.53
0.53
0.53
0.53
0.53
0.53
0.53
1.76
0.53
0.53
1.69
1.79
0.53
1.50
1.36
0.53
0.53
0.53
0.53
1.80
1.50
2.58
0.53
0.53
0.936
0.608
1.544
0.328
64.996
129.992
194.988

-------
                                                               TABLE 2
                                              ANALYTICAL DATA PRECISION FOR KO61
N>
   Average
   StdDev
                         CALCIUM
7.13
7.08
7.07
7.03
7.13
7.13
7.01
7.05
7.04
7.15
7.08
7.15
7.00
7.08

7.08
0.049
ARSENIC
   ppm

   0.17
   0.16
   0.15
   0.17
   0.15
   0.15
   0.19
   0.15
   0.18
   0.17
   0.17
   0.18
   0.18
   0.17

   0.17
   0.013
                                         IRON
                  NICKEL
                    ppm
42.8
43.0
42.8
42.9
42.8
42.9
43.1
42.9
42.9
43.0
43.0
43.4
43.2
43.5
43.01
0.210
103
88.9
88.9
88.9
107
99.2
122
119
117
93.3
97.3
108
102
111
103.25
10.875
                                                                                                                COPPER
                                           0.19
                                           0.17
                                           0.17
                                           0.17
                                           0.15
                                           0.17
                                           0.18
                                           0.19
                                           0.21
                                           0.17
                                           0.15
                                           0.18
                                           0.23
                                           0.16

                                           0.18
                                           0.021
                                                                                        ZINC
9.09
9.15
9.01
9.05
9.08
9.15
9.13
9.13
9.18
9.03
9.07
9.24
9.24
9.15

9.12
OfflB
   RSD%:
0.69
   7.6
0.48
                                                                                            10.5
                                                                                      1.16
                                                                                         0.74

-------
                                             FIGURE 1A
o
cc
UJ
0.
 1.5
 1.4
 1.3
 1.2
 1.1
  i
 0.9
 0.8
 0.7
 0.6
 0.5
 0.4
 0.3
 0.2
 0.1
  0
                                ELEMENTAL DISTRIBUTION PERCENT
                                       24 K061 SAMPLES GENERATOR A
                               LEAD
                                               CHROMIUM
                                                ELEMENT
                                                    COPPER
0.
a.
200
190
180
170
160
150
140
130
120
110
100
 90
 80
 70
 60
 SO
 40
 30
 20
 10
  0
                                            FIGURE IB

                                  ELEMENTAL  DISTRIBUTION  PPM
                                       24 KO61 SAMPLES GENERATOR A
                       CADMIUM    BARIUM     NICKEL     SILVER   MERCURY   SELENIUM
                                                ELEMENT
                                      11-278

-------
                                      FIGURE 2A
                           ELEMENTAL DISTRIBUTION PERCENT
z
Ld

O
1.5 -
1 .4 -
1.3 -
1.2 -
1.1
1 -

0.9 -
0.8 -
0.7 -
o.e -
0.5 -
0.4 -
0.3 -
0.2 -
0.1 -


















sN
\
\.
v\
\
X



\
s/X
X
V,
k \
N^
^
x^
^O

X^
ง^
XX
N V












XX XV
\ \> v x\
LEAD CHROMIUM COPPER
ELEMENT
a
a
                                      FIGURE 2B




                              ELEMENTAL DISTRIBUTION  PPM


                                 24 K061 SAMPLES GENERATOR A

i an —
I JW
180 -
170 -

160 -
150 -
140 -

130 -
120 -
1 10 -
100 -
90 -
80 -
70 -
60 -
50 -
40 -i
30 -
20 -
10 -




















\

\
N


\
\

\
— *r
X
x\

X
x>
X
\Xv
X
\
\X
Xs
\\
s
0 —| 	 1


















X
^
^
\
1^


"s

•i
XN
X
x^
XN
X
\x
N
\
x^
N
\
I 	 1 	 r














X
\
N
X
X
X
X
\











r
\
k
\
x
s.
X
V
\
\
V
\





—^—.
X
\\
\

                                          ELEMENT
                                  1-279

-------
                                          FIGURE 3A
Ui
2.2




  2



1.3




1.6




1.4




1.2




  1




0.8



0.6




0.4




0.2




  0
                              ELEMENTAL DISTRIBUTION PERCENT

                                     4 Koei SAMPLES GENERATOR B
                             LEAD
                              CHROMIUM


                               ELEMENT
                                                               COPPER
o.
a
              600
              500 -
              400 -
              300 -
              200 -
              100 -
                                          FIGURE 3B


                                ELEMENTAL DISTRIBUTION PPM

                                     4 K061 SAMPLES GENERATOR B
        CADMIUM    BARIUM    NICKEL    SILVER



                               ELEMENT
                                                           MERCURY  SELENIUM
                                    11-280

-------
                                           FIGURE 4A
z
UJ
(J
IT
 0.3


0.28


0.26


0.24


0.22


 0.2


0.18


0.16


0.14


0.12


 0.1


O.O8


0.06


0.04


0.02


   0
                               ELEMENTAL DISTRIBUTION  PERCENT

                                      7 FQ06 SAMPLES GENERATOR C
                                        \
                          LEAD
                   T	^-TT

                     CHROMIUM
  T"

COPPER


ELEMENT
                                                             T

                                                           BARIUM
                                                                      NICKEL
Q.
CL
                                            FIGURE 4B


                                  ELEMENTAL DISTRIBUTION  PPM

                                       7 F006 SAMPLES GENERATOR C
100 -
90 -
80 -
70 -
60 -
50 -
40 -
30 -
20 -
10 -
0 -









O<
^x
\\

JvN





— I 	 r 	 1 	 1 	 1 f
CADMIUM SILVER MERCURY SELENIUM
                                                ELEMENT
                                       1-281

-------
QO            SEMI-QUANTITATIVE DETERMINATION OF INORGANIC
           CONSTITUENTS  IN SPECIFIC AND NON-SPECIFIC CATEGORICAL
                         WASTESTREAMS USING EDXRF
       Kat Fennel1. Senior Chemist
       Roger M. Olbrot, Laboratory Manager
       Ted G. Howe, Spectroscopy Group Leader

       USPCI, A Subsidiary of Union Pacific
       Grassy Mountain Facility
       8960 North Highway 40
       Lakepoint, Utah  84074
                                 ABSTRACT
       X-Ray fluorescence spectroscopy was used to determine
       several inorganic constituents and associated interferences
       in F006 and K061 wastestreams.  The primary objective was
       the optimization of stabilization reagents and materials by
       analyzing incoming waste loads prior to stabilization.  By
       applying EDXRF techniques to the pre-accepted waste loads,
       an overall increase in site stabilization efficiency can be
       realized.  Total metal disparities between the pre-
       acceptance sample from the generator and the actual load
       sample are ubiquitous and problematic.  Presently, there
       exists a paucity of quantitative data concerning the role of
       EDXRF as a useful analytical tool applied to the
       environmental analysis  of categorical solid waste.
       Wastestreams that were approved for treatment, stabilization
       and disposal were randomly sampled and subjected to salient
       preparation methods.  Three instrumental calibration
       techniques were investigated using the K061 samples:
       fundamental parameters (FPT), single similar standard and
       simple linear regression calibration. Because of the extreme
       variation of the F006 samples, only two instrumental
       calibration techniques were investigated on these samples:
       FPT, and a sorting program developed for alloy analysis.
       Sample characterization using these techniques can be both
       semi-quantitative and quantitative depending on several
       parameters, notably, the analyte, method of sample
       preparation and interferences.  The sensitivity inherent to
                                   11-282

-------
this analytical technique was sufficient to meet our
objectives and provided a certain degree of direction for
further study.  Moreover, pre-stabilization profiling or
analysis of these types of wastestreams optimizes the use of
proprietary additives while reducing the total volume of
waste placed in the landfill.  A complex database can be
structured and implemented to work in concert with our
rigorous stabilization program furthering our expertise in
the field of waste management.
                             1-283

-------
     IDENTIFYING SOURCES OF ENVIRONMENTAL CONTAMINATION
        THROUGH LASER SAMPLING ICP-MASS SPECTROMETRY


Kenneth J. Fredeen. Sr. Technical Specialist, ICP Worldwide Marketing  Group,
The  Perkin-Elmer Corporation, 761  Main Aye, Norwalk,  CT 06859-0215; Mark
Broadhead, Vice President, Chemical and Mineralogical Services, Inc., 445 West
2700 South, Salt Lake City, UT, 64115


ABSTRACT

Laser sampling ICP-mass spectrometry (LS-ICP-MS) is shown to be an effective
technique  for helping to identify  sources of environmental contamination.  This
rapid, sensitive elemental analysis technique requires little or no sample preparation
and  provides elemental fingerprints of solid samples in a few minutes.  In this study,
LS-ICP-MS was used to  analyze high-volume air filters and sludge samples, with
emphasis placed on obtaining elemental isotopic fingerprints of all the samples and
semiquantitative analysis results for the air filter samples.


INTRODUCTION

Determining  the elemental  composition  of  environmental   contaminants  or
contaminated materials is one of the many concerns of environmental regulatory
agencies and the analytical chemistry community.  For many sample analyses, the
analysis goal is to determine accurately, a specific list of elements.  These analyses
usually have rigid, well-defined analysis protocols and specified maximum levels for
the elements to be determined. For some analyses, however, it can instead be more
advantageous to obtain a full elemental profile of the samples, without the need for
a high degree of accuracy. For other samples, it may also be advantageous to know
the isotopic abundances for specific elements.  Such elemental profiles and isotopic
information  can often be  used  as a "fingerprint" to help identify a source of
environmental contamination.

As an example, the presence of smog and particulate matter in air is an important
environmental  concern.  Determining the relative  elemental compositions of the
smog and particulates will help regulatory agencies decide what elements need to be
monitored and regulated. Also important, however, is identifying the point  sources
of the emissions so that proper controls can be placed on these sources.  This
example can likewise be extended to ground and water contamination.

In this study, laser sampling inductively coupled plasma mass spectrometry (LS-ICP-
MS) has been used to analyze sludge samples of domestic and industrial origins and
high-volume air filters from various locations in a metropolitan area. By comparing
and combining the analysis results with other information, these results can  then be
used to help establish the sources of environmental contamination.

The well-established ICP-MS technique has  the ability to measure rapidly as many
as eighty elements.  Besides the vast numbers of determinations that can be made in
                                  1-284

-------
a relatively short time using the technique, ICP-MS is known for its high sensitivity,
with detection limits generally in the parts-per-trillion range for solution sampling.
As a mass spectrometry-based technique, it also has the advantage of being able to
determine isotope ratios for certain elements present in a sample.

Using laser sampling, as opposed to solution sampling, as the sample introduction
technique,  has the advantage of eliminating the usually tedious sample dissolution
step required for most atomic spectrometry techniques.  This is accomplished  by
focussing a high-power laser beam onto the sample, thereby creating a sample vapor
directly from the solid state. This vapor is then transported to and analyzed by the
ICP-mass spectrometer.  Sample results can normally be obtained in a few minutes.
Among the other  advantages are its ability to provide spatially resolved sample
information in the form  of lateral distributions and depth gradients of elements in a
solid.  An  increasingly important advantage of eliminating the sample dissolution
step is avoiding the  need  to work with concentrated  acids and to  dispose of acid
wastes. A thorough review of the LS-ICP-MS technique has been recently published
[1]-

LS-ICP-MS has been  shown previously to be a  good  technique  for rapid,
semiquantitative analysis of various geological materials [2] and several materials of
environmental  interest  [3],.as well as  many  other sample types.   Given  proper
calibration standards for the elements of interest, LS-ICP-MS can also be used for
quantitative analysis of  many materials. However, like many other solid sampling
techniques, the availability of appropriate solid calibration standards can limit the
quantitative aspect of LS-ICP-MS for some analyses.

The primary interest in this  study was  to use  LS-ICP-MS to obtain elemental
fingerprints of the various samples analyzed.  Such fingerprints indicate the relative
elemental  compositions of the samples without  regard for the absolute element
concentrations.   While some interpretation of the  mass spectra is required to
compensate for spectral interferences,  subsequent quantitation using the element
intensities, and thus calibration of the system, is not required.

Of secondary interest in this study was to quantitate the results from the analysis of
the high-volume air  filters.  In order to accomplish this task, a calibration method
appropriate for laser sampling of the air filters had to be developed.


EXPERIMENTAL

Instrumentation

Two different LS-ICP-MS systems were used for data collection in this study.  The
first system consisted of a Perkin-Elmer SCIEX Model 320 Laser Sampler coupled
to a P-E SCIEX ELAN 500 ICP mass spectrometer. The  laser sampler consists
primarily of a  pulsed, Q-switchable  Nd:YAG  laser; an  enclosed sample  cell
mounted on a three-axis  translation stage; and a closed-circuit video  monitoring
system. The laser and translation stages are all computer controlled. The ICP-MS
was controlled using an  IBM PS/2 Model 70 microcomputer running the P-E
SCIEX ELAN 5000 software under the  Xenix operating system. The laser sampler
                                    1-285

-------
was controlled by an 80286-based microcomputer running the P-E SCIEX laser
sampler software under Windows and DOS.

The second LS-ICP-MS system consisted of a Model 320 Laser Sampler coupled to
a P-E SCIEX ELAN 5000 ICP mass spectrometer. Both the ICP-MS and the laser
sampler were controlled from a single IBM PS/2 Model 70 microcomputer.  The
software used was the same as for the first system except that the laser sampler
software was run under Windows  and VP/ix.  A block diagram of the laser
sampler/ICP-MS system is shown in Figure 1.
         PE-SCIEX ELAN SOOO ICP Man Spcctrofiwtw
     Syซ(ซni Compute
                                                      PE-SCIEX Mod.4 32O Uซw Simpter
Figure 1. Diagram of laser sampling ICP-MS system.

Methodology

Standard and Sample Preparation.   For the sludge analysis, the certified reference
materials  studied were combined  with  an  X-ray fluorescence  binding agent,
SpectroBlend (Chemplex Industries, Inc.), at a sample:binder ratio of 3:1. These
mixtures were shaken well and then pressed into 0.5 g pellets using an IR pellet dye
and a 12-ton press. Had the reference materials not been finely powdered, they first
would have been ground to a 350 to 400 mesh before mixing with the binding agent.

Standards were prepared for the high-volume air filters analysis by evenly loading
measured  amounts of standard reference materials (SRM) onto 1.5 x 3.0 cm pieces
of blank filters  of the same type used for the air sampling. In order to keep the
SRM's from leaving the surface of the filters prematurely when sampled by the
laser, the filter standards were coated with an aerosol-based binding agent. The air
filter samples were cut into 1.5 x  3.0 cm pieces and also coated with the binding
agent before laser sampling. For solution sampling, 3.0 x 3.0 cm pieces of the filter
samples were leached in 20% nitric acid.  The leachate for each sample was then
diluted 20x before analysis. Appropriate blanks were prepared in the same manner
as the respective laser and solution sampling filters samples.

Analysis Procedures - General Standard laser-sampling operating conditions for the
ICP-MS instruments  were used for all the analyses.   For the  semiquantitative
                                   11-286

-------
analyses, the LS-ICP-MS system was calibrated by analyzing a prepared standard for
several certified elements using the TotalQuant routine of the ELAN 5000 software.
The TotalQuant routine then used the results for the certified elements to calibrate
the system for the other elements of interest, for a total of up to eighty elements.
The TotalQuant routine was also used to produce the fingerprint spectra resulting
from the analyses.

Sampling Procedures - Sludges.  The reference material pellets were sampled with
the laser in the Q-switched mode, with an energy of 50 mJ/pulse and a repetition
rate of 10 Hz.  The laser beam was scanned back and forth across the sample on a 5-
mm line for a 20-second  pre-measurement period followed by a  60-second data
collection period. Since only relative sample intensities were desired for  this part of
the study, no concentration calibration was performed.

Sampling Procedures - Air Filters.   The standard and sample air filters were sampled
with the laser in the Q-switched mode, with an energy of approximately 10 mJ/pulse
and a 10-Hz repetition rate.  A 6 x 7 mm area of each filter was sampled using a z-
pattern raster  with the laser beam.   A 30-second pre-measurement sampling time
was used, followed  by 3  successive 60-second  data collection periods for each
determination.
RESULTS AND DISCUSSION

Sludge Samples

Spectral fingerprints for the two sludge materials studied were obtained easily using
the described technique.  Figures 2 and 3 show the single point per dalton "mass
histograms" for the  sludge samples.  These spectra were interpreted and elemental
intensities were automatically calculated by the TotalQuant software routine.   The
element intensities were then normalized to each sample's total element intensity so
that any  differences in laser sampling efficiency could be nullified.  The  resulting
normalized elemental fingerprints for the sludges are shown in Figures 4 and 5.
    7.0
                                160   120   140   160   180   200   220
    1 .0
           20
Figure 2. LS-ICP-MS spectral fingerprint for domestic sludge sample.
                                   11-287

-------
   7.0
                                  Sanple Indusฑ1- IS liidge:
                                     100    120    140
                                         MflSS < ar-iu
                                                                            220   241
Figure 3. LS-ICP-MS spectral fingerprint for industrial sludge sample.
 0.0000001
         H  B   F   A]  Cl  So  Mn Cu Aj Rb Nb  Pd  Sn  Xa  Ce Eu Ho  Lu  Ra  Au  Bi
Figure 4. Normalized elemental fingerprint for domestic sludge sample.
1 -
0.1 -
0.01 -
0.001 -
0.0001 -
0.00001 -
0.000001 -
n nnnnnm -






-4-


1





' t i MM.





tuu.





I





I
l



ii



i




I






I

I





         H  B   F   A/  Cl  Sc  Mn  Cu  As Rb Nb  Pd  Sn  Xs  Ca Eu Ho  Lu  Ra  Au   Si

Figure 5. Normalized elemental fingerprint for industrial sludge sample.
                                       11-288

-------
At first glance, the two sludge elemental fingerprints appear quite similar. However,
upon closer examination, some differences can be seen.  In order to make  these
difference more  apparent, the two  elemental fingerprints were ratioed to one
another.  Figure 6 shows the ratios of the domestic sludge  fingerprint to the
industrial sludge fingerprint. From these ratios, it appears that the domestic sludge
has a higher content of N, F, S, and Se, while the industrial sludge is higher  in In,
Te, Pt, and Bi.
1000000 -
100000 -
10000 -
1000 -
100 -
10 -
1 -
0.1 -

0.01 -
0.001 -

0.0001 -
0.00001 -
0.000001 -
N




I y. JU
s
F


-1


Lj-i_A_A


Se

1 W ' ' ' ••"•ป• — •'' i-w-i mm" • i i ' -m i i i i ||i in i-i III'II1 ' ••!ป•• ||.ปป^"i "••i-i|||l 1 1 |||'i;|'|l !|l
I Ce Bi
1

Te Pt
In


Figure 6. Ratios of normalized results for sludges, domestic:industrial.

Because the ratio of two small numbers can still turn out to be a large (or very
small) number, however, it was necessary to apply a filter function to the fingerprint
data before the ratios were calculated.  Figure  7 shows the domesticrindustrial
sludge ratios after a le-06 filter  function  was applied to the data.   While the
significant elements on the domestic sludge  side didn't change much, the significant
elements for the industrial sludge now appear to be Cd, Sb, Ce, Au, and Bi.

The utility of data such as  these goes beyond the ability to determine rapidly the
content of a sample. For the case  of the domestic sludge, one might expect a high
organic content,  and thus  look for  carbon as a  major constituent.   While the
domestic sludge did contain a large amount of carbon, Figure 7 shows that it was not
much higher than was found in the industrial sludge.  However, while the nitrogen
and sulfur content of the domestic sludge was much lower than the carbon content,
the relative concentrations  of these elements were much higher than those in the
industrial sludge.  Therefore, it should be  possible to  use nitrogen and sulfur as
indicators of the organic content of a sample.  Likewise, it appears that the heavy
metals,  rare earths, and precious metals may be good tracers for industrial waste.
While these data  certainly do not comprise  a comprehensive study, combining such
data with methods of principle components analysis shows great promise for helping
environmental  scientists  to  categorize waste types and even pinpoint  sources of
environmental contamination.
                                    11-289

-------
10000 -
1000 -

100 -

10 -
1 -
0.1 •
0.01 -
0.001 •
0.0001 -
0.00001 -
0.000001 •


N

F
JLJ l_A_J




Se
.111 i ^. i . J. J. t_i t .•!!. i t i_. . i i t ti i _i_ i it _* M tiittttttiiiiitiiitiititJtiLi-ti
B m IgnBTT 1 I l-:im ifB" n I I -m I i ' • f|i ||| • • 111"!*' ' "I'l'II 	 " " w 111' ' ' "ll'll' '•'
Nb Sb Ce Au Bi
•




Figure 7. Domestic:industrial sludge ratios after 1e-06 filter function.
High-Volume Air Filters

In this part of the study, several high-volume air filters from a metropolitan area
were analyzed.  These filters sampled the air in various locations in the area and
were designated as "downtown," "North," "petroleum plant," and "airport."  There
was also a filter that sampled the output from a point emission source.

Obtaining reproducible spectral and elemental fingerprints of the high-volume air
filters was not quite as straightforward initially as is was for the sludge samples.
Because of the fragility of the  glass fiber filters used for this application and the
relatively  low laser power with which they could be ablated, the laser sampling
conditions for this application  had to be carefully controlled.   Once  the  proper
sampling  conditions  were established,  however,    good  fingerprints could  be
produced  readily.  The spectral fingerprint for the "downtown"  filter is  shown in
Figure 8.

Quantitating  the  results from the  air  filter analysis  required  producing  an
appropriate calibration standard and analysis method. Previous  attempts made in
our laboratories,  and those of  other workers [4],  to  provide a  proper calibration
standard have included soaking filters with standard solutions, using a mylar-based
filter standard, and coating filters with slurries of reference materials.

Using the first two methods have the disadvantage  of having the calibration  species
entrained  throughout the filter material, whereas the sample filters have the  analyte
particles mostly on  the surface of the  filter.  Because the filters used  in this
application can contain relatively high levels of some of the elements of interest, the
goal of the laser sampling process is to remove particles from the filter surface while
removing  as little of the filter as possible.  In addition, removing too  much  filter
material can leave filter particles deposited throughout the sample transport line,
                                     II-290

-------
causing memory effects. Another possible problem with the solution-spiked filters is
that the chemical forms of the analyte species in the standards and sample are likely
to be quite different.  This can cause some differences in the sampling efficiencies
for the species,  although  using a Q-switched laser  should  help to  reduce the
differences.  Finally, there has been no proof to date that this calibration method is
valid for the air filters analysis.
      7. a
                              nple Dountoi-in: Laser- Sampling
                                                   160   iaa   2ea   220  2-41
Figure 8. LS-ICP-MS spectral fingerprint for "downtown" air filter.

Advantages of the slurry-coating approach  are  that  the  species  of interest are
deposited onto the surface of the filter  and are more likely to be in the same
chemical form as they are on the sample filters. One potential problem with this
approach,  however, is that some analyte  species could  be leached from the
reference material particles and either be left in the container the slurry was made
in or soaked into the bulk of the filter.  Another practical problem with sampling
slurry-coated  filters with  the laser is that  a relatively wide area, up  to  several
millimeters in diameter, of the reference material is removed by the shock wave
produced by the laser-solid interaction. This may seem to be an advantage  at first,
since the sampling efficiency would be  large.   However,  much of the material
removed in this manner is not  vaporized by the laser, but instead is removed as
particles that are too large to be analyzed well by the ICP-MS or simply fall out of
the carrier stream and are deposited in the sample line.

The approach used in this  study was to load various dry,  powdered reference
materials directly onto blank filters.  The filters,  which were cut to  a specific size,
were weighed before and after the reference materials were added so that a  loading
factor could be calculated for each filter.  This loading factor was then used with the
certified concentrations to determine  the loading,  in ng cm-2, for the elements of
interest on the filter surface.

When filters prepared in this way were sampled with the laser, the problem with the
wide sampling area, described  above,  was encountered.  To counteract this effect,
the filters were coated with an  aerosol-based binding agent and dried. Subsequent
                                     11-291

-------
sampling of the filters was then confined to the immediate area of the laser beam.
Another advantage of this method is that less of the filter substrate was sampled by
the laser, also. Analysis of blank filters coated with the binder only showed that the
binder did not contribute to the background levels of the elements of interest.

Figure 9 shows the semiquantitative analysis results from the laser sampling analysis
of several  SRM-coated filters.  A filter coated  with  NIST SRM  1648,  Urban
Particulate Matter, was used as the calibration standard for this analysis. SRM's on
the analyzed filters included Coal Fly Ash and Estuarine Sediment. Also included
are results from analysis of a filter coated with the  Urban Particulate Matter SRM,
but with a different loading factor than the standard filter.  While  these  results
indicate that  this approach used for calibration has some validity, it is clear that
more work needs to be done to refine the technique.
        100000
             0.1
                                 10
  100

Reference
1000
                                                              10000
                               100000
Figure 9. Semiquantitative analysis results (ng crrr2) for SRM-coated filters.

All of the sample filters were  also analyzed against the SRM 1648 standard filter.
Figure  10 shows the semiquantitative elemental  fingerprint for  the  "downtown"
sample.  Of note are the relatively high results for As, Cd, Hg, and Pb. The other
filters showed similar results. While it was not done in this study, these results could
be compared in the same manner as the sludge results to determine what elements
show the most significant differences and could be used as tracer elements to help
track sources of pollution.

Comparison to Solution Sampling

The analysis of air filters using atomic spectrometry techniques is in wide practice.
While X-ray fluorescence can be used to analyze these filters directly, most analyses
are performed using atomic  absorption and  ICP techniques which  require  the
sample  to be in a solution form. The most widely-used technique for getting  the
samples into solution is to use a nitric acid leaching procedure. In fact, it is because
of this leaching procedure that the  glass fiber filters are used instead of the more
common cellulose fiber filter papers.
                                    1-292

-------
                 N
10000 T

1000 •
100 -

Loading 10 -

1 -
0.1 -
n nt -







i t i i i .
re
S







t-H illi*.







, L. t i L M.M.M,
Cu
Pb

As




•ULA — M — i
c



Jt-l L. .L-J-J*.
;d


|
A— i




t — i
c



tM
Hg
W


. . Mill t I M l


\ \ \ \ I Mi






I
            H  B  F  A)  Cl  Sc Mn Cu As Rb Nb Pd  Sn  Xe Ce Eu Ho Lu Re Au Bi
Figure 10.  Semiquantitative elemental fingerprint for "downtown" air filter.  (Loading in ng crrr2)

The results for laser sampling ICP-MS  were compared  to results obtained  from
solution sampling ICP-MS for the same filter samples.  After nitric  acid leaching
procedures were  performed on the filters, the leachate solutions were diluted to a
factor that would give ICP-MS results in the same intensity range found for  laser
sampling.    Figure  11 shows the solution  sampling  spectral fingerprint for the
"downtown" filter sample.  Many  features of this fingerprint are similar to those
found for the laser sampling fingerprint of this filter (Figure 8).
                          Sanplg Dountoun: Solution Sanpling
     1 .8
                                                         III
                                                  lea   tea
                                                                   220   2-fl
Figure 11. Solution sampling ICP-MS spectral fingerprint for "downtown" air filter.

In order to make comparison of the solution and laser sampling  results easier, the
results were  normalized and  ratioed in the  same manner  as the sludge samples.
Figure 12 shows the ratios comparing the solution and laser sampling results for the
                                     1-293

-------
"downtown" filter sample.  Of particular note from these results is that the laser
seems to have provided a more complete sampling for N, S, Cl, Br, and Hg. It is not
surprising that  these elements were either not leached well from filters using the
nitric acid procedure or did not remain in solution.
10000 i
1000 •

100 •

10 •
1 -
0.1 -

0.01 •
0.001 •
0.0001 -





p








c
\



Mo
Ti

II
I


Ce
Csll
1 II


a Br '
Hg


            N

Figure 12. Ratios of solution sampling:laser sampling results for "downtown" air filter,

Because of specific interest in determining sulfur for the filters, the laser sampling
results for sulfur were examined more closely. This interest is compounded by the
fact that X-ray fluorescence determinations for sulfur are suspect in their accuracy.
Figure 13 shows an intensity versus sulfur loading plot  for  the semiquantitative
analysis of the  SRM-coated filters. While this is not a perfect calibration curve, it
does show promise for laser sampling ICP-MS (or LS-ICP-AES) as a method for
determining sulfur on the filters.

       1000000
Intensity 100000
        10000
            1000
       10000

Sulfur Loading (ng cm-2)
                                                                    100000
Figure 13. Intensity (counts/sec) versus sulfur loading results for SRM-coated filters.
                                     I-294

-------
Solution and laser sampling results were also compared for the point  emission
source filter sample, since this sample contained high levels  of several difference
elements. These results are  shown in Figure 14.  On the solution side, it was found
that Zn and Ba could be determined better by solution sampling than with the laser.
This is because the blank filters used in this application contained high  levels of
these elements which, when sampled by the laser, produced very high background
signals that made them difficult to determine.
100000 -
10000 -
1000 -

100 -

10 -
•
0.1 -
0.01 -
0.001 -
0.0001 -
0.00001 -






I 'ซ,1 1 '
1 l (H 1 !


N
1





Al
1 1 1 1 1 ll
1 I I r t !


J
c

c

p

1

1
a


i

a



Zn
.jUUluUlJL
I1 '
Sc

S


B
Rb

1
I Cs
1 1 1, ซ_il ,,,!,,. -., 1
Pd Sn •
1

e


a






AuHg



Figure 14. Ratios of solution sampling:laser sampling results for point source filter.

For laser sampling of the point source filter, once again the laser seems to be more
proficient for N,  Hg, and the halogens.  Also note that Au  and Pd were sampled
better with the  laser. Figures 15 and 16 show the spectral fingerprint data from 170
to 240  daltons for solution and laser sampling of the point source filter.  The
difference in the  Hg isotope intensities in the  196 - 204 range for the two methods
can be seen clearly, while intensities for Pb (204 - 208), Tl (203  & 205), Re  (185 &
187), W (180 -  186), Th (232), and U (238) are quite similar.  Also, the Au signal at
197  is quite  strong for laser sampling while essentially absent  from the solution
sampling data.

As indicated in  the sludge application, N, S, and  the  precious metals may be
important  elements  for helping to  identify and track  sources of  environmental
contamination. The ability of laser sampling ICP-MS to determine these elements,
even semiquantitatively, could be quite important for this type of application.  For
toxic elements  such  as As, Cd, Pb,  and Hg, the importance  of the more complete
sampling using the laser is even  more apparent.  Had solution sampling techniques
alone been  used for  these  samples,  it  is  possible that important elemental
information would not have been uncovered.
Another advantage of using laser sampling for the analysis of air filters is that by
eliminating the sample leaching step, the need for using the glass fiber filters, as
                                    11-295

-------
opposed to cellulose fiber filters, is also eliminated.  During methods development
for this application with the laser sampler, it was found that cellulose filters coated
with SRM's could be sampled more easily and reproducibly than the glass filters.
The semiquantitative analysis results for laser sampling of an SRM-coated cellulose
filter are shown in Figure 17.  Besides the generally good  agreement between the
observed and certified values, note that zinc could be determined using these filters,
whereas it could not  be determined  on the glass filters  because  of the  high
background levels.
    7.e
                                           Solution Sanuling
    i.e
       IBS
                ise
                          19B
                                   28B      21B
                                    Mass < anu >
                                                      22B
                                                               23B
Figure 15. Spectral fingerprint from 170 - 240 dalton region for solution sampling analysis of point
source filter.
    7.a
    l.B
       163
                isa
                          ise
                                   280      21B
                                    MASS Canu>
                                                      220
                                                               23B
Figure 16. Spectral fingerprint from 170 - 240 dalton region for laser sampling analysis of point
source filter.
                                     II-296

-------
  1000 T
 L 100 -
 O
 a
 d
    10 ••
 n
 g
    0.1 I'*!
        Ba  V  Cr Mn  Co  Ni  Cu  Zn  Go  As   Se  Sr  Mo  Cs  T1  Pb

                              Element

Figure 17. Semiquantitative LS-ICP-MS results (ng crrr2) for an SRM-coated cellulose filter.
SUMMARY

Laser sampling ICP-mass spectrometry can be used to produce full, semiquantitative
elemental profiles for elements present at trace to major levels in a wide variety of
samples. In this study, the LS-ICP-MS technique has been applied to the analysis of
solid  sludge samples and  high-volume  air  filters.   The  elemental  fingerprints
produced  by the technique can be used to help  identify which  elements may be
important to monitor in order to identify sources  of environmental contamination.
The technique also  shows  great promise  in being able to determine  important
elements that are difficult to determine by other techniques.


REFERENCES

[1]  E. R. Denoyer, K. J. Fredeen, and J. W.  Hager^a/. Chem.  63, 445A (1991).

[2]  M. Broadhead, R. Broadhead, and J. W. Hager,^f. Spectrosc. 11, 205 (1990).

[3]  E. R. Denoyer and K.  J. Fredeen,  "Application of Laser Sampling  ICP-Mass
Spectrometry to Environmental Analysis,"  1991 European Winter  Conference on
Plasma Spectrochemistry, Dortmund, Germany.

[4]  D. Potter and R. C. Hutton, "Direct Analysis of Airborne Particulate," Paper No.
1075,   1991  Pittsburgh  Conference  on  Analytical Chemistry   and  Applied
Spectroscopy, Chicago, IL.
                                   11-297

-------
                     ICP/MS ANALYSIS OF
   TOXIC CHARACTERISTIC LEACHING PROCEDURE (TCLP) EXTRACT
                ADVANTAGES AND DISADVANTAGES

          Michael G. Goergen. Violetta F. Murshak,
       Paul Roettger, Isaiah Murshak and Dan Edelman
ABSTRACT

     The use of the ICP/MS analytical technique is growing for
various analytical applications.   The authors of  this work
describe  the  advantages  and  disadvantages of  the  ICP/MS
analysis of toxic  characteristic leaching  procedure  (TCLP)
extract.  This  aqueous extract buffered with acetic acid may
cause   interferences   with  the   ICP/MS   qualitative   and
quantitative  analyses.

     This  study presents  the matrix spike recoveries for all
of  the    RCRA  metals  except mercury (Hg).   These  metals
include:  arsenic  (As), barium  (Ba), cadmium (Cd), lead (Pb),
selenium  (Se), and  silver  (Ag),  as well  as  zinc  (Zn)  and
copper  (Cu).   In  addition,  these  results are  compared  to
ICP/AES recoveries.   Based  on these  recoveries  and  other
observations in our laboratory, we conclude that the ICP/MS is
appropriate and convenient for the analysis of TCLP extract.
INTRODUCTION

     The recent change in toxic characteristic analysis from
EP Toxicity to  the  toxic characteristic  leaching procedure
(TCLP) provides analytical environmental chemists with another
matrix for analysis.  Traditionally environmental laboratories
have performed analyses for metals in the TCLP leachate with
atomic absorption spectrometry and inductively coupled plasma
Atomic Emission Spectrometry (ICP/AES).

     Our laboratory has  used ICP/AES since  1988  to perform
analyses on the TCLP extract.  In 1989 we installed an ICP/MS
to perform analyses  of metals.   Our incentive for installing
the ICP/MS was the low detection  limits the method provides
using traditional aspiration for  sample introduction.   In
addition, the  ICP/MS performs the simultaneous qualitative and
quantitative analyses  of up to 70 or so elements.  This allows
for rapid sample through put.   However, we soon discovered
many other advantages  of ICP/MS and it became the analytical
method of preference for metal  analyses in our laboratory.
                            11-298

-------
     This work discusses observations made in our laboratory
regarding the advantages and disadvantages of analyzing TCLP
extract for metals  with an ICP/MS.  They  are  based on real
life samples submitted  to  our  laboratory by clients.   Thus,
this ICP/MS work did not involve a predesigned experiment as
would be done in academic research.
ANALYTICAL METHOD

     The inductively coupled plasma spectrometer (ICP/MS) in
use at FECL is the SCIEX ELAN model  250, converted to a model
500 capability  (Perkin Elmer,  Norwalk, Connecticut).   FECL
also has a Perkin Elmer Plasma 40  Spectrometer  (ICP).   The
ICP/MS system components include  Xenic System V Software, and
IBM Personal  System  2  computer and  an Epson LQ-850 printer.
The various features and conditions of the  system  are shown in
Table 1.

     For  standard  preparation  equal  volumes  of  10  ppm
multielement  stock solution and 10 ppm silver stock solution
are added  together  resulting  in a  5 ppm  solution  for all
metals.   The stock  solutions  are  obtained from PlasmaChem
Associates, Bradley Beach, New Jersey.
     Under FECL's Standard Operating Procedures, samples of 50
ml each are spiked with 1 ml of 5 ppm  solution  resulting in a
0.1 ppm spike concentration  for the various elements.  To each
of the  samples,  2.5 ml of  70% nitric acid (HNOs) is added.
Spiked  and  unspiked samples are  subsequently  digested in a
microwave oven  for 30 minutes at medium  power.   After the
samples  have cooled  down following  digestion, they all
receive  1 ml of 25 ppm  internal standard.  The samples are
diluted  to  their final volume  of 50  ml  yielding a  0.5 ppm
concentration.    The  samples  are then  ready  for internal
standard ICP/MS  analysis.

     The internal  standard  is  made  up by combining 10 ml of
the 1000 ppm solutions for  each of the five standards, i.e.
In, Se, Y, Rh and Re.  The  total volume is diluted to 400 ml
of 25 ppm concentration.   The internal standards  are supplied
by Inorganic Ventures, Inc.  of Toms River, New Jersey.

     For instrument calibration, two calibration  solutions of
0.50 ppm and 0.10 ppm,  respectively, are  prepared weekly from
the multielement stock solution and internal standard.  For
the 0.5 ppm calibration solution 5 ml  of 10 ppm  multielement
stock solution are mixed with 2 ml of 25 ppm internal standard
and bringing the mixture  to  a final volume of 100 ml.  The 0.1
ppm calibration  solution  combines 10 ml of 1.00  ppm
                             1-299

-------
                          TART.B 1

ICP/MS Operating Conditions
ICP torch
Forward power
Argon flowrate (L/min)
  Outer
  Auxiliary
  Aerosol gas
Sampling position
Sampler
Skimmer
Ion lens settings
  Bessel box barrel
  Bessel box plate lens
  Einzel lenses 1 & 3
  Einzel lens 2
  Bessel box stop
Operating pressure
  Interface
  Quadrupole chamber
Data acquisition
Isotopes monitored
Ames laboratory design  (28); outer
  tube extended 30 mm from inner
  tubes
1.25 kW

12
0.5
1.0 - 1.2
22 mm above load coil, on center
Nickel, 1.2 mm orifice
Platinum, 0.90 mm orifice
Upgraded ion optical system
+30 ฑ 5 V
-13 ฑ 3 V
-70 V
-130 V
-30 ฑ 5 V

1 torr
2.3 x 10-6 torr
Multi-element monitor ing mode, normal
  resolution setting; three
  measurements per peak spaced 0.1
  m/z units about peak top; dwell
  time at  each position  is  20 ms,
  with total measurement time of
  0.17 s allows  detection of three
  analytes  per  injection  without
  missing tops of peaks.
B2cr, escu, eezn, ^B^S, ^BSe,
  107-Ag,  niCd, is^Ba, 208Pb
                            11-300

-------
multielement solution and 2 ml of 25 ppm internal  standard and
diluting to a volume of 100 ml.  A calibration blank and check
standard are also prepared weekly.  The blank is  made up of 5
ml  of 1.00  ppm multielement solution and  2 ml  of  25 ppm
internal  standard.    The  mixture  is  added to  100 ml  of
deionized water  to result  in a  0.05 ppm concentration.   The
check standard of  0.2  ppm is prepared  by diluting  2 ml of 10
ppm multielement solution and 2 ml of 25 ppm internal standard
to 100 ml.   The  check standard is used each time after five to
seven samples are run as an independent calibration solution.

     Operation   of  the  ICP  Atomic   Emission   Spectrometer
(ICP/AES) is similar to that  of  the ICP/MS  except the spiking
level of samples is  higher.  The  ICP samples  are spiked with
5 ml  of 10  ppm multielement stock solution to an element
concentration of 1 ppm.

     We use  EPA method  200.7  for  ICP/AES  analysis  and EPA
method 200.8 for ICP/MS analysis.   Furthermore,  we use EPA
method 1311  for  the  TCLP extraction.

     The data presented herein are grouped by sample type and
type of extraction fluid.  Method 1311 requires Fluid ttl which
is more buffered for samples with low pH (pH <5).   The method
requires Fluid #2 when a special HC1 test yields  a  pH of
greater than five.  Figure 1 shows  a flow chart of the TCLP
procedure.
                             11-301

-------
                           Figure 1
         TCLP EXTRACTION  PROCEDURE
     Determine
     Sample pH
     See Start
     Procedure
     Below
                     filtrate <0.5*
                     of total mass
    Determine
    Sample Type
   solid
          Filter and
          Veigh
          Filtrate
                solid portion
  Place Solid
  Portion into
  Borosilicate  jar
u
   pH > 5
              JLI
   Ho Extraction
   Necessary; Filter
   and Continue as is
Store
Filtered
Liquid
\.pH
Add (2L x KFiltrate)
of Fluid tt2
     Add (2L x XFiltrate)
     of Fluid ttl
     Tumble for
     18 Hours
    Remove from
    Tumbler
 Measure pH
 of Extracted Fluid
 if solid
       Re-add Filtered
       Portion to Jar
   Let Settle
   (as Becessary)
      Filter
   Store 250ml
   of Filtrate in
   Plastic Bottle
      Proceed to
      Digestion
       NOTES:


1.  Fluid #1: 5.7ml HOAc/L
  + 64.3ml NaOH/L.


2. Fluid *2: 5.7ml HOAc/L
                    Start Procedure:


                   Add 5g sample to 96.5ml
                   DI Water. If pH > 5,
                   add 3.5ml HC1. Heat to
                   60 C and recheck pH.
                   Use this pH value to
                   determine extraction
                   fluid needed.
                                  11-302

-------
ANALYTICAL RESULTS

     Table  2 summarizes  the  detection  limits  established
according  to CFR  136  Appendix B  for  ICP/MS and  ICP/AES
analysis of  the  TCLP analytes.  It  also provides detection
limits from  the  analytical methods published  in  EPA method
200.7 and 200.8  for comparison.  Note,  for most of the metals
presented there is a significantly lower detection limit for
the ICP/MS.  Indeed in many  cases,  it is one  hundred times
lower, than ICP/AES detection limits.  But, in every case, the
ICP/MS detection  limit is at  least  a factor  of  ten lower.
These  lower  detection limits  offer  one  of  the  significant
advantages of the ICP/MS.  That is,  the ICP/MS has the ability
to analyze a sample diluted to  the point where matrix effects
are minimized and still yields  an acceptable detection limit.

     Tables  3,  4,  and 5  summarize  the  performance  of the
ICP/MS for  known and  blank  spike  samples.   The  knowns are
standards obtained from outside our laboratory certified for
the concentrations shown in the table.  The blank spikes were
dilutions  of  our  standard  solutions.       The  diluted
concentrations are shown in Table 4.

     Tables  3 and  4  show the accuracy obtained with the
ICP/MS.   Standard  deviations of less  than  nine  percent are
obtained in every  case.  Table 5  shows the precision of the
ICP/MS from  the  analysis of duplicate blank spikes.   These
tables summarize the accuracy and precision that we routinely
observe with the ICP/MS on these standards.

     Tables  6, 7  and 8 show the percent recoveries  for the
metals using the  ICP/MS and ICP/AES.  These tables are grouped
by sample type and the type  of extraction fluid  used in the
TCLP.   In Tables 6 and 8 the standard deviations for all of
the metals, except for arsenic, copper and zinc in Table 6 and
for cadmium and selenium in Table 8 are lower with the ICP/MS.
For the soil matrix in Table  7  ICP-40 provided  lower standard
detections  for  most  of the  metals.   However,  the  ICP/MS
samples were spiked with 0.1  ppm of analyte  compared to 5 ppm
for the ICP/AES analysis.

     The  ICP/AES,  of  course,  is  an established  method for
toxic  characteristic  analyses.    So,   these  observations
indicate that the  ICP/MS is also a  useful  method for these
analyses.

     Tables 9, 10,  11  and  12 show  the spike recoveries from
the  ICP/MS analysis of  TCLP extract  for four of our  most
common  sample types.    Again,  these  tables  group  similar
extraction fluid and sample types.   Table 9 presents the
                             1-303

-------
As

Bo

Cd

Cr

Cu

Pb

Se

Ag

Zn
                        TABLE 2
                Method Detection Limits
ICP-MS(mg/l)
Calculated1 EPA Estimated 2
0.0015   0.0009
0.0015   0.0005
0.0009   0.0001
0.0005   0.00007
0.0012   0.00003
0.0007   0.00008
0.0047   0.0050
0.0032   0.00005
0.0032   0.0002
                                       ICP-40(mg/l)
                                      Calculated1  EPA Estimated 3
0.20
0.053
0.02
0.01
0.04
0.01
O.OB
0.50
0.05
0.02
0.002
0.004
0.007
0.006
0.042
0.075
0.007
0.002
 1.  Using CFR136 Appendix B. Based on 10 Blank Spikes of O.OIppm

 2.  From EPA Method 200.8.

 3.  From EPA Method 200.7.
                           11-304

-------
                               TABLE 3

               Quality Control Knovns by ICP/MS X Recovery

ICP/MS
                               QC-ICP InowoB

        •g/1   SET II  SET 12  SIT 13  SET 14  SET 15    Mean    S*
As
Ba
Cd
Cr
Cn
Pb
Se
Ag
Zn
0
0
0
0
0
0
0
0
0
.2
.2
.2
.2
.2
.2
.2
.2
.2
97
99
98
97
99
101
101
103
100
104
106
103
104
102
103
104
100
96
96
98
95
97
98
101
98
94
97
96
105
98
98
96
96
113
111
95
101
98
102
99
99
101
104
114
107
98.
102.
98.
99.
98.
100.
104.
102.
97.
80
00
50
00
75
25
00
00
00
3.
4.
3.
2.
2.
2.
5.
8.
5.
56
06
36
92
17
61
61
57
34
                               TABLE 4

               Blank Spikes by ICP/MS X Recovery

As
Ba
Cd
Cr
Cn
Pb
Se
Ag
Zn
•g/1
SET fl SET 12 SET t3 SET 14 SET 15
0.05
0.
0.
0.
0.
0.
0.
0.
0.
05
05
05
05
05
05
05
05
104
104
110
100
108
108
108
110
108
100
102
100
98
99
100
114
100
111
96
104
100
94
100
104
98
107
95
108
105
96
95
98
98
98
100
99
94
102
99
100
99
100
97
96
100
Mean
100.
103.
101.
96.
101.
102.
104.
104.
103.
40
75
50
75
25
50
50
25
25
5
1
5
2
4
4
7
6
6
S
.73
.40
.32
.89
.12
.04
.80
.02
.70
                               TABLES

               Blank Spike Difference in Duplicates by ICP/MS
As        0.05
Ba        0.05
Cd        0.05
Cr        0.05
Cu        0.05
Pb        0.05
Se        0.05
Ag        0.05
Zn        0.05       0     4.8     1.6       0     6.8   1.60  3.26

*S = Standard Deviation
1 SET 12 SET 13 SET 14 SET 15
0
0
0
0
0
0
0
0
0
3.4
0
0
0
0
0
0
0
0
0
0
1.9
0
0
0
0
4.1
0
0
0
0
0
0

1


5
1


0
.2
0
0
.9
.0
0
0
Mean
0
1
0
0
0
0
0
0
.00
.88
.00
.00
.47
.00
.00
.00
S
0.00
1
0
0
2
0
0
0
.92
.00
.00
.83
.50
.00
.00
                               11-305

-------
ICP-40 & ICP/MS
                                               TABLE  6

                Spike Becoveries in % for TCLP Analysis of Industrial Faint Sludge
                                Comparison Between  ICP/MS and ICP-40
SET 14
ICP/MS
IHDUSTilAL PAIHT SLUDGE
Em.l     1       1
         6711-E1 6711-E2
Mean
S*
Saiple
As
Ba
Cd
Cr
Cu
Pb
Se
Ag
Zn
1
180
80
100
110
80
90
175
82
170
2
168 	
92 	
100 	
98 	
98 	
84 	
160 	
80 	
90

	 174.00
	 86.00
	 100.00
	 104.00
	 89.00
	 87.00
	 167.50
	 81.00
130.00

8.49
8.49
0.00
8.49
12.73
4.24
10.61
1.41
56.57
                                                            ICP-40
                                                            INDUSTRIAL PAINT SLUDGE
                                                            EXTLI     1       1
                                                                    6711-E1 6711-E2
Mean
S*
Saiple
As
Ba
Cd
Cr
Cu
Pb
Se
Ag
Zn
1
85
114
86
98
93
90
130
10
94
2
88
88
75
83 	
78 	
69
107
7
80

	 86.50
	 101.00
	 80.50
	 90.50
	 85.50
79.50
118.50
	 8.50
87.00

2.12
18.38
7.78
10.61
10.61
14.85
16.26
2.12
9.90
                                               TABLE?

                        Spike Becoveries  in X  for TCLP Analysis of Soils
                                Comparison Between  ICP/MS and ICP-40
ICP/MS
SOILS
EITE.I
Saiple
As
Ba
Cd
Cr
Cu
Pb
Se
Ag
Zn
2 2 2
6704-E1 6704-E2 6734-E1
1
149
R!J
106
110
94
99
159
81
190
2
136
100
100
94
85
93
131
80
70

118
90
80
76
73
75
123
38
80
Mean

134.33
95.00
95.33
93.33
84.00
89.00
137.67
66.33
113.33
S*

15.57
7.07
13.61
17.01
10.54
12.49
18.90
24.54
66.58
                                                            ICP-40
                                                            SOILS
                                                            EITB.t     2       2       2
                                                                    6704-E1 6704-E2 6734-E1
                                                                       S*
Saiple
As
Ba
Cd
Cr
Cu
Pb
Se
Ag
Zn
1
99
169
90
94
90
88
87
64
107
2
110
108
91
100
93
80
105
13
91

87
71
78
74
77
73
116
107
74

98.67
116.00
86.33
89.33
86.67
80.33
102.67
61.33
90.67

11.50
49.49
7.23
13.61
8.50
7.51
14.64
47.06
16.50
*S : Standard Deviation
                                                  11-306

-------
                                             TABLE 8

                      Spike Recoveries in X for TCLP Analysis of WTP Sludge
                              Comparison Between  ICP/MS and ICP-40

SIT 15
ICP/MS                                                    ICP-40
WTP SLUDGE                                               WTP SLODGE
im.S   1111                        EXTR.I     1111
      6613-11 6613-E2 6613-13 6613-E4    Mean     S*                6613-E1  6613-E2 6613-E3 6613-E4     Mean       S*
Saiple   1       2                                        Saiple     1       2
As        110     105     100     104  104.75   4.11       As          104      94      103       76   94.25    12.97
Ba        100     100     100     100  100.00   0.00       Ba            58      74       89       59   70.00    14.63
Cd         78      74      80      80   78.00   2.83       Cd            86      85       85       81   84.25     2.22
Cr        141     141     141     135  139.50   3.00       Cr          100      96       89       87   93.00     6.06
Cn         90      84      77      75   81.50   6.86       Cu          104      85       89       85   90.75     9.03
Pb         77      75      78      77   76.75   1.26       Pb            86      87       75       78   81.50     5.92
Se         89      81     128     130  107.00  25.63       Se          100      100       86       90   94.00     7.12
Ag         68      69      70      70   69.25   0.96       Ag            47      15       20        4   21.50    18.27
Zn        100     100     100     100  100.00   0.00       Zn            85      62       70       67   71.00     9.90

*S = Standard Deviation
                                                        1-307

-------
                                                     TABLE 9

                          Spike Recoveries in X on TCLP Extract of soils by ICP/MS
                                    with Mean and Standard Deviation(s)

SOILS
ERR*  2222      22222222222222      Mean    Sซ

As    112  117   119   130  179  104   124   146  105  121  116  106   94   112   97  118   136  149   121.39  20.24
Ba    140  220   270   100  117  100   100   100  100  100  100   72   100   100  100   90   100  149   118.18  48.78
Cd    100  96   103   102  116  104    94   112   95  109  103   99   101    98  103   80   100  106   101.17   7.54
Cr    135  145   130   118  133  126   115   145  128  128  126  109   103   114  107   76    94  110   119.00  17.13
Cn     95  103   100    82   89  100    82    89   76   83   92   78   76    79   78   73    85   94    86.33   9.02
Pb    100  60    73    76   79   95    74    84   74   81   83   79   82    77   82   75    93   99    81.44   9.79
Se    147  142   153   157  199  130   127   116   96  116  106  103   98   100   99  123   131  159   127.89  26.88
Ag     79  70    77    71   72   90    86    96   91   86   90   79   91    76   89   38    80   81    80.11  12.67
Zn    100  80    95    80  105  100    59    72   50   86  140  100   71    54   80   80    70  190    89.56  31.98

                                                     TABLE 10

                          Spike Recoveries on TCLP Extract of WTP sludge
                                    with Mean and Standard Deviation!s)
WTP SLUDGE
EM*
As
Ba
Cd
Cr
Co
Pb
Se
Ag
Zn
1
94
106
103
94
91
79
118
114
185
1
97
35
102
100
88
82
122
109
85
1
102
99
100
101
86
80
135
105
80
1
137
108
100
117
92
84
154
115
88
1
117
100
92
111
76
55
148
84
160
1
135
100
99
128
85
70
141
101
100
1
130
94
106
104
86
52
86
86
100
1
80
62
82
106
80
82
72
74
100
1
110
100
78
141
90
77
89
68
100
1
105
100
74
141
84
75
81
69
100
1
100
100
80
141
77
78
128
70
100
1
104
100
80
135
75
77
130
70
100
Mean S*ซ
109 16.7
94 18.7
91 11.2
118 17.2
84 5.7
74 9.9
117 26.8
89 18.1
108 30.0
                     - TCLP Extraction Fluid f (i.e. Fluid II or Fluid 12)
               **S    = Standard Deviation
                                                     11-308

-------
                           TABLE 11

     Spike Recoveries on TCLP Extract of Industrial sludge
          with Mean and Standard Deviation(s)

     INDUSTRIAL SLUDGE
     EXTR*  1     1     1111   Mean   S**

     As     135   116  104  120  180  168   137  27.8
     Ba     110   100   98  100   80   92    97   9.1
     Cd      95   104  104   91  100  100    99   4.7
     Cr     130   100  106  104  110   98   108  10.6
     Cu      89    92   90   96   80   98    91   5.8
     Pb      74   120   72   72   90   84    85  16.9
     Se     147    96  114  125  175  160   136  27.2
     Ag      91   100   92   84   82   80    88   6.9
     Zn     100    96  100   34  170   90    98  39.5

                           TABLE 12

     Spike Recoveries on TCLP Extract of Industrial sludge
                with Mean and Standard Deviation(s)

                INDUSTRIAL SLUDGE
                EXTR*   2    2  Mean  S**

                As     138   91  115 23.5
                Ba
                Cd
                Cr
                Cu
                Pb
                Se
                Ag
                Zn
*EXTR = TCLP Extraction Fluid # (i.e. Fluid #1 or Fluid
**S    = Standard Deviation
100
102
86
100
86
145
100
100
100
100
89
80
77
89
109
100
100
101
88
90
82
117
105
100
0.0
1.0
1.5
10.0
4.5
28.0
4.5
0.0
                              1-309

-------
percent spike recoveries for soils  extracted with Fluid #2.
The highest standard deviation  of all the metals shown is for
barium.  It is 1.5 times the next  higher standard deviation
(44 vs. 32).  However, in Tables 10,  11  and 12 the standard
deviations  for Barium are more  than two  times lower-   We
cannot explain this.

     The standard deviation  for wastewater  treatment plant
sludge (Table 10) and industrial sludge extracted with Fluid
#1 (Table  11) and Fluid #2 (Table 12) show standard deviations
less than  30 percent for all metals.  This is at sample spike
levels of  0.1 ppm analyte concentration.

     The percent  spike recoveries presented here are typical
of those we observe in our TCLP analysis.   Please note that
none   of   these   samples  have   been  tested   for  matrix
interferences.     Indeed,   the  samples  themselves  have  a
significant impact on  the recovery  of the metals.  The ICP/MS
performs about the same for each of the four sample types.
OQHCTJJSIOMS

     Based on the data summarized  in  this work and on other
observations made in our laboratory we find the ICP/MS to be
an  appropriate   and  convenient  method  for analyzing  TCLP
extracts.   We find the advantages of the method out-weigh the
disadvantages.  The major advantages and disadvantages of the
ICP/MS analysis  which we've identified are summarized in Table
13.
                            f 1-310

-------
                          TABT.K ia

              ICP/MS Analysis of TCLP Extracts

                Advantages and Disadvantages
Advantages:
       Fast - 3 readings per sample for 10 metals in less than
       20 seconds
       Able to verify calibrations at any place in the run
       and then continue
       Recalibration to compensate for drift by internal
       standards takes only a few seconds
       High sensitivity allows diluting to minimize matrix
       effects
       Avoids atomic emission interferences
       Uses only 2 ml of sample for analysis
       Able to screen samples semi-quantitatively
       High sensitivity requires significant diluting of
       higher analyte concentrations to be within the
       calibration range (or switch to high concentration
       mode)
       Operator must be knowledgeable in recognizing and
       correcting interferences
                            11-311

-------
oc                               SEVENTH  ANNUAL  WASTE  TESTING
ฐ^                             AND QUALITY ASSURANCE SYMPOSIUM
                                            ABSTRACT
         Submitted  by:
         Larry B. Lobring,  Chief
         Inorganic  Chemistry  Branch,  Chemistry Research Division
         Environmental  Monitoring  Systems  Laboratory  -  Cincinnati
         Office of  Research and Development
         U.  S.  Environmental  Protection Agency
         (513)  569-7372,  FTS  684-7372
         Title:
         Chromium VI;   An Overview of Its  Relevant  Environmental Occurrence,  Analytical
         Methods of Quantitation,  and Report on Recent  Ion  Chromatography Methods
         Development and  Validation Activities.
         This  presentation  covers  the various forms of  chromium found in nature and
         those that are significant in environmental  samples and to human and ecosystem
         health.  The  interconversion of trivalent  and  hexavalent  chromium in the
         environment and  related problems  associated  with sample collection,
         preservation  and quantitation of  the various species is discussed.   Topics
         covered include  the  current  analytical methodology that utilizes
         chelation/extraction or coprecipitation  with iron  or lead.  These approaches
         have  several  potential chemical interferences  or deficiencies that  are
         discussed.
         A description  of recent  methods  development  studies,  utilizing ion
         Chromatography and  inductively coupled  plasma mass spectroscopy,  for the
         determination  of total and  hexavalent chromium in incinerator particulate
         emissions is presented.  The ion  Chromatography method developed in this study
         was adapted for use in aqueous environmental  samples  and is now available for
         use in the Environmental  Protection  Agency's  compliance monitoring programs.
         The water method is identified as  Method 218.6, " Determination of Dissolved
                                          11-312

-------
Hexavalent Chromium in Drinking Water, Ground Water and Industrial Effluents
by Ion Chromatography". Results of a recently completed multi-laboratory
method validation study conducted in cooperation with ASTM are presented.

Additional efforts needed in the area of sample processing to extend
application of the technique to a wider variety of sample types will also be
presented.
                                  1-313

-------
og    Rapid High Performance Microwave Digestion

       Ron Rubin. Michael Moses,  Questron Corporation, PO Box 2387,
       Princeton, KJ  08543-2387

       Microwave Digestion Techniques have reduced the time required to
       place a sample in solution.  However, there are still limitations  in
       sample handling, cooling and recoveries of elements*  In this paper
       we will present several different types of Digestion Systems and show
       how each of them addresses the above mentioned problems.  To be
       covered in the study are:  Closed Vessel Microwave Ovens; Open Vessel
       Microwave; High Pressure conventional Digestion, and High Pressure
       Microwave Digestion.  Comparisons are made based upon what we
       consider the two most important operating factors:  throughput and
       recovery.  Throughput encompasses all of the cost factors such as
       speed of digestion, speed of cooling, number of samples per batch,
       amount of reagents and operator time*  Recovery, especially its
       reproducibility, defines the success or failure of  the procedure.
                                   1-314

-------
87
                  THE PERFORMANCE OF A LOW COST ICP-MS
            FOR THE ROUTINE ANALYSIS  OF  ENVIRONMENTAL  SAMPLES

R. Craig Seeley. Thomas M. Rettberg,  Peter D.  Blair,  Fisons Instruments,
24 Commerce Center, Cherry Hill Drive, Danvers, Massachusetts  01923


ABSTRACT

Inductively Coupled Plasma Mass Spectrometry  (ICP-MS) is  rapidly being
recognized as the choice of instrumentation for trace element analysis of
environmental samples.  It possesses  a number of distinct advantages when
compared  with  other  established  techniques:    sensitivity,  speed  and
versatility.

This presentation will demonstrate the performance and cost effectiveness
of  the  new Fisons PQe  ICP-MS.   The  system  was designed to meet  the
specific  and varying needs  of laboratories  concerned  with  analyzing
environmental samples.  Data will be presented on water and soil samples
according  to  USEPA Methodology  6020 showing  the  PQe's  multi-element
capability  and  determination of  all elements  in a  single acquisition.
Detection limit studies  will  be presented along with information on the
profit potential of the  PQe  when used to perform contract  analysis of
environmental samples.


INTRODUCTION

For a laboratory to enter  the Environmental Protection Agency's Contract
Laboratory Program, it must have the  capability to determine 22 metals in
water and soils.  Until  recently, most laboratories have been  using a
combination  of ICP-OES   and  GFAA   instrumentation.    The  choice  of
instrumentation is  made  on  the  basis  of  the  CLP contract  required
detection limits (CRDL's).  The ICP-MS technique  is particularly suitable
for  environmental  analysis   due   to   its   exceptional  multi-element
sensitivity.  ICP-MS can meet or exceed the CRDL's for all 22 elements in
Method 6020 (with the exception of Se in soil)  and perform the analysis in
one sample cycle.  Two instruments are now combined in one.


PROFIT POTENTIAL AND DESIGN

Laboratories performing  high  through-put routine environmental analysis
have been reluctant  to  invest in such  technology  because of  the high
capital  cost  of  current  ICP-MS  instrumentation.    However,  using  a
completely new  approach to  ICP-MS, Fisons has  introduced the new PQe.  A
customized,  low-cost  ICP-MS  instrument  specifically   aimed  at  the
environmental market.   The instrument is based  on a  radical  new design
which emphasizes robustness of hardware and simplicity of operation.  A
completely new  mass spectrometer and detector system is employed and the
benefits to the analyst are detection limits and a dynamic range more than
sufficient  to   meet  EPA  legislative  requirements1.     Profitability
projections will be presented in this poster/paper to help illustrate the
cost effectiveness of the  new ICP-MS design.


ANALYTICAL REQUIREMENTS FOR WATER ANALYSIS

The  term "water analysis" covers a wide variety  of sample  types  and
matrices.     In  particular,   environmental   water   samples   may  vary
significantly in terms  of inorganic  and organic dissolved solid content
                                          1-315

-------
and suspended material.  The PQe has a range of features which enable it
to deal routinely with  this wide range of sample types, for example matrix
independent calibration and wide linear dynamic range.

Matrix independent calibration
The PQe requires only a single set of calibration standards, even for the
analysis of different  sample matrix types,  such as  rain water,  riverine
waters and effluent.  Matrix independent  calibration obviates the need to
run different  standards for each  type  of  matrix,  or  perform  standard
additions on each sample, thus saving valuable analysis time.

Wide Linear Dynamic Range
Environmental water samples may  include  analytes  at high concentrations
e.g. Na, Mg, K and Ca as well as  the  trace and ultra-trace components e.g.
Cr, Cd, Tl and U.  For efficient sample analysis,  it  is essential that all
the elements  of interest  should be  determined  in  the same  solution,
without the need for preconcentration, separation or dilution.   The wide
linear dynamic range of the PQe allows the determination of major, minor
and trace elements  in a single acquisition, without the need for operator
input.  Furthermore,  there is no necessity to match the concentration of
each  analyte  in  the  calibration  standard  to  the  expected  sample
concentration,   thus  simplifying  calibration  procedures  and  further
improving sample through-put2.


ANALYTICAL PERFORMANCE

A series of experiments were carried out  to assess the performance of the
PQe in terms of  accuracy, precision,  spike  recovery,  stability,  dynamic
range and detection limits.  Data will be presented on certified materials
as well as routine water and soil digested samples.


SUMMARY

The  EPA  Contract  Laboratory  Program  protocol  for  inorganics  is  a
complicated program to  enter successfully.   With the help  of  low cost,
simplified, and  high sample through-put instrumentation,  it becomes  a
straight forward and profitable task.
References:

^e,  C.T.,  et al,  1991  Pittsburgh Conference, March 3-8,  1991

2PQe Technical Note 2, VG  Elemental,  Winsford, Cheshire, UK,  1990
                                  11-316

-------
Robotics For Automated Digestion Of Environmental Samples

C.Balas7 Questron Svsteiaa. A. Grfrllo,Questron Corporation, P.O. Box
2387, Princeton, NJ  08543-2387
Microwave Ovens are  ideal for preparation of Environmental samples
for metals analysis.  However, the oven does present problems of
vessel handling and  storage.   A new robotic system, utilizing
several microwave  stations, has been configured to digest samples, at
a sufficiently high  rate, to enable the digestion to keep up with the
pace of a simultaneous  ICP  system.  Protocols  for many different
types of samples can be stored, recalled/ and  implemented,  in order
to allow the  robotics to accept many different samples of various
sizes and consistencies. In our paper we will describe  the software
and protocols and  show  how  they can be utilised to accommodate  the
day to day changes in the types and quantities of samples encountered
in the typical environmental  laboratory.
                             1-317

-------
QQ      APPLICATION OF LASER SAMPLING ICP-MASS SPECTROMETRY TO ENVI-
0       RONMENTAL ANALYSIS
                        K. J. Fredeen, R. J. Thomas
          Perkin-Elmer Corporation
          Norwalk, Connecticut
          Laser Sampling ICP-Mass Spectrometry is increasingly becoming recognized as an
          analytical tool for the direct analysis of solid samples.  Early work with LS-ICP-MS
          focused mainly on geological  and metallurgical type applications mainly because of
          the ability to bypass the  lengthy sample dissolution stage.

          However as the technique progresses, other application areas for LS 1C? MS  nre
          becoming more and more attractive.  One such  area is in the analysis of environ-
          mental  type applications.  The ability to  bind and/or press  samples into a small
          pellet allows LS-ICP-MS to be used for the analysis of samples such  as urban
          particles or  river sediments.

          This work will discuss some of the capabilities and limitations of LS-ICP-MS for
          the analysis of these type of  environmental samples.  In addition, approaches to
          the difficult problem of sampling some of these materials will be discussed.
          RT:td.329
                                              11-318

-------
REGULATORY
COMPLIANCE

-------
QQ             Status of Developing Land Disposal Restrictions
                             for Superfund Soils


                                Richard Troast
                                Carolyn Offutt

                    U.  S. Environmental Protection Agency
                 Office of Solid Waste and Emergency Response
                                Washington,  DC

                              Joan O'Neill Knapp

                       CDM Federal Programs Corporation
                              Fairfax, Virginia
                                  ABSTRACT



      RCRA Land Disposal  Restrictions  (LDRs)  for  contaminated  soil  and

      debris at Superfund sites  are  currently being developed.   This

      paper will present  the  current status  of the  EPA sponsored

      testing  and  the  design  of  an integrated data  base for  both

      technology transfer and the development of  the LDRs.
                                         /


      The unique physical and chemical characteristics of  Superfund

      soil and debris  make these wastes more difficult to  treat  than

      more homogeneous industrial process wastes.   The National

      Contingency  Plan acknowledges  that Best Demonstrated Available

      Technology  (BOAT)  standards are generally inappropriate  for

      Superfund soils.  In response  to this,  EPA  is in the process  of

      developing  separate LDR standards for  contaminated soil  and

      debris  (CSD).  LDRs for CSD are being  developed under  section

      3004 of  the  Hazardous and  Solid Waste  Amendments of  1984 to RCRA.

                                       1

                                   11-321

-------
Until the final CSD standards are in place,  treatability variance



levels,  also based on the actual treatment of soil,  will be



used.      In addition,  the paper will discuss some preliminary



findings on the treatment of debris, and the analytical methods



used for determining the BDAT for CSD.







1.0  INTRODUCTION







Section 3004(m) of the Resource Conservation and Recovery Act



(RCRA) mandates that the U. S. Environmental Protection Agency



(EPA) require treatment of hazardous wastes prior to land



disposal.  Known as the "land disposal restrictions" (LDRs),



these regulations were designed for industrial process wastes



defined to be hazardous under RCRA.  They apply as well to



contaminated soil/ sludge and debris from RCRA facilities and



Superfund sites.  RCRA requirements for treatment are mandatory



and self-implementing at all RCRA regulated facilities, but apply



at a CERCLA site only if a) the waste is a RCRA listed or



characteristic waste; b) the CERCLA activity constitutes



treatment of RCRA hazardous waste, as defined under RCRA; and c)



the treatment activity constitutes  "placement."







The Office of Solid Waste (OSW) is responsible under EPA's Office



of Solid Waste and Emergency Response (OSWER), for responding to



directives under RCRA, and therefore, prepares and presents the



LDR standards to the regulated community.





                                2



                            11-322

-------
The Office of Emergency and Remedial Response (OERR) is



responsible under OSWER for responding to directives under the



Comprehensive Environmental Response, Compensation, and Liability



Act (CERCLA), or Superfund activities.  As the majority of soil



and debris contaminated with hazardous wastes are found on



Superfund sites, LDRs have a profound potential effect on the



government's efforts at site remediation.







OSWER has recognized that contaminated soil is more difficult to



treat than RCRA industrial process wastes, and that it is not



likely that these wastes can be treated to meet the LDRs



developed for RCRA listed wastes because of the physical and



chemical complexity of contaminated soils.  In response, OSWER



initiated a program to develop Treatability Variances, which are



alternate treatment standards based on actual treatment of



Superfund and RCRA soil and debris.  Data was collected, and in



1989, treatability variance levels were established for soils



utilizing 67 data sets (Superfund LDR Guides #6A and #6B).







OSWER developed a strategy for calculating variance levels from a



quantity-limited data base.  The data are categorized into 13



"contaminant groups" which are groups of contaminants having



similar chemical and physical characteristics.  Examples of



contaminant groups include non-polar halogenated aromatics, and



PCBs/dioxins/furans and their precursors.  The variance levels
                                3




                            11-323

-------
that were developed quantified the effectiveness of various
available technologies on the contaminant groups.

EPA OSWER determined that the existing soil treatment data base
was not comprehensive enough to support a formal set of LDRs for
CSD.  Several available technologies had insufficient performance
data to develop regulations.  EPA therefore implemented a
research program to obtain all of the necessary data to support
the development of LDRs for CSD.   In 1988, OSWER's OERR, OSW, and
Technology Innovation Office (TIO), and the Office of Research
and Development's (ORD) Risk Reduction Engineering Laboratory
(ORD-RREL) in Cincinnati, Ohio established a work group to
develop BOAT standards for CSD.  The work group objectives
include a review of the current data base, recommendations for
additional studies on treatment performance, implementation of
treatability studies, collection of new available data, and
development of BOAT regulations based upon new and available
data.  There has been significant progress with these efforts.

2.0  DATA COLLECTION/DATA BASE DESIGN AND OPERATIONS

OSWER, in its initial data collection effort, collected and
examined over 500 studies conducted by the EPA, federal agencies,
industries and universities.  These studies formed the basis for
the development of treatability variances.  Of these studies, 67
met the criteria established for the development of variance

                                4
                            11-324

-------
levels for contaminated soils.  The established criteria required



that the: (1) data be of sufficient quality; and (2) untreated



and treated soil contamination be measured.  The current criteria



for setting final LDR treatment standards are more rigorous than



the criteria for variance levels.  They require more



documentation of quality assurance/quality control (QA/QC)



procedures as well as bench, pilot and full-scale testing data.



A formal data summary form  (DSF) has now been developed by OSWER



to extract pertinent data from all studies reviewed for inclusion



into the data base.







EPA utilizes a four-tiered project category approach in its QA



program in order to more effectively focus QA.  Category I



involves the most stringent QA approach, whereas Category IV



represents the least stringent.  Category  II projects are those



producing results that complement other inputs and are designed



for use in rulemaking, regulation making,  or policy making.



Therefore, all data used to support the CSD LDRs should have a



Category II objective designed into the QA project plan (QAPjP).







After a thorough QA review using the established criteria, only



13 of the 67 studies used for variance levels were determined to



be adequate for consideration in the development of LDR treatment



standards.  However, all studies reporting data are accepted as



Category IV data and included in the data  base for technology



transfer purposes.





                                5



                            11-325

-------
Lack of soil treatment data prompted an aggressive data



collection effort by OSWER and ORD.   Figure 1 shows the system



for data collection and treatment research in the CSD program.



Additional data will be collected from recent remedial/removal



actions/ including DOD and DOE actions/ SITE program



demonstrations, and treatability tests conducted by the CSD



program.  Currently the new data base is planned to contain not



only the original data base/ but studies that have been collected



since the variance levels were published as well.  OSWER will



also use the data base to manage technology transfer information



collected during this project.







This new EPA data base, the Superfund Soil Data Management System



(DMS) is an important tool for fostering technology transfer



involving contaminated soil/ debris and sludge and relating the



information to applicable LDRs of HSWA which are applicable or



relevant and appropriate requirements  (ARARs) to Superfund



actions.  The Superfund Soil DMS will allow maximum utility of



the data obtained from any source.  Data meeting a minimal



criterial will be included in the data base.







The data base construction allows for easy user access and



tailoring of reports to individuals' needs.  Sorting will allow



questions concerning technology, waste characteristics/ soil



matrix and other parameters to be addressed.
                                6



                            11-326

-------
CO
ro
                            ORIGINAL SUPERFUND
                              SOIL TREATMENT
                                 DATA BASE
                                                ALTERNATE
                                               TREATABILITY
                                                 VARIANCE
                                                  LEVELS
   GUIDANCE
  DOCUMENT &
OSWER DIRECTIVE
                                                         TREATMENT
                                                           TESTS
EXISTING
 DATA
                           REVISED
                          DATA BASE
           WEATHERED
             SLUDGE
                                            VARIABILITY
                       CSD
                       LDRs
                                 TECHNOLOGY
                                   TRANSFER
                   Figure 1. EPA OSWER Data Collection and Research Approach

-------
Outside access to the Superfund Soil DMS will be through a



central EPA system.  At this time this is envisioned to be the



Agency's ATTIC System which is being managed at the Agency's



Edison NJ laboratory facility.







3.0  SOIL TREATMENT TESTS







The CSD Program reviewed existing data and identified



technologies that lacked treatment performance data, but would be



available technologies for treating CSD (Table 1).  Twelve



treatment tests are planned.  The technologies that will be



tested include slurry bioremediation, low temperature thermal



desorption, chemical extraction, soil washing, and stabilization



(Table 2).  The technologies are applied to different types of



soils and wastes.  For example, the biotreatment tests will be



conducted on three soil types.  The soil classifications range



from sandy to clay type soils.  In addition, different types of



wastes, including wastes high in PNAs, PCBs and metals, will be



tested.  The stabilization technology will be tested as both a



primary technology and as a residual treatment.







The treatability tests will be conducted according to the OSW



Quality Assurance Project Plan for Characterization Sampling and



Treatment Tests Conducted for the Contaminated Soil and Debris



Program (QAPP) and site specific Sampling and Analysis Plans.



The individual sampling plans specify holding times, analytical





                                7



                            11-328

-------
CO
                                 JHORKMEHATION
                                        IMMOBILIZATION
BBCHLORJNATTON
       TREATABIL1TY
       GROUP
                                                                                                                                    TECHNOLOGIES
         NON-POLAR HALOGENATED
              AROMATICS
                 (W01)
           fCB*. HALOGENATED
           D1OJONS, FVRANS, AND
            THEIR PRECURSORS
                 (W02)
          HALOGENATED PHENOLS.
         CRBSOLS, AMINES, THIOLS,
            AND OTHER POLAR
            AROMATICS (W03)
             HALOGENATED
          ALIPHATIC COMPOUNDS
                 (W04)
          HALOGENATED CYCLIC
          AUPHATICS, ETHERS,
          ESTERS, AND KETONES
                (Wป5)
          NITRATED COMPOUNDS
                 (W0ซ)
  HETEROCYCUCSAND
SIMPLE NON-HALOGENATED
     AROMATICS
        (WOT)
             POLYNUCLEAR
              AROMATICS
                 (W0ป)
              OTHERPOLAR
            NON-HALOGENATED
           ORGANIC COMPOUNDS
                 (W0ป)
             NON-VOLATILE
                METALS
                 (W10)
               VOLATILE
                METALS
                 (Wll)
                 HZ)
                EXISTING DATA EXPECTED
                                                              CSftDTESTDATA
                                  X
                                                                                                 INDICATES TECHNOLOGY IS NOT EXPECTED TO BE EFFECTIVE
                                                 Table 1: Available Soil Treatment Technologies

-------
     SITE
 SOIL TYPE
 CONTAMINANTS
  TECHNOLOGY
Jennison-Wright
Jennison-Wright
Jennison-Wright
Bayou Bonfouca
Bayou Bonfouca
New Hampshire
Brown's Battery
Burlington Northern
Burlington Northern
Ninth Ave.
MIDCO
C&R Battery
Clayey
Clayey
Clayey
Silty
Silty
Silty
Silty
Silty, Sandy
Silty, Sandy
Sandy
Sandy
Sandy
Organics
Organics
Organics
Organics
Organics
Metals
Metals
Organics, Metals
Organics, Metals
Organics, Metals
Organics, Metals
Metals
Bioremediation
LTTD
Solvent Extraction
LTTD
Solvent Extraction
Soil Washing
Stabilization
Bioremediation
LTTD
Bioremediation
Sandy
Soil Washing
                               Table 2. Planned Treatment Tests

-------
methods, chain of custody, and quality control measures, such as



blanks and spikes.  The tests will include measurements of



contaminant concentrations before and after treatment, and



measurements of the waste characteristics that affect the



performance of soil treatment technologies.  Examples of waste



characteristics that affect treatment performance such as



moisture content, oxidation/reduction potential, and particle



size distribution are listed in the QAPjP.







4.0  DEBRIS







OSWER collected existing data on debris treatment in their data



collection program.  The study determined that debris could



constitute as much as fifty percent of the contaminated media,



such as at a wood preserving site.  The study also found that the



sampling procedures were not well documented.  Recognizing the



importance of debris, the CSD Program has implemented a



comprehensive review of debris sampling, analysis and treatment.



The characteristics of debris that have been determined to affect



treatment include permeability and destructibility.  The



potential treatment technologies have been generalized into three



categories for debris: 1) destruction, 2) extraction and removal,



and 3) sealing/solidification (Table 3).  The Agency will discuss



the use of specified-technology standards for debris remediation



in an upcoming Advanced Notice of Proposed Rulemaking (ANPRM).
                                8



                             1-331

-------
"^^^^CONTAMINANT
DEBRK^^GROUPS
MATRICES ^^
PERMEABLE
DESTRUCTIBLE
PERMEABLE
NON-DESTRUCTIBLE
NON-PERMEABLE
DESTRUCTIBLE
NON-PERMEABLE
NON-DESTRUCTIBLE
ORGANICS *
(EXCEPT NITRATED
ORGANICS)
Destruction
Chemical Extraction,
Physical Removal,
Sealing/Solidification
Chemical Extraction,
Physical Removal,
Destruction,
Sealing/Solidification
Chemical Extraction,
Physical Removal,
Sealing/Solidification
NITRATED
COMPOUNDS
Destruction
Chemical Extraction,
Physical Removal
Chemical Extraction,
Physical Removal,
Destruction
Chemical Extraction,
Physical Removal
METALS
Chemical Extraction,
Physical Removal,
Sealing/Solidification
Chemical Extraction,
Physical Removal,
Sealing/Solidification
Chemical Extraction,
Physical Removal,
Sealing/Solidification
Chemical Extraction,
Physical Removal,
Sealing/Solidification
CYANIDE
Destruction
Chemical Extraction,
Physical Removal,
Sealing/Solidification
Chemical Extraction,
Physical Removal,
Destruction,
Sealing/Solidification
Chemical Extraction,
Physical Removal,
Sealing/Solidification
ft
ro
   * Organics include volatile, acid extractable, and base neutral organics, pesticides, dioxins and PCBs
              Table 3.  Potential Management Practices for Debris Decontamination

-------
5.0  SLUDGE

The previous OSWER survey of Superfund sludge data found that
sludges are not consistently defined in the literature.  However,
sludges, when identified, had higher concentrations of
contaminants than soils, and as a result, did not meet variance
level standards as frequently as soil.  Of the OSWER survey data,
55% of the sludges treated met variance levels, while 78% of the
soils treated met variance levels.  OSWER believes that to fully
characterize the treatment of sludge much additional work will be
required.  To this end, OSWER, in conjunction with ORD, is fully
characterizing sludges from several hazardous waste sites on
sludge later this year.  In addition, EPA is holding a symposium
during the summer of 1991 to broaden the background information
and share collective views on this topic.  Additional information
on the symposium will be made available to any interested parties
by contacting the authors of this paper.

6.0  VARIABILITY

The OSWER study of Superfund soil treatability has found an order
of magnitude difference in treatability between remedy selection
testing and full scale treatment.  As a result, treatability
tests must achieve an order of magnitude better treatment than
the standards in order to achieve compliance with the full scale
process.  The factors that affect treatment effectiveness include

                                9
                            11-333

-------
mixing effectiveness, homogeneity of the soil matrix, feed



specifications, and contaminant concentrations.  Variability of



the treatment results for the relatively homogeneous RCRA waste



streams has been accounted for by using classical statistics



which assume a less variable data set than Superfund soils.







EPA has begun a study to determine whether the soil matrix



presents unique problems in specific treatment methods and types



of wastes.  EPA's study will use "clean" soils of similar



characteristics as the contaminated soil and artificially "mark"



the soil with a non-hazardous contaminant.  Soils will then be



mixed and analyzed to determine the efficiency of mixing as a



treatment condition.   The results of the study are expected to



show whether variability mixing effectiveness exists as a



function of soil type, equipment scale or moisture content, which



is representative of  different treatment technologies.  The



results of the study  are not expected to conclusively show what



the variability function is or to allow for a direct correlation



into the LDR.   Additional experimentation will be required to



assess the magnitude  of the variability as it impacts on the



treatment standards for contaminated soil.







7.0  CONCLUSIONS







The current schedule  provides for completion of data collection



and data analysis in  the fall of 1991.  We are soliciting





                                10



                            11-334

-------
existing treatment data, and new tests which may meet these



needs.  We welcome comments on the above described program to



advance this study effort on soils, sludges, debris and



variability.  If you have data, comments or questions regarding



the LDRs for contaminated Superfund soils and debris please



contact:
      Richard Troast



Project Manager, CSD Program



       703-308-8323
   Carolyn K. Offutt



Chief, Special Projects



   and Support Staff



      703-308-8330
            Hazardous Site Control Division (OS 220W)



               U.S.  Environmental  Protection Agency



                        2800  Crystal  Drive



                    Arlington,  Virginia  22207
                                11
                            1-335

-------
8.0  References







1.  Superfund LDR Guide #6A:   Obtaining a Soil and Debris



Treatability Variance for Remedial Actions.  1989, Revised 1990.



Office of Solid Waste and Emergency Response, USEPA.  Directive:



9347.3-06FS.







2.  Superfund LDR Guide #6B:   Obtaining a Soil and Debris



Treatability Variance for Removal Actions.  September, 1990.



Office of Solid Waste and Emergency Response, USEPA.  Directive:



9347.3-06BFS.







3.  Quality Assurance Project Plan for Characterization Sampling



and Treatment Tests Conducted for the Contaminated Soil and



Debris (CSD) Program.  November,  1990.  Office of Solid Waste and



Emergency Response, USEPA.
                                12



                            11-336

-------
91         Certification Protocol for Meeting Organic Treatment Standards
                                   for Incineration Ash
         SEVENTH ANNUAL WASTE TESTING & QUALITY ASSURANCE
                                      SYMPOSIUM
                                 American Chemical Society
                                     Washington, D. C.

                                        July 8-12, 1991
        William R. Schofield, PhD, PE, Schofield Environmental Associates, 1500 Marina Bay Drive, Suite
        1612, Kemah, Texas, 77565( formerly Technical Manager, Chemical Waste Management, Inc., Texas
        Facilities); John W. Kolopanis, Director, Technical Services, Chemical Waste Management Inc., 150
        W. 137th Street, Riverdale, Illinois, 60627; Teresa S. Johnson, Area General Mgr., Chemical Waste
        Management, Inc., 2700 N. S. 48th Street, Pompano Beach, Florida, 33073
                                          11-337

-------
   Certification Protocol for Meeting Organic Treatment Standards
                          for Incineration Ash
William R. Schofield. Technical Manager, Chemical Waste Management, Inc., P. O.
Box 2563, Port Arthur, Texas 77643-2563; John W. Kolopanis, Director, Technical
Services, Chemical Waste Management Inc., 150 W. 137th Street, Riverdale, Illinois,
60627; Teresa S. Johnson, Area General Mgr., Chemical Waste Management, Inc.,
2700 N. S. 48th Street, Pompano Beach, Florida, 33073

ABSTRACT

The Hazardous and Solid Waste Amendment of 1984  (HSWA) of the Resource
Conservation and Recovery Act (RCRA) requires the treatment of hazardous waste
to a specified  treatment standard prior to land disposal.  Testing to verify that
treatment residuals  (i.e., incinerator ash and scrubber sludge/filter  cake) meet
treatment standards is  an expensive and time consuming  process, especially for
commercial incinerators in which each batch of residuals has a different set of EPA
waste codes and consequently different treatment standards.

The challenge is to develop a testing approach or protocol which will simultaneously
provide:     (1)  a  high level of assurance  that  treatment  standards are being
consistently met while (2)  holding testing and residual storage costs and testing
turnaround time at reasonable levels and (3) insuring that permitted  residual storage
limits are met.

Chemical Waste Management has developed a practical testing protocol based on
EPA developed or supported concepts which is sufficiently flexible to fit the widely
varying incinerator facilities within our  system.  In concept, the universe  of EPA
waste codes is  divided into "treatability groups" based on chemical and  physical
similarities.  Each treatability group is then represented by  one or  more "indicator
waste code(s)" selected on the basis of: (1) treatment standard chemical species and
acceptance levels, (2)  volume of waste with that  code needing treatment, (3)
volatility and thermal stability of the chemical species present in the waste, (4)
matrices effects and (5)  related issues.

A demonstration or "trial burn" is conducted in which the  incinerator is operated
within an "operating envelope" and under a "quality assurance/quality control system"
                                     1
                                   11-338

-------
which will assure operation within the envelope.

If the incineration residuals meet treatment standards for the carefully selected
indicator codes, then compliance with treatment standards for the subject chemical
species  are assumed for  other codes within the treatability  group which have
treatment standards ar or above  the  level detected on the  demonstration test
residuals.

Spot tests  are conducted on  at  least a quarterly basis  to  confirm continued
compliance.  New demo burns are initiated whenever the operating  envelope is
changed in a manner in which would be detrimental to the destruction or removal
of organics from residuals.

This approach has been successfully used with all three CWM incinerator facilities.
Other aspects of the protocol will be discussed in the presentation.

INTRODUCTION - DEFINITION OF THE PROBLEM

With the implementation  of the HSWA (Hazardous and Solid  Waste Amendment
of 1984) of the RCRA (Resource Conservation and Recovery Act) landbans and the
gradual elimination of the remaining variances, a large  fraction of the hazardous
waste generated  in  the U. S. must be treated  to meet stringent BDAT (Best
Demonstrated Available Technology)  treatment standards before it can be land
disposed.  For hazardous waste containing organics, sufficient treatment frequently
requires incineration to destroy the organic and cyanide compounds. The resulting
solid residuals (ash and scrubber cake)  are'then stabilized to chemically immobilize
any regulated metals present prior to landfilling.

A great deal of effort has been invested in the hazardous waste management industry
to develop a practical and  reliable method to verify that each treatment step has met
the HSWA treatment standards. In the case  of organics contaminated waste, this is
frequently  a two stage process:  (1) verification that the incinerator ash and filter
cake meet  organic and cyanide standards for all EPA waste codes present followed
by (2)  verification that metal mobility or leachability has been sufficiently reduced
in the stabilization process to meet TCLP (Toxic Constituent Leaching Procedure)
limits for metals.

The focus of this paper is  the certification protocol for organics in residuals from a
commercial hazardous waste incinerator; a simplified version of this protocol would
apply to captive incinerators. The protocol is described in toto; however, no attempt
will be made to cover every possible contingency which can occur when attempting
to satisfy, with total regulatory compliance, a program as exacting and complex as the
EPA landban regulations.
                                    11-339

-------
The  challenge in developing a practical and reliable protocol can be  seen by
considering the following facts:

1.  The EPA landban program presently contains a total of 1,467 variations of waste
codes, categories, and subcategories.  There are typically 20 to 40 EPA waste codes
associated with a given batch of solid residuals from a commercial incinerator; these
codes can change dramatically with time.  Consequently, the required treatment
standards which the residuals must meet also change with time.

2.   EPA has established treatment standards for one or more  organic/cyanide
compounds (target compounds) for each waste code which requires incineration.
(There is typically one target organic compound for U,  P, and D codes, 5 to 12 for
each F and K code and  from dozens of target compounds for F001-5 codes to
hundreds for F039 codes.)

3.  Frequently the same target compound will  appear associated with two or more
waste codes present in a batch of residuals and typically the treatment standard level
will vary from code to code even for the same target compound. Thus, the lowest
treatment standard present must be simultaneously met for every target compound
present before a batch of residuals can be certified as having met BDAT.

4.  There is no on-line method of testing incineration residual for organics and
cyanides - one or more different extraction protocols must be completed on each
sample, typically, followed by multiple GC scans, GC/MS volatile and semi-volatile
scans and other testing depending on which EPA waste codes are present.

This an analogous situation to the use of EPA Modified Method 5 sampling of an
incinerator stack in a trial bum and subsequent extractions and analyses as a means
of verifying that the incinerator met the required minimum destruction and removal
efficiency (DRE) during the trial burn; thus, CWM as  the permittee is authorized
to infer that the unit is meeting DRE requirements during subsequent commercial
operations as long as the unit is operating within the permitted operating envelope.

5.  The  sampling  and analysis turnaround time  for  HSWA residual testing is
extremely slow, very complicated and disruptive to the residual management process
(e.g., one week, under ideal circumstances, to a more  typical 30  to 60 days).  In
addition, the cost is quite expensive (e.g., $3,000 - 10,000 per event).

6.  HSWA requires the treater (in this  case  the incinerator owner/operator) to
"certify under penalty of law" to the land  disposal facility  where the ash will be
landfilled that the waste has met applicable organics treatment standards. Thus,
                                   11-340

-------
boxes of ash can not be shipped until the certification can be completed.

7.  A commercial incinerator typically generates 2-6 boxes of ash and other solid
residuals during a normal operating day and with larger volumes being frequently
generated during  maintenance turnarounds.  As a point of interest, the CWM Port
Arthur incinerator  has  the  physical capacity and permitted authority to process
sufficient waste to produce up to 20 boxes of residuals per day.

A quick review of these facts indicates that if every box of residuals must be sampled
and analyzed  for the applicable target compounds, this would result in ongoing
inventories of at least 50 - 100 boxes of uncertified residuals and would disrupt the
ability to  manage residual  in a timely and environmentally sound manner.  In
addition, analysis of each box would result in analytical cost in the order of hundreds
of thousands to millions of dollars a year, diverting resources from areas that would
afford more protection to the environment.

DEVELOPMENT OF A CERTIFICATION PROTOCOL - CONCEPTS AND RATIONALE

The factors which affect the degree to which organics are destroyed on or vaporized
from  solid residuals in a given hazardous waste (typically rotary kiln) incinerator
include (see Figure 1 for a schematic diagram of an incinerator process):

1.  Residence time of the solid in the hot zone (i.e., rotary kiln length, waste loading
   and RPM).

2.  Temperature in the hot zone (i.e., kiln temperature).

3.  Oxygen concentration (i.e., % excess air).

4.  Degree of agitation of the organic contaminated solid (i.e., kiln RPM)

5.  Solids  loading (i.e., feed  rates of solids bearing waste).

6.  Volatility  of organic compound (i.e., vapor pressure of target compounds).

7.  Thermal stability (i.e., thermodynamic stability of target compounds).

8.  Nature of solid substrate or matrix factor (i.e., Is the waste liquid, sludge or solid?
   Is  contamination a surface or depth phenomenon?).
                                    1-341

-------
WASTE
WASTE,

WASTE,
ROTARY
  KILN
                 AFTER
                BURNER

                      ASH
                                                                       STACK
                                                      RECYCLE LIQUID
QUENCH
        JJ
                                                  ABSORBER
                                          I
                                           I
 IONIZING
  WET
SCRUBBER
                                                             SUSPENDED
                                                               SOLIDS
                                                              REMOVAL
                                                                 I

                                 FAN
                                               FILTER CAKE
                  FIGURE 1   A schematic diagram of a typical hazardous
                     waste incinerator with a wet, acid gas absorber.

-------
The process variables (i.e, temperature, oxygen concentration, solids residence time,
loading, and  agitation)  are controlled directly  or indirectly by the permit limits
and/or the incinerator  design.  The matrix factor is waste stream specific and
volatility and thermal stability is determined by the molecular structure of the target
compound.  Thus, a protocol could be based on:  (1)  direct measurement and
control of the pertinent process variables while controlling (2)  the waste streams
selected to provide a range of matrix types (i.e., solid organics, inorganic solids
contaminated by organics, particle size, liquids, etc., and (3)  waste codes selected to
provide target compounds covering a range of volatilities and thermal stabilities.

A calibration test (or demonstration burn) would then be conducted which would
establish that the treatment (incinerator) system meets the treatment standards for
the codes present while operating within permit/design limits.  The idea being that
an incinerator destroys  organic compounds without regard to the waste codes
associated with those compounds.

This approach is quite analogous to the EPA trial burn approach to demonstrating
that a given  incinerator will meet DRE requirements as long as the  incinerator
operates within its permitted operating envelope.

In developing the RCRA/HSWA landban treatment standards EPA drew on several
concepts which are equally useful  here. These concepts are:  "treatability groups"
based on chemical  and physical  similarities among  wastes  with certain codes,
"transference of data" on treatment efficiency in an incinerator from waste code to
another code in the same treatability group, i.e.,  the use of one waste  code as  an
"indicator code" for other codes within the same treatability group.  CWM has used
these concepts to develop a certification protocol.

The universe of all EPA waste codes was divided into 16 treatability groups along the
same chemical similarity lines EPA used in the RCRA/HSWA third third regulations
(see Table I for a listing of waste codes in each treatability group).

From each treatability group one (or more) code(s) were selected as indicator codes
based on the following criteria:

1. The number and type of target compounds for that code.

2. Treatment standard levels for that code.

3. The thermal stability and incinerability of the  target compound(s) of that code
compared to  the stability and incinerability of the target compounds of other codes
                                    11-343

-------
TABLE 1    EPA waste codes1 assigned to each treatability group.
GROUP 1 - Solvents and Dioxin:  P001-05, F020-23, P026-28

GROUP 2 - Halogenated Pesticides: D012-17, K032-34, K041, K097-98, P004, P037, P050-51, P059-60,
           P123, U036, U060-61, U128-130, U142, U240, U247

GROUP 3 - Chlorobenzene:  K042, K085, K105, UQ37, U070-72, U127, U183, U185, U207

GROUP 4 - Halogenated Phenolics: U039, U048, U081-82

GROUP 5 - Brominated Organics: U029-30, U066-68, U225

GROUP 6 - Miscellaneous Halogenated Organics: P024, U024-25, U027, U043,
           U045, U047, U075, U121, U138, U158, U192

GROUP 7 - Aromatic & Other Hydrocarbons: U019, U220, U239

GROUP 8 - Polymiclear Aromatic Hydrocarbons: K001, K015, K022, K035, K048-52, K060, K087,
           U005, U018, U022, U050-51, U063, U120, U137, U157, U165

GROUP 9 -Phenolics: P020, P047-48, U052, U101, U170, U188

GROUP 10 - Oxygenated Hydrocarbons & Heterocyclic: K023-24, K086, KO93-94, U002, U004, U028,
           U031, U069, U088, U102, U107-08, U112, U117-18, U140, U159, U161-62, U190

GROUP 11 - Organo-Nitrogen Compounds: K011, K013-14, K083, K101-04, P069, P077, P101, U003,
           U007, U009, U012, U105-06, Ulll, U152, U169, U172, U174, U179-81, U196

GROUP 12 - Halogenated Aliphatic: F025, K009-10.X016-21, K028-30, K073, K095-96, U043-44, U076-
           80, U083-84, U131, U184-85, U208-11, U226-28, U243

GROUP 13 - Other Chlorinated Organics:  P024, K043, K099

GROUP 14 - Organo-Sulfur Compounds: K036-38, K040, P039, P071, P089, P094, P097, U235


GROUP 15 - Pharmaceuticals: U141, U155, U187, U203

GROUP 16 - Cyanide: F006-12, F019, P013, P021, P029-30, P063, P074, P098-99, P104, P106, P121
1 Waste codes with BDAT specified technology of incineration (INCIN) are not listed.
                                        1-344

-------
within that group.

4. The volatility of the target compounds of this code compared to the other codes
in the group.

5. The volume of waste with this code in the market place, i.e., the commercial
significance of the code compared to other codes in the group. In this and numerous
other ways CWM  has attempted to conduct these tests in a manner which would
simulate routine, day to day operating conditions.

Once this list of indicator codes is developed one or more waste streams are selected
for each code  based  on  commercial  significance and matrix  effects.   Waste
inventories are collected and used in a trial burn type demonstration in which the
waste is burned in the treatment process under normal operating envelop conditions
and with a defensible QA/QC program in effect.

In the case of the  CWM Port Arthur incinerator the incinerator is controlled by a
computer which continuously monitors all permit limited parameters.  If a single
operating parameter moves outside of its permitted range or operating envelope, the
computer automatically discontinues waste feeds.

In planning the demonstration test, we generally will want wide ranges of thermal
stabilities, volatilities, matrix effects and treatment standard levels represented;
however, for  the sake of minimizing uncertainty or risk, we have tended to  choose
waste codes with less volatile, highly stable target compounds in solid substrates with
quite low treatment standard levels.  (See Table II for the selected indicator codes
for each treatability group.)

The concept  is that waste with the indicator codes will incinerate  in a like manner
to other wastes within  the same  treatability group,  i.e.,  we can transfer  data
concerning how well the treatment process incinerated one waste code to the other
codes with chemical similarities.

Once the demonstration test is completed and the resulting residuals are carefully
sampled and  analyzed (in triplicate), then waste codes from a represented treatability
group with treatment standards at or above the level of target compound(s) detected
in the demonstration test residuals can be certified as long as the process is operated
within the operating envelope and the QA/QC system is maintained.   Conversely,
codes with treatment standards which are  lower than  the residual  concentrations
found in the  demonstration burn could not be certified without process adjustments
as needed followed by a successful new demonstration test.
                                      8
                                    11-345

-------
TABLE II Indicator code(s) selected for each treatability group.
        TREATABILITY GROUP              INDICATOR CODE(S)
1.  SOLVENT & DIOXIN                              F001-5
2.  HALOGENATED PESTICIDES                      U129, P123
3.  CHLOROBENZENE                               U070, UO72
4.  HALOGENATED PHENOLICS
5.  BROMINATED ORGANICS                         U029
6.  MISC. HALOGENATED ORGANICS                  (K019, F001, F002)1
7.  AROMATIC & OTHER HYDROCARBONS            U220
8.  PNA HYDROCARBONS                            U165
9.  PHENOLICS                                     P020
10. OXYGENATED HC & HETEROCYCLIC              U002, U069, U190
11. ORGANO-NTTROGEN                             K011, K013, U012
12. HALOGENATED ALIPHATICS                      K019, K020
13. OTHER CHLORINATED  ORGANICS
14. ORGANO-SULFUR                               P039, P071, P089, P094
15. PHARMACEUTICALS
16. CYANIDES                                      F0007, F008
                               11-346

-------
TABLE III Permit and other limitations which the 1990 operating envelope for the
            CWM Port Arthur incinerator
       HOURLY TOTALS

       CHLORINE
       SULFUR
       ORGANIC  HALOGEN
       ASH (SCC)
       TOTAL  ORGANIC CONTENT
       POLYCHLORINATED BIPHENYLS

       WASTE AND FUEL FEEDS
       ENERGETIC LIQUIDS (KILN)
       ENERGETIC LIQUIDS (SCC)
       ENERGETIC SLUDGE
       NON-ENERGETIC SLUDGE
       ENERGETIC SOLIDS
       NON-ENERGETIC SOLIDS
       AQUEOUS WASTE

       MINIMUM  KILN HEAT VALUE
       MAXIMUM KILN HEAT VALUE
       MINIMUM  SCC HEAT VALUE
       MAXIMUM SCC HEAT VALUE
       MINIMUM  TOTAL HEAT VALUE
       MAXIMUM TOTAL HEAT VALUE

       ACRYLAMIDE
       CHLOROMETHYLMETHYLETHER
       1,2-DIBROMO-S-CHLOROPROPANE
       SYM-DICHLOROMETHYLETHER
       DICTROTOPHOS
       DIMETHYL CARBAMOYL CHLORIDE
       DIPHENYLMETHANE DIISOCYANATE
       ISOPROPYL MERCAPTAN
       ISOPHORONE DIISOCYANATE
       N-NITROSODIETHANOLAMINE
       N-NITROSODIETHYLAMINE
       PHOSPHINE
       LEAD (FEED RATE LIMITS)
       CADMIUM
       VANADIUM
       MERCURY
       ARSENIC
       BERYLLIUM
       CHROMIUM
       NICKEL
  PERMIT LIMIT
1,690
250
1,352
240
20,000
3,172
50,270
3,000
8,900
5,300
10,000
3,000
41,475
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
(determined by DCS)

  47.9 MM BTU/HR
  79.5 MM BTU/HR
  35.0 MM BTU/HR
  77.2 MM BTU/HR
  79.0 MM BTU/HR
  150.0 MM BTU/HR
900
660
79
79
1,330
660
1,330
240
1,330
130
130
1,660
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
                                             CWM OPERATIONAL LIMITATIONS
1,350.0
 240.0
 240.0
 240.0
  48.0
  17.5
 900.0
 135.0
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
1 Although K019, FD01, and PD02 are EPA waste codes which do not appear in treatability Group 6, these codes have
components in common with codes from Group 6. Since K019, F001, and F002 were fed during the demonstration test and the
residual values for the target compounds were found to be lower than all the treatment standard levels specified for these three
codes and for Group 6 codes, the common compounds could and have been used to certify Group 6.
                                             10
                                            1-347

-------
A SPECIFIC EXAMPLE - PORT ARTHUR INCINERATOR, 1990

Shortly  after  the third landbans  became  effective in  1990  a broad based
demonstration burn was conducted on the CWM Port Arthur (Texas) incinerator;
the results are summarized here as an example.

Appropriate streams could not be found for groups 4, 13, and  15.  The test took
place over approximately 24 hours. Table III defines the operating envelope in effect
at the time of the test. Three samples were carefully collected to represent three
eight hour time periods, independently extracted and analyzed per SW846.

Organic concentrations were below the practical quantitation levels (QL) in almost
all cases; exceptions were:
                                      MIN TREAT1
                                PQL    STANDARD
                                                         TEST RESULTS
0.010
0.005
0.005
0.005
0.100
0.010
0.005
0.010
0.330
0.100
1.000
0.590
28.000
5.600
65.000
170.000
360.000
33.000
0.100
3.600
0.100
57.000
0.010
7.865
2
0.055
2
2
0.070
0.020
2
0.100
1.800
0.043
6.100
0.105
0.115
1.145
2
0.200
0.020
2

2
0.011
0.955
2
0.108
2
0.560
0.115
2
0.363
2
1.300
TARGET COMPOUND	

ACETONE
TRICHLOROTRIFLUOROETHAN
CARBON TETRACHLORIDE
IODOMETHANE
ISOBUTYL ALCOHOL
ETHYL CYANIDE
TRICHLOROFLUOROMETHANE
DISULFOTON
DIS-N-BUTYL PHTHALATE
FAMPHUR
TOTAL CYANIDE

All numbers are expressed as TCA (rag/Kg) except Acetone which is expressed as TCLP (mg/L).

1 These are the lowest treatment standards within the landban program for these target compounds
regardless of waste code.
2 The results were below the practical quantitation level (PQL) of the analytical instruments used.

Inspection of these data indicates that this test was successful on all compounds of
the codes/treatability groups  tested.  Thus, based on these results the Port Arthur
facility was able to certify that ash generated with codes from the tested treatability
groups meets all applicable organic and cyanide treatment standards as long as the
unit is within the operating envelope and the QA/QC program is maintained.

An additional demonstration burn would be required to qualify codes/treatability
groups not covered in this initial effort or to requalify the treatment system if one or
more conditions were  changed in a manner which could reduce the incinerator
system's ability to produce clean treatment residuals.

                                     11
                                    11-348

-------
92                                       ABSTRACT

                              Factors Affecting the Admissibility
                       and Weight of Environmental Data as Evidence


             Jeffrey C. Worthington. Director of Quality Assurance; Kerri G. Luka, Audit
             Programs Manager, TechLaw, Inc. 12600 West Colfax Avenue, Suite C-310,
             Lakewood, Colorado 80215.

       Many factors may affect the potential admissibiiity of environmental data in litigation.
       These factors include but are not limited to:

             0  Integrity of the sampling method

             0  Integrity of the analytical method

             0  Comparability to other sets of environmental data

             0  Documented sample custody

             0  Documented quality control results

             0  Authenticity of the data

       These same factors may also enter into the weight of the data as evidence. For example,
       some data may include rigorous  quality control  including the  use  of performance
       evaluation samples with each batch of samples from the field; other  data may include less
       rigorous quality control.  The first set of data may be given greater weight by the trier of
       fact.

       The admissibiiity and weight of environmental data evidence may figure prominently into
       pre-trial settlement  discussions.   Data is  not often accepted at "face value".  Litigants
       usually need to address all the issues concerning the environmental data before proceeding
       to other litigation matters.

       The authors present a discussion of these  factors and summarize several cases where the
        admissibiiity and weight of the data as evidence were items of concern.
                                           1-349

-------
    REVIEW OF GROUNDWATER MONITORING REQUIREMENTS AT RCRA SITES

William G. Stelz. CPG. RCRA Enforcement Division, Office Of Waste
Programs Enforcement, U.S. Environmental Protection Agency, 401 M
Street S.W., Washington, D. C. 20460


ABSTRACT

This paper summarizes the groundwater monitoring requirements for
RCRA facilities under both interim status and operating permit
conditions.  In addition, it highlights the major differences in
the regulations for facilities subject to both interim status as
well as permit requirements.  Along with this overview, this
paper addresses how these regulations are enforced and what
mechanisms are set up to ensure that facilities are in compliance
with the groundwater requirements under the RCRA program.


INTRODUCTION

Subtitle C of the Resource Conservation and Recovery Act of 1976
(RCRA)  regulates hazardous waste treatment, storage, and disposal
facilities (TSDFs) .   Section 3004 of RCRA requires owners and
operators of hazardous waste TSDFs to comply with standards
established by EPA.   Section 3005 provides for implementation of
these standards under permits issued to owners and operators by
EPA or authorized States.  Section 3005 also provides that owners
and operators of existing facilities that comply with applicable
notice requirements may operate as " interim status" facilities
until a permit is issued or denied.  Owners and operators of
interim status facilities also must comply with standards set
under Section 3004.

EPA promulgated regulations for permitted facilities in 1982 (47
FR 32274, July 26, 1982), codified in 40 CFR part 264, Subpart F
and 40 CFR part 270, Subpart B.  These regulations establish
programs for protecting groundwater from releases of hazardous
wastes or constituents from treatment, storage, and disposal
units .
BASIC GROUNDWATER MONITORING REQUIREMENTS

The basic groundwater monitoring program under RCRA consists of
three main components: Interim status requirements, permit
application requirements and operating permit requirements (see
figure 1) .  Each of these components contains specific
requirements and is designed to follow a sequence of applications
as a facility moves into different segments of the regulatory
process .
                            1-350

-------
INTERIM STATUS GROUNDWATER MONITORING REQUIREMENTS

On May 19, 1980, EPA promulgated comprehensive standards under 40
CFR part 265 for owners and operators of hazardous waste
treatment, storage, and disposal facilities (TSDF's) that qualify
for interim status.  A facility owner or operator who has fully
complied with the requirements for interim status specified in
Section 3005(e) of RCRA and 40 CFR 270.70 may comply with the
part 265 regulations in lieu of part 264 pending final
disposition of the permit application.  Part 265 Subpart F
contains groundwater monitoring requirements applicable to owners
and operators of interim status landfills, surface impoundments,
and land treatment facilities.  The goal of the interim status
groundwater monitoring program is to evaluate the impact that the
facility may have on the uppermost aquifer underlying the site.
The regulations establish a two-stage groundwater monitoring
program designed to detect and characterize the migration of any
wastes that may have contaminated the groundwater.  Stage I
consists of a detection monitoring phase where the objective is
to determine if hazardous wastes have leached into the uppermost
aquifer in quantities sufficient to cause a significant change in
groundwater quality.  Stage II is an assessment monitoring phase
that is initiated when a significant change in water quality has
been detected at a hazardous waste facility and contamination is
suspected.  The assessment monitoring program is directed at
characterizing the rate and extent of contaminant migration.
Assessment monitoring under Section 265 entails a determination
of both the vertical and horizontal concentration profiles of all
hazardous waste constituents in the plume(s) that escape from the
hazardous waste management areas.  Figures 2, 3 and 4 outline the
major features of interim status groundwater monitoring.


PART 270 - PERMIT REQUIREMENTS

Part 270.14(c) establishes permit application requirements (Part
B), that an owner/operator must submit in order for EPA to
determine if the facility is in compliance with the part 264
standards.  Part 270.14(c) requires the applicant to establish
the nature of the facility's impact on the groundwater, as well
as the hydrogeologic characteristics of the site's subsurface and
the extent of the waste management area.


OPERATING PERMIT REQUIREMENTS

The part 264 Subpart F groundwater monitoring requirements apply
to owner/operators that treat, store, and or dispose of hazardous
waste in surface impoundments, waste piles, land treatment units,
or landfills that receive waste after July 26, 1982.  Such units
are referred to as "regulated units."  These requirements are
                            11-351

-------
effective only at facilities that have failed to qualify for
interim status (see Part 270).   An interim status land TSD
facility would not be subject to part 264 standards until a
permit is issued to that facility and unless waste was received
after July 26, 1982.

Groundwater Monitoring at permitted facilities has three phases:

1. Detection - of indicator parameters,  waste constituents, or
reaction by-products in the uppermost aquifer.
2. Compliance Monitoring - to better define the extent of aquifer
contamination by identifying which hazardous waste constituents
are present in the groundwater and by describing the shape and
concentration of the contaminant plume
3. Corrective Action - to remove hazardous waste constituents
from the groundwater or to treat them in place.

Typically, a facility employs a detection monitoring program
until there's a statistically significant increase in that
program's parameters or constituents, after which a compliance
monitoring program begins.  If there's a statistically
significant increase in the concentrations established in the
compliance monitoring program,  i.e., if the groundwater
protection standards have been exceeded, the facility must enter
a corrective action program.  However, a facility need not begin
with detection monitoring - if there is existing evidence of
groundwater contamination (such as from an interim status
monitoring program),  the facility can be put directly into a
compliance or corrective action program when the facility's
permit is issued.  Figures 5 and 6 illustrate the main components
of groundwater monitoring for facilities with operating permits.
SUMMARY

The part 264 groundwater monitoring standards differ from those
in part 265 in that the part 264 standards are more flexible and
go beyond just contaminant assessment and allow for corrective
action to be directly incorporated; whereas under interim status,
corrective action has to be achieved via another mechanism such
as from an enforcement order (e.g., a 3008(h)).
Instead of testing for specific parameters as in part 265, part
264 requires the Regional Administrator to specify parameters and
hazardous waste constituents to be monitored on a site-by-site
basis.  In each phase of the groundwater monitoring program under
part 264, the number, depth, and location of wells must yield
representative samples of groundwater.  In addition, under part
264, a groundwater protection standard is set up for each
constituent found in the groundwater, and if exceeded, corrective
action is initiated.  Figure 7 summarizes the various options for
groundwater monitoring for land disposal facilities.
                            1-352

-------
REFERENCES

Code of Regulations, Protection of Environment, 40, parts 190 to
399, revised July 1, 1990.

Solid Waste Disposal Act, as amended by the Resource Conservation
and Recovery Act of 1976, as amended by the Hazardous and Solid
Waste Amendments of 1984, (42 U.S.C. 6905, 6912(a), 6921, 6924,
6925, and 6935).

U.S. Environmental Protection Agency. 1986.  RCRA Ground-Water
Monitoring Technical Enforcement Guidance Document.  OSWER-9950.1
                             11-353

-------
       Basic Groundwater Monitoring Requirements
CO
Groundwater monitoring requirements include:

    Interim status requirements
    (40 CFR Part 265 Subpart F)

    Permit application requirements for groundwater monitoring
    (40 CFR Part 270.14 (c) Part B)

    Operating permit requirements
    (40 CFR Part 264 Subpart F)

-------
 Interim Status  Groundwater Monitoring
CO
en
01
Groundwater monitoring program includes:

    •   Monitoring wells
          1 hydraulically upgradient
          3 downgradient
       Sampling/analysis plan
       Statistical comparison test
       Outline of groundwater quality
       assessment program
               (Based on regulations at 40 CFR 265.91)

-------
                  Interim  Status
      Groundwater Monitoring (cont'd)

             Is Contamination Present During Interim Status?
          NO
en
Continue monitoring
program until
closure or permit
application
issuance.
                                      YES
Implement groundwater
assessment program as
stipulated by submitted
outline.

Continue to make
assessment quarterly
until closure or permit
issuance.

          (Based on regulations at 40 CFR 265.90 through 265.94)

-------
    Monitoring Over Time
  Quarterly monitoring for:
          Drinking water standards
          Groundwater quality parameters
          Indicator parameters
          Water elevations
  Semiannual monitoring for:

         • Indicator parameters
         • Water elevations
  Annual monitoring for:

         • Groundwater quality parameters
(Based on regulations at 40 CFR 265.90 through 265.94)

-------
Groundwater Monitoring During Permit

      Groundwater monitoring program requirements:
          • Specify the point of compliance
          • Sufficient wells properly located to yield both
           - background groundwater quality and
           - water quality passing the point of compliance
          • Consistent sampling/analysis procedures
          • Determination of groundwater elevations during all
           sampling periods
          • Background groundwater quality
          • Statistical comparison procedure

-------
        Potential Results of Monitoring
             Groundwater monitoring program requirements
CO
en
CD
   No contamination
Hazardous
constituents
detected at point
of compliance
Hazardous
constituents
detected at point
of compliance
and downgradient
of facility boundary -
exceeding the
groundwater
protection standard.
   Detection
   monitoring
   (40 CFR 264.98)
  Compliance
  monitoring
  (40 CFR 264.99)
   Corrective
   Action
   (40 CFR 264.100)

-------
                         Options  for  LDFs
                                       INTERIM
                                       STATUS
                                       FACILITY
       Operating Permit
   No release to
   groundwater
 Detection
Monitoring
                                  GROUND-WATER MONITORING
                                            Closure
Release In excess of the groundwater
protection standard at or beyond the
point of compliance
   Detection monitoring
   (I.e., with no groundwater
   contamination)
   Assessment monitoring
   (l.e.,wlth groundwater
   contamination)
    Corrective
 Action Monitoring
Clean Closure
       Release in excess of background at
       the point of compliance
                                                    NO MONITORING
Closure With Waste
     In Place:
   Post-Closure
   Requirements
                                               30 YEARS OF MONITORING
                1
            Compliance
            Monitoring
                                                       i
                                                  Post-Closure
                                                     Permit

-------
94                                      ABSTRACT

                         The Paperless Environmental Laboratory:
                                   A Plan for Realization


             Jeffrey C. Worthinyton. Director of Quality Assurance; George A, Duba,
             PhD, Vice President;  TechLaw, Inc.,  12600 West Coifax, Suite  C-310,
             Lakcwoo<:>- Colorado 8021S.

       Many laboratories are buying, installing, or modifying their current Laboratory Information
       Management Systems (LfMS) to produce all the documents necessary to effect the smooth
       flow  of  samples through the laboratory.   Laboratory managers  and analysts most
       comfortable with keyboards hope to make all paper disappear on the work bench by using
       direct data entry.

       Users of data from environmental laboratories often include attorneys who may need to
       demonstrate sample  custody  and integrity of the sample data in order  to admit  the
       information into court.  These data users  are less than comfortable seeing hand-written
       documents disappear from the laboratory to be replaced by electronic records.
          % author presents guidelines for the development of a paperless laboratory system. The
       ฃ.  ""nes include consideration for laboratory management issues and litigation related
                                          1-361

-------
95                            DATA MANAGEMENT ISSUES
                                          IN THE
                              HAZARDOUS WASTE INDUSTRY
         Gerald Austiff. CRM, Technical Records Archivist, Marty Cahill, Manager-Waste
         Analysis Plans Group, John Krecisz, Technical Manager-Incineration,   Chemical
         Waste Management, 150 West 137th Street, Riverdale, IL 60627

         ABSTRACT

         One of the most direct and yet unaddressed consequences  of increased Federal,
         State,  and Local Government regulation of U.S. Industry in the later half of the
         twentieth century has been the added responsibility of creating those documents and
         data necessary to verify compliance with these regulations. For an industry such as
         hazardous waste management,  the  responsibilities of mandatory records creation
         have proven to be especially great.

         What has not necessarily followed, however, is the development of records and data
         management systems proportional to the importance that information serves in the
         operation of a hazardous waste facility.  In today's business  climate, however, the
         opposite is equally true-the lack of management attention to the records and data
         that is routinely produced by the organization can cost plenty, both in terms of
         dollars and in reduced productivity.

         This paper will address the data management issues facing every company in the
         hazardous waste industry  and outline a records management strategy  that  such
         companies must consider not only to avoid costly fines/penalties, but to turn their
         records and data into a  positive asset.

         INTRODUCTION

         As one of the  most heavily-regulated sectors of the world economy, the hazardous
         waste  industry  has  many  specific and long-term  records/data—management
         requirements which must be met in order to be allowed to continue to conduct
         business in its operating jurisdiction.  The ability to create those documents and data
         required for waste profiling, analyses of waste samples, and facility operation has
         been  greatly  enhanced   by  sophisticated  laboratory  equipment   and   the
         computerization of manual recordkeeping practices in general.  This development
         has not, however, resulted in an equal ability to provide  long-term protection and
                                           11-362

-------
retrievability of this information.  To the contrary, the burden to keep analytical
records and data for the time periods required by facility permits and government
rules  has become even more challenging  and  elevates  the  task of records
management to a higher management priority that it  may have been raised to in
the past.

RETENTION REQUIREMENTS

All hazardous waste  facilities  are  subject to a  number  of Federal  and  State
Regulations which impact recordkeeping and data management. Some of the most
important recordkeeping regulations are:

                                  RCRA

40CFR, Parts 264.16 and 265.16
    -  Requires  that Training Records on current personnel must be kept until
       264.16&265.16 closure of the facility.

40CFR, Parts 164.73 and 165.73
    -  Requires  the owner or operator to keep the written operating record at his
       facility   until closure.  Monitoring data at disposal facilities must be kept
       throughout the post-closure period.

40CFR, Part 265.94
    -  The owner or operator (for ground water monitoring purposes)  must keep
       records   of required  analyses throughout the active life of the facility, and,
       for disposal facilities,  throughout the post-closure care period as well.

                                   TSCA

40CFR, Part 761.180 (Subpart J)
    -   Documents that include  the;  dates, ID of facility & owner of Facility from
       whom  whom  PCBs  were received,  Dates of PCB  disposal or  transfer,
       summary  of total    weight of PCBs, and total number of PCB articles
       received or transferred for  5 years after the facility is no longer used for the
       storage or disposal of PCBs (Chemical landfills must keep this documentation
       at least 20 years after the landfill is no longer used for the disposal of PCBs.
       Incineration facilities  must collect and maintain  data  on PCB incineration
       rates & quantities,  combustion temperatures, stack emissions,  monitoring
       data for 5 years from the date of collection.
                                   11-363

-------
                                                                      Page 3
40CFR, Part 761.180(a)
    -  Requires that owners or operators of chemical waste landfills collect and
      maintain water analyses & operating records for at least 20 years after the
      landfill is no longer used for the disposal of PCBs.

                                   OSHA

29CFR, Part 1904.6
   -  Requires that "...logs and summaries of occupational injuries, supplemental
      records of each occupational injury, and annual summaries of injuries ... be
      maintained for 5 years following the end of the year to which they relate."
      Failure to maintain these records  shall be punished by a fine of not more
      than $10,000, or by imprisonment,  for not more than 6 months, or both. (29
      CFR Part 1904.9)

ACCOUNTABILITY

For the hazardous waste facility, recordkeeping and data management is clearly a
long-term responsibility that, if neglected, will result in substantial fines. An analysis
of administrative  actions  initiated  against   regulated  facilities by  the  U.S.
Environmental Protection Agency reveals that from the period 1972-1989 there were
12,250 actions which resulted in over $105 million in penalties where inadequate
recordkeeping was cited as one of the major violations.1

To illustrate the degree to which the recordkeeping practices of a hazardous waste
facility can be held accountable by Federal Regulators, consider the activities of the
National Enforcement  Investigations Center (NEIC) which provides the  U.S.
Environmental Protection Agency's Office of Legal and Enforcement Council with
technical information and evidence in support of potential enforcement actions on
a site's violations of permit conditions or federal regulation.

The scope of an NEIC investigation will generally involve the request to have access
to all  records maintained at the facility.  The  NEIC project team will gain consent
to enter the facility from  the owner or operator and schedule  a date  for the
FY  1989 Enforcement  Accomplishments Report. U.S. Environmental  Protection Agency, Office of
Enforcement, Compliance Evaluation Branch.
                                    11-364

-------
                                                                     Page 4
investigation to begin. NEIC actions can be of 2-3 weeks in duration and usually
require the assistance of facility management staff to obtain requested records and
provide additional information.  Records/data requested generally originate with
randomly selected Waste Manifest Files.  The NEIC team will request all related
paperwork and data associated with select Manifests. Objectives of this request are
to:

    1)  Track the movement of waste  streams through the facility
    2)  Verify that all documentation is traceable to original manifest, and is
       logically filed and retrievable.

Related paperwork that must be produced for the NEIC investigators  includes the
operational records (weight tickets, time & date stamps, logbook pages, records
which  detail  the  movement  of the waste, charts from emission monitoring  for
incinerators, location of drums), the laboratory data (analytical raw data, instrument
readouts, result summaries, log books, QA/QC checks, and QC tests),
and residue management records  (for incinerators).

In order for a facility to successfully met the demands of an NEIC  investigation and
provide timely retrieval of requested records, the facility's  recordkeeping  system
must be  in order  to demonstrate to regulators that it  is in compliance with the
recordkeeping requirements of the Code of Federal Regulations and their operating
permit. It would be extremely damaging for the facility to be unable to produce a
complete tracking record for a Waste Manifest- with the result being additional fines
and disruption of normal activities. Beyond this specific example, try to imagine the
impacts to a company's operation if a body of records and  data  were lost  due to
fire, flood, theft, or slow deterioration  in poor storage conditions.  It is for these
reasons  that  a systematic-  proactive  approach has been  developed by  many
companies to provide protection to critical records and data.   The  need for such  an
approach for a hazardous waste TSD facility is no less important.
DATA MANAGEMENT ISSUES FOR HAZARDOUS WASTE FACILITIES

The objectives of a records/data management program are simply stated to:

    1) Furnish accurate and complete information when it is required
                                   11-365

-------
                                                                          PageS
         to manage and operate the organization efficiently.

      2) Process recorded information as efficiently as possible.

      3) Render maximum service to  the customer(user of the  records)2

    In addition to these goals, the organization must also ensure that  the records and
   data they have on hand will be admissible in court to defend their actions/decisions
   which may have occurred much  earlier in  time.   The  existence of  a record
   management program satisfies the Uniform Rules of Evidence requirement that a
   process be in place to produce an accurate result and that the records created by
   the organization are trustworthy.3 The organization's  records management program
   must have written procedures, training, and regular audits in  order to demonstrate
   that the organization carefully developed its records program, that staff was fully
   aware of the  recordkeeping requirements, and that  the  procedures were actually
   followed by organization's staff.

   The components of a records management program  for a hazardous waste facility
   must take into account the following requirements:

      1) Long term retention of data  (over 30 years),

      2) Ability  to  retrieve analytical records  and data  based on waste manifest
         numbers, customer  profile IDs, Dates of tests,

      3) Timely  responses for customer requests for  information to  decision waste
         streams, and to   recertify wastes for final disposition.

   The requirement to maintain facility operating records and analytical data for such
   long periods of time makes it difficult to rely  exclusively on  a paper-based system
   of recordkeeping to stay in compliance with federal regulation.  First, paper simply
   will not last  as long  as the  law requires. Secondly, paper-based recordkeeping
2
   Information and Records Management, Robek, Brown^nd Maedke, Glencoe Publishing Co. 1987

   Donald S. Skupsky, Legal Requirements For Microfilm, Computer and Optical Disk Records, Information
   Requirements Clearinghouse 1991.
                                       11-366

-------
                                                                    Page 6
systems take up a large amount of space-space that is  limited or unavailable in a
laboratory environment.   Lastly,  paper-based  systems   provide no information
protection to the organization in instances of fires, floods or misfiling of data. It is
for these reasons that a forward-thinking records management program should plan
and provide for the long-term preservation and security of analytical  data  by
replacing (or reducing) the organization's dependance on paper-based systems with
other media.
In those instances where it is impossible to incorporate microfilm or image scanning
technology, the organization must provide for controlled climate, secure storage that
meets federal guidelines for fire protection.

LABORATORIES

Records  management responsibilities for any organization presents  a significant
challenge to management. In the hazardous waste industry, however, the presence
of laboratories dictates an even higher degree of data complexity that the program
must address.  The sophistication  of modern-day laboratory instruments and their
ability to produce data means that the records program will have to take into
account many forms of output that become part of the analytical record of the waste
disposal  decision.  Examples of different media produced by laboratories are:

   1)  Perkin Elmer 5000 writes data on 5 1/4" floppy disks

   2)  Jarell Ash ICAP writes data on  RLO1K-DC disk packs or 158 mb tape
       cassettes

   3)  Leeman ICAP writes data on paper tapes

   4)  Hg CV Instrument produces data compilations on thermal paper.

   5)  INCOS Mass Spectrometry Instruments backup  data onto  45 mb tape
       cassettes w/ IDOS as the system's operating system.

   6)  Parr Bomb Calorimeter Instrument produces data in the form of 3 and 1/2"
       paper rolls.

   7)  Logbooks which provide indexes to the computer media noted above.
                                    11-367

-------
                                                                      Page 7
The  presence of such  varied  forms of  information  media  dictates two   key
components of data management strategy for a laboratory environment.  One, the
nature of electronic media enables the creator to routinely back-up or duplicate data
from the instrument's hard disk onto a portable media such as floppy disks and
transfer the back-up media to an off-site storage facility or on-site vault designed
for computer media. Secondly, although it is relatively easy to create back-ups, the
media itself has never been considered an archival storage media and is subject to
data loss over a period of time. It is for this reason that a sound data management
policy must provide  for periodic conversions of data from old media to new media
to arrest any possible loss of information due to the age of the original magnetic
storage device. The  greatest threat to the retrievability of electronic data, however,
is neither a physical  calamity or human error. The greatest challenge to maintaining
control of electronic media is the continual hardware/software technology migration
that the computer industry is subject to.  New hardware means different size  tape
drives and  new operating systems.  New software releases are not automatically
compatible with  earlier versions.  For information  that must  be  maintained and
made available for  periods in excess of 30 years, the organization can certainly
expect to have a significantly different computer hardware configuration and new
software requirements than originally where put into place.

In order to ensure  that data remains accessible to the organization, the persons
responsible for data  management  must rigorously review the impacts of new
hardware and software purchases on data recovery and make necessary conversions
before the old equipment leaves the site.

Another major issue facing any  organization which desires  to  develop  a  data
management program is in the selection of media to provide long-term protection
for their information. Storage space reductions, time to access files, admissability
of media in  legal actions, cost, longevity of media, and the type of information
being recorded  are all  factors with varying degrees of priority for different
organizations.  For  the hazardous waste industry, however, primary consideration
should be given to  the media type which  satisfies  its need to keep information
secure for the required periods of retention.  As previously noted,  paper-based
systems  are vulnerable to natural  disasters, take an increasingly larger amount  of
space away from staff and equipment,  are subject to misfiles, and, perhaps worst
of all, as soon as the file leaves the desk of the user, represents a loss of staff time
to retrieve the file for reference purposes.
                                    1-368

-------
                                                                     PageS
  comparative analysis of information media types follows:
  Media Type
Disadvantages
                               Advantages
  Paper     -Traditionally preferred in
               legal actions
             -Requires no special viewing
              device
             -Comfort level of users is high
                                            -Requires large storage area
                                            -Deteriorates over time
                                            -Easy to misfile
                                            -Can only be indexed one way
  Microfilm -All courts & governmental      -The equipment needed to read
              agencies will accept as evidence  film is bulky and must be in
             -Recognized by ANSI as an
              archival media (silver based
              film will last 200 years)
             -Reduces storage space by 95%
                                             a common access area
                                            -Film (without computer aided
                                             computer aided software) is
                                             slow to load and retrieve the
                                             the desired image.
                                            -The hard copy record must be
                                             sent off site to be photographed
                                             and processed before the film is
                                             available.  Information is not
                                             available for this period of time.

Optical Disk -Provides the fastest, most flexible   -Admissibility in court not
Imaging      access to documents (files can be
Technology   indexed numerous ways)
             -With the existence of a PC or
              terminal on a desk the information
              is sent to the user in seconds.
             -As with traditional computer media,
              optical disks can be easily dupli-
              cated for offsite security
             -If indexed properly, it is
              impossible to misfile or lose
              a file
                                                established
                                              -No industry standards
                                    1-369

-------
                                                                      Page 9
              -One 12" Optical Disk can hold up to
               260,000 documents (118 cu. ft. or
               15 legal sized file cabinets)

The media that the organization selects to establish a records management program
will have to be evaluated against these characteristics-with the ultimate decision
based upon its own particular concerns. The rate at which companies that are
currently receiving and maintaining great amount of information  are installing
imaging systems, however, demonstrates that its ability to send information
directly to users in very short periods of time establishes it as the office technology
of the  future.  The U.S.  Environmental  Protection Agency has issued  a position
relative to  this technology-stating that it is permissible to  maintain compliance
information on electronic imaging systems, but due to the lack of industry standards
on optical disk technology, recommends that the original paper records also be
maintained. The  legal admissibility  and industry  standards  concerns  are being
addressed at the  present time  and will soon not be  obstacles in evaluating the
suitability of this technology for an organization in the hazardous waste field.

ESTABLISHING  THE RECORDS MANAGEMENT PROGRAM

A records & data management program for an organization in the hazardous waste
industry must have as its principle objectives the following:
   1) The protection and security of analytical data created in support of
      waste treatment and disposal decisions.

   2) The maintenance of the facility operating record and the ability to
      track waste streams & verify that all permit requirements have been
      satisfied in treatment and final disposition.

   3) The ability for staff to quickly access records and data to respond to
      customer or regulator inquiries.
The facility must establish a written procedure or program for the management of
its official records and data. The first place to start is by  conducting a records
survey (inventory of facility records).   The survey will  identify the  number of
different record groups, the volume(# of file drawers or storage boxes), the  current
                                    1-370

-------
                                                                         Page 10
   media(paper, magnetic tape,  etc.), the present  location(s)  of records/data, the
   interrelationships between different facility groups in creating records, and how the
   records and data are used by the facility. After completion of the initial survey, the
   person responsible must research and establish the legal retention requirements and
   any existing company policies  which will determine how long the records must be
   maintained.

   The  survey when in  final form  should  be  reviewed by  the company's  upper
   management  for approval and designation as  official company policy in  regard to
   recordkeeping requirements.   The  survey is now  the facility's  official records
   retention schedule and is a key component of the  requirement to have a written
   plan or  program in  place to insure that  the facility's  records are deemed
   "trustworthy."    The  facility's  recordkeeping program  will   also  require  the
   development   of a  corporate-wide directive  or  procedure  which  establishes the
   standards that must  be followed in maintaining those records related to disposal
   decisions, waste receiving,  processing, disbursement of waste product, supporting
   analytical data, and quality assurance.

   The recordkeeping policy should address the following topics:

      1)  Permit Requirements-the policy must  include a statement that  all
          record and data will be maintained in accordance with facility permit
          conditions.

      2)  Retrieval Requirements-the policy  must make clear the requirement
          to keep records in  sufficient  detail in order to be able to retrieve
          analytical data  and  quality assurance  records for individual waste
          samples and  manifests for the duration  of  the  records retention
          period.

      3)  Records Storage  Area-the  policy  should  state  that  all records  be
          maintained in  secure  storage under  conditions  that  will  prevent
          deterioration  of the information for the    duration of the retention
          period4. The conditions for  storage could be those  established by the
4   Good Automated Laboratory Practices. Office of Information Resource Management, U.S. Environmental
   Protection Agency.
                                        1-371

-------
                                                                     Page 11
      National Archives and Records Service5,  for electronic  media the
      storage facility  standards outlined by  the   National Bureau  of
      Standards  in  Special  Publication 500-101  "Care and  Handling  of
      Computer  Magnetic Storage Media," and the standards  set by the
      National Fire Protection  Association publications           NFPA
      232AM "Archives and  Records Centers" and NFPA 232 "Protection
      of Records."

   4) Analytical Records-the policy must state that analytical  data must be
      logically filed, manual entries  be made  in  permanent,  reproducible
      ink,  must be dated and signed or initialed by the technician.   Any
      changes to the data must be crossed out with a single line, dated and
      initialed, with no obscuring of the corrected data.

   5) Logbooks-must be used whenever information cannot be recorded on
      the analytical data, loose  paper must be permanently affixed to the
      logbook & dated/initiated.

   6) OA  Records-must be  retained as outlined by the  facility operating
      permit and company policy.

Once the  records  survey  and  the  Standard  Operating  Procedure  have  been
completed, the person responsible  for the program must investigate the use of
microfilm or optical disk technologies and the suitability of each for converting the
facility's data/records into a media that will provide for longevity and security of
analytical information.  Either technology will reduce storage  space requirements,
allow the duplication and off-site storage of back-up documentation, and insure the
integrity of information for the terms of the retention periods. The ability of optical
disk technology to  provide  rapid  access to  detailed inquiries,  however, has
established this information  media as  the preferred method of managing  large
amounts  of data for those organizations which  wish to remain responsive to their
clients and have a significant  recordkeeping burden.

A model of a typical imaging system follows:
Center Operations Division, Office of Federal Records Centers, National Archives and Records Service,
General Services  Administration(August 1976)
                                    11-372

-------
                                                                Page 12
                                           Manage/Store
  Recommended Configuration
            for
Chemical Waste Management, Inc.
                                1-373

-------
                                                                     Page 13
For a waste disposal decision process, an organization could electronically scan the
Generator's Waste Profile  Sheet, all  analytical  test results, and the Disposal
Decision.  An electronic folder would be created and indexed by; Customer Profile
Number, Customer Name, Laboratory Sample ID  Number, and storage box # for
the original paper documents. The electronic file could now be retrieved by any of
the index keys in a matter of seconds.  The original paper file  could be transferred
to archival storage as the electronic data should satisfy all subsequent information
needs. The data written on the imaging system's optical disk is  routinely  duplicated
on a separate disk for security purposes and stored in an off-site location. Customer
service is  enhanced when a customer requests a copy of an  entire file or only a
specific document, the imaging system has the capability to  transmit a facsimile
directly from the provider's PC to the requestor's  facsimile transmission device. If
Waste Decision Files are indexed by date of decision, the user group would have the
ability to retrieve all files prior to the expiration of the original decision in order to
recertify waste streams for disposal.

The effect of changes in governmental regulation might dictate the reconsideration
of a number of Waste Decision Files and imaging technology  has the capability to
retrieve electronic files by "key word" searches.  For disposal sites, having imaging
technology would enable the facility to index  all required documents to the original
waste manifest number. Raw Data if it is maintained separately from  the rest of
the Decision File  could be indexed to  the Customer's  name and the Testing
Laboratory's sample control number system.  With this technology, an organization
could truly have control of their files and put their information to  work for them.

CONCLUSION

Proper data management techniques for a hazardous waste facility must be based
on the realization that  required recordkeeping is not only the obligation to create
certain  forms  and data,  but that  this information must be retrievable for the
duration of legal periods of retention.  This requirement dictates  that the facility
apply a  systematic approach to record/data creation, active use, and long-term
storage. The belief that the filing of a  record in a file cabinet or  storage box has
provided adequate protection to the company's interests has been the source of
much later grief and   unnecessary  expense. As  is the case with any regulated
industry,  the data that is maintained for compliance purposes is (and will be for
long periods of time) an extremely valuable  asset  to a company and requires the
                                   11-374

-------
                                                                   Page 14


implementation of a program to guarantee the longevity of the records/data not
only for  compliance reasons, but also to become a positive asset in company
operations.
                                    11-375

-------
AIR/GROUNDWATER

-------
95                     NEW DIRECTIONS IN RCRA GROUND-WATER
                                  MONITORING REGULATIONS
        James R. Brown. Environmental Scientist, Vernon B. Myers, Chief, Monitoring and
        Technology Section, Office of Solid Waste (OS-341), U.S. Environmental Protection
        Agency, Washington, D.C., 20460; Ann E. Johnson, Hydrogeologist, Regulatory Analysis
        and Support Division, Science Applications International Corporation, McLean, Virginia
        22102
        ABSTRACT

        EPA soon expects to issue a Notice of Proposed Rulemaking (NPRM) in the Federal Register concerning
        amendments to the ground-water monitoring requirements for land-based hazardous waste treatment, storage,
        and disposal facilities (TSDFs) that are regulated under Subtitle C of the Resource Conservation and Recovery
        Act (RCRA).  The notice will propose amendments to the list of ground-water monitoring constituents for
        TSDFs, Appendix IX to Title  40, Code of Federal Regulations. Part 264, ("Appendix IX"), and require that
        certain procedures be used in the design, installation, and operation of ground-water monitoring systems at
        TSDFs.  The proposed changes to Appendix IX include the addition of a required list of detection monitoring
        analytes  (Appendix IX-A), the deletion or addition of several Appendix IX compounds due to analytical
        considerations, and a site-specific variance from the annual Appendix IX analysis requirement during compliance
        monitoring. The proposed standards for ground-water monitoring procedures will be specified in revisions to
        Chapter  Eleven of  the  U.S.  EPA  document  SW-846,  Test  Methods for  Evaluating  Solid Waste,
        Physical/Chemical Methods," (Third Edition), or more generally referred to as "Chapter Eleven of SW-846."
        Chapter Eleven of SW-846 specifies requirements concerning the characterization of site hydrogeology, placement
        of detection monitoring wells, monitoring well design and construction, and ground-water sampling and analysis
        programs.   All hydrogeologic investigations and monitoring activities must comply with the methods and
        procedures required in Chapter Eleven of SW-846.

        The proposed  requirements in Chapter Eleven of SW-846 represent EPA's establishment of qualitative data
        quality objectives (DQOs)  for the RCRA ground-water monitoring program.   At the same time, EPA's
        Environmental Monitoring Systems Laboratory in Las Vegas, Nevada (EMSL-LV) is in the process of evaluating
        the efficacy of establishing quantitative DQOs for ground-water monitoring system performance. If appropriate,
        quantitative DQOs would allow  EPA to determine the minimum  number and location of monitoring wells
        required to achieve a specified probability of leak detection. This paper will summarize some of the methods
        that have been investigated to establish quantitative DQOs for RCRA ground-water monitoring.
        INTRODUCTION

        Subtitle C of RCRA  creates a comprehensive program for the safe  management  of
        hazardous waste. Owners and operators of facilities that treat, store or dispose of hazardous
        waste must comply with standards established by EPA that are "necessary to protect human
        health and the environment." Implementation of these standards occurs through permits
        issued to owners and operators by EPA or authorized States.

        Standards for protecting ground water from releases of hazardous wastes from permitted
        TSDFs were promulgated by EPA in 1982 (47 FR 32274; July 26, 1982),  and are codified
                                               1-379

-------
at 40 CFR Part 264,  Subpart  F ("Subpart F1).  Subpart F requires  TSDF owners and
operators to characterize their site's hydrogeology, install and maintain a ground-water
monitoring system,  and to sample  and  analyze ground water at specific intervals to
determine whether hazardous wastes or hazardous waste constituents from the facility are
contaminating ground  water.

The Subpart F requirements consist of a three-phase ground-water monitoring program:
detection monitoring,  compliance monitoring, and corrective action.  The first phase,
detection monitoring,  involves at least semi-annual monitoring of "indicator" parameters,
waste constituents, or reaction products specified in the facility permit that provide a reliable
indication of the presence of hazardous constituents in ground water. Owners and operators
employ detection monitoring at new land disposal facilities and at land disposal facilities not
believed to be releasing contaminants to ground water.  If monitoring indicates that the
concentration of a monitored constituent has shown a statistically significant increase over
background concentrations, then EPA requires analysis for all Appendix IX constituents and
the facility enters compliance monitoring.

Compliance monitoring, the second  phase of  ground-water monitoring, requires at least
semi-annual monitoring for constituents identified in the facility permit, including  those
constituents detected in ground-water during the detection monitoring program.  A facility
in compliance monitoring must also monitor ground water for all Appendix IX constituents
at least annually and  report the concentration of any new constituent detected to the
Regional Administrator. All detected Appendix IX constituents are then monitored at least
semi-annually during  compliance monitoring.   The concentrations  of all compliance
monitoring constituents are compared  to concentration  limits  specified in the facility's
permit. Concentration limits are an element of the  facility's ground-water  protection
standard used to determine if ground-water contamination has occurred.

If any compliance  monitoring constituent  shows a  statistically significant increase in
concentration above  the concentration limits  set forth in the facility's ground-water
protection standard, the facility enters  the  third phase of ground-water monitoring,
corrective action. In corrective action, the facility owner or operator is required to "remove
or treat hi place" all constituents that exceed the allowed concentration limits specified in
the facility permit.  The monitoring associated with corrective action must demonstrate the
effectiveness of the clean-up and must be able to determine whether any other constituents
are entering the ground water at concentrations above the concentration limits.

The 1982 regulations required that contaminated  ground water be  analyzed for all
constituents contained in Appendix VDI to Part 261 ("Appendix VIII"). While appropriate
for hazardous waste listing purposes, the Appendix VIII list presents a number of difficulties
when used for purposes of ground-water monitoring (RMAL, 1984;  U.S. EPA, 1987c).
These difficulties include practical and analytical problems such as monitoring for  large
categories of chemicals, lack of availability of some analytical standards, and the lack of
reliable analytical methods for many constituents. Other problems relate to the dissociation
                                   11-380

-------
or actual decomposition of many Appendix VQI constituents in water, rendering monitoring
for these constituents impractical. To address these analytical problems, EPA proposed on
July 24,1986 to replace the requirement to monitor for all Appendix VIII constituents with
a requirement to monitor for a new series of ground-water  monitoring analytes listed in
Appendix IX. Appendix IX was promulgated as a final rule on July 9, 1987 (52 FR 25942),
and included those constituents in Appendix VIII that had available analytical methods for
the ground-water matrix, plus 17 constituents routinely monitored in the Superfund program.
DISCUSSION OF PROPOSED CHANGES TO SUBPART F

Detection Monitoring Analytes (Appendix IX-A)

The Agency expects to propose amendments to the Subpart F regulations to change the
provisions governing the selection of detection monitoring analytes.   The regulations
currently require an owner or operator of a facility in detection monitoring to monitor for
indicator parameters,  waste constituents,  or reaction products that provide a reliable
indication of the presence of hazardous constituents in ground water. Studies have shown
that volatile organic compounds (VOCs) serve as reliable leak indicators at hazardous waste
TSDFs because they frequently occur in leachate and contaminated ground water (Eckel et
al., 1985; Plumb, 1987; Lawless, 1987; Rosenfeld, 1990). Inorganic constituents (e.g., metals)
have also been reported to occur in leachate from hazardous waste TSDFs (Bramlett et al.,
1987; WMI,  1990 and 1991), and in ground water in the vicinity of TSDFs (Lawless, 1987).
In consideration of these data, the Agency expects to propose a list of detection monitoring
constituents  known as Appendix IX-A.

The Appendix IX-A constituents are a subset of the Appendix IX constituents, and consist
of 48 VOCs and 16 metals that the Agency believes would serve as good "release indicators"
for hazardous waste disposal sites that receive a variety of wastes.  The  specific VOCs
contained in Appendix IX-A were  chosen primarily by determining which VOCs could be
identified by gas chromatography/mass spectroscopy (GC/MS) with a reasonable degree of
precision and accuracy (Lawless, 1990). Other considerations regarding the selection  of
VOCs was their reported frequency of occurrence in leachate and ground water (discussed
above). The GC/MS  method recommended minimizes the number of separate analyses
required to determine the concentration of many VOCs in a ground-water sample (e.g., all
VOCs in Appendix IX-A can be determined in a single GC/MS analysis). Thus, the use of
GC/MS procedures for ground-water analyses provides  reliable  results  and  conserves
analytical resources. Because of these advantages, EPA assumed that the newly proposed
SW-846 GC/MS Method 8260, a modification of Method 8240 (54 FR 3213; January 1989)
in SW-846 would be the standard method used for this analysis.  All but two (i.e., barium
and vanadium) of the 16 metals on Appendix DC-A have been on EPA's Priority Pollutant
List since 1979 (U.S.  EPA,  1979), and can be analyzed  by inductively coupled plasma
emission spectroscopy (ICP) or atomic absorption spectroscopy (AA).  The analytes that
comprise Appendix IX-A are listed in Table 1.
                                   11-381

-------
Although Appendk IX-A may be appropriate for the majority of TSDFs, situations may
arise that would warrant further tailoring of the list to meet site-specific concerns.  For
example, a facility that manages only smelter ash may not normally need to monitor for
VOCs, because VOCs would not likely be  present in the ash, and therefore, would not
provide a  reliable indication that the regulated unit was releasing hazardous wastes to
ground water. In order to retain flexibility in the current regulations, alternative provisions
will allow the Regional Administrator to  add or delete constituents from Appendix IX-A
after considering the waste managed hi the  regulated unit.
Proposed Revisions to Appendix IX

Under the Subpart F ground-water monitoring program, Appendk IX is the "master" list of
ground-water monitoring analytes. Appendix IX constituents are monitored at facilities that
are in compliance monitoring or corrective action. Appendix IX contains 222 constituents
that in 1987 had analytical methods that were verified to a sufficient degree, and that were
amenable  to ground water monitoring on a routine basis (U.S. EPA, 1987c).  The 222
constituents on Appendix IX consist of 17 metals and metalloids, 2 inorganic ions, 6 classes
of organic compounds (i.e., chlordanes, toxaphenes, PCBs, PCDDs, PCDFs, and xylenes),
and 197 specific organic chemicals.

EPA expects to propose removing eleven analytes from the current Appendix IX.   The
eleven constituents proposed for deletion were chosen on the basis of new analytical data
that EPA has generated or received since the Appendix IX  rule was first promulgated in
1987.  These new data indicate that, for the eleven compounds proposed for deletion, the
analytical procedures described in SW-846 do not provide consistently acceptable results in
terms of method performance for determining their concentration in ground water (Lawless,
1990). In addition, 4-nitroquinoline 1-oxide is being proposed for deletion from Appendix
IX because SW-846 does not  provide QC criteria or accuracy and precision data for its
analysis. Further, since 4-nitroquinoline 1- oxide is an experimental pharmaceutical, it was
not produced in commercial chemical quantities and therefore has a low frequency of
occurrence hi ground water near TSDFs (Plumb, 1991).  The chemical compounds proposed
for deletion from Appendix IX are listed in Table 2.

The Agency expects to propose the addition of six constituents to Appendix  IX.  All six
constituents  are members of the volatile organic class of compounds, and, as discussed
earlier, VOCs have been shown to serve as good release indicators. In addition,  all six
VOCs are amenable to analysis by GC/MS (Method 8260 in SW-846), and are included in
Appendix VIE to Part 261 as part of small classes of hazardous constituents.  Furthermore,
five of  the  six  compounds are halogenated alkanes, many of which are suspected
carcinogens. The six constituents suggested for addition to Appendix IX are listed in Table
3.
                                   11-382

-------
                                             TABLE 1
          LIST OF CONSTITUENTS FOR DETECTION MONITORING rAPPENDIX IX-A1
Common Name

1.  Acetone
2.  Acrylonitrile
3.  Benzene
4.  Bromochloromethane
5.  Bromodichlorornethane
6.  Bromoform; Tribromomethane
7.  Carbon disulfide
8.  Carbon tetrachloride
9.  Chlorobenzene

10.  Chloroethane; Ethyl chloride
11.  Chloroform; Trichloromethane

12.  Dibromochloromethane
        Chlorodibromomethane
13.  l,2-Dibromo-3-chloropropane
14.  1,2-Dibromoethane;
       Ethylene dibromide
15.  1,2-Dichlorobenzene
16.  1,4-Dichlorobenzene
17.  trans-l,4-Dichloro-2-butene
18.  1,1-Dichloroethane; Ethyldidene
        chloride
19.  1,2-Dichloroethane;
        Ethylene dichloride
20.  1,1-Dichloroethylene;
        1,1-Dichloroethene;
        Vinylidene chloride
21.  cis-l,2-Dichloroethylene;
        cis-l,2-Dichloroethene
22.  trans-l,2-Dichloroethylene;
        trans-1,2-Dichloroethene
23.  1,2-Dichloropropane;
        Propylene dichloride
24.  tis-l,3-Dichloropropene
25.  trans-13-Dkhloropropene
26.  Ethylbenzene
27.  2-Hexanone;  Methyl butyl ketone
28.  Methyl bromide; Bromomethane
29.  Methyl chloride; Chloromethane
30.  Methylene bromide;
        Dibromomethane
31.  Methylene chloride;
        Dichloromethane
32.  Methyl ethyl  ketone; MEK;
        2-Butanone
33.  Methyl iodide; lodomethane
34.  4-Methyl-2-pentanone
        Methyl isobutyl ketone
CASRN2

67-64-1
107-13-1
71-43-2
74-97-5
75-27-4
75-25-2
75-15-0
56-23-5
108-90-7

75-00-3
67-66-3

124-48-1

96-12-8
106-93-4

95-50-1
106-46-7
110-57-6
75-34-3

107-06-2

75-35-4
 156-59-2

 156-60-5

 78-87-5

 10061-01-5
 10061-02-6
 100-41-4
 591-78-6
 74-83-9
 74-87-3
 74-95-3

 75-09-2

 78-93-3

 74-88-4
 108-10-1
Common Name1               CASRN2

35. Styrene                   100-42-5
36. 1,1,1,2-Tetrachloroethane   630-20-6
37. 1,1,2,2-Tetrachloroethane   79-34-5
38. Tetrachloroethylene;        127-1&45
        Tetrachloroethene;
        Perchloroethylene
39. Toluene                   108-88-3
40. 1,2,3-Trichlorobenzene     87-61-6
41. 1,1,1-Trichloroethane;      71-55-6
        Methylchloroform
42. 1,1,2-Trichloroethane       79-00-5
43. Trichloroethylene;          79-01-6
        Trichloroethene
44. Trichlorofluoromethane;    75-69-4
        CFC-11
45. 1,2,3-Trichloropropane     96-18-4
46. Vinyl Acetate              108-05-4
47. Vinyl Chloride             75-01-4
48. Xylene (Total)             1330-20-7
49. Antimony                 (Total)
50. Arsenic                   (Total)
51. Barium                   (Total)
52. Beryllium                 (Total)
53. Cadmium                 (Total)
54. Chromium                (Total)
55. Cobalt                   (Total)
56. Copper                   (Total)
57. Lead                     (Total)
58. Mercury                  (Total)
59. Nickel                    (Total)
60. Selenium                 (Total)
61. Silver                    (Total)
62. Thallium                 (Total)
63. Vanadium                (Total)
64. Zinc                      (Total)
                                              1-383

                                                 5

-------
                                   TABLE 2
           Chemical Compounds Proposed for Deletion from Appendix IX

            Common Name1                       CAS RN2

      1.  Aniline                                 62-53-3
      2.  Aramite                                 140-57-8
      3.  alpha, alpha-Dimethylphenethylamine      122-09-8
      4.  1,4-Dioxane                             123-91-1
      5.  Hexachlorophene                        70-30-4
      6.  4-Nitroquinoline 1-oxide                  56-57-5
      7.  N-Nitrosomorpholine                     59-89-2
      8.  Pentachloroethane                       76-01-7
      9.  2-Picoline                               109-06-8
      10. Pyridine                                110-86-1
      11. Tetraethyl dithiopyrophosphate            3689-24-5
                                   TABLES
            Chemical Compounds Proposed for Addition to Appendix IX
            Common Name1

      1.  Bromochloromethane
      2.  cis-l,2-Dichloroethylene
      3.  1,3-Dicbloropropane
      4.  2,2-Dichloropropane
      5.  1,1-Dichloropropene
      6.  1,2,3-Trichlorobenzene
CASRN2   APPENDIX WI REFERENCE

74-97-5      Halomethane, N.O.S.
156-59-2    1,2-Dichoroethylene
142-28-9    Dichloropropane, N.O.S.
594-20-7    Dichloropropane, N.O.S.
563-58-6    Dichloropropene, N.O.S.
87-61-6      Chlorobenzene, N.O.S.
   1   Common  names  are  those  widely  used in government regulations,  scientific
publications, and commerce; synonyms exist for many chemicals.

   2 Chemical Abstracts Service registry number.  Where "Total" is entered, all analytes
in the ground water that contain this constituent are included.
                                     11-384

-------
Variance From the Annual Appendix IX Analysis During Compliance Monitoring

During compliance monitoring, the owner or operator is required to monitor for parameters
identified in the facility permit at specified frequencies.  In addition, ง 264.99(g) requires
the owner or operator to analyze samples from all wells located at the point of compliance
for all Appendix IX constituents at least annually. If additional constituents are found, then
the owner or operator must report their concentrations and list these constituents in the
facility permit. All additional constituents that are subsequently listed in the facility permit
as a result of the annual Appendix  IX monitoring, also form the basis for compliance
monitoring, and are sampled and analyzed at least semiannually.

Experience has shown that for some hazardous waste TSDFs,  annual monitoring for all
Appendix IX constituents may not be necessary.  Certain analytes  such as EPA's Priority
Pollutants have been shown in studies to have a higher frequency of occurrence in leachate
and contaminated ground water than do other constituents in Appendix IX (WMI, 1990 and
1991; Plumb, 1991). Furthermore, in each of these studies, "non-priority pollutant" Appendix
IX constituents were not detected in the absence of priority pollutants.  This  suggests that
routine monitoring for non-priority pollutants may not be necessary at every TSDF. In light
of this new  information,  EPA expects to propose a site-specific variance to the annual
Appendix IX monitoring requirements under certain circumstances.  Such a variance could
involve performing an abbreviated Appendix IX analysis on an annual basis.  To exclude a
constituent from the annual Appendix IX analysis, the owner or operator would be required
to demonstrate that the  constituent  could not be present  in the waste managed by the
facility (either as a constituent of the waste, or as a reaction product), and is not present in
the facility's soil and ground water. The benefits of the variance would be realized primarily
for those constituents that require special analytical methods (e.g., TCDD) rather than for
those that are amenable to analytical "scan" techniques such as  SW-846 method 8260.

EPA expects that any variance from the annual monitoring  requirements for  Appendix IX
constituents would not relieve the owner or operator from ever monitoring for the excluded
constituent(s).  The initial,  full Appendix IX analysis would still be required in detection
monitoring.  Retention of this requirement is necessary to characterize the nature and extent
of a release and could be used to demonstrate that the excluded constituents are not present
in ground water at the facility. In addition, if a successful demonstration is made and the
Regional Administrator excludes constituents from  the annual Appendix IX compliance
monitoring requirements, the  owner or operator would be  required to monitor for all
Appendix IX constituents (including the excluded constituents) at least once every five years,
when the permit is  usually reviewed or renewed.  If at any time a facility that  has a
monitoring exclusion began to  receive or generate wastes that contain  any excluded
Appendix IX constituents, the facility would be required to resume annual monitoring for
the appropriate analytes. Likewise, these steps would need to be followed if a treatment
process was  modified, or  a new one begun, that resulted in the production of the excluded
constituents.
                                   11-385

-------
CHAPTER ELEVEN of SW-846: GROUND-WATER MONITORING

One of the major topics that arose during EPA's recent review of over forty TSDF operating
permits (Permit Quality Review) throughout the nation was the importance that each of the
numerous  steps taken by an  owner or operator to comply with RCRA ground-water
monitoring requirements has in determining the quality of the data produced by monitoring
networks (U.S. EPA,  199 la).  One  example  of  this is  the value of an  adequate
characterization of site hydrogeology. Improper site characterization can lead to incorrect
placement of monitoring wells, or a failure to recognize ground-water flow paths and
contaminant migration pathways.

In addition, inspections conducted by EPA's Hazardous Waste Ground-Water Task Force
(HWGWTF) during the years 1984 to 1987 identified deficiencies in existing ground-water
monitoring systems and determined that many of the deficiencies resulted from owners and
operators collecting poor quality hydrogeologic data, collecting inadequate quantities of
hydrogeologic data (or misinterpreting such data), and using improper sampling and analysis
techniques (U.S. EPA,  1988).  The deficiencies almost always involved technical areas for
which die RCRA regulations provided the least specificity, but that were covered extensively
in  non-binding  EPA  guidance documents  (e.g., subjects  such  as  hydrogeologic
characterization, well construction and location, and ground-water sample collection).  The
HWGWTF and  Permit Quality  Review  experience highlighted  the need to develop
nationally consistent regulatory requirements addressing the process an owner/operator must
follow to characterize site hydrogeology, to design and construct a ground-water monitoring
system, and to collect and analyze ground-water samples.

As a result, EPA expects to propose to require owners and operators to use the methods
described  in proposed  revisions to  Chapter  Eleven of SW-846 when conducting
hydrogeologic investigations, designing and constructing monitoring systems, and performing
ground-water sampling and analyses. Chapter Eleven of SW-846 embodies the Agency's
best judgment and current understanding regarding ground-water monitoring techniques, and
addresses  a  variety of  ground-water monitoring techniques  and  procedures  including:
hydrogeologic characterization, well placement, well design, well drilling, well completion,
well casing materials, well development, well purging, sampling equipment and methods, and
sample handling. EPA does not expect that any  new burdens will be placed on the vast
majority of owners and operators by requiring them to conform to the methods discussed
in SW-846 because  these techniques and methods are based on widely accepted practices
of most geologists and ground-water professionals. Furthermore, for each phase of ground-
water monitoring system design and operation, Chapter Eleven of SW-846 generally offers
several methods that are acceptable depending on the specific hydrogeologic setting of a
facility, the waste management practices, and the waste characteristics. Where specific
techniques or procedures are not provided because of the complexity and site-specific nature
of ground-water monitoring programs, Chapter Eleven of SW-846 provides discussion and
technical guidance on the available alternatives. In these cases, there is a significant amount
of flexibility allowed in the choice of methods used for ground-water monitoring  system
design and operation.
                                  11-386

-------
Data Quality Objectives for Ground-water Monitoring

The Agency has begun using DQOs for evaluating remedial response activities associated
with  Superfund sites to define the type and quality of data required to support specific
regulatory decisions (EPA, 1987a  and 1987b).  DQOs include both qualitative  and
quantitative specifications. Many of the current ground-water monitoring requirements that
are specified in Subpart F are broad-based performance standards that allow the Regional
Administrator, RCRA Part B permit writers, and owners and operators of hazardous waste
TSDF's to account for site-specific factors when designing a ground-water monitoring system
that meets  the requirements  of the RCRA regulations.  The wide variations  in waste
management practices  coupled with diverse hydrogeologic  settings and geochemical
environments across the United States make it difficult to promulgate a regulation specifying
a minimum number of monitoring wells and their location that would be applicable to all
facilities. EPA instead  has relied on technical guidance documents (e.g., U.S. EPA 1986;
U.S. EPA, 1989) and the experience of permit writers to implement these types of general
performance standards on a site-specific scale. Presently, given the current "state-of-the-art"
of ground-water monitoring practices, a qualitative approach to defining the adequacy of
ground-water monitoring systems is the norm. However, significant efforts are underway at
EPA to develop quantitative approaches for designing ground-water monitoring systems.

The Agency continues to focus on efforts  that will improve both the type and quality of
RCRA ground-water monitoring data.  Changes to Appendix IX and the creation of
Appendix IX-A will improve the type of data collected, by changing the constituents for
which owners/operators must monitor. A variance to the annual Appendix IX compliance
monitoring requirement will ensure that meaningful data are collected. The incorporation
of Chapter Eleven of SW-846 into the Part 264 and Part 270 ground-water monitoring
requirements will offer more prescriptive directions on what methods and procedures should
be used in the design and operation of ground-water monitoring systems.  These are part
of the Agency's efforts to establish qualitative data quality objectives for the RCRA ground-
water monitoring program.

Data quality for ground-water sampling and analysis activities is also addressed in Chapter
One  of SW-846 titled, "Quality Control." Chapter One of SW-846 identifies the minimum
quality control (QC) components to be used when performing all RCRA sampling and
analysis activities, and includes the QC information which must be documented.  Chapter
One of SW-846 provides guidance on the development of quality assurance project plans for
field and laboratory work that is conducted in support of the RCRA program. Chapter One
was part of the first update package to SW-846,  third  edition, and  is  mandatory for
compliance with RCRA sampling and analysis requirements.
Quantitative Data Quality Objectives for RCRA Ground-Water Monitoring

The Agency is assessing the feasibility of establishing quantitative DQOs for ground-water
monitoring under 40 CFR Part 264, Subpart F. Quantitative DQOs would be developed for
                                   11-387

-------
each phase of monitoring and would establish numeric standards that specify the level of
performance for RCRA ground-water monitoring systems.  Thus, quantitative DQOs for
detection monitoring could, for example, require that ground-water monitoring networks
achieve a specified  probability of  detecting contamination.   Quantitative  DQOs for
compliance monitoring could require that ground-water monitoring networks achieve a
specified  probability  of  characterizing  the extent  of ground-water  contamination.
Quantitative DQOs for corrective action could require that ground-water remediation efforts
achieve clean-up standards within a specified probability. A similar approach has been used
to support decisions concerning the design of remedial actions for contaminated soils at
Superfund sites (Neptune, et al., 1990).  In all phases of RCRA ground-water monitoring,
quantitative DQOs would allow the Agency to specify the exact number and location of
monitoring wells, and number of ground-water samples, required to achieve a desired level
of performance.

The Agency's Office of Research  and  Development is  investigating the  efficacy  of
establishing quantitatively-based DQOs for ground-water monitoring (U.S. EPA, 199 Ib).
Research plans are oriented toward developing a process aimed at defining, with a specified
probability, that a monitoring well system will detect a release from a TSDF. This process
will still involve the collection of detailed site-specific hydrogeologic data to support the
development of a conceptual model.  This data may then be integrated with a conditional
simulation model and/or  a  contaminant fate and transport model that would predict
preferential flow  paths of contaminant migration and estimate the probability of leak
detection based on monitoring well network configuration.

Relatively early research  performed by  Massmann  and Freeze  (1987), calculated the
probability of contaminant plume detection by monitoring networks. As noted by Meyer and
Brill (1988), however,  these  investigations  stopped  short  of  optimizing ground-water
monitoring network performance (in terms of the probability of detecting a contaminant
plume) by failing  to generate alternative networks that are more efficient with respect to
contaminant plume detection. Meyer and Brill utilized Monte Carlo simulations of plume
releases to develop a method for optimizing the location of monitoring wells.

Quantitative monitoring network design methods offer intriguing advantages over their
qualitative alternatives, and are beginning to find applications at hazardous waste sites. A
two-dimensional deterministic model based on the work of Meyer and Brill has been used
to predict low density, aqueous-phase contaminant plume detection in unconfined aquifers
(Wilson, et al., 1991).  This model offers a quantitative yet user-friendly approach to
monitoring network design. Other applications of quantitative monitoring network design
will likely continue to surface in the literature.

A more recent development of a procedure to  estimate the probability of contaminant
plume detection  uses  geostatistical conditional simulation  and parameter estimation
sequentially to generate contaminant migration pathways (Weber, et al., 1991). Recognizing
that aquifer heterogeneities and the high cost of hydraulic conductivity measurements often
inhibit adequate site characterization, these researchers utilized hydraulic head and available
                                   11-388

-------
hydraulic conductivity measurements to estimate the distribution of flow paths (Figure 1),
and developed a relationship between the probability of plume detection and monitoring
system cost (Figure 2).  The procedure is  also amenable to conditional simulation  of
hydraulic conductivity if sufficient measurements are  available to perform geostatistical
analysis (Weber, et at., 1991).

The  results of research efforts  like those described  above could provide EPA with  a
quantitative means for specifying DQOs for Subpart F  ground-water monitoring networks.
Current limitations of ground-water monitoring, subsurface characterization, and modeling
techniques, however, make it difficult to develop quantitative DQOs (most of the current
applications utilize two-dimensional models).   For example, it may not be  possible  or
practical to design a monitoring system that will detect releases at  a desired probability of
contaminant plume detection.  Before a probability statement can be made, population
characteristics should be known (or assumed to be known). In the context of ground-water
monitoring at TSDFs, the population consists of all possible contaminant migration pathways
in the subsurface.  To characterize this population,  very detailed site characterization
methods and analyses are required. Consequently, a central issue involves the level of detail
that a site characterization must include to define all of the population characteristics. As
discussed above however, surrogate parameters (i.e.,  hydraulic head measurements) for
hydraulic conductivity have been  used successfully to define the spatial distribution  of
contaminant migration pathways and evaluate monitoring well performance where data is
sparse and collection methods are expensive (Weber, et al. 1991).

EPA  will  continue to support the development of quantitative DQOs for ground-water
monitoring under Subpart F as technical advances allow. EPA will use such information to
assess the feasibility of developing quantitative DQOs  for ground-water monitoring. If an
acceptable procedure is developed for establishing quantitative DQOs, it will be proposed
in the Federal Register and formally opened to .public comment.
                                     1-389

-------
FIGURES
                                           Estimated Probability
                                          attribution of Flow Piths
Monitoring Network Cos
Thousands of Dollars
                60


                50


                40


                30


                20


                10
                          Figure 1 (After Weber, et al., 1991)
                                     40
60
80
100
                  Estimated Percent Probability of Contaminant Plume Detection
                      (Based on cost of $5,000 per well, 300 ft spacing)
                          Figure 2 (After Weber, et al., 1991)
                                11-390

-------
SUMMARY

EPA expects  to propose amendments to the Subpart F requirements for ground-water
monitoring at TSDFs.  The proposal will amend  the  list of ground-water monitoring
constituents (Appendix IX) based on analytical considerations; create a subset Detection
Monitoring list of constituents that have been shown to provide a reliable indication of
releases from TSDFs; issue an exclusion variance from the annual Appendix IX Compliance
Monitoring requirement;  and require owners and  operators of TSDFs  to  comply with
ground-water monitoring methods and procedures contained in a revision to Chapter Eleven
of SW-846.

EPA is also conducting research on quantifying ground-water monitoring network design
efficiency. Research efforts are investigating the efficacy of optimizing the probability of
contaminant plume  detection for a  given monitoring well configuration.   The desired
outcome  of this research would allow  for the establishment of quantitative DQOs for
ground-water  monitoring.
ACKNOWLEDGEMENTS

This paper was  written,  in part,  by members of U.S. EPA's Office of Solid  Waste,
Washington, D.C.  It has not  been reviewed by the Agency and the contents  do  not
necessarily reflect the views and policies of EPA. Mention of trade names, commercial
products, or publications does not constitute endorsement or recommendation for  use.
REFERENCES

Bramlett,  J.,  Furman,  C,  Johnson,  A., Ellis, W.D.,  Nelson,  H., Vick, W.H.,  1987.
 Composition of Leachate from Actual Hazardous Waste Sites.  Project Report for U.S.
 EPA (EPA/600.S2-87/043).

Code of Federal Regulations, Title 40, Part 261, Section 261.20. Appendix VIII, Hazardous
 Constituents.

Eckel, W.P.,  D.P.  Trees, and  S.P. Kovel,  1985.  Distribution of Chemicals and  Toxic
 Materials Found at Hazardous Waste Dump Sites. Proceedings,  National Conference on
 Hazardous Wastes and Environmental Emergencies. Control Research Institute.  Silver
 Spring, Maryland. May. pp. 250-257.

Garman, J., Freund, T., and Lawless, E., 1987. Testing for Groundwater Contamination at
 Hazardous Waste Sites. Journal of Chromatographic Science, vol. 25, pp. 328-337.

Lawless, E.,  1987. Preliminary Report on the Analysis of Groundwater Monitoring Data.
 Letter Report to J. Garman, Work Assignment 12, Contract No. 68-01-7310.
                                  11-391

-------
Lawless,  E.,   1990.  Technical  Background  Document.  Ground-Water  Monitoring
 Constituents: Appendix IX-A and Appendix IX. Work Assignment 68, Contract No. 68-01-
 7287.

Massmann,  J.,  and  R.A.  Freeze,  1987. Groundwater  Contamination  from  Waste
 Management  Sites:  The  Interaction  Between  Risk-Based Engineering Design  and
 Regulatory Policy, 1, Methodology, Water Resour. Res., 23(2), pp. 351-367.

Meyer, P.D., and  E.D. Brill, 1988. A  Method for  Locating Wells in a Groundwater
 Monitoring Network Under Conditions of Uncertainty. Water Resour. Res., 24(8), pp. 1277-
 1282.

Neptune, D., Brantly, E.P., Messner, M J., Michael, D.I., 1990. Quantitative Decision Making
 in Superfund: A Data Quality Objectives Case Study. Hazardous Materials Control, v. 3,
 no. 3, pp. 19-27.

Plumb R.H., and A.M. Pitchford, 1985. Volatile Organic Scans: Implications for Ground-
 Water Monitoring. Conference on Petroleum Hydrocarbons and Organic Chemicals in
 Ground-Water-Prevention, Detection, and Restoration.  Houston, Texas. American
 Petroleum Institute and National Water Well Association, pp. 207-222, November 13-15.

Plumb, R.H., 1987. A Comparison  of Ground-Water Monitoring Data from RCRA and
 CERCLA Sites. Ground Water Monitoring Review, v. 8, pp. 94-100.

Plumb, R.H., 1991. The Occurrence of Appendix IX Organic Constituents in Disposal Site
 Ground Water. Ground Water Monitoring Review, v. XI, no. 2, pp. 157-164.

Rocky Mountain Analytical Laboratory,  1984. "Evaluation of the Applicability of SW-846
 Manual to Support all RCRA Subtitle C Testing."  A report to American Petroleum
 Institute on  Hazardous Waste  Management System;  Ground  Water  Testing  and
 Monitoring Activities. Rocky Mountain Analytical Laboratory, Arvada, Colorado, 148 pp.
 December 20.

Rosenfeld, J.  K.,  1990. Ground-Water  Contamination at  Hazardous  Waste Disposal
 Facilities.  Proceedings of the National Water Well Association Conference on Ground
 Water Geochemistry. February, 1990. Kansas City, Missouri. Ground Water Management,
 v. 1,  pp. 237-250.

Test Methods  for  Evaluating Solid Waste, Physical/Chemical Methods, SW-846. U.S.
 Government Printing Office, Washington, D.C. Order No. 955-001 00000-1.

U.S. Environmental Protection Agency,  1979. Water-Related Environmental Fate of 129
 Priority Pollutants. Volume I. EPA-440/4-79-029a.
                                  11-392

-------
U.S. Environmental Protection Agency,  1982. Hazardous  Waste Management System;
 Standards Applicable to Owners and Operators of Hazardous Waste Treatment, Storage,
 and Disposal Facilities; and EPA Administered Permit Programs. Federal Register, vol. 47,
 no. 143, July 26, 1982. pp. 32274-32388.

U.S. Environmental Protection Agency, 1986. RCRA Ground Water Monitoring Technical
 Enforcement Guidance Document. U.S. EPA Office of Waste Programs Enforcement,
 Washington, D.C., 208 pp. and App.

U.S. Environmental Protection Agency, 1987a. Data Quality Objectives for Remedial
 Response Activities (Development Process). EPA/540/G-87/003.

U.S. Environmental Protection Agency, 1987b. Data Quality Objectives for Remedial
 Response Activities (Example Scenario: RI/FS Activities at a Site with Contaminated
 Soils and Ground Water). EPA/540/G-87/004.

U.S. Environmental Protection Agency, 1987c. List (Phase 1) of Hazardous Constituents for
 Ground-Water Monitoring. Federal Register, vol. 52, no. 131, July 9,1987. pp. 25942-25953.

U.S. Environmental Protection Agency, 1988. Hazardous Waste Ground Water Task Force:
 1987 Status Report and 1988/1989 Program Recommendations. 34pp.

U.S. Environmental Protection Agency, 1989.  Handbook of Suggested Practices for the
 Design and Installation of Ground-Water Monitoring Wells. Cooperative Agreement No.
 CR-812350-01. (EPA/600/4-89/034).

U.S. Environmental Protection Agency, 199 la. Permit Quality Review Revised Draft Report
 Office of Solid Waste, Permits and State Programs Division, (in preparation).

U.S. Environmental Protection Agency, 1991b. Subsurface Monitoring Research Activities:
 EPA/600/9-91/003. 45pp.

Waste  Management  of North America, Inc.,  1990.  Leachate Characterization Study.
 Wastewater Group. Project No. 307CO. 7  pp. and Tables, App.

Waste  Management  of North America, Inc.,  1991.  Leachate Characterization Study.
 Wastewater Group. Project No. 307CO. 30 pp. and App.

Weber,  D., Easley, D., and Englund, E., 1991. Probability of Plume Interception Using
 Conditional Simulation of Hydraulic Head  and Inverse Modeling. Mathematical Geology,
 Vol. 23,  No. 2, p. 219-239.

Wilson, C.R., Eichenberger, C.M.,  Kindred,  J.S., Jackson, R.M., and Mercer,  R.B.,
 1991. "Efficiency Based Monitoring System Design." Presentation at the American Society
 of Civil Engineers' Energy in the 90's Conference. March. Pittsburgh, Pennsylvania.
                                   11-393

-------
Q7        DE-MYSTIFYING THE PROBLEM OF FILTERED VS UNFILTERED SAMPLES

          Richard D. Brown. Lead Scientist, Hazardous Waste Systems Department, Energy, Resource and
          Environmental Systems Division, Center For Civil Systems, The MITRE Corporation, McLean,
          Virginia  22102-3481

          ABSTRACT

          A controversy has persisted for at least a decade concerning the filtration versus non-filtration
          of ground water  samples,  particularly with respect to samples used for  metals  analyses.
          Renewed emphasis on the collection of data of high quality and an ever increasing need to
          better understand the subsurface environment have resulted in a resurgence of attention on the
          representativeness  of ground water samples to adequately reflect the level of threat to public
          health and the environment  Often, it is difficult to differentiate the contribution of metals from
          natural sources, incomplete  purging  or disturbance of sediments during sampling, or releases
          from an abandoned or uncontrolled  hazardous waste site.   This  paper examines the various
          facets of the problem, discusses options  for filtration  versus  non-filtration when collecting
          samples for different purposes, and clarifies the relative importance of various fractions of a
          sample (i.e.,  suspended  solids, colloids, dissolved solids, and colloids and dissolved  solids
          adsorbed on  suspended solids)  in understanding  the  subsurface environment  The paper also
          discusses the benefits and drawbacks  associated with the related issues of acidification, transport
          and storage temperatures, use of filters of varying porosity, field versus laboratory filtration, and
          the development of a well to a turbidity standard.  The above issues also  are discussed within
          the context of comparing the sample data to health  and environmental benchmarks, both  for
          ground water and for surface water samples.  Possible solutions to the problem are suggested.

          INTRODUCTION

          A topic of fervent  debate  when discussing ground water  sampling plans often focuses  on
          whether "to filter or not to filter" collected samples.  One viewpoint is that filtration results in
          a substantial physical and chemical  modification of  the  sample.  Another perspective is that
          filtration allows data users to concentrate  only on those  contaminants which are actually
          dissolved, excluding  any substances  which may be adsorbed on, or conveyed by, paniculate
          matter in suspension.  Both positions have merits and the collection of filtered or unfiltered
          samples (or  both)  may be  suitable  dependent  on the questions  which need to be resolved
          (Nielsen  1991).

          The reasons  for filtration of ground  water samples include:

           •     Removal of suspended solids to permit analyses only of the dissolved fraction
                 of substances in the sample,  reflecting drinking water quality as delivered

           •     Removal of any interference  caused by suspended particles (e.g., when ultraviolet
                 spectrophotometric screening techniques are used

           •     Analysis of "clear" samples,  required when using delicate instrumentation easily
                 clogged by sediment-laden samples

           •     Separate analyses of constituents associated with suspended solids

           •     Determination of the percent of suspended solids

          The disadvantages associated with filtration include:
                                                  11-394

-------
 •     Chemical changes in the sample due to changes in partial pressure of dissolved
       gases during filtration under positive pressure or a vacuum

 •     Volatile organic substances may be lost to the atmosphere during filtration

       Aeration of sample during filtration can cause precipitation of metals

 •     Possible inadvertent removal of substances, both organic and inorganic, that tend
       to adsorb on suspended particles

 •     Increased opportunities for sample contamination, especially if filtration is
       conducted in the field

 •     Practical difficulties in the field when filtering during sub-zero temperatures and
       when filtering sediment-laden  water

Generally, the problem of deciding "to filter or not to filter" is associated with the analysis of
metals.  Most ground water samples  collected for the analysis of organic compounds  are not
filtered because:

 •     Many organic hazardous substances are not natural components of ground water,
       therefore, the analyst is interested in the total sample concentration

 •     Most volatile organic compounds can easily be lost during filtration

 •     Since water solubility and partition coefficients vary among most organic
       substances, there is no compelling reason to differentiate between the suspended
       solid and dissolved paniculate fractions of a  sample collected for the routine
       analysis of organic substances.

 •     Except for variations for some pesticides  and PCBs, concentrations of organic
       hazardous substances in ground water do not vary as markedly as metals in proportion
       to the amount of sediment in  a sample.

Thus, the problem of "to filter or not to filter" relates primarily to a perceived need to filter
samples of ground  water to be used for the analysis of metals. The problem manifests itself
in the form of artifacts in ground water monitoring data which cannot easily be explained within
the context  of having  intentionally collected representative samples.  For example, very high
metal concentrations have been observed in samples collected to determine contamination from
a  waste  site when the  metals could  not be  attributable to that specific source.   Sometimes,
background concentrations would be highly elevated, but levels near a source would be at trace
levels.  Very high concentrations (e.g., 640,000  ugA  aluminum,  1,000  ugA nickel,  500 ugA
chromium)  of metals  commonly found  in  soil have been observed in ground  water,  when
normally such concentrations are  low  (e.g., 200 ug/1,  40 ug/l, and 10 ugA, respectively) in clear
ground water.

Generally, there appears to be a direct relationship between high levels of metals and high
levels of suspended solids in the samples, independent of a  sample  being representative of
background or site contamination.  High levels  of  suspended solids are suspected to be the
source of the high concentrations of metals.   The presence of high levels of suspended solids
in ground water samples  complicates efforts to  establish representativeness  of samples and
attribution to sources  of contamination.  In the evaluation of data from such samples, it  is
                                         11-395

-------
difficult to differentiate the contribution of metals from natural sources, the incomplete purging
or disturbance of sediments during sampling, or releases from a site.

BACKGROUND

There are several terms relevant to  a discussion of  contaminant particle fractions in water.
Under current "Standard Methods" (American Public Health Association, et al.  1989), waterbome
solids are divided into two components.  One component, "suspended solids," is retained during
nitration of water through a 0.45 micron (a micron is one millionth of a meter)  filter.  The
second component, "dissolved solids," passes through the filter.  In addition to suspended solids
and dissolved solids, there is a fraction of suspended  solids termed "colloids."

Suspended Solids

Suspended solids are waterbome particles which do not pass through a filter used to  produce
a filtrate containing only dissolved solids. In static water, large suspended solids will  settle to
form sediments.  When sediments are disturbed, such as during the purging of a well, they will
form suspended  solids.

Generally, large-sized suspended solids (e.g., greater than 10 microns in diameter) are not found
in ground water. The exception to this norm is the ground water of Karst areas where surface
debris  and  soil  particles  can enter the system  through sink  holes.   A rapid discharge rate
through caverns and crevices can entrain more large particles through erosion of soft limestone.

Naturally occurring solids, such as clay particles and quartz silicates, move as suspensates in
ground water. At  some locations and at certain times, naturally  occurring metallic hazardous
substance(s) of concern can  be found at relatively high concentrations in ground water.  This
is particularly true for metals found in surface  water  and  ground waters of mineralized areas.
Examples of these areas  include  locations of ultra-basic rocks rich in nickel and chromium,
basaltic and some sedimentary rocks high in zinc and copper, and galena-bearing rocks rich in
lead.  A major fraction of the metals in the ground water of these areas is the suspended solids
present as eroded components of the parent material  (rock and overburden).   Because eroded
particles, in the  form of sediments, can become suspended in wells during sampling,  they are
a major focus of concern.

Colloids

Colloids are  extremely small solid particles which will  not readily settle out of a  solution.
Colloids dispersed in water scatter light even though they are too small to be  seen by the naked
eye.  They are intermediate in size between true dissolved solids and large suspended solids
which are visible to the naked eye.

Colloids vary in size.  They are classified according to size, but there  is  not a  uniform
definition with respect to their lower  or upper limit  The scale used by the U.S.  Department
of Agriculture and the Soil Science Society of  America defines colloids as clay particles with
diameters less than two microns,  but which will not pass  through a 0.45 micron filter used to
extricate dissolved solids from a water sample.  This classification is equivalent to  particles
smaller than fines described  by the U.S. Army Corps of Engineers.  The Wentworth scale used
for sediments, which is a logarithmic scale in that each grade limit is twice as large as the next
smaller grade limit,  defines  clay  particles to be smaller than 3.9 microns (Blatt et al.  1972).
Since  colloids are retained  by a 0.45 micron  filter,  they are  considered to be a fraction of
suspended solids.
                                         11-396

-------
In surface waters, where water movement inhibits settling, colloids have been considered to be
somewhat larger, encompassing  small microorganisms such as bacteria, protozoans, and small
unicellular diatoms.  In  this context,  colloids are considered  to be particles smaller than
10 microns in diameter (Stumm and Morgan 1981).

Where mixtures of pure chemicals are studied under laboratory conditions, colloids may be
viewed as particles  smaller that one micron in diameter.  Such very fine particles may take up
to one year to settle from suspension.

Studies have shown that colloids can facilitate the  transport of contaminants in ground water.
There is  evidence that colloids in excess of 1 micron may not only be mobile in ground water
but also  may move faster than the average ground water flow in porous media as the result of
such effects as size exclusion from smaller spaces  (Puls 1990).

Colloids  have demonstrated strong binding and sorption capacities for inorganic contaminants.
As much as 42 percent plutonium in a  release has been found to be mobilized as colloids
sorbed on suspended solids (Champ et  al.  1982).  High  metal  concentrations, as much as
200 parts per billion of copper, lead, and cadmium, were found to be associated with colloidal
particles  CTillekeratne et al.  1986).  Other  studies have  shown  a strong  affinity  for metal
sorption  onto colloidal particles in  ground water (Gschwend and Reynolds  1987, Enfield and
Bengtsson 1988, Puls and Bonn 1988, Puls 1990).

Dissolved Solids

Dissolved solids are extremely fine particles that pass through a filter with a pore size of 0.45
microns.  Such particles will not settle  from a water sample, but will  remain in a vessel after
evaporation of  a  sample and its  subsequent drying in  an  oven  (American  Public  Health
Association,  et al.  1989).  However, for the purposes of this paper the term dissolved solids
will include the  "volatile solids" which are ignited and some mineral salts which are volatilized
during a dissolved  solids determination.

Strictly speaking, dissolved  solids include only chemical species in solution.   However, the use
of a 0.45 micron filter to remove suspended solids  means that colloidal particles less than 0.45
microns  in diameter are usually characterized as dissolved solids.  This convention was adopted
as a consensus standard representing a compromise  between complete removal of all paniculate
material  and the speed with which filtration may  be  completed.  Thus, some colloidal metal
particles have  been  shown to  pass through  a 0.45 micron  filter, leading to an  order of
magnitude or more  error  in using  0.45 micron  filtration as  an  operational definition  for
"dissolved"  (Puls and Barcelona 1989a).

Dissolved solids represent the aqueous phase  of transport of substances in ground  water.  It
should be kept  in  mind that  there is a dynamic solid-solution  equilibrium in water, wherein
elements move  from  solution to colloids and larger solids and back again depending on
physical, chemical, and microbiological factors. Thus, metals may exist at  one location in an
aquifer in the dissolved state and at another location, or at the same  location at a later point
in time,  as colloidal metal oxides, metal hydroxides, metal carbonates, or chelated metals bound
in  organic matrices.  In fact,  they all  can be present at the same place  and time, all in
equilibrium with one another.

Interaction of Fractions

In ground  water,  the  three  fractions  of particles  (suspended solids, colloidal fraction  of
suspended solids, and dissolved solids) can exist simultaneously.  Also, colloids and dissolved
                                           1-397

-------
substances may be found adsorbed on suspended solids.  In addition, dissolved substances can
be adsorbed on colloidal particles.

The relative states (adsorption, de-sorption, solution) of the metals can change abruptly due to
the actions  of physical,  chemical,  and  microbiological factors.   For  example, an  acidic
environment may lead to the decomposition of metal laden suspended solids (e.g., natural clay
and silicate particles), thereby releasing natural metals into solution.

Fractions of Interest in Ground Water Assessment

Current filtration procedures (using a 0.45 micron filter)  exclude most colloids and suspended
solids  from ground water samples,  leaving  for analysis the  aqueous phase  containing  the
dissolved fraction of the hazardous substances of concern.  Filtration  is useful because  some
suspended solids, such as well sediments  inadvertently collected during sampling, may not be
desired and require removal through filtration.  However, the removal of colloids may not be
desired because of their reported capacity to adsorb and transport contaminants in the subsurface
environment.

With  respect  to  colloids,  a recent  article by  Puls (1990) summarized the  importance  of
delimiting their fraction with regard  to hazardous waste site assessment activities:

 "Inherent  in  these discussions  [concerning  colloids]  is  the  concept  of 'dissolved'  vs.
 'paniculate' and  the  rather arbitrary separation technique of using a  0.45  micron filter,
 commonly used in data collection activities in the laboratory and in the field.  If colloids  as
 large  as  1 to 2  microns are mobile  and capable  of  transporting  contaminants for  large
 distances, then our sampling protocols must make allowances for this component of transport."

Thus,  in summary, the desired  ground  water fractions of primary  interest for evaluating
contamination  of ground water are:

 •     Natural, large-sized suspended solids such as found in Karst environments

 •     Dissolved solids and colloids

 •     Dissolved solids and colloids adsorbed on suspended solids.

Not  desired are  large  suspended sediments  artificially  introduced  into  the  sample during
collection activities.

SAMPLE PROCESSING AND FILTRATION PRACTICES

Ground water samples are collected from active drinking  water wells,  standby wells, and
monitoring wells.   Commonly used sampling devices include electrical submersible pumps,
positive-displacement  bladder pumps, bailers, and  suction-lift pumps.   The type of sampling
device used is based on the rate of well purging possible in view of available well yield, well
diameter, limitations in the lift capability of the device, and the sensitivity of selected chemical
species to the method of sample collection and delivery to a sampling container  (Keith 1988).

Metals samples usually are acidified with nitric acid in the field to pH<2.  The purpose of the
acidification is to inhibit dissolved and colloidal particles from adsorbing onto solids and the
surface of the sample container and  forming precipitates (e.g., hydroxides or hydrated oxides).
                                         11-398

-------
Under the Contract Laboratory Program  (CLP) of the U.S. Environmental Protection Agency
(EPA), metals samples  are acid digested  in  the  laboratory prior to  analysis.   This entails
treatment with acid (and also with hydrogen peroxide if analyzed by furnace atomic absorption)
and heat (95ฐC) to oxidize organic materials.  The sample is then filtered (or  alternatively
centrifuged or allowed to settle by  gravity) to remove insoluble material (EPA  1987a).

For the initial screening of hazardous waste sites, the EPA has recommended that the Total
Recoverable Metals Method,  a method performed on an unfiltered sample, be the standard
technique  in determining metal concentrations in  ground  water.  This technique presumably
releases the loosely bound metals from the paniculate fraction but does not totally destroy the
matrix.  This is viewed as preferable to a dissolved metals analysis on filtered samples, which,
by contrast, does not account for those metals that are adsorbed to the soil matrix and which
may move back and forth in equilibrium with the  ground  water, resulting in an underestimate
of chemical concentrations in ground water from an unfiltered tap (EPA 1989).

However,  the Agency  has recognized the need for filtering when a sample is highly turbid.
For example, if silt persistently appears  in a  sample  because of well construction or design,
and the situation cannot be corrected, then it may be worthwhile to perform both the dissolved
(filtered) and total metals (unfiltered) analyses.  If  filtration occurs (i.e., a dissolved metals test
is to  be performed), the metals  samples are  to be filtered immediately  on-site by  the  field
sampler before adding preservative (EPA 1987a).

Sampling  protocols in general  practice  often recommend that samples  from ground  water
monitoring wells to be used for  metals  analyses be  field-filtered  under pressure before
preservation and analysis.  The filtered  samples  collected for  metals are usually  acidified.
Acidification of unfiltered samples can lead to dissolution of minerals from suspended clays.
The sample should be filtered as soon as possible after it is collected,  preferably in  the field.
Where field filtration is not practical, the sample should be filtered as soon as it is received in
the laboratory  (American Public  Health Association, et al. 1989, EPA 1976).

POTENTIAL SOURCES OF  PROBLEMS

This section contains a  brief discussion  of the predominant mechanisms  wherein undesirable
suspended  solids,  in  the form  of fine  particles, become entrained  in well water.   The
predominant  mechanisms  are   through  inadequate  well  construction,  development,   and
maintenance and well purging and  sampling.

Well  Construction. Development, and Maintenance

The proper construction and development of monitoring wells is essential to the collection of
representative  water samples.  Improperly developed monitoring wells will produce samples
containing suspended sediments that may both bias chemical analyses of collected samples and
cause clogging of field filtering mechanisms (EPA 1987b).

When constructing monitoring wells, the drilling process may cross contaminate aquifers with
loosened fine particles of topsoil, possibly laden with agricultural or industrial chemicals (Keith
1990).  Installation of a screen  with oversized slots, a poorly  designed filter pack, improper
screen placement,  and removal of cement  holding the sand grains together around  the well
screen also contribute to the movement of fine-grained materials into a well.

Monitoring wells must be developed to provide water free of suspended solids.  There are many
ways to develop wells.  The first step in a common method of well development involves the
movement of water at alternatively high and low velocity into and out of the well screen and
                                        11-399

-------
gravel pack to break down the mud cake on the well bore and loosen fine particles  in the
borehole.  This step is followed by pumping to remove these materials from the well and the
immediate area outside of the well screen.  If the flushing procedure  is too harsh, the  gravel
pack may be dislodged leading to possible screen damage or creation of a conduit for small
fines to enter the screen.   Inadequate  pumping will leave  sediments  in  the well.   These
sediments can become entrained as suspended solids in samples.

Improper maintenance can lead to the incrustation of carbonates, metal hydroxides, and biofilms
of iron bacteria which can slough off as  suspended  solids (Driscoll 1989). These incrustations
can markedly affect the chemistry of the well water.

Purging and Sampling

Because biochemical and geochemical reactions and other factors alter the quality of water
stored  in  a well  casing, the stored water  must be  removed  before obtaining a  sample
representative of the quality of water in  the aquifer. The amount of water to be purged from
a well prior to sample collection varies  from  well to well.  If a sample is collected too early
before complete purging, it may not reflect the quality of water in the aquifer. If collected too
late, water or contaminants from areas removed from the well can be drawn into the sample,
possibly resulting in a sample which is not representative of aquifer quality at the well location.
Often, samples are collected after a standard number of well volumes are purged (e.g., 2 to 10)
or when the purged water appears to become "stabilized", determined by the presence of water
that appears to be  unclouded (Brown and Egan 1989, EPA 1983).

Well water that appears to be  clear  may contain paniculate matter in suspension, particularly
if the water is from new or little used wells, such as ground water monitoring wells or standby
municipal supply wells. The amount of sediment discharged from a well is affected by the type
of pump, well construction, size and type of screen, the purging  rate, and other factors.  Often,
fine grained materials near a well intake erode due to water pressure and  well construction.
These pass through a well screen and accumulate as  sediments  in the bottom or on the sides
of the well casing. When a bailer or pump intake is activated for sampling, the sediments can
be disturbed and entrained as suspended sediments in the water sample (Brown and Egan 1989,
Bloese 1983).

Bailers are commonly used for both purging and sampling water from small diameter, shallow
wells  because of their relatively low cost and portability.  However, without very  careful
control, the movement of a bailer often  mixes well water, resulting in a potential for aeration
and degassing of the sample.  The aeration is the result of repeated submergence and removal
of the  bailer during sampling, which  may result in turbulent flow of water in the wellbore.
Further aeration can occur as a result of pouring the collected sample from the top of the bailer
into the sample bottle (Keith 1988).  Such aeration and degassing causes physical and chemical
changes in  water  quality, creating  suspended  solids in the form of hydroxides and other
precipitates.

Bottom-draw bailers, and suction-lift,  gas-displacement,  and other types of pumps have been
used to minimize the problems of aeration and turbidity.  All of these devices have drawbacks
(e.g., slow withdrawal rate, degassing),  compared to the  simplicity of the bailer (Keith 1988,
EPA 1983).

Water samples containing suspended sediments derived from well disturbance do not represent
true ground water quality. The results of analyses  of metals from such samples (if unfiltered)
would be biased high relative to true levels  in ground  water,  due to metal release from the
disturbed sediments.
                                         11-400

-------
Laboratory Storage and Pre-treatment

The accepted time limit for the storage of metals samples is 180 days (EPA 1987a, EPA 1976).
During this period of time, the acidic environment of the sample may cause decomposition of
the suspended  solids, thereby releasing metals into solution.  Also, heat and acid of the
laboratory pre-treatment procedure may release metals through decomposition of the suspended
solids.  These  processes  do not affect determination of a "total  metals" concentration in a
sample.  However, the effects (i.e.,  release of metals  into solution) of these two processes
negate the  ability to obtain representative differentiation of the colloidal and dissolved fractions.

Variable Practices

The difficulty of obtaining a representative ground water sample in light of the suspended solids
problem is complicated by the lack of consistency in sample filtration and  sample acidification.
Delays  in  filtration and preservation  and the sequencing of each  process  result in additional
complications.   Currently, there is no commonly followed  practice for the filtering of ground
water samples (Puls and Barcelona 1989b).

Quality control is not implemented uniformly with respect to the preservation of a ground water
sample with  acid.  Often, it is standard practice to preserve a sample by adding  a  standard
amount of acid (e.g., 5 drops), with the intent of creating a  pH<2 in the sample.  However, due
to variation in the buffering capacity of ground waters in different parts  of the country, the
pH of the a sample may vary  from <2 to >5 following addition of  the acid. The pH is seldom
verified with a pH meter and corrected to <2.

In an examination of field quality control methods in general practice, Keith (1988) found a
number of procedures and areas  of disparity at the time of sampling and sample preservation
that contribute  to  variances in the quality of the collected  water.  These practices include:

 •         aeration and degassing of sample during field filtration

 •         delaying acidification

 •         delaying filtration  or  filtering after  acidification

           lack of necessary  temperature reduction for successful
           stabilization of certain samples (e.g., mercury,  chromium,
           cyanide) during transport.

Delay in the preservation of  metals  samples can lead  to  substantial variation in the reported
concentration.  For example, an experiment has shown that the concentration of iron in  a sample
acidified immediately after collection was  11.6 mg/1; whereas, the concentration of a  duplicate
sample acidified seven hours  after collection was 0.33 mg/1.  Replicate samples from another
site,  acidified  in the same manner showed similar results  (5.74 to less  than 0.08 mg/1).
Significant changes were also observed for other metals (Keith 1988).

A major concern related to the timing of acidification of a metals  sample relates to aeration of
the sample.  When ground water is in a reduced state,  the addition of oxygen can cause metal
precipitation.  Aeration of the sample can occur during transfer from the  sampling  device to a
sampling bottle, transfer to a holding  container prior to filtration, or during filtration. If fixation
of the metals in the  sample by addition of acid occurs after filtration, metal precipitates (e.g.,
metal hydroxides and metals  adsorbed to the hydroxides) could be removed by the initial field
 filtering and not be available for laboratory analysis.  The turbulence and associated aeration
                                          11-401

-------
of a metals sample during filtering affects sample quality much more than simply holding a
sample which has not been acidified,  hi fact, studies indicate that aeration of the sample during
nitration  can have  as much  affect on sample  quality  as  the sampling  activity  itself.
Acidification immediately after sample collection and prior to  filtering minimizes  precipitate
formation after sampling (Brown and Egan  1989, Keith 1988).

POSSIBLE SOLUTIONS  TO THE PROBLEM

Attainment of Turbidity Standard Prior to Sampling

It may be possible to restrict the entrainment of suspended solids into a sample.  A monitoring
well can be developed in such a way that it is  basically free of sediments from construction
activities.   Although not always possible  and only  if it  has been properly  designed  and
developed, a monitoring well can be maintained in such a way that  the screen does not become
clogged and the incrustation of carbonates, metal hydroxides, and biofilms  of iron bacteria are
controlled.   However, such chemical and  physical maintenance  techniques  are difficult to
perform without destroying the representativeness of samples. Excepting severe damage during
well development, a monitoring well  can  be purged and sampled in such a way that its clarity
is equal to that of drinking water (e.g.,  maximum contaminant level turbidity  standard of 5
nephelometric turbidity units).

There are precedents relating to the establishment of clarity hi well water before sampling.  The
goal of several Federal  ground water  sampling  programs  (e.g.,  EPA monitoring program
objectives  under the Resource Conservation and Recovery Act,  the Air Force Installation
Restoration Program, and the Superfund  remedial program) is  to develop, purge, and  sample
monitoring wells in such a way as to assure clarity in the collected samples of water (Puls and
Barcelona  1989b, EPA 1989).

Due to time and resource constraints, there  are several problems inherent in the attainment of
a turbidity standard of clarity before sampling during a site inspection.  These problems include:

  •     Duration of sampling.  Some  monitoring wells are so laden with fine sediments
       that purging rates  need  to be as low as two liters per hour.  Some monitoring
       wells may require up  to seven hours (Keith  1988)  to complete an adequate
       purging and sampling effort, a time and resource requirement which may not be
       achievable under the conditions of an initial ground water  screening.

  •     Verification.  A  frequent nephelometric measurement would be  required to
       confirm attainment of a turbidity standard.  Although relatively easy to perform,
       this would be a burdensome task for site inspection personnel, given limited time
       and resources.  The additional sample handling could increase the probability of
       sample contamination and alteration of the chemical characteristics of the sample.

  •     Well Development  The development of a well for the purposes of producing
       water of potable quality is very time consuming and relatively costly compared
       to  the time and resource  constraints associated  with the installation of a
       monitoring well for screening purposes.

  •    Well Maintenance.  Monitoring  wells  installed for site  inspections may be
       sampled once after installation and never again.  They  may remain unattended
       for many  months or even years  between sampling events.  Without periodic
       screen and gravel pack cleaning,  treatment for incrustation and biofouUng,  and
       other maintenance activities,  clarity  of samples cannot be assured.
                                        11-402

-------
These problems can be overcome in a long-term monitoring program where wells are sampled
periodically (e.g., every quarter).  Drawdown rates for such wells  are known and the proper
well-specific optimal purge volumes determined through records of periodic  sampling.  The
routine, repetitious sampling regimen of such long-term monitoring programs allows for the
provision of adequate time for sampling and well maintenance activities.  The time and resource
constraints of a screening process does not allow  such extensive quality control procedures.

Field Filtration

Field filtration of ground  water  samples to  separate  colloidal and  dissolved solids  from
suspended  sediments is desirable, preferably with vacuum filtration  to expedite the  filtering
process.  The disadvantage of field filtration relates to quality control.  Under field conditions
it would be difficult to avoid  sample  contamination while  coping  with several procedures
inherent in the filtering process.  The filter disks  need to be washed  with successive volumes
of distilled water (American Public Health Association, et al. 1989) and then prewashed with
sample water to equilibrate the  filter disks with sample water (disk will initially sorb certain
metals).   However, this problem  could  be overcome through  use of  pre-washed  disposable
filtration devices.  Problems also  arise  with  control  over  the build up of a "filter cake" and
resultant clogging of filters associated with high  concentrations of suspended solids.   During
sampling, handling, and filtration, aeration could result in  unintentional metal precipitation.

Field filtration has become a routine practice in some monitoring  programs, but an exacting
expectation for sample representativeness and quantitation may preclude  field filtering due to
the above mentioned quality control problems.  The additional resource burden associated with
the filtration of ground water samples in the field may be excessive, given the limited resources
available for site inspections.

Laboratory Filtration

Filtration in  a fixed laboratory, such as  a laboratory under the CLP, is an attractive alternative
compared with field filtration. Conditions are conducive for controlled  analytical measurement
and sample handling.

The disadvantage to filtration in the laboratory relates to the time lag  from sample collection
to analysis, the greater this time lag, the more the entrained sediments become dissolved by the
acidic preservative. Filtration in the laboratory would involve immediate analysis vs. the current
practice of metals sample storage for a prolonged  period of time, bringing about a new concept
in metals analysis.   However,  the analysis of metals samples  upon receipt by the laboratory
is logistically feasible, because the CLP requires  a rapid turn around for the analysis of other
types of hazardous substances.

Preservation and Storage

The filtration of colloids and dissolved solids from large suspended solids requires special care
in sample preservation and  storage  in order to minimize degradation  of the large suspended
solids by  acid.  If filtration is conducted in the field,  dissolution of the suspended  solids
fraction is minimal.  However, if laboratory filtration is performed, special care must be taken
to minimize chemical reactions after acidification (which "fixes"  dissolved solids already in
solution).

One apparently ideal method of minimizing the chemical reactions which can breakdown the
suspended solids in acidified metals samples is to lower the temperature of the samples.  This
                                         11-403

-------
is a standard technique  for the preservation and storage of many types of samples.   Water
reaches its maximum density at 4ฐC.  For many decades, it is at this temperature that samples
of waterbome coliform bacteria, pathogens, and organisms were stored for immediate transfer
to a laboratory for culturing, plating, and  analysis.  Field  samples easily are maintained at a
temperature of 4ฐC by means of an ice  slurry (or wet ice)  in  an ice chest.  Maintaining
chemical samples in an ice  slurry is common  practice for certain metals  affected by biotic
activity (e.g., mercury, chromium, colorimetric  analysis of copper, cyanides) and for volatile
substances.  When metals samples at ambient temperature  are  placed in an ice slurry (or wet
ice), the samples attain a temperature of 4ฐC within three hours (Keith 1988).   This cooling
process could aid in stabilizing the various metal fractions until receipt at the laboratory for
analysis.

There are a few disadvantages associated with the cooling of metals samples. One disadvantage
is that a decrease  in temperature of a  sample will increase its oxygen saturation level,
contributing to aeration of the sample and possible hydroxide formatioa  However, acidification
of the sample should mitigate problems  associated with such aeration.  The cooling of a ground
water sample from a 20ฐC temperature  of  a warm,  shallow aquifer to 4ฐC can raise its pH by
as much as one-half of a pH unit (Diehl 1970).  However, if the sample is acidified properly
to a pH<2, any change in the pH due  to cooling should have insignificant effects on precipitate
formation.

Freezing  metal  samples is another alternative.   Freezing samples  will minimize chemical
reactions and inhibit breakdown of  suspended solids, but presents  several problems.   The
freezing action (unless flash frozen; e.g.,  with liquid nitrogen) can create a phase  separation
wherein water free of acid becomes frozen first leaving the remaining liquid more acidified,
possibly  creating problems  in metals  recovery in  the  laboratory  (e.g.,  during the CLP
pie-treatment  analysis. The field logistical requirements for freezing involve special transport
and handling  of the freezing agent (e.g., dry ice), extra cost of materials and equipment, and
special training of field personnel.   The receipt,  storage, handling,  and thawing of  frozen
samples in the laboratory may present added logistical and analytical problems.

SUGGESTED FILTRATION PRACTICES

It is recommended that ground water metals samples be acidified immediately upon collection
in the field and cooled to a temperature of 4ฐC for transport to a fixed laboratory for analysis.
The acidification to pH<2 should be verified in the field prior to cooling the samples.

The metals samples should be filtered for the separation and analysis of colloidal and dissolved
solids immediately upon receipt at the fixed laboratory. After filtration, the filtrate should be
acid  and heat pre-treated  using  the  current CLP  procedure  for the pre-treatment of  metals
samples.

The filter pore size used for filtration should be large enough to allow the bulk of the colloidal
particles to be recovered, but small enough as to exclude larger suspended sediments.   A
commercially available,  acid resistent, 5 micron pore size filter is available in standard sizes
(e.g.,  2.2 cm to  4.7 cm) and is recommended.   A  larger  pore size (10 micron) filter is
available, but is not recommended for the  size range of colloids associated with ground water.
It may be  possible  that a more preferable 2  micron  pore size,  acid  resistent, filter is
commercially available,  but its availability needs to be confirmed.

Field filtration is not recommended for ground water  metals samples.  Should a decision be
made  to  filter  ground  water metals  samples in the  field,  the  following  procedure  is
recommended.  Immediately upon collection,  the  samples should be  subjected to mild acid
                                         11-404

-------
treatment  (e.g., nitric acid  pH 3 to 4) for 10 minutes to free sorted dissolved and colloidal
contaminants from large  suspended solids.   Then, filtration should be  preformed using a
5 micron (or 2 micron if available) pore size filter.  The filtrate should be acidified immediately
upon collection  in the field and  cooled to a temperature of 4ฐC for transport  to  a fixed
laboratory for analysis.  The acidification of the filtrate to pH<2 should be verified  in the field
prior to cooling  the samples.

Coupled with the above, relevant site inspection guidance should be developed, focusing on the
use of various techniques to minimize the entrainment of suspended sediments in ground water
metals samples.

The problem of sediments entrained in ground water samples is associated with an overestimate
of the concentration  of metals in ground water.   Samples collected for  organic  compounds
analyses should  not be  filtered.  This is consistent with common practice to not filter samples
collected for the analysis of organic compounds (Keith 1991).

Ground water samples in Karst areas should not be filtered. The presence of suspended solids
larger  than colloids  is an intrinsic  feature  of these  systems  and is indicative  of  natural
background levels.

Both filtered and unfiltered surface water samples (split samples) should be used for metals
analyses.  Data from unfiltered samples should be used for comparison with benchmarks such
as Ambient Water Quality Criteria (AWQC)  which represent unfiltered concentrations.  Data
from filtered samples should be  used for comparison  with  benchmarks such as Maximum
Contaminant Levels (MCLs) which represent water delivered to a user of a public water supply.
Large  suspended  solids have  been removed from such  delivered  water by various means
including sand filters, flocculation, and gravity settling in storage facilities.  Even in private,
rural water supplies, paniculate matter is  removed by settling in household compression tanks,
gravity and pressure filters, zeolite softeners,  and other ion-exchange units for the removal of
unwanted hardness.

SUGGESTED CONFIRMING STUDIES

A  number of  studies need  to  be  conducted  to confirm  that the  recommended  sample
preservation and filtration  procedures are appropriate.   The studies should be conducted by a
laboratory familiar with  the  filtration of colloids  and dissolved solids  from ground water
samples containing high concentrations  of suspended solids.  The  following are  some of a
number of questions  which should be addressed by such studies.

        What is  the most appropriate type of filter and  filter pore  size in terms of
        availability and applicability, given acidified conditions and the need to extricate
        colloidal and dissolved metal particles from ground water samples?

        What portion of the total metals concentration of a sample  is associated  with
        suspended solids greater than 5 microns and greater than 2 microns in diameter?

        Do colloids represent a  significant amount of  the metals concentration  of a
        sample (excluding  dissolved solids)?

        Does the cooling of acidified metals samples significantly reduce the breakdown
        of suspended solids?
                                          11-405

-------
       Arc the quality control problems associated with field filtration too  great to
       warrant field filtration in lieu of laboratory filtration?

 •     Do the metal concentrations degrade significantly during transport between the
       field and the laboratory and during temporary storage prior to filtration?

 •     If samples are not analyzed immediately in a fixed laboratory, will the suspended
       solids be significantly degraded, given acidification and cold (4ฐC) storage?

RAMIFICATIONS OF SUGGESTIONS

The use of filtration to separate the colloidal and dissolved metal fractions for analysis, thereby
removing large suspended solids from the sample, represents a "forced" control over a sampling
problem which may not be controllable in the field. The resultant data, derived from analysis
of the filtrate, would be more representative of conditions  representing background and site
contamination.

Requiring  filtration would eliminate the occurrence (though infrequent) when samples were
collected to represent background and site contamination, where one of the samples was filtered
and the other sample was not filtered.

Data from filtered ground water samples are more appropriate  for comparison  against drinking
water benchmarks.  This comparability  of sample comparisons applies to both surface  water
as well as ground water.

The use of filtration recognizes the fact that under the screening conditions of a site inspection,
the problem of the entrainment of suspended sediments in ground water metals samples is not
easily solved by  quality control procedures.  The time and resource constraints of a site
inspection also may preclude field filtration in lieu of filtration in a fixed laboratory.

Filtration  will increase the number of and types of metals samples to be collected and tracked.
For example, hi surface water, filtered and unfiltered metals samples will need to be collected
at each sampling  point through the use of split  samples.  In Karst aquifers, metals  samples
would not be filtered.  For surface water, the data user must be assured that  filtered  samples
are compared with filtered samples and vice versa. All reported water data  will need  to be
flagged with respect to whether the samples were filtered or unfiltered.

The requirements  for site inspection personnel would be increased through the implementation
of a filtration policy.  Field acidification would require verification.  Samples will need  to be
cooled and maintained at a temperature of 4ฐC and rapid transport  to the laboratory  assured.
Improved quality assurance and quality control requirements relating to purging and sampling
may be required.

Contracts with fixed laboratories  may need to specify a new pre-treatment  protocol in the
statement of work for inorganic analyses.  The new laboratory procedure would shorten the
holding time for metals samples from  180  days to less than 48 hours, resulting in a marked
change in routine laboratory  procedures.

The removal of colloidal particles represents a new filtration practice involving more extensive
quality control procedures.  Although the precedent of filtration is firmly  established, the
separation of colloidal particles would represent  a new way of thinking in contrast with the
traditional viewpoint of "dissolved" versus "suspended" solids.
                                       11-406

-------
SUMMARY

A new approach to ground water sampling and analysis of metals is proposed.  The approach
involves the separation of colloids and dissolved solids from ground water samples by means
of filtration within a fixed laboratory.  This approach will not eliminate all  uncertainties, but
represents a "forced" control over a sampling problem which may  not  be controllable in the
field during initial screening investigations of an uncontrolled hazardous waste site.

REFERENCES

American Public Health Association,  American Water Works Association, and  the  Water
Pollution Control Federation.  1989.  Standard Methods For the Examination  of  Water and
Wastewater, 17th Edition.  Washington, DC

Blatt, Harvey, Gerard Middleton, and Raymond Murray.  1972.  Origin of Sedimentary Rocks.
Prentice-Hall, Englewood Cliffs, New Jersey

Bloese, Rod.  1983.  Field Filtering Ground Water Samples. Internal EPA memorandum to Joe
Petrilli, dated Marck 22, 1983

Brown, Richard D. and David E. Egan.  1989.  Site Assessment Media-Specific Considerations:
Lessons  Learned from  a Data User (Seminar  presentation).   Detailed outline published  in
Proceedings of The 6th National Conference On Hazardous Wastes and Hazardous Materials,
The Hazardous Materials Control Research Institute, Greenbelt, Maryland

Champ, D.R., W.F.,  Merritt, and J.L. Young.  1982. Potential for Rapid Transport of Pu in
Ground Water as Demonstrated by Core Column Studies.  In, Scientific Basis for Radioactive
Waste Management,  Vol.5.   Elsevier Science Publishers, New York

Diehl, Harvey.   1970.  Quantitative Analysis: Elementary Principles and Practice.  Oakland
Street Science Press, Ames, Iowa

Driscoll, Fletcher G.  1989. Groundwater and  Wells (2nd. Ed.).   Johnson  Filtration Systems
Inc., St.  Paul, Minnesota

Enfield,  C.G.  and  G. Bengtsson.    1988.    Macromolecular Transport  of Hydrophobic
Contaminants in Aqueous Environments.   Ground Water 26(1): 64-70.

Gschwend, P.M. and M.D. Reynolds.  1987.  Mono-disperse Ferrous Phosphate  Colloids in An
Anoxic Ground  Water Plume.  J. of Contaminant Hydrol. 1: 309-327.

Keith, Lawrence H.  1991.   Environmental Sampling and Analysis: A Practical Guide.   Lewis
Publishers, Chelsea,  Michigan

Keith, Lawrence H.  1990.  Environmental Sampling: A Summary.  Environmental Science and
Technology 25(5):610-617

Keith, Lawrence H.  1988.  Principles of Environmental Sampling.  American Chemical Society,
Washington, DC

Nielsen, David.   1991. Practical Handbook of Ground-Water Monitoring.  Lewis Publishers,
 Chelsea, Michigan
                                         11-407

-------
Puls, R.W. 1990. Colloidal Considerations in Groundwater Sampling and Contaminant Transport
Predictions. Nuclear Safety 31(1): 58-65

Puls, Robert W. and Michael J. Barcelona.  1989a.  Filtration of Ground Water Samples for
Metals Analysis.  Hazardous Waste & Hazardous Materials 6(4): 385-393

Puls, Robert W. and Michael J. Barcelona.  1989b.  Superfund Ground Water Issue: Ground
Water Sampling for Metals Analyses.  EPA/540/4-89/001.  Center for Environmental Research
Information, U.S. Environmental Protection Agency, Cincinnati, Ohio

Puls, Robert W. and Hinrich L.  Bonn.  1988.  Sorption of Cadmium, Nickel, and  Zinc by
Kaolinite and Montmorillonite Suspensions.  Soil Sci. Amer. J. 52(5): 1289-1292

Stumm,  W. and  J. Morgan.   Aquatic Chemistry: An Introduction Emphasizing Chemical
Equilibria in Natural Waters (2nd ed.).  John Wiley and Sons, Inc., New York

Tillekeratne, S., T. Miwa, and A. Mizuike.  1986. Determination of Traces of Heavy Metals
in Positively Charged Inorganic Colloids in Freshwater.  Mikrochimica Acta B:  289-296.

U.S. Environmental Protection Agency.   1989.   Risk Assessment Guidance For Superfund:
Human Health  Evaluation Manual, Part A.  EPA/540/1-89/002.  Office of Solid Waste  and
Emergency Response, Washington, DC

U.S. Environmental Protection Agency.   1987a.  Statement  of Work: Inorganic Analyses.
Contract Laboratory Program, Washington, DC

U.S. Environmental Protection Agency. 1987b.  Handbook: Ground Water. EPA/625/6-87/016.
Robert S. Kerr Environmental Research Laboratory, Ada, Oklahoma

U.S. Environmental Protection Agency.  1983.  Characterization of Hazardous Waste Sites - A
Methods Manual: Volume II.  Available Sampling Methods. EPA-600/4-83-040. Environmental
Monitoring Systems Laboratory, Las Vegas, Nevada.

U.S. Environmental Protection Agency.  1976.  Manual of Methods for Chemical Analysis of
Water and Wastes.  EPA-625-/6-74-003a.  Environmental Monitoring and Support Laboratory,
Cincinnati, Ohio
                                       1-408

-------
QO                     DETERMINATION OF TARGET ORGANICS
90                    IN AIR USING LONG PATH SPECTROSCOPY

                            Richard D.  Spear Ph.D.  0)
                 Pamela D. Greenlaw (2), Raymond J.  Bath Ph.D. (2)


       The  EPA  Region   II,   Environmental  Services  Division   (ESD),
       Surveillance  and  Monitoring Branch (SMB)  has recently acquired  a
       transportable system to perform long path  remote sensing of air
       contaminants.  This remote sensing system consists of spectrometers
       which identify and guantify target organic chemicals in ambient air
       Pathlengths,   up   to  500  meters,   are  defined  by  use  of   a
       retroreflector,  a  specially constructed  mirror  assembly which
       reflects and collimates the signals generated by the spectrometers.
       The spectrometers used  are:  a  Fourier transform  infrared  (FTIR);
       with a resolution of 0.5 cm-(1) and a liquid nitrogen cooled mercury-
       cadmium-telluride  (MCT)  photodetector and a long  path ultraviolet
       (LPUV) with a prism monochromator and a photo diode array detector.
       With meteorological monitoring, this system can be used to  monitor
       the air for many  environmental  applications:   site investigations
       for  Hazardous  Ranking  System  (HRS);   fenceline  monitoring  of
       industrial  sites; off-site  health  and  safety monitoring  during
       remediation   or   removal  projects;  monitoring   of  lagoons  for
       potential  air release;  and in emergency  response to  community
       complaints  on air quality.  This  paper will present the  design,
       application  and  interpretation of  data  for the  EPA Region II,
       ESD/SMB, LPUV/FTIR.
       (1) U.S. Environmental Protection Agency, Region II, Edison, NJ 08837

       C2) NUS Corporation,  1090 King Georges Post Road, Edison, NJ  08837
                                    11-409

-------
99       MEASUREMENT OF TOXIC ORGANIC COMPOUNDS IN LANDFILL GAS
            SAMPLES USING CRYOGENIC TRAPPING AND FULL SCAN GC/MS
                                     Steven D. Hoyt
                             Environmental Analytical Service
                                   170-C Granada Drive
                                San Luis Obispo, CA 93401
                                     (805)541-3666
              A Nutech automated cryogenic concentrator with adjustable sample volume
        loops is used for analyzing landfill gas samples using full scan GC/MS and selected
        ion monitoring (SIM).  This method is able  to quantitate VOC compounds over
        concentration ranges of 0.5 ppbv to 1000 ppbv. Landfill samples can be effectively
        collected in evacuated  SUMMA passivated canisters and most VOC compounds
        have a holding time of 14 days. A 0.5 to 500 ml landfill gas sample is loaded into the
        Nutech Automatic Concentrator and then analyzed with an HP 5890 GC using a 30
        meter DB-5 fused silica capillary column connected directly to the source of an HP
        5790 MSD.  The capillary column is temperature programmed from -40 to 150 C to
        analyze compounds from F-12 to trichlorobenzene. The relative standard deviation
        for the method is less than 10% for most compounds and the MDL is about 0.5 ppbv
        depending on  sample size  and the  carbon dioxide content of the sample. The
        sampling methods, instrument modifications  for analyzing landfill gases  will be
        discussed along with the examples of data, and the limitations of the method.
                                         11-410

-------
100            THE DETERMINATION OF THE HEAT OF COMBUSTION AND
                    WATER CONTENT OF INCINERATOR FEEDS USING
                             NEAR INFRARED SPECTROSCOPY

         Dr. Nilesh K. Shah. Senior Analytical Chemist, Methods Development, Dr. Peter A.
         Pospisil, Manager, Methods Development, Rita A. Atwood, Analytical Chemist,
         Chemical Waste Management  Inc., Technical Center, 150 West  137th  Street,
         Riverdale, Illinois 60627;

         Dr.  David L. Wetzel, Research Analytical Chemist, Arnold J. Eilert, Associate
         Analytical Chemist, Kansas State University, Schellenberger Hall, Manhattan,
         Kansas 66502

         ABSTRACT

         Near Infrared Spectroscopy (NIR) allows the simultaneous determination of the
         heat of combustion and moisture content of a broad range of heterogeneous
         incinerator feeds, with no sample preparation.

         RCRA regulations require the determination of the  heat of  combustion on all
         incinerator feeds to determine  if they are  above the 5000 BTU/lb level.  Water
         content is necessary for  proper operating conditions of the incinerator.  To satisfy
         these requirements, a large number of samples are currently analyzed using both
         bomb calorimetry and Karl Fischer titration, which are labor  intensive and time
         consuming methods.  The  NIR  procedure  utilizing  selected absorption  bands
         eliminates  all  sample  preparation,  while  simultaneously  determining  both
         parameters.

         NIR technology was used to generate heat of combustion and moisture data on 73%
         of 564 incinerator feeds  at a 90% success level, subsequent to software screening to
         classify the incinerator feeds into physico-chemically unique types.  The 73% can be
         increased to 95%  and  the success level  increased, by consolidating  feed type
         calibration  curves  and  by improving the prescreening  software.   Additional
         parameters may be added as the database is expanded.  The runtime of two minutes
         per sample entails an 80% analytical cost savings.

         INTRODUCTION

         RCRA regulations require the determination of the  heat of combustion on all
         incinerator feeds to determine if they are above the 5000 BTU/lb level.  Chemical
         Waste  Management Inc. incineration facilities receive a broad range of liquid
         hydrocarbon-based wastes  requiring incineration.     Incinerator   feed  type
         compositions cover very wide  ranges of constituents with  heats of combustion
         ranging from  1,000 to 20,000  BTU/lb, water contents  from 0.1% to 100%  and
         halogen contents from 0.1% to 70%.  Heat of combustion and water content are
         critical sample composition parameters that affect incinerator performance  and
         blend feeds before incineration.   Because these  analytical parameters critically
         affect incinerator performance and efficiency, each feed requires chemical analysis
         using conventional bomb calorimetry and Karl Fischer titration methods, which are
         both labor intensive and time consuming.
                                            1-411

-------
                                                                      Page 2


Near infrared reflectance (NIR) spectroscopy is a rapid and sensitive measurement
technique that  has found many  applications  in analyzing  agricultural  and
pharmaceutical materials [1,2].  The near infrared spectral region, which spans from
1100 to 2500 nm, has high information content in the form of many overlapped
bands  arising  from  overtones  and combinations of X-H  stretching modes  of
vibration.  The widespread use of NIR spectroscopy can be  greatly attributed to the
introduction of powerful computerized  data processing techniques for interpretation
of complicated NIR spectra. The development of this technique for quantitation is
primarily due to the availability and use of multilinear regression analysis; however,
quantitation is limited  to  samples  of controlled  composition.   The qualitative
information available in the NIR spectral  region  is used by pattern recognition
techniques for the identification and classification of samples of unknown origin.

This paper reports a classic example of using near infrared spectroscopy and
chemometrics methods for analyzing hazardous wastes. Because of the nature and
spectroscopic complexity of hazardous wastes,  a two-step chemometrics  approach
must be used  to successfully extract  useful information from  the  near infrared
spectra. The first step is to extract qualitative spectroscopic features  from the near
infrared spectra for pattern recognition analysis.  The second step,  then,  is
quantitation of heat or combustion and water content for multivariate calibrations.
Mahalanqbis  distance   pattern recognition  analysis is  used  to  develop  the
classification models from near infrared spectra.  Multivariate calibration models
are developed by multilinear regression analysis for each of the defined classes.  A
reasonable degree of accuracy is obtained in predicting the  heat of combustion and
water content of liquid incinerator feeds provided appropriate calibration is used.

THEORY

Symbols and Notations

The following discussion explains the symbols and notations used in this paper  to
describe the theory of Mahalanobis  distance pattern  recognition analysis  and
Multilinear regression analysis.  Bold letters  are used to denote matrices and lower-
case letters to denote scalars (italic) and vectors (bold). A vector is always a column
vector if no transpose is  attached.  Transposed vectors are denoted by single quote
('). The symbol x is for a NIR spectrum and c is the concentration of the chemical
constituent of interest.   In addition, i is  the number of  training set samples
(observations)  and k is  the number of spectral values (wavelengths).  With this
notation, the model consists of i observations of k dimensions and the two sets  of
data are denoted by c and X.

The training set is defined as the samples that are used to develop the classification
and calibration models.   The test set  is defined as the  samples that are used  to
evaluate the classification and calibration models and are samples that are not used
in the training set.
                                   11-412

-------
                                                                      Page3


M ahalanobis distance pattern recognition analysis

The Mahalanobis distance technique assumes a multivariate normal  distribution
N(p, E) for the class population.  The class model consists of a single point in
multidimensional space, the class centroid /*.  The distance between an zth sample,
x,-, to the centroid is given by the generalized squared distance:


                         MDJ  = (x../0'E-1 (x.-,,) ................................................. 1


where E  is  the training set's  variance-covariance matrix which explains the
dispersion of data around the centroid.  In practice, the true  centroid /* and the
variance-covariance matrix E of the class population are unknown and, therefore,
must be estimated by the mean vector ~x and the variance-covariance matrix S from
a sample training set of n.  The sample Mahalanobis distance can then be calculated
from equation 2:


                         MD?  = (x; -It)' S"1 (x. -Ic) ................................................ 2

and
                         s =
Geometrically, the Mahalanobis distance class model is an ellipsoid-shaped cluster
with the population mean at its centroid.  A spectrum is classified as a member of a
group if the Mahalanobis distance is less than 6 as compared to the Mahalanobis
distance for that sample with  other groups. An excellent review of the theory  of
Mahalanobis distances is given by Mark and Tunnell [3].

Multilinear Regression Analysis

Regression  analysis is  used  for predicting  BTU  values from a collection  of
independent variables such as  wavelengths. The procedure consists of two phases:
calibration and  prediction  [4].   A data  matrix is  constructed from the NIR
instrument response X (absorbance)  at  selected wavelengths for  a given  set  of
calibration samples.  A vector  of heat of combustion values c is then formed using
an independent method such as bomb calorimetry method.

One of the objectives of the  calibration phase is to develop a model that relates the
NIR spectra to the heat of  combustion values  obtained by the bomb calorimetry
method.   In regression analysis,  a linear  combination of the variables in X  is
calculated such that the model's estimates of the heat of combustion values of the c
in the calibration set are as close to the known values of c as possible (minimizes the
errors in reproducing c).  Mathematically, the linear regression model with a single
response (BTU value) can be explained by equation 4:


-------
                                                                    Page 4
where ft is a vector of regression coefficients and c is a vector of errors or residuals
associated with the regression model. To "fit" the model given in equation 4 to the
known values of c, we must determine the values for the regression coefficients /9
and the residual e consistent with the available data.

The  method of least squares selects regression coefficients  estimates, p, using
equation 5:

                       p = (X'X)'1 X'c	5

and the estimated response c using equation 6:

                             a  =
The  regression estimates p  are consistent with  data  whose  sum  of  squared
differences (ซ ) from the observed c is as small as possible.
                             n
where deviations hase separation. A QA/QC program has also
been  developed  for the Bran+Luebbe NIR spectrometer during the method
development process.

A Bomem MB155  FTIR/NIR connected  to a  Compaq  386  20 Mhz personal
computer was used to acquire NIR spectra.  The absorbance data were collected in
the NIR spectral range from  10,000 to 4,000 cms'1 (or 1100 to 2500 nm). Sixteen
scans at 8 cms'l resolution were averaged for Fourier data processing. Using the
complete NIR spectrum  range provided visual information for identifying spectral
patterns responsible for C-H and O-H overtone bands.

The data for pattern recognition and the calibration models were  acquired using a
Bran+Luebbe NIR 400 filter instrument and a Compaq 386 20 Mhz computer.  The
                                   11-414

-------
                                                                     PageS


Bran+Luebbe 400 instrument consists of 19 filters mounted on a filter wheel which
were configured with different wavelengths. The NIR spectral region selected for
the present work  ranged from 1600 to 2450 nm since most  of the useful spectral
information was present in this region. The samples were analyzed using a stainless
steel cell  covered with  a quartz plate  to  obtain a thin layer of film which was
measured in reflectance mode by the filter instrument.

Customized software written in Microsoft Quick Basic program was used for the
Technicon 400 instrument data acquisition.   The data were  then imported to
Bran+Luebbe  IDAS  software  for developing  Mahalanobis  distance  pattern
recognition and multilinear regression calibration models. Two separate equation
files, written in ASCII format, were read by a custom software to predict samples of
unknown origin. The custom software first classified the "spectroscopic type" of the
sample based on  the Mahalanobis distance pattern recognition analysis and then
used the  appropriate  calibration  to obtain  a quantitative results for heat of
combustion and water content.

RESULTS AND DISCUSSION

Conventional Methods Overview

The standard technique for determining the heat of combustion of liquid incinerator
feeds is the bomb calorimetric method.  The heat of combustion, measured in
British Thermal Units per pound (BTU/lb), is determined by burning a previously
weighed sample in an oxygen calorimeter under controlled conditions. The energy
required to raise  the temperature of a given volume of water is  measured by
observing the temperature before firing the bomb and after a stable temperature is
reached.  These observations  are made  and recorded by the calorimetry apparatus
which also reports the heat of combustion (BTU/lb).

The standard technique for determining the water  content of liquid incinerator
feeds is the Karl Fischer titration method. The percent water content is determined
by titrating a known amount  of sample with standardized  Karl Fischer Reagent
(KFR) to its endpoint.  When there is an excess of KFR, the  solution color changes
to a dark brown due to presence of free iodine.  The  Karl  Fischer reagent  is
standardized by titrating KFR with a known amount of water. Using an automatic
titrator, the endpoint of the reaction can also be electrometrically determined.

NIR Spectroscopy

Near infrared spectroscopy   is based  upon molecular heteroatom vibrations
producing a  charge distribution, which interacts  with  electromagnetic  radiation.
The interaction intensity is directly proportional to the dipole moment of the
molecular bond, and produces the characteristic absorption patterns representative
of the chemical composition of the sample. The mid-infrared region (25.0 to 2.5 pm
or 400 to 4000 cms"1) is the most well known range of analysis of organic materials.
The sharp spectral bands produced by the fundamental vibrational frequency of the
heteroatom bonds are  directly related to skeletal  and functional  structures of
organic compounds.  Near infrared absorption bands are produced by vibrational
overtones, and for each  mid-infrared band there are four to seven near infrared


                                    11-415

-------
                                                                      Page 6
overtones.  This crowding produces broad plateaus arising from superimposed
harmonics.  The loss of structural information is offset by a gain in signal to noise
ratio. This makes NIR spectroscopy especially suited for the analysis of hazardous
waste liquid incinerator feeds.

The near infrared region (1100 to 2500 nm) is attractive for heat of combustion and
water analysis because most of the absorption bands observed in this region arise
from overtones and combinations of C-H and  O-H  stretching vibrations.  Near
infrared spectra of three types of  liquid incinerator feeds are shown in Figure 1.
The spectra show prominent bands for each type of incinerator feeds.  For example,
type 6 feeds have a broad band at 1940 nm that is characteristics of O-H stretching
and second overtone vibration. Absorption bands are particularly strong above 2300
nm due to presence of two or more types of hydrogen bonded molecular complexes.
Type 1 feeds are primarily fuel oil (hydrocarbon) types of hazardous waste and,
therefore, the NIR spectra of such type materials contain  a broad C-H overtone
band around 1720 nm. PCB type of materials are responsible for peaks at 1650 and
2175 nm in type 2 feeds. The absorptivity of these bands is largely independent of
the remainder of the molecule, but does  depend on  the concentration  of the
absorbing functional group and, therefore, can be used  for predicting the heat of
combustion and water content of the liquid incinerator feeds.
               1.5 -
                1 -
                         Near Infrared Spectrum
                        of Incinerator Feed Types
                16OO
                           iaoo
                                     8OOO

                                 Wavelength (nm)
                                                          24 OO
              Figure 1: Near Infrared Spectra of Incinerator Feeds

Incinerator feeds have been identified into seven types based upon their  NIR
spectral patterns.  Table I is a summary of matrix types responsible for the seven
groups of incinerator feed types.  The distribution of the seven groups of incinerator
feeds analyzed by the NIR spectroscopy is shown in Figure 2.
                                   11-416

-------
                                                                       Page?
         Table I: Summary of matrix types for pattern recognition analysis
Feed Type
1
2
3
4
5
6
7
Matrix Type
Fuel Oils
Haloeenated solvents (e.\ PCB)
Unknown
Aromatic solvents
Unknown
Polar solvents (e.g , methanoH ??
Aqueous solution (e.g.. water)
                                           50.89%
                14.54%
7.98%
   2.13%
                                           14.18%
Di

D2



• 4

• 5



 17
     Figure 2: Distribution of seven feed types analyzed by NIR spectroscopy

Near infrared analysis depends on the development of an empirical linear equation,
in which the constituent concentration is related to some combination of optical
measurements, usually expressed  in  absorbance  or reflectance.   To  use this
empirical approach, the analyst must have  a set of samples having known values
generated by another method (training set samples).  From this set of knowns, the
system is trained through an iterative process. Using regressive and correlative data
processing, the analyst generates a multiterm linear expression making suitable use
of the analytical data.  With sufficient experimentation and statistical treatment  of
the data, this produces a final working calibration curve.

Mahalanobis Distance Pattern Recognition Analysis

In the Mahalanobis distance classification technique, two or more wavelengths are
used for classification  of samples.  The classification of spectra was based on the
generalized square distance of an observation  from the centroid of a cluster.  In
addition, only  one mathematical model was constructed for all incinerator  feed
types.   In  our present work, four wavelengths  gave  adequate discrimination  to
identify seven groups of incinerator feeds based on their NIR spectral patterns and
Mahalanobis distance pattern recognition analysis. In Figure 3, a three dimensional
                                     11-417

-------
                                                                     PageS
plot  of wavelength 2100, wavelength  2139 and  wavelength 2348  shows good
discrimination  between various  classes.  Adding a fourth wavelength 2310, the
pattern recognition model was able to classify various feed types more accurately.
The legends 1... 7 used in Figure 3 are explained in Table I.
                                     3D Scatter Plot
                                   of Incinerator Feeds
           WL 21001.2
                 OS
      /•••'T~M^
               4 *y
             ซ 4 ?ป*
"^ *   'Xli/Jj
                                           '^S4i
                                           4  1
4ฅ*-—
 ki       ••••ซ....
                                                  1.2
                                                   WL2139
                                           0.4
  Figure 3:  3-Dimensional scatter plot of wavelength 2100 vs wavelength 2348 and
            wavelength 2139. See Table I for explanation of legends

The mathematical model for the Mahalanobis distance pattern recognition consists
of two  matrices: the  group-mean  matrix  and  the  mversed  pooled  variance-
covariance matrix.  Using the model, the Mahalanobis  distances between groups
were calculated for the training set data from which the  model was developed.  In
addition, greater the  Mahalanobis distance  between  groups,  the greater the
difference in their patterns.  The results for Mahalanobis distances between groups
are summarized in Table  n.  According to Table II, only group 1 and group 4 are
close to each other, suggesting a similarity in spectral patterns between them.

     Table II: Mahalanobis distances between groups used in the training set

                                       to
  from

Group 1
Group 2
Grouo 3
Grouo 4
fJrQUD 5
Group 6
Group 2
5.2952





Group 3
7.1974
5.8535




Group 4
32439
4.9845
9.0612



Group 5
62445
5.4749
7.5224
5.5473


Group 6
61238
7.7708
11.6443
5.0761
10 2863

Group 7
7.3523
8.4692
97214
7.6695
10.2894
6.1324
                                   11-418

-------
                                                                      Page 9
Multilinear Regression Analysis

Regressions were performed on the absorbance data (log 1/R) without any prior
data  pretreatment.    Individual  models,   and  the  corresponding  regression
coefficients, were developed and evaluated to achieve maximum inference from the
regression analysis.  Three types of calibration curves for three groups  of liquid
incinerator feeds; type  1,  type 2 and type 4,  have been developed for heat of
combustion and water content  determination. These three groups comprise about
73% of liquid incinerator feeds analyzed by this technique.  Table III summarizes
the regression statistics on each constituent for the three incinerator feed types.

 Table III: Regression statistics for heat of combustion and water content equations

Tvpel BTII
H2O
Tvpe2 BTU
H20
Tvpe4 BTU
H20
F ratio
295.479
40.517
162.045
25.431
23.735
31.462
Corr Coef
0.968
0.821
0.950
0.768
0.861
0.890
SEE (W\
0.645
2.229
0.575
0.345
1.421
0.916
SEP (%\
0.730
2.301
0.589
0.398
1.436
0.932
Range*
Ilr000-20r000
0.1-10.0
5000-11 000
0.1-1.0
9.000-18rOOO
0.1-10
            Range for BTU is BTU/lb and % moisture for water content

In general,  the regression statistics given  in Table  III are used to evaluate the
validity of the regression model. The F-ratio for regression is a quality measure for
the regression that puts an overall goodness of the regression into one number.  A
high value of "F1 is indicative of a good fit obtained from many samples with a small
number of wavelengths. These kinds of calibrations will be more robust against
small variations and time.  The multiple correlation coefficient is a measure of error
versus  total variation and should tend to unity.  For a given range, the standard of
error of estimate (SEE)  and  standard error of prediction (SEP) evaluates the
calibration and the prediction model and should be as small as possible.

Besides evaluating the regression statistics given in Table III, the residuals must also
be examined to evaluate the adequacy of the regression model.  Figure 4 is a plot of
NIR predicted heat of combustion (BTU/lb)  values against the actual heat  of
combustion  values  obtained  using  bomb  calorimetry  procedure  for  type  1
incinerator  feeds.  All sample informations on  lack of fit is contained in the
residuals.
                                    11-419

-------
                                                                    Page 10
                           10     12    14    16

                                   Actual BTU
18
20
 Figure 4: Plot of NIR predicted BTU versus actual BTU from Bomb Calorimetry

If the regression model is valid, the residuals are the estimates of the model error,
which are assumed to have a normal distribution around the mean (/z =  0) and
constant variance. Figure 5 is a plot of residuals against the predicted BTU values
for type 1 feeds. According to Figure 5, the residual plot for type 1 feeds has  a mean
equal to zero and a constant variance, suggesting the robust nature of the calibration
curve.  Residual plots were also evaluated for type 2 and type 4 feeds before using
the calibration models to predict the heat of combustion and the water contents of
incinerator feeds.
                                  Predicted BTU
    Figure 5: Plot of residuals versus predicted BTU for type 1 incinerator feed

QA/QC Procedure

Instrument  performance  parameters must be  evaluated on  a daily basis to
demonstrate that the instrument is performing properly. Two instrument diagnostic
checks and one  instrument performance standard  were  developed  for QA/QC
procedure.  The  two instrument diagnostic checks are checking for the front end
board of the instrument and the amount of light energy passing through the filters.
                                   11-420

-------
                                                                   Page 11


Kerosene  was chosen  as  the  instrument performance standard because of its
consistency and wide use in bomb calorimetry procedure.

In addition to the instrument reliability, the method must also be proved reliable.
Method performance is monitored throughout the day by using a quality control
(QC) check sample. A  QC check sample is a material which represents the sample
matrix being analyzed.  Sample duplicate and fortified samples are used to measure
the precision and accuracy or the method.

CONCLUSIONS

NIR technology can be  used for analyzing incinerator feeds for heat of combustion
and water  content.  About 73% of  the incinerator feeds have been successfully
analyzed by the NIR technology.  Additional calibration curves will increase the
percent of samples analyzed as the database is  expanded.  The  elimination of
conventional bomb calorimeter  and Karl Fischer  titration for sample preparation
drastically  reduces the analytical  costs by  streamlining sample analyses.   The
runtime of two minutes  per analysis entails an 80% cost savings.

REFERENCES

1. Wetzel, D.L.;AnaL Chem., 1983,55,1165A-1176A.
2. Shah, N.K.; Gemperline, PJ.;Anal Chem., 1990,62, 465-470.
3. Mark, H.L.; Tunnell, D.;AnaL Chem., 1985, 57,1449-1456.
4. Johnson, R.A; Wichern, D.W.;  Applied Multivariate Statistical Analysis, Prentice
Hall: NJ, 1988, pp 300-314.
                                    1-421

-------
           SOURCE SAMPLING AND ANALYSIS GUIDANCE
                    A METHODS DIRECTORY1

Merrill D. Jackson and  Larry D. Johnson,  Quality Assurance
Division,   Atmospheric  Research  and  Exposure  Assessment
Laboratory,  U. S. Environmental Protection Agency,  Research
Triangle Park, North Carolina 27711; Kim W. Baughman, Ruby H.
James and Ralph B. Spafford,  2000 Ninth Avenue South, Southern
Research Institute,  Birmingham,  Alabama 35255

ABSTRACT

Sampling and analytical methodologies are  needed  by EPA and
industry for testing stationary  sources for specific organic
compounds  such as those listed under the Resource Conservation
and Recovery Act  (RCRA) Appendix VIII and Appendix IX and the
Clean Air  Act of  1990.

A computerized directory, Problem POHC Reference  Directory,
has been  developed  that supplies  information  on available
field sampling and analytical methodology for each compound in
those  lists.    Existing EPA  methods  are  referenced  if
applicable,   along with  their  validation  status.  At  the
present, the data base is strongly oriented toward combustion
sources.  The base may  be searched on the  basis  of several
parameters including name,  Chemical Abstracts  Service (CAS)
number, physical  properties,  thermal stability,  combustion
rank,  or general  problem areas in  sampling or analysis.  The
methods directory is menu driven and requires no programming
ability; however, some  familiarity  wit dBASE  III+  would be
helpful.

INTRODUCTION

There are  a large number of chemical compounds listed under
Appendix VIII1 and Appendix IX2 of  RCRA and the Clean Air Act
of  19903,   that  are regulated  by  the  U.S.  Environmental
Protection  Agency (EPA).    EPA has  several  sampling  and
analytical methods  which are validated for  many  of  these
compounds. Other  of the listed compounds may be analyzed by
these methods but they have not been validated.  EPA or State
permit writers and industry personnel may  not be familiar with
each compound and its status;  therefore,  a data base of each
compound listed along with its methodology, has been prepared.
If  the methodology has  been validated  for  a compound,  a
reference  is given; however, if no method has been validated,
1. This paper has been reviewed in accordance with the U.S.
Environmental Protection  Agency's peer and  administrative
review policies and approved for presentation and publication.
The mention of trade  names or commercial products does not
constitute endorsement or recommendation for use.
                            11-422

-------
the best method to try is indicated. Since the data base was
originally developed  for use  with incinerators, it  has an
orientation towards combustion methodology.

COMPUTER AND SOFTWARE REQUIREMENTS

An IBM PC or compatible system with a hard disk using DOS 2.0
or higher is required.   Version 1.0 of our program requires
have dBASE  III+ in order to  run and  is available  from the
National Technical Information Service (NTIS) under the name
"Problem  POHC  Reference Directory"6.   This  version  contains
only  the compounds  listed  under  RCRA in Appendix  VIII.
Version  2.0  will  also include  the compounds  listed under
Appendix IX and the Clean Air Act of 1990, and it  is scheduled
to be released  shortly. It will be titled "Source  Sampling and
Analysis  Guidance, Version  2.0" and will  be available from
NTIS.  We plan to have Version  2.0 in the compiled format;
therefore, this version  of our program will not need a data
base program such as dBASE I11+ or IV  in order to run.

DATA BASE CONTENTS

The  following   information  for  each   compound  is  given if
available:  (1)  name  of compound  (The  Appendix  VIII name is
given first with either the Appendix IX or the Clean Air Act
name given next.  If a common name that had not been used is
known, then  it is  given also.),(2) the CAS registry number,
(3)  chemical  formula,  (4)  molecular  weight,   (5)  compound
class,  (6)  University  of  Dayton  Research  Institute  (UDRI)
thermal stability class and ranking4,  (7) heat of combustion,
(8) combustion ranking5,  (9) boiling, melting and  flash points
and  water solubility,  (10)   information  on toxicity,   (11)
sampling and analysis methods,  (12) validation status of the
compound in the methods, (13) general  and specific problems,
(14) a description of the problems, and  (15)  solutions (if
known) .  The data in the base is not complete by any means and
is constantly being revised.  Yearly updates are planned.

RUNNING THE PROGRAM

The first screen seen  after  opening the program is the main
menu shown in Figure 1.

Selection of an option will start a new sequence.  Selection
of option 1  will print the entire  data base (warning: This
will take about 1.5-2 hours.).  This option will probably be
used only once  to provide a complete hardcopy of everything in
the data base;  additional copies can be photocopied. Selection
of option 4 will print a list of all the compounds with their
CAS numbers and data base record number. This is a very useful
tool to have available since  the data base record number is
                            11-423

-------
needed when using option number 2.

               Figure 1.  Opening Screen
                         MAIN MENU

         1.  PRINT ALL RECORDS IN DATABASE
         2.  PRINT A SPECIFIC DATABASE RECORD
         3.  LIST COMPOUNDS BY PHYSICAL PROPERTY,
            THERMAL STABILITY,  OR COMBUSTION RANK
         4.  LIST COMPOUNDS  BY NAME AND/OR CAS REGISTRY NUMBER
         5.  LIST COMPOUNDS BY PROBLEM AREAS
         6.  EXIT

ENTER YOUR CHOICE (1-6)  FOR THE ABOVE:
Using selection  number 2  will  bring up  the Records  Menu
(Figure  2).

                    Figure 2.  Records Menu
     PRINT A SPECIFIED DATABASE RECORD.
     SPECIFY THE RECORD TO BE PRINTED BY:

         1. RECORD NUMBER
         2. COMPOUND NAME
         3. CAS REGISTRY NUMBER
               OR
         4. EXIT TO MAIN MENU

ENTER YOUR CHOICE  (1-4)  FOR THE ABOVE:
Upon the entry of choice 1, 2,  or 3, the question "DO YOU WANT
A HARD COPY OF  THE DATA? (Y/N)" will  appear. Selection "yes"
will create a printed copy, whereas  a "no"  answer will only
bring the data  on  screen. The  search routine is such that the
record number is the fastest way to locate an entry; however,
if you do not know the data base record number, you may search
by either the name of the compound or  its CAS Registry Number.
The Records Menu will probably be the most used menu since it
provides the complete information on  a given compound.

Selection of the  third option  on the  Main Menu brings up the
Specific Compounds Menu (Figure 3).
                            11-424

-------
               Figure 3. Specific Compounds Menu
         LIST COMPOUNDS ON THE BASIS OF:

         1. UDRI THERMAL STABILITY CLASS
         2. UDRI THERMAL STABILITY RANKING
         3. MOLECULAR WEIGHT
         4. BOILING POINT
         5. MELTING POINT
         6. COMBUSTION RANK

         7. COMBINATION OF ANY TWO PROPERTIES,

         8. RETURN TO MAIN MENU
After selecting any of options 1-6,  the user will be prompted
to input  a  range for that option before  again asking if he
wants a  hard copy. Selection  of  number 7 will  result in a
request  for  the  two properties  and  the  range   for each
property.  This search and listing option  can be particularly
helpful  in  Principal  Organic Hazardous  Constituent  (POHC)
selection for trial burns,  since  compounds can be listed by
incinerability category and by physical properties.

The fourth selection on the Main Menu (Figure 1) will provide
an alphabetical  list  of  the  compounds  with  the  data base
number. This provides  you with the easiest method of searching
with option number 1 of the Records Menu  (Figure 2).

The Problem Menu (Figure 4) is selected from option 5  of the
Main Menu.

                    Figure 4. Problem Menu
         1. LIST ALL PROBLEM COMPOUNDS
         2. LIST COMPOUNDS BY GENERAL PROBLEM
         3. LIST COMPOUNDS BY SPECIFIC PROBLEM

         4. RETURN TO MAIN MENU

     ENTER YOUR CHOICE  (1-4) FOR THE ABOVE:




The first option will list every compound that is recorded to
                             1-425

-------
have any problem. The second choice brings up the screen shown
in Figure 5.

                    Figure 5.  General Problem Types
         1.  ANALYSIS
         2.  HAZARDOUS
         3.  SAMPLING

     SPECIFY GENERAL PROBLEM TYPE (1, 2, OR 3):
A selection here will list all problem compounds in the area
selected.   The third choice on the  Problem Menu probably is
the most useful  one since  it allows a more limited selection.
The menu which goes with the third choice is shown in Figure
6.

              Figure 6. Specific Problem Types
GENERAL PROBLEM
SPECIFIC PROBLEMS
1. ANALYSIS
2. HAZARDOUS
3. SAMPLING
         E. SENSITIVITY
         F. RECOVERY
         G. DECOMPOSITION
A. CHROMATOGRAPHY
B. INTERFERENCE
C. WATER SOLUBLE
D. BLANK

A. CORROSIVE
B. EXPLOSIVE
C. INCOMPATIBILITY
D. TOXIC

A. BLANK
B. BREAKTHROUGH
C. DECOMPOSITION
D. REACTIVE
SPECIFIED GENERAL PROBLEM TYPE (1,2,  OR 3)
After the user  selects the general  type from  the Specific
Problem Types menu,  then  the program  prompts  the  user to
select a specific  problem type  from the selections  on the
right.
                            11-426

-------
Samples  of printouts  of individual  compounds showing  the
actual information available are presented in Figures 7 and 8.
On Figure 7,  points of  interest are that hexachlorobenzene is
listed on both Appendix VIII and the Clean Air act of 1990 but
not on  Appendix IX. It  has  a UDRI class  and ranking.  Only
compounds  listed  on Appendix VIII have UDRI  ratings  at the
present time.  The record also indicates that we have several
areas not filled in yet.  The data base is not complete, and
data will be  added as we become aware of it.  The sampling and
analytical methods for  this compound are listed as suggestions
since they have not been validated. The heat of combustion is
listed  for help in determining  which compounds in a waste
mixture should  be  selected  as POHCs.  Figure 8 shows a fully
documented compound,  benzene. The  sampling  and  analytical
methods have  been validated, and the references are given. The
specific problem type is a blank  problem, and suggestions are
given on how to overcome this problem.

SUMMARY

A  data  base  program  listing sampling and  analysis methods
along with several characteristics of  each compound  listed
under RCRA Appendix VIII,  is available for  use  with dBASE
III+. The  data base permits  those personnel  who  need field
sampling and analytical procedures for regulation purposes to
have  a  single  reference for this  information.    A  second
version covering  RCRA  Appendix VIII,  Appendix IX,  and Clean
Air Act  1990 compounds will  be  available  in late 1991. The
second  version will be  a compiled program,  which  will not
reguire  any  additional  software (ie  dBASE  III+  or  IV)  to
operate.

REFERENCES

1.   Code of  Federal Regulations, 40, Part 261, Appendix VIII,
     p 90-98, July  1,  1990.

2.   Code of Federal Regulations, 40, Part 261, Appendix IX,
     p 98-117,  July 1, 1990.

3.   Clean Air  Act, Title III, Public Law 101-549, 1990.

4.   Guidance on Setting Permit Conditions and Reporting Trial
     Burn  Results.   Volume  II  of   the  Hazardous   Waste
     Incineration  Guidance  Series    p  105-123,  EPA-625/6-
     89/019,  January 1989.

5.   Guidance Manual for Hazardous Waste Incinerator Permits,
     Mitre Corp., NTIS PB84-100577, July 1983.
                             11-427

-------
6.    Baughman,   K.W.,   R.H.  James,  R.B.  Spafford   and  C.H.
      Duffey,  Problem   POHC  Reference  Directory,  EPA-600/3-
      90/094,  January 1991.
         Figure 7.   Data  Output  for Hexachlorobenzene

RECORD  NUMBER:        361              DATE OF LATEST ENTRY:    04/10/91

COMPOUND:   Hexachlorobenzene


CAS REGISTRY NO:   118-74-1
FORMULA:   C6-(C1)6
MOLECULAR WEIGHT:     284.80
COMPOUND CLASS:    Chlorinated aromatic
APPENDIX 8?  Y         APPENDIX 9?    N      CLEAN AIR ACT OF 1990?   Y

UDRI THERMAL STABILITY CLASS:   1
UDRI THERMAL STABILITY RANKING:    31

BOILING POINT, CELSIUS:    323
MELTING POINT, CELSIUS:    231
FLASH POINT, CELSIUS:
SOLUBILITY, IN HATER:   Insol  0.035  ppm

HEAT OF COMBUSTION, KCAL/MOLE:        567.70
COMBUSTION RANKING:     65

TOXICITY DATA:

SAMPLING METHOD:   SW-846 No. 0010  (MM5)

ANALYSIS METHOD:
SW-846  No. 8270  (Extraction,  GC/MS)

VALIDATION STATUS:

GENERAL PROBLEM TYPE(S):

SPECIFIC PROBLEM TYPE(S):

DESCRIPTION OF PROBLEMS:

SOLUTIONS:
                                 11-428

-------
               Figure 8.    Data Output  for Benzene
RECORD NUMBER:         77              DATE OF LATEST ENTRY:    12/13/90

COMPOUND:   Benzene


CAS REGISTRY NO:   71-43-2
FORMULA:   C6-H6
MOLECULAR WEIGHT:     78.11
COMPOUND CLASS:    Aromatic hydrocarbon
APPENDIX 8?  Y         APPENDIX 9?    Y      CLEAN AIR ACT OF 1990?    Y

UDRI THERMAL STABILITY CLASS:   1
UDRI THERMAL STABILITY RANKING:     3

BOILING POINT, CELSIUS:   80.1
MELTING POINT, CELSIUS:   5.5
FLASH POINT, CELSIUS:     -11.00
SOLUBILITY, IN WATER:   Sol

HEAT OF COMBUSTION, KCAL/MOLE:         780.96
COMBUSTION RANKING:     47

TOXICITY DATA:  Cancer suspect agent; flammable liquid

SAMPLING METHOD:   SW-846 No. 0030 (VOST)

ANALYSIS METHOD:
SW-846 No. 5040 or Draft No. 5041{Therm. Desorb./P and Trap-GC\MS)

VALIDATION STATUS:
The VOST  method has been  validated  for this compound  (See "Validation
Studies of the Protocol for the VOST" JAPCA Vol. 37 No. 4 388-394, 1987).
(Also see "Recovery of POHCs and PICs from a VOST" EPA-600/7-86-025.)

GENERAL PROBLEM TYPE(S):   Sampling

SPECIFIC PROBLEM TYPE(S):  Blank

DESCRIPTION OF PROBLEMS:
Cancer suspect.
Blank problem with Tenax
Benzene is a common PIC.  This may complicate interpretation of results,
and make it  difficult to achieve  acceptable ORE with  low waste  feed
concentrations.

SOLUTIONS:
Level of  lab blank should be determined in advance.  Calculations should
be based on waste feed concentration to determine if blank level will  be
a significant problem.  Benzene should not be chosen as a POHC at very low
waste  feed levels because  it  is  likely to make  blank  or  PIC problem
significant.
                                  1-429

-------
    A FIELD INVESTIGATION OF GROUNDWATER MONITORING WELL
                     PURGING TECHNIQUES

Van  Maltbv.   Research  Scientist,  Jay  P.  Unwin,  Regional
Manager, National Council of the Paper Industry for Air and
Stream Improvement, Inc., Central-Lake States Regional Center,
(NCASI), Western Michigan University, Kalamazoo, MI  49008.

ABSTRACT

A field investigation of commonly used monitoring well purging
techniques was conducted under different conditions including
type of pump, pump  inlet  location,  and the use of packers.
Tracers  including deionized  water,  fluorescent  dye,  and
lithium chloride were used to define the  amount of stagnant
water at any  given time  in the pump  discharge.   Tests were
conducted in  shallow  5  cm  (2 in) diameter wells.  The effects
of drawdown were examined.

All runs conducted in the absence of drawdown with the pump
inlet in a fixed position at or  above the screen  showed a
highly variable and unpredictable inclusion of stagnant water.
The use of packers did not completely  prevent the inclusion of
stagnant water into the pump inlet.  The inclusion of stagnant
water into a  sample was minimized by purging  from some dis-
tance above  the  screen followed by  relocation of  the pump
inlet into the screen for sample collection.  In wells where
drawdown occurred during purging, stagnant water inclusion was
minimized by reduced  pumping  rates  to   allow for  sample
collection during periods  of  well   recharge.   Real  time
monitoring of indicator parameters  such as pH, temperature and
specific conductance was not generally successful in indicat-
ing when purging was  complete.

MONITORING WELL PURGING

It  is generally  recognized  that the  composition  of  the
stagnant water within  a  monitoring  well above the  screened
section is probably not representative of the overall ground-
water quality at the sampling site.  The water standing in the
well casing  is commonly referred to  as being  stagnant, that
is, the water has been isolated  from the  aquifer  at  least
since the last time  the well was sampled.   During that time,
the chemical  quality of the stagnant water may have changed by
(a) direct introduction of foreign material  into the well, (b)
interactions  with the well casing or at the  interface with the
atmosphere,  or (c) biological  activity.    Even without such
alterations,  the  stagnant water would not reflect any changes
in the groundwater quality that may  have occurred since the
last time the well  was sampled.   Because  the investigator
cannot be certain which,   if  any,  of these  influences  has
occurred or whether inclusion of some  of the stagnant water in
a sample from the well would significantly change the conclu-
                            11-430

-------
sions that might be drawn from the data, the safe thing to do
is to prevent or minimize such inclusion.

One presently used purging technique presented as a coopera-
tive agreement by the Illinois States Water Survey (ISWS) and
Illinois State  Geological  Survey (ISGS)   (1)  is  to  pump the
well and calculate the percentage of water at any given time
in a pump discharge that can  be attributed to drawdown.  This
approach is based  on the knowledge  of time-drawdown charac-
teristics of a well and does  not  account for contributions of
stagnant water from any source other than drawdown.   Another
commonly used purging technique presented by the U.S. Geologi-
cal  Survey (USGS)  (2)  is to pump a  well until  indicator
parameters such as  pH,  temperature,  and  conductivity stabi-
lize.   This  approach ignores the  possibility  that  a  near
constant contribution of stagnant water into  the sample may
result in stabilized readings for the observed  parameters.  It
also fails to account for contributions of  stagnant water that
are too small to notably affect the measured parameters, but
which may  significantly alter the  outcome of  an  analysis.
Probably, the most commonly used  purging practice is to purge
an arbitrary number of bore volumes  (well  casing) with little
or no regard to drawdown or  indicator parameters.

This  research reflects  the   need  for documentary  evidence
regarding the hydraulic behavior of a monitoring well during
purging.    By  examining  truly  trace  concentrations,  the
fraction of stagnant water entering a pump  inlet can be better
defined as  a  function of bore volumes pumped (or  time)  and
inlet position.

CRITICAL REVIEW OF EXISTING  INFORMATION  ON MONITORING WELL
PURGING

Illinois State Water  Survey  and Illinois State Geological
Survey    ISWS  and  ISGS have published "Procedures  for the
Collection of Representative  Water Quality Data From Monitor-
ing  Wells"  (1)  which  describes in  detail  guidelines  for
monitoring well purging.   The basic assumption  made in this
research  was that  during the  initial pumping  of  a  small
diameter monitoring well, a  significant fraction of the pump
discharge comes from stagnant water within the  well casing.
This  effect  is  due  to drawdown.    The  procedure   uses  an
equation which develops time-drawdown data  based on individual
monitoring well hydrologic data  obtained  during pump tests.
The resulting theoretical drawdown  curve  is used to predict
the  time at  which  the effects of casing  storage due  to
drawdown become negligible.   This curve is  intended to be used
as a guideline,  in conjunction with the observation of indica-
tor parameters  for the selection of  an appropriate pumping
rate and number of bore volumes to be pumped prior to sample
                            11-431

-------
collection.

Six monitoring wells at different locations within the State
of Illinois were  used for the  development of  this purging
protocol.    NCASI  encountered  several  difficulties  while
reviewing the  results of the study on these six wells.  These
difficulties  included  (a)  reproduction  of the  theoretical
drawdown curves presented, and (b) interpretation and conclu-
sions drawn from the monitored  indicator  parameters.   NCASI
and others have been  unable to reproduce any of the published
theoretical  drawdown  curves for six wells examined by ISWS and
ISGS.  For each well, ISWS and ISGS have provided theoretical
drawdown  curves  derived  from  the  Papadopulos  and  Cooper
equations which generally show good agreement with the actual
drawdown curves presented.  NCASI has used these equations in
the manner described by ISWS  and ISGS  to produce theoretical
drawdown  curves  which bear  little  resemblance  to  those
published.  A careful examination by NCASI  has  not revealed
the reason for these discrepancies.

The cooperative agreement examined the effects of well purging
on the chemical composition for six monitoring wells.  Five of
the six wells  were described  as  having site specific limita-
tions  which hindered  interpretation  of the  results.   The
single well  (Site  5), in which a clearly indicated effect of
purging on indicator parameters was noted,  directly contra-
dicted information from the pump test portion of  the study.
In spite of limitations described for each  of the six sites
examined,  ISWS and ISGS concluded that "the chemical data from
this portion of the study have verified the theoretical ratios
of aquifer to  stored water predicted during the  pump tests".
A  subsequent  publication by  the  ISWS  "Practical  Guide  For
Ground-Water Sampling  (3) has  endorsed the above  mentioned
purging protocol.

United States  Geological Survey  The United States Geological
Survey (USGS)  (2) states that  in  order to obtain a representa-
tive sample  from an aquifer at a given  location,  a well must
be pumped  until indicator parameters such as pH,  temperature,
and conductivity are constant.  Measurement of drawdown during
the purging period is recommended because changes in the indi-
cator parameters may reflect water from different zones of the
aquifer being  drawn into the well.   This procedure is stated
as the minimum required precaution for insuring that a sample
adequately represents the water  quality in the aquifer.

Guidelines  for indicator  parameter   stability  have  been
presented by Gibs  and Imbrigiotta (4) .  Research  by Slawson
et al. (5) examined the variability of  indicator  parameters
and other constituents in well  discharges during  continuous
pumping.   Appreciable changes were observed in several of the
                            11-432

-------
parameters,  most notably conductivity.   These changes were
attributed  to naturally  occurring vertical  and horizontal
variability within the aquifer from natural  influences.  Such
a situation would not allow for the universal use of indicator
parameters to  determine  when a groundwater sample should be
collected.

Consideration  should also be given to the general nature of
the specified  indicator parameters.  Conductivity and pH are
may be affected by changes in temperature  and pressure during
sampling.   Pressure changes can  cause rapid  degassing of
carbon dioxide and other  gases that could affect sample pH and
specific conductance.

NCASI  Laboratory Purging Investigations  NCASI conducted a
laboratory  investigation  to  examine  factors  other  than
drawdown that could cause stagnant water to enter a pump inlet
(6) .   All tests were conducted  at constant  head,  thereby
disregarding  the  effects of  drawdown.    In  the  research,
stagnant water in a well  column was spiked with  a fluorescent
dye.  Care was taken to minimize density differences induced
by either temperature gradients between stagnant and aquifer
water or concentration induced density gradients.   The well
was then  sampled with the  pump  inlet  in various positions
while  the tracer concentration  in the  pump  effluent  was
continuously monitored.  Results demonstrated that an average
of about  2  to 4 percent  of  the water pumped from locations
above the  screen and an  average  of about  1 percent  of the
water pumped  from within the screen  of the monitoring well
came from  the stagnant water located  above the pump inlet.
The mechanism that  caused  the overlying stagnant  water to
reach  the  pump  inlet was  not investigated,  though  mixing
caused by turbulence around the pump inlet was hypothesized.

University of  Waterloo   Robin and Gillham  (7)  conducted a
study using non-reactive  tracers to judge the effectiveness of
various purging procedures.   The  results  suggest a  sharp
interface and little mixing  between fresh water  in the screen
or below a pump  inlet and the stagnant water in the casing.
For wells not completely evacuated, pumping from immediately
beneath the air/water interface for 2  or 3  bore volumes was
deemed sufficient to collect a representative sample.   Three
tracers  were  used  for  the  study:  deionized  water,  NaCl
(conductivity), and bromide.   Of these, NaCl was demonstrated
to be an inappropriate tracer due to mixing caused by density
differences between  the  tracer and the fresh water.   While
deionized water  was  determined to  be  an  appropriate  tracer
(verified with bromide)  an even greater  density difference
existed between deionized water and the natural groundwater at
the site.  Although  the  deionized  water was less dense than
the groundwater,  the effect of this  density difference may
                             1-433

-------
have acted to discourage mixing around the pump inlet.


EXPERIMENTAL PROCEDURE

Wells

Two shallow monitoring wells were selected for this research.
Both wells were constructed of 5-cm ID PVC  (2-in) with commer-
cially available PVC  screened sections (Timco  0.010  slot).
Both wells  are  situated in  a shallow glacial  unconfined
aquifer composed of sand and clay.   Each well is approximately
9.2 m (30 ft)  deep  and has a standing water level within 1.5
m  (5 ft)  of the  ground  surface.  Although these  wells were
only approximately 62 m  (200  ft)  apart,   local variability
within the  aquifer  accounted  for  marked  difference  in the
hydraulic performance  of each well. At the purging rates used
in this research, one  well experienced very minimal drawdown
(less than 0.6 cm,  0.25 in),  and  the other well experienced
extreme drawdown  and could easily  be  pumped dry.

Equipment

Two  pumps were  used  for  purging in this  research.   The
majority of the purging  runs were conducted using  an above
ground peristaltic  pump  (Masterflex,  #70-15  head)   with  a
maximum flow rate of approximately 800 ml/min.   Several runs
were conducted with a  submersible pump (Keck #84) with a flow
rate of approximately  4.5 L/min.

The  fluorescent  tracer  concentration in  the   purging  pump
discharge was detected with a fluorometer  (Turner #111)  with
a flow-through cell for continuous measurement.  Conductivity,
drawdown, temperature,  and pH  were   monitored  continuously
using methods described elsewhere (7).   Drawdown was moni-
tored with a submersible pressure  transducer.   All data from
the instruments with the exception of pH were  recorded on a
portable computer.  Values for pH  were recorded manually due
to pH  signal  recording  difficulties.   A portable  electric
generator provided  electrical power where  needed.

The amount of stagnant  water  in  the pump discharge  at any
given time was measured directly by the use of one or more of
the following tracers:   Rhodamine WT,  lithium  (as  lithium
chloride), or deionized  water.  Rhodamine  WT was used as the
tracer of choice during this research. Rhodamine WT is a non-
toxic fluorescent xanthene dye  commonly used in percolation
studies,  potable water  systems, and   surface water  systems.
Additionally,  Rhodamine  WT exhibits  both  low reactivity and
sorption tendencies (8)  making  it well  suited  for this res-
earch.  The initial concentration after the dye had been added
                            11-434

-------
to the  stagnant water was  approximately  200 ppb.   At this
concentration in water, the dye  imparts  no color detectable
with the naked  eye.   In several  of  the purging tests in the
well in which  drawdown occurred, the  presence of turbidity
interfered with the detection of the fluorescent dye.  In this
situation, lithium was used  as the tracer.   For  several of the
tests,  deionized water was  used as a  tracer in conjunction
with Rhodamine WT.

General Procedures

At the start of each purging test, the  static water level was
measured.   Tracer was added  to the stagnant water in the well
in a  manner that  resulted in  a homogeneous  concentration
within the stagnant water column, without migration into the
screen area.   To accomplish this,  an  inflatable  packer was
used to hydraulically isolate the screened portion of the well
from the cased  portion above.  The  packer was  designed as a
flow-through device so that  water could be collected from the
screen area  during the time the packer was  inflated and in
position.

The  stagnant water above  the   packer was  pumped out  and
collected in a container at the surface.   Aquifer water from
the screen  area was  pumped to  the surface at the  planned
purging rate in order to  zero  the fluorometer  and obtain
background  readings for  pH,  conductivity  and temperature.
With  the   fluorometer zeroed  for  the aquifer water,  the
previously  collected storage water was  pumped through the
fluorometer in a closed loop system. Tracer was added to the
casing water until the fluorometer  readout was 100 percent.
This casing water  containing the tracer was poured back into
the well  to a  water  level  slightly below the static water
level so that the volume displaced by the packer would be re-
placed  by aquifer water  moving  into  the  screened  section
rather than  by spiked water moving down  into the screened
section when the packer was removed.

The stagnant water containing  the  tracer  was  kept isolated
from the  screened  section  with the  inflated  packer  for a
minimum of 12 hours to ensure that  the undisturbed tempera-
ture-depth profile of  the stagnant  water  would be re-estab-
lished.   A  preliminary investigation  of the wells used for
this research,  revealed  that the temperature-depth  profile
determined after a month  of non-pumping,  would be re-estab-
lished approximately 8 hours after having been disturbed.

A  run  was  initiated  by  slowly deflating  the packer  and
carefully removing it from the  well.    The pump  inlet was
placed at  a predetermined  depth  for  the run.    Pumping and
automated data collection  were started.  Fluorometry readings
                            11-435

-------
were recorded every two seconds,  other readings except pH were
recorded every five seconds.  Readings for  pH  were recorded
every 105 seconds.   The computer  and  software allowed for
real-time observation of data throughout  the purging run.

At the  end  of  a test,  the pump  inlet  was  raised to the
air/water interface while pumping continued until the fluorom-
eter readings  returned to  background.   At this  point new
tracer could be added for the next run.

For the  runs  where  it  was applicable,  the number of  bore
volumes pumped was based on the volume between the pump inlet
(bottom of sample line)  and the  top of  the  well screen.   For
the runs where the pump inlet was placed immediately above the
screen top,  the duration of a run was measured in the number
of liters pumped,  rather  than  bore volumes.   Because the
measured tracer concentration in the aquifer water was  zero
for all runs, the fraction of the pumped volume that came from
the stagnant zone above the pump  inlet is simply the ratio of
the measured concentration of tracer in  the pumped water to
the initial  concentration in the stagnant zone.

Two packers  were investigated to determine their  effect on
stagnant water  concentration during purging  and  sampling.
Such packers form a seal against  the inside of the well casing
to isolate the stagnant water above from  the pump inlet.  In
effect, the  volume  of standing water above  the pump inlet is
reduced.   One  packer  was  a commercially available  unit
attached to  the top of the Keck  pump.  The  other packer was
laboratory made and designed as  a flow  through device.

The effect of purging and sampling  in a  well experiencing a
significant  degree  of drawdown was investigated.   During
drawdown, the  balance of the water in the pump discharge not
accounted by flow into the well from the screen comes from the
stagnant casing  water  above the pump  inlet.   A  sample was
collected from such a well by purging the  well at a  rate great
enough to produce  drawdown and  thus reducing the level of
stagnant water.   Sample  collection occurred  at  a  reduced
flowrate during water level recovery. Purging and sampling in
this manner  allows the stagnant water/aquifer water interface
to move  upward and away from the pump  inlet,  reducing the
chance  for  stagnant water  to become  captured by the  pump
inlet.  As  noted previously, due to excessive turbidity in
this well, lithium chloride at an initial concentration of 62
mg/L was used  as the tracer.

The following  purging configurations were investigated during
this research:  (a) fixed pump inlet positions at approximately
5 cm (2 in)  below the air/water interface, mid-casing, and at
the top of the well screen, (b)  a comparison of Rhodamine WT
                            11-436

-------
and deionized water as tracers with the  pump inlet reposi-
tioned between the screen and the air/water interface several
times  during  the  run,  (c)  purging  from  above  the screen,
followed  by  sampling  within  the  screen,  (d) packers,  (e)
drawdown and recover.   Details specific to  each test  are given
in the section below.

RESULTS AND DISCUSSION

Drawdown did not become significant during  the  following tests
(maximum 0.6 cm,  0.25  in) at the specified purging rates.

Peristaltic Pump. Inlet Near Static Water Level

Figure 1 graphically presents  the  results of  a purging test
with the pump inlet located approximately 5 cm (2 in)  below
the static water  level in the  monitoring  well.   The purging
rate during this test was 1.01 L/min.

During most of the time required  to  remove the initial bore
volume of water from the well  (14.3 minutes),  the concentra-
tion of stagnant water in the pump  inlet was 100 percent.  As
the column of  fresh water moving up the  well casing approached
the pump  inlet,  there  was a  corresponding rapid decrease in
the stagnant  water concentration detected in the pump dis-
charge.   During  the removal of subsequent bore  volumes the
concentration of stagnant water continued  to decrease but was
still detectable in the pump discharge for a  relatively long
time  (9.0 bore volumes,  128  minutes).   At  this  point the
majority of the stagnant water had been removed from  below the
pump inlet, however,  it may not have  been entirely removed
from  above the  pump  inlet.    Earlier research  (9)  using
visible  dye  concentrations  in  transparent  well  casings
provided evidence that the intermittent inclusion of stagnant
water detected in the pump discharge  may have  been related to
turbulence and subsequent mixing around the pump inlet.  The
reason for  the relatively small concentrations  of stagnant
water detected at  the  end of the  run is probably related to
the fairly  small  volume of stagnant water above the inlet,
(0.1L).  This volume would most likely have been diluted by
mixing during the pumping of 9 bore volumes (130L) .   At 9 bore
volumes,  the  pump  inlet  was  lowered into the  screen  area.
With the pump inlet located in  this position,  one bore volume
of water (14.4 liters)  was collected  without the detection of
stagnant water in the pump discharge.   Figure  2 presents a
five minute expanded segment of Figure 1  between 4.0 to 4.3
bore  volumes.   It  shows the intermittent  nature of  the
intrusion of  stagnant  water  into the pump inlet.   The bore
volume axis has  been converted to time.   While  the contri-
bution of stagnant water is relatively small at this  point, it
is nonetheless measurable, and  for intervals up to 30 seconds.
                             1-437

-------
                      456
                       BORE VOLUMES
                10
     Fig.  1- Inlet near static water level
              BORE VOLUME 4.0 TO 4.3
         57.5  58   58.5   59  59.5  60
                        TIME, (min)
                                        -A-
60.5  61   61.5   62
Fig.  2- Expanded view,  bore  volume 4.0  to 4.3
                       11-438

-------


I
a
M
^
c
o
1
>-
>
JH
o
a
-7

o


1 U
9 -
J

8-

7 -

6-
4-

3-

2-


1 -
n -

DEPTH



pH

\^/^
\ CONDUCTIVITY




•
TEMPERATURE
. — L. -A
r \ — . — . — ^ — ~ — ^— — ~ —
'

-/4
-23

-22

-21

-20
-19
-18
-17
-16

-15

-1 4
1 *T
-13
-1 O





j:

x-
uj
Q
^^s
O,

a."
u
i—




         0
456
BORE VOLUMES
10
                Pig. 3- Indicator parameters
One implication of this is that approximately 500ml of water
could be collected as  a  groundwater sample over a 30 second
time interval.  Depending on the purpose for which the sample
was being collected,  inclusion of almost 1 percent stagnant
water could be problematic.

As shown in  Figure 3, temperature,  pH, and conductivity during
this run displayed an initial response during the removal of
the first bore volume.  However, they exhibited little or no
change  during  the pumping  of  subsequent  bore volumes.   A
visual inspection of these parameters between bore volumes 1
and  2  indicate  little if  any change,  whereas  the  tracer
concentration shown  in Figure 1  indicates  a  stagnant water
concentration decreasing between 11 and
2 percent.   These parameters  were too  insensitive  for the
detection of trace concentrations  of stagnant water in this
run.  An examination of these parameters for all of the other
purging runs revealed similar results.

Peristaltic Pump. Inlet at Mid-casing
                             11-439

-------
Figure 4 displays the results of a purging test with the pump
inlet located in the mid-casing position with equal amounts of
                        PUMPING TIME 90 MINUTES
                                    INLET LOWERED TO SCREEN

                                          AT 10.6 B.V.
                            567
                           BORE: VOLUMES
10   11  12
                 Fig.  4- Inlet at mid-casing
water above  and below  the  inlet.    Examination of  Figure  4
reveals results similar to the purging run displayed in Figure
1 in that the  majority of the stagnant water below  the pump
inlet was  removed  with the  initial bore  volume of  water.
While the stagnant water concentration tapered off more slowly
between bore volumes 2  and 4 with the  inlet position located
in the mid-bore position, the subsequent bore volumes pumped
after the fourth bore volume appear to  have elevated  stagnant
water concentrations  when compared  to  the run displayed  in
Figure 1.

Peristaltic Pump, Inlet at Screen Top

Figure 5 displays the results of a purging run  with  the pump
inlet situated immediately above the well  screen.
Because there was no  volume  of  storage water below  the pump
inlet,  the lower  x-axis in  Figure  5 refers to  the number  of
liters pumped.  The absence  of  storage water below  the pump
inlet is evidenced  by the lack of an initial interval during
                             1-440

-------
  20

  18-

  16 -


z 12-
<
i 10-
              LJ

              <
             O
             ex.
             LJ
             D-
6 -



2-

0
0
                               PUMPING TIME 80 MINUTES
                                           INLET RAISED 10 FEET
                          IMLET LOWERED TO SCREEN
i
             20
                      30    40
                     LITERS PUMPED
                                                50
60
70
                       Fig. 5- Inlet at top of  screen
        which the pump discharge was 100 percent stagnant water.  This
        indicates that the pump inlet was positioned  at  the stagnant
        water/fresh water  interface.   The  results suggest  that the
        stagnant water detected in  the  pump discharge during  the 80
        minute  pumping   period was  caused  by  the  stagnant  water
        overlying the  inlet mixing  with the  fresh  water  from the
        screen.   At  60 minutes, the pump inlet was lowered 5 cm (2 in)
        to the top of the  screen  with little effect  on  the stagnant
        water concentration.  At  69  minutes,  the inlet was raised 3 m
        (10 ft)  which increased the  stagnant water concentration to
        100 percent.

        Source of Stagnant Water

        The results  of the preceding three purging trials suggest that
        the stagnant water  detected in the pump discharge was from the
        dyed water overlying the pump inlet.  The reasons are twofold.
        First,  the  concentration  of stagnant  water detected  in the
        pump discharge was  apparently affected by the volume  of dye
        overlying the  inlet.  A mathematical average  of  the stagnant
        water concentration detected between the  fifth and the sixth
                                    11-441

-------
bore  volumes was  determined  to  be 0.026  percent,  0.199
percent,  and 0.103 percent for Figures 1, 4, and 5 respective-
ly (Fig.5 concentration determined by examining the data be-
tween 38 and 45 minutes, an  interval  equal  to that selected
for the concentration  determined in Figure 4) .   One expla-
nation for the average  stagnant water concentration in Figure
1 being approximately one order of magnitude smaller than that
of Figures 4 and 5  is  the smaller volume of  stagnant water
(0.1L -vs- 7.2L and 14.4L) available for mixing above the pump
inlet.  The much smaller volume would have been diluted to a
greater extent over a given  time period,  accounting for the
smaller percentage  of stagnant water in  the pump discharge.
Second, the results of the run displayed  in  Figure 5 verify
that the pump inlet was  positioned at the stagnant water/fresh
water interface.   The  detection  of  stagnant water  for the
duration of the  80 minute run indicates that the tracer above
the pump inlet was  being captured.

Multiple Inlet Positions.  Dual Tracers

In an attempt to repeat a purging test conducted by Robin and
Gillham (6),  deionized  water  was  used as a tracer in addition
to Rhodamine WT.   During this  purging  test, varying inlet
positions were used starting from the screen top.  The purging
rate was 1.1 L/min.   The water above the screen was replaced
with  deionized  water  containing  Rhodamine  WT  using  the
inflatable packer in the  manner described previously.   The
well was pumped  with the inlet immediately above the screen
while the fluorescence  and conductivity were being monitored.
The well  was  pumped in this manner until the  fluorescence
approached zero. The inlet was then raised 1.2  m (4 ft) and
pumping was continued at the new level until the fluorescence
approached zero.   This process  was  repeated several times
until the air/water interface was reached.

The results  of  this test  are presented  in  Figure  6.   In
general,  the shapes  of the  curves  obtained with  the fluo-
rescent dye are very similar to the initial portions of curves
shown in Figures 1,  4,  and 5  in that removal  of the majority
of'the stagnant  water below the inlet occurs  fairly quickly.
The curves from the deionized water conductivity measurements
however,   contradict  those for  the  fluorescent  dye.   The
deionized water  tracer  suggests  background water quality was
reached within a few minutes of  each  relocation  of the pump
inlet, whereas  the  curves from  the  fluorescent  dye  tracer
still show  stagnant  water detectable in  the percent range
several minutes  later.   The pattern of response  for each of
the tracers was  repeated each time the pump inlet was raised.

Deionized water  may be  too insensitive a tracer when used to
differentiate water sources in a monitoring well.  Conclusions
                            11-442

-------
                                   CONDUCTIVITY
                                  (DEIONIZED WATER)
               10
20
30    40
TIME, (min)
50
60
                                                    1.2
                                                   -1
                    STAGNANT WATER

                    (FLUORESCENT DYE)
70
                                  TJ

                                  O
                                  i_
                                  D>
                                  .*
                                  O
                                  o
                                  .a
                                   _
                                   W
                                   o
                                   0)
                                   E
                                                       O
                                                       O
  Fig.  6- Comparison of Rhodamine WT  and deionized water as
                           tracers
as to required purging volumes based on deionized water tracer
tests may, therefore, be inappropriately low  for many  purpos-
es.
Peristaltic  Pump.  Pump and Lower

Figure 7 displays  the results of a purging test during which
the pump  inlet was initially positioned  five feet above the
screen for approximately five bore  volumes (18.4L) and then
repositioned into  the screen  for  sample  collection.   The
advantage  of purging  and sampling  in this fashion  is that the
entire column of stagnant water need not be  purged  in order to
collect a  representative groundwater  sample.   This test was
conducted  in triplicate with similar curves observed for all
three runs.  During the removal of the initial five bore vol-
umes of water  a clean zone generally  free  of stagnant water
was developed below the pump inlet.   After apparently lowering
the pump  inlet,  a  minimum 40 liter  pumping period followed,
during which stagnant water was not detected in the discharge.
                             II-443

-------
                        PUMPING TIME 105 MINUTES
                   INLET LOWERED TO SCREEN

                       (18,4 LITERS)
                                         INLET RAISED
             10  20  30  40  50  60  70  80  90  100  110
                            LITERS PUMPED
                    Fig.  7-  Pump and lower
Stagnant water was eventually detected in the pump discharge.
This occurred  while the pump  inlet was still  in the screen
area meaning  that the  stagnant water had  migrated over the
five foot  separation between the  initial  inlet position and
the screen.   The mechanism  for this migration  has not been
investigated, but is likely due to a combination of diffusion
and advection.

PACKERS

Figures 8  (Keck pump)  and  9  (laboratory manufactured packer)
present the results of purging runs using packers.  Tracer was
added  to  the well  for  each trial  in  the  manner described
earlier.  At the start of each test the packer used for tracer
addition was removed.  The  packer used for sampling was slowly
lowered so that the  pump inlet  for each trial was positioned
immediately above the screen.   The packer  was  inflated and
pumping was initiated.

Neither packer functioned as  anticipated. This is evidenced by
the unexpected contribution of stagnant  water to the pump dis-
                             II-444

-------


Of.
Ld
^_

1—
z
z
o
1—
z
Ld
o
cr
1 1
LlJ
Q_



ZU
18-
16-

14-
12-
10-
8-
6-

4-
2-
o-
1







PUMPING IIME /4 WINUILS









,[
^iJlliilwiiilLi,, i, i, Li J. L i
i i
0 50 100 150 200 250 300 350 4C
                           LITERS PUMPED
                Fig.  8-  Keck Pump with packer
charge.  While stagnant water was detected in both runs, there
are marked differences between Figures 8 and 9.  Both curves
exhibit  the  same  general shape.  However, the  curve for the
Keck pump (Figure 8)  is  dramatically shifted to the  right.  A
portion of the curve from Figure 8 is superimposed onto Figure
9 for  comparison.   There are three  possible causes for the
higher than expected contribution of stagnant water in these
tests.  First, leakage past the packer induced by a decrease
in head below the packer during pumping may  have caused stag-
nant casing  water  to  migrate past the packer  into  the pump
inlet.  Such a leak could have been caused by an irregularity
or crack in the casing wall.  Second,  in  the case of the Keck
pump and packer, a  significant volume of water is displaced
(2.66 L) when the pump is lowered into position.  This
volume is equivalent to  a water level rise of 1.31 m   (4.3 ft)
in a 5 cm (2 in) well casing.  It is assumed therefore, that
stagnant water was displaced  downward through the screen into
the aquifer.  This may have caused the dramatic shift to the
right of the curve for the Keck pump.   Finally,  stagnant water
held between the packer and the pump inlet may have been drawn
in  by  the  same  phenomenon  observed  in  other tests  were
                             1-445

-------
20

18-

16
      <

      <
        8-
      LJ c J
      O 6 I
      o:
        2-

        0
         0
                       PERCENT STAGNANT WATER
                       FROM KECK PUMP/PACKER
                            INLET RAISED TO
                            NEAR SURFACE
                            AT 28 MM.
                    _ ....1 ....ป- A ซ
                           j
                   10       15
                    LITERS PUMPED
20
25
            Fig. 9- Peristaltic pump with packer
stagnant water was above the pump inlet.  The  inlet on the
laboratory manufactured  flow-through packer  was immediately
below the packer.   However the Keck pump inlet was situated
approximately 30 cm (12  in)  below the bottom of the packer.
This may explain why the test with the Keck pump showed much
more frequent and higher  spikes  of stagnant water than did the
flow-through packer.

Purging/Sampling With Drawdown

Figure 10 displays stagnant water concentrations for a well
sampled  by  the  drawdown  and  recover  method  described
earlier.  The pump inlet was positioned at the screen top for
the duration of the test.   The initial pumping rate for the
first 25 minutes  was approximately  750 ml/min.   During the
initial portion of this time period when the rate of drawdown
was the  greatest  (as  indicated  by  the  slope of  the curve
marked  "depth")  a relatively  larger  portion  of the  pump
discharge would  have  been  from  stagnant water.    This   is
confirmed by the larger lithium concentrations.  As the rate
of drawdown decreases  (indicated by decreasing slope of the
                             11-446

-------
                10
20      30     40
    TIME, (min)
50
60
           Pig.  10- Purging/sampling with drawdown
"depth"  curve)  the lithium  concentration  also decreases,
indicating a  smaller fraction of stagnant water in the pump
discharge.  At 25  minutes the pumping rate was decreased to
300 ml/min to allow the well to  recharge.   Evidence of the
stagnant water/fresh water interface, moving upward away from
the pump inlet is shown by the rapid decrease  in  lithium con-
centration.  Sample collection during the period of recovery
would minimize  the inclusion of stagnant water.

The principle illustrated here is related to that used in the
ISWS procedure  (1).  However, since the mathematical methods
used there were inadequate to predict the drawdown for this
well, it was necessary to measure drawdown directly in order
to determine when a sample could be  obtained.  Direct measure-
ment is  not  difficult in  such  a low yielding well,  and it
could be part of the sampling protocol for such a well.

SUMMARY AND CONCLUSIONS

The  results  of  the  research presented  here show  that the
procedures used to sample a  well can have an effect on the
                             1-447

-------
amount of stagnant water in a groundwater  sample.   All runs
conducted in the absence of drawdown with the pump inlet in a
fixed position at or above the screen showed a highly variable
and unpredictable inclusion of stagnant water.  This inclusion
of stagnant water may have been  caused by  turbulence around
the pump  inlet.   Packers  were  not  generally  effective  in
preventing the  inclusion of  stagnant  water  into the pump
inlet. Deionized water as a tracer in  this  study was general-
ly  ineffective   and  possibly misleading  when  compared  to
Rhodamine WT.  The results of tests conducted with deionized
water as a tracer which suggest  pumping 2  to  3  bore volumes
from near the air/water interface may be inappropriately low.
Real  time  monitoring of  indicator parameters  such as  pH,
temperature and  conductivity was  not successful  in indicating
when purging was complete.   Research by Gibs and Imbrigiotta
(4) resulted in a similar conclusion.

The inclusion of stagnant water into a  sample was minimized by
purging  from some  distance  above  the  screen  followed  by
relocation of the pump  inlet into  the  screen for  sample
collection.   In  wells where drawdown occurred during purging,
stagnant water  inclusion  was minimized by reduced  pumping
rates to allow  for sample collection  during  well  recharge.
The Illinois  State  Water Survey procedure for calculating the
effect of drawdown was found not to be  usable for the wells in
this research.

REFERENCES

(1)  Illinois State Water Survey  and Illinois State Geologi-
     cal  Survey, Cooperative Groundwater Report 7,  "Proce-
     dures for the Collection of  Representative Water Qual-
     ity Data from Monitoring Wells",  Gibb, J.P.,  Schuller,
     R.M.,  and Griffin, R.A.,  State of Illinois Department
     of Energy and Natural Resources,  1981.

(2)  Wood,  W.W., "Guidelines for  Collection and Field Analy-
     sis of Groundwater Samples  for Selected Unstable Con-
     stituents", U.S.  Geological  Survey Techniques of Water
     Resources Investigations, Book 1, p24, 1976.

(3)  Barcelona,  M.J.,  Gibb,  J.P., Helfrich, J.A.,  and Gar-
     ske, E.E.,  Practical  Guide  for Ground-Water Sampling.
     Illinois State Water Survey, November  1985.

(4)  Gibs,  J., and Imbrigiotta, T.E,  "Well-Purging Criteria
     for Sampling Purgeable Organic Compounds",  Ground
     Water,  Volume 28, Number 1,  January-February 1990.
                            11-448

-------
(5)   Slawson,  G.C.,  Jr, Kelly,  K.E.,  and Everett, L.G.,
     "Evaluation of Ground-Water Pumping and Bailing Meth-
     ods-Application in the Oil Shale Industry",  Ground
     Water Monitoring Review.  Summer 1982.

(6)   Unwin, J.P., and Maltby,  C.V.,  "Investigations and
     Techniques for Purging Ground-Water Monitoring Wells
     and Sampling Ground Water for Volatile Organic Com-
     pounds",  Ground-Water Contamination: Field Methodsf
     Collins,  A.G.,  and Johnson, A.I.,  Eds.,  ASTM STP 963,
     1988.

(7)   Robin, M.J.L.,  and Gillham, R.W.,  "Field Evaluation of
     Well Purging Procedures",  Ground Water Monitoring
     Review. Fall 1987.

(7)   "Guide to Groundwater Sampling", NCASI Technical Bul-
     letin No. 362,  January 1982.

(8)   Replogle, J.A., Myers, L.E., and Brust, K.J., "Flow
     Measurements with Fluorescent Tracers", Journal of
     Hydraulics Division. September 1966.

(9)   Unwin, J.P., and Huis, D., "A Laboratory Investigation
     of the Purging Behavior of Small-Diameter Monitoring
     Wells", Proceedings of the Third National Symposium on
     Aquifer Restoration and Ground-Water Monitoring. Niel-
     sen, D.M., Eds.,  National Water Well Association,  May,
     1983.
                             1-449

-------
103    ANALYSIS  OF POLYCHLORINATED BIPHENYLS IN WATER  AND  STACK
        EMISSIONS  BY  HIGH  RESOLUTION  GAS  CHROMATOGRAPHY/  HIGH
        RESOLUTION MASS SPECTROMETRY.


        Edwin A Marti,  Hani S Karam, Jakal Amin,  Triangle Laborato-
        ries,  Inc., 801-10 Capitola Dr., Research  Triangle  Park,
        North  Carolina  27713;  Timothy J Yagley,  Alan  F  Weston,
        Occidental Chemical Corporation, Niagara  Falls, New York.


        ABSTRACT

        Polychlorinated  biphenyls (PCBs) in environmental  samples
        are generally analyzed by high resolution gas  chromatogra-
        phy/low  resolution mass spectrometry (HRGC/LRMS)  or  high
        resolution gas  chromatography with electron capture  detec-
        tion  (HRGC/ECD).    Detection limits reported  using  these
        techniques for  water samples are on the order of 50-500 ppt
        for  the mono-deca PCBs (HRGC/LRMS) or 50-100 ppt for  Aro-
        clor characterization by GC/ECD.  HRGC/LRMS analysis of air
        samples  (collected  on  XAD-2)  typically  show  detection
        levels of 50 to 500 ng for the mono-deca  PCBs.

        High  resolution GC/high resolution MS (HRGC/HRMS) is  used
        routinely  for  the analysis of polychlorinated dioxins  and
        furans (PCDDs/PCDFs) in water and air samples, with  detec-
        tion  limits as low as 10 parts per quadrillion  (ppq)  for
        water and 50 picograms (pg) for air.

        This HRGC/HRMS  technique has recently been utilized for the
        analysis  of  PCBs in water and air samples and the  sample
        results indicate that the detection limits of these species
        are  at least two orders of magnitude lower  than  achieved
        using  the  low resolution  mass  spectrometric  technique.
        Using this technique, PCBs are reported as totals for  each
        congener  group  (mono-deca) as well as  congener  specific
        analysis for 11 congeners, seven of which are quantified by
        isotope dilution mass spectrometry.

        INTRODUCTION

        The  past few years have witnessed an increasing  need  for
        new methodologies that are capable of measuring very  small
        quantities  of  toxic substances in various matrices,  i.e.,
        low parts per trillion (ppt) for soils, parts per  quadril-
        lion  (ppq)  for water and picograms (pg) for  air  samples
        collected on solid absorbents.

        The proposed method determines polychlorinated biphenyls in
                                   II -450

-------
sample  extracts representing one liter of water  or  stack
emission  (air) samples collected on XAD-2.   The method  of
analysis  utilizes high resolution gas  chromatography/high
resolution mass spectrometry (HRGC/HRMS) operated at a  re
solving power of 8,000 to 10,000 in the selected ion  moni-
toring  (SIM) mode. This method was based on isotope  dilu-
tion  mass spectrometry during which nine 13Ci2  -  labeled
internal  standards were used to characterize and  quantify
all 209 PCB congeners.  By using published retention  times
of  the 209 congeners 1f five retention windows  bracketing
the  ten  congener groups could be monitored  to  determine
total PCBs by isomer groups (mono through deca) as well  as
specifically quantify eleven PCB congeners.

EXPERIMENTAL METHOD

For  stack  emission sampling of  stationary  sources,  the
XAD-2  resin was spiked with 10 ng of  surrogate  standards
prior  to sampling (Table 1).  Following the sampling  ses-
sion,  the samples (XAD-2, glass-fiber filter,  front  half
and  back half solvent rinses, impinger water and  impinger
rinses)  were returned to the laboratory.  The  front  half
and back half rinses were concentrated, then placed  inside
a Soxhlet extractor along with the rest of the solid  frac-
tions of the sampling train.  The sample was spiked with 10
ng  of PCB internal standards (Table 1), then  Soxhlet  ex-
tracted  with 750 mL of methylene chloride.    The  impinger
water  was spiked with 10 ng of alternate surrogate  stand-
ards (Table 1).  The water was then liquid-liquid extracted
in a separatory funnel using 3 X 60 mL methylene  chloride.
Both  the  impinger water extract and the  Soxhlet  extract
were  concentrated  then combined. ' The extract  was  split
50:50,  with  one-half being archived and  the  other  half
subjected to an acid/base wash cleanup.  The sample extract
was then concentrated to a final volume of 50 microliters.

For water samples, one liter of sample was spiked with  the
nine internal standards (in acetone) at 10 ng.  The  sample
was  allowed to equilibrate for one hour.  The  sample  was
then  extracted with 3 X 60 mL portions of methylene  chlo-
ride in a separatory funnel.  The extract was  concentrated
using  a  K-D apparatus and put through an  acid/base  wash
cleanup.   The  extract was then concentrated  to  a  final
volume  of  100 microliters.  Before analysis  of  the  PCB
extracts, 5 ng of recovery standards (Table 1) are added to
the extracts.

CALIBRATION

The mass spectrometer response was calibrated by using  the
                           11-451

-------
set of five initial calibration solutions  shown  in  Table
1.   Each solution was analyzed once and the analyte  rela-
tive response factors (RRF) were calculated.

An acceptable calibration must meet the following criteria:

  1) The percent relative standard deviations (RSD) for the
    mean  response factors from each of the unlabeled  ana-
    lytes  must be less than 25 or 30 percent depending  on
    the analyte (Table 2).

  2) The  signal-to-noise  ratio (S/N) for the  GC  signals
    present in every selected ion current profile must be >
    10:1.

  3) The ion abundance ratios must be within the  specified
    control limits (Table 3).

A continuing calibration was demonstrated every 12 hours by
injecting  one uL of solution number 2 from Table  1.   The
RRFs are calculated and compared to the mean RRFs  obtained
during  the initial calibration procedure.   An  acceptable
continuing  calibration run must meet the following  crite-
ria:

 1) The  measured  RRFs (for the unlabeled  PCBs)  obtained
    during the continuing calibration run must be within 25
    or 30 percent depending on the analyte (Table 2) of the
    mean values established during the initial calibration.

 2) The  ion-abundance ratios  must be within  the  allowed
    control limits listed in Table 3.

 3) The  signal-to-noise  ratio  (S/N) for  the  GC  signal
    present in every selected ion current profile must be >^
    10:1.

At  the beginning of every 12-hour shift during which  sam-
ples  are  analyzed the fused-silica  capillary  GC  column
performance was verified by injecting a 1-uL aliquot of the
PCB window defining mixture (Table 4).  This was  necessary
to  identify  the various retention time windows  for  each
group of analytes, which are grouped in five mass  descrip-
tors.   Figure 1 shows the tetra-PCB first and last  eluter
chromatogram  with  the  corresponding  tetra-PCB  internal
standard.

RESULTS AND DISCUSSION

In  order  to evaluate the analytical method's  ability  to
                           11-452

-------
detect and quantify small quantities of analyte present  in
water,  three sets of five samples at 0.5, 5.0 and  25  ppt
were  analyzed.  The results for the 0.5 ppt  matrix  spike
are  given in Table 5 and on Figures 2 and 3. The  MDL  and
LOD values were calculated for the lowest point (0.5  ng/L)
samples.  The  mean %Accuracy for the most  congeners  were
approximately equal to 100%. The mean %Accuracy ranged from
75.6%  for  2255-T-PCB  to 100%  for  2234455-Hp-PCB.   The
%Recovery  of the internal standards range from 44.62%  for
22455-Pe-PCB to 280.0%  for 334455-Hx-PCB. The high  recov-
ery  of  the  carbon labeled 334455-Hx-PCB was  due  to  an
interference problem. The %RPD range from 0.0% for 2234455-
Hp-PCB  to -13.6% for 223344556-Nona-PCB.  The  results  of
5.0  ng/L and 25.0 ng/L were also similar to 0.5 ng/L  sam-
ples.  For both 5.0 ng/L and 25.0 ng/L the internal  stand-
ards recoveries of 22455-Pe-PCB were the lowest and highest
for 334455-Hx-PCB.  For 5.0 ng/L mean %Accuracy range  from
66.60% for 2255-T-PCB to 119.20% for 244-Tr-PCB.  For  25.0
ng/L  the mean %Accuracy range from 2255-T-PCB for  111.12%
for 3344-T-PCB.  The %RPD ranged from -0.80% for  44-Di-PCB
to  -33.40% for 2255-T-PCB for 5.0 ng/L samples.  The  %RPD
range  from 2.56% for 244-Tr-PCB to -28.20% for  2255-T-PCB
of 25.0 ng/L samples.

In all three points, the ฑ3C12-22455-Pe-PCB gave the lowest
recovery.   One  possible explanation for the low  recovery
might  be the compound was not in the same mass  descriptor
as the recovery standard.  Another explanation is that  the
concentration of carbon-labeled standards are not  measured
using  isotope  dilution method.   When  the  corresponding
analyte, 22455-Pe-PCB was measured using the isotope  dilu-
tion method, the mean %Accuracies were 103.6%,110.20%,  and
93.84%  for  0.5  ng/L, 5.0 ng/L, and  25.0  ng/L  samples,
respectively.  The  concentrations  of  224455-Hx-PCB  were
computed using 13d2-(245)3-Hp-PCB with mean %Accuracies of
86.80%, 87.02%, and 78.79% for 0.5 ng/L, 5.0 ng/L, and 25.0
ng/L,   respectively.   This   was   done   because    when
13Ci2-334455-Hx-PCB was used to compute the analyte concen-
tration,  the results were erratic due to the high  percent
recovery  of the internal standard caused by  an  interfer-
ence.  The 13Ci2-224455-Hx-PCB was not used to compute  the
corresponding analyte concentration because  carbon-labeled
standard was used as the recovery standard.  In the future,
the  analyte,  224455-Hx-PCB, will be  computed  using  the
corresponding  internal standard (13C12-224455-Hx-PCB)  and
the T3Ci2-334455-Hx-PCB will be used as the recovery stand-
ard.


The  MDL values for 0.5 ng/L samples were calculated  using
                            11-453

-------
the formula:


           MDL = S * ttm_a.,  i_ซ _ 0.99)
                 where S = Standard Deviation
                       t = Student t.

For the present study, the MDL value for the 0.5 ppt  spike
was  0.072 ppt.  A more rigorous determination of  the  MDL
can  be determined spiking seven replicate samples at  0.05
ppt.   In  lieu of this, the Limit of Detection  (LOD)  can
still  be  calculated with this data set.   Using  So  (the
value of the standard deviation as concentration approaches
zero), the LOD (Limit of Detection) was computed using  the
formula:

             LOD = 3 * S0.

The LOD for the mono-PCB isomer was 0.036 ng/L.

The matrix spike evaluation for the stack emission  samples
is  currently  in progress.  The preliminary  results  show
recoveries  between 80 and 140% for the eleven  PCB  target
analytes with %RPDs ranging from 5 to 40% for three  matrix
spikes.

CONCLUSIONS

The  extraction, cleanup and analysis procedures  described
in this method for the trace analysis of mono through  deca
polychlorinated  biphenyls  in water are adequate  for  the
isolation  and measurement of individual PCBs to  detection
limits in the low ppq range.

The  Limit of Detection (LOD) was calculated to be  36  ppq
(parts per quadrillion) for the mono-PCB isomer.

The  overall  accuracy of the method, as  determined  by  a
series of five matrix spikes at three different  concentra-
tion  levels  (33 total measurements), was  94.5%  [ranging
from 66.6% for the 2255-tetrachlorobiphenyl to 120% for the
3344-tetrachlorobiphenyl both at the 5 ppt spike level].

The precision of the method, as calculated from the mean of
the  33  analyses,  was 6.4%  relative  standard  deviation
[ranging  from  2.49% for the  3344-tetrachlorobiphenyl  to
19.7% for 2255-tetrachlorobiphenyl both at the 5 ppt  spike
level].
                            11-454

-------
Table 1. Composition of the Initial Calibration Solutions
^ ^" ^"^ ™ป •• •ป •ป •ป •ป •ป •ป •• •ป •ป ^ Bป BB BB •ป *• BB BB BB BB BB BB BBBB BM BB BB BB BB BB ••ป BB Bป BB BB ^ ^ ^ ^ ^B BB BB BB BB BB BB BB BB BB BB •• •• •
Compound                      Concentrations (pg/uL)



Solution Number                         1    234
Un labeled Analytes
2 -Chlor obipheny 1
44 ' -Dichlorobiphenyl
244 ' -Trichlorobiphenyl
22 '55' -Tetrachlorobiphenyl
33 ' 44 ' -Tetrachlorobiphenyl
22 ' 455 ' -Pentachlorobiphenyl
22 ' 44 ' 55 ' -Hexachlorobiphenyl
22 ' 344 ' 55 ' -Heptachlorobiphenyl
22 ' 33 ' 44 ' 55 ' -Octachlorobiphenyl
22 ' 33 ' 44 ' 55 ' 6-Nonachlorobiphenyl
Decachlorobiphenyl

0.5
0.5
0.5
1.0
1.0
1.0
1.0
1.5
1.5
2.5
2.5

5
5
5
10
10
10
10
15
15
25
25

10
10
10
20
20
20
20
30
30
50
50

50
50
50
100
100
100
100
150
150
250
250

100
100
100
200
200
200
200
300
300
500
500
Internal Standards  (13Ci2)

4-Chlorobiphenyl(3y
44'-Dichlorobiphenyl
244'-Trichlorobiphenyl
33'4,4 '-Tetrachlorobiphenyl
22'455'-Pentachlorobiphenyl
22'44'55'-Hexachlorobiphenyl<4>
22'344'55'-Heptachlorobiphenyl
22'33'44'55'-Octachlorobiphenyl(2 >
Decachlorobiphenyl
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
Surrogate Standard  (13Ci2)

33'55'-Tetrachlorobiphenyl
33'44'5-Pentachlorobiphenyl
22'344'5'-Hexachlorobiphenyl
22'33'55'66'-Octachlorobiphenyl

Alternate Standard  (13C12)

2 2'3 3'4 4'-Hexachlorobiphenyl
100  100 100
100  100 100
100  100 100
100  100 100
100  100
100  100
100  100
100  100
100  100 100  100  100
Recovery Standards  (13Ci2)

22'55'-Tetrachlorobiphenyl
33'44'55'-Hexachlorobiphenyl{4
200  200 200  200  200
200  200 200  200  200
                             11-455

-------
The method evaluation of stack emission samples is current-
ly in progress with results expected at the time of  publi-
cation of this paper.

REFERENCES

(1)  Mullin,  M. D.f Pochini, C. M., McCrindle, S.,  Romkes,
M., Safe, S.  H. and L. M. Safe (1984). High Resolution  PCB
Analysis:  Synthesis and Chromatographic Properties of  All
209 PCB Congeners.  Environ. Sci. Technol., Vol. 18, No.  6,
p. 468-476.
                            11-456

-------
Table 1 (continued):

Notes:

1) Based on 100 uL final extract volume,this corresponds to a
   calibration Range from 50 pg to 10 ng for mono-PCB).

2) The labeled octa-PCB (Internal standard) is used to compute
   response factors of unlabeled nona-PCBs.

3) The Mono-chloro-Biphenyl internal standard is a 13C6 and not
      al3p
        *-12 •

4) The 22'44'55'-hexa-PCB (internal standard) and 33'44'55'-hexa-PCB
  (recovery  standard)  reflect the current  method.   The  original
  method validation had them switched.
                             1-457

-------
Table  2.   Initial  and  Continuing  Calibrations  Response  Factors
  Minimum Requirements
Compound
Relative Response Factors
 I-Cal        Con-Cal
 %RSD         %Delta
2-chlorobiphenyl
44 ' -dichlorobiphenyl
244 ' -trichlorobiphenyl
22 ' 55 ' -tetrachlorobiphenyl
33 ' 44 ' -tetrachlorobiphenyl
22 '455 ' -pentachlorobiphenyl
22 ' 44 ' 55 ' -hexachlorobiphenyl
22 ' 344 ' 55 ' -heptachlorobiphenyl
22 ' 33 ' 44 ' 55 ' -octachlorobiphenyl
22'33'44'55' 6-nonachlorobiphenyl
decachlorobiphenyl
30
25
25
30
25
30
30
25
25
30
25
30
25
25
30
25
30
30
25
25
30
25
13C6-4-Chlorobiphenyl
13Ci2-44'-Dichlorobiphenyl
13Ci2-244'-Trichlorobiphenyl
i3Cj.2-33'44'-Tetrachlorobiphenyl
X3Ci2-22'455'-Pentachlorobiphenyl
13C12-22'44'55'-Hexachlorobiphenyl
13CX2-22'344'55'-Heptachlorobiphenyl
13Cj.2-22 ' 33' 44' 55' -Octachlorobiphenyl
13Ci2-Decachlorobiphenyl
       30
       30
       30
       25
       30
       30
       30
       25
       30
                                                            30
                                                            30
                                                            30
                                                            25
                                                            30
                                                            30
                                                            30
                                                            25
                                                            30
13CX2-33'55'-Tetrachlorobiphenyl
X3Ci2-33'44'5-Pentachlorobiphenyl
13C12-22'344'5'-Hexachlorobiphenyl
i3Ci2-22'33'55'66'-Octachlorobiphenyl

13C12-22'33'44'-Hexachlorobiphenyl
                                        25
                                             25
                                             25
                                             25
                                             25
                 25
                       25
                       25
                       25
                       25
Notes:
1)  Isomers  that  have 25% criteria are
  X3Ci2-labeled standards.
    those  with  corresponding
2)  The labeled tetra-PCB (IS) will be in a different mass
    descriptor than the unlabeled tetra-PCB analyte.
                             11-458

-------
Table 3. Ion-Abundance Ratio Acceptable  Ranges
Number of
Halogen
Atoms
 Ion Type
Theoretical
   Ratio
           Control Limits
                              Lower
                   Upper
 1 Cl
 2 Cl
 3 Cl
 4 Cl
 5 Cl
 6 Cl
 7 Cl
 8 Cl
 9 Cl
10 Cl
  M/M+2
  M/M+2
  M/M+2
  M/M+2
  M/M+2
M+2/M+4
M+2/M+4
M+2/M+4
M+2/M+4
M+4/M+6
   3,
   1,
   1,
   0,
   0,
   1,
   1
08
54
03
77
61
24
04
   0.89
   0.78
   1.18
2.62
1.31
0.87
0.65
0,
1.
  52
 ,05
0.88
0.76
0.66
1.00
3.54
1.77
1.18
0.89
0.70
1.43
1.20
1.02
0.90
1.36
                             11-459

-------
Table 4. PCB Window Defining Mix
PCS isomer                      Group Number
2-chloro (F)                             1
4-Chloro (L)                             1
2,6-dichloro (F)                         2
4,4'-dichloro (L)                        2
2,4,6-trichloro (F)                      2
2,3,5-trichloro                          2
2,2',6,6'-tetrachloro (F)                2
3,4,4'-trichloro (L)                     3
2,3,3',4-tetrachloro                     3
3,3'4,4'-tetrachloro (L)                 3
2,2',4,6,6'-pentachloro (F)              3
2,2',4,4',6/6'-hexachloro (F)            3
3,3',4,4',5-pentachloro (L)              4
2,2',3,4,4',6-hexachloro                 4
2,2',3,4,5,6'-hexachloro                 4
3,3',4,4',5,5'-hexachloro (L)            4
2,2',3,4',5,6,6'-heptachloro (F)         4
2,2',3,3',4,4',5-heptachloro             4
2,2',3,3',5,5',6,6'-octachloro  (F)       4
2,3,3',4,4',5/5'-heptachloro (L)         5
2,2',3,3',4,4',5,5'-octachloro  JL)       5
2,2' ,3,3',4,5,5',6,6'-nonachloro  (F)     5
2,2',3,3',4/4',5,5',6-nonachloro  (L)     5
decachloro                               5
Note:  The table contains the order of elution  for  specific
       isomers.
                            11-460

-------
Table 5   Matrix Spike Replicate Analytical  Results
                          (0.50 ppt Spike)

Analytes
2 -Mo
44-Di
244-Tr
2255-T
3344-T
22455-Pe
224455-Hx*
2234455-Hp
22334455-Oc
223344556-No
Dec a
Cj.2
Internal
Standards
4 -Mo
44-Di
244-Tr
3344-T
22455-Pe
224455-Hx
2234455-Hp
22334455-Oc
Dec a
MSI

0.46
0.54
0.55
0.72
1.20
1.10
0.86
1.50
1.70
2.10
2.60


MS2
<
0.44
0.53
0.53
0.77
1.00
1.00
0.88
1.40
1.50
2.20
2.50


MS 3
[ Concentration
0.45
0.57
0.53
0.73
1.00
0.98
0.88
1.50
1.60
2.10
2.60


MS4
in ppt)
0.48
0.59
0.53
0.77
1.20
1.00
0.89
1.60
1.70
2.20
2.80


MS 5

0.49
0.57
0.59
0.79
1.20
1.10
0.83
1.50
1.70
2.20
2.70


Mean

0.46
0.56
0.55
0.76
1.12
1.04
0.87
1.50
1.64
2.16
2.64


%Recoveries
68.5
140.0
90.1
116.0
56.6
301.0
98.4
122.0
90.3
78.0
149.0
105.0
125.0
52.8
211.0
103.0
114.0
92.3
78.1
174.0
112.0
123.0
42.8
228.0
100.0
112.0
88.0
73.2
161.0
105.0
115.0
34.3
246.0
101.0
116.0
90.0
85.3
157.0
107.0
124.0
36.6
414.0
107.0
131.0
89.1
76.6
156.2
103.8
120.6
44.6
280.0
101.9
119.0
89.9
*13Ci2-(245)3-Hp was used as the  internal standard.
                             1-461

-------
                  riiป
                                    >•• ilf.BZ*j r:j.2ff.fS23 T-.f
                                     L*ปOK*TOXJfS. INC. : :ซ HXS rCB:
100
 P5j
 to.
 •sj
 •0:
 75:
 70.
 <5.
 CO.
 55.
 50.
 45.
 40.
 35.
 30.
 25:
 20.
 ซ
 10
   5
   0
      22:00    23:00    24:00     25:00    26:00     27:00    21:00     29:00    30:00     31:
rilป:*H2125 Aey:2f-tOO(-fl  07:12:33 tUft 301.9626 F:3
Sปmplป Tuct:SANPXC riiป TซJrt:ntZANCLE IABORATOXZES, INC.: :KI RES PCS; :
                         24 10
            r-r-fcf
          <—
               23:03
                     23:41
                                                               L 3.3'.4,4'-T-PCS " 0.01 ppt
                              t
 P5j
 *0j
 ป5J
 •OJ
 75J
 70J
 C5.
 ซ0
 55.
 50j
 45.
 40:
 3S
 30
 25
  20.
  153
  10.
                                Cซrbon-I*b.iaปd-2, 2' 5, S' -T-PCB
                                                                                    30:10
                                                 Cซrlx>ij-I*b.Jl.<(-3, 3', 4, 4'-T-PCB
       22:00
                23:00
                         24:00
                                   25:00
                                            26:00
                                                      27:00
                                                               28:00
                                                                         29:00
                                                                                  30:00
00  XTNC

 S.9E7
.S.SE7
.I.OC7
.7.6E7
.7.1E7
.6.7E7
.6.3C7
L5.8E7
 5.4E7
 4.9E7
 4.SE7
.4.0E7
.3. 6E7
.3.1C7
.2.7E7
.2.2E7
.1.BE7
 .1.3C7
 .t.PEC
 .4.see
                                                                                                O.OZO
                                                                                            31: 00  TIH.
                                           Figure  1
                                                 11-462

-------
 PCB HIGH RESOLUTION GC/MS ANALYSIS
    WATER MATRIX SPIKE REPLICATE RESULTS
Q.
CL
o
LLJ
O
z
o
o

     ACTUAL SPIKE
MEAN
%RPD
: %RSD (N = 5)
                                   0

                Figure 2

-------
 PCB HIGH RESOLUTION GC/MS ANALYSIS  I
     WATER MATRIX SPIKE REPLICATE RESULTS
HI ACTUAL SPIKE
f///A

MEAN
'/'#

%RPD
III!
%RSD
CL
Q.
o
\-
2
LLJ
O
Z
O
o
10 b
  LLJ
  O
  DC
  LU
  DL
                 Figure 3

-------
      CONTINUOUS ANALYSIS  OF  VOCs  IN AIR USING A NEW,
        PHENYL-METHYL SILICONE  STATIONARY PHASE FOR
                HIGH-RESOLUTION CAPILLARY GC

Rene P. M. Dooper and Nico Vonk. Chrompack Inc.,  1130 Route
202 South, Raritan, New Jersey 08869.   Henk J.  Th.  Bloemen,
RIVM, P. O. Box 1, Bilthoven, The Netherlands

INTRODUCTION

Monitoring (very) volatile organic compounds in outdoor air
becomes  more and  more important, as  these compounds  are
involved in  smog formation and are known ozone precursors.
Also  the  US  1990 Clean air  Act clearly  indicates  the need
for accurate data.  In air, volatile organic components are
numerous  and often very  similar  to each  other.    In most
cases  they  occur   in  the   gaseous   phase   at  ambient
temperature at the (sub) part-per-billion level.   For these
reasons, chromatography is the most suitable method.  These
considerations  and  the  specifications  for  a  monitoring
system have formed the basis for the design of a monitoring
system  for volatile  organic  compounds  in air,  described in
this  paper—the  VOC  Air Analyzer  (Chrompack International,
Middelburg, The Netherlands) .

INSTRUMENTAL

The VOC Air  Analyzer is a system for unattended continuous
automatic analysis of air containing  (very) low  levels of
organic components.   The VOC Air  Analyzer takes samples at
regular  intervals  over  a  selected  time.    The  volatile
organic  components are concentrated on  an  adsorbent tube.
When  sampling  has been  completed the  adsorbent  tube  is
heated  to  release the components and  transport these to a
liquid  nitrogen-cooled  fused silica  trap.   Here the sample
components are  refocussed in  a narrow band.  The  trap is
then  flash  heated by  which  the sample is  introduced into
the capillary column for analysis.

A schematical presentation of the VOC Air Analyzer is given
in figure 1.  It has been designed to simultaneously sample
and   analyze  air  using  (cryo)   adsorption  and  thermal
desorption techniques.   Control of the heating and cooling
devices,  valves,  and  sampling   pump,   as  well  as  the
synchronization  with  the  gas  chromatograph   and  the
integrator-data  processor,  is  performed  by  the  VOCAA-
controller.

In  the sample  collection mode,  air  is drawn through  the
valve  and the  adsorption  trap by means  of a  sample pump
with  a flow  ranging from 10 to 70 mL/min.   The adsorption
trap  filled  with  appropriate  trapping  materials  such  as
Tenax,  Carbosieve, or  Carbotrap,  or  a  composition of these
materials, is cooled using liquid nitrogen to a temperature
ranging from -20ฐC to ambient.
                            11-465

-------
Just before  the end  of a  sampling period,  the  capillary
trap is cooled to a temperature with the range of -180ฐC to
subambient (cooling rate 100ฐC/min).

In  the sample  transfer mode,  the  trapped  compounds  are
transferred from the adsorption  trap to the capillary trap
using  a  reversed  carrier  gas  flow.   The  capillary trap
consists of a wide  bore fused silica capillary coated with
liquid phase and/or filled with adsorbent.  The transfer is
induced  by  heating  the  adsorption  trap   (heating  rate
190ฐC/min)  and  switching of the valve.   Only in  this mode
are the sampler and the injector connected.   While in this
mode, the temperature  of the capillary  trap is maintained
at  low temperature set during the  precool.   To minimize
discrimination of  the  higher boiling  compounds and sample
transfer time,  the transfer  flow is higher  than  the flow
defined in the  restriction of the  capillary column.   This
is  achieved  by  opening the  desorption  vent.    After  the
completion of the transfer, the valve is switched again.

To  remove  any remaining compounds  the  adsorption  trap is
heated for a short  period  to  a temperature higher than the
one in the sample transfer  mode  and again using a reversed
carrier  gas  flow.     Before   sampling  is   restarted,  the
adsorption  trap  is cooled  to  the  desired  temperature.
Analysis  time  is   optimized  to  allow  separation  of  the
components of interest,  cool  down to,  and equilibration at
the initial temperature  setting  before the sampling period
is over.   In this way a continuous,  unattended operation is
possible,  based on a one hour cycle time.

It  is  necessary  for   most  air samplers   to  remove  the
moisture  in  the   sample stream,  as  it might block  the
adsorption tube or the cold trap with ice.  Moisture can be
selectively removed on-line  by  passing the  sample stream
through a  two-stage  dryer  [Nafion   tubes, DuPont  Corp.
(Wilmington,  Delaware),  see  the  right  part of  Figure 1]
reducing the dew  point to -55ฐC.   This  two-stage dryer is
self-regenerating.     The  drying  force  is  the  moisture
gradient generated  by  the  underpressure (0.1 atm absolute)
in  the first stage.   The  dry  air  stream  is used  in  the
second stage  to  dry   the  sample stream.    If  dry air is
available,  then a  single  Perma-Pure  (Nafion  tube)  dryer
system can be  used, which also removes  the  water from the
sample stream through  the  semi-permeable wall of  the dryer
tube.   The dry air stream is  typically 4-5  times higher
than  the   air   sample  stream to get  the   desired  drying
effect.   The  complete  set  of  dryers,  pump,  valves,  and
Nafion tubes  is built  in  a new dryer/pump  unit,  which is
controlled  by  the   control  unit   of  the  VOCAA.    A
disadvantage  of the   Nafion  dryer   is  that it  partially
eliminate polar  components which might be  present  in  the
air  samples.    If  these  types  of  compounds  have  to be
analyzed,   an  option  is built  in  to  collect  the  samples
without the dryer in line in the sample stream, after which
the  adsorption  tube is heated  from -20ฐC up to  +10ฐC  and
                            11-466

-------
pre-flushed with  carrier  gas  (helium)  to remove the water.
In  this  way,  however,  a  substantial  part  of the  C2-C4
hydrocarbons  in also  flushed of  the  adsorption tube  and
lost  for analysis.   For  the separation both  thick  film,
apolar  phases  capillary columns  (such  as CP-Sil 5  CB,  50 m
x 0.32  mm i.d.  df  = 5.0  /urn)  and  A1203/KC1 PLOT columns can
be  used.   Dual detection (FID  and ECD) should  be  used by
splitting the  effluent from  the analytical column to  the
two detectors.  In order  to cover the  whole range  from C2-
C12 hydrocarbons, the apolar column must start at -20ฐC and
programmed up  to  210ฐC.    Monitoring compounds  in  relation
to  the biospheric  ozone  formation,  such as  aliphatic  and
olefinic  hydrocarbons  as  well   as  the  alkyl  aromatics,
required  a chromatographic column with a high resolution of
the  very volatile organic  compounds,  such  a  ethene  and
ethane.   For  this purpose,  the A1203/KC1 PLOT column is
selected.   Temperature  program  then   can  start at  40ฐC,
which  eliminates  the use of a  cryogenic  unit  in  the  gas
chromatograph.  On  this column, however, it is difficult to
analyze   some  of  the halogenated hydrocarbons,  such  as
unsaturated freons.

The above-described analyzer  is  used  by the Dutch National
Air Quality  Monitoring Network  and by the EPA during  the
Summer  1990 ozone precursor study in Atlanta, Georgia.   The
VOC  Air Analyzer  can  be   used   where   (very)   low
concentrations  of  volatile organic components  in  air  have
to be monitored continuously without operators present.

Th range  of components that can be analyzed is:
1.   Hydrocarbons C2-C^g
2.   Halogenated hydrocarbons up to trichlorobenzene
3.   Aromatic hydrocarbons up to trimethyl benzene

The application field is  air pollution control and in  some
cases   industrial   hygiene  (especially  where  levels   to
control are in the  ppb range).

Table  1.  shows  the  composition of   an  EPA  calibration
mixture,  which can be separated  on   a 5  /im  CP-Sil  5  CB
column  under the  above described conditions.   For reasons
of  sensitivity  and  identity-conformation  dual  detection
should be used.

RESULTS AND DISCUSSION

Figures 2A and  2B  show the results of  a 200 mL outdoor air
sample  at Raritan, New Jersey,  being  the FID  and  the  ECD
signals.   Here the apolar column program was  started at
40ฐC and  programmed up to 200ฐC.   In Figure 3  the  increase
in  retention  and  separation  is  shown, using  an A12O3/KC1
PLOT   column   under  the   same  conditions.    Typical
concentrations  of  the  components are in  the  0.1-10  ppb
range.    Under  the  conditions   described  in  Figure  3,  a
continuous monitoring of  air was realized during  several
months.   Part  of  the quantitative results are  plotted in
                            11-467

-------
Figure 4.  Similar plots are made  for C4,  C5,  C6, C7-8 and
aromatic components.

A  new method  for  the analysis  of C2-C6  hydrocarbons and
halogenated compounds,  including C5-C12   hydrocarbons,  is
to  inject  and  separate  the collected  air samples  on two
capillary columns simultaneously.  Such a combination could
be  an A12O3/KC1 PLOT  column and  an apolar  liquid phase,
such as CP-Sil 5 CB (thick film).  In this way the analysis
can start at 40ฐC-45ฐC, so cryogenic cooling of the GC oven
is  avoided,  which  gives  a substantial  reduction  in the
liquid  nitrogen consumption.    In order  to  optimize the
separation of  the chlorinated  compounds  a new,  slightly
more polar liquid stationary phase was developed, CP-Sil 13
CB.   This  is  also  a  polysiloxane  phase,  containing  an
average  of  14% phenyl/86%  methyl groups  in  the polymer.
CP-Sil   13   CB  has  an  excellent   selectivity  for  the
halogenated hydrocarbons, as mentioned  in  EPA 624 and 502-
2.   By  combining  FID and  ECD  detection,  some  co-eluting
peaks  can  be  quantified  independently.    The  stationary
phase does not contain electro-negative groups.   Combined
with the low bleeding of the column  this  results in a very
stable baseline on the ECD trade.

CONCLUSION

The most significant  characteristic of the VOC Air Analyzer
is  the  simultaneous  sampling  and  analysis  of  (very)
volatile organic  compounds  at  a  frequency of  1  hour  or
less.   Using  thick film WCOT  or porous PLOT  columns, the
compounds that can be monitored  range from the unsaturated
and   saturated  alkanes,   benzene,   and   the  substituted
aromatics and  various  halogenated  compounds.   The  high
resolution power of  capillary  columns allows  high quality
identification  and  quantitation and produces  information
concerning individual  compounds  relevant  in  atmospheric
processes.
                           11-468

-------
FIGURE CAPTURES.
Figure 1.


Figure 2A.
Figure 2B.
Schematic  diagram of  the Chrompack  VOC  Air
Analyzer, including the new pump-dryer unit.

FID  signal  of a  200 mL  outdoor air  sample,
collected  at Raritan,  New Jersey.    Column:
CP-Sil 5  CB.   Temperature:   40ฐC (4 min)  to
200ฐC.     Sample   collection:     -20ฐC.
Desorption:  220ฐC.  Peak identification:
1 = benzene; 2 = toluene;  3 =  ethylbenzene;
4 = p.m-xylene;  5 = 0-xylene.
                  200 mL  outdoor air
                   in   Figure   2A.
sample.
  Peak
Figure 3.

Figure 4.
  BCD signal  of  a
  Conditions  as
  identification:
  1 = trichlorofluoromethane; 2 = methylbromide;
  3 = trichloroethane;  4 = carbontetrachloride;
  5 = trichloroethene;  6 = tetrachloroethene;
  7 = tetrachloroethane; 8 = hexachlorobutadiene

(see original)

  Plot of  the C2-C3  concentration fluctuation
  during  a continuous monitoring of outdoor air
  at Bilthoven,  The  Netherlands.   Sample cycle
  time was one hour,  sample volume:  333 mL.
                            11-469

-------
                                                                                          SAMPLE PUMP
                                                                                           GD—ป
  iAHPLEl OUT
 SAMPLE  COLLLCTtON,*
4 DRYER AIR IN
                                                                                                       DRYER AIR OUT
VENT
                                                                                           PUMP, CAL.SAMPLE,
                                                                                           BRYER OPTION
     CALIBRATION
     SAMPLE
     CYLINDER
               COLUMN*

-------
                               60 COMPONENT CALIBRATION MIXTURE
 (1)   Acetylene
  2)   Ethylene
  3)   Ethane
  4)   Propylene
 (5)   Propane
 (6)   Isobutane
 (7)   1-Butene
 (8)   n-Butane
 (9)   trans-2-Butene
(10   cis-2-Butene
(11   3-Methyl-1-Butene
(12   Isopentane
(13)   1-Pentene
(14)   n-Pentane
(15)   Isoprene
(16)   trans-2-Pentene
(17)   cis-2-Pentene
(18)   2-Methyl-2-butene
 19)   2,2-Dimethylbutane
 20)   Cyclopentene
 21)   4-Methyl-l-Pentene
(22)   Cyclopentane
(23)   2,3-Dimethylbutane
      2-Methylpentane
      3-Methylpentane
      2-Methyl-1-pentene
      n-Hexane
      Chloroform
      trans-2-Hexene
      cis-2-Hexene
(31)   Methylcyclopentane
 32)   2,4-Dimethylpentane
 33)   1,1,1-Trichloroethane
 34)   Benzene
 35   Carbon tetrachloride
 36   Cyclohexane
 37   2-Methylhexane
 38)   2,3-Dimethylpentane
(39)   3-Methylhexane
(40)   Trichloroethylene
      2,2,4-Trimethy1pentane
      n-Heptane
      Methylcyclohexane
      2,3,4-Trimethy1pentane
      Toluene
(46)   2-Methylheptane
(47)   3-Methylheptane
(48)   n-Octane
(49)   Perchloroethylene
(50)   Ethyl benzene
(51)   p-Xylene
 '52)   Styrene
 53)   o-Xylene
 54)  n-Nonane
 (55)  Isopropylbenzene
 (56)  n-Propylbenzene
 (57)  a-Pinene
 (58   1,3,5-Trimethylbenzene
 (59   1,2,4-Trimethy1 benzene
 (60   /J-Pinene
                                   1-471

-------
ro

E
\
z
QL
h-
Z
u
o
z
O
O
            VERY  VOLATILE  ORGANIC  COMPOUNDS
                                C2-C3
            ethane
ethene
                        HOURS PAST 00:00 MAY 29 1990

                                       3
O  propane
propene
propadiene

-------
*
*
* 2A
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* * ^
* * <*•
* *
* * 'T1
* * o
* *
* * in
* * T-ซ X
* * ..
* * o
* * M ^ ^
* * .. ,Tl O
V 5*C O 'Tป jT|
* * -r-l _| .
* * ! ฃti
* * O M
* * "••>• ro  U. U.
























































ID


mm

P
O

4^
f^
3
C

•*J

-------
*
*
*
*
*2B
*
*
*
*
*
*
*
*
*
*
*
*
*
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *

* *

jjc sje







































1-1
Tf
••
iT)
6

(\i
C"D v.
••
ro
c\i _t ,-1
•• <7* O
O* 'T1 ฃCf
1-1 i-t >
1 Lฃ
O ^
^ i a
^ o
i-i ni ••
'T>
1 ซ ซ C
O O
CO 01 0) "i
i 4* e 4*
^"* fl3 fl3 ^
O TJ C 3
C
.. Q| QJ (U
i— I i— 1 4*
ai -r-i ---I .p
-P U. U. 
-------
           13
       10
      9
LjJl
11 12
uu
            14
                     Figure 3 Chromatogram of an air sample. Sample: Outdoor air at Bilthoven
                     Station, June 22,1990, at 15.00 hr. Column: fused silica 25m x 0.53mm
                     i.d. AI2O3/KCI, df =  10 \JJTI. Temperature: 40 ฐC, isothermal 1 min, pro-
                     grammed to200 "Cat 10ฐC/min, 200 ฐC, isothermal30min. Carrier: He
                     0.4 bar. Detector: FID 1, 10E-12 A, 2 mV full scale, 275 ฐC. Adsorption
                     trap: Carbosieve SHI, Carbotrap, Carbotrap C (7.5 x 0.29 cm). Capillary
                     trap: Poraplot U fused silica 0.53 mm i.d., 20 \im. Sample coll.: -20 ฐC,
                     35 min.  Sample flow: 9.5 mL/min. Sample vol: 333 mL Sample desor.:
                     250 ฐC,  5 min, desorption vent flow 2.5 mL/min. Cryofocusing: -150 ฐC,
                     5 min. Injection: 125 ฐC, 10 min. Back flushing: 270  ฐC, 10 min, flow: 20
                     mUmin. Valve: 200 ฐC. Injection block: 200 ฐC. Peak identification: 1 =
                     ethane;  2 = ethene; 3 = propane; 4 = propene; 5  = 1-butane; 6  =
                     prodadiene; 7 = n-butane; 8 = trans-2-butene; 9 = 1-butene; 10 =
                     1-butene; 11 = cis-2-butene; 12 = cyclopentane; 13= 1-pentane; 14 =
                     n-pentane;  ISA  = 3-methyl-1-butene; 15B = trans-2-pentene; 15C =
                     2-methyl-2-butene; 15D =  1-pentene; 15E= 2-methyl-1 -butene; 15F =
                     cis-2-pentene; 21A  = methylcyclopentane; 21B = cyclohexane; 21C =
                     2-methylpentane; 21D = 3-methylpentane; 25 = n-hexane; 26  = n-hep-
                     tane; 27 =  benzene; 28  = n-octane; 29 =  toluene.
                                                 29
                              27
                                                                          27min.

-------
GENERAL

-------
•J05             DEVELOPING A UNIFORM APPROACH FOR COMPLYING WITH EPA METHODS

         Peggy Sleevi.  Corporate Director of Quality Assurance,  Enseco,  Incorporated,
         2612 Olde Stone Road, Midlothian, Virginia 23113;  Deborah Loring,  Director of
         Quality Assurance, Enseco-East,  2200 Cottontail  Lane,  Somerset,  New Jersey
         08873;  Jerry Parr, Chief Organic Scientist, Enseco Incorporated,  4955 Yarrow
         Street, Arvada, Colorado 80002;  Nancy Rothman,  Chief Scientist,  Enseco
         Incorporated,  205 Alewife Brook Parkway, Cambridge, Massachusetts 012138

         ABSTRACT

         Since the late 1970's, EPA has developed methods using GC and GC/MS technology
         to support regulatory initiatives.  These methods have been promulgated,
         distributed, and used as contract mechanisms.  Commercial laboratories have
         faced a bewildering array of "approved" methods, generally utilizing identical
         technology but varying in detail.

         As stated in a recent EPA report to the U.S. Congress "Improved coordination
         is needed in the Agency's methods development program", Enseco supports the
         activities of the Environmental Methods Management Committee, created to
         respond to EPA's recommendation and has drafted an approach which results in
         regulatory control combined with analytical flexibility.  The approach
         controls critical method elements such as the procedural details, calibration,
         and quality control requirements but eliminates superficial differences that
         currently exist in EPA methods.

         Using as a model the methods available to analyze volatile organics by purge
         and trap GC/MS (624, 524.2, 8240, 8260, etc.),  information is presented
         comparing and contrasting the differences between EPA methods from various
         sources.  Data will be presented discussing the impact of varying the method
         details.  Finally, an approach will be presented which discusses how
         laboratories can balance productivity, technical enhancement and method
         compliance  issues.

         INTRODUCTION

         Since the late 1970's, EPA has developed methods using GC and GC/MS technology
         to support  regulatory initiatives.  These methods have been promulgated,
         distributed, and used as contract mechanisms.   Commercial laboratories have
         faced a bewildering array of "approved" methods, generally utilizing identical
         technology  but varying in detail.

         In response to a recent EPA report  to the U.S.  Congress  (Adequacy,
         Availability and Comparability of Testing Procedures for the Analysis of
         Pollutants  Established under section 304(h) of  the Federal Water Pollution
         Control Act, also referred to as the 518 Report), EPA has formulated the
         Environmental Methods Management Committee  (EMMC) to address the issue of
         methods consolidation.  The Committee's efforts have been described to reduce
         the number  of method variations  labs must integrate and to allow more methods
         to be developed.  The methods integration group has stipulated that quality
         control is  an intrinsic part of  the methods  (1).  The approach to
         consolidation must eliminate superficial differences which result in
         laboratories needing to run duplicative methods with differing requirements
                                           11-479

-------
which do not impact the quality of the data.  Since these methods are
performed in response to regulatory requirements, it is incumbent upon
laboratories to meet the specific requirements of each method, irregardless of
the technical merit.  This results in redundancy of effort and increased
analytical costs to ensure that specific method details are met.

As a solution to this dilemma,  EPA has indicated that the various methods
utilizing the same technological approach will be combined into one master
method by using the "best practices" from each method.  We are concerned that
this approach will result in methods which have such stringent criteria as to
be virtually unusable.  Furthermore, the methods will not have the flexibility
to meet the Agency's various regulatory needs.

This "best practices" approach  is also contrary to an earlier statement by EPA
in Environmental Lab.  As stated by David Friedman:

         "The approach we have  been taking when promulgating
         analytical methods often has been counter productive.  It
         has stifled creativity; it has led to poor analytical
         results;  and it has, in some cases raised the cost of
         testing ... We have  to move toward performance
         standards, not design  standards.  We must specify what
         needs to be done, including data quality objectives, not
         how to do it." (2)

We are presenting here an alternative approach to EPAs recommendation to
use the "best practice" from each method to get one method.  We propose a
minimum acceptable practice (MAP) to be used in concert with Data Quality
Objectives (DQOs)  to define the analytical requirements for each project.
The DQOs and analytical requirements must be documented in a project
specific QAPjP which is agreed  to as part of the project planning process.

In addressing this problem we have evaluated what aspects of a method are
critical to the execution of the method.  Minimum QC criteria should be
specified outside of the method as proposed for SW-846 (55 Federal Register
4440).  The ASTM document "Standard Practice for Generation of
Environmental Data Related to Waste Management Activities:  QA/QC Planning
and Implementation" addresses the minimum acceptable practices to assure
the quality of field and analytical activities.  Types of control samples
that are used to monitor method performance are described externally to the
methods and apply universally to all techniques amenable to such external
controls.  Additional use of matrix-specific QC must be related to the
project needs based on the DQOs and not be mandated as a laboratory
exercise.  The analyte list and QC sample elements are therefore not
mandatory elements of the method.  Calibration criteria, sample size and
preparation procedures are, however, inherent method elements.  The
specifics of these method elements may be variable and subject to
validation.
                                 11-480

-------
Supporting information with respect to analytes which may be determined by
a method, in addition to any performance criteria obtained under specific
conditions should be included as appendices to the method.

Such an approach will improve the quality and usability of data while
providing a cost-effective means to generate environmental data.  It will
also allow for technical enhancements to a method to allow for innovation
and improvement.  Under the current system you can follow the method
exactly and obtain poor quality, unusable data.  If you deviate from the
method to improve the quality and usability of the data you may be guilty
of non-compliance.  We must ensure data integrity by requiring those
elements that are critical for integrity, while recognizing those elements
that are not critical.

Each laboratory must be required to demonstrate method proficiency based on
the specifics of the method as performed in that laboratory.  This will
allow for differences in GC columns, temperature programs, and target
analytes to suit the project requirements, and assure the performance of
quality control as applied to that method.  Each time a change is made to
how the method is executed or if additional analytes are to be included a
rigorous validation procedure must be performed.  A proposal for initial
demonstration of proficiency and validation is described in this paper.

In this paper we address the applicability of this approach to the methods
for the analysis of volatiles by GC/MS, specifically Methods 624, 524.2,
8240 and 8260.  The procedural differences in these methods have been
adequately described elsewhere and were not the focus of our efforts (3).

MINIMUM ACCEPTABLE PRACTICES

An analytical procedure should provide enough detail to allow an
experienced laboratory unfamiliar with the procedure to generate equivalent
data.  Thus, extensive procedural details are required to be written into
the method.  Examples of this level of detail include concentration of
calibration standards, mass range, extraction solvent, sample size,
internal standards, usable method performance data, etc.  However, very
few, if any of these details should be mandatory.  Rather, as discussed in
more detail later, this descriptive information provides the basis for the
performance data presented in the method.  Other techniques can be used
provided they result in equivalent or better performance.

This section presents two examples of the problems with the current
approach and presents a proposed solution.

The first example relates to the retention time window used for identifying
compounds by GC/MS.  The following "requirements" were found:

Method 524:        "11.2.1 The GC retention time of the sample component
(1983)             must be within t s of the time observed for that same
                   compound when a calibration solution was analyzed.
                   Calculate the value of t with the equation:

                   t = (RT)l/3
                                  11-481

-------
                   where RT =  observed retention time (in seconds)  of the
                   compound when a calibration  solution was analyzed." (4)

Method 524.2:       "11.1.1  The GC retention time of the sample component
(Rev. 2.0,  1986)*  must be  within 10 s of the time observed for that for
                   that same compound when a calibration solution was
                   analyzed."  (5)

* Note: As  a comment on the current disarray of methods, we found four
versions of Method 524.2, two  of which on the surface were stated to be the
same (Revision 2.0) and are required in the regulations, but are
substantially different. Currently both the 1986 and 1988 versions are
promulgated (40CFR141.24).   However, the 1988 revision is required for
compounds 9 through 18 and  the 1986 version required for the remaining
compounds (56 Federal  Register 3526).  Thus, the entire volatile analyte
list for drinking water must be performed using two distinct versions of
the method  with different requirements.  Furthermore, Revision 3.0,
available in the public domain, is not an approved method.

Method 524.2:       "11.6 The GC retention time  of the sample component
(Rev. 2.0,  1988)   should be within three standard deviations of the mean
                   retention time of the compound in the calibration
                   mixture." (6)

Method 624:        "12.1.2  The retention time must fall within + 30s of the
                   retention time of the authentic compound." (7)

Method 8240:       "7.5.1.1.1  The sample component RRT must compare within
                   + 0.06RRT units of the RRT of the standard component.
                   For reference, the standard  must be run within the same
                   12 hours as the sample." (8)

As shown above, our review  found five different "requirements" for
establishing this retention time window.  Complete and full method
compliance  would require that  laboratories assure that the "correct"
approach was used based on  the method purportedly used.  Since the
retention time window is a  minor component of identification in GC/MS, we
believe that these "requirements" are generally ignored.  As an interesting
exercise, we evaluated the  data from a volatile calibration standard
processed using current GC/MS  target compound software.  We believe that
the system  would correctly  identify the compounds using any of the
definitions.

A more appropriate wording  of  this section would be:

         Identification of  target compounds is  based upon both
         retention time and mass spectral agreement.  The data
         contained in this  method were based on a + 0.06 RRT
         window.  Other approaches may be used  if they provide
         equivalent performance.

As another  example, consider the language in Section 7.2.3 of Method 8240
which states "Prior to use, condition the trap  daily for 10 minutes while
                                 11-482

-------
backflushing at 180ฐC with the column at 220ฐC."  In this one sentence,
there are six requirements, "Prior to use", "daily", "10 minutes",
"backflushing", "180ฐC", and "220ฐC".  Accordingly,  if someone was  to
develop a different trap which could be conditioned at 170ฐC for 8  minutes,
it could not be used.  Again, we would rewrite this sentence to read:

         The trap must be conditioned to maintain performance.
         It is recommended that the trap be backflushed at 180ฐC
         for 10 minutes.  Other approaches can be used provided
         the trap maintains performance as measured by analyses
         of QC samples.

In summary, there are many imperatives contained in methods.  We should
always be questioning the intent of the imperative and verifying that it
is a requirement.  Is there no alternative?  Do we really mean "must"?
Or, would we allow a deviation?  If the "requirement" is merely a
description of what was done during method development/implementation,
then it should not be a requirement.  We do believe that the imperative
writing style adds clarity.  For example, "Purge 5 mis" is much clearer
than "5 mis is purged".  However, these imperative statements should be
followed by language that indicates other approaches could be performed.

A review of EPA methods for measuring volatile organics by purge and trap
GC/MS indicate that there are thousands of requirements.  For example, we
found 28 requirements in Section 7.3 of Method 8240 relating to daily
calibration.  This is in addition to the other 50 or so calibration
requirements in Method 8000 and the requirement for initial calibration  in
section 7.2.  We believe these requirements could be reduced to a few
critical elements, minimum acceptable practices, and all others reworded
to indicate the conditions used.  For example, for calibration by GC/MS  we
believe the key requirements are:

    1.   The mass spectrometer must generate reliable mass spectra, as
         demonstrated by the measurement of a reference mass marker
         compound such as bromofluorobenzene.

    2.   A predictable relationship between response and concentration must
         be established.  This calibration response must be used to define
         the upper and lower limits of quantitation.

    3.   The calibration must be shown to be in control during instrument
         operation.

How then do we assure data quality and comparability?  Two ways.  First,
data quality objectives, established for each and every analysis, are used
to specify the requirements expected of the method.  For example, if my
objective is to measure vinyl chloride in groundwater at 10 ppb with a
precision of less than 25%, I would establish the procedural details around
this objective.  I might for example analyze a larger sample, use vinyl
chloride as my matrix spike and specify a 15% RSD for initial calibration.
                                  11-483

-------
If my objective was  to survey groundwater for the Appendix IX list of
volatiles,  I  might use representative compounds as matrix spikes and have a
larger %RSD for initial  calibration,  i.e.,  use the default conditions in
the method.

If the method is written to allow for the most general  objective, then
requirements  can be  superimposed for more specific objectives.

The second  process to assure data quality and comparability is to establish
a rigorous  method validation process.  This process,  described later, will
allow virtually any  change to the method, as long as  the change is
validated.  Otherwise, the information contained in the method become the
requirement.   The two processes, DQOs and method validation,  must be used
in tandem.

This approach is not significantly different from the approach in EPA-
600/8-83-020, "Guidelines and Format for EMSL-Cincinnati Methods". (4)  In
fact, the original Method 524, appended to EPA's report, was  the closest
example to  our approach of the methods surveyed.  Unfortunately, many of
the niceties  of this method were eliminated in subsequent revisions.  For
example, the  original method allowed for alternate traps if "it has been
evaluated and found  to perform satisfactorily".  This language was
eliminated  from the  revisions.

DATA QUALITY  OBJECTIVES

In 1984, the  Quality Assurance Management Staff (QAMS)  at EPA proposed that
the design  of environmental data collection programs  be based on the
development of Data  Quality Objectives (DQOs).  DQOs  are statements of the
level of uncertainty that a decision maker is willing to accept in results
derived from  environmental data, when used in a regulatory or programmatic
decision (9).  The DQO process is designed to ensure that the quality of
data is compatible with the requirements of the decision making process.
By utilizing  this process, the participating parties  can design an
environmental data collection program and its associated QA/QC program
which results in data which satisfies the needs of decision makers in a
cost effective manner.

In the DQO  process,  the decision maker must describe  the decision, why data
are needed  and background on the problem.  The type of information needed
for the decision is  described with respect to the scope and type of data
required.  The use of the data must be defined, along with the importance
of the data for making a decision.  The consequences  of an incorrect
decision resulting from inadequate environmental data are also described.
The extent  to which  false positives and false negatives can be tolerated
must be defined. In addition, a description of the available resources to
fund the project must be stipulated.
                                 11-484

-------
The second stage of the DQO process impacts the laboratory.  It is in this
portion that the specific data required are defined with respect to
analytes, matrices, spatial and temporal requirements.  The results to be
derived from the data are stipulated.  That is, are the data to be assessed
for a particular statistic, for values relative to a regulatory action
level, or for determination of baseline or background levels.  The desired
performance with respect to precision and accuracy, and acceptable levels
of false positives or false negatives are detailed.  It is this portion of
the DQO process which should effect the specific method requirements and
the laboratory should have the flexibility to establish and validate method
criteria to meet the needs of the project.

Matrix-specific QC should be defined in conjunction with these elements
rather than being stipulated as an inherent part of the method.  The same
method can be used to meet a regulatory limit or provide data to determine
baseline levels and therefore the frequency and makeup of matrix-specific
QC (spike levels and spike components) should be controlled by the type of
information being sought, not specified in the method itself.

Every project should go through this evaluation and specification process
in which the DQOs are clearly defined.  The consensus of the lab, the data
user and the regulator must be forged before the work begins.  The process
should be formalized in a project-specific Quality Assurance Project Plan
(QAPJP).

We therefore believe that the methods should have fewer elements relating
to the project objectives and that these elements should be addressed by
the DQO process.  For example, we believe that analytes should not be
listed in the method.  Rather, a list of compounds evaluated by the method
and their performance (precision, bias, detection limits) should be
contained as an Appendix to the method.  (As a side benefit, this list
could be expanded to include additional analytes and/or additional
performance data without rewriting the method.)  As another example, the
components used as matrix spikes, the spike levels and spike frequency
should not be specified in the method, but in the project objectives.  For
example, we are continually amazed at the number of customers who are
interested only in PCBs but who require representative pesticide compounds
be spiked, because the compounds are listed in Method 8080.

By segregating project objectives from method requirements, data users will
be forced to make decisions based on their objectives, and not on some
default condition in the method.  For example, the ASTM Standard Practice
addresses the issue of matrix spikes by first defining a matrix spike as
"an aliquot of sample spiked with a known concentration of target
analyte(s)" and then requiring matrix spikes to be analyzed "based on the
DQOs of the data collection activity".  (10)

Thus, the data user must decide on the compounds and the frequency.
Obviously, the laboratory must be more involved in the overall process.
However, we believe that this approach will improve the quality and
                                  11-485

-------
usability of the data.   As presented by Jim Barren of OERR at the
Analytical Methods Caucus in San Diego in March 1991, EPA has focused
almost entirely on data authenticity. (11)  This approach implies that
"quality" is achieved when the method is followed exactly and that "fraud"
is committed when the method details are not performed.  As stated by Mr.
Barron, the method can  be followed exactly and unusable data of poor
quality can be generated.  Conversely, useable data of high quality
obtained by a non-approved modification of the method can be rejected.  We
believe that our approach will not only improve the quality and usability
of the data but will  also restrict charges of fraud to those instances
where actual fraud occurs.

This approach obviously increases the complexity of the laboratory work and
could result in bottlenecks which would prevent work from ever being
performed.  We believe  the solution to this dilemma is to establish default
conditions which are  used in many situations, such as those determined from
a general survey.  In these situations, the default conditions could be
written into the method.  For example, the default condition for an
Appendix IX volatile  analysis could be the analytical conditions in the
method, representative  target compounds for matrix spikes, matrix spikes
every twenty samples  per project, a five point calibration for all
compounds with a 30%  RSD for all compounds with two allowed out.  The
default conditions for  analysis of TCLP leachate for toxicity
characteristic compounds might involve modifying the sample size relative
to the method, using  all target compounds as matrix spikes, spiking every
sample and requiring  a  25% RSD for every compound with a three point
calibration.

This approach requires  that the laboratory and the data user agree on the
objective, prior to the initiation of the project.  We have developed a
project initiation checklist which addresses these types of issues.  In our
process, we seek to obtain consensus from the data users on the following
issues:

    o    Sample containers, preservatives, holding times
    o    Operational  details - sample size, calibration, etc.
    o    Quality control samples - frequency, spike components, spike
         levels, control limits
    o    Detection limit requirements
    o    Report format, content

In those situations where the objective are not clearly known, we have
established default conditions which relate to these issues.  These default
conditions are documented in our laboratory QAPP, in method SOP's, and in
other reference documents.  For example, we have a document which lists the
spike components and  spike levels for routine methods and another document
which lists analytes  and reporting limits for every test performed.
                                 11-486

-------
METHOD VALIDATION

As discussed in the Section 518 report, a validated method is a method
based on sound technical principles that can be used routinely to achieve
some base level of performance. (12)  We believe that method performance
data obtained by the author of the method should be included in an appendix
so that potential users can evaluate the usefulness of the method.

We also believe that laboratories who propose to use the method must
demonstrate a basic capability to generate comparable data using the
method.  While we believe that the existing approach in Methods 624 and
8240 adequately address this issue, we would recommend a more rigorous
approach.  We propose that the method contain performance data for a
limited number of analytes, such as those on the priority pollutant or
target compound lists.  This performance data would be used by laboratories
to assess their performance.  Laboratories wishing to use the method would
be required to analyze seven spiked samples (ideally standard reference
materials) at concentrations spanning the working range of the method.
Statistical tests (f-test and t-test) would then be performed to
demonstrate equivalency.

This process would demonstrate that a laboratory has the basic capability
to perform the method for a limited number of analytes.  On going quality
control activities would then demonstrate the laboratory's performance on a
continual basis.

The more important issue relates to expanding the analyte list or to
modifying the method.  We believe that the methods should be sufficiently
flexible to allow for extensive modification.  However, to provide a
measure of control, we recommend that any change to the method and any
addition to the basic analyte list be permitted only if a rigorous
validation is performed.  While on the surface this recommendation may seem
overly stringent, if the proposed change will substantially improve the
method (better quality, cheaper, faster, safer, etc.) then the effort will
be justified.  Otherwise, there is no need to change.

Method validation must address method characteristics such as:

         o    detection limit,
         o    working range,
         o    precision,
         o    ruggedness,
         o    matrix,
         o    analytes,
         o    comparability, and
         o    bias

Adding a new analyte is distinctly different from modifying an existing
validated method.  In the first case, very little if anything is known
about the performance and thus a more extensive validation must be
performed.  We are developing an internal approach which involves four
activities.  The initial activity is to establish the working instrumental
range.  Once the working range is known, spiked samples, spanning the
working range are carried through the method.  If this process is


                                  11-487

-------
successful, ruggedness testing is performed.  Finally, blind spiked samples
are analyzed. (13)

From these analyses, sufficient information will be provided to thoroughly
document all of the method characteristics for the new analytes.

For an existing analyte in an existing method, the method characteristics
should be known.  The key when changing the method conditions is to
demonstrate that the change did not significantly affect the performance.
The process involved would require 14 analyses of a spiked sample, seven
analyses under the existing method conditions and seven using the proposed
modification.  Statistical tests (f-test and t-test) would then be
performed to demonstrate equivalency.

The previous section discussed replacing method requirements with method
descriptions.  These method descriptions would be the default requirements
unless an equivalency study was performed.  Thus, for many of the method
details, a proposed change would not be justified relative to the efforts
involved.  However, if the change was important, e.g. packed column versus
capillary column there would be a system to allow for the change.

No equivalency process will address every sample and every method condition
that may be experienced.  The purpose of this validation process is to
demonstrate that the proposed change is fundamentally sound.  Laboratory
controls and project specific quality control activities are used on an
ongoing basis to assess the quality of the laboratory work.

SUMMARY

As David Friedman discussed in his Environmental Lab article, this approach
will give analysts (and laboratories) more freedom.  For this approach to
work, the laboratories and data users must therefore accept more
responsibility to ensure that the work is performed correctly.   Freedom to
change methods will not result in constantly varying methods.  This
approach will not meet the laboratory needs any better than the current
requirements.  The driving force for this approach is the achievement of
technically sound, defensible data that does not burden the laboratory with
overly restrictive requirements.

We recognize that rewriting all of the existing methods to incorporate this
approach is a formidable challenge.  In the interim, we would request the
EMMC to introduce language into each method which will allow deviations
based on DQOs and validation data.

Finally, the establishment of DQOs  prior to initiation of the work, must be
mandated by EPA.  It can no longer be acceptable for data users to request
an analysis from a laboratory without specifying the requirements.
                                 11-488

-------
REFERENCES

(1)  Fisk, Joan, "Methods Consolidation Efforts Between Superfund and
     RCRA," presented at the 14th Annual EPA Conference on Analysis of
     Pollutants in the Environment.

(2)  Friedman, David, "New Directions in Method Development,"
     Environmental Lab, August/September 1990.

(3)  Mealy, R.6., "Towards a Unified Approach", Environmental Lab,
     September 1989

(4)  EPA-600/8-83-020, "Guidelines and  Format for EMSL-Cincinnati
     Methods", August 1983.

(5)  "Methods for the Determination of  Organic Compounds in Finished
     Drinking Water and Raw Source Water," EMSL, Cincinnati, September
     1986.

(6)  EPA/600/4-88/039 "Methods for the  Determination of Organic Compounds
     in Drinking Water,"  EMSL, Cincinnati, December 1988.

(7)  40CFR Part 136

(8)  SW-846, Third Edition, November  1986.

(9)  Development of Data Quality Objectives,  Description of Stages  I and
     II, QAMS, July 16, 1986.

(10) ASTM ES 16, "Standard Practice for Generation  of  Environmental Data
     Related to Waste Management Activities:  QA/QC Planning and
     Implementation."

(11) Barron, Jim, "Quality Assurance  Issues"  USEPA  Analytical Methods
     Caucus, March 1991.

(12) EPA/600/9-87/030, "Availability, Adequacy, and Comparability of
     Testing Procedures for the Analysis of Pollutants Established  Under
     Section 304(h) of the Federal Water Pollution  Control Act," September
     1988.

(13) Winkler, P., et al, "Assuring Reliable Laboratory Data Via Rigorous
     Method Validation," in Preparation.
                                  11-489

-------
HQg            PERFORMING TCLP ANALYSES TO GET MEANINGFUL DATA
       Kyle Dolbow, Ph.D.  and  Jody Price,  IEA Inc.  of New  Jersey,  62L
       Route 10, Whippany, NJ 07981

       The Hazardous Waste Regulations which were promulgated in March of
       1990 have  had significant  impact  on  waste  generators and  the
       commercial laboratory.  The  major change in the regulation is the
       substitution of  the  Toxicity  Characteristic Leaching  Procedure
       (TCLP)  for the Extraction Procedure Toxicity test.   Upon closer
       examination of the regulation, IEA  Inc.  of  New Jersey discovered
       several difficulties for waste generators and many new challenges
       for the laboratory.  Typical problems are:
            1.0  The  waste generator  may not  be  able to tell  how many
                 analyses will  be  reguired (and  the total  cost)  until
                 after samples have been submitted to the laboratory and
                 initial testing has been performed.
            2.0  With many types of  samples, matrix interferences severely
                 affect the analysis of one or more fractions.
            3.0  For  multi-phase samples,  data from up to  six complete
                 TCLP analyses  has to  be  combined through  defined and
                 complicated formulas to a final set of numbers which is
                 then   compared  to   regulatory  levels   to  make   a
                 hazardous/nonhazardous decision.
            4.0  The  regulatory levels  for the individual  analytes in a
                 particular fraction, such as semivolatiles,  cover a range
                 greater than the linearity of the instrument used to do
                                   11-490

-------
          the analysis.
     5.0  Matrix spike  levels  and other QC requirements are much
          different from that of  "normal" SW846 methods.

All of the above concerns  caused IEA,  Inc.  of New Jersey to take
an integrated approach to  TCLP analysis.  This approach includes
all laboratory  staff  understanding TCLP data quality objectives,
client communication,  optimized sample preparation methods, TCLP
specific  analytical  schemes,  and  an  automated computer  data
handling strategy.  The end result is a high quality product which
specifically  addresses the  data quality objectives  of TCLP and
presents  the  results in a format that  is  both  easy  to read and
understand.   This  paper  describes  this integrated  approach  in
detail.
                             11-491

-------
107    TOTAL CYANIDE BY PHOTOLYSIS







        Jenner Gutierrez,  SAIC,  8400 Wespark Drive,  Mclean,  VA  22102










         ABSTRACT









         The purpose of this study was to develop and examine  a viable method which




         could provide better quantification of the  total cyanide  (both soluble  and




         insoluble organometallic complexes) content in a waste.   The methods which




         have been employed comprise the derivatization and  gas chromatographic




         separation of benzylic nitrile derivatives.  The results  have shown promise




         in that the reaction occurs spontaneously and at room temperature.  Present




         studies are currently aimed at reducing the overall detection limit and at




         determining the degradation efficiency of the cyano-metallic complexes  through




         photolysis.  At the conclusion of the study, in-house samples will be prepared




         to compare the prospective with the current EPA methods SW-846 9010 and 9012.
                                          11-492

-------
-I QO   AMMONIA and TOTAL KJELDAHL NITROGEN DETERMINATIONS USING FLOW
       INJECTION ANALYSIS WITH GAS DIFFUSION

       J. Philip Calvi, Perstorp Analytical, Inc.,   2875  C Towerview
       Road,  Herndon,  VA   22071; Bernard Bubnis,  Novatek,  10 West
       Rose Avenue, Oxford, OH  45056;  Jan-Ake Persson,  Tecator AB,
       Box 70, S-26321  Hoganas,  Sweden.

       Abstract

       Free ammonia and total Kjeldahl  nitrogen (TKN)  in wastewater
       have been determined by flow injection analysis (FIA) using a
       gas permeable membrane.  Final effluent from ten sample sites
       representing  the  top  five  standard  industrial  code  (SIC)
       classifications  for the  nitrogen parameters  were  tested.
       Results  comparing this  new method with the established EPA
       methodologies are  presented.

       FIA methods introduce  an aliquot of sample using an injection
       valve.  The valve generally is capable  of delivering 40 to 200
       uL of  sample into a reaction stream which  produces ammonia
       gas.   The gas permeates an in-line membrane to enter  into an
       indicating acceptor  stream where a color change takes place.

       Data  indicate that  the gas  diffusion methods give  results
       equal  to or  better than  currently approved  EPA protocols.
       Operating  ranges were determined to be  from 0.02 to 10 mg/L
       with a method detection limit  (MDL) of 0.006 mg/L for  ammonia
       and from 0.2  to  10 mg/L with  a MDL of  0.02  mg/L for TKN.

       The gas diffusion technique is  simpler  than other automated
       nitrogen procedures.  It does not require harmful chemicals;
       is not sensitive  to Kjeldahl digest  pH;  and  is capable of
       producing  a  result in 70 seconds.

       Introduction

       The growing  concern about the environment and  the quality of
       drinking  water  have  caused  a  substantial  increase  in the
       number and frequency  of  analyses performed by laboratories.
       The nitrogen parameters (ammonia and TKN)  are  a particularly
       important  indicator  of the quality of water and soils  and are
       monitored  routinely.    Traditionally these parameters are
       measured   using   distillation  techniques,   ion selective
       electrodes and a variety of automated colorimetric procedures.
                                    1-493

-------
A number of drawbacks  exist using the current  EPA approved
chemistries.   The  phenate method  for  ammonia and  TKN (EPA
350.1/351.1)  requires  the use  of a  high concentration  of
phenol.  Reagent toxicity and disposal issues certainly need
attention when this method is used.  Similarly, the salicylate
methodology for  TKN (EPA 351.2) uses sodium nitroferricyanide
which is also classified as a hazardous  chemical.  In addition
to the health risks associated with the phenol reagents used
in the EPA  methods  350  and 351, the smell of the reagents can
be a nuisance.

The automated EPA  chemistries can be  difficult to operate.
Both  methods  are  temperature  dependent  requiring  close
control.   The  salicylate method  is prone to precipitation
problems which cause clogging of the  reagent channels.  The
most  troublesome  aspect to using  these  methods  is  the
influence of pH on method sensitivity.   Strong buffers are
required to maintain  pH  control.    The TKN  analysis  is
particularly a  problem in terms of final sample pH. During
sample digestion, the organic matter in a sample will consume
acid.  Therefore it is possible that the indivdual digestion
tubes can contain slight acid variations.
Block digestion as a sample  preparation for FIA gas diffusion
methods proved  to be  efficient  and in  general troublefree
during the testing phase of this work.

The  FIA gas  diffusion  technique  addresses the  automated
chemistry  method drawbacks.  The  reagents  used are  sodium
hydroxide,  water and a  colorimetric indicator.  An  aliquot of
sample is injected into a carrier stream which is merged with
sodium  hydroxide to  raise  the  pH  of the  sample.   Under
alkaline conditions  ammonium ion becomes ammonia  gas which
passes  through  an  in-line  gas permeable membrane  into an
interference free colorimetric indicator.  The color change of
the indicator is monitored  at 590  nm and is proportional to
the  amount of  ammonia  that passed  through the  membrane,
Variations in the acid  content of samples is overcome by using
an excess amount of sodium hydroxide.  The cycle time  of the
method  is 70 seconds.

In 1990, Tecator AB (Sweden)  field tested and submitted the
gas diffusion method to the EPA for nationwide method approval
and  inclusion  in  the  Federal Register.   This process was
performed   as   outlined  in  the  EPA   bulletin   entitled
"Requirements  for Alternate  Test Procedures for  Inorganic
Parameters  in  Non-Continuous  National  Pollutant  Discharge
Elimination System Monitoring".  The results of this study are
summarized in this paper.
                             11-494

-------
Sample Site Selection

In  accordance with  the June  20,  1990  protocol,  ten  (10)
industrial sites representing five (5) SIC listings for each
parameter supplied by EPA were  sampled  (2  per SIC  site)  and
are listed in Table 1.
Table 1  SIC Listings for Nitrogen Parameters

Ammonia                            TKN
SIC    Industry
4952   sewerage
2621   paper mills
2869   industrial chemicals
4911   electrical services
2911   petroleum refining
SIC    Industry
4952   sewerage
2621   paper mills
2869   indust. chemicals
1475   phosphate rock
2611   pulp mills
Reference Methods

The methods chosen to compare  the automated FIA procedures
were the ion selective electrode procedure EPA methods 350.3
and 351.4.   These methods  were chosen since the  EPA had a
large data  base to compare  results generated by the proposed
methods.  In the  case of TKN, the electrode method requires
block  digestion.     This feature  was  appealing  since  our
procedure  of choice  is  block  digestion  and not  the macro
digestion   procedure  commonly  used  with the  distillation
methods.  Further, EPA was adamant  that no deviations from the
referenced  methods take place.  Since  the  only other block
digestion   sample  preparation  procedure  approved used  the
salicylate  chemistry, our  choice  was to use  the electrode
procedures.

Experimental

Sample Collection  and Preservation

Samples  were  collected  in  either  glass   or  polyethylene
containers.  The  samples were acidified by the addition of a
sufficient  amount of  cone H2S04 to lower the pH to <2 followed
by refrigeration  at  4ฐC. Using this procedure,  the maximum
allowable holding  time is 28  days.
                             11-495

-------
Analysis Requirements

ATP submission to the EPA can be undertaken for nationwide use
(NW) or limited use  (LU) by EPA Regional, State or commerical
labs and individual dischargers.  Table 2 describes the number
and type of required analyses and quality control checks that
need to be presented in the ATP submission.

In  summary,  for nationwide ATP approval a  method submittal
must include 250 analysis results  from the  top five (5) SIC
classifications  for a particular parameter (125 each for the
approved and proposed methodologies)  and  38 quality control
checks for a total of 288 pieces of data.
************************************************************

Table 2  Effluent Sample and Subsample Analytical Requirements

Type   Applicant    Analyses          Quality Control
          unspiked  spiked   total   known  unknown   total

NW     Any   10      240      250     25      13        38

LU     EPA    5      120      125     13       7        20
     Regional
     State or
    commerical

LU   Indiv.   5       60       65      7       4        11
   discharger
Digestion Procedure

A block digester was pre-heated to 160 ฐC.  100 mL of sample or
an aliquot of  sample  diluted to  100 mL was  placed  in each
digestion tube.    Sulfuric acid-mercuric  sulfate-potassium
sulfate solution was added to each sample tube.  Two boiling
rods were placed in each tube.  A fume exhaust manifold was
placed over the digestion tubes which were then lowered into
the preheated block for one hour.   The block temperature was
then raised to 380ฐC for one and  one half hours.  The tubes
were removed from the digester; the boiling rods rinsed and
the residue diluted to volume.
                            11-496

-------
Apparatus

     FIA System

          Tecator FIA System to include:

          Injection valve  capable  of injecting 40 to 200 uL
          samples
          Tecator gas diffusion chemifold
          Thermostat
          590 nm detection system


     Digestion Apparatus

          Tecator aluminum block digester  (6 or 20 place)
          Fume removal manifold
          Digestion tubes, 250 mL
          Boiling rods


     FIA System Operating  Information

          Injection time             20s
          Cycle time                 70s
          Analysis rate              50/hr
          Sample loop                40 -  200  uL
          Temperature                3 0 ฐ C
          Wavelength                 590 nm
          Pathlength                 10 mm

     Flow Diagram Information  (See Figure  1)

          Mixing coil:   100 cm x  0.7 mm id
                          (temperature - 30"C)

          Flow rates:    Sample   (blk/blk) =1.4  mL/min
                         Reagent 1 (or/or) = 1.8  mL/min
                         Reagent 2 (or/wh) =0.9  mL/min
                         Indicator (blk/blk) = 1.4 mL/min
          Reagents:
                               Ammonia             TKN

                Reagent 1      water              5 N NaOH

                Reagent 2      0.5  N NaOH         5 N NaOH
                             11-497

-------
                                                              Revised  3/91
 Figure 1
Ammonia/TKN  Flow Injection  Gas  Diffusion  Manifold
 Reagent 1
 Reagent 2
 Indicator
                              Thermostat
                          /    Gas
                         •/   Diffusion
                                                    Cell
HA SYSTEM OPERATING INFORMATION
Injection time:
Cycle time:
Analysis rate:
Sample loop:
Temperature:
Wavelength:
Pathlength:
Evaluation:
Gain:
20s
70s
50/hr
30ฐC
590 nm
10 mm
peak height
1
FLOW DIAGRAM INFORMATION

Mixing coil 1:        100 cm x 0.7 mm i.d. (temperature ซ 30ฐC)
 Flow rates:
Sample (blk/blk) = 1.4 mL/min
Reagent 1, NaOH (or/or) - 1.8 mL/min *
Reagent 2, NaOH (or/wh) = 0.9 mL/min
Indicator (blk/blk) = 1.4 mL/min
  * For Aanonia Reagent  1  = Water
                                   11-498

-------
Quality Assurance

Each laboratory using these methods in regulated environmental
monitoring   is  required  to   operate  a   formal   quality
assurance/control program.  The minimum initial requirements
of  this  program   consist   of  the  demonstration  of  the
laboratory's capability with these methods.  On a continuing
basis,the laboratory should  check its performance (accuracy
and  precision)  by analyzing  reagent  blanks  and  check
standards,   fortified   blanks   and/or   fortified   samples
preferably at a minimum frequency of 10% of the total samples
analyzed by the methods.  The laboratory should maintain the
performance  records that  define  the  quality  of the  data
generated with the method.

Method Detection Limit  (HDL)

The procedure for determining the MDL is outlined in 40 CRF,
Part 136, Appendix B, Rev. 1.11.  A refined second document in
Environmental Science and Technology  (1981, 12, 1427) by EPA
personnel further explains the MDL calculation.   The MDL is
defined as the minimum concentration of a substance that can
be measured and reported with 99% confidence that the analyte
concentration  is  greater than  zero and  is  determined from
analysis of a sample in a given matrix containing the analyte.
The MDL for ammonia  and TKN was experimentally calcuated to be
0.0064 and 0.023 mg/L respectively  (Table 3).

Results

Linear Range

Calibration  curves  for  the  ammonia  and  TKN  gas  diffusion
methods  were linear over  the  concentration  ranges  tested.
Figure 2  is  the calibration  curve for ammonia in the 0.02 -
2.0 mg/L range  (gain setting =5).  It  is described by y = -
0.0092 + 0.5578x R = 1.00.  A second calibration for ammonia
in the 0.2 - 10 mg/L range (gain setting = 1) was run and is
described by y = -0.0046 + 0.0969x R =  1.00.  The TKN method
was linear  over the 0.2 - 10 mg/L  range.   The TKN curve is
described as y = -0.003 + 0.0713x R = 1.00.

Accuracy and Precision

Data indicate that the gas diffusion methodology is capable of
measuring ammonia  and  TKN levels  over  the indicated ranges
with high accuracy and good precision.   Results are presented
in Tables 4 and  5.   Analysis  of EPA  unknown  samples were
carried out over the course of the work (3 months).  Results
are presented in Tables 6 and 7.  EPA has yet to declare the
value but the five  (5)  individual ampules of each unknown is
consistent over the three (3) month testing period.
                            11-499

-------
Figure 2
   1.0-
   0.8-
   0.6-
i
8
   0.4-
   0.2-
   0.0
              NH3, T-W-0020-1, High Range (Revised 3/91)
                                y = - 0.0046 + 0.0969x  R . 1.00
                                                   8          10
                         Concentration (mg/L)
                              II-500

-------
SIC Sample Analysis

A sample data set which compares the EPA accepted methodology
(ISE)  and the  gas  diffusion method for ammonia for  a  SIC
sample  is presented in  Table 8.  Table 9 shows  similar data
for a TKN SIC  sample.

Compiled data  for the entire data  set  are shown in Tables 10
and 11.
Table 3  MDL Information

                               TKN                  Ammonia

     Mean  (n = 7)              0.329                0.0175

     St  Dev                   0.0063                0.0023

     S 2                        0.00011               0.000006

     Su2                        0.000039              0.000005
      b

     F.95C6.6,                   4'28                 4'28

     S 2/S *                    2.78                 1.18
      a ' b

     S                         0.0087                0.0024
      pooled

     MDL (mg/L)               0.023                0.0064

************************************************************
************************************************************

Table  4    Ammonia Accuracy  and Precision

Known      Mean      St Dev     RSD (%)      Bias    % Recovery
0.04
0.25
0.51
1.00
4.00
9.05
0.04
0.24
0.51
0.99
4.08
8.90
0.01
0.02
0.01
0.02
0.04
0.05
25.00
8.33
1.96
2.02
0.98
0.56
0.0
-0.01
0.0
-0.01
+0.08
-0.15
100.0
96.0
100.0
99.0
101.7
98.3
 ************************************************************
                              1-501

-------
************************************************************
Table 5 TKN Accuracy and Precision

Known     Mean      St Dev    RSD (%)
                           Bias    % Recovery
0.20
0.30
0.90
2.50
4.00
8.00
0.20
0.30
0.91
2.53
3.95
7.99
0.02
0.03
0.01
0.01
0.03
0.25
10.00
10.00
1.10
0.40
0.76
3.13
0.0
0.0
+0.01
+0.03
-0.05
-0.01
100.0
100.0
101.1
101.2
98.8
99.9
************************************************************

Table 6  Analysis of EPA Ammonia Unknown Samples
Trial 1
Trial 2
Trial 3
Trial 4
Trial 5

Mean
St Dev
Unknown fl

  20.66
  21.13
  20.69
  20.35
  20.91

  20.75
   0.26
Unknown #2

   3.55
   3.49
   3.46
   3.42
   3.53

   3.49
   0.05
                                             Unknown  f3
 0.98
 1.
 1,
 ,00
 ,18
0.96
0.93
 1.01
 0.09
Table 7  Analysis of EPA TKN Unknown Samples

               Unknown fl     Unknown I2      unknown #3
Trial 1
Trial 2
Trial 3
Trial 4
Trial 5

Mean
St Dev
   0.73
   0.72
   0.75
   0.76
   0.71

   0.73
   0.02
   11.37
   10.30
   10.60
   11.00
   11.00

   10.85
    0.37
12.59
12.91
13.34
13.05
13.00

12.98
 0.24
************************************************************
                            11-502

-------
                       Tecator AB Alternate Test Procedure Data
                         Nitrogen,  Ammonia,  Method T-W-0020-1
                  Automated FIA Gas Diffusion Reference No. N90 0018
Table 8

                Sample A            Sample B
                                  ISE       FIA
                                  388       379
                                  388       388
                                  387       381
                                  388       383
                                  < 1        3
Source:  SIC 4952 Sewerage Plant #1
Note:  Ammonia results in mg/L; ISE Method EPA 350.3
s
CO




Technique
Trial 1
Trial 2
Trial 3
Mean
St Dev
ISE
0.19
0.17
0.16
0.17
0.01
FIA
0.20
0.20
0.20
0.20
0.0
   Sample c
ISE       FIA
809       794
780       802
789       800
793       799
 10        3
 Sample D
ISE     FIA
1208   1189
1194
1183
1195
  9
1183
1187
1186
  2

-------
                       Tecator AB Alternate Test Procedure  Data
                           Nitrogen,  TKN,  Method T-W-0021-1
       Block Digestion, Automated FIA Gas Diffusion EPA Reference No. N90 0015
Table 9
Sample A
Technique
Trial 1
Trial 2
Trial 3
Mean
St Dev
I8E
0.18
0.19
0.18
0.18
0.0
FIA
0.14
0.17
0.17
0.16
0.01
Sample B
ISE
8.2
7.9
8.2
8.1
0.1
FIA
8.0
8.0
8.1
8.0
0.01
Sample C
ISE
16.1
16.7
15.7
16.2
0.4
FIA
16.1
15.8
16.4
16.1
0.2
Sample D
ISE
24.2
24.8
24.6
24.5
0.2
FIA
24.2
24.4
23.4
24.0
0.4
Source:  SIC 4952 Sewerage Plant #1
Note:  TKN results in mg/L; ISE Method EPA 351.4

-------
Table 10
                    Combined Ammonia Data
SIC 4952  Sewerage
     Technique         Mean

     FIA                0.16
     ISE                0.16
     FIA                393
     ISE                399
     FIA                800
     ISE                782
     FIA               1202
     ISE               1182

SIC 2621  Paper Mills

     FIA                0.16
     ISE                0.14
     FIA                 387
     ISE                 398
     FIA                 771
     ISE                 764
     FIA                1209
     ISE                1196

SIC 2869  Industrial Chemicals
     FIA
     ISE
     FIA
     ISE
     FIA
     ISE
     FIA
     ISE
0.19
0.17
 388
 398
 794
 793
1197
1198
SIC 4911  Electrical Services
     FIA
     ISE
     FIA
     ISE
     FIA
     ISE
     FIA
     ISE
 0.18
 0.16
  403
  388
  808
  794
 1202
 1191
SIC 2911  Petroleum Refining
     FIA
     ISE
     FIA
     ISE
     FIA
     ISE
     FIA
     ISE
 0.20
 0.15
  392
  392
  806
  799
 1206
 1200
                Confidence Intervals
                    -3s       +3s
0.12
0.10
357
361
791
722
1150
1148
0.24
0.22
429
437
809
859
1254
1226
                    0.01
                    0.01
                     355
                     367
                     712
                     676
                    1160
                    1151
0.12
0.02
 401
 373
 780
 768
1196
1160
          0.31
          0.29
           419
           429
           830
           852
          1258
          1241
0.13
0.08
333
387
768
775
1177
1182
0.25
0.26
443
409
820
811
1217
1214
0.24
0.34
 405
 403
 836
 820
1208
1222
0.08
0.12
386
371
782
772
1146
1156
0.32
0.18
398
413
830
826
1264
1244
Note:  Ammonia  results  in mg/L

************************************************************
                            11-505

-------
************************************************************
Table 11
SIC 4952  Sewerage
                     Combined TKN Data
     Technique

     FIA
     ISE
     FIA
     ISE
     FIA
     ISE
     FIA
     ISE
 Mean

 0.14
 0.16
 8.0
 8.1
16.2
16.2
24.1
24.5
SIC 2621  Paper Mills
     FIA
     ISE
     FIA
     ISE
     FIA
     ISE
     FIA
     ISE
 0.14
 0.13
 8.0
 8.0
16.2
16.1
24.0
23.8
SIC 2869  Industrial Chemicals
     FIA
     ISE
     FIA
     ISE
     FIA
     ISE
     FIA
     ISE
 0.18
 0.17
 8.1
 8.0
16.0
16.2
24.1
23.9
SIC 1475  Phosphate Rock
     FIA
     ISE
     FIA
     ISE
     FIA
     ISE
     FIA
     ISE
 0.18
 0.14
 7.8
 8.2
16.2
16.2
23.8
24.0
SIC 2611  Pulp Mills
     FIA
     ISE
     FIA
     ISE
     FIA
     ISE
     FIA
     ISE
 0.13
 0.18
 8.1
 8.2
16.4
16.3
24.2
24.0
Confidence Intervals
   -3s         +3s
   0.05
   0.08
   7.1
   7.1
  15.4
  12.2
  22.3
  22.3
   0.11
   0.01
   8.0
   7.3
  15.6
  14.3
  23.3
  21.5
   0.02
   0.12
   7.6
   7.0
  15.3
  14.0
  23.7
  22.4
 0.23
 0.24
 9.1
 9.1
17.0
20.2
25.9
26.7
 0.17
 0.25
 8.0
 8.7
16.8
17.9
24.7
26.1
0.15
0.05
7.9
7.6
15.5
14.9
23.6
22.8
0.21
0.29
8.3
8.4
16.5
17.5
24.6
25.0
0.15
0.01
6.8
6.9
15.6
15.7
21.9
22.8
0.21
0.28
.8
.5
16.8
16.7
25.7
25.2
 0.24
 0.24
 8.6
 9.4
17.5
18.6
24.7
25.6
Note:  TKN results in mg/L
                           11-506

-------
Discussion

Precision and accuracy for the FIA gas diffusion methods are
shown in Tables  4  and 5.  For  ammonia  samples the standard
deviation did not exceed  0.05 and recovery ranged from 96.0 to
101.7 percent.  For TKN the standard deviation did not exceed
0.25 and recovery ranged form 98.8 to 101.2 percent.  Tables
6 and  7  show analyses of  EPA unknowns for  ammonia and TKN
respectively.   Again, very good  precision for  the FIA gas
diffusion technique is exhibited.

Tables 8  and 9  show  complete  data sets  for both  the ISE
reference methods and the FIA methods from one of the sewerage
sites (SIC 4952).  Four samples  were tested in triplicated by
the  ISE  and FIA procedures for both TKN  and  ammonia.   The
equivalency  of  the FIA  gas  diffusion technique to the ISE
reference method is obvious.

Tables 10 and 11 show the  combined data for each of the two
sites for each parameter (TKN  and ammonia)  and each  SIC code.
In almost all cases the precision of the FIA gas  diffusion
technique  is equivalent  or  better  than  the  ISE   reference
technique.

Conclusion

The FIA gas diffusion technique  for testing ammonia and TKN in
wastewater has been shown to provide equivalent results to the
EPA reference ISE methods  (350.3 and 351.4).   At a rate of 50
samples per  hour it offers an automated approach to TKN and
ammonia analysis of wastewater  additionally  it does not have
the  drawbacks of the phenate  and salicylate methods (350.1,
351.1 and 351.2) which were stated in the  introduction.

References

"Methods for Chemical Analysis  of Waster and Wastes".  USEPA
March 1983,  EPA 600/4-79-020

Requirements for Approval of Alternate  Test Procedures for
Inorganic  Parameters  in  Non-Continuous  National   Pollutant
Discharge Elimination System Monitoring,  June 20, 1990, Nancy
S. Ulmer and Larry  B. Lobring, Inorganic Chemistry Branch,
Chemistry  Research Division, USEPA  Office  of Research and
Development,  Environmental Monitoring   Systems Laboratory,
Cincinnati,  Ohio   45268

40 CFR Part  136 Appendix B page 510 - 512

Environmental Science & Technology 1981,12,1427,  John Glaser,
Denis Foerst, Gerald  McKee, Stephen Quave, William  Budde
                             11-507

-------
109
      AN OBJECTIVE CRITERION FOR TERMINATING PERMEABILITY TESTS

Mark  S.  Meyers,  P.E.,  Geotechnical  Engineer,  U.S.  Army  Corps  of
Engineers,  St.  Paul District, 1421 USPO, St. Paul, MN  55101
        ABSTRACT

        In  present day  (1991)  geotechnical  engineering laboratories,  when
        permeability  testing is  performed,  whether  the  testing is  being
        performed for research purposes or design purposes,  the criteria used
        to   terminate  the  permeability  tests  vary  from  laboratory  to
        laboratory. While  an experienced engineer  may be  able  to determine
        when equilibrium has been reached, his  judgment is based solely on
        his  past experience  with  the type of soil/permeant/permeability test
        combination he is using:  thus it is a subjective judgment.

        Subjective judgments are used in engineering every day, especially in
        the  field of geotechnical engineering, with its highly variable soil
        types and soil conditions.   Most laboratory tests  used to determine
        distinct  properties   of a  given soil  have  an objective criterion
        associated with  the  tests.   For exanple,  the Proctor test uses a
        specific energy input;  the plastic limit uses a 1/8-inch thread; the
        liquid  limit uses  the number of blows to close a given width groove
        along a 1/2-inch length of the groove; and shear strength tests use a
        given percent strain depending on the use and type of soil.

        In  recognition of the  uncertainties and  inadequacy  of  a  totally
        subjective test termination  criterion, this paper will investigate a
        new  and more generally applicable objective criterion for terminating
        a permeability test.  This approach is  applicable  to the soil types
        commonly tested in a geotechnical or materials testing laboratory.

        The  termination  criteria developed are  not intended to replace the
        judgment of the engineer or researcher.  The termination criteria are
        tools to be used as  an  objective confirmation of the judgment of the
        engineer  or  researcher  in  determining that  a   test  has  reached
        equilibrium.   In no  case  should  judgment  be overridden  if  the
        engineer or researcher feels the test has reached equilibrium.

        Background

        The  use of relatively  impervious compacted clays  and grouted sands
        for such  purposes  as liners in hazardous and toxic waste landfills,
        as   cores  in earth  and  earth-rock  dams,  and  as  cutoffs for dams
        requires  a thorough  understanding of the capability of the  soil to
        satisfactorily perform  its intended function.  The major indicator of
        the  ease  with which water  is able  to travel through  a  soil is
        referred to as the permeability of the soil.

         Permeability and Hydraulic Conductivity

        The  terms  permeability and   hydraulic  conductivity  are  often
         interchanged  in the field  of  geotechnical  engineering.   From the


                                          11-508

-------
fluid flow aspect, this  is only correct if the fluids are held at 20
C.    The  fundamental difference  is   that to  calculate  intrinsic
permeability, the temperature (and thus viscosity)  of the permeant is
taken into  account.   The  use of the term "permeability11 throughout
this paper is to be  taken  as hydraulic conductivity,  as no effort is
usually  taken  in  soil  mechanics  laboratories  to  measure  fluid
temperature.

Definition of Permeability

Permeability can be  defined as  the discharge velocity through a unit
area of  soil under a hydraulic gradient of unity  (Cedergren, 1989).
It  is more commonly known as the coefficient 'k1  in Darcy's law for
laminar  flow in a soil media.    Darcy's  law can be stated by the
equation
                              Q = kiAt                        Eq. 1

where Q  is the quantity of seepage flowing through a cross section of
soil  having an  area A  normal  to the direction  of flow,  under a
gradient i,  during a period of time  t.   If  the terms in Equation 1
are rearranged,  Equation 2  is  obtained, which is  the basis for the
experimental determinations of permeability  that  measure the amount
of  seepage over a period of  time under  a given gradient.

                              k = Q/iAt                       Eq. 2

The coefficient of permeability has units of velocity and is usually
expressed in centimeters  per second,   cm/s,  for soils having a low
coefficient  of  permeability.    Other  commonly used units  include
ft/day and ft/yr.

The coefficient of permeability is usually assumed to be constant for
a given  soil type.   However,  it  can vary widely for a given soil type
or  other material  depending on  a number of factors,  as discussed in
detail   by  Taylor   (1948),  Daniel  (1985),  Bodocsi   (1988),  Bowers
 (1988),  Carson  (1988), Cedergren (1989), and Conrad (1991).

Laboratory Methods for Determining Permeability

The two  most commonly used methods for determining the coefficient of
permeability of a soil sample  in a laboratory are the constant head
permeability test  and the falling-head permeability test, utilizing
either a rigid wall or a flexible wall  permeamster.  A discussion on
the constant head and  falling-head tests  can be  found in Cedergren
 (1989).   Daniel,  et  al.,  (1985),  Carpenter (1986), and Evans  (1986)
discuss  permeameters in  detail.

Terminating A Permeability Test

During a permeability test the  measured permeability of a soil sample
often undergoes a  prolonged period of  transitional behavior before an
equilibrium value of permeability is  reached.   The  decision as to
when  an  equilibrium value has been reached is not straightforward.
 In   general,  no  concensus  criteria  exist  for  terminating  the


                                 11-509

-------
permeability test procedure.

Die existence of a standardized objective  criteria would allow for
more  reliable   comparisons  of  permeability  between   independent
research laboratories and commercial soil testing laboratories,  would
aid  in  the elimination  of  inconsistencies  in  permeability  test
results, and would eliminate unnecessary  testing  time  (Pierce and
Witter, 1986).

Review of the Literature

A review of the literature has revealed several  termination criteria
in use at the present time, with subjective judgment being the most
common method used.

Subjective Judgment

The method  of  subjective judgment can be  described as a method in
which  a subjective  decision   is   made   regarding  whether  the
permeability of a test sample has reached equilibrium by examining a
plot of  permeability vs. time  for the test.  A horizontal plot is
sometimes used as  an indicator  of  equilibrium.    However, the soil
being tested may result in a permeability  on the order  of say 10~7
cm/s.    A horizontal  plot on  this   scale may be  interpreted  as
equilibrium, even  when  the slope has  some  small deviation  from the
horizontal.    At this order of magnitude  of permeability,  a very
slight deviation in the permeability plot may  in fact be an increase
or decrease of several percent.  In other words, equilibrium may not
have been reached.  Conversely,  if the soil  type were such that a
value  of permeability on the  order  of 10~3 cm/s resulted, a  large
deviation from  the horizontal  may  in  fact represent  only a  small
percent  increase  or  decrease  in  permeability.    In  the  latter
scenario, equilibrium may have been reached and  the engineer may not
realize this due to the visual  appearance of the  plot.

•Die  previous  paragraph   used  percent  increase   or   decrease  in
permeability as  a  measure of  comparison   of  the  data points from
reading  to   reading  during  a  permeability test  to  determine  if
equilibrium has been reached.   This is similar to the method used by
MoCandless  (1988).   In this termination criterion,  if the  value of
permeability does not vary by more than a predetermined percent for a
set  number  of  readings,  it  is  judged that equilibrium has been
reached.    The   difficulties  inherent  in   this   method are  how  to
determine an acceptable  percent  change  in  permeability over the
course  of  several readings and how to  arrive  at the number  of
readings  over  which  the  termination  analysis is to be made.   For
soils  with  a low value of permeability, a  relatively  small percent
change might be selected.  The  question still remains:  "How small of
a  percent change  is  acceptable?11   Even when combining  the percent
change over several readings with a visual  examination  of a plot of
the permeability test data, questions  arise as to whether equilibrium
has been reached.   Conversely, for a  soil with a higher permeability
value,  the  question  becomes:    "How  large of  a percent  change is
acceptable?11


                                 11-510

-------
Other Methods Used to Terminate a Permeability Test

Pierce and Witter (1986) use a cxaribined criterion,  specifying that at
least one pore volume of permeant must have passed through the sample
and the slope of a plot of log permeability vs.  number of cumulative
pore volumes cannot be shown, by the use of a regression analysis, to
differ significantly  from zero.  Although  this  criterion is  one of
the most  promising termination methods to  be developed, the  use of
linear regression for determining the slope of the plot has several
statistical shortcomings.

Bryant and Bodocsi  (1986)  suggest the use of an  adaptation of Mann's
test  for monotone  trend  and  also discuss  the use  of a  Bayesian
analysis.    Other  termination  criteria,   used  individually  or  in
combination with  other criteria, include:  a predetermined number of
pore volumes passing  through the sample;  a  change in permeability by
several orders  of magnitude;  a  predetermined concentration  of  the
effluent  (if a chemical permeant is used); a plot of log permeability
vs. some  measure of  time  becoming horizontal; a value of k greater
than  10~7 cm/s  being reached  (for a soil to be used as an approved
EPA cover or liner);  or the passing of at  least two pore volumes of
permeant through a soil sample in conjunction with a horizontal plot
of log  permeability vs. some  measure of time.  Although several of
the criteria used show promise and attempt to overcome some of the
shortcomings of  subjective judgment, these  criteria  do in fact also
exhibit some disadvantages.

The most promising termination criterion appears to be that suggested
by  Bryant  and  Bodocsi  (1986).    This source  has  developed  the
groundwork  for  a statistically based  termination criterion which
overcomes  the statistical disadvantages  of the method developed by
Pierce and Witter.   The method  developed by Bryant and Bodocsi uses
Mann's test  for  monotone trend to  determine if  equilibrium has been
reached within predetermined statistical levels of significance and
bounds.

Application of Mann's Test For Monotone Trend

Advantages Over Regression Analysis

Mann's  statistic  (Bryant  and  Bodocsi,  1986)  is  designed  to  be
sensitive  to  any  increasing  trend,  and   by altering the  method
slightly,  the statistic is  also sensitive to any  decreasing trend.
Conversely, linear regression may be relatively insensitive in a case
where  permeability  is  increasing  at a  decreasing  rate,  such  as
happens  when  the  permeability  test  is  approaching  equilibrium.
Mann's   statistic   is   less   sensitive   to  occasional   unusual
observations,  which   occur  in  most  permeability  tests.     These
occasional unusual  observations might unduly affect a slope cxanputed
by least  squares regression analysis such as  in  Pierce and Witter's
method.  Finally, Mann's statistic does not require the assumption of
normally   distributed  within  test  errors.    Psegression  analysis
requires the latter (Bodocsi and Bryant, 1986).
                                 11-511

-------
General Example

In  order to  apply  Mann's test for  monotone  trend to  a  set  of
permeability test data, Bryant and Bodocsi have heuristically adapted
it  for  use  in a  sequential manner.  For this general example,  the
procedure will be described as if a permeability test were being run
for  a  soil  in  which the test  results would  indicate  a  general
increase in permeability  with time  as  the test approaches a  final
equilibrium value.    The  4  steps   involved  in the  procedure  are
discussed below.

   1.  The permeability test  is permitted to run over a preliminary
period of time to allow at least two pore volumes of flow to pass
through the test  sample.

   2.  Following the preliminary testing period, permeability values
are measured at n equally spaced points in time (t = 1, 2,  ..., n).
Mann's test for  monotone  trend,  a  nonparametric  statistical  test,
allows  for the   use  of  data  which  is  not normally distributed
(contains less than approximately 30 data points).  The use of too
few data points  provides  insignificant results.   Bryant  (1986-1987)
recommends the use of 15 data  points at the start of the procedure.

Equally spaced test  readings should be used.   This is a generally
accepted  practice in most   laboratories,  with readings  taken  at
approximately  the same time every day.   As long as the time between
test  readings  does  not vary  drastically, the use of  approximately
equal time intervals  should be adequate.  For test samples  exhibiting
a  permeability  less  than  10~10  cm/s,  the  pore  volumes  of  flow
permeating a test sample  during a reading interval may be difficult
to  determine,  depending on the experimental apparatus being  used and
the environment in which the  apparatus  is being  used.  Carson (1988)
indicates that a value  of permeability of  1CT12 cm/s may be the
lowest  value  of permeability  which  can   be   obtained  with  a
conventional permeability testing apparatus.   The  use of  a  constant
flow  permeameter may eliminate this lower  bound.   Hie permeability
values are converted to Iog10 permeabilities, denoted as y^,  in this
step.

    3.   Mann's test is used  to  test a  null  hypothesis of  no  trend
 (i.e., the plot of yt vs.  time is horizontal and equilibrium has been
reached) against the  alternative hypothesis of an increasing trend in
 (i.e.,  the value of y^  continues to  increase  with  time)  at  a
specified level  of  significance  aj_.    If the null hypothesis  of no
trend is rejected, then the oldest test  observation  is deleted from
the data set and a new observation is made.   Step 3 is repeated until
the null hypothesis  of  no  trend  is   accepted.    Step   4  is  then
conducted.

ax  represents the probability of a  Type I error  occurring.  To guard
against a Type I error,  a^ is chosen to be small.   The actual  value
of  a-ฑ  is selected  to be 0.01,  approximately representing a 99%
confidence level that a Type I error will not occur.
                                11-512

-------
After the  first n data points for analysis are selected, Mann's test
statistic  is confuted by calculating the tied ranks and the bivariate
ranks for  the 2 x n matrix X representing the row x column matrix of
the first  n data  points of time and permeability-  Ihe matrix X is
                           xl  X2 X3
                                         Yn
This results in a 3 x n matrix F.  Row 1 of F contains the tied ranks
R!J of the first row of X; row 2 of F contains the tied ranks R2-j of
the second row of X; row  3  of  F contains the bivariate ranks Q-H of
X.                                                               -1

The tied rank Ry of an element X^ of a vector is defined as

                       Rij  = 0.5 +   u(Xi - Xj)            Eq. 3

where

                                  1, if t < 0

                       u(t) =     0, if t = 0

                                 -1, if t > 0

The bivariate rank Q^j of  a pair of elements (X^Y^) is defined as

                  Qij  = 0.75 +  U(Xi - Xj)u(Yi  -  Yj)       Eq. 4

where u(t)  is defined as in Equation 6.

Kendall's Tau for Step 3,  T3, is

                  T3 =  (4  F3j - n2 -3n) / (n(n-l))       Eq. 5

where  F3j  is the sum of the elements of row 3 of F.


Mann's statistic for Step 3 is found by dividing T3 by the variance
of Kendall's T, ST, which  is

                 ST = 2(2 x n +  5) / (9  x n x (n  - 1))     Eq. 6

Mann's statistic for Step  3, denoted by Z3/ is now

                              Z3 = T^STJI                  Eq. 7
Zo is  now compared to the critical value  of Z,  Zc, for  a level of
significance aฑ.   If Z3 is greater  than Zc,   the null hypothesis is
rejected.   The next step  is  to add a new data point after the next
reading and drop the oldest data point.

   4.  Step 4  requires that  the  experimenter specify  an upper bound

                                  11-513

-------
on the slope of y^ vs.  time plot from horizontal at the termination
of the permeability test.  This bound is referred to as B*, which is
greater than zero for a  general  increase in permeability with time.

In Step 4,  the data  is adjusted by cxsnputing a value of yt*, which is

                           Yt* - Yt - B*t                Eg-  8

Mann's test is now performed on the adjusted test data, as in  Step 3.
If the null hypothesis of no trend for the  adjusted data is accepted
at a level of significance a2 against an alternative hypothesis  of a
downward  trend   in  the  adjusted  upper   bound   data,   then   the
permeability test is terminated:  equilibrium has  been reached.   If
the null hypothesis is  rejected,  an additional observation is made,
but the  oldest observation  in the  data set is not  deleted.    The
sample size is thereby increased by one data point.  Return to Step 3
and repeat  the procedure until the  null hypothesis  is accepted in
Step 4.

For acceptance of the null  hypothesis  for Step 4, Z4 must be  less
than Zc  at  a level if significance  a2 •   When this occurs, the  null
hypothesis for Step 4 is accepted.  The permeability test can now be
terminated  and  the permeability value  for  the sample  is   that
determined for the terminating observation.

a2 represents the probability of a Type II error.   To guard against a
Type II error, a2 ^ chosen to be small.  The actual value of a2 is
selected to be  0.05,   a  level  of  significance  commonly used  in
engineering statistical  applications  (Brubaker and  McGuen, 1990) .

B* is a measure of the degree of  trend the experimenter judges to be
practically, as opposed  to statistically,  significant.  B  represents
an   upper   bound  (for  tests  exhibiting  generally  increasing
permeabilities) of the  slope of  the permeability  plot, below which
the experimenter wants  the final slope of the permeability plot to
be, in order to  be  considered for hypothesis testing  in Step 4.  B*
cannot be selected statistically  per se;  it can however be selected
practically, using the  results of past tests and  appropriate levels
of significance  aฑ  and  a2.   A goal of this research is to select
practical values  of B*  to use  in Step  4.  B* is proposed to be
dependent on soil type.

A  typical  statistical  hypothesis  test   would  terminate   testing
immediately  upon acceptance of  the  null  hypothesis  in  Step  3.
Acceptance  of  the null  hypothesis in Step 3  does not necessarily
provide  statistically strong evidence that  a trend does not exist.
It only implies that such a hypothesis can be maintained.
Step 4 arHซ  a check to determine whether the procedure is sensitive
enough to detect trends  of a meaningful magnitude.  With relatively
noisy data, as is the case for permeability tests, the insignificance
of  a  hypothesis test  for trend  does  not  provide  a  reasonable
termination  criterion.   The  insignificance  of  the   test must  be
combined with a mechanism which  ensures  achievement of  an adequate

                                11-514

-------
sensitivity  against trends  of  a meaningful  magnitude.    To guard
against a  Type II error, the selection of B*  is critical.   Step  4
also increases the sample size to increase the sensitivity of  future
tests.

In summary,  the procedure terminates  a permeability test after  two
statistical criteria have been met:

   1.  A hypothesis that the slope of  a plot of yt vs.  time is equal
to zero can be maintained, and

   2.  A hypothesis that the slope of  a plot of y^ vs.  time is equal
to or greater than B* (for tests exhibiting generally increasing
permeabilities) or is equal  to or less than B* (for tests exhibiting
generally decreasing permeabilities) is almost certainly false.

The Permeability Plot

A  typical  permeability plot consists  of  the dependent  variable,
permeability,  plotted  on the ordinate scale,  usually as Iog10'  vs.
the independent variable, time, plotted on the abscissa.  The measure
of time should be such that an equal interval of time is obtained for
each  data point.   The  two  most  common measures  of time used  for
permeability tests and  the associated  permeability plot are raw  time
on  test,  measured  in  consistent units  of hours,  days,  etc.,  and
cumulative pore volumes of flow passing through the sample.

The use of cumulative pore volumes of flow was thought to be superior
to  the use of  raw time.    At  the start of a  permeability test,
readings are taken at  specified periods of time,  which  are usually
dependent  on soil type  and  the  judgment of the  experimenter.   As a
test progresses, an increased or decreased number of pore volumes of
flow will  pass through the  sample during the  specified time period.
If the time interval between readings is maintained, especially for
soils with low permeabilities, reading the difference in fluid levels
in the standpipes becomes difficult,  depending upon the permeability
apparatus  being used.   This introduces errors  into the calculated
value  of  permeability for a  reading interval.   If the time interval
is  modified,   resulting  in reading  intervals  which  allow   for
approximately  similar volumes of permeant to pass through the sample
between readings, a regular  interval measure of time is obtained.

Bodocsi and Bowers  (1989)  and Carson (1988)  indicate that the use of
cumulative  pore volumes  of  flow as a measure of time results  in a
difficult  analysis of  graphical plots of log permeability  vs.  time
when  the material  has  a  low value of permeability.   Permeability
plots  using pore volumes of flow on the abscissa become vertical as
the   permeability  test  progresses,    indicating  a  decrease   in
permeability and  the  associated  reduction of  volume of  permeant
passing  through the sample.  Raw  time is found  to  be the superior
measure   of  time  for materials  with  extremely  low  values  of
permeability.  Where equipment restrictions do not apply and readings
can be taken at equally spaced time intervals, a time scale using raw
time  should be used.   The final determination of which scale to use


                                 11-515

-------
on the abscissa rests with the experimenter.

Bodocsi and Bryant  (1986) state empirical and practical reasons  for
the use of a  Iog10  scale on the ordinate.  This practice  is used in
this paper.
In summary,  a plot of log permeability vs.  some measure of time is to
be analyzed  to determine if the slope of the permeability plot at
equilibrium  is  essentially zero,  within  practical and  statistical
means.

Parameter Selection

Several  parameters  are  required  to  perform the  various   steps
discussed for determining when  to terminate a permeability test.  The
statistical levels of significance ai and a2 and the  upper or lower
bound on the slope of the permeability plot, B*,  are used to guide
the experimenter as to whether  or not the  permeability test should be
terminated.   The methodology used to select applicable parameters to
use in the algorithm is discussed.

Methodology

Bodocsi and Bryant (1986) recommend a practical selection of B* based
on the results of  past permeability  tests and  appropriate levels of
significance.  This methodology requires data from  many permeability
tests and some indication of when the permeability  test  should have
been terminated.

Bodocsi, et al. (1986),  Bowers, et al. (1988),  and  Carson (1988) ran
a large number of permeability tests.  The test data includes data
using water as a permeant to determine the baseline  permeability of a
sample  and  using  various  chemicals  as  permeants  to determine the
affects of chemical permeants  on various grouts.   McCandless  (1988)
ran  several  permeability tests to  determine  the  acceptability of
various solidification/stabilization  mixes.   Eighty five  data sets
were selected  for  analysis  in this work.   Most of the permeability
plots   for   the  data   had   an   approximate  horizontal  segment,
representing an apparent  termination  time.

A panel of five experts was assembled to determine  when to terminate
a permeability test, based on the permeability plot for a given data
set.    These individuals  have  a strong  experience  background in
permeability testing.   They have struggled with  permeability test
data in the past in trying to determine if a given  permeability test
should be stopped.

Each expert reviewed the  history of the permeability plot for each of
the 85  data  sets,  eliminating data points which were thought to be
non-representative of the  history of the  test (i.e., bad reading,
apparatus problems,  sample deterioration,  etc.).    The expert then
determined when  he, as  an experimenter,   would have  terminated the
permeability test.   The expert  was asked to consider only the portion


                                11-516

-------
of the plot which would have been available during the actual test.
In other words, "future" readings were not to be considered.

Several panel members were intimately  familiar with many of the data
sets.  The data sets were given generic names,  Data Set 1, Data  Set
2, and so  on,  in an attempt  to mask the actual identity of the data.
Those panel members familiar with the  data indicated that  several of
the plots  looked  familiar, but an effort was made to use an unbiased
judgment in determining the termination point.

The panel  members were asked  to  indicate which method they used to
select the termination  point.   Some form  of  subjective judgment  was
used by all panel members.

Final Data Sets and Ranges of Permeability

The  data  sets  supplied to  the  panel  members  were modified  to
eliminate  the data points  thought to be non-representative of  the
test history.   The data  files were then set up for use in computer
runs through a FORTRAN  computer program which mimics the SAS routine
used by Bryant to calculate the required statistical information.

Five ranges of permeability data  were used,  based on the apparent
order of magnitude of the termination points selected by  the panel.
The panel  agreed  on the termination point  in most cases.   In several
instances, the data set was broken up into two separate data sets, to
reflect the selection of one termination point by two or three of the
panel members and the selection of  another termination point a large
number of readings  away from the former point by the remaining two or
three panel members.  The ranges of permeability for which  parameters
will be selected are 10~6, 10~7, 10~s,  10"9, and 10~10 cm/s.

Initial Runs To Select Parameters

Several data  sets were run to test the significance of a^ and 32 on
the  selected  termination point.   For all  computer  runs,  B*  was
allowed to vary for each combination of the  levels of  significance,
resulting  in  a total of 500 combinations  of a^, 32, and B* for  each
data  set.    The  computer program was modified to print  a summary
table, listing the  combination of the parameters used to indicate the
first point at which the  procedure  would stop the permeability test.
For the  data sets  analyzed, B* was the more  significant  parameter,
resulting  in  a change  in the  selected termination point  of several
data points for a  change in B* of  only 0.0001.   Clearly,  B* is  the
more  significant  parameter,  regardless   of the  selection of  the
combination of levels of significance.

Selection of B*

With aj and a2 set at  0.01  and 0.05 respectively,  the  determination
of B*  proceeded  as follows.   The  computer program was modified to
print a  summary table  indicating the  first point at which the  null
hypothesis  for  Step 4 was accepted, along with  the value  of B* used
for that run.   B* was varied from 0.0001 to 0.10.   Each data set was

                                 11-517

-------
evaluated,  with the values of B* and permeability being tabulated for
the termination point selected by each panel member.

Several data sets gave an indication that the value of B* used in the
analysis is not significant or the values used for B* were not small
enough or  large enough to be  sensitive  to the  data  set.  In other
words, the  test was shown to terminate at the  same data point for
every value of  B* used,  or the test was shown to terminate prior to
or long after reaching the termination points selected by the panel.
These data  sets will have to  be reanalyzed using values of  B  less
than 0.0001 and greater than 0.10.   Therefore,  at the time  of this
writing, the results  should be considered preliminary.

Discussion

A plot of log permeability vs.  log B*  was generated using the  Harvard
Graphics software.   This plot illustrates apparent relationships of
log permeability vs.  log B*,  depending upon the order of magnitude of
permeability.   The relationships appear to be more sensitive  for the
higher orders of magnitude of permeability.

The apparent relationships between B* and Permeability were analyzed
using simple linear regression with Iog10 transformations on the data
for three orders of magnitude of permeability:  10~7;  10~9; and 10~10
cm/s.   The relationships using the data  transformations indicate
correlations  of 0.528,  0.725,  and  0.117,  respectively for  these
orders  of  magnitude   of  permeability.    Obviously,   other  data
transformations will need to used to determine  the most significant
relationship.

At this time,  there does not  appear to be  a  general relationship
between B* and permeability.   The relationship appears to be  limited
to distinct ranges of permeability.  Future analyses will clarify the
extent  and  significance   of  these   relationships  and   develop
mathematical equations to use in the termination procedure to select
an appropriate value of B*.  These equations will then be programmed
into  the computer  routine  to  intrinsically  select  B* during the
analyses, while allowing override by the user.

Acknowledgements

The  author  wishes  to  thank  the  USEPA  AWBERC   Risk  Reduction
Engineering Laboratory  for their partial  support of this research.
The author  also wishes  to thank Dr. Andrew Bodocsi Dr. Mark  Bowers,
and Mr. Richard McCandless of the University of Cincinnati, Mr. David
Carson  of  USEPA in Cincinnati,  and Dr.  Earl McCullough  of  the
University  of  Wisconsin at  Platteville  for  their assistance  in
evaluating permeability plots.
                                 11-518

-------
References

Bodocsi, A., M.T. Bowers, and R.A.  Sherer,  "Permeability of Grouts
   Subjected to  Chemicals,"  In:   Proceedings of the Specialty
   Conference   on   Environmental   Engineering.   Environmental
   Engineering Division  of The American Society of Civil  Engineers,
   June 1986, pp. 99-105.

Bodocsi, A., M.T. Bowers, and R.A.  Sherer,  "Reactivity of  Various
   Grouts to Hazardous Wastes and Leachates,"  Final Report, United
   States Environmental  Protection  Agency,  Contract No. 68-06-3210,
   Work Assignment 13, Cincinnati, Ohio, February,  1988.

Bodocsi,  A.,   and  J.L.  Bryant,   "Precision and  Reliability  of
   Laboratory  Permeability Measurements,"  EPA/600/S2-86-097,  United
   States Environmental Protection Agency, Cincinnati, Ohio, 1987.

Bowers, M.T.,  A. Bodocsi,  and D.A.  Carson,  "Reactivity of Various
   Grouts  to  Hazardous Wastes  and Leachates  -  Phase  IV,"  Final
   Report,  United States  Environmental Protection  Agency,  Contract
   Nos.  68-03-3210-13 and 68-03-3379-06, Cincinnati,  Ohio, January,
   1988.

Brubaker, K.L.,  and R.H. McGuen,  "Level of Significance Selection in
   Engineering   Analysis,"     Journal  of   Professional   Issues  in
   Engineering.  American Society of Civil Engineers, Volume 116, No.
   4, October 1990,  pp.  375-387.

Bryant, J.L.,  Private discussions with the author,  Department of
   Quantitative  Analysis, University of Cincinnati, Cincinnati, Ohio,
   1986 through  1987.

Carpenter,  G.W., and R.W.  Stephenson, "Permeability  Testing in the
   Triaxial Cell,"  ASTM  Geotechnical  Testing Journal.  Volume  9,
   Number 1, March,  1986, pp. 3-9.

Carson,  D.A.,  "Hydraulic Conductivity of Modified Cement and Polymer
   Based Grouted  Soils When  Exposed  to  Hazardous  Chemicals",  a
   thesis  presented to the Department of Civil  and Environmental
   Engineering in partial  fulfillment of  the requirements  for the
   degree   of  Master  of  Science,   University   of  Cincinnati,
   Cincinnati, Ohio, 1988.

Cedergren,  H.R., Seepage.  Drainage, and Flow Nets. 3rd Edition. John
   Wiley &  Sons, New York, New  York, 1989.

Conrad,   D.J.,  S.A.  Shumborski,  L.Z.  Florence,  and  A.J.  Liem,
   "Assessment  of  the  Parameters  Affecting  The  Measurement  of
   Hydraulic  Conductivity  for  Solidified/Stabilized Wastes,"  In:
   Remedial Action. Treatment,  and  Disposal  of  Hazardous Waste.
   Proceedings  of   the  17th Annual REEL  Hazardous  Waste Research
   Symposium.   EPA/600/9-91/002,  United  States  Environmental
   Protection Agency, Cincinnati, Ohio, April  1991, pp. 543-559.
                                  11-519

-------
Daniel, D.E., D.C. Anderson,  and S.S. Boynton,  "Fixed-Wall vs.
   Flexible-Wall  Permeameters,"  Hydraulic   Barriers  in  Soil  and
   Rock. ASM Special  Technical Publication  874,  A.I.  Johnson,  R.K.
   Frobel,  N.J.  Cayalli,   and  C.B.  Pettersson,  Editors,  American
   Society for  Testing Materials,  Philadelphia, PA,  June,  1985, pp.
   107-126.

Evans, J.C., H.-Y. Fang, "Triaxial Equipment for Permeability Testing
   with Hazardous and Toxic Permeants,' ASTM  Geotechnical Testing
   Journal. Volume 9, Number 3, September 1986, pp. 126-132.

Pierce,  J.J.,   and  K.A.  witter,   "Termination  Criteria  for  Clay
   Permeability  Testing,"  Journal  of  Geotechnical   Engineering.
   American Society of Civil Engineers, Volume 112,  No. 9, September
   1986.

SAS Institute,  Inc.,  SAS User's Guide;  Basics.  1982  Edition. SAS
   Institute, Inc., Gary, NC,  1982.

SAS Institute,  Inc.,  SAS User's Guide:  Statistics. 1982 Edition. SAS
   Institute, Inc., Gary, NC,  1982.

Taylor, D.W., Fundamentals of Soil Mechanics.  John Wiley & Sons, New
   York, New York, 1948.

Notation

&1  =  statistical level of significance for Step 3
a2  =  statistical level of significance for Step 4
A   =  cross-sectional area of soil sample
B*  =  bound on slope of permeability plot used in Step 4
i   =  hydraulic gradient
k   =  coefficient of permeability
n   =  number of data points used in the test
Q   =  flow rate of permeant through a soil  sample
Qi  =  bivariate rank of a pair of elements  X^,Yj[
Ri  =  tied rank of an element X^
Sji  =  variance of Kendall's Tau
T   =  Kendall's Tau
T3  =  Kendall's Tau for Step  3
T4  =  Kendall's Tau for Step  4
t   =  time
y-j.  =  Iog10 permeability for  a time t
yt  =  adjusted Iog10 permeability for use in Step 4
Z   =  Mann's Test Statistic
Z3  =  Mann's test statistic for Step 3
Z4  =  Mann's test statistic for Step 4
ZQ  =  critical  value of Z for use in hypothesis testing
                                H-520

-------
•J 1 Q           SAMPLING  AND ANALYSIS PLANS  TO EVALUATE THE  PERFORMANCE OF
                                LEAD-BASED PAINT ABATEMENT


         Benjamin S. Lim. Ph.D., Randy Cramer, Ph.D., Field Studies Branch,  John
         Schwemberger, M.S.,  Design and Development  Branch,  U.S.  Environmental
         Protection Agency,  Washington,  D.C. 20460;  Bruce  Buxton,  Ph.D.,  Steve
         Rust,  Ph.D.,  Bob  Lordo,  Battelle,  2101  Wilson Boulevard,  Suite  800,
         Arlington, Virginia 22201-3008;  Gary Dewalt, Ph.D.,  James McHugh,  CIH,
         Midwest Research Institute, 425 Volker Boulevard, Kansas City,  Missouri
         64110.
         Abstract

         The U.S.  Department of Housing and Urban  Development conducted a lead
         paint abatement demonstration at 169 houses from five metropolitan areas.
         The U.S.  Environmental Protection Agency  plans  to  conduct a follow-up
         study at  these  houses  to measure  levels of lead in dust and soil.  Six
         types of  interior locations will be sampled  for dust.   Three types of
         exterior  locations will be  sampled  for soil.   Dust and  soil  will be
         chemically analyzed for lead.  In general,  soil and  dust  samples will be
         digested  by  nitric acid and hydrogen peroxide,  and analyzed by ICP or
         graphite  furnace AA.   Soil  samples  will be sieved and  dried before
         digestion.  Dust  results will be reported  as  a loading (pg/square foot)
         and  a  concentration  (ptg/g).    Soil  results will be  reported as  a
         concentration  (pg/g).
         Introduction

         In response to  requirements mandated by the Lead-Based Paint Poisoning
         Prevention Act,  as amended by Section 566 of the Housing and Community
         Development  Act  of 1987,  the U.S.  Department  of  Housing  and Urban
         Development  (HUD)  carried out  a  lead paint  abatement  demonstration
         project in FHA re-possessed housing.  The demonstration was conducted in
         five metropolitan  areas across the country.  Single family FHA houses in
         these cities that were owned by the department were tested for lead-based
         paint.  Homes that met certain criteria were chosen for the lead paint
         abatement project.   This  HUD  project  is now virtually  completed.

         Under an interagency memorandum of understanding,  the U.S. Environmental
         Protection Agency is providing technical  support to  HUD  on lead-based
         paint  issues.    EPA  plans to conduct  a  follow-up  study  to  the  HUD
         abatement  demonstration  in  order  to  measure  the levels  of  lead in
         household dust  and exterior soil  in the years following abatement.  The
         purpose  of  the study is to assess  the long-term   efficacy  of  the
                                           11-521

-------
abatement  methods  used  in  the  HUD  demonstration  project.     This
information is needed before  the nation embarks on  a  costly  abatement
program.
Study Homes

In the HUD Demonstration, 169 houses were abated  for  lead paint.   Both
interior and exterior  housing components were  abated.    Six  different
methods of lead paint abatement were used in the project (encapsulation,
enclosure, heat gun stripping, chemical stripping, mechanical stripping,
and component replacement).   The first two methods  cover existing lead
paint, the last four remove it.   An  individual home was  likely  to be
treated by more  than one abatement method.

In order  to  have enough houses  for  the statistical analysis in  the
follow-up   study,   houses   have   been    classified    as    either
encapsulate/enclosure houses or removal houses.  Classification was made
on the basis of  the square footage abated in  the  interior of the house
by the encapsulate/enclosure methods  (encapsulation and enclosure)  and
the removal methods  (heat gun stripping,  chemical stripping, mechanical
stripping, and component replacement).  Interior abatement was chosen for
classification of  houses  because of an a priori assumption that interior
lead paint abatements have the most impact on interior dust levels.

Interior dust and  exterior soil will be collected  at each house that has
been re-sold, re-occupied,  and recruited for the study.   Six interior
locations will be  sampled for dust:  floors, window sills, window stools,
inside entryways, air  ducts,  and upholstered  furniture/rugs/carpets.
Three exterior locations will be sampled for  soil:   outside entryways,
along the house foundation, and near the property line.  The selection
of locations will  be discussed in the next section.
Selection Of Sample Locations

For the follow-up study, two rooms in each house will  be selected for
sampling.   Rooms will be  selected so  that the predominant abatement
method used in the room matches the predominant interior abatement method
for the  house.   In each room,  a  floor  section,  a window sill,  and a
window stool will be sampled. An  air duct will be sampled in each room,
if an air  duct  is present.   In addition, one carpet, rug, or piece of
upholstered furniture will be sampled in each room, pending availability.
Finally, the interior of two entryways will be sampled.

Table  1  summarizes  the environmental  sampling  planned  for  the study,
including  both  regular  samples  (vacuum dust and soil cores) and field
quality  control samples  (wipe dust,  field  blanks,  and  side-by-side
                                 11-522

-------
samples) intended  to assess sampling variability  and  potential  sample
contamination.  As shown in this table, a total of  23 samples will  be
collected from  each house during each  sampling campaign, with a grand
total  of over  6,000  samples  being  collected in  all three  sampling
campaigns.

The objectives of the Abatement Performance Study include both assessing
long-term  performance   of  abatement  methods  and  investigating  the
contribution to interior dust  lead levels from other sources.  The role
of each type of sample listed in Table 5 for meeting these objectives is
as follows:

•     Vacuum dust from floors — Provides primary measure of performance
      for interior  abatement;

•     Vacuum  dust  from  window  sills — Provides primary  measure  of
      performance  for  interior abatement;

•     Vacuum dust  from window  stools  —-  Provides measure of performance
      for  interior  abatement,  possible measure  of  performance  for
      exterior  abatement,  and  possible transport  of exterior soil from
      outside to inside  the house;

•     Rugs,  upholstery,  and  air ducts  —  Provides measure  of  source
      contribution to  interior dust lead levels;

•     Entryway  floor  —  Provides  measure  of  possible  transport  of
      exterior  soil from outside to inside the house;

•     Soil cores — Provides primary  measure of performance of exterior
      abatement, and measure of possible transport  of exterior soil lead
      into the  house;

•     Wipe dust from floors —  Provides consistency check against earlier
      results from HUD Demonstration  and other studies;

•     Field   blanks   —  Provides   assessment  of   potential  sample
      contamination; and

•     Side-by-side  samples   —  Provides   assessment  of  sampling
      variability.
                                  1-523

-------
          TABLB 1.   SUMMARY OF ENVIRONMENTAL SAMPLING PLANNED
                  FOR THE ABATEMENT PERFORMANCE STUDY
                         Samples     Total for       Total for
Sample Type              Per House   One Campaign*  Three Campaigns


Regular Samples

1.  Vacuum dust
    a.  Perimeter floor       2           180            540
    b.  Window sill           2           180            540
    c.  Window stool          2           180            540
    d.  Rug/Upholstery        2           180            540
    e.  Air ducts             2           180            540
    f.  Entryway floor        2           180            540

2.  Soil cores
    a.  Near foundation       2           180            540
    b.  Property boundary     2           180            540
    c.  Entryway              2           180            540

Quality Control Samples

3.  Wipe dust
    a.  Floor                 1            90            270

4.  Field blanks
    a.  Vacuum dust           1            90            270
    b.  Soil cores            1            90            270

5.  Side-by-aide samples
    a.  Vacuum dust floor     1            90            270
    b.  Soil cores            1          	90.            270

    Total samples            23          2070           6210


* As Burning an average of 90 houses  sampled  in each campaign (i.e., 105,
90,  and  75  houses  in  the   first,   second,   and  third  campaigns,
respectively).
                                 11-524

-------
Interior Dust

It is anticipated that results from the Abatement Performance Study may
be compared  with earlier  results from  the  HUD Demonstration and  HUD
National Survey.  For this reason the sampling and analytical methods for
the Abatement Performance  Study  have  been selected to match as closely
as possible the methods used in these  earlier two studies.  The sampling
and  analytical  methods planned for  interior  dust  sampling in  the
Abatement  Performance  Study  are  summarized  in   Tables   2  and  3
respectively.   Some important points  to note in  these  tables are the
following:

•     Sampling will be performed in two different rooms of each house —
      for abated houses this will provide a measure of the variability
      in abatement performance within  a  house, while  for control houses
      this will  provide a measure of the variability  in background lead
      levels within  a house.  Rooms  in  abated houses will be selected
      according  to the  largest  square  footage  abated and the highest
      percentage abated by  the  predominant abatement method  for the
      house.

•     Sampling  will be  performed in each room  separately for floors,
      window  sills,  and window  stools — for abated houses, this will
      provide  a means  to  assess differences in  the  way  an  abatement
      method  may perform  on different  structural components,  and for
      control houses this  will provide a further measure of  the within-
      house variability of background lead levels.

•     Sampling  will  also be performed in each room separately from one
      rug  or upholstered  furniture  piece,  and  one  air  duct; in cases
      where  more than  one such  component is  available  in a room, the
      specific  component  for sampling  will  be  randomly selected  from
      those  available.

•     Vacuum sampling, rather than wipe sampling,  is  the primary method
      planned for interior dust  — as noted earlier,  this  method allows
      for  measurement  of  lead  on  a  concentration  basis  so   that
      comparisons among abatement methods,  houses, and  across time can
      be made,  controlling for  potentially biasing effects  due  to
      variations in the total amount of dust present;  vacuum sampling
      also allows rugs and upholstered  furniture to be sampled.
                                   1-525

-------
                                     TABLE 2.   SAMPLING METHODS FOR INTERIOR OUST
                     OTS Abatement
                  Performance Study
                                                                 HUD
                                                            Demonstration
                                          HUD
                                    National Survey
Sampling device
Sampling Area
                     Vacuum  (Oast rotary pump,
                     modified 37-mm mixed cellulose
                     ester filter cassette)

                     4 square feet (floors, rugs/
                     upholstery) Entire area
                     (window sills, stools)
ro
Samples Collected Total of 12 samples;
                  One window sill (two rooms)
                  One window stool (two rooms)
                  One perimeter floor location
                    (two rooms)
                  One front entryway floor
                  One back entryway floor
                  Two area rug/upholstered
                    furniture
                  Two air ducts
Compositing       Will be determined after review
                  of pilot sampling results
Chubs Thick Baby Wipes
with Aloe (5-3/4x8")
One square foot (floors,
window sills, window
stools)

Three samples per abated
area;
One window sill per abated
  area
One window stool per
  abated area
One floor per abated area
                                                         None
Vacuum (Cast rotary pump, modified
37-mm mixed cellulose ester
filter cassette)

Four square feet (floors)
Entire area (window stools,
sills)

Total of at least 7 samples;
One floor at front (or most
  heavily used) entryway
One floor in wet room
One floor in dry room
Each window stool in wet
   room
Each window stool in dry
   room
Each window sill in wet room
Each window sill in dry room

None

-------
                                        TABLE 3.   ANALYTICAL METHODS FOR INTERIOR DUST
                         OTS Abatement
                      Performance Study
                                                              HUD
                                                         Demonstration
                                                                              HUD
                                                                        National Survey
en
ฃ3
    Sample
    preparation
    summary
Instrumental
Technique

Est. LOQ

Data reporting

QA/QC Notes:
                  Filter digested in HNO-,/
                  H202 Diluted to 25 mL
Graphite furnace atomic
absorption

15 pg/g or 0.15 /jg/sample

ftg/g and fjg/ft2

NIST Buffalo River sediment
(SRM 2704) and Estuarine
sediment (SRM 1646) used
for reference materials.
Wipe ashed at 550-600 C
for 2 hrs. Acid digested
in HNO3/H202 Diluted to 10
mL

Flame atomic absorption
                                                          ^/g/ sample
                                                          No reference material used
                                                          Used side-by-side sampling
                                                          for duplicates
                                                                 Filter digested  in HNO,/
                                                                 H2O2 Diluted to  25 mL
Graphite furnace atomic
absorption

0.15 /jg/sample

/jg/sample

In-house spiked soil used
for reference material

-------
For the exterior, two  aides  of the house  will  be selected at  random.
Foundation samples will be taken one foot  from the foundation of  the
house on the two sides of the house selected.   Foundation  samples will
consist of five equally spaced samples along the side of the house.  The
five equally spaced samples will be composited into a single soil sample.
On the  same two sides of  the house selected  for foundation  samples,
samples will be collected near  the property boundary.  Two randomized
positions along  the property  boundary will  be chosen.   Boundary samples
will consist of  a composite of three  soil samples  collected at  the
vertices of an equilateral triangle with a side length of 20  inches.
Finally, soil samples will be collected outside  the same entryways  for
which interior dust samples were collected.  Entryway  soil  samples will
consist of three soil samples  collected at the vertices of an equilateral
triangle with side length  of  20 inches.
Exterior Soil

The HUD  Demonstration evaluated  the abatement  of  both interior  and
exterior painted  surfaces,  and  in  fact,  for  many  houses  exterior
abatement was the most significant activity performed.  Furthermore,  the
same abatement method might be expected  to perform quite differently on
interior and exterior surfaces.   Therefore, the Abatement  Performance
Study will  evaluate  both  interior and exterior abatement.

If an  abatement method fails to completely  control an exterior lead-based
paint hazard, then the resulting effect  would  most likely be seen as an
increase in soil  lead concentrations close to the  foundation of  the
house.  Therefore, exterior soil sampling will provide the primary means
for assessing the performance of exterior abatement.  In this assessment,
lead  concentrations  measured  in  soil   samples  taken  close  to  the
foundation  will  be compared with those measured in samples taken at  the
property boundary which are as far as possible from the foundation,  and
therefore,  primarily affected  by only background sources of lead, rather
than lead-based paint abatement.  As with interior dust sampling, results
of  soil  sampling  from the Abatement Performance Study  will  also  be
compared with earlier results from the  HUD Demonstration  and National
Survey.  Therefore,  the sampling and analytical methods for soil in  the
Abatement Performance Study have been selected to  closely correspond to
those used  in these earlier two studies.  Those methods are summarized
in Tables 4 and 5, where the following important points should be noted:

•     Soil  samples will be collected both at the foundation of each house
      and at the property boundary — for abated houses this will provide
      a measure of both soil potentially  contaminated by abatement (i.e.,
      at the  foundation)  and  soil contaminated  mostly by  background
      sources (i.e., at the property boundary); for control houses this
      will  provide a measure of the spatial variations in background soil
      lead  levels.
                                   8
                                 11-528

-------
                                        TABLE 4.   SAMPLING METHODS  FOR EXTERIOR SOIL
                         OTS Abatement
                      Performance Study
                                            HUD
                                       Demonstration
                                          HUD
                                    National Survey
    Sampling device
1-inch ID soil recovery probe,
top 0.5 inch of soil is taken.
_   Samples Collected Total of 6 samplest
en
    Compositing
One taken 1 ft from foundation
 (two opposite sides of unit)
One at property boundary
 (two opposite sides of unit)
One at front entryway
One at back entryway
Foundation samples will be a
composite of 5 uniformly-spaced
cores.  Boundary and entryway
samples will be a composite of
3 cores spaced 20 inches apart.
0.75 inch ID tube (0.5
square inch surface area),
top 0.5 inch of soil is
taken.

Total of 4 samples (both
SPR 24x1-1/8 soil probe, top 2-3
cm of soil is taken.
                                                                 Total of 3 samples;
                                                          before and after abatement)t One taken 1 ft from foundation
                                                          One taken 1 ft from
                                                          foundation (all 4 sides
                                                          of the unit)
All samples are a composite
of 5 uniformly-spaced cores
along the length of the
wall.
 (where exterior XRF occurred,
 or if no XRF, then at a wall
 selected randomly)
One taken halfway between XRF-
sampled wall and property boundary
One at entryway

All samples are a composite of 3
cores spaced 20 inches apart.

-------
                                    TABLE  5.   ANALYTICAL METHODS FOR EXTERIOR SOIL
                     OTS Abatement
                  Performance Study
                                            HUD
                                       Demonstration
                                          HUD
                                    National Survey
Sample
preparation
summary
Instrumental
Technique

EST. LOQ

Data reporting

QA/QC Notes:
Sample drying and homogeniza-
 tion Digest 0.5 g using
 HNO3H2O2 Dilute to 50 mL
Inductively coupled plasma
atomic emission spectrometry
pg/gram dry wt

NIST Buffalo River sediment
(SRM 2704) and Eatuarine
sediment (SRW 1646) used for
reference materials
Oven dry, sieve
Oven dry at 105 C for 24
 hrs.
1 gram digested in HNO?
Dilute to 100 mL

Flame atomic absorption
6 pg/g

pg/gram dry wt

Reference material not
specified
0.5 gram digested in HNOi/
H2O2 Diluted to 50 mL
Inductively coupled plasma
atomic emission spectrometry
pg/gram

In-house spiked soil used
for reference material.
                                                          10

-------
      Samples will be collected from two opposite sides of the  house —
      for abated houses this will provide a measure of the variability
      in abatement performance,  while for  control  houses  this will
      provide another measure  of the spatial variations  in  background
      soil lead levels.  In selecting  sides of  the house  for sampling,
      priority will  be  given  to  sides including  the largest  square
      footage abated and the highest percentage abated by the predominant
      abatement method for the house.

      Samples will be collected  immediately outside the front  and rear
      entryways — for both abated and control houses this will provide
      a means for assessing possible transport of exterior lead into the
      house.
Summary

Sampling and analysis methods described in this manuscript are currently
being tested  in a pilot  study.   There  may be changes to  the methods
described after the pilot is completed.
                                11
                                  11-531

-------
"Ml      FURTHER EVALUATION OF THE CAGE  MODIFICATION  TO THE TCLP



          Paul White, SAIC,  8400 Westpark Drive, Mclean,  VA  22102
          ABSTRACT
                The purpose of WA 17 was to further evaluate the proposed cage modification to
          Method  1311.  The  new proposal incorporates the use  of a stainless steel cage for the
          testing of solidified/stabilized waste without prior particle size reduction. Central to this
          issue is whether the cage approximates the level of stress that  a stabilized waste would
          undergo if disposed in a landfill.
                To evaluate the utility  of the cage, wastes were collected from several  waste
          generators, including electroplating operations, secondary lead smelters, and creosote wood
          preservers. The wastes were  stabilized by addition of cement.  The stabilized wastes were
          tested for compaction strength to provide a standard by which to assess the level of stress
          imparted by the cage. It is though that low strength stabilized wastes should be significantly
          degraded while  high strength formulations should be less  degraded.  Extractions  were
          conducted with the cage, and hard plastic bottles to directly compare the level of stress
          imparted by each extraction method.
             It was found that the bottle and cage were equivalent with  respect to the amount of
          degradation observed for low strength stabilized wastes i.e. all were degraded completely.
          High strength wastes showed that the cage was more aggressive than the bottle and that
          waste stability in the TCLP extraction fluid was equally important in predicting the degree
          of degradation of the waste.  One of the stabilized wastes did show a correlation between
          high compaction strength and survivability in the cage.  As the compaction strength of the
          formulations  decreased, the amount of sample degradation increased.
             In general, the proposed cage modification would provide a aggressive challenge to the
          stability of a  stabilized waste without prior size reduction.
                                             11-532

-------
112      COMPARATIVE  STUDY  OF  EPA TCLP  AND  CALIFORNIA W.E.T

                               FOR METALS IN  DIFFERENT MATRICES.


                  G.S.Sivia. M.S.Iskander and J.T. Coons,
                  Hazardous Materials Laboratory. California Department Of Health Services,
                  2151  Berkeley Way.  Berkeley, California 94704.


                     ABSTRACT;

                     EPA implemented the Toxicity Characteristic Leaching Procedure (TCLP) to simulate leaching of hazardous
                     waste and to identify additional characteristics of waste, primarily organic constituents in RCRA waste in
                     September, 1990. But for hazardous waste characterization particularly for metals, California has a Waste
                     Extraction Test (WET) which covers seventeen metals including eight regulated by EPA under TCLP.

                     Comparative studies were carried out to evaluate the TCLP procedure against California W.E.T for leaching
                     of EPA regulated metals (Ag, As, Ba, Cd, Cr, Pb,  Hg,  Se). Originally, EPA incinerator ash sample (EPA
                     Interlaboratory Study XX sample) was used for comparison, and it was found that California WET gave higher
                     results than the TCLP for the regulated metals extracted.

                     Later some of the  actual hazardous waste soil, sludge and liquid samples received  at HML (Hazardous
                     Materials Laboratory), Berkeley, Ca. from different contaminated sites around California were extracted using
                     both TCLP and Calif. WET  procedures and were analyzed by ICPAES (Inductively Coupled Plasma Atomic
                     Emission Spectroscopy).

                     California WET gave consistently higher results for all eight EPA regulated metals in all the matrices tested.
                     Also California WET seems to be a more aggressive test than TCLP, even if the TCLP results are doubled to
                     account for the different extraction ratio which is 1-2O for TCLP and 1-10 for California WET. There was no
                     apparent relationship between soluble metals in TCLP extract as a percentage of W.E.T or as a percentage
                     of total metals in different samples.

                     California Waste  Extraction Test is also applicable to other metals (Be, Co, Cu, Mo, Ni, Sb, Tl, V, Zn) which
                     are  regulated under Title 22 , California Administrative Code, but not under EPA RCRA regulations. When
                     results for these  metals by the two extraction protocols were compared, Calif. WET came out superior than
                     TCLP.

                     In addition, California WET also has advantages over EPA-TCLP procedure because it is simpler in that it does
                     not  require sample digestion after extraction, no pH  measurement before extraction and no pre-selection of
                     extraction solution.

                     MTROOUCTION:

                     The most significant risk from  the hazardous waste results from the leaching of toxic  constituents into
                     groundwater. The EPA designed Extraction Procedure Toxicity Test (EP Tox.) to simulate the leaching of solid
                     hazardous waste co-disposed with municipal waste in a sanitary landfill and to asses the potential  impact
                     of the leachate on ground water contamination. But  since EP Toxicity test has a limited applicability due to
                     short list of constituents , EPA proposed a "second generation"  extraction procedure TCLP as a replacement
                     to address the shortcomings of EP Toxicity. The TCLP protocol includes the  expanded list of regulated
                     contaminants from the fourteen listed in the EP Toxicity protocol to a total of fifty-two which includes eight
                     metals. California has a equivalent  extraction test (WET) for soluble metals under its code of regulation "Title
                     22" which includes seventeen metals. There are many contrasts among these three procedures, which are
                     listed in the figure 2. Maximum  contaminants levels  for seventeen metals are listed in figure 3.
                     This study was designed to compare the extraction efficiencies  of telp and Calif, wet test. This comparison
                     was accomplished in two  ways.  In the  first, the  metal extraction effectiveness of  the two extraction
                     procedures was  evaluated on a  EPA  incinerator ash  for some metallic contaminants (listed in Table 3) and
                                                           I-533

-------
SCHIM additional metal*. The second phase included evaluating the efficiency of the extraction on actual soil
and sludge sample* received at HML. Berkeley, Ca. The effect of sample digestion was also investigated on
extracts from both methods using EPA 3010 digestion procedure (as recommended in tclp protocol).

METHODS:

SAMPLE PRHปARATION:

Incinerator ash sample was very homogeneous, so no sample preparation was done. HML soil samples were
grinded and passed through 10 mesh sieve  to get a homogeneous sample before extraction. The liquid
samples (containing <0.5% nonfilterable solids) were filtered through appropriate filter papers ( 0.45 urn for
wet & 0.6 urn for tclp) while sludge samples were filtered through appropriate filters; filtrate was saved and
solid part  was extracted with proper extraction fluid, filtered and combined  with original filtrate  before
analysis.

TCLP Procedure :

Incinerator ash sample received as a part of EPA Interiaboratory study xx was used in the  preliminary
investigation.  TCLP protocol was followed as outlined in tdp flowchart (Figure 1). In order to select the
proper extraction fluid for tdp . sample pH was determined . The sample pH in reagent water was 5.69, but
after adding 3.5 ml of 1.0 N Hd to the sample solution, heating to 50 C for 10 minutes and cooling, the pH
came down to < 5.0 (pH 4.70). so extraction  fluid #1 was used. Extraction  fluid #1 is  a acetate buffer
which is prepared by adding 5.7 ml of glacial acetic add to about 900 ml deionized water, then adding 64.3
ml of 1.0 N NaoH. and diluting to a volume of 1 liter. The pH of fluid was 4.93.

25 gm of ash sample was extracted in triplicate  for 18 hrs. over a rotary extractor at 30 r.p.m. with  500ml
of extraction fluid. The sample* were filtered through 0.6 um glass microfiber filters (14.2 cm) under pressure
with nitrogen (in Millipore Hazardous Waste Filtration System OM 100). The filtrate for each replicate was
divided in to two portions; one portion was analyzed as such , while the other portion was digested using
EPA 3010 digestion procedure (SW 846. 3rd edition ,1986).  Both the extracts were analyzed  for soluble
metals with ICPAES (Inductively Coupled Plasma Atomic Emission Spectroscopy).
25-50 gms of HML •oil and sludge samples, and  10O ml of liquid sample were used for TCLP extraction. The
above TCLP protocol was followed.
                                     ;(WCTJ:

California wet does not need pH test of the sample before extraction. Wet extraction fluid (dtrate buffer) was
prepared by adding 42.0 gm of monohydrate citric add in 950 ml of deionized water and then adjusting the
pH to 5.0 by adding 50 % NaoH and making the volume to 1 liter. Prior to use in the extraction step, buffer
was deoxygenated by purging with nitrogen gas. 25-50 gms of incinerator ash and HML samples were
extracted with 250-500 ml of citrate buffer on a regular mechanical shaker for 48 hrs. After extraction the
fluid was filtered through 0.45 um filter paper . The filtrate was divided into two  portions; one part was
analyzed as such while other part was digested using EPA 3010 digestion procedure. Both these extracts
were analyzed for soluble metals by ICPAES.

TOTAL META' ft •

Incinerator ash and all other samples were also digested using EPA 3050 digestion procedure (SW 846. 3rd
edition, 1986) and analyzed for total metals for comparison purposes. Yttrium was used as internal standard
in aB samples and standards to compensate for viscosity differences in different matrices before analysis.
Icp was used for analysis of all the samples except Hg.

HP ANALYSIS;

Hg analysis on soil and sludge (total Hg) and undigested portions of TCLP and WET extracts ( soluble Hg)
was done by Cold Vapor technique ( EPA 7470. SW 846. 3rd edition. 1986 ).

QUALITY CONTROL:

Method blank, method spike , matrix spike duplicates, qc check sample, and EPA reference standards (ICAP-
19, ICAP-7. WP-287) were analyzed with each set of samples as a quality control check on analysis. Percent
red, rpd, and % recovery were calculated as a means of determining the precision and accuracy of the data.
                                           11-534

-------
RESULTS:

The results of Incinerator ash sample for total, and soluble metals in wet and tclp extracts are summarized
in figure 4. The mean and % rsd of three replicates are listed . The concentration of total metals in ash
sample varies widely from less than detection limit for selenium to 55000 mg/kg for Calcium. Although there
is 10 mg/kg of Ag in ash sample, but none is extracted in tclp or wet extract. Arsenic and Chromium are
present at about the same level (69.4 and 62.6 mg/kg respectively) as total metals in the ash sample, but
As is not extracted at all and Cr is only 0.2% in TCLP extract, as compared to 5.2% As and 1.3% Cr in wet
extract as a %age of total metal content. In general, soluble metals extracted  by TCLP as a percentage of
total metals varied from non-detected (As. Se, Ag) to maximum of 4.8% for Cd ( Figure.7a), where as wet
extracted quite a higher concentration of As (5.2%), Pb (6.5%) and Cd (7.3%). This data indicates that WET
extracted consistently higher amounts of soluble metals as compared to tclp for all the metals analyzed in
this experiment. Of particular interest are As and Pb which were extracted in significantly higher amounts
by wet  method as compared to tclp. Using soil and sludge matrices besides  Incinerator ash gave  similar
results.  Results of HMUM542 (sludge) and HML# 1543 (soil) samples are discussed here. Both these
samples came from Empire Mine , Grass Valley. Ca. and were high in total As and Pb, and also had Ba and
Cd above STLC ( soluble threshold limit concentration). The pH of the samples was between 5-6. Both these
samples were digested for total metals as well as extracted by tclp and wet methods for soluble metals. The
data has been graphically presented (Figures 8,9). In  order to fit the data to scale, bars for total metal
represents only 10% of total metals concentration (Figure 8) while 5% for As and Pb . and 1% for Ba and
Cd (Figure 9). Interpretation of this data showed the same trend that wet test gave quite higher results than
tclp for both soil and sludge samples for As, Ba, Cd, and Pb. Total Ag, Cr, and Se were less than detection
limit and consequently none was extracted in tclp and wet extracts. Both these samples were high in total
as well  as soluble Hg than threshold limits (figure 3).  Total, wet, and tclp concentrations of Hg in  sludge
sample was 32.9, 2.65, and 0.53 mg/kg while in soil 95.0, 3.21, and 0.39 mg/kg respectively (Figure 8a).
In addition,  studies were carried out to find out if digestion after extraction makes any differences in the
recovery of  metals. A set of soil samples received at HML from a contaminated shipyard in San Francisco,
and another set of sludge samples from metal recycler. Short Scrap Iron and Metal Inc. Redding were used.
These samples were analyzed for total and soluble metals. A portion of the extracts from both the TCLP and
wet extracts was taken and digested with  EPA 3010 ( as recommended  in tclp protocol) and other half
portion was analyzed as such. The results were compared of both digested and  undigested extracts for both
tdp and wet methods (Figure 10,111.The data indicate that there is no differences in digested and undigested
extracts for recovery of any of these metals tested. Although digestion step for the extract is only a part of
tdp protocol, but was tried in the wet method too for these samples . Out  of four soil and two sludge
samples tested in this experiment , none of the metals showed a significant difference in digested and
undigested recoveries.
TCLP results when multiplied with a factor of two  in order to take into account the dilution factor of(1-20)
compared to WET (which is 1-10) in a soil  sample from Orange County Steel and Salvage, Anaheim, Ca.
were still lower than wet (FigureTb). Similarly some of the soil and sludge samples received at HML from
Empire Mine, Grass Valley, Ca. were high in total Pb and As (Figure 8,9), but soluble Pb (wet) extracted in
sludge sample was about 8.3  % of total where as in  soil it was about 5% , although total Pb was about
6000 mg/kg in both the sample matrices. Higher cone, of soluble Pb (wet) in sludge sample may be due to
the fact that part of sludge  sample was liquid which  had more soluble Pb and was recombined with the
extract  after the  solid  portion was  extracted with appropriate extraction  fluid.  Arsenic shows the same
pattern, since total As ranged from 1630-1750 mg/kg in sludge and soil, respectively (Figure 8a), but soluble
As (wet) in  sludge was 1.2% and 0.7% in soil as percentage of total As. In essence the data shows that
both As and Pb as soluble metals ( by wet method)  are present in higher concentration in sludge sample than
in soil,  but  percentage wise Pb is extracted more than As by the same method in the same matrices.
Although same relationship is true in tdp extraction procedure for sludge and  soil i.e Pb is extracted more
by tdp in sludge than soil, but As is higher in soil  sample than sludge though total As amount is the same
in both  matrices.  There seems to be no consistent relationship of wet and tclp soluble metals extracted as
percentage of total metals. The reasons for higher recovery in wet than tdp seems to  be associated with
type of buffer and extraction time. Citrate buffer used in wet is more aggressive than acetate buffer and also
longer extraction  time of 48 hrs in wet than 18 hrs. in tclp might be a factor in solubilizing more metals.
For liquids (containing <0.5 % nonfilterable solids), a HML sample (F1783) which was high in silver did not
show any significant difference between tdp and wet soluble silver (Figure 8a), since the sample was not
subjected to extraction and was only filtered through specified filter papers in each method.

QUALITY CONTROL:

A comprehensive QC guidelines were followed to validate the data  for predsion and accuracy (figure 5,6).
All  the  QC samples show  very  good  precision  and  accuracy. Method blank results donot show any
contamination
                                              11-535

-------
duplicate matrix spikes where each matrix was spiked before digestion or extraction. In method spikes,
reagents or extraction fluids were spiked before extraction or digestion. Method spikes recoveries for tclp
and wet varies from 80-95 % except for Ag . Relative percent difference (RPD) for duplicate method spikes
in both extracts range from 0.2-13 %. except for Ag in tdp which is 24.8 (figure 5). For total metal analysis,
RPD for matrix spike duplicates on incinerator ash sample is under 13 % except for Ag and Ba, and %
recoveries of matrix spikes are in high eighties. Low Ag recovery may be due to addition of Hd in method
EPA-3050. Same may be the reason for bad precision (high rpd) in matrix spike duplicates (total metals) for
Ag. But these deviations do not affect our results because total Ag was below detection limit in all soil and
sludge samples and was present at such a low level in ash sample that it was not extracted in tdp and wet
extracts. Low matrix spike recovery for Ba in total metal  determination may be  due to the precipitation of
Ba as Baso4 in the ash sample, mhouse HML soil qc check sample with known values was also digested and
analyzed and % recoveries varied from 98-114%. To check the accuracy of instrumental  analysis of the
samples, EPA reference standards (lcap-19, lcap-7) were analyzed along with the samples and percent
recoveries ranged from 99-1 10% (figure 5). Matrix spikes  recoveries of Pfa in total and wet extract does not
showup due to high  concentration of Pb present in these samples, and also cone, beyond the calibration
curve (100 ppm) of Icp instrument.

Tdp and wet extracts were also post spiksd at  10mg/kg and 4mg/kg level. Percent recoveries for pre-spikes
in wet ranged from 42-76% while in post spike varied 74-121%. Tdp pre-spike recoveries ranged from 1 1-
83% while post spikes were 76-107%. In general, post-spikes recoveries were good for both extraction
methods, but pre-spikes recoveries were better In wet extracts than tdp.  Precision was good in both tdp
(under 15%) and wet (under 6%) (figure 6). Both the precision and accuracy were better in  Calif, wet than
tdp. For Hg analysis, method spike for wet and tdp gave recoveries 71.6 and 128%, respectively, when
spiked at 1.0 mgykg. EPA WP-287 reference standard (T.V. 0.1 mg/kg) gave 108 % recovery when analyzed
by cold vapor along with soil and sludge samples (Rgure 8a).

CONCLUSION:

Differences exist between tdp and wet methods in terms  of solubilizing metals in different matrices of soil
.sludge, and incinerator ash. Wet gave higher results for all EPA regulated metals and some additional metals
tested. Wet results for metals were still higher even when tdp results were multiplied with a factor of two
to account for difference in dilution factor for both the methods. Also wet  procedure is simpler than TCLP,
that H does not require no pre-selection of extraction fluid,  no pH determination of sample, and also no after
digestion of extract and thus saves lots of total analysis  time for routine samples. Although digestion of
samples after extraction is part of tdp method, but in the samples tested it did not make  any significant
difference in soluble metals recovered whether extract was digested or not, both in tdp and wet methods.
There is no apparent relationship between tdp or wet in soluble metals extracted as a percentage of total
metals in different matrices.
The authors want to thank Emery G. Lee, Public Health Chemist at HML for analyzing the extracts for Hg
analysis by cold vapor technique and for help in preparation of certain slides by photographing the graphs
from the computer screen.
1. Test Methods for Evaluating Solid Wastes: Physical / Chemical Methods, U.S.Environmental protection
  Agency, Office of Solid Waste, Washington, DC, SW 846. Vol. 1A, 3rd edition, Sept. 1986.

2. Federal Register. - Extraction Procedure Toxidty Characteristics" May 19,1980. 45, 33063-33285

3. Federal Register. " Toxidty Characteristic Leaching Procedure "  November 7,1986, 51., 40572-40654

4. Bricka. R.Mark. Teresa T. Holmes and M. John Cullinane Jr.  1988. A Comparative Evaluation Of The
  USEPA TCLP and EP Extraction Procedures. US Army Engineer Waterways Experiment Station, Vicksburg,
  Mississippi 39180.

5. CaBfomia Code of Regulations. Title 22, Vol.29, Article 11,  Sections 66699, 66700, Environment
  Health, p679-681. published by Bardays Law Publishers, 400 Oyster Point Bl, P.O.Box 3066, South San
  Francisco, CA.S4080.
                                            11-536

-------
                  TCLP Flowchart for Metals
Wet Waste Sample  *	
<0.5% non-filterable
  Solids
                         Representative
                          Waste Sample
                             dry waste
Liquid/Solid
Separation:
0.6 to 0.8
Glass Fiber
Filtration
 TCLP
          "discard
          solid
                                 sample
                     	> Sludge
                     Contains >0.5%
                     non-filterable
                         solids
                                               solid
      liquid
                       Reduce Particle Size
                       if >9.5mm in narrowest
                       dimension or surface
                          area < 3.1 cm2
                      Liquid/Solid
                       Separation:
                      0.6 to 0.8 ;x
                       Glass fiber
                        Filtration
extract
                                Pre Screening
                            to Select Extraction
                                   Fluid
                                                                   liquid
                                                              Store at 4ฐ C
                                Liquid Solid
                                Separation:
                                0.6 to 0.8 jra
                                Glass Fiber
                                Filtration
                                  TCLP
                                           -^-discard
                                            solid
                                         liquid
extract
                               -> Analytical
                                I   Methods
                                 FIGURE  1
                                    I-537

-------
             Major  Differences  Among
        the three Extraction Procedures
       WJE.T.

1.  One set  extracting
   solution.    Citrate
   Buffer pH 5.0
2.  Sample to extraction
   fluid ratio is 1:10
       TCLP

Extraction fluid selection
depends on sample pH:
a. Acetate buffer pH
  4.93 ฑ 0.05
b. Acetic acid solution
  pH 2.88 ฑ 0.05

Sample to extraction fluid
ratio is 1:20
       E.P.TOX

One extraction solution:
distilled deionized H2O +
0.5 N acetic acid  to pH
5.0 ฑ 0.2
Sample to extraction fluid
ratio is 1:20
3.  Does  not  specify
   extraction  vessel
   design
4.  Requires use of 0.45
   nm membrane filter
   for  extract  after
   extraction

5.  Uses  mechanical
   shaker for extraction
6.  Extraction period of
   48 hours

7.  No monitoring of pH
   required  during
   extraction
TCLP requires extraction
bottles  made  of  glass,
polypropylene,   high
density polyethylene for
non-volatiles

TCLP requires use of 0.6
to 0.8 /im glass fiber filter
Protocol does not specify
reaction vessel design
Requires rotary agitation
in end over end fashion
at 30 ฑ 2 r.p.m.
18 ฑ 2 hours
No  monitoring  of  pH
required   during
extraction
Requires use of 0.45
cellulose triacetate filters
Allows  either  a
blade/stirred open vessel
or a rotary end over end
agitator

24 hours
Requires monitoring and
adjustment of pH to 5.0
during extraction
8.  Does  not  require
   acid  digestion after
   extraction for metals
Requires acid digestion
after extraction for metals
other than mercury
Requires acid digestion
of extract  for  metals
other than mercury
                            FIGURE  2

                                 M-538

-------
      Maximum Concentration of Metallic
Contaminants for Characteristic of EP Toxicity,
         TCLP,  and California W.E.T.
Contaminant
Arsenic
Barium
Cadmium
Chromium
Lead
Mercury
Selenium
Silver
Antimony
Beryllium
Cobalt
Copper
Molybdenum
Nickel
Thallium
Vanadium
Zinc
                      Maximum Concentration
                              mg/L

                                      5.0
                                      100.0
                                      1 .0
                                      5.0
                                      5.0
                                      0.2
                                      1.0
                                      5.0
                       California Wet Only
                                      15.0
                                      0.75
                                      80.0
                                      25.0
                                      350.0
                                      20.0
                                      7.0
                                      24.0
                                      250.0
                   FIGURE 3
                        I-539

-------
Department of Health Services
Hazardous Materials Laboratory
Inorganic Section
                       California W.E.T. vs T.C.L.P. Comparison Study
                                      EPA Incinerator Ash
                                      Summary of Results
             Total Metals
               (mg/kg)
Ag
As
Ba
Ca
Cd
Cr
Mg
Ni
Pb
Se
Zn
A
9.03
73.8
404
54900
368
58.2
6890
27.2
7290
<3.0
23500
B
8.73
69.8
401
55700
372
64.0
7290
31.9
7390
<3.0
24400
C
13.0
64.6
316
54500
392
65.6
7150
28.6
7320
<3.0
24300
Mean
10.3
69.4
374
55000
377
62.6
7110
29.23
7330
<3.0
24100
RSD
23.2
6.6
13.4
1.11
3.4
6.22
2.85
8.25
0.70
0.00
2.05
    Soluble Metals by
TCLP Extraction  (mg/L)
  Soluble Metals by
California W.E.T. (mg/L)
A
<0.01
<0.03
0.18
604
17.8
0.14
89.8
0.17
40.0
<0.06
402
B
<0.01
<0.03
0.17
593
18.4
0.11
88.7
0.20
37.7
<0.06
407
C
<0.01
<0.03
0.19
597
18.5
0.12
90.0
0.18
37.0
<0.06
411
Mean
<0.01
<0.03
0.18
598
18.2
0.12
89.5
0.18
38.2
<0.06
407
RSD
0.00
0.00
5.56
0.93
2.08
12.4
0.78
8.33
4.11
0.00
1.11
A
<0.01
3.47
1.02
3000
26.3
0.80
241
0.52
496
<0.06
1410
B
<0.01
3.62
0.96
3010
27.7
0.84
241
0.49
472
<0.06
1420
C
<0.01
3.69
1.03
2973
28.1
0.82
238
0.49
457
<0.06
1390
Mean
<0.01
3.59
1.00
2990
27.4
0.82
240
0.50
475
<0.06
1410
RSD
0.00
3.13
3.77
0.64
3.45
2.44
0.72
3.46
4.14
0.00
1.09
                                                                                 Gurmail S. Sivia
                                                                                  September, 1990
                                           FIGURE  4

-------
State of California
Department of Health Services
Hazardous  Materials Laboratory
Inorganic Sectio
'buality Control for yV.E.T. vs  TCLP Study
II

Ag-Silver
As-Arsenic
Ba-Barium
Cd-Cadmium
Cr-Chromium
Ni-Nickel
Pb-Lead
Se-Selenium
Zn-Zinc
Method ''
Blank
<0.01
<0.03
0.01
0.02
<0.06
<0.02
0.08
<0.06
0.04
EPA "
ICAP-19
found
	
0.99
	
1.07
1.04
1.02
1.10
1.06
1.10
true
	
1.00
	
1.00
1.00
1.00
1.00
1.00
1.00
%
Recovery
	
99.0
	
107
104
102
110
106
, 110
EPA "
ICAP-7
found
1.00
	
1.02
	
	
	
	
	
	
true
1.00
	
1.00
	
	
	
	
	
	
%
Recovery
100
	
102
	
	
	
	
	
	
HML ''
Soil QC
found
51.3
47.1
41.7
24.1
49.5
29.7
49.9
54.1
61.1
true
50.9
45.3
39.8
24.3
43.3
30.1
50.7
52.4
55.0
%
Recovery
101
104
104
99.2
114
98.7
98.4
103
111

Spiked Duplicates for Total Metals determinations

Ag-Silver
As-Arsenic
Ba-Barium
Cd-Cadmium
Cr-Chromium
Pb-Lead
Se-Selenium
Unspiked
(mean)
0.21
1.39
7.48
7.54
1.25
147
<0.06
Spike
A
0.53
9.64
11.4
15.5
9.46
*
8.33
Spike
B
0.19
10.5
5.61
16.4
10.1
*
9.47
RPD
94.4
8.54
68.1
5.64
6.54
*
12.8
Spike
added
10
10
10
10
10

10
%Recovery
Mean
1.5
86.8
10.3
84.1
85.3

89.0
Method Spiked Duplicates
% Recovery
(means of 2)
TCLP W.E.T
18.2
95.0
92.0
93.0
88.7
82.7
88.6
67.8
90.8
80.0
80.0
84.1
79.1
92.4
R.P.D.
TCLP W.E.T.
24.8
3.41
4.22
4.86
6.61
9.08
12.9
7.82
0.54
2.26
2.36
0.24
6.75
1.28
  Units are mg/L or mg/kg
  Sample is EPA Incinerator Ash
                               FIGURE  5

-------
Stati of California
Department of Health Servioet
Hazardous Materials Laboratory

Inorganic Section
California W.E.T.
                                                      TCLP vs California W.E.T. Study
                                                             Quality Assurance
Method
Blank
Ag
As
Ba
Cd
Cr
Pb
Se
0.01
0.04
0.01
0.01
0.06
0.02
0.06
Duplicate Spiked Samples
Unspiked
mean
<0.01
3.59
1.00
27.4
0.82
475
<0.06
Spike
A
<0.01
10.2
5.02
32.8
7.58
470
7.45
spike
<0.01
10.7
5.33
33.3
7.73
473
7.79
RPD
0.00
4.78
5.99
1.51
1.96
0.63
4.46
Spike
added
10
10
10
10
10
*
10
% Recoveries
A B mean
0.00
66.1
40.2
54.0
67.6
*
74.5
0.00
71.1
43.3
59.0
69.1
*
77.9
0.00
68.6
41.8
56.5
68.4

76.2
Post Spike
Spike
Result
7.40
13.9
11.1
35.9
10.3
*
12.1
Spike
added
10
10
10
10
10

10
%
Recovery
74.0
102
101
96.0
95.2

121
TCLP
                                                                                                                * beyond calibration of ICP
Duplicate Spiked Samples

Ag
AS
Ba
Cd
Cr
Pb
Se
Unspiked
mean
<0.01
<0.03
0.18
18.2
0.12
38.2
<0.06
Spike
A
<0.01
1.12
1.13
26.8
2.67
43.2
2.70
Spike
B
<0.01
1.06
0.97
26.1
2.48
42.8
2.48
RPD
0.00
5.5
15.2
2.6
7.4
0.9
8.5
Spike
Added
10
10
10
10
10
10
10
% Recoveries
A B Mean
0.1
11.2
9.50
86.0
25.5
50.0
26.4
0.1
10.6
7.90
79.0
23.6
46.0
24.2
0.1
10.9
8.70
82.5
24.6
48.0
25.3
Post Spike
Spiked
3.05
4.02
3.98
21.5
3.95
42.3
4.29
Spike
Added
4.00
4.00
4.00
4.00
4.00
4.00
4.00
%
Recovery
76.3
101
95.2
96.8
95.8
102
107
  Units are mg/L or mg/kg
  Semple Is EPA Incinerator Ash
                                                            FIGURE   6

-------
to
u
o.
o
        California  Wet  and TCLP Results
       as  %  of Total  (EPA  incinerator ash)
o  10.00-1
    7.5O-
    5.00-
    2.50
o.oo
          Aa    As    Ba    Cd   Cr    Pb    Se
           TCLP
                              WET
                                                         WET  and  TCLP  X  2
                                                         Comparison of  Results
                                                  20 -i
      Soil Sample from
Orange County  Steel A: Salvage
        Anaheim, CA
                                                             Cd     Cr
                                                               EUmenl
                                                          b.
                                                                                   IฑJ WET


                                                                                      TCLP X 2
                    Pb
      a.
                                   FIGURE  7

-------
         Total, WET,  TCLP
        Sludge  sample F1542
   in
   •f
Ul
700

600

500

400

300

200

100
        /
Site: Empire Mine
 Grass  Valley, CA
          As-Arsenic     Pb-Lซad
                 Elซmซnt
                            10% of Tofal

                            WET

                            TCLP
                                                        I
Ol
-f

Dl
                                                        ii

                                                        a:
                                                        Total,WET,  TCLP
                                                       Sludge sample F1542
                                                          20
                                                          10
       Site:  Empire Mine
        Grass  Valley, CA
                                                                               (X50)
                                                         Ba-Barlum    Cd-Cadmlum
                                                                Elซmซnf
10% of Total

W.E.T.

TCLP
                                      FIGURE 8

-------
s
en
                                 Comparison Of Total, Tclp, And California Wet
                                             Extracts (mg/kg)
      HML NUMBER
      SAMPLE TYPE
     AS-ARSENIC
     BA-BARIUM
     CD-CADMIUM
     CR-CHROMIUM
     PB-LEAD
     SE-SELENIUM
     AG-SILVER
     Hg-Mercury
              F1542
             SLUDGE
TOTAL
1630
44.4
95.1
< 9.40
5760
< 25.5
< 3.90
32.9
WET
19.6
1.33
0.05
< 0.19
478
< 0.51
< 0.08
2.65
TCLP
< 0.19
0.22
< 0.03
< 0.19
64.2
< 0.51
< 0.08
0.53
                               F1543
                                SOIL
TOTAL
1750
105
121
9.40
6190
25.5
3.90
95.0
WET
11.8
0.89
0.37
< 0.19
311
< 0.51
< 0.08
3.21
TCLP
0.76
< 0.13
0.18
< 0.19
25.7
< 0.51
< 0.08
0.39
      EPA WP-287
      Hg-Standard
                     Quality Control  ( Hg-Analysis )

                       True Value   Result   %  Recovery
                                        Soluble Silver by  "WET" and "TCLP"
                                         (filtered through .45 or .6 - .8 micron)
0.100
0.500
0.108
0.470
      Mtd-Spike(Wet)  Spiked at i.o mg/kg
      Mtd-Spike(Tclp) Spiked at 1.0 mg/kg
108
94

71.6
128
4.00
3.20

2.40
1.60
0.80
0.00
/


/


Liquid Samples
Site: Sierra Medical.
Fresno


v



/



V


/
Em WET
KSSSSj TCLP

FI7S3
SompU Numbtr
                                       FIGURE  8A

-------
       Total,  WET,  TCLP
        Soil  sample  F1543
in
rt
-V.
OI
E
   400
   300
200
         Site: Empire  Mine
           Grass  Vajlev,—-,
     As-Arsenic
                    Pb-Lซad
               Eltmtnl
5% of Total

WET

TCLP
                                                          Total,  WET, TCLP
                                                           Soil  sample F1543
                                                           2

                                                      at
                                                      E
                                                         B
                                                         r.
                                                             ,
                                                            Site: Empire  Mine
                                                             Grass  Valley,  CA
                          Ba—Barium    Cd-Cadmlum
                                 EUmปnt
    1% of Total

l:::::x::l WET
                                     FIGURE   9

-------
     California Department of Health Services
     Hazardous Materials Laboratory

     Inorganic Section
          Comparison Between Digested  and Undigested
                             TCLP  Extracts
HML NUMBER

As-Arsenic
Ba-Barium
Cd-Cadmium
Cr-Chromium
Pb-Lead
Se-Selenium
Ag-Silver
Triple A Hunter's Point, San Francisco (Soil samples)
: C863 C864 C866 C871
D
< 0.19
3.88
< 0.03
< 0.19
1.40
< 0.51
< 0.08
UD
< 0.19
3.82
< 0.03
< 0.19
1.33
< 0.51
< 0.08
D
< 0.19
1.68
0.03
< 0.19
5.37
< 0.51
< 0.08
UD
< 0.19
1.66
0.04
< 0.19
5.31
< 0.51
< 0.08
D
< 0.19
1.33
< 0.04
< 0.19
2.88
< 0.51
< 0.08
UD
< 0.19
1.32
< 0.03
< 0.19
2.82
< 0.51
< 0.08
D
< 0.19
1.07
< 0.03
< 0.19
1.27
< 0.51
< 0.08
UD
< 0.19
1.04
< 0.03
< 0.19
1.20
< 0.51
< 0.08
8
Short Scrap Iron & Metal, Inc., Redding
(Sludge
HML NUMBER :

As-Arsenic
Ba-Barium
Cd-Cadmium
Cr-Chromium
Pb-Lead
Se-Selenium
Ag-Silver
F2541
D
< 0.19 <
1.67
0.18
< 0.19 <
1.28
< 0.51 <
< 0.08 <

UD
0.19
1.68
0.18
0.19
1.21
0.51
0.08
samples)
F2542
D
< 0.19 <
4.26
0.11
< 0.19 <
0.56
< 0.51 <
< 0.08 <


UD
0.19
4.26
0.11
0.19
0.55
0.51
0.08
     Notes: D = Digested, UD = Undigested.  Mean of two replicates reported,
                               FIGURE 10

-------
California Department of Health Services
Hazardous Materials Laboratory
Inorganic Section
                 Comparison Of Digested And Undigested
                            California Wet  Extracts
HML Number

As-Arsenic
Ba-Barium
Cd-Cadmium
Cr-Chromium
Pb-Lead
Se- Selenium
Ag-Silver
Triple A Hunter's Point, San Francisco
( Soil samples )
: C863 C864 C866
D
< 0.19
14.9
0.07
2.88
15.1
< 0.51
< 0.08
UD
< 0.19
14.9
0.08
2.92
15.1
< 0.51
< 0.08
D
< 0.19
7.87
0.08
3.53
21.07
< 0.51
< 0.08
UD
< 0.19
7.65
0.08
3.40
21.0
< 0.51
< 0.08
D
< 0.19
3.24
0.07
1.03
26.2
< 0.51
< 0.08
UD
0.20
3.22
0.07
1.01
26.5
0.51
0.08
C871
D
< 0.19
3.18
0.06
1.04
7.83
< 0.51
< 0.08
UD
< 0.19
3.15
0.07
1.05
7.95
< 0.51
< 0.08
Short

HML Number :

As-Arsenic
Ba-Barium
Cd-Cadmium
Cr-Chromium
Pb-Lead
Se-Selenium
Ag-Silver
Scrap Iron

And Metal
(Sludge
F2541
D
< 0.19
5.62
0.70
0.48
27.5
< 0.51
< 0.08
UD
< 0.19
5.58
0.70
0.47
27.2
< 0.51
< 0.08
Inc. , Redding
samples)
F2542
D
< 0.19 <
24.7
0.75
0.55
17.3
< 0.51 <
< 0.08 <



UD
0.19
24.7
0.78
0.53
17.4
0.51
0.08
Notes: D = Digested, UD = Undigested. Mean of three replicates reported.
                                    FIGURE  11

-------
AUTHOR INDEX

-------
AUTHOR INDEX
Author
                                 Paper
                               Number
Author
                                Paper
                              Number
Author
 Paper
Number
Abdel-Hamid, M.
Actor D..
Alchowiak,J
Allison, J
                                    13
                                   ....1
                                    19
                                    24
                                 *•••ป•{) 1
Amide, E.N. ---------------- 63
Amin,J..._ _______________________________ 103
Anderau, C. .............................................28
Anderson, D. A........... ____ „...„. ----- .......10
Anderson, D. R. --------------------------- 68
Ashraf-Khorassani, M ---- 71
Atwood,R.A..
Austiff,G.A...
Baker, R.D....
Bates, C	
Barone, G..ซ...ซ—
Bath,R.J	
Baughman, K. W..
Beaty,R.D	
Beckert, W. F.	
Bellar,T.A	
Bencivengo, D. J™
Benedicto, J-•
Berges,J.A..
Betowski, L. D.....
Blair, P.O.,
                                  ..100
                                  ..-95
                                  .-52
                                    88
                              ______ 98
                              ----- 101
                              ________ 29
                              ...49,56
                              ------ 44
Bloemen, H. J. Th. ..„
BoUman, M....^.........^
Boyer,D.S..
Breen, J. J.ซ~~.
Broadhead, M..
Brown, J. R. ....
Brown, R. D.....
Bruce, M. L	
Bubnis, B	
Buote, B..
Burnetti,J...
Butler, L.C	
Buxton,B	
Bychowski, J. T..
Calvi,J.P.	
Carter, J..
                              >*•ป•**• •••4D
                                   .49
                              	11,63
                                   ป53
                                  .....87
                                  ...104
                                  ....35
 Chiang, T.C.H....
 Chong, P..
 Coakley.W...
 Colby, B.N..
 Coons, J. T.
 Cornell, J. L..
 Cotter, R.
 Craig, C. A
Cramer, R..
Cunningham,!...
Davis, C.B.	
Denoyer, E. R..
Dewalt,G	
Dfllard,J.W...
Dodhiwala, N...
Doeffinger, J. ...
Dogruel, D. ~~.
                                   -71
                           	37
                           >••••••***•*• ปu O
-------
Author
                                      Paper
                                   Number
                                               Author
                                      Paper
                                   Number
                                               Author
     Paper
   Number
Naser,Z	
Newberry.W.R.....
Nwosu, J................
O'Donnell, A. D	
Offutt, C. K.	.	....
Olbrot, R. M	

Paessun, M. A................
Pan, J. C	
Panholzer, F...................
**in% J* ••••••••••••••••••*••••••••
Persson, J.-A. ...
Petersen, J. ........
Pickering, M.V..
                                    ........14
                                    	35
                                    ........68
                                         90
                          ........................82
                          ....	.	35
                                    	g
                          	59
                          ....................... 79
                                       ..108
                                         77
                                      i...ซ / /
                                      .....51
                                      .....77
                                      k*ปM<3v
                          ,...57,60,81,100
                          	72
Pohcowicz, C	
Pospisil, P. A.
Prashar, S	,
Price,J.
Ray, L.	
Rettberg, T. M.
•Kicimrosoiiy j^* /v. •••••••••••••••••••••••••.•••••••••ซy
Richter,B. E	72
Robertson, G. L....................11,16,26,63
                                       ...34
                                       ...87
                                     ปซซt**v4
                                     	71
Rosselli, A. C	
JCotlini&ii* Wง ••*<*•••••••••••*••*•••**•*••ป•••*•••*•••• Iu5
Rubin, R................................................ซ86
Rust, S. ...•..........••..•..••...••..••..•...•.....•.....110
Ryan, J. F............ป...........................32,48
•MyilUy fป W* ...•.•.••••.••••••••••••••••••^••••••••••••••4O
Rynaski, A		72
Schaleger, L	
Schalk,A.
uCOOIIClCl) iTป Kซ *•*••••••••••**•**ป*•••••••••<
odineif F 't t*. •*••••• ...ป•• ..••..•......•••.•aปi
Schwemberger, J.	110
                                        .65
                                        .79
                                        ,91
                                        .75
                            ......................87
                            ...................... / /

                                     ......... /
                                     ....112
                                     ,...105
Seeley, R. C...
Settle, F. A—
Shah, N. K	
Simes, G. F.
Sivia, G. S	
Sleevi, P. —......
Smith, D. .................................................61
Spafford, R. B.
Spear, R. D......
Spurlin, S. R. ....
Stainken, D. M.
Stanton, L	,
SteIz,W.G.
Stephens, M. W.
Stock, M...
Syhre, D.
Tatro, M. E. ...
Taylor, C.
                                                                                   .......42
Thomas, R.
Thomas, R. J.
Tilbury, M. D.
Towa, B	
Troast,R.
Tucker, E. S.
                                               Turman, K. ................
                                               Unwin,j. P.  .................
                                ..............81
                                	37
                                                                                              VanKley,H.,
                                                                                              Vandermark, T. L.
                                                                                              Vanderveer, E. P..
                                                                                              Vargo,C	
                                                                                              Vasavada, S	
                                                                                              Venna, V. L.	
                                                                                              Villalpbos, K	
                                                                                              Voice, T. C	,
                                                                                              Vonk,N	
                                                                                              Voyksner, R. D. ...
                                                                                              Wakakuwa, J. ......
                                                                                              Walter, P.	
                                                                                              Weesner,F.J	
                                                                                              Weichert, B. A.....
                                                                                              Weitz, S.

                                                                                              Wentworth, N. W.
                                                                                              Weston, A. F.	
                                                                                              Wetzel,D.L	
                                                                                              White, P.  	
                                                                                              Wilborn, D	
                                                                                              Williams, B. —.....
                                                                                              Williams, L	
                                                                                              Winslow, M. G..
                                                                                              Wittwer, T.	.....
                                                                                              Woolfenden, E. .....
                                                                                              Worthington,J.C.
                                                                                              Xiques, D. R..........
                                                                                              Yagley,T.J.	
                                                                                              Zweidinger, R. A.  .
       ...81
       .....7
       ..38
 	69
 	65
 ............ 78

 .............48
 ...........104
 	53,70
 ...5,21,22
 ••••••••••••• / /
 	62

   •••••••••
-------