r ,. I r’ cU
E 1 .

                         printed on recycled paper

                                 TABLE  OF CONTENTS
Paper                                                                                            Page
Number                                                                                          Number


I.        Field QA and Representative Sampling: Findings of a National Workshop. L. R. Williams                   1

2.        Quantitative In-Situ Soil Gas Sampling. G.E. Robitaille, K.T. Lang, K.P. Lucero                           2

3.        Field and Laboratory Methods for a Superfund Ecological Risk Assessment at Milltown Reservoir
         Wetlands. G. Under, M. Bollman, C. Gillett, J. Nwosu, S. Ott. D. Wilborn, G. Henderson, C. Callahan          15

4.        Use of HgI2 Spectrometers in Detection of Lead in Paint. J.S. Iwanczyk, Y.J. Wang, W.R. Graham            28

5.        Immunoassay Methods for Environmental Field Screening. S.B. Friedman                             43

6.        Cost Effective PCB Investigation Utilizing Immunoassay. J.S. Smith, E. Brozowski, I.E. Rhodes              58

7.        Current Status of Lipid Containing Semipermeable Membrane Devices as Environmental Dosimeters.
         J.D. Petty, J.N. Huckins, J.A. Lebo, J.L. Zajicek                                                      74

8.        Development of a Monoclonal Antibody Immunoassay for the  Detection of Polyaromatic
         Hydrocarbons In Soil. R.L. Allen, T.M. Stewart, T.A. Withers. W.B. Manning, S.B. Friedman                 86

9.        The Field Screening of a Large Site for Pentachlorophenol Contamination Using an
         Immunoassay-Based Analytical Method. K.R. Carter                                                87

10.       The Application of Immunoassay-Based  Field Analytical Methods. K.R. Carter                           90

11.       The Assessment of a Site for PCB Contamination Using an Immunoassay-Based Field
         Analytical Method. K.R. Carter                                                                  98

12.       A Mixed Waste Methods Compendium: DOE Analytical Methods for Environmental and
         Waste Management Samples. S.C. Goheen, M. McCulloch, S.K. Fadeff, R.M. Bean, R.G. Riley              105

13.       Analysis of PCB's by Enzyme Immunoassay. R.O. Harrison, R.E. Carlson                               120

14.       Quantification of 2,4-D and Related Chlorophenoxyacetic Acid Herbicides by a  Magnetic
         Particle-Based Solid-Phase ELISA. F.M Rubio, T.S. Lawruk, C.S. Hottenstein, D.P. Herzog, J.R. Flecker      129


15.       Automating Analytical Quality Control on a Local Area Network Based Lims System.
         R.D. Beaty, M.R. Schoen                                                                      140

16.       Performance Evaluation of Analytical Methods for RCRA Regulatory Programs for the
         Petroleum Industry. G. Walters, D. Kocurek                                                      154

17.       The Importance of Selecting Representative Samples in the Development of a Successful
         Treatability Study. LJV. Hines, S.E. Morrissette                                                    175

18. Method Modificiation . ..Or Deviation? Addressing Data Comparability. R.C. Mealy
19. Laboratory Auditing as a Quality Control Procedure to Evaluate Achievement of Data
Quality Objectives. IA. Phillips 197
20. Total Quality Management in the Environmental Laboratory. F.J. Unangst 203
21. Designing and Implementing a Representative Sampling and Analysis Plan for PCBs in
Non Homogeneous Industrial Scrap. D.N. Spels, M.P. Hoyt 214
22. Advances in Data Quality Objectives Guidance for Superfund Remedial Investigation. D. Neptune 229
23. Performance Evaluation Samples—What Do They Tell Us?. G.L. Robertson, J. Lee 230
24. Database Trends in Problems Observed in Contract Laboratory Program (CLP) Inorganic Data Reviews.
KA.Aleckson 231
25. Use of Organic Tape and Data Reviews in Technical and Quality Assurance Oversight of Superfund
Contract Laboratories. J. Armour, V. Dandge, D. Hewetson 232
26. Data Validation Procedures for Dioxin/Furan Data. L I. Dupes 233
27. Cost Effectiveness of Data Quality Objectives in Environmental Monitoring Performance-Based
Methodology. M.Dymerskl 243
28. ENSR ’s Automated Data Validation System. M. Fend, M. Hoyt, J. Kamofsky 244
29. Using Handheld Computers to Improve The Quality of Field Data Documentation and Transmittal.
B. Hill, J. Holley 245
30. Laboratory Automation System for Standard and Reagent Preparation. M. Hummel, L. Lindquist,
R. Doyi, F. Dias, B. Warden 251
31. Cost-Effective Monitoring Program—A Success Story. K. Johnson 260
32. Waste Minimization Program at Environmental Laboratories. H.C. Mehra 262
33. Laboratory Compliance Through Self-Audit. G. L Poftng, R.A. Thompson 263
34. Data Validation Guidance for EPA Organic and Inorganic Analytical Methods. A. Rosecrance 268
35. Use of Performance Evaluation Samples in Assessing Environmental Data Quality. D. Sims Dwight,
P. Zawodny 283
36. Quality Assurance Practices for Data Entry and Electronic Data Transfer. LM. Tomczak 287
37. The Impact of the GC/MS Raw Data Audit in the Monitoring of the USEPA Contract Laboratories.
M.K. Wolf, Di. Elkins 303
39. Status of the RCRA Organic Program. B. Lesnlk 304
40. Determining TCLP Volatiles in Waste Oil at Regulatory Levels. P. Marsden, B.W. Colby, C.L Helms 305
41. Volatiles in Soils. P. Kester 320

42. Vacuum Distillation: An Alternative to the Analysis of Volatile Analytes of Conventional
Purge and Trap. D.R. Youngman, M.H. Hiatt 321
43. Open Tubular Solid Phase Extraction. P. Del Mar 322
44. Accelerated Liquid/Liquid Extraction with Reduced Solvent Volume. M.L. Bruce,
M.W. Stephens, I. Carl, B. Killough, T. Zine, D. Burkitt 335
45. A Rugged and Cost-Effective Method for the Determination of Explosive Residues in
Environmental Water Samples. M.G. Winslow, B.A. Weichert, R. Baker 341
46. A Solid Phase Extraction Disk Method for the Extraction of Explosives from Water.
G. Le Brun, P. Rethwill, I. Matteson 351
47. Analysis of Environmental Samples for Methylmercury Using Gas Chromatography with
Electron Capture Detection. D.K. Root, W.W. Li 352
48. A Practical Means of Measurement of Purgeable Organic Halides at Low Concentrations by
Oxidative Pyrolysis/ Coulometric Titration. J.R. Simon, Jr., J. Lowry, C. Ramsey 365
49. Ion Trap Mass Spectrometry for Pesticide Analysis and Other Applications. M. ZImmerman 376
50. Extending the Detection Limits of Quadrupole GC/MS Systems in Environmental Analysis:
Common Problems and Novel Solutions. L.C. Doherty 385
51. Environmental Applications of Particle Beam LC/MS. DR. Doerge 397
52. The Determination of Semi-Volatile Organic Compounds in Analytical Extracts Using Split
Injection Technique with an Ion Trap GCIMS. R. Brittain, N. Kirshen, E. Almasi 412
53. Analysis of Pesticides in Groudwater Using Atmospheric Pressure Chemical. D.R. Doerge 413
54. The Use of Supercritical Fluid Extraction/GC-MS for the Quantitative Determination of PAHs in Soil.
J.M. Levy 426
55. Supercritical Fluid Extraction (SFE) of Total Petroleum Hydrocarbons (TPHS) with Analysis
by Infrared Spectroscopy. M.L. Bruce, M.W. Stephens 427
56. Using a Vacuum Centrifuge as an Alternative to Kudema-Danish Distillation, Rotary Evaporators
and Nitrogen Blowdown. B.N. Colby, C.S. Parsons 437
57. An lnterlaboratory Evaluation of a Method To Extract Total Petroleum Hydrocarbons from
Solid Samples Using Supercritical Carbon Dioxide. W.F. Beckert 442
58. Concentration of Water Soluble Volatile Organic Compounds from Solidsamples by
Azeotropic Microdistillation. ML Bruce, R. Tomayko, M.W. Stephens 443
59. GC/MS Identification of Artifacts Formed During Sample Preparation Using EPA Methods 625 and 8270.
P.H. Chen, W.A. VanAusdale, W.S.Keeran, D.F. Roberts 447
60. Analyses of 2,4-D,2,4,5-T, and Silvex in Hazardous Wastes and TCLP Extracts Using HPLC. CR. Hecht,
F. Thomas, J.W. Kolopanis 457
61. An Interlaboratory Calibration Study of a Thermospray-Liquid Chromatography/Mass Spectrometry
Method for the Detection of Carbamates in Environmental Matrices. T.L Jones, V. Lopez-Avila 471
62. The Determination of Sub Part Per Billion Levels of Volatile Organic Compounds in Air by
Pre-Concentration From Small Sample Volumes. N. Kirshen, E. Almasi 472

63. Supercritical Fluid Extraction of Chiorophenoxy Acid Herbicides from Soil Samples. V. Lopez-Avlla,
I. Benedicto, N.S. Dodhiwala, W.F. Beckert 483
64. Molecular Profiling Oily Wastes by Thermal Extraction—Gas Chromatography-Mass Spectrometry:
Use in Treatability Studies. C. Sutton 498
65. Isolation of PCDD/PCDF from Complex Sample Extracts With an Automated Modular Liquid
Chromatographic System Using Disposable Prepacked Chromatographic Columns. T.O. Tiernan,
H. Shiridian, D.j. Wagel, G.F. VanNess, J.G. Soich, J.H. Garrett 499
66. The Efficacy of Microwave Hot Acid Leaching for Determination of Metals. L Jassle, E. Hasty,
R. Revesz, H.M. Kingston 500
67. A New Device for High Pressure Microwave Digestion. R. Kramer 509
68. Microwave Digestion of Incinerator Samples. 1.. Yang 510
69. Capillary Ion Analysis: A New Method for Determining Ions in Water and Solid
Waste Leachates. J.P. Romano 519
70. Use of Capillary Electrophoresis for the Determination of Organic and Inorganic Anions in
Ground Water at a Superfund Site. LW. Strattan, M. Siao 520
71. The Determination of Hg and Other Trace Elements in Soil Using Neutron Activation Analysis.
L Robinson, F.F. Dyer, D.W. Combs, J.W. Wade, NA Teasley, I.E. Canton, A.L. Ondracek, J.R. Stokely 532
72. Use of Transportable X-Ray Fluorescence for Determination of Selected Aqueous Trace Contaminants.
R.E. Phillips, R.J. Bath, P.D. Greenlaw, J. Hudek, R.D. Spear 533
73. Comparative Evaluation of Sample Preparation Methods for the Determination of
Metal Analytes in High Concentration Environmental Samples. C. Jones, W.R. Newberry, X. Suarez 534
74. Ion-Exchange Solid Phase Extraction: Factors Influencing Retention and Elution And
Their Application In Method Development. A.J. SpilkIn 535
75. Applications of X-Ray Fluorescence Spectroscopy to Hazardous Waste Analyses. D. Kendall 539
76. Handbook and Database of RCRA Ground-Water Monitoring Constituents: Chemical and
Physical Properties. A.E. Johnson, R. Carlston, J.R. Brown, V.B. Myers 540
77. Rationale for the Design of Cost-Effective Groundwater Monitoring Systems in Limestone and
Dolomite Terranes: Cost-Effective as Conceived is Not Cost-Effective as Built if the System Design
and Sampling Frequency Inadequately Consider Site Hydrogeology. J.F. Quinlan, G.J. Davies 552
78. Releasable Cyanide and Sulfide; Dysfunctional Regulation. J. Loway, J. Fowler, C. Ramsey, M. Siao 571
79. Design Considerations for an Automated On-Line Air Sampling System. J. Ryan, I. Seeley 576
80. Review of Army Research on Well Casings Used in Groundwater Monitoring. LV. Parker, A.D. Hewitt 582


Quality Assurance
Liewellyn R. Williams, PhD, Senior Science Advisor, U.S.
Environmental Protection Agency, EMSL-Las Vegas, Nev 4a 89193
In recognition of an urgent need to share information on
controlling in-field sources of data variability (often the largest
components of variability in environmental measurements), a
workshop was convened April 29—May 1, 1992, in Las Vegas, Nevada.
Th. purpose of the workshop was to identify practical approaches
to, and gaps in our knowledge of sa p1ing, sampling design, and
field quality assurance for environmental studies, particularly as
applied to waste sites, heterogeneous waste streams, and
contpminated soils or sediments. Scientists, •n’;ineers, and
quality professionali from various ag.ncies in the Poderal, State
and private sectors qath.r.d. This paper rspor-ts consensus
findings and workgroup recommendations for issues r.ated to: (1)
the design and implementation of statistically valid sampling
approaches; (2) the application of the “Data Quality Objectives”
process to field operations; (3) training and ed.ucational needs. for
field personnel; (4) quality control and quality assurance
techniques and sample types applicable to field operations: (5)
“best estimator” parameters for total measurement error: and.. (6)
th. complex problems posed by highly heterogeneous environmental
media or atypical distributions of toxicants or ot .er hazardous

2 Quantitative In Situ Soil Gas Sampling
George E. Robitaille, U.S. Army Toxic and Hazardous Materials
Agency, Technical Support Division, CETHA-TS-C, Aberdeen Proving
Ground, Maryland 21010-5401 and D.P. Lucero, Lucero Associates,
18421 Cedar Drive, Triangle, Virginia 22172
In conjunction with a triservice (Army, Navy, and Air Force)
program to develop a cone penetrometer with associated sensors
and detectors, a prototype soil gas sampling system has been
fabricated and functionally tested. The system, referred to as
TerraTrog, quantitatively samples hazardous soil gases and
vapors. TerraTrog can be deployed by a cone penetrometer to
depths of 100 ft far less expensively than drilling monitoring
wells. The device may be permanently implanted or may be
retrieved and deployed at multiple locations using the cone
TerraTrog comprises two modules: an implant of small dimensions
containing a gas-permeable membrane of high diffusion impedance
(located at subsurface levels) and a sampling and calibration
interface with a pneumatic manifold (located at ground level).
Unlike conventional non-quantitative soil gas sampling techniques
requiring vacuum to operate, TerraTrog relies only on soil gas
diffusion for subsurface soil gas collection and a carrier gas
stream flowing at a slight positive pressure for lifting the
sample to the surface. Because the sampling is diffusion-limited
by a membrane of known impedance, the sampling rate and sample
size are independent of soil permeability. Sampling does not
deplete the local soil gas or vapor, guaranteeing the accuracy of
measurements made with the device even after long periods of
continuous sampling. The system has a 15 mm. maximum time-rate-
of - response.
Functional and performance testing has been performed with
trichioroethylene in soil, water, and air, using a Photovac 10S70
portable gas chromatograph. The implant has been demonstrated to
operate as designed, i.e., is diffusion-limited with implant
response directly proportional to external soil gas partial
A major problem in the cleanup process or assessment of sites
contaminated by hazardous waste and toxic chemicals stems from
the lack of information or misinformation regarding site
subsurface characteristics, composition, and aerial and
volumetric extent. Performing a general prospecting survey of
the site hazardous fluids and their mobility or stability is of
significant value in developing preliminary overall containment

and treatment plans (1). A network of relatively low-cost
implanted gas samplers deployed throughout the site’s vadose and
peripheral zones as well as adjacent aquifers and high
permeability strata can be utilized effectively for site
prospecting and characterization. The notion of an implanted
sampler network is a viable concept only if waste
characterization data can be provided quickly and inexpensively
and if the sampler can provide samples of all hazardous soil
fluids and contaminants and can interface at the dump site with a
variety of analyzers or monitors and secondary samplers.
This gas sampler system, called TerraTrog (2) for easy reference,
is described below and addresses the above requirements
satisfactorily, offering features that promote simple, low cost
sampler deployment; minimal soil disturbance during deployment;
minimal sample extraction during each sampling episode, providing
a correspondingly more representative sample of soil gases;
minimal hardware; and small dimensions. The TerraTrog implant
has a 1-in, lateral dimension and can be deployed by cone
penetrometers available commercially (3). In addition, sampler
operation is independent of the soil permeability over a range of
0.1 to 1000 mD, and therefore, quantitative data are obtained for
sandy as well as clay soil types. These operational features
also render the sample obtained independent of sampling chamber
volume, line length, sampling pump head, and corresponding
pressure losses.
F i u r TB AT G S S

TerraTrog comprises two modules: the subsurface implant and the
surface control interface. Figure 1 illustrates the system
deployed in the soil; deployment in groundwater is analogous
exactly. Gases enter the implant at flow rates proportional to
the individual gas partial pressures and the partial and vapor
pressures of dissolved and pure liquids, respectively, regardless
of the soil permeability. The soil gases are lifted to the
surface by the carrier gas stream, which enters the surface
module and flows at a controlled and measured flow rate to and
through the implant and returning to the surface as shown. Soil
gas analysis and monitoring is accomplished by the analyzer
attached to the carrier gas stream return line at the interface.
The analyzer/monitor and carrier gas used are compatible with all
aspects of TerraTrog and the data quality requirements of the
application. In addition, a secondary sampling device (adsortion
tubes, bubbler, etc.) may be attached to the interface, and soil
gas may be collected in batches for subsequent laboratory
analysis. With a sufficiently large carrier gas stream flow
rate, one or more analyzers/monitors and/or one or more secondary
samplers can be attached to the carrier gas outlet of the
interface and can be operated concurrently.
- - ane Gas n
CaLattcn Gas i
_Top eaøer an
Gas Mar to1d
%eta WasneL
‘flyet Seek
Men rw e S t
— Ca raticn Gas
- Gffuser Cap
I Hoadeq
F ig re 2 Sa1p ( ki Iant

The cross-sectional illustration of the implant (Figure 2)
depicts a cylindrical array of eight metal rods approximately 6
in. long, contained within a 1-in, diameter envelope. These are
surrounded externally by a 1-in, diameter, 0.002-in, thick Teflon
tubular membrane. The rods provide mechanical support for the
tubular membrane.
The surface control interface module provides the means to
accurately control the flow of the inert carrier gas. In
addition, there are fittings for the carrier and calibration gas
supply lines and the respective pneumatic lines to the implant.
The carrier gas return line connects to a manifold which can be
connected to an analyzer(s) and/or secondary sampling devices.
Implant operation is based on a flow of soil gases by diffusion
through the semi-permeable tubular membrane of Figure 2 . In
addition, the soil gas flow rate is diffusion limited by the
membrane and consequently independent of the soil permeability.
As carrier gas flows through the implant, the concentration of
the soil gas species at the surface is a ratio of the two gas
flow rates:
[ G] = (Q /Q,) 1o (1)
[ G] = soil gas species concentration in the carrier gas
stream at the interface module, parts per billion
(ppb, v/v);
= soil gas species flow rate into the implant, std
mi/mm; and
carrier gas flow rate, std mi/mm.
The carrier gas flow rate is measured at the surface interface
module. The soil gas species flow rate is the product of the
soil gas species membrane conductance and partial pressure in the
surrounding soil. By lumping the membrane and carrier gas
parameters into the term, y, the soil gas partial pressure is
related to IG] as follows (7,8):
= y (G]
= soil gas species partial pressure in the soil, torr.

The system response time is approximately 100-300 s. For
terratrog operating with a 50-std mi/mm carrier gas flow rate
stream, 0.0625-in inside diameter pneumatic lines, and an implant
50 feet below the surface, the lag time of the system is
approximately 73 seconds.
Equations 1 and 2 describe the soil gas species concentration at
the surface interface for TerraTrog operating in the dynamic
sampling mode, i . e., the operating mode in which the carrier gas
flows continuously through the implant. The implant can also be
used in the static sampling mode, i.e., the operating mode in
which the carrier gas does not flow (Q = 0) for a prescribed
period of time preceding dynamic sampling but flows only after
the equilibrium condition described below is attained. Note that
soil gas flow into the implant will continue, regardless, until
the soil gas partial pressure difference across the tubular
membrane is zero. At this point, the net flow of soil gas into
the implant is zero, and an equilibrium soil gas concentration
internally and externally of the tubular membrane is obtained.
After this equilibrium is attained, the carrier is used to lift
the soil gas accumulated in the implant.
For the initial condition, where the soil gas partial pressure,
P , measured at the surface interface is zero (Ps, the sampler
implant intergral soil gas partial pressure is zero), the time
required to obtain the static equilibrium condition with the soil
gas pressure, i.e., P,g = P , is 7 days (7,8) for the implant
dimensions listed above and a soil gas permeability coefficient
of 8 x 1010 std ml/min-cm 2 -torr/cm (2).
Aside from fundamental system analytical and monitoring
performance requirements, the system design constraints are
established by reliability and service life requirements and
deployment flexibility. TerraTrog reliability corresponds
generally and most importantly with the exigencies of maintaining
the relationship of soil gas species partial pressure, P , and
the measured soil gas species concentration, [ G], described by
equation 2. Adherence to this relationship is predicated on the
design and operational integrity of the tubular membrane and the
pneumatic lines leading to the surface. It is essential that the
soil gas flow into the implant by a diffusion process only,
therefore, the tubular membrane must be free of tears, punctures,
and holes and other pneumatic leaks. Thus, pre- and post-
assembly inspection of the tubular membrane as well as an implant
leak check is required. The tubular membrane must not be damaged
during the deployment and operational processes.

The maximum typical soil gas sample flow rate into the implant is
approximately 0.01 std jzl/min for arbitrary but realistic
conditions. This estimate is based on a membrane material with
= 2 x i0 ml/min-cm 2 -torr/cm at 20°C (9). In a relative
sense, it is a very small sample, yet large enough to produce a
[ G] for many soil gas species within the response range of many
gas phase analyzers/monitors that may be attached to the
There are three important aspects to the relatively small sample
size or flow rate: First, the disturbance to the soil is
minimized; consequently, a more representative sample is obtained
independent of soil fluid conditions. Second, for soil strata in
and around dump sites, the soil gas flow rate into the implant is
diffusion limited by the tubular membrane and is independent of
the gas permeability of the surrounding soil. Thirdly and most
importantly, these conditions lead to a quantitative measurement
of the soil gas partial pressure.
The soil gas flow rate is proportional to the soil gas species
pressure only, without regard to the form of the sample, i.e.,
gas phase, liquid phase, or dissolved gas/liquid phase. For
example, the implant can obtain information regarding dissolved
trichioroethylene (TCE) in water or TCE saturated in water, and
insoluble gases contained in the water. Furthermore, the implant
also functions as described immersed completely in an aquifer or
other body of water or liquid.
In a relative sense, the actual analysis of the transport gas
output stream from the interface panel is the most simple and
direct procedure of the entire system. A variety of analyzers,
monitors, and secondary sampling devices can be used singularly
or simultaneously. The user, however, must establish preliminary
requirements for the target species and the lower detection
limits of the analytical devices contemplated, i.e., it is
essential to consider the analyzer performance specifications to
specify and adjust the operating conditions of TerraTrog
Laboratory Performance
Representative data for the TerraTrog time-rate-of-response in
static sampling mode to dissolved phase TCE in water is shown in
Figure 3. The internal concentration reaches equilibrium with
the external TCE concentration in 7 days (168 hr.).

1_. . ?t . Is”. iIQ. .1 51.1
Figur. 5 Implant I.apons . to TCE in Air
t.rr.Tz .s Ill 003-ZOOS
! 21 (21 .sl )
3.11 ft... (a C l . , 1st
S Li. Ibd . D m i.
C.gri.. G .s Air
20s1.ys.t ICsss
/ I I
1 2 3 4 3
I .. . . .. C.rst.r 01.. 5 . ... if QI. .151.1
Figur. 4 Implant Reapona. to TCE in Soil
Figure 6 Implant Responae to TCE in Vat.r
1 4C
T.rr.2ro 3/l
S Lir ftd.
IC! 20 (20 00 /1 )
is lI.t . 22 C
?bot. .a.. 10370
71. .. 1. hE
Figure 3 Implant Time-Rate-of-Response
to ICE in Water
1 ... C.sst.. ?1 l .. IIQr. .J.sl.i
?.m?00S SfL 051-2042
T 221 p (12.4 5/s3)
leAt.. 22 C
S iIOS ft04 . D I 5
C. Ms
...iy.er, sss 10570
T. g.T .s IlL
£ 5si 0 5 0tr
S pU . . ftd .:
C..xl.t GO .:
* 0 .1 7 w :
005-2 10 1
! 100 — (100 COIL)
(5 Vt* . 22 C
0 e
0000C 10570
5 .
i ... .. C.rrt.. fl.r 5.4.. 11g.. .151.4

Representative dynamic sampling mode data for the implant
response, [ G], to TCE as a function of inverse carrier flow rate,
l/Q 0 , is shown in Figures 4, 5, and 6. A multipoint calibration
curve, Figure 7, shows the relationship between implant response,
tG), and external dissolved phase TCE concentration in water.
Figures 6, 7, and 8 show the implant response is linear
regardless of whether the implant is deployed in soil, water, or
air. Additionally, the flow rate of TCE into the implant, Q 1 , is
constant when the implant is sampled in dynamic mode in an
environment of constant external TCE concentration. The sample
flow rate, Q,, is relatively small, ranging from 9.2 x 10 std.
l/min for dissolved phase TCE at 100 ppbm external concentration
in water, to 3.7 xlO 3 std. zl/min for gas phase TCE at 227 ppmv
external concentration in air. Figure 7 shows that the implant
response, and hence Q , varies linearly with the external TCE
concentration and therefore with the external TCE partial
pressure. The data demonstrate that Q 1 is dependent only on the
permeability, of the implant gas permeable membrane, and the
external TCE partial pressure, or concentration (6,7).
Therefore, the implant operation is diffusion-limited by the
implant gas permeable membrane and the implant response is
directly proportional to the external TCE partial pressure in
soil, water, or air, exactly as described by equations 1 and 2.
The multipoint calibration curve, Figure 7, can be used with the
implant to directly measure the TCE concentration in contaminated
groundwater in the field. For example, a user would deploy the
implant to the desired depth in a monitoring well or other body
of water and establish a carrier flow rate of 40 mi/mm. The
implant response would be measured using a conventional gas-phase
TCE analyzer, such as the Photovac 10S70. The implant response
would then be located on the vertical axis of the calibration
curve, and the corresponding TCE concentration in the groundwater
read off the horizontal axis of the curve.Since the implant
response is shown to be linear in the multipoint calibration
curve of Figure 7, it may be replaced with a single point
calibration which yields a linear calibration factor, K:
K = IC ]/ [ G)
K = implant linear calibration factor, ppbm/ppbv
[ Ce) = external TCE concentration, ppbm
[ G) = implant response, ppbv
The linear calibration factor may be used exactly as the
multipoint calibration curve to make direct field measurements.

T erraTres S/I 003-2006
It Utter. 22 C /
re StIed 1’
S pUes M.d. Dyti.
12 terrier 0.. Air
C.rrt.r 0.. FL. .. 8.1 . 0 .I/j
An. Ly e.r Thet .. 10370
I L t C .Libr.tj .. FreIer E
I t S
/ Z i..t leepes..,
,_ / — 12.82
10 3 )0
Vat., T a Cwtrnt.a. pp
Figure 7 Callbration Curve for I p1ant Response
to TCE in Water Versus TCE Concentration
Preliminary field charterization work was performed with
TerraTrog deployed in groundwater wells at the U.S. Army Phoenix
Nike Site, Baltimore, County, Maryland and in compacted soil at
the Department of Energy (DOE) Savannah River Site (SRS).
TerraTrog was used at both sites to sample primarily for
trichioroethylene (TCE).
Groundwater Wells
Two terraTrog implant modules were deployed in groundwater wells.
Each implant module was suspended approximately 40 ft in the well
2 to 3 ft below the water level over periods of several months.
One unit developed leaks through punctures in the membrane and
was repaired on several occasions. The implant modules were all
leak checked prior to deployment and it was surmised that the
membrane was damaged during deployment. After several attempts
the implant module was deployed successfully.
The learning process continued in analytical attempts to analyze
the groundwater for TCB with a portable GC (Photovac model 10S70+
with a CPSi5CB-lOm by 0.53mm column) and with sorbent sampling
tubes. Analytical success was finally achieved with the portable
GC in that dynamic and static measurements were obtained and the
results of both measurements correlated well. TerrTrog used a
Helium carrier gas at 15 ml/min. The GC used an Ultra Zero Air
Carrier at 5 mi/mm with the oven temperature at 40C and was
calibrated at 92 and 900 ppb. A 500ul syringe was used to obtain
the discrete samples with the TerraTrog surface module and inject
it into the GC. The Helium flow was shutoff after each sample as
taken and started again to obtain the discrete samples from the

TerraTrog surface module and inject it into the GC. The Helium
flow was shutoff after each sample was taken and started again to
obtain the next sample. The TCE retention time was 142s. Nine
separate samples were taken over a 40 minute period as shown
below and in figure 8.
TIME (mm)
CONO (ppb)
(pp .)
1.5 3.0 4.5 8.0 9.0 15 25 34 40
256 982 1280 1080 1040 497 145 91 119
1400 -
The total volume of the implant module and the return line to the
surface module is approximately 100 ml. Approximately 6.7
minutes are required for one TerraTrog volume exchange. Thus,
the static sample “peak” should appear between 2-6 minutes as
determined by TerraTrog volume and the carrier gas flow rate.
The data above shows a broad peak at 4.5 minutes tailing off
slowly for about 15 minutes and attains a steady state level near
From the static data considerations, it is determined that the
groundwater headspace TCE partial pressure is at least l26Oppb.
A simple plot of the data and an extrapolation to the peak yields
a concentration near l3BOppb.
From the dynamic data considerations and the calibration curve of
Figure 3 corrected for the reduced TerraTrog carrier gas flow
rate, i.e., 15 mi/mm and K = 16.) it is determined that the
groundwater TCE concentration is 1.46ppmm. At this stage of the
field testing the dynamic and static measurements correlate well.
10 20 30 40 60
Figure 8 ICE Concentration versus has

Compacted Soil
Deployment and retrieval experiments were performed to assess the
implant field worthiness and to obtain TCE samples for on site
analysis by a portable GC (Photovac Model lOS7O with a CPSi5CB-
lOm by 0.5mm column) operating with 10 ml/min of Ultra Zero Air
and an oven temperature of 25C.
The implant was deployed in sandy-gravely soil to a 3ft depth and
retrieved 20 hr later. Inspection of the implant-tool assembly
showed accumulation of dense compacted soil containing gravel on
the conetip shoulder. Small rocks (0.1 to 0.05 in. average
diameter) with sharp edges were lodged adjacent to the membrane.
A post-deployment leak check of the implant-tool assembly showed
a small leak at the base of the TerraTrog implant. Upon removal
of the implant from the tool, a small rock (0.05 in. diameter)
was found to have punctured a small hole in the membrane.
The implant was deployed in grassy clay-like soil to a 4 ft depth
and retrieved 14 hr later. No leaks in the membrane were
detected and the implant was fully functional. A TCE response
was obtained with 40 ml/min TerraTrog carrier gas flow rate.
Inspection of the implant-tool assembly showed an accumulation of
moderately dense, compacted sandy-clay soil that was visibly
moist on the shoulder of the penetrometer tip.
The implant was deployed a third time in highly compacted sandy
soil to a 14 ft depth and implanted permanently. Twelve hours
after deployment the GC was connected to the TerraTrog surface
module and 0.O5ppm TCE level responses were obtained from the
carrier gas. From these data it was estimated the TCE headspace
partial pressure was 10 to 2oppm.
A second set of field experiments were performed 22 weeks later.
The sample analysis was performed in real time with an ion trap
mass spectrometer supplied and operated by the Oak Ridge National
Laboratory (ORNL). In addition, discrete samples were taken by
in-line sorption tubes developed by ORNL. Static samples were not
obtained because of an operational error made in connecting the
TerraTrog sample return line to the real-time monitor sample
inlet fitting. However, soil gas dynamic samples of TCE,
Benzene, Toulene and Xylene were obtained and analyzed with the
real time monitor as were discrete results correlated with 10%
for each species. However, it was possible to estimate the soil
headspace concentration for only TCE and Benzene because the
implant gas conductance for the remaining species has not been
determined. It was estimated that the TCE and Benzene soil
headspace partial pressures are 62.8 and 13.2ppmv respectively.

The development of the TerraTrog is viewed as having potential
for future use in the evaluation of hazardous waste sites. The
potential utility of the device includes not only initial site
assessment, but possibly of more importance, its use in the
routine monitoring that is essential to the long term assessment
of a site before, during and after remedial activities are
accomplished. Although initially designed to be used in a cone
penetrometer, the utility of the device for routine groundwater
monitoring is also recognized due to its small diameter and
ability to descend down standard well casings.
ACKNOWLEDGEMENTS: Mark Wise, Ceril Thompson and Roger Jenkins
from Oak Ridge National Laboratories for their assistance in
collecting data at the Savannah River Site. Mark Kiuger and Steve
Kane from Viking Instruments for their assistance with data
collection at the Phoenix Nike site.
(1) S.C. Golian, E.E. Dodge, and B. Bixier, “Conducting Remedial
Investigations on Feasibility Studies Under CERCLA,”
Superfund ‘88 Proceedings of the 9th National Conference ,
Hazardous Materials Control Research Institute, Silver
Spring, MD, p. 1, 1988.
(2) K.T. Lang, D.T. Scarborough, M.E. Glover, and D.P. Lucero,
“Quantitative Soil Gas Sampler Implant for Monitoring Dump
Site Subsurface Hazardous Fluids,” “Proceedings of the Second
International Symposium f or Hazardous Waste and Toxic
Chemicals,” Las Vegas, NV, 12-14, February 1991, p. 423.
(3) L.R. Taylor and N. Berzins, “Subsurface Contamination
Screening by Combined Soil Gas/Groundwater Survey
Procedure,” Superfund ‘88 Proceedings of the 9th National
Conference , Hazardous Materials Control Research Institute,
Silver Spring, MD, p. 158, 1988.
(4) R.M. Barrer, Diffusion In and Through Solids , Cambridge
University Press, London, Chapter 1, 1981.
(5) D.P. Lucero, “Performance Characteristics of Permeation
Thbes,” Analytical Chemistry 43 , p. 1744, 1971.
(6) D.P. Lucero, “Ultra Low-level Calibration Gas Generation by
Multi-Stage Dilution Techniques,” Calibration in Air
Monitoring , ASTM STP 598, merican Society for Testing
Materials, Philadelphia, PA, pp. 301-319, 1976.

(7) D.P. Lucero, “Soil Gas Sampler Implant for Monitoring Dump
Site Subsurface Hazardous Fluids,” Proceedings of the 6th
National Conference on Hazardous Wastes and Hazardous
Materials , New Orleans, Louisiana, April 12-14, 1989.
(8) DP. Lucero, “A Soil Gas Sampler Implant for Monitoring Dump
Site Subsurface Hazardous Fluids,” Hazardous Materials
Control ., 5, p. 36, 1990.
(9) A.D. Little, Development of a vapor Phase Calibrator for the
APIMS-1 Analyzer , ANXTH-TE-TR-84305, U.S. Army Materiel
Command, Aberdeen Proving Ground, MD, September 1984.

Greg Linder , Mike Boilman, Chris Gillett, Julius Nwosu, Suean Ott, David
Wilborn, ManTech Environmental Technology; Gray Henderson, University of
Missouri; and Clarence Callahan, US Environmental Protection Agency, Environmental
Research Laboratory, 200 S.W. 35th Street, Corvallis OR.
ABSTRACT : Within ecological contexts, biological evaluations in the field and
laboratory should be considered critical components in the ecological risk assessment
process for Superfund, since integrated approaches to hazard evaluation consider
contaminant bioavailability and subtle expressions of adverse biological effects associated
with chronic exposures. Depending upon habitat type, field and laboratory methods have
been developed for hazard evaluation which lend themselves directly to the Superfund
ecological risk assessment process. For example, wetlands in mining districts in the
western U.S are frequently impacted by heavy metal-laden sediments. The present study
summarizes a baseline ecological risk assessment completed by U.S. EPA Region 8 for
a Superfund site located at Militown Reservoir wetlands (MRW) in western Montana.
As part of the ecological risk assessment, a variety of biological test methods [ e.g.,
terrestrial and aquatic testsl were critical to the wetland evaluation. For evaluating heavy
metal effects at MRW, various field and laboratory methods were included as part of the
ecological assessment. These included:
• Preliminary food-web contamination survey
• Vegetation tests in emergent and upland habitats
• Heavy metal uptake studies using plants
• Soil macroinvertebrate tests
• Sediment macroinvertebrate studies
• Studies using amphibians
• Studies using bacterial tests
• Soil and chemical analysis
In conjunction with chemical analyses, these biological and ecological evaluations yielded
an integrated evaluation of ecological effects and exposure at MRW. The data gathered
from laboratory and field work at MRW has reduced the uncertainty associated with the

baseline ecological risk assessment, and suggested that biological and ecological effects
were subtle in their expression in the wetland.
Milltown Reservoir and its associated wetlands are located on the Clark Fork River in
western Montana, six miles east of Missoula, Montana. The reservoir was formed in
1907 following the construction of a hydroelectric facility located on the Clark Fork
River immediately downstream from its confluence with the Blackfoot River. Since
construction of the dam, a wetland habitat has been created. Because of the upstream
mining activities on the Clark Fork River, however, Milltown Reservoir has accumulated
a large volume of heavy metal-laden sediment. The Milltown Reservoir wetland was
initially identified under CERCLA [ Comprehensive Environmental Response,
Compensation, and Uabiity Acti in 1981 after community well-water samples were
found to have arsenic levels that ranged from 0.22 to 0.51 mgIL; the EPA
recommendation for potable water supplies suggested that arsenic not exceed 0.05 mgIL.
Within an ecological context, however, the impact of the contaminated sediments on the
wetlands was unclear; the laboratory and field investigations evaluated the extent of
contaminant and its impact on the indigenous wildlife and vegetation characteristic of the
site. For the wetlands evaluation, heavy metals appear the most critical contaminant
which must be considered in the ecological assessment at Milltown; those of primaiy
interest include arsenic, cadmium, copper, lead and zinc (Woessner, t al. 1984).
Assessment strategy. In general, an ecological assessment may be considered an
integrated evaluation of biological effects derived through measurements of ecological
effects and exposure (U.S. EPA 1988; 1989), and an overview of the assessment strategy
used at MRW during field and laboratory operations is summarized in Figure 1. Within
the context of Superfund ecological risk assessm , chemical and biological interactions
associated with contaminant exposures in soil or sediment may be evaluated according to
various assessment strategies. For example, both chemically-based and toxicity-based
approaches have made significant contributions to ecological assessments for hazardous
waste sites (Parkhurst, al. 1989). From an ecotoxicological perspective, ecological
effects and exposure assessments are complex interrelated functions which yield estimates
of h ard associated with environmental contaminants in various matrices sampled at a
site. Ecological assessments at Superfund sites reflect the site-specific demands required
by waste sites, and represent integrated evaluations of ecological effects, induding
toxicity, and exposure. Depending upon the site, both laboratory tests and field methods
will be required for an ecological assessment.
Toxicity assessments are derived from acute tests as well as subacute and chronic tests
which measure biological endpoints other than death, and are generally completed as part
of the ecological effects assessment. These toxicity assessments are usually derived from
laboratory-generated data, but j gu toxicity assessments, while not as well developed

Figure 1. Ecological effects and exposure assessments as components of the ecological
assessment process for h27ardous waste sites (US EPA 1991).
• Qualitatively evaluate contaminant release, migration, and fate
• Identify;
- Contaminants of eco’ogical concern - Exposure pathways
• Receptors - Known effects
• Select endpoints of concern
• Specify objectives and scope
• Quantify release, migration, and fate ASSESSMENT
• Characterize receptors • Literature
• Measure or estimate Toxicity testing
exposure point concentrations • Field studies
• Current adverse effects
• Future adverse effects
• Uncertainty analysis
• Ecological significance

as laboratory toxicity tests, are becoming more prominent in the ecological assessment
process (Warren-Hicks, et a!. 1989). In zLu methods more closely infer a linkage
between toxicity and exposure functions, and reduce the problems associated with lab-to-
field extrapolations of toxicity data. Ecological effects assessments also rely upon field
methods which measure ecological endpoints, either on-site or at reference sites, and yield
swvey data relevant to estimates of adverse ecological effects associated with a waste site.
Depending upon the environmental matrix being tested, site-specific toxicity assessments
may be derived using various tests( .g., Horning and Weber 1985; Peltier and Weber
1985; Weber, al. 1988; Greene, 1. 1988), and may include invertebrate and
vertebrate, algal, plant, and microbial test systems. These toxicity assessment tools
potentially yield information regarding acute biological responses elicited by site-samples
or their derivatives, and may suggest longer-term biological effects (e.g., genotoxicity
or teratogenicity) potentially associated with subacute and chronic exposures to complex
chemical mixtures characteristic of h rdous waste sites. Linkages among chemical
contaminants, adverse ecological effects, and ecological toxicity (Parkhurst, al. 1989;
Stevens, al 1989) may subsequently be evaluated during the site-assessment process
using quantitative or statistical methods (Figure 2). When toxicity assessments are
combined with (1) chemical analyses which evaluate pertinent site samples and (2) field
surveys which measure ecological endpoints, higher level biological organization (e.g.,
populations and communities) ecological risk assessments may be derived with less
uncertainty to assure that sound ecological management practices are implemented during
the remedial investigation/feasibility study process.
MRW assessment methods. For an ecological assessment at MRW, various tasks were
completed during the preliminary (Eli 199(k) and definitive year’s work. These ta .ckc
• Preliminary food-web contamination survey [ field and laboratory methods]
• Vegetation evaluations [ laboratory root elongation tests on groundwater;
emergent and upland vegetation tests in field and laboratory]
• Earthworm evaluations [ field and laboratory methods]
• Preliminary and scoping year studies using amphibians [ field and
laboratory methods]
• Preliminary and scoping year studies using bacterial tests (laboratory tests]
• Plant uptake studies [ field/greenhouse methods]
• Plant fluorescence evaluations [ field and laboratory methods]

• Soil characterization and chemical analysis Isoils, sediments, and biota;
laboratory methods]
Whether completed during the preliminary or definitive year’s studies, the data gathered
from laboratory and field work at MRW has been summarized as a contribution to the
baseline ecological risk assessment.
Figure 2. Sources of information (toxicity, chemical, and
ecological) that contribute to an ecological assessment.
Toxicity data Chemical data Ecological data
‘V V V
Statistical or Quantitative
Contribution to ecological
risk assessment
Preliminkry Food-Web Contamination Survey Ifleld and laboratory methodsi. In
preliminary food-web analyses for small mammals at MRW, there appears to be no overt
indication of metal biomagrnfication in the small mammal community at Miitown.
Additionally, in support of screening level risk calculations, the comparative literature
suggests that no overt problem is indicated by tissue metal residues in either carcass or
selected organs of the herbivores collected at Militown during the preliminary field
season. Observations regarding health and status of resident small mammal populations
was also considered in the evaluation of small mammal risks associated with metal
exposures. Upon gross examination, external features were generally unremarkable in
all animals collected at MRW.
From a similar food-web contamination evaluation for fur-bearers (beaver and muskrat)
the comparative data for metal loading in terrestrial and semi-aquatic mammals suggests
that no adult muskrat or beaver would approach critical whole-carcass burdens, although
no empirical site-specific data were collected at MRW to support this position. As with
herbivorous small mammals, no target organ metal loads were considered quantitatively
in these screening calculations. If target organs were considered (e.g., kidney), metal
residues could be relatively greater than in carcass. But, small mammal data collected
at MRW do not suggest that target organ toxicity or biomagnification are being
expressed. Again, calculated metal concentrations in muscle tissue are relatively low
which may suggest that the metal bioavailability in plant tissue (e.g., cattail tuber) is
relatively limited, or that exposure is minimized due to selective feeding.

Vegetation Evaluations [ laboratory root elongation tests on groundwater; emergent
and upland vegetation tests in field and laboratoryl. During the scoping year’s field
and laboratory efforts, seed germination testing was completed using site-soils. Both on-
site and laboratory methods were employed in these preliminary soil contamination
evaluations, and no overt expressions of phytotoxicity were observed. During the
definitive year’s studies, rather than continue work with these two phytotoxicity test
methods, alternative test methods were used in evaluating MRW. One, groundwater
samples were collected at selected sites across the wetland and were tested using the root
elongation procedure. Two, emergent zones in the wetlands were evaluated using
laboratory and field methods that tested native marsh plants as well as a standard
submerged aquatic vascular plant ( Potamogeton pectinatus and a standard test species
Hydrilla verticillata) .
The root elongation tests that were completed on groundwater samples collected at MRW
were consistently not acutely toxic, suggesting that groundwater coincident with the
rhizosphere would generally not be overtly phytotoxic. These groundwater samples were
collected from deposition zones in the braided stretches of the Clark Fork River, and
while generally not acutely toxic, some statistically significant biological activity was
noted with respect to root elongation inhibition, however, at some sampling locations at
MRW. These groundwater samples were clearly inhibitory to root elongation as
measured using the standard test species. Field surveys completed in conjunction with
the sampling and laboratory testing, however, did not suggest that these biological effects
were currently being expressed at these sampling locations.
In addition to root elongation testing for groundwater evaluations, field and laboratory
testing was completed with an indigenous emergent vascular plant, sago pondweed
( Potamogeton pectinatus) , as well as a standard test species Hydrilla verticillata during
the definitive year’s operations. In general, emergent vegetation testing in laboratory and
field suggested that effects, when expressed, were not acute but sublethal in expression.
The jfl testing with sago pondweed ( . pectinatus ) indicated no adverse effects for
growth endpoints (root and shoot length), and physiological markers (peroxidase activity
[ PODJ) indicative of plant stress were similarly unremarkable. Indigenous plant samples
( Elodea sp.) were also collected concurrent with ifl iiu testing and were analyzed for
POD activity. Statistically significant differences among sites were noted across MRW
sampling locations which may be indicative of general plant stress. In parallel laboratory
exposures, no consistent pattern was noted with respect to growth endpoints (root and
shoot lengths and chlorophyll a) for either test species (E. pectinatus and if. verticillata )
when tested with bulk sediments collected from MRW sites. While morphologic
endpoints related to growth (e.g., shoot and root length) suggested no acute toxicity in
laboratory or field exposures with MRW sediments, differences in POD activity across
MRW may reflect the spatial variability in metals that are differentially bioavailable.

Root elongation tests on soil eluates. Eluates prepared from site soil samples were
tested using root elongation as the toxicity endpoint. While qualitative differences were
observed with respect to the inhibition of root growth, no soil-derived eluate expressed
statistically significant results following 120-hr incubation.
Earthworm evaluations [ field and laboratory methodsi. Earthworm tests - - both on-
site and laboratory -- expressed no acute toxicity. Subtle biological response data (e.g.,
morphologic and dermatopathologic effects) from laboratory and field tests were
frequently expressed in soils with elevated total metals, however. For example, on-site
testing with earthworms suggested that soil from some sampling locations was associated
with sublethal effects in exposed earthworms. These sampling locations had also been
identified in the root elongation tests completed with groundwater samples collected via
hand-driven well points. Again, no acute toxicity was expressed, and differential
bioavailability of metals may be considered a possible source of these spatially variable
expressions of subacute and chronic effects.
Preliminary and scoping year studies using amphibians [ field and laboratory
methods . Laboratory tests completed on Militown surface water grab samples collected
during the preliminary year expressed spatially variable, but sampling-location consistent
results as summarized by ETI (1991c). No overt toxicity was expressed by 96-hour
tadpoles when exposed to site samples in the laboratory, and in conjunction with
laboratory work using defmed metal mixtures and single-metal exposures, chemical
analysis of these site-samples had also suggested that metal concentrations were not
sufficient to mediate acute effects. Subacute and chronic effects measured on Militown
surface water grab samples were generally expressed in altered growth, though only a
limited number of embryos appeared to present those growth effects in a statistically
significant manner. These subacute, or teratogenic, endpoints were frequently subtle
(e.g., mild abdominal edema, hyperpigmentation) though gross malformations were
occasionally expressed. No contaminant-specific malformations were noted in these
exposures. Amphibian testing was suspended during the definitive year’s operations and
replaced in the site assessment by the emergent vegetation evaluations.
Preliminary and scoping year studies using bacterial tests. As with the amphibian
evaluations, MicrotoxR afforded an opportunity to characterize potential short-term effects
associated with water soluble constituents derived from site soils and surface water grab
samples during the preliminary year’s work. None of the surface water samples
expressed adverse biological effects on screening, and only soil sample eluates yielded
responses in screening tests [ undiluted samples at highest dilution possible] that required
definitive tests being completed. Four soil eluates presented sufficient biological effects
to warrant additional testing and calculation of EC s. While Microtox indicated that
some biological activity was associated with either surface water grab samples or soil
eluates, those limited number of responsive samples were consistent with other biological
tests completed during the preliminary field and laboratory season. The relative

agreement among the various test methods used during preliminary studies in FY 90
suggested that Microtox become a secondary evaluation method during the definitive
year’s studies, and consequently more ecologically relevant methods were applied during
FY 91 field and laboratory operations.
Plant uptake studies [ field/greenhouse methodsi. The preliminary data collection for
plant uptake studies established baseline information regarding soil fertility; the definitive
year’s work involved two interrelated greenhouse exposures using field-collected soil or
exposed sediment samples. The greenhouse studies considered [ 11 qualitative screening
tests which addressed age-related changes in metal disposition in plant tissue and [ 2J
quantitative studies which addressed metal uptake by representative garden species
following growth typical of a domestic garden.
The qualitative study regarding normal plant growth from germination through seedling
and mature plant to senescence suggested that, if germination and early seedling survival
were not impaired, and a typical plant life cycle could be completed under ideal garden
conditions. However, the quantitative early seedling growth and plant vigor test which
was completed in conjunction with the metal uptake studies suggested that soils collected
from some depositional areas may exert biological effects that would not be detected in
seed germination and root elongation tests. For example, companion work completed on
MRW samples with these routine plant toxicity tests suggested no overt adverse effects
associated with soils collected from old “ox bow” reaches along the river, but in historic
depositional areas growth reduction was indicated. It should be noted, however, these
samples were few in number, and while reductions in biomass in both species tested
(lettuce and radish) were consistent, additional samples would have to be evaluated to
determine the spatial pattern of these effects. Root elongation tests completed on sample
splits also indicated some statistically significant reductions in growth. Contrasted to
these controlled greenhouse and laboratory exposures with commercial garden varieties,
field surveys failed to indicate widespread vegetation response in the sampled area. In
part, the physicochemical characteristics of the soils across MRW may explain the
variation expressed in growth reduction, particularly in those depositional areas
occasionally associated with reduced plant vigor in laboratory tests. The geochemical
heterogeneity of soils across MRW was apparent, even in sampling units that were similar
with respect to texture, for example. While all MRW soils are xerofluvents, the soils
ranged in texture from barns to sandy barns and presented cation exchange capacities
(CECs) that could, in part, contribute to the differential bioavailabiity of metals.
Generally, soils with higher CECs (. 3OmEq/l00gram soil) were more likely to be
associated with subacute effects, and frequently expressed adverse effects in plant tests.
Variability at relatively lower CECs could be associated, in part, by subtle differences
in soil texture, as well as organic content and geochemistry.
These interacting soil matrix characteristics undoubtedly contribute to the apparent
heterogeneity in metal uptake in plants -- both native emergents, for example, and the

garden species tests under greenhouse conditions. In plant uptake studies, metals did
accumulate in plant tissues with the trend, not surprisingly, clearly suggesting that roots
would accumulate metals to a greater extent than leaves. The pattern of metal
accumulation in roots differed, in part, as a function of soil type. For example, root
crops grown in a relatively high clay soil did not have as much metal associated with its
epidermis as did plants grown in a lighter, loam soil. Nonetheless, these roots did
accumulate metal in the parenchyma tissue of the root core. To fully characterize the
biological disposition of metals in garden crops, additional time course work would be
required, and in order to adequately address human health risks, metal bioavailability in
plant tissues should be address. The human health implications of this differential plant
uptake were not considered in this work.
Plant fluorescence evaluations Efield and laboratory method]. Definitive year’s
studies with indigenous flora were variable in expression, and the preliminary year’s work
undoubtedly captures a similar variability with respect to soil matrix and metal
interactions. When considered as a qualitative measure of plant health and as a
supplement to the emergent vegetation and root elongation evaluations completed at
MRW, plant fluorescence data suggests that no overt phytotoxicity is being expressed,
but subtle indications of plant stress may be indicated at various depositional areas at
Soil and chemical analysis Isoils, sediments, and blots; laboratory methods].
Within an ecological assessment for MRW, biological effects measured in toxicity tests
must consider interactions, or potential interactions between the soil matrix and metals
in the soil, particularly with respect to bioavailability of metals in soils. Soil pH, percent
total nitrogen LTNI and percent organic material [ OM] were measured on soil samples
collected at MRW during the definitive year’s work. Soil texture analysis was also
completed, as was DTPA [ diethylenetriaminepentaacetic acid, O.025M] metals as a
physicochemical analog of bioavailable metals. With these soil matrix characteristics
available, potential confounding effects associated with soil physicochemical properties
could be identified, and metal toxicity interpretations were more adequately developed.
Soils at MRW were, not surprisingly, heterogeneous across the site, but within sampling
unit variability was relatively limited though subtle soil geochemical differences were
indicated when plant testing was considered. Soils throughout MRW were xerofluvents
and ranged from barns to silty barns to occasional silty clay barns. Organic material
was variable across MRW but was relatively invariant within defined sampling strata.
Total nitrogen, soil pH, and cation exchange capacity were similarly variable but
relatively similar within strata. Overall, these physical characteristics were invaluable to
interpretations of the biological effects associated with soil metal burdens.

The ecological assessment at MRW suggests that no overt toxicity or adverse biological
effects are being expressed at the current time. Consistently, and regardless the field or
laboratoiy test methods used, biological assessments at MRW were unable to characterize
acute toxicity. With the exception of subtle biological effects noted in samples collected
from depositional areas, future remediation plans should consider the potential biological
and ecological impacts associated with remediation efforts in light of the current impacts
associated with elevated metals in soils and sediments. While the current subtle effects
and potential future effects should not be understated, currently any widespread physical
alteration of wetland habitats may not be justified. Future site monitoring should address
potential problems associated with remediation both at MRW and at upstream operable
units, and in particular should consider the long-term effects associated with vegetation
exposures in the depositional areas. Also, indirect effects, e.g., habitat alteration
associated with reduced vegetative growth, associated with future management plans
(e.g., “walk away” or sediment dredging) should be considered as a potential
consequence of the in aifla metals that presently occur in the soils and sediments at MRW.
For those biological and, by inference, ecological effects that were considered during the
course of the work at MRW, uncertainty within the risk assessment process must be given
high regard. For example, whether “sentinel species” (Lower and Kendall 1990) are
appropriate to an ecological assessment, and whether adequate surrogate test species (in
either laboratory or field) were used in the biological assessment have been and will
continue to be, the proximate sources of uncertainty in the ecological risk assessment
process; the work at MRW is no exception and the uncertainty associated with those tools
used during the ecological assessment must be considered. At MRW specifically:
the preliminary food-web contamination survey suggested that
bioaccumulation of metals was evident in emergent vegetation in some
reaches of the deposition zones within MRW, but biomagnification or
trophic level transfer of metals to herbivores did not appear to be a
problem for the endpoints considered. Field surveys at MRW did not
contradict these assessments, but the sparse comparative toxicity data base,
particularly for chronic endpoints must be considered as part of the
uncertainty in this regard.
• Vegetation tests, particularly in evaluating water collected from the
rhizosphere, suggested that no acute effects were associated with
groundwater or surface water, but subtle growth-related effects were not
infrequent in samples collected from deposition zones at MRW. These
subtle indications were noted In both laboratory tests using emergent
vegetation and in standard root elongation tests. Again, field surveys
found no overt expression of altered vegetation patterns and reduced cover,

for example, was evident only in those areas that had been previously
physically manipulated.
Provided earthworms are good “sentinel species” to assess soil health,
earthworm evaluations in both field and laboratory were consistently
negative, suggesting that soil macroinvertebrates may not be at great risk
as long as the current soil conditions exist. Soil microbial communities,
however, were not adequately described and should be evaluated when
methods are available.
• Preliminary and scoping year studies using amphibians suggested that
subtle biological effects may potentially be expressed at MRW, but field
surveys did not support any conclusions that those effects would be
prominent nor potentially adverse.
• Preliminary and scoping year studies using bacterial tests were consistent
with the balance of biological test methods used at MRW, but may not be
representative of the soil community that occurs at MRW. Biological and
ecological assessments for soils are currently characterized with relatively
high uncertainty.
• Plant uptake studies suggest that garden crops, like those native plants
collected as part of the food-web contaminant evaluation, accumulate
metals differently, depending upon the plant species and anatomical feature
considered (e.g., root versus leaf). While empirical data illustrate the
accumulation of metal in plants growing in MRW soil, the relationship
between metal loads in plant tissue and consumer (e.g., small mammals
or humans) are quite heterogeneous and potentially a source of much
• Plant fluorescence evaluations were primarily supportive of more definitive
plant tests and at present should be considered as less sensitive than
“whole-plant” test endpoints that integrate biological responses over longer
periods of time.
• Characterization and metal analysis of MRW soils, sediments, and biota
clearly indicated that metals have accumulated in various environmental
matrices, and that MRW is spatially quite heterogeneous with respect to
metal deposition. Within sample unit variation was relatively less than
across MRW variation. Soils within sample units were relatively
homogeneous, although sufficient variation was evident that accounted for,
in part, the variability noted in biological samples (e.g., emergent plants
and terrestrial invertebrates) collected at stations within the sample units.

Environmental Toxicology international, Inc. (ETI). 1991a. Militown Reservoir
sediments site baseline risk assessment: Ecological risk assessment work plan. TZ4-
C08012-WP-H1416. Prepared for U.S Environmental Protection Agency, Region 8.
Helena, MT under U.S. EPA Technical Enforcement Support Contract.
Environmental Toxicology International, Inc. (ETI). 199 lb. Milltown Reservoir
sediments site baseline risk assessment: Sampling and analysis plan. TZ4-C08012WP-
H 1417. Prepared for U.S Environmental Protection Agency, Region 8. Helena, MT
under U.S. EPA Technical Enforcement Support Contract.
Environmental Toxicology International, Inc. (ETI). 1991c. Preliminary field survey
and laboratory evaluations at Milltown Reservoir. TZ4-C08012-FO-H1287. Prepared
for U.S Environmental Protection Agency, Region 8. Helena, MT under U.S. EPA
Technical Enforcement Support Contract.
Greene, J.C., C. Bartels, W. Warren-Hicks, B. Parkhurst, G. Linder, S. Peterson, and
W. Miller. 1989. Protocols for Short Term Toxicity Screening of Hazardous Waste
Sites. EPA/600/3-881029. U.S. Environmental Protection Agency, Environmental
Research Laboratory, Corvallis, OR.
Horning, W.B., II, and C.I. Weber. 1985. Short-term methods for estimating the
chronic toxicity of effluents and receiving waters to freshwater organisms. EPA/600/4-
85/014. Environmental monitoring and Support Laboratory, Office of Research and
Development, U.S. Environmental Protection Agency, Cincinnati, OH.
Lower, W.R. and R.J. Kendall. 1990. Sentinel species and sentinel bioassays. in
Biomarkers of environmental contamination. McCarthy, J.F. and L.R. Shugart (eds.).
1990. Lewis Publishers, Inc. Chelsea, Michigan. pp. 309-33!.
Parkhurst, B, G. Limier, K. McBee, G. Bitton, B. Dutka, and C. Hendricks. 1989.
Toxicity tests. in W. Warren-Hicks, B. Parkhurst, and S. Baker, Jr. (eds.). Ecological
assessment of hazardous waste sites. EPA/600/3-89/0l 3. U.S. Environmental Protection
Agency, Environmental Research Laboratory, Corvallis, OR.
Peltier, W. and C.I. Weber. 1985. Methods for Measuring the Acute Toxicity of
Effluents to Aquatic Organisms. Third Edition. EPA/600/4-85/013. Environmental
Monitoring and Support Laboratory, Office of Research and Development, U.S.
Environmental Protection Agency, Cincinnati, OH.
Stevens, D. 1989. Field sampling design. In W. Warren-Hicks, B. Parkhurst, and S.
Baker, Jr. (eds.). Ecological assessment of hazardous waste sites. EPA/600/3-89/013.

U.S. Environmental Protection Agency, Environmental Research Laboratory, Corvallis,
Stevens, D., C i. Linder, and W. Warren-Hicks. 1989. Data interpretation. I a W.
Warren-Hicks, B. Parkhurst, and S. Baker, Jr. (eds.). Ecological assessment of
hazardous waste sites. EPA/60013-89/013. U.S. Environmental Protection Agency,
Environmental Research Laboratory, Corvallis, OR.
Suter, G.W. 1990. Endpoints for regional ecological risk assessments. Environ.
Manage. 14:9-23.
Suter, G.W. and J.M. Loar. 1992. Weighing the ecological risk of hazardous waste
sites. Environ. Sci. Tech. 26:432-438.
U.S. EPA. 1991. Ecological assessment of Superfund sites: An overview. Ia ECO
Update, Vol. 1, Number 2. U .S. Environmental Protection Agency, Office of Solid
Waste and Emergency Response, Office of Emergency and Remedial Response,
Hazardous Site Evaluation Division (OS-230), Washington, D.C. Publication 9345.0-
Warren-Hicks, W., B. Parkhurst, and S. Baker, Jr. (eds.). 1989. Ecological assessment
of h izardous waste sites. EPAI600I3-891013. U.S. Environmental Protection Agency,
Environmental Research Laboratory, Corvallis, OR.
Weber, C.!., W.H. Peltier, T.J. Norberg-King, W.B. Horning II, F.A. Kessler, JR.
Menkedick, T.W. Neilieisel, P.A. Lewis, D.J. Klemm, Q.H. Pickering, E.L. Robinson,
J.M. Lazorchak, L.J. Wymer, and R.W. Freyberg. 1989. Short-term methods for
estimating the chronic toxicity of effluents and surface waters to freshwater organisms.
Second Edition. Environmental Monitoring Systems Laboratory, U.S. Environmental
Protection Agency, Cincinnati, OH. 600/4-89/001.

.7.5. Iwanczyk, Y.J. Wang and W.R. Graham
Xsirius, Inc., 4640 Admiralty Way, Suite 214, Marina del
Rey, CA 90292
Abstract : A laboratory study of Hg1 2 spectrometers for use
in the in—situ determination of lead on painted surfaces has
been conducted. The energy resolution attainable with Hg1 2
detectors in the energy region corres ondinq to lead K x—
rays has been measured. Sources of ° 9 Cd and 57 Co
have been used for this purpose. Energy resolutions of 880
eV (FWHM), 1370 eV (FWHM) and 1940 eV (FWHM) were obtained
at energies of 60 key, 88 key and 122 key, respectively.
Measurements on pure lead and thin film standards, ranging
from 0.5 mg Pb/cm 2 to 2 mg Pb/cm 2 , have been conducted.
Spectra from these samples, excited by 109 Cd and 57 Co
sources, are illustrated in the paper. Well separated Kai,
Ka2, Kp 1 , Kp 2 and L 1 , La, Lfl, L lines compare very
favorable to those measured with a cryogenically cooled Ge
detector. Future plans for the development of a portable
Hg1 2 XRF lead analyzer are also discussed.
There is a growing awareness about the health hazards
related to lead—based paint poisoning that can occur in many
old houses. Recent legislation required the U.S. Department
of Housing and Urban Development (HUD) to establish
procedures to remove, as far as practicable, such poisoning
hazards. In response to this legislation, HUD promulgated a
regulation which requires abatement to eliminate paint
poisoning hazards in housing in which the concentration of
lead in paint equals or exceeds 1 mg/cm 2 [ 1]. In addition,
the legislation requires HUD to periodically review and
reduce the level below 1 mg/cm 2 , to the extent that reliable
technology makes feasible the detection of a lower level,
and medical evidence supports the imposition of a lower
Because of the large number of homes that may be involved in
testing, it is very important to develop instruments and
testing methods which will provide for the reliable, rapid,
and inexpensive measurement of lead in lead—based paints,

and that also could be transferable for the measurement of
lead in household dust and urban soil. There are basically
two general types of instrumentation and techniques that are
presently in use: 1) analytical laboratory methods, 2) in-
situ field methods.
Analytical Laboratory Methods
This group include Flame Atomic Absorption Spectroscopy
[ 2,3), Inductively Coupled Plasma [ 4], Voltammetry [ 5],
Laboratory X—Ray Fluorescence [ 6], Neutron Activation
Analysis [ 7), Mass Spectrometry [ 4], and Ion Selective
Electrodes [ 4]. All the above techniques, alone or in
combination, can provide the required sensitivity and
accuracy. The major drawback with all these laboratory
methods is in the problem of collecting representative
samples. For example, it is necessary to ensure that all
layers making up the paint film are collected in proper and
relative amounts. Additionally, all lab methods require the
destruction of part of the painted surface in the sample
collection process. The time and labor involved in sample
collection, transportation and preparation (many techniques
require dissolution of the lead) makes the laboratory
techniques very expensive and slow, which practically
eliminates them as methods of choice for the application
described above.
In-situ Field Methods
Field methods, by contrast, potentially offer lower costs
and shorter times to complete the analyses. There are two
general types of screening tests, a) chemical spot tests
[ 8,9) and b) tests carried out with portable testing
equipment, e.g., portable x—ray fluorescence spectrometry
(XRF). Although some of the spot tests were reported to
have adequate detection limits, the tests were designed for
lead ions in solution. The process of producing a solution
of lead ions, starting from a dried paint film, normally
involves the use of concentrated acids, and hence would not
be readily performed in the field. There have been certain
successes reported with another method using sodium sulfide
and scratching the paint film [ 8]. This method is capable
of detecting lead in new paint films for concentrations
greater than 0.5%, however it is limited to use on white,
or light colored, films and by its susceptibility to
interference by other metal ions, e.g., mercury. In
addition, the effects of a non—homogeneous distribution of
lead throughout the film, the composition, age, and exposure
history of the paint, and the possible overcoating of many
layers of new paints, will all increase the error and lower
the precision and accuracy of any spot test method.

The only type of portable instrument for non—destructive
detection of lead in paint films reported in the literature
is an x—ray fluorescence (XRF) spectrometer [ 10,11]. The
XRF technique can potentially satisfy all the requirements
for the measurement of the lead content in paints. This
non—destructive technique can provide very quickly (in
seconds) information about lead concentrations without any
sample preparation. In addition, using information from the
characteristic L and K x—ray lines of lead simultaneously,
it is possible to determine whether lead is present on the
surface or it is buried under layers of paint.
Laboratory XRF units employing cryogenically cooled
detectors are broadly used and provide high accuracy and
precision. Currently, however, there are no truly portable,
battery operated, hand held XRF units available on the
market that can offer similar parameters. It was shown in
the literature [ 12) that it is mandatory to use the lead K
x—rays in the determination of lead concentrations. The
lead L x—rays can serve only as supplementary additional
information due to their strong attenuation by paint
overlays. The high energy of Pb K x-rays (about 75 key)
requires detectors built from high atomic number materials
in order to assure a good detection efficiency, which in
turn additionally narrows the choice of a detector.
Presently available field instrumentation can be divided
into the following categories: a) poor detection efficiency
and poor energy resolution (this include systems using gas
filled proportional counters and room temperature silicon
detectors) [ 13,14], b) good efficiency but poor energy
resolution (this includes systems employing NaI(Tl) and
other scintillating detectors) [ 15,16,17,18] and finally c)
good efficiency and resolution, but a serious lack of
portability (this include systems utilizing cryogenically
cooled GefLi) and High-Purity Ge detectors) [ 12,18,19,20].
Without the need for cryogenics, Hg 1 2 -based XRF
spectrometers are extremely attractive, due to their high
energy resolution, small size, low power requirements and
the possibility of constructing truly portable instruments
for lead paint detection. Hg1 2 detectors, due to the very
high atomic numbers of its constituents, strongly absorb x—
rays, assuring almost 100% detection efficiency for both the
K and L characteristic x—rays of lead. Mercuric iodide
portable XRF units seem to be ideally suited for detecting
lead concentrations in urban housing, offering very
convenient, non—destructive and reliable field methods
suitable for use by non—technical personnel. Such an

instrument has a significant potential to improve the
precision and accuracy over the existing field portable
instrumentation, particularly at critical decision points
such as the 1.0 mg/cm 2 level of lead in paint. Also, it may
substantially reduce the time and cost of the analysis.
With small modifications, the instrument also can be
transferable for the measurement of lead in household dust
and urban soil, with the addition of inexpensive sample
preparation techniques and target holders.
X—ray secondary—emission spectrometry, or x—ray fluorescence
spectrometry, is a non—destructive instrumental method of
qualitative and quantitative elemental analysis. In XRF, a
collection of atoms, the sample, is excited by an external
source (e.g., x—rays, electrons, etc.). This primary
radiation is incident on the sample, where it interacts to
excite inner shell electrons, which then de—excite to
produce fluorescent x—rays, whose energies are uniquely
characteristic of the elemental identity of the atom
emitting them. Therefore, if these emitted fluorescent x-
rays are collected, one may use them as an indication of the
composition of the sample. Primary excitation photons
incident on the sample interact either by the photoelectric
effect to produce the desired characteristic x-rays, or by
scattering — mainly from the atoms in the low—atomic—number
substrate. These scattered x—rays constitute an unwanted
background that sets the detection limit for the
fluorescence measurement. Thus, excitation, detection, and
spectrum analysis are the three major steps in energy
dispersive XRF analysis. The selection of the excitation
source, sample—fluorescer—detector geometry, and proper
software for the calibration and analysis of the results are
very important. The analysis method is basically driven,
and often limited, by the types of x-ray detectors available
to sense the fluorescence radiation from the test sample.
The detection efficiency and energy resolution of the
detector used determines whether it is possible to achieve
the required detection limits, precision and accuracy. As
noted, the ratio of the signal to the unwanted scattered
background must be higher for better energy resolution.
Better energy resolution helps also to separate signals from
neighboring chemical elements, whose atomic numbers are
close to that of lead.

The function of the excitation source is to excite the
characteristic x—rays in the spectrum via the x—ray
fluorescence process. Several types of sources have been
used, including nuclear sources, Bremsstrahlung radiation,
secondary fluorescence, charged particles, and synchrotron
sources. The latter three will not be further considered
here, since they are not practical for in—situ applications.
For the best portability, isotope excitation sources are
preferable due to their relatively small size.
Monoenergetic excitation energies of isotope sources produce
the unmodified (Rayleigh) and modified (Compton) scattering
peaks just at and below the incident energy. For good
sensitivity, the fluorescence x—rays of the elements must
not overlap the scatter peaks. This effect favors the use
of monoenergetic x—ray excitation sources, rather than
broadband excitation, which would distribute the scattered
radiation over the entire range of energies. Disadvantages
of isotope sources include a lack of ability to adjust their
energies, and special requirements on their handling,
shielding and accountability. Also, many of them have
relatively short half lives, requiring periodic replacement
of the source and continuous software adjustment for
changing intensities.
Recently, significant progress has been made in the area of
miniature x—ray generators. The advantages of an x—ray tube
source include its higher intensity and controllable output,
and its higher safety factor. In order to efficiently
excite the characteristic K lines of lead, the x—ray tube
would need to operate at bias voltages well in excess of 100
kV. There are such x—ray generators operating from
batteries (e.g. Kevex Model PXS6; size 18” x 7.3” x 7.3”;
weight 3Olbs; bias voltage up to 130kV ), but they do not
satisfy the criteria of small size and weight.
As a result of the above analysis a conclusion can be drawn
that for the construction of a truly portable instrument,
isotope excitation sources represent the best choice. For
the excitation of K and L lines of lead there are two
practical sources available: 57 Co and 109 Cd. Major gamma and
x—ray lines of these sources, with respect to the absorption
edges of lead, are described in Table I.

57 Co
109 ca
Major gamma radiations
Major gamma, x—radiations
time: days
Lead Absorption Edges
*See the definition in “Table of Radioactive Isotopes”,
Edgardo Browne and Richard B. Firestone, John Wiley & Sons.
The 109 Cd source offers several advantages over the 57 Co
source. First, a 1 - 09 Cd source has a longer half—life time
than a 57 Co source (453 days vs. 270 days). Therefore,
longer periods between source replacement and system
recalibration are possible with the 109 Cd isotope source.
Second, the radiation energy of 109 Cd is lower than that of
57 Co (88 key vs. 122 key), allowing easier shielding for
personnel protection and beam confinement. The third
advantage is related to certain properties of paint and its
substrates. In general, there are mainly low—Z elements in
the paint and substrate materials. In contrast to heavy
metals, these low Z elements cause a strong Compton
scattering interaction with the incident primary gamma-ray
photon. The background from such scattered photons may
seriously limit the lower detection limit of the system.
While the 122 key photons of 57 Co are well above the K-shell
binding energy of lead and provide a good source for Pb x—
ray excitation, the scattered energy from this source is in
the range 83 key to 122 key (corresponding to the scattering
angle of 1800 to 00). In order to separate the scattered
radiation from the Pb K x-rays in the energy region (73 key-
87 key), the system excitation geometry has to be modified
to produce a scattering angle of close to 00, i.e., a
grazing angle scattering. Hence, the scattering peak can be

shifted to a higher energy region. However, this is an
inefficient excitation geometry, and the low energy tail of
the direct radiation from the source will overlap the Pb K
x—ray region. For a 109 Cd source (88 key gamma line), on the
other hand, the majority of the scattered energy is in the
range 65 key to 75 key, corresponding to scattering angles
of 1800 to 90g. In this case, the 1800 backscatter angle
yields the best system geometry, and the Compton scattering
peak from this angle can be shifted to the lowest possible
energy region.
The broad band gap of Hg1 2 (2.2 eV) results in a low
detector leakage at room temperature (typically below 1 pA),
and with a construction designed to keep detector
capacitance in the vicinity of 1 pF, the electronic noise
from such detectors is very low. In a working model of an
Hg 1 2 spectrometer built for NASA’s Comet Rendezvous Asteroid
Flyby Mission, a total energy resolution of 198 eV (FWHM)
has been obtained for the 5.9 key K line of Manganese. The
noise contribution for this system was about 152 eV (FWHN)
[ 21). Also, because of the high atomic numbers of its
constituents, Hg1 2 strongly absorbs x-rays and so exhibits
almost 100 % detection efficiency up to energies of
characteristic x-rays of lead (75 key).
A field-portable XRF system, based upon the use of mercuric
iodide x—ray detectors, has several critical advantages,
foremost of which is its ability to operate without a supply
of liquid nitrogen cryogens (as would be the case for a Ge
x—ray detector), and with much better energy resolution than
scintillator—based detectors, or room—temperature—operable
Si(Li] detectors and proportional counters. Hg1 2 XRF
spectrometers have the advantages of small size and low
power needs, with obtainable energy resolution approaching
those of cryogenically cooled Ge systems, making possible
truly portable, hand held, automated instruments.
The energy resolution attainable with Hg1 2 detectors in the
energy region corresponding to lead K x—ra ’s has been
measured. For this purpose, sources of 24l , 9 Cd and 57 Co
have been used. Energy resolutions of 880 eV (FWHM), 1370 eV
(FWHN) and 1940 eV (FWHM) were obtained at energies of 60
key, 88 key and 122 key respectively. Figure 1 presents a

gamma—ray spectrum obtained with an source. Figure 2
is the spectrum obtained with a 109 Cd source.
A series of XRF measurements have been performed recently to
verify the feasibility of Hg1 2 detectors for use in XRF
analyzer application for lead detection. Experiments were
performed with two isotope sources: 109 Cd and 57 Co. The
‘ 09 Cd annular shaped source was 9 years old with an activity
of about 0.15 mCi. The activity of the decayed Mossbauer
57 Co source was about 3.6 mCi. The setup for using the
annular - 09 Cd source is illustrated in figure 3. The setup
for the 57 Co excitation is illustrated in figure 4. A
Tennelec TC244 spectroscopy amplifier with a shaping tine of
12 &s has been used for the experiments. The spectra were
collected using multichannel analyzer consisting of a
Nucleus Personal Computer Analyzer (PCA) card installed in
an IBM PC-type computer.
Figure 5 shows a spectrum of a lead sample excited by the
source. Figure 6 presents a spectrum from the sane
sample, but excited with the 57 Co source. In both cases one
can see the clearly separated Kai , Ka2, K 1 , and lines
of lead. Figure 5 shows much more pronounced L lines than
can be seen in Fig. 6, due to the very intense 22 key
excitation energy present in the 109 Cd source.
In addition to measurements on the pure Pb sample, tests
were also made on thin film samples with low lead
concentrations. Thin lead coating standards made on Mylar
films and backed with plastic substrates, ranging from 0.5
mg Pb/cm 2 to 2 mg Pb/cni 2 , were obtained from the National
Institute of Standards and Technology. The annular shaped
-° 9 Cd source discussed earlier was used for these thin film
measurements. The backscatter peak of 109 Cd falls at
approximately 66 key, with a sharp high energy edge which is
below the K peaks of Pb. Figures 7 and 8 show spectra taken
from a standard sample with a 1.025 mg/cm 2 concentration of
lead, which is at the detection level required by HUD. Lead
K lines are shown in Figure 7, while Figure 8 presents the
Pb L x-ray lines region. Besides lead Kai, Ka2, Kp 1 , Kp 2 and
L 1 , La, L , L lines, one also observes counts from Compton
and Rayleigh scattering. Note that the lead K characteristic
line spectra measured with an Hg1 2 detector (presented in
figures 5, 6 and 7) compare very favorable to those measured
with a cryogenically cooled Ge detector and presented in
References 13,19, and 20.

The preliminary experimental results show the great promise
of Hg1 2 detectors for use in XRF analyzer applications for
lead detection. A truly portable instrument for lead paint
detection can be implemented by the use of such Hg 1 2
detectors. Such an instrument can easily meet the
requirement for a critical detection limit of 1.0 mg Pb/cm 2 ,
which is the regulatory (HUD) limit for the lead
concentration in paints. In addition to being a qualitative
screening device, the Hg1 2 XRF lead analyzer can also be
used to determine the lead concentration quantitatively by
using the information obtained from the spectra. To do so,
specialized software can be used in the data analysis which
will convert the measured Pb x-ray intensities directly to
lead concentrations in the paint. This software should meet
the following requirements: a) Minimize influence of matrix
effects of paints and different substrates by utilizing
information from the entire energy spectra b) Best use of
the information from both the L and K characteristic x-rays
to obtain the highest precision and to determine the
relative location of the lead in the depth of the film.
Further experiments related to quantitative analysis,
geometry optimization for the detector-sample-excitation
source, and the miniaturization of the whole system are
planned as future tasks. In particular, the measured
detection limits, accuracy and precision will be studied and
compared with those obtained for existing portable XRF lead
analyzers in order to show clear advantages of Hg1 2 based
The research work described in this paper was supported by
the National Aeronautics and Space Administration, Contracts
NASW-4432 and NAS5-33045; NIH Grant No. 5ROl GM37161. In
addition the authors wish to acknowledge the support and
helpful assistance of Mr. Bart Dancy in the construction of
the Hg1 2 spectrometer probes, Ms. Fanny Riquelme for the
fabrication of the HgI detectors, and Mr. Wayne Schnepple
for assistance in preparation of this paper.
1. Federal Register, 53, No. 108, June 6, 1988, 20791

2. K. A. Hausknecht, E. A. Ryan, L. P. Leonard,
“Determination of Lead in Paint Chips Using a Modified
Ashing Procedure and Atomic Absorption Spectrophotoiuetry”,
Atomic Spectroscopy 3(2), 1982, 53-55.
3. “Standard Test Method for Low Concentrations of Lead,
Cadmium, and Cobalt in Paint by Atomic Absorption
Spectroscopy”, ASTM D 3335—85a, Annual Book of ASTM
Standards, 6.01, American Society for Testing and Materials,
Philadelphia, PA, 19103, 1988, 507—509.
4. D. A. Skoog, Principles of Instrumental Analysis, Third
Edition, CBS College Publishing, 383 Madison Ave., New York,
NY, 1985.
5. P.C. Lai, K. W. Fung, “Determination of Lead in Paint
by Differential-Pulse Anodic-Stripping Voltainmetry”,
Analyst, 103, 1978, 1244—1248.
6. G. S. Kuntz, R. L. R. Towns, “Determination of Lead in
Paint by Energy Dispersive X—Ray Fluorescence Spectrometry”,
Journal of Coatings Technology, 54(687), 1982, 63—69.
7. G. 3. Lutz, “Determination of Lead in Paint with Fast
Neutrons from a Californiuin-252 Source”, Analytical
Chemistry, 46, 1974, 618—620.
8. H. P. Vind, C. W. Mathews, “Field Test for Detecting
Lead-Based Paints”, Technical Note N-l455, Civil Engineering
Laboratory, Port Hueneme, CA, 1976.
9. F. Feigl, V. Anger, R. E. Oesper, Spot Tests in
Inorganic Analysis, Elsevier Publishing Company, London,
1972, 285—286.
10. W. E. Byrd, M. E. McKnight, “Potential Methods for
Measuring and Detecting Lead in Existing Paint Films: A
Literature Review”, NISTIR 89-4205, National Institute of
Standards and Technology, Gaithersburg, MD 20899, Jan 1990.
11. M. E. McKnight, W. E. Byrd, “Screening Procedures for
Detecting Lead in Existing Paint Films: A Literature
Review”, NISTIR 89-4044, National Institute of Standards and
Technology, Gaithersburg, MD 20899, Jan 1990.
12. J. L. Campbell, L. A. Crosse, “Portable Instrumentation
for the Determination of Lead in Painted Surfaces”, Bulletin
of Environmental Contamination and Toxicology, 16(4), 1976,
13. G. R. Laurer, T. J. Kneip, R. E. Albert, F. S. Kent,
“Insitu Determination of Lead on Painted Surfaces for the
Prevention of Childhood Lead Poisoning”, Applications of Low
Energy X and Gamma , C. A. , ed., Gordon and
Breach, New York, NY, 1971, 289—301.

14. M. E. McKnight, W. E. Byrd, W. E. Roberts, E. S.
Lagergren, “Methods for Measuring Lead Concentrations in
Paint Films”, NISTIR 89-4209, National Institute of
Standards and Technology, Gaithersburg, MD 20899, Dec. 1989.
15. J. C. Spurgeon, “Response Characteristics of a Portable
X—Ray Fluorescence Lead Detector: Detection of Lead in
Paint”, NBSIR 73—231, June 1973, National Institute of
Standards and Technology (formerly The National Bureau of
Standards), Gaithersburg, MD 20899.
16. A. P. Cramp, H. W. Berger, “Evaluation of New Portable
X—Ray Fluorescent Lead Analyzers for Measuring Lead in
Paint”, NBSIR 78—1466, May 1978, National Institute of
Standards and Technology (formerly The National Bureau of
Standards), Gaithersburg, MD 20899.
17. D. Barltrop, C. L. Harford, N. J. P. Killala, “The
Determination of Lead in Paint Films with a Portable Isotope
Fluorescence Analyzer”, Bulletin of Environmental
Contamination and Toxicology, 6(6), 1971, 502-508.
18. S. D. Rasberry, “Investigation of Portable X-Ray
Fluorescence Analyzers for Determining Lead on Paints
Surfaces”, Applied Spectroscopy, 27(2), 1973, 102-108.
19. G. R. Laurer, T. J. Kneip, M. Eisenbud, R. E. Albert,
N. Nelson, “New York City Health Research Council X-Ray
Fluorescence Analyzer for Lead”, Isotopes and Radiation
Technology, 9(3), 1972, 275—277.
20. G. R. Laurer, T. J. Kneip, R. E. Albert, F. S. Kent,
“X—Ray Fluorescence: Detection of Lead in Wall Paint”,
Scince, 172(3982), 1971, 466—468.
21. J. S. Iwanczyk, Y. J. Wang, J. G. Bradley, J. N.
Conley, A. L. Albee, and T. E. Economou, “Performance and
Durability of Hg1 2 X-Ray Detectors for Space Missions”, IEEE
Trans. Nuclear Science 36, No. 1, 841—845 (1989)

2500 I I
59.6 keV
z N —L lines
1000 FWHM
880 eV
V -
O 20 40 60 80 100 120
ENERGY (key)
Figure 1. The spectrum from an source.
7 88keV
‘° 9 Cd SOURCE
2000 FWHM
0 50 100 150 200
Figure 2. The spectrum from a 109 Cd source.

iii H
Figure 3. Geometry of detection system with the 109 Cd
Figure 4. Geometry of detection system with the 57 Co

1 50
1 2000
Figure 5. The spectrum of a lead sample excited by a 109 Cd
C ’)
0 20 40 60 80 100 120
Figure 6. The spectrum of a lead sample excited by a 57 Co
0 20 40 60 80 100 120

40 120
Figure 7. The Pb—K line spectrum of 8 1.025 mg Pb/cm 2 thin
film sample excited by a 9 Cd source.
Figure 8. The Pb—L line spectrum of 1.025 mg Pb/cm 2 thin
film sample excited by a 9 Cd source.
C l ,
C -,
60 80 100
0 5 10 15 20 25 30

Stephen B. Friedman, Ph.D .
Vice President, Research and Development
EnSys Inc.
P.O. Box 14063
Research Triangle Park, North Carolina, 27709
Effective field screening methods can increase the efficiency of site
management and improve overall data quality when used to supplement
the services of regional laboratories. The development of these
methods, however, begins with the selection of a technology that will
be compatible with numerous compounds and matrixes and yet be
simple, effective and rugged enough to be incorporated into a protocol
for use in the field.
We have developed several immunoassay-based field screening
methods for the detection of pentachiorophenol, PCB’s, and petroleum
contamination, in solid and liquid waste samples. Performance and
stability evaluations of these methods have been evaluated within our
own laboratories and in numerous external field studies. These studies
have demonstrated the effectiveness of these methods when used in
field screening applications. The immunochemistry underlying these
methods, and the preparation of the reagents, will be reviewed. A
description of the protocol and the characteristics that minimize the
incidence of false negative results will be presented. Their general
performance characteristics and application to site characterization
will be discussed.
Testing is an essential, and integral, component of all environmental
protection and restoration activities. It is the rate limiting element
that influences the time, cost and overall efficiency of project

The management of toxic waste sites usuafly involves a progression
through the phases of identification, characterization, remediation and
monitoring, with testing being performed during each phase.
Reference laboratory methods can effectively identify and quantify
unknown compounds in a sample, but become relatively inefficient when
used to rapidly locate contamination (i.e. mapping), and assist in
remediation and monitoring activities. The complexity of laboratory
protocols, and the laboratories proximity to the test site, delays the
availability of information and increases the cost of data. The ultimate
cost is in the time required by the field crews. Effective field
screening methods can increase the efficiency of the clean-up process
by providing an on-site, high-throughput, and cost-effective way to
locate contamination and manage its remediation.
The EPA has long promoted and supported the concept of screening
methods to supplement laboratory analysis and increase overall
efficiency. The need for more effective methods has been recognized in
the Superfund Amendments and Reauthorization Act of 1986 which
specifies the development and evaluation of alternative time and cost-
saving methods that will assist in the eventual remediation of the
nations Superfund sites.
We have developed several field screening methods that are being
used to detect Pentachiorophenol, PCB’s, Petroleum Products, and PAH’s
in both soil and water matrixes, and on solid surfaces. Our objective
has been to develop reliable and cost effective methods for obtaining
the data needed for site investigation, remediation and monitoring
activities, waste screening, process control, and monitoring activities
to maintain regulatory compliance. Our approach was to develop
methods that were consistent with a list of essential screening
Essential Field Screening Characteristics
Screening methods need to provide fast, simple, cost-effective and
reliable information when operated under field conditions. The
reagents and equipment should be portable and stable at ambient
conditions, and the claims relating to performance should accurately
reflect anticipated field use. The methods should be able to rapidly
provide an ample quantity of data, and the protocol should be simple to

perform and safe to use. Performance characteristics relative to
sensitivity, freedom from matrix interferences and cross-reacting
compounds, and correlation to an acceptable reference method should be
carefully evaluated. Developers must maintain high, and consistent,
quality standards relative to the consistency of their manufacturing
protocols, the adequacy of in-process and pre-release quality control
methods, and the reliability of their product claims. A characteristic
of particular significance for screening methods is that they exhibit a
very low frequency of false negative results.
Screening methods detect contamination at specified
concentrations. The concentration may relate to a hazardous threshold,
a clean-up target, or a process-control parameter. The potential
implications of false negative data far outweigh those of false positive
results. The consequence of a false positive, while a costly problem
that needs to be minimized, results in additional testing or treatment.
False negative data, however, provides the erroneous perception of a
clean site, and may have serious environmental and legal consequences.
Safeguards that minimize the incidence of false negative results are
imperative. Appropriate control over the frequency of false positive
data needs to be established and maintained.
Immunoassay Applications
The field of immunochemistry, and the development of immunoassay
technology, has been evolving since the late 19th century. However, the
majority of these methods have been developed for use by the medical
community. These methods have achieved a reputation for reliability
and cost-effectiveness. Literally hundreds of immunoassay’s have been
developed for Drugs of Abuse testing, Therapeutic Drug Monitoring (e.g.
digitalis derivatives, anti-asthma formulations, anti-epileptic
reagents, antibiotics), pregnancy testing, hormone testing (e.g.
thyroxine, thyroid stimulating hormone), tests for pathological markers
(e.g. lactic dehydrogenase isozymes, creatine kinase isozymes), tests
for acute phase proteins (e.g. carcinoembryonic antigen, alpha
fetoprotein) and tests for of tumor marker proteins.
Environmental applications have been explored for the better part of
a decade and a number of immunoassay methods have been developed 123 .
Most have been used for the detection of herbicides and pesticides in

aqueous matrixes. The application of immunoassay technology to the
testing of solid waste, complex matrixes, and highly lipophilic
compounds, has provided unique challenges for the chemistry. 4 The
feasibility of developing such methods, however, has been
demonstrated with immunoassay’s for Dioxin 56 and in the screening
methods developed by EnSys.
Historical Prospective
The history of immunoassay technology can be traced to 1900 when
Karl Landsteiner described the A, B and Zero (0) blood types after
observing the agglutination reaction (i.e. aggregation) that resulted
when he mixed the erythrocytes and serum from several of his co-
workers on a slide 7 . His observation became the basis for present day
blood typing methods. Landsteiner remained a dominant figure in
immunology for the next 40 years performing numerous experiments
that demonstrated the extraordinary specificity of the antibody binding
reaction. He introduced the term hapten to define compounds that are
unable to directly stimulate antibody production when injected into an
animal, but are capable of binding to an antibody if they are produced by
an alternate means. Most environmental chemicals are haptens, and
although potentially toxic, will not stimulate the immune system to
respond 8 .
For 50 years following Landsteiner’s discovery, immunoassay
technology continued to rely upon the binding and cross-linking ability
of an antibody to cause agglutination, cell lysis, and protein
flocculation reactions. These methods were relatively insensitive
when compared to the immunoassay methods of today, and better suited
to the analysis of larger compounds and organisms (e.g. bacteria,
proteins). A major advance occurred in the 1950’s when Drs. Berson
and Yalow, while investigating the metabolism of radiolabelled insulin
administered to diabetic patients, observed the production of anti-
insulin antibodies in the serum of insulin-treated diabetics. They
described a radioimmunoassay (i.e. RIA) method in 1959 that used anti-
insulin antibody molecules, and radiolabelled insulin, in a highly
sensitive procedure to quantify insulin in the serum of patients. The
method used a competitive antibody binding reaction, where
radiolabelled insulin and sample insulin compete for a limited number
of antibody binding sites.’° In 1977, Rosalyn Yalow was awarded the

Nobel Prize in Medicine for her work on the development of the
radioimmunoassay method for peptide hormones RIA rapidly became
a universally accepted method that demonstrated exceptional
specificity, sensitivity, and simplicity.
A simpler, safer, and more convenient immunoassay was reported in
1971, when two independent research teams, Engvall and Perlmann, 12
and Van Weeman and Shuurs, 13 simultaneously disclosed a competitive
immunoassay method that used an enzyme-labelled conjugate instead of
a radiolabelled-conjugate, to produce a test that generated a visible
end-point signal. The new ELISA (i.e. enzyme linked immunosorbent
assay) method eliminated the problems associated with the safety,
disposal and detection of radioactive reagents. The method offered long
term stability, the opportunity to generate quantifiable data using
instruments commonly available in most laboratories, and a mechanism
to develop separation-free (i.e. homogeneous) procedures and simple
qualitative screening tests.
Current immunoassay technology benefits from the diversity of
detection systems that have been developed that use enzyme-catalyzed
chromogenic reactions, radionuclides, chemiluminescence,
fluorescence, fluorescence polarization and a variety of potentiometric
and optical biosensor techniques. Improvements in the sensitivity
achieved by these methods has necessitated the generation of new
descriptive nomenclature for methods that can detect ligands at
“zeptomolar” (102h1 600 molecules) concentrations. 14
Enzyme Immunoassay Chemistry and Protocol
Immunoassay methods combines the specific binding characteristics
of an antibody molecule with a read-out system that is used to detect
and quantify compounds. Antibodies are binding proteins that are
produced by the immune system of vertebrates in response to
substances that are perceived as foreign. The immunoassay methods
we have developed use antibody reagents, and a chromogenic detection
system, that specifically bind and detect hazardous chemicals in both
solid and liquid waste samples.
The EnSys chemistry uses two basic reagents, namely, an antibody,
for example anti-pentachlorophenol, and an enzyme conjugate reagent

composed of, for example, PCP molecules covalently bound to the
enzyme horseradish peroxidase.
The Antibody Molecule
The physiological role of antibody, or immunoglobulin, molecules is
to bind, and thereby label for destruction, a foreign substance within an
organism. Antibody molecules are synthesized by a subset of
lymphocytes, termed B lymphocytes, that become activated to produce
antibody after exposure to substances having prerequisite size,
complexity and “foreignness” to the host organism. They are large,
polymeric proteins (i.e. 1.5 x i0 d), that can be c’assified into sub-
populations on the basis of their sequence, size and number of sub-
units. Five major populations, or isotypes, exist that carry the
designations of 1gM, IgA, lgD, lgG and IgE, with immunoglobulin G (lgG)
usually found in the highest concentration. 15
Binding affinity and specificity is influenced by the chemistry and
conformation of a binding cleft at the N terminal end of the molecule,
that exists between juxtaposed, and convoluted, portions of “heavy”
and “light” potypeptide chains. The amino acid sequence, and therefore
the conformation at the N terminus, is highly variable and influences
the binding specificity of the molecule. 16 Studies have demonstrated
that binding is a function of the conformational complimentarity that
exists between the target ligand and the antibody binding site 17 , and
that the “goodness” of fit relates to the interaction between the
electron cloud within the binding site and the bound ligand. Antibody
binding is not covalent, and the affinity or strength of binding, is a
function of cooperative hydrophobic, Van der Waals, electrostatic and
hydrogen bonding interactions 18 In general, equilibrium constants for
the most avid antibody binding reactions do not exceed 1012 L/M. 19
To induce the formation of antibodies that can be used to detect, for
example, pentachIoropt enol, molecules of PCP must first be derivatized
and coupled to large carner molecules such as albumin, hemocyanin or
thyroglobulin. The increased size and complexity of this “immunogen”,
once injected, is sufficient to stimulate the immune system to produce
an antibody response. The effectiveness of the immunogen is
influenced by the surface density of the chemical on the carrier (i.e.

epitope density), the nature of the bridge chemistry used, the
immunization protocol, immunogen concentration, incorporated
adjuvants (i.e. immune response stimulants), and the species of the
host animal. Significant progress has been made in deciphering the
mechanisms of the humoral immune response, but a great deal is still
not understood. Experience and good fortune continues to play a
significant role in the production of effective antibody reagents for
test kit development.
The Enzyme Conjugate Reagent
The enzyme-conjugate reagent for the EnSys PCP detection method
is synthesized by derivatizing PCP and coupling it to the enzyme
horseradish peroxidase. Numerous functional groups on enzyme
molecules (e.g. amino, sulfhydryl, carboxyl, carboxamide, tyrosyl,
sugars) offer convenient points for the attachment of ligand molecules.
Enzymes enhance the sensitivity of the method by the catalytic
amplification of the detection signal. A single molecule of the
enzymes commonly used in immunoassay methods will convert
approximately 1 o6 molecules of a substrate into a product within one
minute at ambient temperatures. 2 ° Catalysis is a function of the
conformation at the enzymes catalytic site, and it is this conformation,
and the alignment of certain amino acid residues at spatially
significant positions, that influences its rate and selectivity. The
catalytic site is maintained by both non-covalent (i.e. hydrophobic,
hydrogen bonding, ionic and Vander waals interactions) and covalent
(i.e. disulfide) forces, and can be influenced by temperature, the binding
of ions, chaotropic agents, detergents, lipids, etc.. It is therefore
important to normalize and correct for anticipated variations in the
reaction environment.
An effective enzyme conjugate reagent must retain the ability to
both bind antibody, and hydrolyze a substrate reagent into a detectable
signal. Some of the more commonly used enzymes include horseradish
peroxidase (i.e. HRP), alkaline phosphatase, and b-galactosidase. Each
is compatible with a wide variety of different substrates offering
unique choices in kinetics and hydrolysis products. The EnSys methods
use a hydrogen peroxide substrate solution with a tetramethylbenzidine
chromogen that produces a blue chromophore in the presence of
horseradish peroxidase enzyme.

Analysis of Solid Waste
Immunoassay methods have predominantly been used to test liquid
matrixes such as blood, urine, and water. The testing of solid waste
requires that the issues of sample collection, dispersion, extraction
and clarification be addressed and integrated with the immunoassay
component. A reproducible, particulate-free, leachate must first be
produced. The extraction and recovery of a compound from soil requires
the selection of an appropriate solvent system, adequate samp’e
dispersion, sufficient time for partitioning, non-invasive clarification
and compatibility with the subsequent immunochemistry. For polar
compounds, buffers, detergents (e.g. Tweens, Tritons, etc.) or solvents,
used together, or in combination, have proven to be effective for
extraction. Analytical methods for the analysis of solid waste rely
upon gravimetrically collected samples, and results are reported in
gravimetric units. Volumetric sampling for solid waste should be
avoided because of the potential bias that may be caused by the
specific gravity of the sample.
Solid waste analysis using the EnSys system involves the
gravimetric collection of a 10 g sample using a small battery-operated
balance. The sample is transferred into a dispersion vial containing a
solvent and dispersing pellets, and is subjected to a one minute manual
agitation for adequate dispersion and partitioning of the ligand.
Filtration of the sample suspension to produce a particulate-free
leachate is accomplished using a fingertip-operated filter unit fitted
with run-adsorbing fitters. The clarified leachate is next analyzed
using the immunoassay component.
Immunoassay Component
The EnSys immunoassay chemistry is explained using the following
pentachlorophenol model. Anti-PCP antibody is immobilized to the
bottom of polystyrene tubes at a pre-defined concentration (see figure).
The concentration and affinity of the antibody for the sample molecules
and enzyme conjugate molecules directly influences the overall
sensitivity of the final method. High and equivalent affinity, and
minimal non-specific signal generation, will usually produce assays
having superior sensitivity.

For this illustration we will simultaneously test a negative PCP
sample, a sample containing >5 ppb PCP, and a standard solution
containing the equivalent of 5 ppb PCP. We begin by adding the samples,
and standard to separate, and identical, anti-PCP antibody-coated test
tubes. To each tubes we also add an equal volume of HRP-PCP enzyme
conjugate solution. The three tubes are then allowed to incubate at
ambient conditions for ten minutes.
During the incubation period sample PCP molecules and PCP-HRP
conjugate molecules compete for the limited number of antibody
binding sites that are available on the bottom of each of the tubes. The
antibody concentration present is insufficient to permit the binding of
all of the sample PCP and PCP-enzyme conjugate molecules
simultaneously, and a situation somewhat analogous to the game of
musical chairs exists, with the limited antibody binding capacity
serving as the chairs in this example. The concentration of enzyme
conjugate immobilized in each tube is inversely proportional to the
concentration of PCP in the sample or standard. The PCP in the
“standard” tube, limits the binding of enzyme conjugate, the
“negative” sample permits more conjugate to bind (i.e. relative to the
standard), and the “positive” sample limits the binding of the conjugate
(i.e. relative to the standard). At the end of the 10 minute incubation
period, the tubes are washed leaving only the enzyme conjugate that
was retained by the antibody on the bottom of each tube.
The enzyme-conjugate remaining is next used to produce a
detectable signal. Upon addition of the substrate/chromogen reagents
(i.e. H Q and tetramethylbenzidine), the enzyme molecules catalyze
the formation of a blue product. The color that is generated is directly
proportional to the concentration of enzyme in each tube.
Therefore, the negative sample tube containing < 5 ppb of PCP (i.e.
tube 2) rapidly produces a solution that is visibly darker (i.e. greater
absorbance) than the standard tube. The positive test sample in tube
#3, that contains > 5 ppb of PCP, produces a solution having less color
(i.e. lower absorbance) than the standard tube. By comparing the
absorbance of the sample tubes to the absorbance of the standard tube
that contained 5 ppb of PCP using the battery-operated comparative
photometer offered with the system, sample contamination can

empirically be determined. In competitive ELISA methods, the final
absorbance produced is inversely proportional, and logarithmically
related, to the ligand concentration present in the initial test sample.
Screening Method Characteristics
We have developed six screening methods that share several common
performance characteristics (see figure). Each of the methods can
process multiple samples in less than 30 minutes. A single individual
has been able to analyze fifty samples within one day of testing. The
methods are self-contained, field-compatible, do not require
refrigeration or use hazardous components. The detection level for
each can be set at the users discretion, with the maximum obtainable
sensitivity consistent with significant regulatory levels. The PCB-RISc
methods will detect PCB contamination in soil at a concentration of 5
ppm, and at 10 ug/100 cm 2 when using the “wipe” method for solid
surfaces. The Penta-RlSc method screens for PCP in water to a
concentrations of 5 ppb, and to 500 ppb when using the soil analysis
method. The Petro-RlSc method will detect petroleum product
contamination in soil at, or above, a concentration of 100 ppm, and the
PAH-RlSc method detects contamination in soil at, or above, 10 ppm.
Each method is configured to permit multilevel analysis of samples in
order to facilitate the construction of concentration profile maps.
Screening Applications
Numerous applications for these screening methods have been
demonstrated during field trial and site investigational activities. In
general, these methods have been used to facilitate site
characterization and remediation activities, and assist in a variety of
monitoring programs.
The EnSys immunoassay screening methods have been used to locate
contamination. Their relatively low cost has amplified the quantity of
data generated and has provided the information needed to produce
high-resolution site maps quickly. The information, provided in real
time, has increased the efficiency of field operating crews, and has
provided a mechanism for the selective screening of samples destined
for subsequent laboratory analysis. The costly analysis of
uncontaminated samples, and the time required to complete site

investigational activities, has been significantly decreased.
The methods have accelerated the progress of remediation activities.
Field crews have used them to follow contamination plumes and
indicate when excavation can be discontinued. They have been used to
assist in the design of effective remediation protocols.
In support of monitoring programs, these methods have helped to
maintain compliance with appropriate discharge levels, and screen
waste prior to acceptance for treatment, storage, transport or disposal.
The advantages of immunoassay technology can be attributed to the
underlying lock and key binding principle and its compatibility with
aqueous matrixes. The method does not involve, nor require, the
chromatographic separation of sample components, nor does it require
that compounds absorb light of a specific wavelength for detection.
Interferences from other compounds are considerably less of a problem
because of the conformational nature of the antibody binding process.
Sample processing time is significantly reduced, and the direct testing
of aqueous samples, or water-soluble leachates of soil, can be
performed. The technology offers a unique, and conservative, approach
to field screening. The incidence of false negative data is
exceptionally low. Aspects that tend to interfere with immunoassay
methods tends to cause an overestimation of contamination, or false
positive result.
The advantages of these methods relates to their specificity and
compatibility with aqueous matrixes. Their disadvantages relate to
these same characteristics. As a screening method for specific
compounds these methods excel, but become less efficient when
multiresidue analysis of samples is required. The development of tests
for highly lipophilic ligands, and matrixes, offer unique challenges
because of the phase disparity issues that exist.
As a technical platform immunoassay technology offers significant
versatility and performance advantages. These methods offer a
convenient and effective new tool that can enhance the efficiency of
site management activities and the utilization of our national
laboratory system.

Hammock, B.D.; Gee, S.J.; Showing, P.Y.K.; Miyamoto, T.; Goodrow, M.H.;
Van Emon, J.; Seiber, J.N. Utility of Immunoassay in Pesticide Trace
Analysis. In “Pesticide Science and Biotechnology; Greenhalgh, R., Roberts,
T.R., Eds.; Blackwell Scientific: Ottawa, 1987; pp 309-316
2 Hammock, B.D.; Mumma, R.O. Potential of Immunochemical Technology
for Pesticide Analysis. In Pesticide Analytical Methodology; Harvey, J.,
Zweig, G., Eds.; ACS Symposium Series 136; American Chemical Society:
Washington, DC, 1980; pp 321-352
Mumma, R.O.; Brady, J.F.; Immunological Assays for Agrochemicals; In
Pesticide Analytical Methodology; Harvey, J., Zweig, G., Eds.; ACS
Symposium Series 136; American Chemical Society: Washington, DC, 1980;
pp 341-348
Albro, P.W.; Luster, M.I., Chae, K., Chaudhary, S.K., Clark, G., Lawson,
L.D., Corbett, J.T., McKinney, J.D .; Toxicology and Applied Pharmacology;
50:137-146 (1979)
Vanderlaan, M.; Stanker, L.H.; Watkins, B.E.; Improvement and
Application of an lmmunoassay for Screening Environmental Samples for
Dioxin Contamination Environmental Toxicology and Chemistry, 7:859-870,
Stanker, L.H.; Watkins, B., Rogers, N.; Vanderlaan, M.; Monoclonal
Antibodies for Dioxin: Antibody Characterization and Assay Development;
Toxicology, 45:229-243, 1987
Landsteiner, K. “Uber Aggluntinationserscheinungen Normalen
Menschlichen Blutes.” Wein. Kiln. Wschr. 14:1132,1901
Basic and Clinical immunology, Stites, D.P.; Terr, A.l. eds.; Appleton
and Lange, Connecticut, 7th Edition, 1991
Ritzman, SE: Behring Diagnostic Manual on Proteinology and
Immunoassays, 2nd Edition, Behring Diagnostics, 1977
Principles of Competitive Protein Binding Assays, Second Edition;
ODell, W.D., Franchimont, P (editors); Chapter 1, Introduction and General
Principles; ODell, W.D.; John Wiley and Sons, New York (publishers);
Basic and Clinical Immunology, Stites, D.P.; Terr, A.l. eds.; Appleton
and Lange, Connecticut, 7th Edition, 1991
Engvall, E. and Pertmann, P : Enzyme-Linked immunosorbent assay
(ELISA). Quantitative assay of immunoglobulin G; Immunochem., 8:871-
874, 1971
Van Weemen, B.K. and Schuurs, A.H.W.N.: lmmunoassay using antigen-
enzyme conjugates; FEBS Letters, 15: 232-236, 1971

14 Proceeding of the Twenty-Third Annual Oak Ridge Conference on
Advanced Analytical Concepts for the Clinical Laboratory; April 11, 12,
1991; Clinical Chemistry; 37: 9, 1991
15 Immunology; Roitt l.M., Brostoff J., Male D.K.; C.V. Mosby Co. and
Gower Medical Publishing; St. Louis, Mo.; 1985
Capra JD, Kehoe JM: Hypervariable regions, idiotypy, and the antibody-
combining site. Adv. lmmunolo., 1975:20:1
Wu TI, Kabat EA: An analysis of the variable regions of Bence Jones
proteins and myeloma light chains and their implications for antibody
complimentarity. J. Exp. Med. 1970; 132:211
Basic and Clinical Immunology; Stites D.P., Terr A.l.; Seventh edition;
Appleton and Lange(pub.), E. Norwalk, Conn.; 1991
“Immunology; Roth, l.M., Brostoff, J.; Male, D.K. eds.; C.V. Mosby
Company, St. Louis, Missouri, 1990, pp 7.1
Immunoassays for the 80’s; Ekins, A., Chapter 2; Voller A., Bartlett A.,
Bidwell D. eds; University Park Press, Baltimore, Md., 1981
Ngo T.T.; Enzyme Mediated Immunoassay: An Overview;

Incubation and competitive binding reaction
EnSys Enzyme
Immunoassay Chemistry
1. Components in ELISA chemistry
2. Enzyme addition
5 & 6. Color deveiopmeni

( T i
EnSys Field
Screening Test Kit

James Smith , Chemist and Eugene Brozowski, Analyst, Trillium,
Inc. Coatesville, Pennsylvania 19320 (215) 383—7233 FAX (215)
383—7907, and John E. Rhodes, Environmental Engineer, Rhodes
Engineering, 505 South Leola Road, Moorestown, New Jersey
08057, (609) 273—9517 FAX (609) 273—9518.
The discovery of PCBs at a site inevitably leads to the
expenditure of a major amount of money on analytical
measurements. The most cost effective approach to the site
investigation and the PCB remediation is to accomplish the
analyses at the site with a minimum turnaround time. This
approach would limit the number of samples taken during the
site investigation and focus sampling to the “hot spots” in an
efficient manner. The same approach can facilitate the
removal of contaminated soil as well as limit the amounts of
soil reniediated without costly delays waiting for analytical
results. This approach has been used by many contractors in
field laboratories.
Most field PCB analyses have been based on the gas
chromatographic methods developed by Dr. Tom Spittler of U.S.
EPA Region I laboratory. 3 We feel that the introduction of
the immunoassay for PCBS has increased the efficiency of the
field laboratory by decreasing the turnaround time of the
analyses with consequent decreases in cost. A single field
laboratory analyst can complete approximately 40 to 50
inimunoassays in an 12-hour shift. A typical single analyst
12—hour shift using the Tom Spittler methods can extract,
clean-up, and analyze approximately 10 to 20 % of the total
number of samples that can be tested by the iminunoassay
The immunoassay for PCBs is a colorimetric test and is
utilized as a positive or as a negative measurement. It is
highly specific for PCBs in that the test is usually free from
false positives. The test can be conducted by anyone with
minimal training with few false positives or false negatives
when compared to acceptable and consensus field and laboratory
In the case study to be described in this paper, we used the
PCB immunoassay by EnSys for a “go/no go” at a 5 ppm (mg/Kg)
concentration (as received basis). Utilizing a grid, the
sampling team obtained surface soil samples at an efficient
pace by defining the PCB contamination versus area using the

iminunoassay results. Each sample delivery group consisted of
a soil blank and a soil blank spike at 5 ppm (ug/Kg dry weight
basis). The rapid immunoassay turnaround time of 20 minutes
allowed better solution of the next group of samples. The
field results were verified by analyzing 10-20% of the samples
by subsequent field gas chromatographic analyses and/or
laboratory analyses by approved methods.
Our results indicate the cost effectiveness of the immunoassay
to both PCB site investigations and reinediations. The cost of
an immunoassay was $28.50 per test. We foresee numerous cost—
effective uses for PCB immunoassays including the use of the
50 ppm assay for better definition of “hot spots.”
The immunoassay technique is based on the specific binding of
the analyte with an antibody. The first step in producing a
test for an analyte involves the cultivating of the antibody
that will bind with a high degree of affinity and specificity
to the analyte when the analyte is present in very low
concentrations. A reporter conjugate must be present. This
is made up of a ligand attached to an enzyme molecule.
Antibodies are specific binding proteins that are produced in
response to a foreign substance, and that bind and label that
substance - antibody for disposal by the immune system. It is
believed that the antibody’s binding is mainly a function of
conforinational complimentarily between the target analyte and
the antibody’s binding site. Antibodies are the key
ingredient for an immunoassay. They are produced by
vertabrate organisms. The immune systems of animals will only
respond to bacteria, virus and pollen. These “invaders” are
macromolecules or micro organisms compared to target analytes
of environmental concern. The target environmental analytes
are unable to directly stimulate the production of antibodies
and large immunogen cojugates must be prepared in order to
trigger an antibody response.
In order to produce antibodies that bind to the environmental
target analyte, a derivative of the target analyte is made and
chemically bonded to a macromolecule, usually a protein,
called a carrier molecule. This synthesized immunogen is now
effective in stimulating an animal’s immune system and
antibodies are generated and collected for use in the
The PCB immunoassay incorporates chromogenic reactions as a
detection system. The chromogenic ELISA (enzyme-linked
immunosorbent assay) chemistry uses two essential reagents, an

antibody and an enzyme conjugate reagent. The enzyme
conjugate is a PCB molecule linked to horseradish peroxidase.
This conjugate must be able to both bind to the antibody via
its attached target analyte and convert a substrate into a
detectable color. The horseradish peroxidase molecule
catalyzes the conversion of a colorless chromogen
tetrainethylbenzidine to a blue derivative in the presence of
the substrate, hydrogen peroxide.
The Test
Tubes are coated with the antibody. The sample extract, or
known, standard is added with a predetermined amount of the
conjugate containing the horseradish peroxidase (HRP). The
sample PCBs and the conjugate HRP compete for the available
sites of the antibodies on the tube coating. After this
competition or incubation is complete, the sample and
conjugate — HRP are washed out of the tubes. The color
reaction mixture is added and developed. The more PCB
molecules present in the tube the fewer enzyme conjugate
molecules that will bind and thus less blue color will
develop. The more PCB molecules bound to the antiboclys, the
less blue color is developed. By utilizing a standard PCB
concentration one can colorimatrically determine if the sample
contained a concentration greater than or less than that of
the standard.
This inununoassay system was used to carry out a site
investigation for PCBs using a 5 ppm “go - no go” test kit
developed by EnSys Inc. of Research Triangle Park, North
This work was not a test but a real project. The immunoassay
was added to the site investigation in order to reduce the
cost of the project. The site consisted of approximately an
acre with three masonry buildings. In the most recent past,
this site was a heating oil storage and transfer facility for
a small local home distributor. In an economical move, the
heating oil firm ceased the use of this facility and removed
the large above ground storage tanks. To complete the site
clean—up in preparation for sale or lease, several small
gasoline underground storage tanks were removed. Testing that
accompanied the tank removal did indicate some areas with high
total petroleum hydrocarbons. In the negotiations with the
state UST agency, there was a review of the data by an expert
hired by the law firm representing the heating oil company.
The tentatively identified compounds (TICs) that were
previously ignored by everyone indicated approximately 1000
ppm of PCB in the semivolatile organic analyses. This

observation was quickly verified by a gas chroTnatographic
pesticide — PCB analysis. The state agency was informed
immediately and the planning for a PCB site assessment began.
The site assessment for PCBs was planned using a 20 foot
square grid except for the gasoline underground storage tank
area. This was considered to be the “hot spot” and the grid
was reduced to 10 foot squares. Along the northern border a
row of 10 foot squares were placed next to an electric sub-
station because it might be a possible source of
The site assessment was planned for three days. It included
the sampling of surficial soil in the center of each square
from the surface to a depth of six inches. All of the soils
collected from the 10 foot squares of the grid would be
analyzed using the immunoassay tests. Twenty samples would be
confirmed using the field gas chromatographic method developed
by Dr. Spittler. At least twenty samples would be chosen for
laboratory analysis. These laboratory analyses were performed
by Envirotech Research of Edison, New Jersey, utilizing method
For the iminunoassay, the quality control samples that were
used to ensure that the tests were being run correctly were
blank soils, blank spiked soils and replicates of field
samples. Blank soil was prepared by air drying and sieving to
homogenize the soil. The blank spike soil was a portion of
this blank soil spiked with approximately 20 ppm (mg/Kg) of
Aroclor 1242. This concentration was confirmed by gas
In the three days of testing, a blank soil was run at the
beginning and at the end of a 12 hour work day. Table 1 shows
that all soil blanks gave satisfactory results.
The blank spikes were run at the same frequency as the blank
soils. Table 2 shows that these blank spike results were
satisfactory. However, we did learn that a single technician
running 50 plus immunoassays and 7—8 gas chromatographic runs
for twelve hours on three consecutive days leads to “burn
out.” Thus, there was an omission of a blank spike from our
However, there was an extra replicate as indicated on Table 3.
These replicate analysis did show the necessity to check at
least one sample, K—i, by gas chromatography.

The field gas chromatographic method was a great help in
making us feel comfortable with the immunoassay results.
Tables 4,5, and 6 give the field gas chromatographic results
versus the immunoassay results. This comparison indicates
that the immunoassay may give false positive results at or
near its detection limit of 5 ppm (mg/Kg-wet weight).
The laboratory gas chromatographic method results are compared
to the iminunoassay results in Tables 7,8 and 9. This
comparison gives the same conclusion concerning the
effectiveness of the immunoassay test for PCBs in soil. There
are approximately 20% false positives and all of these samples
have PCB concentrations near the immunoassay detection limit.
There were a total of 151 field samples run by the immunoassay
methodology for PCBs. Twenty one more tests were run on
quality control samples. Mapping the results on the site grid
and the comparison of the immunoassay results with verified
laboratory gas chromatographic results gave adequate
information for a remedial action work plan for the PCB
contamination. It clearly shows that the source of PCBs is
not the electric substation. The data indicates that the
suspected “gasoline” underground storage tanks were used for
the storage of PCB5.
Most importantly for the owner of this site was that the cost
of this site assessment was less than 50% of the projected
costs of a conventional assessment. The savings were realized
in the cost of analyses as well as the time spent on the site.
The added benefit was the ability to select new samples based
on the results of previously run samples. The analytical
costs are given in Table 10.
The false positives were not an important problem in this
work. If there was a false negative, it was not a factor
either. However, we believe that false negatives in this type
of project are inclined to be due to a heterogeneous soil
sample with a concentration near the detection limit.
This program was a success. The imxnunoassay will become an
important methodology in environmental testing in the near

1 Spittler, Thomas M., Ph.D., Field Measurement of PCBs in
i1 and Sediment Using a Portable Gas Chroiuatograph,’ t U.S.
Environmental Protection Agency, Region I, Lexington,

All positive sample values indicate a “not detectecf’
(ND) with a detection limit of 5 parts per million (ppm)
Blank No. Value
1 +0.18
2 +0.14
3 +0.32
4 +0.30
5 +0.27
6 +0.05

20 mg/Kg (Air Dried) PCB 1242
All negative sample values indicate that the sample
contains 5 mg/Kg or more of PCBs.
Blank Spike No. Value
1 -0.33
2 -0.16
3 -0.11
4 -0.10
5 -0.23

Sample Location Run 1 Run 2 Run 3 Run 4 Run 5
D-9 + 0.04 +0.22
G-10 -0.12 -0.07
H-3 +0.15 +0.22
14-14 + 0.24 +0.28
K-i - 0.01 - 0.01 +0.22 +0.09 +0.01
K-5 - 0.43 - 0.25
K-b +0.17 +0.21
Positive values: <5 mg/Kg (wet weight)
Negative va/ues: >5 mg/Kg (wet weight)

‘HITS” (Negative Values)
Immuno GC Value
Sample Location Assay Value mg/Kg (wet wt. )
D-6 -0.47 >400
F-9 -0.09 16
E-1 -0.55 100
J-5 -0.39 18
J-4 -0.56 120
1-7 -0.23 24
D-7 -0.42 40
D-4 -0.51 400
1-5 -0.60 150

CLEAW (Positive Values)
Immuno GC Value
Sample Location Assay Value mg/Kg (wet wtJ
D-10 +0.06 ND
B-6 +0.36 ND
C-8 0.00 2
B-10 +0.07 ND
B-B +0.38 2
H-20 +0.37 ND
J-14 +0.28 ND
ND is not detected with a detection limit of 1 mg/Kg (wet weight).

oo PS ”
Immuno GC Value
Sample Location Assay Value mg/Kg (wet wt. )
G-i0 -0.12 and -0.07 ND
K-i -0.01 and -0.01 6 and 8
and +0.22 and
+0.09 and +0.01
K-9 -0.01 1
D-8 -0.06 4
3 False Positives
1 False Negative
ND is not detected with a detection limit of 1 mg/Kg (wet weight).

(Sum of PCB 1248 and PCB 1260)
“HITS’ (Negative Values)
lmmuno GC Value
Sample Location Assay Value mg/Kg (dry wt. )
B-i -0.51 730
D-4 -0.54 960
D-6 -0.47 700
E-1 -0.55 640
H-6 -0.45 210
J-5 -0.39 36
S-4 -0.22 5

(Sum of PCB 1248 and PCB 1260)
aCLEANI (Positive Values)
Immuno GC Value
SamDIe Location Assay Value mg/Kg (dry wt. )
A-15 +0.32 0.4
A-16 +0.33 0.8
A-17 +0.37 ND
A-18 +0.06 2.7
A-26 + 0.34 0.5
B-6 + 0.36 1.3
B-b +0.07 0.6
L-14 +0.15 0.6
0-24 +0.67 ND
V-12 +0.63 ND
ND is not detected with a detection limit of 0.1 mg/Kg (dry weight).

(Sum of PCB 1248 and PCB 1260)
.0OPs TM
Immuno GC Value
Sample Location Assay Value mg/Kg (dry wt.. )
G-1O -0.12 and -0.07 ND
K-i -0.01 and -0.01 1.5
and + 0.22 and
+0.09 and +0.01
1 (9 -0.01 1.4
P-2 -0.23 1.7
3 False Positive Values
1 False Negative Value
ND is not detected with a detection limit of 0.1 mg/Kg (dry weight).

immuno Assay $5,160
Field GC Analyses 2,560
(including technician for field GC)
Laboratory GC Analyses* 4,050
Total Analytical invoice: $11,770
*lncludes 2 duplicates, 2 matrix spike/matrix spike
duplicates, and data package.

J.D. Petty , J.N. Huckins, J.A. Lebo, J.L. Zajicek, and J.C. Meadows, National Fisheries
Contaminant Research Center, U.S. Fish and Wildlife Service, 4200 New Haven Road,
Columbia, MO 65201 USA.
Layflat polyethylene and capillary silastic® tubes containing pure triolein (lipid) are shown
to have considerable promise as passive in-situ samplers for nonpolar organic
contaminants in water and potentially from the vapor phase. These semipermeable
membrane devices (SPMDs) appear to simulate the bioconcentration of hydrophobic
contaminants by aquatic organisms. Bioavailable organic contaminants in water passively
diffuse through the thin membranes into the enclosed lipid. Contaminant molecules
associated with particles and dissolved organic carbon are excluded because the
transport corridors through these dense polymers are <10 A in cross- sectional diameter.
Equilibrium SPMD-triotein-water partition coefficients ( I c) of test chemicals as determined
by static tests closely correspond to the octanol-water partition coefficients of the same
chemicals, resulting in enriched analyte concentrations in triolein. Moreover, enrichment
of hydrophobic organic contaminants (eg. PCBs) from airborne residues appears to be
extremely facile. Thus, SPMDs with 1 .0-g of triolein achieve sampling rates for PCB
residues from water of 5 L/day at 18°C and . 4,000 L/day from air at 26°C.
Contaminants concentrated in the lipid are recovered from intact SPMDs (polyethylene)
using organic solvent dialysis.
Passive air monitors or dosimeters are widely accepted for determining occupational
exposure of workers to ambient organic vapors (1,2). They use diffusion coefficients of
vapors through a quiescent air gap or a permeable membrane to obtain a sampling rate
that is proportional to exposure time and to the mean airborne concentrations. However,
this integrative approach has seldom been applied to contaminants in aquatic
environments. Huckins et.al. (3) developed a lipid-containing semipermeable membrane
device (SPMD) for in-situ integrative monitoring of aquatic contaminants. The SPMD is

conceptually similar to passive air monitors and fills a gap in current analytical and
biomonitoring techniques for organic contaminants.
The SPMD (3) is constructed from virgin (no additives) layflat tubing of low density
polyethylene (PE), but other nonporous polymers such as polypropylene (PP), silastic,
plasma-treated silicon or silicon-PP laminates (4) can be used. Polymeric films used in
SPMDs are referred to as nonporous, although transient cavities with diameters up to Ca.
ioA are formed by random thermal motions of polymer chains (5,6). Due to the
extremely small and dynamic nature of cavities or transport corridors in most of these
membranes, permeant molecules are considered to be “solubilized” by the polymer (7,8).
Because the cross-sectional diameters of many environmental contaminants (9) approach
the maximum size of membrane transport corridors, only dissolved organics should
diffuse through the membrane and be concentrated.
SPMDs contain a thin film or small-diameter plug of a high molecular weight ( 600
daltons) neutral lipid, such as triolein; purified lipid extracted from a representative
organism, or lipid-like synthetic compounds. The capacity of triolein-containing SPMDs
to concentrate an organic chemical is delineated by its K (equilibrium triolein-water
partition coefficient). Contaminant residues sequestered in triolein are readily recovered
from intact PE SPMDs by dialysis in organic solvent (10).
Upid-containing SPMDs have been used in several environmental settings (3,11-13), and
their ability to sequester trace concentrations of both persistent (polychiorinated biphenyls
(PCBs) etc.) and biodegradable (PAH5) organic contaminants has been demonstrated.
Prest et al. (11) compared the devices to freshwater bivalves ( Corbicula fluminea ) for
environmental contaminant monitoring in the Sacramento and San Joaquin Delta. He
recommended further development of the SPMD approach based on ease of use,
interpretability of data, and low concentrations of interfering substances in SPMD blanks
relative to control bivalves. Also, Gray and Spacie (14) found that SPMD K s of lindane
and trifluralin were much closer to lindane’s and trifluralin’s fish bioconcentration factors
than the concentration factors in hexane-filled dialysis (cellulose) tubes (15). They
concluded that PE-SPMDs have greater promise for field application because of their
durable construction and high contaminant concentration potential.
The use of SPMDs for monitoring and removal of organic vapors also holds considerable
promise. The high surface area of the PE membrane relative to the low volume of
enclosed triolein permits very high sampling rates of ambient vapors. In addition, the
presence of a thin liquid phase on the exterior PE membrane surface, consisting of low
molecular weight lipid impurities in 95% triolein, appears to enhance the uptake of organic
vapors by SPMDs. Presented herein are the results of selected studies from our ongoing
research designed to define the applicability of the SPMD technology for sampling both
aqueous environments and air-borne hydrophobic pollutants.

The experimental details of the preparation and exposures of SPMDs and subsequent
analysis of SPMD sample dialysates have been published previously (3, 10, 12, 13, 16).
We choose an urban creek for the first field application of SPMDs that specifically targeted
polycyclic aromatic hydrocarbons (PAH) residues. The creek runs through a local
Midwestern city, and our primary criterion in choosing it for sampling sites was its
proximity to our laboratory. During low water conditions, the depth of the creek ranges
from about 5 to 80 cm. The creek bottom consists of rock, gravel, and sand.
For sampling the creek, we deployed the SPMDs in the protective shrouds of galvanized
steel. Use of the shrouds enabled us to place SPMDs in the horizontal orientations
necessary for sampling this shallow stream. The shrouds protected the SPMDs from
abrasions and protected the sequestered PAH5 from light. Exposures were conducted
for 21 days.
As part of our ongoing research designed to delineate the functional processes of
contaminant uptake by the SPMDs and to define the kinetics of contaminant enrichment,
SPMD membranes consisting of PE layflat tubing and silastic tubing were compared in
flow through exposures to 2,2’,5,5’-tetrachlorobiphenyl (TCB). In general silastic tubing
has greater polymeric free volume than PE and even though the wall thickness of SPMD
silastic tubing was 3.2.-fold greater than the PE tubing, it was expected to concentrate the
2,2’ ,5,5’-TCB at approximately the same rate. This arises because the permeability
coefficient of nonelectrolyte organics in silastic membranes are, generally much greater
than the corresponding values for low density PE membranes of the same thickness.
Figure 1 and 2 illustrate the uptake of 2,2’,5,5’-TCB by the silastic and PE SPMD
configurations. Following seven days of exposure, the concentration factor of 2,2’,5,5’-
TCB in the triolein of silastic SPMDs was approximately 6,000 whereas the concentration
factor in triolein of PE SPMDs for identical exposure conditions was approximately 20,000.
These data indicate that for PCBs the permeability or sampling rate of silastic and PE
SPMDs of the same membrane thickness, may be similar.
Huckins et al. (3) suggested that membrane permeation is the rate-limiting step, ie.
highest resistance to mass transfer, in the uptake of several hydrophobic organic
contaminants (log Iç,,s 5 to 8) by SPMD lipid. Membrane control of analyte sampling
rate is required to accurately estimate average concentrations of contaminants in water
or air (unless equilibrium is achieved during exposure interval), because the variance

associated with intrinsic properties of SPMDs (membrane and triolein) can be minimized,
whereas the minimization of variance associated with systems controlled (extrinsically) by
boundary layers (3,7), ect., Is very difficult. The membrane control design also enables
the use of less complex kinetic models to estimate analyte water concentrations from
SPMD residue data.
Depending on the physicochemical properties of the contaminant and the components
of the selected SPMD design, equilibrium or uptake kinetic modeling approaches may be
needed (16) for the estimation of contaminant concentrations in water. Assuming the
exposure time is sufficient to reach 0.9 of the K of a particular contaminant (limitations
of this approach are delineated by Huckins et al. [ 16]), then its water concentration can
be estimated by the pseudo-equilibrium relationship
C = C /0.9
where C is contaminant concentration in water and C is its concentration in SPMD
triolein. The PE membrane represents a major portion of the SPMD mass and it
contains a significant amount of sequestered contaminant residues. Calculation of C
from contaminant residues recovered in dialysates can be done by using the following
C = Amd/ (M + KmtMm),
where nd is analyte mass in the SPMD diatysate M and Mm are the masses of the
triolein and the membrane, and K is the distribution coefficient of a contaminant in
the membrane and triolein phases.
For contaminants with high ic’s (. . 2 X 10k), kinetic modeling approaches are usually
more appropriate. Based on the definition of solute transport in polymers, linear
SPMD uptake can be described as
C = (P}
fitted to data from the flow-through exposures of SPMDs to C 14 -2,2’,5,5’-TCB and the
model estimate of C s were within 50% of the measured values.
To determine the effectiveness of SPMDs for sampling pollutants in the environment,
SPMDs were deployed in a small urban stream suspected to be contaminated with
PAHs. The reproducibility of the SPMDs for monitoring aqueous PAH residues is
exemplified by Figure 3. in Figure 3 we present mirror imaged chromatograms
representing duplicate SPMDs deployed in the urban creek. The SPMDs were
exposed for 21 days. Because of the intrinsic reproducibility of the SPMD uptake of
PAHs these simple devices are extremely useful for determining the relative degree of
PAH contamination in aquatic sites where conditions such as temperature and
periphytic growth are similar. Research on the effects of temperature variation and the
efficacy of antifouling agents is underway.
Another area of our current research is the use of the SPMD technology for
concentrating hydrophobic organic contaminants from the atmosphere. It has long
been recognized that areial transport of organic contaminants occurs on a global basis
(18,19). Such materials exist in both the vapor phase and associated with particulate
A variety of techniques have been employed to measure both wet and dry deposition
of atmospheric contaminants (20,21). These techniques include glycerol-coated plates
or pans, Teflon® sheets, containers of water, polyurethane foam plugs, carbon or
other adsorbent traps. While great progress has been made in improving active air
samplers, these devices suffer the consequences of complexity and mechanical
operation. Passive air samplers, either diffusion or permeation type (e.g. personal
monitors) are attractive due to their simplicity but generally have low sampling rates
(total for device) because of the low surface area of the tube (air diffusion zone) or
membrane sampler relative to the volume of trapping media. SPMDs have high
surface area (membrane and their exterior lipid film) relative to the volume of trapping
media, permitting membrane control of sampling rates, while achieving very high total
sampling rates.
We exposed SPMDs to the laboratory air at the National Fisheries Contaminant
Research Center (NFCRC). The SPMDs were exposed for a period of 14 days and
subsequently analyzed by gas chromatography/electron capture detection. The
chromatogram of an air sample and a control SPMD sample are presented in Figure 4.
The major portion of the electron capture response from the air sample is associated
with PCBs. Indeed, the PCBs sequested represent 7pg total residue.
In order to estimate the sampling rate of the SPMDs exposed to the laboratory air, we
used a standard method for determining PCB5 in air (23) as a benchmark. Thus, the
NIOSH method resulted in a determination of 30 ng/m 3 in the laboratory atmosphere.
Using the 7 pg total PCB residue sequestered in a sampler array consisting of 3

SPMDs ( 0.46 g triolein/SPMD) and the 30 ng/m 3 determined by the standard
NIOSH method, it can easily be ascertained that each SPMD sampled approximately 4
m 3 of air per day. This is an extremely facile sampling rate for a passive sampler. We
are currently exploring the kinetics of SPMD uptake of organic pollutants and will field
test the technique in the near future.
The SPMD technology has been demonstrated to be a simple and highly efficient
method for sequestering hydrophobic organic contaminants from both water and air.
The SPMDs provide a means of estimating the presence and relative amounts of a
broad spectrum of bioavailable (assuming no food contribution) organic contaminants.
The SPMD technique allows the determination of episodic contamination and
pollutants (e.g. PAH5) rapidly metabolized by sentinel organisms.
The extension of the SPMD technology into the area of air sampling potentially
provides an efficient method for determining non-particulate bound residues without
the problems associated with filtration. The passive nature of the SPMD technology
lends itself to a wide variety of sampling situations.
1) Underhill, D.W.; Feigley, G.E. Anal. Chem. 1991, 63, 1011-1013.
2) Fowler, W.K. Am. Lab 1982, 14, 80-87.
3) Huckins, J.N.; Tubergen, N.W.; Manuweera, G.K. Chemosphere 1990, 20, 533-
4) Zander, A.K.; Chen, J.S. and Semmens, M.J. Wat. Res. 1992, 26, 129-237.
5) Lieb, W.R.; Stein, W.D. Nature 1969, 224, 240-243.
6) Flynn, G.L.; Yalkowsky, S.H. J. Pharm. Sd. 1972, 651, 838-852.
7) Comyn, J., Editor, “Polymer Permeability” 1985; Elsevier Applied Science
Publishers LTD, New York, NY, p 383.
8) Hwang, S.T.; Kammermeyer, K., Editors, “Membranes in Separations” 1975;
Robert E. Krieger Publishing Company, Inc. Malabar, FL, p 559.

9) Opperhuizen, A; V8kJB, EW.; Gobas, FA.P.C.; Liem, D.A.K.; Steen, J.M.D.
Chemosphere 1 6 , 14, 87i-t
10) Huckins, J.N.; Tubergen, M.W.; Lebo, J.A.; Gale, R.W.; Schwartz, T.R. J. Assoc.
Off. Anal. Chem. 1990, 73, 290-293.
11) Prest, H.F.; Burns, S.A.; Jarman, W.M. 12th Meeting of the Society of
Environmental Toxicology and Chemistry, EMAP Session. 1991, Poster No. 323, p
12) Huckins, J.N.; Meadows, J.C.; Manuweera, G.K.; Lebo, J.A. 12th Meeting of the
Society of Environmental Toxicology and Chemistry, EMAP Session. 1991, Poster
No. 315, p 217.
13) Lebo, J.A.; Zajicek, J.L.; Huckins, J.N.; Petty, J.D. “A Demonstration Method for
the Application of Semipermeable Membrane Devices to In Situ Monitoring of
Environmental Wa rs for Po4ycycUc Aromatic Hydrocarbons” Chemosphere.
1992, In Press.
14) Gray, M.A.; Spacie, A. 12th Meeting of the Society of Environmental Toxicology
and Chemistry, EMAP Session. 1991, Poster No. 272, p 206.
15) Sôdergren, A. En*on. Sd. Techno. 1987, 21, 855-859.
16) Huckins, J.N.; Manuweera, G.K.; Petty, J.D.; Lebo, J.A.; Mackay, D. “Lipid
Containing Semipermeable Devices for In-Situ Estimation of Time Weighted
Aqueous Concentrations of Organic Contaminants; Background, Theory and
Model Developmert’ 1 2, StA rnitted to ES&T.
17) Neff, J.M. 1985. G.M. Rand and S.R. Petrocefli, Editors. “Fundamentals of
Aquatic Toxicology.” Hemisphere Publishing Corp., Washington, D.C., USA, p
18) HeImig, D., J. MCifler and W. Klein. Chemosphere. 1989, 19, 1399-1412.
19) Bidleman, T.F. and G.E. Olney, Science. 1974, 183 , 516-518.
20) Swackhamer, D.L., 8.0. McVeety and R.A. Hites, Environ. Sd. Technol. 1988, 22,
21) Broman, D., C. Nat, C. Rotif and V. ZehOhr, Environ. Sci. Technol. 1991, 25, 1841-

22) ElIer, P.M., NIOSH Manual of Analytical Methods. 3 ed. Cincinnati: U.S.
Department of Health and Human Services. Pubi c Service, Center for
Disease Control, National Institute for Occupational Safety and Health, 1984.
DHHS (NIOSH) publication No. 84-100: 5503-1-5503-5.

Figure 1: Relative concentration factors of 2,2’,5,5’-TCB in
triolein (tube contents), silastic tube containing
triolein (tube membrane) and silastic strips.
Time (Days)
0 2
4 6

c i)
C l )
E- -H rI
I .Q
Lfl 4-
.4. )
U) W d
9- I -
04- i
4 - i a ) a)
o b
o r1
o —
r 1
w o
. .H .,.
H a)
4JrH 4 j
a) -ic
0 5 10 15 20 25 30
Time (Days)

Figure 3:
Chromatograms of dialysates from SPMDs exposed for
21 days in a small urban stream (top) and a
control pond. Mirror Image Traces: two SPNDs
deployed contiguously for 21 days. Bottom Trace:
SPMD exposed in a control pond. Labeled Peaks:
A) naphthalene; B) fluorene; *) l-methylfluorene
(internal standard); C) phenanthrene;
D) fluoranthene; E) pyrene;
F) benz(a)anthracene; G) chrysene;
H) benzo (b) fluoranthene and benzo(k) fluoranthene;
and I) benzo [ a)pyrene
0 10 20 30 40 50 60

TIME 0.00
Chroinatograias of dialysates from SPMDs exposed to
laboratory air for 14 days (top) and blank SPMDs
(bottoi ).
TIME 0.00
Figure 4:

Randy L. Allen , Thomas N. Stewart, Tracy A. Withers, Wayne B.
Manning, Steve B. Friedman, EnSys, Inc., P.O. Box 14063, Research
Triangle Park, North Carolina 27709.
A sensitive and rapid enzyme immunoassay (EIA) was developed using
a inonoclonal antibody (MAb) reagent that detects polyaromatic
hydrocarbons in soil. Derivatives of naphthalene, methyl
naphthalene, phenanthrene and acenaphthene were synthesized with
various types of spacers and conjugated to either bovine serum
albumen (BSA) or bovine thyroglobulin (BTG). A total of 16
different conjugates were used for immunizing both Balb/c and Swiss
Webster mice. Mice responding to immunizations were selected as
splenocyte donors for MAb production. A panel of MAbs were
produced that recognized PAH compounds in a competitive ELk.
An antibody specific for phenanthrene (F35—2Z11) was chosen for
immunoassay development. The detection of phenanthrene in soil
provided a reliable marker for PAH compounds because of its
relatively low volatility and low susceptibility to leaching. An
enzyme—hapten conjugate was prepared for 14kb F35-2Z11 that
demonstrated suitable characteristics of sensitivity, cross—
reactivity and compatibility with extraction buffers. The
resulting EIA demonstrated a sensitivity of less than loppm for
phenanthrene in soil. Extracts of negative soil samples from
different types of soil did not significantly alter the performance
of the assay. Interference from polychiorinated biphenyls, penta—
chiorophenols and petroleum products were negligible. This method
will offer speed and cost-effectiveness over current testing
methods of PAH compounds in contaminated soil which will result in
an increased efficiency of site management.

9 The Field Screening of a Large Site for Pentachiorophenol
Contamination Using an Immunoassay-Based Analytical Method
Kevin R. Carter, EnSys Inc., Research Triangle Park, NC 27709
The contamination of soil and groundwater by pentachlorophenol as a result
of wood preserving operations throughout the country has been a prominent
environmental problem. A new method for the measurement of
pentachiorophenol in environmental samples has been developed that
employs immunoassay technology. Its high specificity, sensitivity, and
freedom from matrix interferences as well as its speed and ease of use make it
ideal for use as a field analysis tool to aid in the delineation of contaminated
areas of soil at former wood treating sites. The application of this method in
the field can result in the saving of both time and money. The results of a
comparison between immunoassay-based testing performed in the field and
GC analysis conducted in a field laboratory on a large wood treating facility
showed that the field method can be used to accurately assess
pentachlorophenol contamination.
Pentachiorophenol (PCP) contamination of soil and water around operating
and closed wood preserving plants has received much attention from
regulatory authorities due to the potential health hazard posed by PCP and its
co-contaminants, such as octachiorodioxin. The specific evaluation of PCP
contamination in environmental samples has previously required the use of
GC or GC/MS methods that are costly and subject to the normal delays
associated with laboratory analytical methods. The recent development of
rugged, accurate field analytical tests for PCP in soil and water has enabled the
rapid screening of contaminated sites, as well as timely remediation
monitoring, and pre-closure verification of clean-up (Mapes j.., 1992).
These tests are based on immunoassay technology and rely on the specific
binding properties of a biological compound known as an antibody to
recognize the target analyte in an environmental sample. The presence of
the analyte is visualized through a colorimetric enzymatic reaction and the
results are compared with those obtained with a standard. The semi-
quantitative result is interpreted with the aid of a battery-powered
comparative photometer.

The EPA’s Environmental Response Team conducted extensive site
characterization at several former wood preserving facilities in the Southeast
during the summer of 1991. The first plant to be investigated (in Brunswick,
GA) had recently ceased operations. Several areas on the 50 acre plant
grounds were heavily contaminated with pentachlorophenol, creosote, and
CCA (chromated copper arsenate). These included the area immediately
surrounding the treating equipment, treated pole storage barns, the transfer
yard, and a wastewater lagoon. The initial characterization was focused on
rapidly assessing soil contamination from each of the wood-preserving
chemicals. A mobile lab was established on-site at the facility in Brunswick
to analyze samples by GC from all of the wood preserving facilities. Because
the cost for this approach was so high and the need in the future for on-site
analytical capability was foreseen, the immunoassay-based test developed by
EnSys was evaluated on this project in parallel with the mobile lab.
The Brunswick site was sampled for PCP contamination assessment using a
100 foot grid with samples taken at the surface and at several depths. The
EnSys immunoassay test system was employed to screen 200 soil samples as
they were collected prior to GC analysis. The advantages of the screening test
in this application included the ability of two field operators to screen up to
100 samples per day where two analytical chemists would only be able to
analyze 25 samples per day for pentachiorophenol. In addition, the EPA
contract field operators needed no sophisticated equipment and little training
to use the field test. The test was capable of ranging the pentachiorophenol
concentration in soil over several orders of magnitude, with a minimum
detection limit of 0.5 ppm. For this project the test was used to assess
pentachiorophenol concentration at thresholds of 25 and 100 ppm.
A high degree of correlation was observed between the GC-FID results and
those obtained on-site with the immunoassay-based field test. Of the 176
samples analyzed using both methods, 147 (83%) exhibited excellent
agreement. There were 21 (12%) false positives and 8 (5%) false negatives
recorded. Several of these (6), however, must be considered provisional,
because for these the field immunoassay results were within 25% of the GC
results. The especially low incidence of false negative results highlights the
utility of immunoassay-based field tests for the delineation of
uncontaminated areas. It should be noted that the GC method was subject to
inaccuracy primarily due to interferences caused by high levels of creosote
and fuel oil in many samples, which makes 100% correlation between the two
methods unlikely in any case. This side-by-side comparison conducted under
actual field conditions demonstrates the accuracy of the screening method.

A field analytical method for the semi-quantitative measurement of
pentachiorophenol based on immunoassay technology has been tested at a
wood preserving facility and the results compared to those obtained with GC-
RD. The correlation observed was very good with an acceptably low level of
false negative results.
Mapes, J. P., K. D. McKenzie, L. R. McClelland, S. Movassaghi, R. A. Reddy, R.
L. Allen, and S. B. Friedman. PENTA RISc- Soil - A Rapid, On-Site Screening
Test for Pentachiorophenol in Soil. Bulletin of Environmental
Contamination and Toxicology, in press.

10 The Assessment of a Site for PCB Contamination Using an
Immunoassay Based Field Analytical Method
Kevin R. Carter, EnSys Inc., Research Triangle Park, NC 27709
The application of field analytical methods to site assessment projects can
result in substantial savings of time and money relative to the conventional
practice of collecting samples and submitting them for laboratory analysis.
Where large numbers of samples must be collected and analyzed, the
analytical turnaround time can severely handicap the progress of a project. A
PCB-spedfic immunassay-based field analytical method was used to evaluate
several hundred soil samples from a site on which electrical transformers
had been stored. The field analysis of over one thousand samples from this
site allowed the contaminated areas to be delineated in a relatively short time
period. A PCB contamination map of the site was created that showed the
pattern of PCB contamination clearly. A series of quality assurance/quality
control (QA/QC) procedures were implemented to support the field analytical
method. The QA data provided the necessary confidence in the field results.
Immunoassay-based field analytical methods can be used to substantial time
and money advantage in site assessment projects relative to the conventional
practice of collecting samples and submitting them for laboratory analysis.
While laboratory methods can provide definitive identification and
quantitation of hazardous chemicals in soil and water, they require highly
trained personnel, expensive capital equipment, and relatively laborious
quality assurance/quality control (QA/QC) procedures to achieve these
results. These factors contribute to substantial delays in obtaining results
following the submission of a sample for analysis. Delays stemming from
the laboratory analysis turnaround seriously reduce the time-effectiveness of
site assessment work. This has a more pronounced effect where large
numbers of samples must be collected and analyzed.
In order to complete the site assessment phase of an environmental clean-up
project where hundreds of samples are required in a timely fashion, either
the samples must be composited, sent to several analytical laboratories, or a
field method must be employed. Although it certainly has valid
applications, compositing of samples often creates detection limit problems.
The use of several laboratories presents a significant QA!QC management

problem. With the appropriate simple quality assurance/quality control
measures, immunoassay-based field screening of environmental samples can
be accomplished at a reduced cost without an accompanying decrease in the
quality of the data necessary to support the decision-making process.
Immunoassay technology has only recently been applied to in-field sample
analysis for environmental contaminants (US EPA, 1990 and Goolsby
1991). Immunoassays rely on the specific binding properties of an antibody to
identify contaminated samples. The presence of the target contaminant is
visualized through a colorimetric enzymatic reaction and the semi-
quantitative results are interpreted with the aid of a portable comparative
Immunoassay-based methods are appropiate for field analysis because:
—the reaction is extremely specific for the target contaminant
—there are few matrix interferences
—the test procedure is simple and rapid
—the method limits false negatives.
Immunoassays can, with a high degree of confidence, establish that sites or
areas are free from contamination. Extensive testing by environmental
consulting firms and by the EPA have shown that immunoassay-based
products can correctly identify over 90% of samples tested as clean versus
contaminated (Taylor ., 1991).
A west coast military base recently proposed to convert a surplus/salvage
equipment storage area to another use. Because the 1.5 acre site had
previously been used to store utility transformers and there was anecdotal
evidence for transformer leakage, the state specified a phase 2 site assessment
plan that included a total of up to 3700 samples to be collected and analyzed
for PCBs. This sampling frequency was necessary to ensure that a small spill
was detected. Since there was no existing record of where transformers had
been stored on the site and surface staining provided no clues, the use of an
intensive grid sampling design was required.
The environmental consultant was faced with the problem of analyzing a

large number of samples on a limited time scale with a budget that did not
match these needs. The solution to the problem was a field analytic l
method that is specific for PCBs and provides the capability to analyze 35-50
samples per day on-site. Originally, a field method based on the detection of
chlorine chemically stripped from PCBs was considered, but rejected due to
the presence of significant levels of chloride ion in the soil that would act as a
positive interferent. The immunoassay-based method was selected because
of its accuracy, PCB-specificity, and sensitivity. This immunoassay field
analytical system is manufactured by EnSys Inc.
The environmental consultant set up the two meter grid required by the
sampling plan and a two-person crew collected surface samples for screening
by the field method. The first round of samples was collected at eight meter
intervals to roughly define contaminated areas. The action level for the site
was 25 ppm PCBs. Analyses of each sample were made at 5 and 50 ppm to
allow the mapping of contaminated zones. One large area and a few smaller
areas were found to be contaminated with PCBs at levels greater than 50 ppm.
Several samples that were shown to contain greater than 50 ppm PCBs on the
eight meter grid were sent to a laboratory for confirmatory analysis by GC
(results from the first batch are shown in Table 1). Over 75% of the field
testing results were confirmed by the laboratory analysis. Where agreement
between the two methods was not observed only false positives were
obtained. The absence of false negatives is an important criterion for a
successful field method.
To identify smaller areas of contamination and more sharply delineate the
larger areas, samples were next collected and analyzed from points located at
four meter intervals. Finally, a relatively small number of samples were
collected at the two meter interval to sharply define any contaminated areas
identified in the broader screens. The results of the field analytical testing are
depicted in Figure 1.
In any project where field data are to be used for decision-making it is
imperative that QA/QC be performed in the field to provide documentation
regarding the validity of the field analytical data. Several different types of
field QA/QC data were collected. The semi-quantative PCB field test kit
contains duplicate calibration standards that are tested in parallel with each
group of three samples. The evaluation of duplicate standards serves to
provide internal test system quality control. The results are deemed valid if
the difference in the results for the standards are within a predefined limit.
In addition, during the course of this project the field test operators routinely
tested performance evaluation (PE) samples supplied for the purpose of
confirming that false negative/false positive specifications for the field test

were met (Table 2). This was primarily focused on the need to verify that
samples with PCB concentrations near the action level, 25 ppm, were
identified properly. This included ensuring that no false negative results at
the 5 ppm test level were recorded for samples containing PCBs. The results
of the PE sample analyses verified the expected performance relative to false
negatives. The repeatability of the test was documented through the testing
of duplicate subsamples at the rate of one for every twenty samples.
Because the PCB contamination covered areas substantially larger than the
smallest grid interval, testing of samples from all of the nodes at that interval
was unnecessary. The total number of samples screened over a 4 week period
by the two-person field screening crew was slightly over 1000. More than half
of those samples tested negative for PCBs. The ability to quickly identify
areas contaminated with PCBs allowed the environmental consultant to
focus on delineating the areas with PCB soil levels in excess of 5 ppm and 50
The project might have proceeded quite differently at a sustantially greater
cost had conventional laboratory-based GC analysis been employed. The
performance of 1200 field analyses at $34 per analysis (including labor) and 150
laboratory confirmatory analyses at $150 per sample comes to under $64,000.
By way of comparison, a strictly laboratory-based approach to this project
would have cost $180,000 (1200 samples at $150 per sample). Due to the need
for an approved laboratory and limited available laboratory capacity the
project would have required an estimated 4 months to obtain all of the
analytical data. This contrasts sharply to the 4 weeks required for the
completion of the site assessment using the field analytical test.
Following the field analysis of over one thousand samples using an
immunassay-based field method, several areas of PCB contamination were
clearly and completely delineated. The data obtained allowed a “map” of PCB
contamination on the site to be constructed in a short period of time. Quality
assurance data were collected during the course of the project to document
the quality of the field analytical testing.
US Environmental Protection Agency. Evaluation of the Westinghouse
Bioanalytic Systems PCP Immunoassay. EPA 600/X-90/146, 1990.

Goolsby, D. A., E. M. Thurman, M. L. Clark, and M. L. Fomes. Immunoassay
as a Screening Tool for Triazine Herbicides in Streams: Comparison with Gas
Chromatographic-Mass Spectrometric Methods. in Immunoassays for Trace
Chemical Analysis , M. Vanderlaan, L. H. Stanker, V. E.Watkins, and D. W.
Roberts, eds. American Chemical Society, Washington, D.C., 1991.
M. J. Taylor, S. D. Wesson, and N. Monacella. Laboratory Evaluation of a
Rapid Immunoassay Screen Test for Determination of Polychiorinated
Biphenyls in Soil. GRI Contract No. 5089-253-1836, Task 4 (to be published),
June 1991.

Table I
QA Data: Confirmatory Laboratory Analyses
Sample Field Test Results GC-ECD Results
( ppm) ( ppm )
2-52 50 3310
4-28 50 1440
4-48 50 2900
6-24 5 50 492
6-32 50 616
6-36 50 237
8-48 50 40
10-32 50 112
10-36 50 145
12-48 50 19
14-32  5  50 114
20-4 5 <50 <0.5
24-40 50 86
28-4 <5 <50 <0.5
32-4 <5 <50 <0.5
34-32 50 3.5
38-36 <5 <50 4.5

Figure 1
PCB Contamination Map of Military Site
* 16
0 4 8 12 16 20 24
28 32 36 40
u.a U
a a..
a. •
44 48 52 56 60
. U..
50 PPM

Table 2
QA Data: Performance Evaluation Sample Results
Sample Type Field Test Results
negative soil <5 <50
negative soil <5 <50
negative soil  5 <50
negative soil <5 <50
negative soil <5 <50
negative soil <5 <50
negative soil <5 <50
negative soil <5 <50
27.5 ppm Aroclor 1260 soil  5 <50
27.5 ppm Aroclor 1260 soil  5  50
27.5 ppm Aroclor 1260 soil  5 <50
27.5 ppm Aroclor 1260 soil  5 <50
27.5 ppm Aroclor 1260 soil  5 <50
27.5 ppm Aroclor 1260 soil  5 <50
27.5 ppm Aroclor 1260 soil  5  50
27.5 ppm Aroclor 1260 soil  5 <50
24.7 ppm Aroclor 1260 soil  5  50
24.7 ppm Aroclor 1260 soil  5 <50
24.7 ppm Aroclor 1260 soil  5 <50
24.7 ppm Aroclor 1260 soil  5  50
24.7 ppm Aroclor 1260 soil  5 <50

11 The Application of Immunoassay -Based Field Analytical Methods
Kevin R. Carter. EnSys Inc., Research Triangle Park, NC 27709
The application of field analytical methods to hazardous waste site
assessment can significantly enhance the quality of the site assessment effort
and result in substantial savings of time relative to the conventional practice
of collecting samples and submitting them for laboratory analysis. With the
appropriate field method and adequate quality assurance/quality control
(QA/QC) this can be accomplished without an accompanying decrease in the
quality of the data necessary to support the decision-making process.
The recent development of several low-cost, rapid, analyte-specific field
analytical tests based on immunoassay technology has provided on-site tools
for the cost-effective assessment of soil and water contamination. Extensive
application of immunoassay-based field analytical methods for
pentachiorophenol, polychiorinated biphenyls, and petroleum hydrocarbons
have shown them to be accurate, rugged, and reliable. In order to fully
realize the advantages offered by these tests, a project must be designed from
the beginning with field analysis in mind. There are two principal
operational areas in which the typical sampling or remediation plan is
significantly impacted by the use of a field analytical tool: the sampling
frequency and quality assurance/quality control procedures.
The optimum sampling frequency for projects where field screening is
employed is usually greater than it would have been using conventional
practice. The rapid feedback regarding location and level of contamination
that is available and the lower cost per sample collected and analyzed argues
that a greater frequency of sampling, overall, is of benefit to the quality of the
site assessment or remediation effort. Examples of modified sampling plans
appropriate to immunoassay-based field testing will be presented and
In order for the results obtained with any analytical method to be reliable and
accurate, a sampling QA/QC plan with data quality objectives (DQO) must be
developed and followed. A specific plan should be developed based on the
goals of the project for which it is to be used and the decisions that are
anticipated to follow from the results. QA/QC procedures have been
developed for immunoassay-based analytical methods to enable their
application in the field with resulting data of documented quality.

In many cases, the application of field analytical methods to site assessments
can result in substantial savings of time and money relative to the
conventional practice of collecting samples and submitting them for
laboratory analysis. In order to fully realize these advantages, however, the
project plan must be designed from the beginning with field analysis in mind.
There are two principal operational areas in which the typical sampling plan
is significantly impacted by the use of a field analytical tool: the sampling
frequency and quality assurance/quality control (QA/QC) procedures.
Immunoassay-based field tests possess the basic technical qualifications to
make them extremely useful for the assessment of soil contamination.
The environmental immunoassay technique relies on a molecule referred to
as an antibody that is developed to have a high degree of affinity for the target
analyte. The high specificity and high affinity of the antibody is coupled with
a very sensitive colorimetric reaction that provides visualization of the
result. All of this chemistry is accomplished with a small number of
solutions that are applied to the processed sample or a dilution thereof. Soil
samples require a simple extraction step and subsequent filtration of the
extractant. A wide range of analyte concentration in samples is
accommodated through conventional serial dilutions. Extraction,
normalization, and sample dilutions can all be preformatted for ease of use in
the field.
The attributes that make immunoassay tests ideal for field screening include:
• immunoassay-based tests are analyte-specific;
• they are accurate and precise;
• they are easy to use;
• they are rapid (<30 minutes);
• immunoassay-based tests are not significantly affected by the composition
of the sample (soil or water) or the presence of most other compounds.
In addition to these characteristics, a properly designed immunoassay-based
field screening method should possess a very low false negative detection
rate. This property has the practical consequence of not treating a substantial
volume of contaminated soil as though it were “clean.” Viewed another
way, this impacts the conduct of the site assessment in a positive fashion. It
is not necessary to subject an area that has been screened and found to be free

of contamination relative to the action level to a large amount of
confirmatory sampling and laboratory testing.
The optimum sampling frequency for projects where field screening is
employed is usually greater than it would have been using conventional site
assessment practice. The rapid feedback regarding location and level of
contamination that is available and the lower cost per sample collected and
analyzed argues that a greater frequency of sampling, overall, is of benefit to
the quality of the site assessment effort. It is widely acknowledged that in soil
matrices the chief impediment to obtaining an accurate picture of
contaminant distribution is the heterogeneity associated with the matrix.
One approach to overcoming this problem is simply collecting and analyzing
more samples. However, this solution carries with it a proportionally higher
cost. The use of field methods makes it possible to get the additional data
points within the original cost constraints. In addition, this information can
be provided in the course of a single field mobilization with no added
expense for demobilization and redeployment of the sampling team for
successive sampling trips.
The application of an immunoassay-based field method requires a somewhat
different approach in the design of a sampling plan. It is impossible to create
a generic sampling plan to address the use of field analytical methods, because
any sampling plan depends on several factors that are site-, analyte-, and
matrix-specific. These factors include the pattern and magnitude of
contamination and the relative cost of sampling/analysis relative to the
anticipated cost of the remedy. Several general guidelines do exist, however.
The most effective use of the strengths of field screening is obtained when the
sampling/analysis activity is conducted to increasingly smaller intervals as it
progresses, in an interactive fashion. The starting sampling interval is
determined by consideration of many factors specific to the situation.
For example, a large site with relatively large areas of soil contamination
might initially be sampled at the nodes of a 100 foot grid. Node samples
testing “clean” would require no additional sampling in their adjacent areas.
On the other hand, node samples analyzing above the action level would
trigger the collection of additional samples at points 50 feet from the node in
the direction of adjacent clean nodes. This type of iterative increase in
sampling resolution allows the boundaries of contaminated areas to be
defined relatively precisely with fewer samples (and at a lower total cost) than
would be required to achieve the same degree of precision with uniform grid

sampling in a single trip, sampling/laboratory analysis scenario. But the
increase in precision is obtained with more samples than would be
collected/analyzed for the same cost using conventional laboratory analysis to
evaluate the extent of contamination using a fixed grid sampling plan. As a
result, the quality of the site assessment is enhanced by field testing without
additional expenditure. The same principle of iterative sampling and field
testing can be applied where soil contamination is expected to be more
localized, as well. However in this case, a smaller beginning node spacing
would be utilized.
In order for the results obtained with any analytical method to be judged
reliable and accurate, a sampling QA/QC plan must be developed and
followed. The quality of data is dependent both on its analytical accuracy and
precision and its utility to the user in making a decision with a defined degree
of confidence. The use of a “high-powered” analytical technique does not
ensure that the data derived therefrom are either accurate or useful. To
obtain the maximum value from the money spent to collect data, it is
necessary to define data quality in such a way as to not only specify accuracy
and precision limits, but to include the intended use of the data, as well.
To ensure that data derived from field measurements using immunoassay-
based methods are adequate for the task for which they are being obtained it is
useful to follow the guidelines developed for the Superfund program by the
EPA Quality Assurance Management Staff. A sampling QA/QC plan with
data quality objectives (DQO) should be developed based on the goals of the
project for which it is to be used. The EPA has developed the data quality
objectives concept to enable the project manager to ensure that all
measurements made in association with an environmental project produce
data of established quality. The key factor in establishing the quality of data
generated in the course of an environmental project is the level of QA/QC
performed. The decisions to be made with the data and the consequences of
failure dictate the level of QA/QC required for a given sampling plan. The
EPA has defined three levels of QA/QC for the Superfund program (Ryti and
Neptune, 1991) and Removal program (US EPA, 1990) that are valuable for
other types of projects, as well. They are summarized briefly below:
QA1: This is a screening objective that is typically used for rapid, preliminary
assessment of site contamination by field methods. Data collected for this
objective have neither definitive identification of contaminants nor
quantitation of their concentrations. No QA data are collected; a calibration

or performance check of the method is required along with verification of the
detection level.
QA2: This is an objective that calls for the verification of the analytical results
obtained by either field or laboratory methods. This objective is most
frequently applied to field methods or quick laboratory methods. A portion
of the results (lO%) are selected for qualitative (identification) and
quantitative (concentration level) verification. Assuming that the portion
selected for verification is representative of the whole, the project manager is
given a measure of confidence regarding the whole body of data.
QA3: This is a rigorous objective that specifies the assessment of the accuracy
and and precision of the analytical determination, as well as the verification
of the identity of the analyte. This level of QAIQC is usually only performed
on data collected using laboratory methods. It provides the highest level of
confidence, allowing critical decisions to be made based on results obtained.
Field analytical methods based on immunoassay techniques can readily be
incorporated into projects to meet the data quality objectives for site
assessments. Due to their analyte-specific nature, immunoassay-based field
analytical methods are best-suited to situations where the nature of the
contamination has already been determined using more general methods,
such as CC/MS. Only where strong historical evidence exists for
predominance of a single analyte should an immunoassay-based test be used
for a Phase I type of site assessment.
The requirements necessary to satisfy QAI level objectives are quite readily
achieved using immunoassay-based field methods. In addition to sample
documentation only calibration need be performed. Regular calibration is
always a component of a properly designed immunoassay-based field method.
The decisions resulting from data generated as a consequence of a Phase 2 site
assessment activity usually dictate that several elements of a QA2 level
QA/QC plan be implemented in conjunction with immunoassay-based field
testing. Such a plan contains the features detailed below:
A. Sample documentation
1. Location, depth
2. lime and date of collection and field analysis
B. Field analysis documentation - provide raw data, calibration, any
calculations, and final results of field analysis for all samples screened
(including QC samples)

C. Method calibration - frequent (several times daily) calibration; this
should be an integral part of the immunoassay test
D. Site-specific matrix background field analysis - collect and field analyze
uncontaminated sample from site matrix to document matrix effect
B. Duplicate sample field analysis - field analyze duplicate sample to
document method repeatability; at least one of every 20 samples should
be analyzed in duplicate
F. Confirmation of field analysis - provide confirmation of the quantitation
of the analyte via an EPA-approved method different from the field
method on at least 10% of the samples; choose 2% or at least two
representative samples testing below the action level or lowest test level
and 8% or at least two representative samples testing above the action
level; provide chain of custody and documentation such as gas
chromatograms, mass spectra, etc.
G. Performance evaluation sample field analysis (optional, but strongly
recommended) - field analyze performance evaluation sample daily to
document method/operator performance
H. Method blank, rinsate blank field analysis (optional)
I. Matrix spike field analysis (optional) - field analyze matrix spike to
document matrix effect on analyte measurement
In addition to the QA/QC elements above, we have found that a simple
quality control procedure used at the time of the analysis can markedly
improve the performance of immunoassay-based field tests in the hands of
field workers. Due to the temperature dependence of the immunochemistry,
standards are normally tested alongside of samples in the field. For a semi-
quantitative test a standard at a single concentration level is used to simplify
the test for quick, easy field application. Little additional complication and
time results from concurrently testing a duplicate of that standard, but the
duplicate provides an easy field check of the operator’s technique for that set
of samples/dilutions tested.
Following the immunoassay-based field analytical method instructions and
providing for the QA/QC requirements described above will result in cost-
effective data at the level of confidence necessary for most Phase 2 site
assessment-based decision-making.

Field analytical methods provide a means for improving the quality of site
assessment work with no cost or data quality penalty. Immunoassay-based
field analytical methods have performance characteristics that are well-suited
for their use in assessment of soil contamination. With the application of
appropriate QA/QC procedures, immunoassay-based methods can be used to
generate data that meets QAI and QA2 level objectives.
Ryti, R.T. and D. Neptune. Planning Issues for Superfund Site Remediation.
Hazardous Materials Control 4(6), 47-53,1991.
US Environmental Protection Agency. Quality Assurance! Quality Control
Guidance for Removal Activities. EPA/540/G-90/004, 1990.

Steven C. Goheen , Margaret McCulloch, Sandra K. Fadeff, Deborah S. Skiarew, Berta L. Thomas, Roger
M. Bean, Georgia K. Ruebsamen, Shantha M. Anantatmula, and Robert 0. Riley. Pacific Northwest
Laboratorya , Richiand, Washington 99352; Craig S. Leasure. Los Alamos National Laboratory, Los
Alamos, New Mexico 87545; James S. Poppiti, Laboratory Management Branch, U.S. Department of
Energy, Germantown, Maryland 20874
Abstract: The document DOE Methods for Evaluating Environmental and Waste Management Samples
(DOE Methods) is a guidance document planned for release in October 1992, with subsequent updates to be
published every 6 months. DOE Methods contains guidance and validated methods for use by DOE and
DOE contractor laboratories in waste and environmental sampling, and in analyzing the constituents of
waste and environmental samples. The purpose of the document is to provide methods for the sampling and
analysis of radioactive and radioactive-hazardous waste and environmental samples that are not currently
provided by existing guidance manuals (i.e., EPA ’s SW-846). The development of DOE Methods is
supported by the Laboratory Management Division (LMD) of DOE. LMD is charged with providing the
required analytical facilities and analysis capabilities to cost effectively support DOE’s environmental
restoration and waste management (EM) programs. DOE Methods will support the activities that will
determine whether EM actions are needed as defined by the DOE or the U.S. Environmental Protection
Agency (EPA).
A database called the DOE Methods Compendium Database (database) is being developed in parallel with
DOE Methods. Sampling and analytical methods have been and continue to be acquired from DOE and
DOE coniractor labs and entered into the database. Similar methods are consolidated and subjected to a
review, validation, and approval process. Methods meeting certain validation criteria (i.e., QC data
availability, internal/external laboratory verification) are selected for inclusion in DOE Methods.
Until now, DOE and DOE contractor labs have relied on the EPA, other agencies, and professional
societies for sources of compiled sampling and analysis methods. Historically, laboratories within the DOE
complex have developed waste and environmental characterization procedures independent of one another.
This practice has led to the accumulation across the DOE complex of a multitude of site-specific sampling
and analysis procedures. Acquisition and consolidation of these procedures is the goal of the DOE Methods
Compendium Program (DMCP). Through DOE Methods, DMCP will provide guidance for implementing
sound sampling and analysis strategies and providing the most up-to-date methodologies in support of EM
programs. Emphasis will be placed on incorporating radiochemical analysis methods, standard methods that
have been modified to accommodate highly radioactive samples (i.e., methods from the EPA’s SW-846 (1),
Contract Laboratory Program (CLP), and 500 and 600 series), and field and waste sampling methods.
Sampling methods will include those for both low and high levels of radioactivity.
Several documents were used in the development of DOE Methods (1-5) which comply with pertinent
DOE regulations. DOE Methods will be reviewed quarterly and updated biannually. As part of the review
process, new and revised methods and chapters will be subjected to external review each quarter. Reviewers
comments will be incorporated so that the revised methods or chapters can be inserted in the next updated
version of the document.
a Pacific Northwest Laboratory is operated by Battelle Memorial Institute for the U.S. Department of
Energy under contract DE-ACO6 76RL0 1830.

Program Roles and Responsibilities. Figure 1 depicts the overall organizational structure for
the management and conduct of the DMCP. The LMD has overall responsibility for the DMCP. DOE has
engaged the support of several organizations to meet the goal of this program. These groups and their
responsibilities follow:
Los Alamos National Laboratory (LANL) is the lead organization for the technical conduct of the
program. LANL-specific responsibilities include database development, management, and application.
• Pacific Northwest Laboratory (PNL) is the custodian of DOE Methods and is responsible for its
development, periodic updating, and issue. Both LANL and PNL work collaboratively to acquire
procedures from laboratories across the DOE complex to support database and DOE Methods
• Assisting LANL and PNL is the National Air and Radiation Environmental Laboratory (NAREL).
NAREL is responsible for independently assessing the methods, with special attention on providing
guidance to the program on whether the methods are consistent with EPA monitoring and/or regulatory
needs and quality assurance and related ANSI/ASTM standards. NAREL also provides the primary
regulatory interface for radioactive and mixed-waste (or radioactive and hazardous sample) analysis
Database and DOE Methods Dynamics. Since the middle of FY 1991, LANL and PNL have
been acquiring procedures from the DOE and DOE contractor analytical laboratories within the DOE
complex. Figure 2 indicates current LANL and PNL communication networks used to obtain procedures.
PNL’s network encompasses 6 operations offices and currently includes 25 laboratories and 1 program
office. DOE sites targeted by LANL are also of similar scope. To date, the total number of procedures
entered into the database based on the combined efforts of LANL and PNL is approximately 1500.
Figure 3 summarizes the interrelationships of LANL and PNL database methods compendium
development activities. Network-received procedures are entered into the database by LANL staff by
transfonning hardcopy information into text and image files and storing in the database. Both LANL and
PNL maintain indexed hardcopy files of procedures. LANL’s files are used in support of database
development and to respond to user requests for procedures while PNL’s repository is used in the
development of DOE Methods. Hard copies of procedures from the LANL repository can be ordered through
the database.
The database is accessed through the SEARCHmate system, which permits searching by procedure
type or by selected procedure elements (i.e., analyte, matrix, detection limit). Through this process, users
may identify existing procedures that meet their needs, thus avoiding costs associated with developing their
own procedure or using an inferior procedure that would impact data quality. The database may also help to
identify gaps in methodology, thus helping DOE identify where investments are needed in methods
development projects. The database contains over 1,500 full-text procedures, most of which are
radiochemical procedures. Figure 4 shows the distribution of a random sampling of 512 database
procedures. For the development of DOE Methods, PNL is responsible for preparing all chapters and
entering methods in the documenL Network-received procedures are screened through a prioritization
scheme that gives high-level mixed (radioactive-hazardous) waste and environmental methods (sampling and
analysis) the highest priority, followed by low-level mixed waste methods. Methods for sampling and
analysis of radioactive materials have third priority, and methods for non-radioactive samples are given
fourth priority.
In situations where a number of common procedures have been received from DOE laboratories which
are not available in commonly used documents, a single method will be developed that incorporates the
attributes of all procedures in that batch. At the other extreme, unique methods will be evaluated on their
own merit. Developed procedures are subjected to a review, validation, and approval process. A draft plan

has been prepared which defines 1) the mechanism by which procedures from the database become part of
DOE Methods, and 2) the criteria used for establishing their level of validation (Figure 5). The placement
of a procedure in a particular validation level is partially determined by submitted performance data.
Methods that successfully complete the process are reformatted and incorporated into DOE Methods, then,
where appropriate, submitted to EPA for inclusion in their methods documents (i.e., SW-846).
DOE Methods has been structured to complement EPA’s Test Methods for Evaluating Solid Waste
(SW-846), with emphasis on guidance for the sampling and analysis of radioactive and radioactive-hazardous
waste and environmental samples. DOE Methods contains 11 chapters and 2 appendices. Chapter and
appendix content are summarized in Figure 6. Chapters 1-6 provide guidance for the design of an EM
sampling/analytical program. Chapters 7-11 contain sampling and analytical methods that can be used in
EM program conduct. Each method is assigned a number according to the scheme shown in Figure 7.
Performance vs. Prescriptive Methods. DOE Methods will encourage a performance-based
approach to methods application. Performance-based methods specify acceptable performance criteria for an
analysis and allow the analyst flexibility in achieving that performance level. Analysts may use a
recommended method or modify it to achieve a desired level of performance (i.e., any method can be used as
long as its performance meets the data requirements of the project). In contrast, prescriptive methods
specify in detail all the steps that must be followed, including quality control. Such methods are useful
when the choices for analysis need to be limited, as is normally the case when contracting for analytical
services. Most of EPA’s methods, especially those used in the CLP, have been prescriptive.
The methods contained in DOE Methods will resemble prescriptive methods; however, they also will
contain performance criteria. All relevant validation and performance information, to the extent that it is
known, will be included for each method. At a minimum, a method developed by consolidating procedures
will contain single-laboratory performance information. If the methods are not used as written, it will be
necessary to demonstrate that data obtained from a subsequent adaptation will produce data comparable to
that obtained when using recommended or other documented or approved methods. Such a practice increases
the quality of data inte itomparability across the DOE complex. Adoption of a performance-based approach
will allow flexibility in how recommended methods are applied by DOE and DOE contractor labs to meet
DOE needs.
DOE Methods contains guidance and sampling and analysis methods to address DOE’s characterization
needs for EM programs. It fills a gap currently not addressed by EPA guidance, thus aiding DOE in
implementing cost-effective strategies for the monitoring, cleanup, and management of wastes and
environmental contamination unique to the DOE complex. Supporting DOE Methods is a database
configured so that any user from across the DOE complex can access it and review a particular method, or
order a hard copy of a method. The DMCP provides the vehicle for DOE and DOE contractors to submit
procedures to be considered for inclusion in DOE Methods. DOE Methods is expected to become a standard
reference for guidance for the conduct of contaminant characterization at DOE sites.
I EPA Test Methods for Evaluating Solid Waste Physical/Chemical Methods (SW-846), 3rd
Edition, 9/86
2 Pacific Northwest Laboratory Waste Management and Environmental Compliance, PNL-MA-8
3 Pacific Northwest Laboratory Radiation Protection, PNL-MA-6
4 Pacific Northwest Laboratory Health and Safety Management, PNL-MA-43
S Measurement Quality Assurance for Radioassay ANSI Standard N42.2, Laboratories, 1/91.

a — — —
DOE Methods
Compendium Program

BlO. & ENV.

Hardcopy mailing
1. Development
2. Procedure Entry
DOE Methods
Chapter 0ev.!
Methods Entry
Methods Review,
Validation and
Approval Process

Number of Procedures
0 100 200 300 400 500

Number of Procedures
t-JiiUKL ‘lb
0 100 200 300 400

Number of Procedures
Sample Prep
• Counting
0 100 200 300

?ñ >
i ,
o Americium
• Plutonium
• Alpha
• Uranium
Number of Procedures
0 100

Increasing levels
3 of validation
External QC
Data Adequate
Internal QC
Data Adequate

DOE Methods for Evaluating
Environmental and Waste Management Samples
Scope and Application
Program Design
Quality Control Ciiteria
Data Reporting
QC Samples Submitted by Outside Organizations
Chapter Three SAFETY
Scope and Application
Radiation Protection
Dose Equivalent Limits, Controls, Policies, and Records
Hazardous Chemical Storage and Handling
Waste Minimization
Disposal Requirements
Scope and Application
General Considerations
Quality Control
Summary of Methods
Data Reporting
Sampling Methods

Analytical Methods
Radiological Screening
Scope and Application
General Considerations
Quality Control
Summary of Methods
Data Reporting
Scope and Application
General Considerations
Quality Control
Summaiy of Methods
Sampling Considerations
Data Reporting
Method OGO15R: Major Nonhalogenated Volatile Organics in
Radioactive Aqueous Liquids Analyzed by Direct Aqueous Injection
Gas Chromatography (DAI-GC)
Method PVO3OR: Purge-and-Trap
Method PE55OR: Ultrasonic Extraction
Method PE541: Automated Soxhiet Extraction
Method 0M260R: Volatile Organic Compounds by Gas
Chromatography/Mass Spectrometry (GC/MS): Capillary Column
Scope and Application
General Considerations
Quality Control
Summary of Methods
Sampling Considerations
Data Reporting
Method PM100R: Solvent Extraction of Uranium and Thorium from
Radioactive Liquid Wastes using TOPO
Scope and Application
General Consdierations
Quality Control
Summary of Methods
Sampling Consideations
Data Reporting

Scope and Application
Summary of Methods
General Considerations
Sampling Considerations
Quality Control
Definition of Terms
Data Reporting
Method Selection Process
Method Validation Process
Method Approval Scheme

S Sampling
SAxxx Air Sampling
SWxxx Water Sampling
SSxxx Soil Sampling
SDxxx Drum Sampling
O Organics
OPxxx Organic Sample Prep
OGxxx GC Methods
OHxxx HPLC Methods
OMxxx GC-MS Methods
Olxxx GC-IR or IR Methods
OSxxx Screening
OVxxx UV-Vis Methods
OXxxx Organic Halides/Total Halides
M Inorganics
MPxxx Inorganic Sample Prep
MAxxx AAA Methods
MCxxx Ion Chromatography
Mlxxx ICP Methods
MMxxx ICP-MS Methods
MSxxx Screening
R Radionuclides
RPxxx Radionuclide Sample Prep
RGxxx Energy Analysis (GEA)
RAxxx Alpha Energy Analysis (AEA)
RLxxx Liquid Scintillation Counting (LSC)
RSxxx Screening--Alpha, Beta, Gamma
X Miscellaneous Analytes and Tests
XBxxx Biological (COD, etc.)
XOxxx Nonspecific Organics, (TOC, Phenolics)
XPxxx Properties (Corrosivity, Ignitability)
XHxxx pH

Robert 0. Harrison , ImmunoSystems, Inc., Scarborough, Maine
Robert E. Carison, Ecochem Research, Inc., Chaska, Minnesota
A competitive inhibition nzyme ImmunoAssay (EIA) has been developed for the
determination of olychlorinated iphenyls (PCB’s). The test is capable of analyzing for
PCB’s in the field in 15 minutes from a prepared sample, using no specialized equipment.
Validation data have been generated for soils using a rapid extraction procedure which can
be done in the field with an inexpensive extraction device (drying of the soil is not required
before analysis). Applications currently under development for other matrices include
water, transformer oil, sediment, and surface wipes. The test specificity is restricted to
PCB’s, with high sensitivity for Aroclor’s 1016, 1242, 1248, 1254, and 1260. Matrix and
solvent interferences are minimal. The sensitivity and flexibility of the test allow analysis
of PCB’s with a wide variety of sample preparation methods. Sensitivity in the matrix of
interest depends on the matrix and sample preparation method. The present rapid extraction
and EIA, commercially available in kit form, is suitable for PCB screening of soils in many
field and laboratory situations.
Reagent Development
The assay development process followed the general procedure described previously by
Harrison et al. (1988) and Jung et al. (1989). The development of the HA for PCB’s
followed these steps: 1) PCB derivatives were synthesized for conjugation to proteins; 2)
one of these PCB derivatives was conjugated to a carrier protein and the resulting conjugate
was used to immunize animals, which then produced antibodies recognizing both the PCB
derivative and PCB’s; 3) a PCB derivative was conjugated to horseradish peroxidase
(HRP) to make a conjugate which can be captured by anti-PCB antibodies; 4) the PCB-
HRP conjugate was used to screen and select antibodies; 5) the selected system was
optimized for sensitivity and matrix tolerance and characterized for specificity; 6) sample
preparation methods were developed for specific sample types; 7) these methods were
validated using field samples.
PCB E1A Procedure
Figure 1 schematically illustrates the procedure used for the analysis of samples containing
PCB’s. In summary: 1) rabbit antibodies which recognize the PCB structure are
immobilized on the walls of plastic test tubes; 2) samples or calibrators are added to tubes
with Assay Diluent, allowing PCB’s to be captured by the immobilized antibodies
(incubation 1). PCB’s are specifically retained on the solid phase by the anti-PCB antibody
when the rest of the sample is washed away; 3) PC B-enzyme conjugate is added to tubes
and is boupd by the anti-PCB antibody in the same manner as in step 2 (incubation 2). The
unbound conjugate is washed away and the amount retained by the immobilized antibody is

inversely proportional to the amount of PCB bound in step 2; 4) enzyme substrate and
chromogen are added to the tubes or wells for color development by the bound enzyme
(incu/xition 3). The intensity of color is proportional to the amount of captured enzyme and
is inversely proportional to the amount of PCB bound in step 2. Therefore, more color
means less PCB.
Field Soil Extraction
Soil samples were extracted for analysis by the following procedure using a simple
extraction kit available with the HA kit. Soil extraction consists of the following steps: 1)
Weigh soil on the available pocket balance and place in extraction container. 2) Extract soil
by adding an equal amount (wlv; e.g. 5 g soil + 5 mL) of methanol, and shaking
vigorously for two minutes. Filter extract and collect using the filtration device contained
in the soil extraction kit. 3) Analyze extract as described in step 2 of the EIA Procedure
above, using a 1:100 dilution of extract
Other Sample Types
Aroclor 1248 spiked into tap water was analyzed by extracting 500 mL water samples with
a glass barrel solid phase extraction device containing C18 silica. Columns were
conditioned with hexane followed by isopropanol and reagent water, then the sample was
applied to the solid phase extraction device using Teflon tubing. After air drying, the
extracted PCB was eluted with isopropanol, and the eluate was used directly in the assay
by dilution into Assay Diluent.
The HA kit has also been used for surface wipe tests. Standard surface sampling methods
were employed, except for substitution of isopropanol for hexane for wiping. Additional
isopropanol was used to extract the gauze wipe and this extract was used directly in the
assay by dilution into Assay Diluent. Sediment analysis is currently under investigation
using methods similar to the soil procedure described above.
Matrix and Solvent Tolerance
The primary requirement for sample preparation for the PCB BA is that the solvent used to
introduce the sample into the test must be miscible with water. The test offers excellent
resistance to the effects of blank soil extracts made with both DMSO and methanol. The
test tolerates several solvents, including methanol, isopropanol, and DMSO, up to 20%,
with no loss of sensitivity. Thus, adequate flexibility exists for the use of a variety of
solvents and protocols in the development of new applications for different sample
Test Specificity and Sensitivity
The crossreactivity of the test for five commonly detected Aroclors was examined.
Standard solutions of 200 ppm in methanol (Supelco) were used to make serial dilutions in
methanol, which were tested in the EIA as described above. Table 1 shows the broad
specificity of the HA for the most common Aroclors.

Table 1. Crossreactivity of the HA test for five Aroclors commonly found in soil. Values
given are the stock concentrations required (when diluted 1:100 in sample diluent) to give
50% reduction in OD compared to the negative control (50%B 0 ).
Aroclor 1016 1242 1248 1254 1260
ppmofAroclorstock 3.4 4.1 3.4 2.0 4.6
giving 50% B 0
Based on the data of Table 1, using the specified 1:100 extract dilution, a soil extract
concentration of 5 ppm of any of the Aroclors would give a test response of greater than
50% inhibition. Combining these crossreactivity data (least crossreactive Aroclor within
twofold of 1248) and the standard curve of Figure 2 (detection of 0.2 ppm of 1248), yields
a test sensitivity for any of the five Aroclors of 0.5 ppm or better. Specificity was also
tested for selected congeners in the same manner as for the Aroclor’s of Table 1. The
congeners most strongly recognized were 2,2’,5 ,5’ tetrachiorobiphenyl, 2,3’,4,4’,5
pentachiorobiphenyl, and 2,2’,4,4’,5,5’, hexachiorobiphenyl, showing that the Aroclor
specificity reflects the congener specificity. Biphenyl and several chlorinated single ring
compounds were also tested for crossreactivity in the EIA. All of the following
compounds demonstrated less than 0.5% crossreactivity compared to Aroclor 1248: 1,2-
dichlorobenzene, 1 ,3-dichlorobenzene, 1 ,4-dichlorobenzene, 1 ,2,4-trichlorobenzene, 3,3’-
dichlorobenzidine, biphenyl, 2,4-dichiorophenol, 2,5 -dichiorophenol, 2,4,5 -
trichiorophenol, 2,4,6-trichlorophenol, and pentachiorophenol. These results mean that
more than 200 ppm of any of these compounds would be required to give the same test
response as 1 ppm of Aroclor 1248. Also, more than 10,000 ppm of any of these
compounds would be required to give the same test response as 50 ppm of Aroclor 1248.
Method Repeatability
Calibrator solutions of Aroclor 1248 in methanol were diluted 1:100 into Assay Diluent for
HA analysis as shown in Figure 1. The test was performed as described above over two
weeks using three reagent lots. Figure 2 shows the means and standard deviations at four
calibrator concentrations. Similar results were obtained over 5 months with calibrators in
DMSO, using several calibrator preparations from two different sources. For the DMSO
calibrators, pooled intraassay precision at 0, 7, and 45 ppm was 7.5% for 12 assays by
two operators with two reagent lots. Interassay precision of determination of a control
sample of approximately 20 ppm Aroclor 1248 in DMSO was 20% for 46 assays by two
operators with two reagent lots. In another precision study, four field soil samples were
extensively homogenized and analyzed using a methanol soil extraction and methanol
calibrators. Extraction and analysis were performed by three analysts at three sites over 9
days using 3 reagent lots. Overall coefficients of variation averaged 33%, while interassay
precision within lot and within site was as low as 16%.
Sensitivity for PCB’s in Soils
Interference from soil extracts was tested using 12 clean PCB-free soils. Extracts using
either methanol or DMSO showed minimal interference, equivalent to less than 1 ppm for
DMSO and less than 0.5 ppm for methanol. As described in the previous section, all of the
5 Aroclors of Table I can be detected in DMSO or methanol at 0.5 ppm, using the 1:100

dilution into Assay Diluent described above. Thus, 0.5 ppm of any of the 5 Aroclors of
Table 1 would be detected in a DMSO or methanol soil extract. The actual minimum
detection level (MDL) for PCB in soil depends more on the variability of the soil and the
variability and efficiency of the extraction than on the sensitivity of the EIA to PCB or soil
interferences in the EJA.
Recovery of PCB Spikes from Soils
Aroclor 1248 standards in hexane were spiked onto 2 g aliquots of six different PCI3-free
soils at 20 and 100 tg/g. Extractions were performed one day to one week later according
to the protocol described above. The mean recovery was 84% at 20 ppm and 95% at 100
ppm. Similar results were observed for 2 selected soils spiked at 50 ppm and analyzed
after 5-8 months storage at 22° C.
Correlation of GC and EIA Results for Soil analysis
Comparisons were made between the PCB FAA with field extraction and GC-ECD, using
both methods to analyze split field samples from a wide variety of sites. One study
involving three outside groups gave the data shown in Figure 3 over the range from 0.1
ppm to 10,000 ppm. Similar data have been obtained for similar soils using methanol
extraction. One example data set is shown in Figure 4 for samples containing more than
0.5 ppm Aroclor 1260 by GC.
Other Sample Types
Using the water method described above, the sensitivity of the test for Aroclor 1248 in
clean water was estimated at 25 nglL, based on analysis of solid phase extraction device
blanks. Recoveries of Aroclor 1248 at this level were nearly quantitative. The utility of
this method for environmental water samples containing high particle loads is still being
evaluated. Evaluation of surface wipe tests and sediment methods are continuing.
The test is capable of analyzing for PCB’s in soil in the field in approximately 20 minutes,
using no specialized equipment. The test specificity is restricted to PCB’s, with Aroclor’s
1016, 1242, 1248, 1254, and 1260 recognized best. Congener specificity of the test
reflects the Aroclor specificity. The sensitivity of the test for PCB in soil is better than 3
ppm, but varies with different soil type and Aroclor. Further work in this area will include
improved soil analysis, oil analysis, surface wipe tests, sediment analysis, biological
sample analysis, and field testing and validation for the above applications.
The initial phase of development of this PCB immunoassay was partially supported by the
US EPA through a sub-contract to ECOCHEM from Mid-Pacific Environmental
Laboratories, Inc.

1. Harrison, R.O.; Gee, SJ.; Hammock, B.D. “Immunochemical Methods of Pesticide
Residue Analysis” Chapter 24 in Biotechnology in Crop Protection: ACS Symposium
Series Vol. 379 , Hedin, P.A., Menn, JJ., Hollingworth, R.M., eds., American
Chemical Society: Washington, DC, 1988.
2. Jung, F.; Gee, SJ.;, Harrison, R.O.; Goodrow, M.H.; Karu, A.E.; Braun, A.L.; Li,
Q.X.; Hammock, B.D. “Use of Immunochemical Techniques for the Analysis of
Pesticides”. Pesticide Science 1989, 26, 303-3 17.

Figure 1. Schematic illustration of the Enzyme Immunoassay for PCB t s. All three
incubations are 5 minutes. Both washes are done with tap water. In summary: 1) rabbit
antibodies which recognize the PCB structure are immobilized on the walls of plastic test
tubes; 2) samples or calibrators are added to tubes with Assay Diluent, allowing PCB’s to
be captured by the immobilized antibodies (incubation 1). PCB’s are specifically retained
by the antibodies on the solid phase when the rest of the sample is washed away (wash 1);
3) PCB-enzyme conjugate is added to tubes and bound in the same manner as in step 2
(incubation 2). The unbound conjugate is washed away and the amount retained by the
immobilized antibody is inversely proportional to the amount of PCB bound in step 2
(wash 2); 4) enzyme substrate and chromogen are added to the tubes or wells for color
development by the bound enzyme (incubation 3). The intensity of color is proportional to
the amount of captured enzyme and is inversely proportional to the amount of PCB bound
in step 2. Therefore, more color means less PCB.
Unbound PCB-HRP Is washed away, leaving an
amount of enzyme thversely proportional to the
PCB concentration In Incubation 1.
Colorless substrate and chromogen are
converted to blue color In proportion to
amount of bound enzyme. Less color means
more PCB . Stop solution ln tivates the HRP,
th iges color to yellow. and stabilizes color.

C r
S • Substrate
C Chromogen

Figure 2. Mean Standard Curve of Aroclor 1248. The x axis values are concentrations of
the standards which were diluted 1:100 in incubation 1 as shown in Figure 1. These values
equate directly to the concentrations in Table 1. These results are from a precision study
which used 3 reagent lots to perform 27 assays over 9 days. Values for mean and SD are
expressed as %B 0 ((OD of standard/OD of negative control)x 100), which is the color
intensity of the standard as a percent of the color intensity of the negative control.
Mean Standard Curve with 3 Reagent Lots
%Bo 50
•_n =27at
A = 0.999
y = -261og(x) + 69
each concentration
1 10
ppm Aroclor 1248

Figure 3. Correlation of GC and FAA Results for Soil Analysis. Results for analyses at
three sites using multiple reagent lots over more than 3 months. Soils were analyzed as
described in the materials and methods section except for the use of DMSO for extraction
and standards, rather than methanol. Values over 50 ppm were determined by analysis of
diluted extracts. Twelve additional samples were analyzed but not plotted because one or
both of their results was expressed as a range only, e.g. <3 ppm or >50 ppm. Three of
these samples were greater than 50 ppm by both methods, while nine were less than 3 ppm
by both methods.
100 1000 10000 100000
ppm by GC
0.1 1 10

Figure 4. Correlation of GC and EIA Results for Soil Analysis. Twelve soil samples
giving results between 0.5 and 100 ppm of Aroclor 1260 by GC were analyzed by EIA as
described in the materials and methods section, using methanol for extraction and
standards. These results indicate similar performance to Figure 3.
ppm 10
1 1 I I liii
ppm by GC
10 100
y = 1.06x -0.62

Timothy S. Lawruk, Charles S. Hottenstein, David P. Herzog,
Fernando M. Rubio , Ohmicron Corporation, 375 Pheasant Run,
Newtown, PA 18940; James R. Fleeker, North Dakota State
University, Biochemistry Dept., P.O. Box 5516, Fargo, ND
58105; J. Christopher Hall, University of Guelph, Dept. of
Environmental Biology, Ontario, Canada N1G 2W1.
The intense pressure for increased pesticide residue
testing in food, water and soil, together with the expense
and delays associated with currently available testing
methods, has focused attention on immunocheinical methods
which are sensitive, reliable, simple, cost—effective and at
the same time provide more rapid results. Some of these
desired features can be met with immunoassays formats
currently available.
The performance characteristics of a magnetic particle—
based solid-phase enzyme-linked immunosorbent assay (ELISA)
requiring no sample preparation for the quantification of
2,4-D and related chiorophenoxy acid herbicides in
groundwater samples are discussed.
Pesticide testing in water, food and soil has increased
dramatically over the past several years due to concerns over
the potential contamination of wells and streams from spills,
spraying and pesticide run-off. 2,4-D, a member of the
chiorophenoxy acid class of compounds (CPHs), is a selective
systematic herbicide commonly used for the post—emergence
control of annual and perennial broadleaved weeds in cereals,
maize, sorghum, grasslands, established turf grass, seed
crops, orchards (specifically pome fruit and stone fruit),
cranberries, asparagus, sugarcane, rice, forestry, and non—
crop land including areas adjacent to water allowing the
control of broad leaved aquatic weeds. 2,4—D is classified
by the EPA as a category III contaminant with a maximum
contaminant level of 70 ppb in water (USEPA, 1991). The
Practical Quantiation Level for 2,4—D is 5 ppb (USEPA, 1991).
2,4—D has been shown to damage liver, kidneys and the nervous
system of lab animals when exposed to large amounts. The
National Cancer Institute has conducted studies showing that
dogs in contact with 2,4-D are twice as likely to develop
non-Hodgkins lymphoma and suggests that 2,4-D may be a human
health hazard.

The principles of enzyme linked immunosorbent assay
(ELISA) have been described (Hammock and Mumma, 1980) and
applied to the detection of 2,4—D in water (Fleeker, 1987;
Hall et al, 1989). Magnetic particle-based ELISA’s have
previously been described and applied to the detection of
pesticide residues (Rubio et al, 1991; Itak et al, 1992;
Lawruk et al, 1992). These ELISA’ s eliminate the
imprecision problems of coated plates and tubes (Harrison et
al, 1989) through the covalent coupling of antibody to the
magnetic solid-phase. The uniform dispersion of particles
throughout the reaction mixture allows for rapid reaction
kinetics and precise addition of antibody.
Amine terminated superparamagnetic particles of
approximately 1 urn diameter were obtained from Advanced
Magnetics, Inc. (Cambridge, MA). Glutaraldehyde (Sigma
Chemical, St. Louis, MO). Rabbit anti-2, 4-D serum (Ohrnicron,
Newtown, PA). 2,4-D-HRP conjugate (available from Ohmicron,
Newtown, PA). Hydrogen peroxide and TMB (Kirkegaard & Perry
Labs, Gaithersburg, MD). 2,4-D and related compounds as well
as non—related cross—reactants (Riedel—de—Haen, Hanover,
The anti-2,4-D coupled magnetic particles were prepared by
glutaraldehyde activation (Weston and Avrameas, 1971). The
unbound glutaraldehyde was removed from the particles by
magnetic separation and washing four times with 2-(N—
inorpholino)ethane sulfonic acid (MES) buffer. The 2,4—D
antiserum was incubated overnight with agitation at room
temperature with the activated particles. The unreacted
glutaraldehyde was quenched with glycine buffer and the
covalently coupled anti—2,4—D particles were washed and
diluted with a Tris-saline/BSA preserved buffer.
Water samples (250 uL) and horseradish peroxidase (HRP)
labeled 2,4-D (250 uL) are incubated for 30 minutes with the
antibody coupled solid-phase (500 uL), (step 1). A magnet-
ic field is applied to the magnetic solid-phase to wash and
remove unbound 2,4-D—HRP and eliminate any potential
interfering substances, (step 2). The enzyme substrate
(hydrogen peroxide) and chroinogen (3,3’,5,5’- tetramethyl-
benzidine [ TMB)) are then added and incubated for 20 minutes,
(step 3). The reaction is stopped with the addition of aç d
and the final colored product is analyzed using the RPA-I
Photometric Analyzer by determining the absorbance at 450 nm.
The observed sample results were compared to a linear
regression line using a log—linear standard curve prepared
from calibrators containing 0, 1.0, 10.0, and 50.0 ppb of
2,4-D. If the assay is performed in the field. 1 .Jon-site), a
battery powered photometer such as the RPA-III is used.

Figure 1 illustrates the mean standard curve for the 2,4-D
calibrators, collected over 55 runs, error bars represent 1
SD. The displacement at the 1.0 ppb level is significant
(87% B/Bo, where B/Bo is the absorbance at 450 nm observed
for a sample or standard divided by the absorbance at the
zero standard). The assay sensitivity based on 90% B/Bo
(Midgley et al, 1969) is 0.7 ppb.
A precision study in which surface and groundwater samples
were spiked with 2,4—D at 3 concentrations, and each assayed
5 times in singlicate per assay on five different days is
shown in Table 1. Coefficients of variation (%CV) within and
between day (Bookbinder and Panosian, 1986) were less than
10% and 12% respectively.
Table 1
Poo l# 1 2 -
Replicates 5 5 5
Days 5 5 5
N 25 25 25
Mean ppb 4.16 19.55 36.06
% CV Intra 9.5 8.7 7.7
% CV Inter 11.5 11.]. 11.3
Correlation of 33 groundwater water samples with values
obtained by the present ELISA method (y) and an established
gas chromatography/mass spectronomy (x) method is illustrated
in Figure 2. The regression analysis yields a correlation of
0.963 and a slope of 0.928 between methods.
Table 2 summarizes the accuracy of the 2,4-D ELISA. Added
amounts of 2,4—D were recovered correctly in all cases with
an average assay recovery of 103%.
Table 2
2,4-D Mean SD %
added (ppb) (ppb) (Dpb) Recov rv
5.0 4.64 0.44 93
15.0 16.81 1.47 112
30.0 32.96 3.32 110
40.0 39.36 3.21 98
Average 103

Values obtained from 3 groundwater samples diluted in the
Diluent/Zero showed agreement between measured and expected
values (Table 3).
Table 3
Observed Expected
Sample 2,4-D ppb 2 ,4-D ipb Linearity
Sample 1 41.82
1:2 22.42 20.91 107
1:4 10.76 10.46 103
1:8 4.48 5.22 86
Sample 2 41.89
1:2 23.57 20.95 112
1:4 10.76 10.47 106
1:8 5.25 5.24 100
Sample 3 30.43
1:2 17.23 15.22 113
1:4 7.96 7.61 104
1:8 4.21 3.80 111
Average 105
Table 4, summarizes the cross—reactivity data using a
variety of chlorophenoxy acid analogues. The percent cross—
reactivity was determined as the amount of analogue required
to achieve 90% B/Bo (Least Detectable Dose). Many non—
structurally related agricultural compounds demonstrated no
reactivity at concentrations up to 10,000 ppb.
Table 4
90% B/Bo 50% B/Bo % Cross
Compound LDD (pDb) LDD reactivity
2,4—D 0.50 15.0 100
2,4-D Propylene
Glycol Ester 0.05 0.79 1962
2,4—D Ethyl Ester 0.05 0.82 1885
2,4-D Isopropyl
Ester 0.07 1.44 1076
2,4—D Methyl Ester 0.12 1.64 945
2,4—D Butyl Ester 0.19 2.40 646
2,4-D Sec-Butyl
Ester 0.13 2.20 705
2,4-D Butoxyethyl
Ester 0.13 3.10 500
2,4,5—T Methyl

Ester 0.98 18.1 106
2,4-D Butoxy-
propylene Ester 1.21 31.1 50
2,4—D Isooctyl
Ester 2.08 30.0 52
2,4,5—T 2.98 190 8.2
2,4—DB 3.95 139 11.2
NCPA 7.8 159 10.0
Silvex Methyl
Ester 12.4 1000 1.9
MCPB 56.8 1470 1.1
4 —Chiorophenoxy-
acetic acid 61.1 1220 1.3
Dichlorprop 117 7500 0.2
Silvex (2,4,5—TP) 167 2060 0.8
Dichiorophenol 217 3570 0.4
Triclopyr 830 NR <0.].
MCPP 1160 NR <0.1
Pentachiorophenol NR NR <0.1
Picloram NR NR <0.1
Table 5 summarizes that no interferences are present up to
the tested levels of various common water components. The
concentrations of the compounds chosen are those that would
most likely exceed levels found in groundwater samples
(American Public Health Association, 1989).
Table 5
Compound Max. Conc. tested Interfered
Nitrate 250 ppm No
Copper 250 ppm No
Nickel 100 ppm No
Thiosulfate 250 ppm No
Sulfite 250 ppm No
Sulfide 250 ppm No
Sulfate 10,000 ppm No
Iron 250 ppm No
Magnesium 250 ppm No
Calcium 250 ppm No
NaCl 1.0 M No
Humic acid 50 ppm No
Silicates 1,000 ppm No
This ELISA demonstrates both the feasibility of using
magnetic particles as a solid—phase in an immunoassay for the
detection of pesticide residues, and its performance
characteristics in the quantitation of 2,4-D in groundwater
samples. The assay compares favorably to GC/MS
determinations and exhibits excellent precision and accuracy

which guarantees consistent monitoring of environmental
samples. The assay sensitivity of 0.7 ppb (90% B/Bo) in
water exceeds the EPA Maximum Contaminent Level of 70 ppb and
the Practical Quantitation Level of 5 ppb. The antibody
employed allows for the detection of 2,4-D and related
chiorophenoxy acids in the presence of other pesticides and
commonly found groundwater components. This ELISA is ideally
suited for the adaptation to on-site monitoring of 2,4—D in
water, providing results in less than 60 minutes.

American Public Health Association (1989) Standard Methods
for Examination of Water and Wastewater. American Public
Health Association, Washington DC.
Bookbinder, M.J., Panosian, K.J., (1986) Correct and
incorrect estimation of within—day and between—day variation,
Clin. Chem., 32:1734—1737.
Fleeker, J. (1987) Two enzyme immunoassays to screen for 2,4—
Dichiorophenoxyacetic acid in water, J. Assoc. Anal. Chem .,,
Hall, J.C., Deschamps, R.J.A., Krieg, K.K. (1989)
Iminunoassays for the detection of 2,4—D and piclorain in river
water and urine, J. Agric. Food Chem., 37:981-984.
Hammock, B.D., Mununa, R.O. (1980) Potential of immunochemical
technology for pesticide analysis. In Pesticide Identif i-
cation at the Residue Level, ACS Symposium Series, Vol. 136
(Harvey, I. & Zweig, G., Eds.) American Chemical Society,
Washington, DC, pp. 321-352.
Harrison, R.O., Braun, A.L., Gee, S.J., O’Brien, D.J.,
Hammock, B.D. (1989) Evaluation of an enzyme-linked
immunosorbent a say (ELISA) for the direct analysis of
molinate (Odram ) in rice field water, Food & Agricultural
Immunology, 1:37—51.
Itak, J.A., Selisker, N.Y., Herzog, D.P. (1992) Development
and evaluation of a magnetic particle based enzyme
immunoassay for aldicarb, aldicarb sulfone and aldicarb
sulfoxide, Chemosphere, 24:11-21.
Lawruk, T.S., Hottenstein, C.S., Herzog, D.P., Rubio, F.M.
(1992) Quantification of alachlor in water by a novel
magnetic particle-based ELISA, Bull. Enviro. Contam. Toxicol.
Vol. 48:643—650.
Nidgley, A.R., Niswender, G.D., Rebar, R.W. (1969) Principles
for the assessment of reliability of radioimniunoassay methods
(precision, accuracy, sensitivity, specificity), Acta
Endocrinologica, 63:163-179.
Rubio, F.M., Itak, J.A., Scutellaro, A.M.., Selisker, M.Y.,
Herzog, D.P. (1991) Performance characteristics of a novel
magnetic particle—based enzyme—linked inununosorbent assay for
the quantitative analysis of atrazine and related triazines
in water samples, Food & Agricultural Immunology, 3:113—125.
USEPA, National Primary Drinking Water Regulations; Final
Rule, Federal Register, 40 CFR parts 141—143, Vol. 56, No.
20, January 30, 1991.

Weston, P.D., Avrameas, S. (1971) Proteins coupled to
polyacrylamide beads using glutaraldehyde, Biochein. Biophys.
Res. Commun., 45:1574—1580.

Magnetic Particle Immunoassay
t 1
c( *
1. Immunological Reaction
3. Color Development
Magnetic Particle with
Antibody Attached
Pesticide Conjugated
with Enzyme
Ch romogen/Su bstrate
Colored Product
2. Separation

2,4—D (ppb)
Figure 1. Dose response curve for 2,4—D.
represents the mean of 55 determinations.
indicate +1- 1 SD about the mean.
Each point
Vertical bars
1 10

C ,,
CC/MS (ppb)
Figure 2. Correlation between 2,4—D concentrations as
determined by ELISA and GC methods. n = 33, r = 0.970, y =
0.938x + 2.21.
0 100 200 300 400


Richard D. Beaty , Ph.D., Director Product Development,
and Mary R. Schoen, Software Engineer, Telecation Inc.,
19423 N. Turkey Creek, Suite C, Morrison, Colorado,
80465. (303) 697—8080
Quality control is a calculation and data management
intensive operation of an analytical laboratory.
Analytical quality control involves some procedures which
are standard, accepted practice, and others which may be
specific to an individual laboratory addressing a special
application. In addition to their own in-house
procedures, a given laboratory may have to comply with a
number of QC protocols simultaneously, including those
defined for “Good Laboratory Practice” (GLP) and special
government regulatory programs with their own QC
An important contribution of a LIMS system is to increase
the overall efficiency and integrity of the laboratory’s
QC program, by automating the data management and
statistical functions which accompany quality control.
The procedures used for analytical quality control can
generally be grouped into two categories: 1) those
involving treatment of actual analytical unknowns and
observing the response of the treated samples; and 2) the
inclusion of additional samples within an analytical run
for purely quality control purposes. Both types involve
the generation of QC data and appropriate statistical
calculations, which provide a measure of the “goodness”
of the data as determined by the QC calculation.
Typically, sample results will be flagged according to
rules defined for the QC measurements which accompany the
sample analysis.
One of the most frequently used and statistically
cumbersome QC procedures is the maintenance of x—bar or
“Shewhart” charts. A Shewhart chart involves mean and
standard deviation calculations on historical data to
determine “warning” and “control” limits, within which
subsequent determinations of the control sample are
Informix & SmartWare are registered trademarks of
Informix Software, Inc.

expected to agree. This routine task involves not only
the statistical calculations, but also the determination
of a “rolling average” of the twenty most recent
determinations. This paper discusses techniques for
dynamically updating the LIMS data to be included in the
most recent Shewhart chart and automatically flagging
data which exceeds warning and control limits.
The total QC process involves assignment of the various
QC procedures to analytical runs, calculation of QC
statistics, flagging data which fails QC protocols, and
maintaining historical QC trends. The ultimate goal of
QC management with a LIMS is to totally automate this
process. This paper will define an approach which not
only achieves this goal for predefined, standardized
protocols, but also allows the flexibility to deal with
emerging techniques and unique QC applications.
Quality control is a calculation and data management
intensive operation of an analytical laboratory.
Analytical quality control involves some procedures which
are standard, accepted practice, and others which may be
specific to an individual laboratory addressing a special
application. In addition to their own in-house
procedures, a given laboratory may have to comply with a
number of QC protocols simultaneously, including those
defined for “Good Laboratory Practice” (GLP) and special
government regulatory programs with their own QC
An important contribution of a LIMS system is to increase
the overall efficiency and integrity of its QC program by
automating the data management and statistical functions
which accompany quality control. The challenge a LIMS
vendor faces, when attempting to automate these
functions, is that not all laboratories conduct the same
types of QC. Even for so—called “standard” techniques,
different laboratories use variations in their
approaches. In general, achieving the greatest amount of
automation requires defining the functions to be
automated, and then programming those functions.
However, a commercial LIMS to be used by many
laboratories must leave functionality open, in order to
accommodate the varying QC procedures from laboratory to
laboratory. This fact precludes “hard coding” all of the
QC functions into the system. This paper will discuss
one technique, which optimizes automation of “standard”
QC procedures, while leaving the system open to user
modification for specific procedures.

Standard practice in LIMS design today calls for use of a
commercial data base controlled by a fourth generation
language. TELECATION has traditionally used Informix’s
“SmartWare” (R) system, which offers not only a powerful
data base and fourth generation language, but also
provides built—in integration and seamless transfer of
data between the data base module and the spreadsheet and
word processor modules. The same fourth generation
programming language which controls the data base is used
to generate turnkey applications involving other modules.
The spreadsheet/graphing module is especially powerful
for QC applications.
For this investigation, TELECATION used Informix’s
“SmartWare II” system, which offers the same degree of
integration between data base and spreadsheet, but with
dramatically expanded data base and programming
capabilities. The LIMS functions were developed and
tested on a local area network consisting of Novell 386
operating in an Ethernet environment. The file server
consisted of an 80386 processor, with 4MB of RAM and a
300MB hard drive. The network workstations consisted of
a variety of 80286 and 80386 machines, each with 4MB of
The procedures used for analytical quality control can
generally be grouped into two categories: 1) those
involving treatment of actual analytical unknowns and
observing the response of the treated samples; and 2) the
inclusion of additional samples within an analytical run
for purely quality control purposes. Both types involve
the generation of QC data and appropriate statistical
calculations which provide a measure of the “goodness” of
the data, as determined by the QC calculation.
Typically, sample results will be flagged according to
rules defined for the QC measurements, which accompany
the sample analysis.
Standard techniques involving the treatment of unknown
samples include running sample duplicates (for
determining the percent difference between the replicate
runs), sample spikes (for determining percent recovery of
the spiked analyte), and matrix spike duplicates (for
determining the percent difference between two similarly
spiked samples). In addition to defining the formulas
for calculation of the QC parameters characteristic of
each technique, analytical data which is outside

predefined limits of acceptability is typically flagged,
to call attention to QC failure. Therefore, the
conditions which initiate data flagging, as well as the
flags, themselves, must somehow be programmed into the QC
automation system.
Standard techniques involving the testing of incremental
QC include the analysis of control samples. These
controls may be monitored over many days for
determination of long term precision. Alternately, the
experimental results of the control sample may be
compared to the “known”, or certified value, as a measure
of the accuracy of the determination.
Depending on the parameters being measured and the
analytical techniques used for these measurements, a
variety of QC measurements selected from the above types
are included in each analytical run. For QC techniques
involving treatment of actual samples, not every sample
in the run is treated for QC measurements, as this would
unduly burden the productivity of the laboratory.
Instead, certain samples are selected as representative
of all samples in the run, and the success or failure of
the QC measured on the selected samples is extrapolated
to indicate the validity of analytical results for all
similar samples in the run. The significance of this
fact to the automation of quality control, is that each
sample measured in a single analytical run must somehow
be linked to all other samples measured in the run, so
that the results of quality control measurements, which
validate the analytical results, can be tracked. The
relational data base of SmartWare II makes this type of
linkage possible.
With the above considerations in mind, a LIMS data base
was designed, which would not only provide all of the
standard data storage and retrieval requirements of a
LIMS, but would also optimize the mechanism for
automating analytical quality control. For storing all
of the basic information about samples and their
analytical results, three data bases were defined. These
data base files were named “Samples”, “Tests”, and
“Parameters”, signifying the basic information stored in

SANPLES.DB - information specific to entire sample
examples: sample ID
collect date/time
login date/time
storage location
TESTS.DB - information regarding test methods to be
run on each sample
examples: test ID
lab section
preparation method
analytical method
hold time
preparation date/time
analysis date/time
PARAMS.DB — detailed information on each parameter to
be determined in a test.
examples: parameter ID
check limits
detection limits
reportable significant figures
reportable decimals
FIGURE 1. Definition of sample-related data bases.
“Samples” is used to store information specific to an
entire sample to be analyzed. “Tests” stores the name of
each test to be run on the sample, along with critical
information about that test. “Parameters” stores the
name of each parameter to be determined in a test, along
with the results of that determination and other critical
parameter-specific details. Figure 1 illustrates the
basic definition of these data bases.
Each “sample” record is uniquely identified by a “lab#”
field. “Test” and “Parameter” records, which are a part
of the sample, are linked to the “Sample” record through
the “lab#” field. Additionally, each “parameter” record
can be linked to the “test”, in which it is determined,
through a “test_id” field. This relationship is
illustrated in Figure 2.

To provide for the needs of run-oriented data entry and
quality control, two more data bases are defined. These
include the “Runs” and “QC Parameters” data bases. Each
record of the “runs” data base contains a “run#” field to
uniquely identify a run, a “test_id” field to identify
the test being run in the “run#”, and other run status
fields. The “QC Parameters” data base contains fields to
store the results of standard QC measurements, and
additional fields to control the calculation of each.
The “QC Parameters” data base is structured to contain
one record for every “parameter” subjected to a quality
control measurement. Definition of the “runs” and “QC
Parameters” data bases is illustrated in Figure 3.
The “parameter” records from unknown samples being
determined in the “run” and the “QC Parameter” records
determined in the “run” are linked together by a common
“run#”, as shown in Figure 4.
[ lab#)
[ test_id] PARANS.DB
FIGURE 2. Relationship of sample-related LIMS data
base files.

RUNS.DB - list of run#’s with run status information
examples: run#
analyst name
analysis date
approval name
approval date
QCPARANS.DB — results of QC measurements for each
examples: duplicate / %difference / flag
spike / %recovery / flag
spike dup I %difference / flag
control / true value / %accuracy
warning & control flags
FIGURE 3. Definition of “Runs” and “QC Parameters”
data bases.
[ run#]
___ I ____
FIGURE 4. Relationship of sample parameters and QC
parameters measured in common analytical run.
The parameters on which QC measurements are to be made,
as well as the QC to be performed on each parameter, will
vary with the nature of the analytical procedure, or
“test”, being run. Therefore, a flexible mechanism is
required to determine both which parameter records are
added and what QC measurements are to be run for each.

Most QC procedures call for adding QC parameters
according to one of two schemes. For those procedures
involving the measurement of existing sample parameters
(such as sample duplicates, spikes, etc.), it is usually
desirable to select a representative sample or samples,
and make appropriate QC measurements on the selected
samples only , for every parameter being measured in the
run. Alternately, many organic analyses call for the
determination of “surrogates”, or parameters which are
normally absent in the original samples. Controlled
amounts of these surrogate parameters are added to all
samples , and their recovery measured. To accommodate the
different schemes for QC parameter addition, an “Add—QC”
menu provides the user with the option of assigning QC
measurements to a run, according to the requirements of
“Representative Samples” or “Surrogate Parameters”. The
differences are summarized in Figure 5. By simply making
the menu choice, the program automatically adds the
necessary parameter records to the QC parameters file.
REPRESENTATIVE-SAMPLES - for selected sample only ,
adds one parameter record to QC Parameters file
for every unique parameter in run.
SURROGATE-PARAMETERS - for every sample in run , adds
surrogate parameter records according to list
defined for test in surrogate reference file.
FIGURE 5. Menu selectable routines for adding QC
parameters to run.
For the “Representative Sample” option, a number of
possible QC measurement techniques might apply. Yet,
every technique may not be used in every case. To allow
easy assignment of the desired QC to be measured on a
selected sample, a simple “checkoff” menu of predefined
QC procedures (Figure 6) is presented to the user who
requests to “ADD QC” to a representative sample. Only
those techniques checked will then be activated for the
selected sample.

Check (x) QC to be assigned to LAB#: 92-2 0945
FIGURE 6. Checkoff menu for user selection of QC
assignments to representative sample.
The heart of the automated QC system is the definition of
the various QC measurements and the associated formulas
which control the calculation of QC variables. The
generation of QC flags, which alert the user to out of
control conditions, must also be defined. To the LIMS
vendor designing a LIMS for use across a wide variety of
application areas, it is also critical to consider the
necessity of user-modifiability, so that QC details can
be configured for individual needs.
To provide both preconfigured automation of standard QC
techniques and user flexibility for tailoring to specific
needs, the details of all QC techniques are contained in
the definition of the “QC Parameters” data base. A view
of the “QC Parameters” data base for a selected parameter
record is shown in Figure 7. In this view, the user may
enter results for sample duplicate, spike, spike
duplicate, blank, and control. Other fields in the view
are either referenced directly through relationships to
other files, or calculated from the entered values
through predefined formulas. The sources of values for
the “QC Parameters” fields are summarized in Figure 8.

DATE 5/14/1992 TEST_ID Inorganics RUN 92050041
Transferred to PINISRED
1.0000 WEIGET 1.0000 SOLIDS 100.00%
MEAN 5.0718 WARNING 0.5449
CONTROL 0.8174
“QC Parameters” data base record.
are automatically calculated from
QC results
predef med
QCLAS7 QC000001O4
LAS! 92-9002
CONILAB# 920501
86.60 W
QC 00 00 00 003B
QC000000004 3
QC0 000 0000B5
QC0000000 086
QC000 0000091
QC0 0 00 0 0 01O6
QC000000 0109
QC00 0 0 00 0115
QC0 000 000119
QC000000 0120
QC 000000 0131
04/ 30/ 1992
05/02/ 1992
05/05/ 1992
05/09/ 1992
4 • 6600
5. 2 100
4. 9500
5. 6600
5. 2200
4. 8 500
5. 1300
5. 0600
5. 7200
5. 1400
5. 0800
4 • 9700

QC Procedure Field Source
CALC RESULT direct link to PABAMS
DIFF% calculated
DUP FLAG calculated
SPIKES SPIKE user entry
SPIKE_ANT user entry
RECOV% calculated
SPIKE_FLAG calculated
SPKDUP% calculated
SPKDUP_FLAG calculated
BLANKS BLANK user entry
TRUE VALUE direct link to ref
ACCURACY% calculated file
CONTROL_FLAG calculated
LONG-TERN PREC. CONTROL HISTORY direct link to history
(X—BAR CHART) MEAN calculated
WARNING window calculated
CONTROL window calculated
FIGURE 8. Source of QC Parameters field information .
It should be noted that since all of the “intelligence”
regarding the determination of QC results is built into
the definition of the QC Parameters view, as opposed to
being hard coded in the program, the system remains open
to modification of formulas for existing QC measurements,
or even addition of completely incremental QC techniques,
which were not originally defined in the system. The
system defined above is preconfigured, and therefore
requires no user intervention to implement. However, the
system may be easily modified and expanded, either by the
vendor, or by the trained user.

The purpose of the “Shewhart,” or x—bar chart, is to
monitor long—term precision. This is done by recording
the results of a control sample over a period of days or
weeks, calculating the mean of the results over that
period of time, and determining “warning” and “control”
limits, defined as 2x and 3x the standard deviation of
results, respectively. The x—bar chart is a particular
challenge to automate fully, since it involves data, not
all from the same run, but from many past runs. To
further complicate the challenge, not all past results
are included in the determination. The x—bar data is
usually calculated only from the last 20 measurements of
the control.
In spite of the difficulty of the challenge, a system
which would automate the x-bar process and automatically
flag controls exceeding warning or control limits would
be immensely valuable to the QC manager, attempting to
make decisions on the validity of current data. This has
been achieved by linking the “QC Parameters” file to a
“Control History” file, which contains past runs of the
sample. The lower half of the “QC Parameters” record,
illustrated in Figure 7, illustrates this link.
From Figure 7, it can be seen that all of the results
from past determinations of the same parameter in the
same control sample are automatically available. In
addition, fields for the calculation of x-bar parameters
(including the mean, warning, and control limits), are
defined in a manner similar to other QC fields, discussed
above. It should be noted that no manual entry is
required , in order to exercise the automatic
determination of warning and control flags. All values
necessary for the process to occur are obtained, either
by direct link to the necessary data file, or by
automatic computation through predefiried calculated
The net result of the design concepts described above can
be seen in Figure 9. Results for sample test parameters
assigned to a run can be entered or reviewed in the
“Sample Parameters” window. By simply selecting
“QC—View” from the menu, the “QC Parameters” accompanying
the run (lower half of Figure 9) are instantly shown.
The percent difference of sample duplicates, percent
recovery of spikes, percent difference of spike
duplicates, and the value of the blank are tabulated for
each parameter measured. The percent accuracy of
laboratory controls is also shown. Finally,
automatically computed flags (“*“ in Figure 9) alert the

05/12/1992 BOLD DAYS 10 BOLD_DATE 05/22/1992
1 1 1TCA
1 1
1 100.00 24.0000 24
1 1
1 100.00 55.0000 55
1 1
1 100.00 62.0000 62
1 1
1 100.00 8.0000 lOU
1 1
1 100.00 44.0000 44
1 1
1 100.00 29.0000 29
1 1
1 100.00 4.0000 lOU
1 1
1 100.00 26.0000 26
1 1
1 100.00 52.0000 52
1 1
1 100.00 12.0000 12
1 1
1 100.00 35.0000 35
1 1
1 100.00 22.0000 22
1 1
1 100.00 59.0000 59
95.00 8.96 2.0000
110.00 10.12 0.0000
118.00 9.21 5.0000
90.00 6.54 4.0000
120.00 6.27 6.0000
93.50 6.93 0.0000 86.67C
108.00 9.67 2.0000
107.00 9.33 2.0000
91.00 7.33 0.0000
41.00* 74.13* 0.0000
156.00* 12.12 4.0000
121.00* 10.53 2.0000
112.00 4.12 0.0000
95.50 7.73 3.0000
88.00 10.12 0.0000 115.79W
185.00* 23.30* 5.0000
94.00 11.76 4.0000
74.00* 15.79 0.0000
The data entry and review screen allows
simultaneous viewing of sample and QC
parameters and QC flags, to alert reviewer to
out of control results.

data reviewer to results which failed quality tests. The
“W” and “C ” flags in the last column, immediately
identify control values which exceed “warning” and
“control” limits, respectively, as determined by the
automatic computation of x—bar information.
The purpose of analytical quality control is to provide a
quantitative tool to the data reviewer, and to assist
that person in making critical decisions regarding the
validity and acceptance of laboratory analyses. In order
for this goal to be achieved, the necessary data must be
easily and quickly accessed by the data reviewer at the
time decisions on data acceptability are being made. The
above described system meets this requirement.
Inasmuch as the QC programs of all laboratories do not
follow a single, predefined standard, it is absolutely
necessary that any commercial LIMS, addressing quality
control automation, do so in a manner which leaves the
system open for user-personalization of the details. The
techniques described in this paper provide the necessary
flexibility to meet this need.
Quality control is a time-consuming task in the operation
of an analytical laboratory. Nothing can compensate for
the time required to conduct QC analyses. However, the
overhead in managing the data, making the calculations,
and providing ready access to the results, can be
expedited through a relational data base LIMS, that has
been designed around the way a testing, laboratory works.
Quality control is one of the most tedious, detailed, and
time—consuming functions of a laboratory. If the purpose
of a LIMS is to increase laboratory efficiency,
automation of analytical quality control is a critical
part of the full-functioning LIMS system.

Gary Walters , Enseco Rocky Mountain Analytical Laboratory, 4955 Yarrow
Street, Arvada, CO 80002; Allen Verstuyft, Ph.D., Chevron Research
Corporation, 576 Standard Avenue, Richmond, CA 94802, Dianna Kocurek,
Tischler / Kocurek, 116 East Main, Round Rock, Texas 78664
ABSTRACT . Monitoring is required by petroleum refineries to support a
variety of RCRA program initiatives. An evaluation was performed to
better define the data quality objectives (DQOs) for three specific
regulations which impact the petroleum industry. Specifically, work was
performed to evaluate the monitoring requirements associated with the
Appendix IX Groundwater monitoring rule, the toxicity characteristic
(TCLP) and the land disposal ban program. Subsequent work documented the
performance of the SW-846 methods relative to the monitoring requirements
for each of these three regulatory programs relative to petroleum industry
needs. In many cases, the performance characteristics can be applied to
samples of a similar matrix to aid in the design of cost effective
programs for similar monitoring programs.
Once the DQOs were defined, laboratory work was performed to document the
performance of the analytical methods for selected target analytes
relative to four matrices of concern to the petroleum industry.
Volatiles, semivolatiles, and metals of concern to the petroleum industry
were evaluated. Analytes were selected from the Appendix IX list and the
“Skinner” list or Refinery list, which is a subset of Appendix VIII
constituents used for delisting refinery wastes, land treatment
demonstrations, and site closures related to petroleum refinery RCRA
programs. The four matrices were soil, treated waste, oily waste, and
TCLP leachates. For semivolatiles, various sample preparation techniques
(Method 3550 with 1:1 acetone:methylene chloride, a modified Method 3550
with methylene chloride only, EPA Handbook for the Analysis of Petroleum
Refinery Residuals) and clean-up procedures (Method 3611 Alumina column
cleanup and Method 3640 GPC column cleanup) were evaluated for the
treated waste and oily waste matrices to determine the relative advantages
and disadvantages of these various options. For each matrix a method
detection limit (MDL) study based on 40 CFR 136, Appendix B was performed
for all target analytes. These MDL values were evaluated by comparing the
precision of each matrix to reagent water.
Petroleum refineries must perform testing under RCRA to respond to various
EPA initiatives. In some cases, the testing requirements are not well
understood either by those who must use the data or by those who perform
the analyses. This work was funded by the American Petroleum Institute to
prepare a series of guidance documents. Currently, a document on
evaluation of analytical methods for Appendix IX analysis [ 1] has been
released and another document on the Toxicity Characteristic Leaching
Procedure is in preparation. These documents contain a summary of the
Practical Quantitation Limits (PQLs) plus analytical precision and

The focus of this work was to develop baseline performance data for other
Appendix IX analytes in matrices other than water: TCLP leachate, clean
soil, treated refinery waste, and oily waste. The primary emphasis was on
analytes of interest to the petroleum refinery industry. In addition,
some alternate extraction and cleanup techniques were investigated for
semivolatile analytes.
Method Perfori ance
The four matrices studied were: 1) an oily waste (mixed separator sludge
and slop oil emulsion), 2) a treated oily waste (the cake from a solvent
extraction process), 3) a clean loamy soil, and 4) the TCLP leaching
solution. Data for reagent water is also presented for comparison;
however, this data was generated for a different study at different
spiking levels. The studies consisted of seven replicates with compounds
spiked at concentrations relative to the nature of the matrix as defined
below. (1,4-Dioxane was spiked 20 times higher than the other volatile
analytes in the first four matrices. In the reagent water matrix,
1,4-dioxane, acetone, and 2-butanone were spiked at higher levels-see
Table 7.) The study analytes are listed in Table 1. The analytes were
selected by the API Environmental Monitoring Task Force as representative
analytes of concern to the petroleum industry.
Semivolatiles Volatiles
The spike levels for the metals varied based on the technique and the
metal and are shown in Table 2. Three metals (antimony, arsenic and
selenium) were evaluated by both Inductively Coupled Plasma - Atomic
Emission Spectroscopy (ICP) and Graphite Furnace Atomic Absorption (GFAA)
The oily and treated waste samples contained background levels of some
target organic compounds. All solid matrices contained background levels
of the metals. Accordingly, an initial analysis of each sample was
performed to determine background concentrations. These data were used to
evaluate the spiking levels. The MDL was calculated as 3.14 times the
standard deviations from the seven replicates, as described in 40 CFR 136
[ 3]. The results for each matrix were evaluated by a comparison of
precision and accuracy.
accuracy data present in SW-846 [ 2]
criteria are all based on the analysis
Data Quality Objectives (DQOs) for
However, they should be used with
quantitation and uncertainty of data for
and 40 CFR Part 136{3]. These
of water , and they are useful as
the analysis of other matrices.
caution in the estimation of
matrices other than water.
Oily Waste
Treated Waste
Clean Soil
TCLP Leachate
Reagent Water
Oily Waste
Treated Waste
Clean Soil
TCLP Leachate
Reagent Water

Metals Acenaphthene
Antimony Anthracene
Arsenic Benzenethiol
Chromium Benzo(a)anthracene
Cobalt Benzo(a)pyrene
Lead Benzo (b) fi uoranthene
Nickel Benzo(k)fluoranthene
Selenium Benzo(g,h,i)perylene
Bis(2-ethylhexyl )phthalate
Volatile Organics Chrysene
m-Cresol (3-methyl phenol)
Acetone o-Cresol (2-methyl phenol)
Benzene p-Cresol (4-methyl phenol)
2-Butanone Dibenz(a, h)anthracene
Carbon disulfide Dibenzofuran
1 ,2-Dibromoethane 7, 12-Dimethylbenz(a)anthracene
1,4-Dioxane 2,4-Dimethyiphenol
Ethyl benzene Fluoranthene
Methylene chloride Fluorene
Styrene Indene
Toluene Indeno(1 ,2 3-cd) pyrene
m-Xylene 3-Methylchloanthrene
o-Xylene 1-Methylnaphthalene
p-Xylene 2-Methylnaphthalene
Naphthal ene
Py rene
Spike Concentration, ppm
Reagent TCLP Treated Oily
ICP Metals Water Leachate Soil Waste Waste
Antimony 0.05 0.5 50 50 50
Arsenic 0.05 0.5 50 50 50
Chromium 0.05 0.2 20 0* 0*
Cobalt 0.05 0.2 20 20 20
Lead 0.05 0.5 50 50 50
Nickel 0.05 0.5 50 50 50
Selenium 0.5 2.0 200 200 200
GFAA Metals
Antimony 0.02 0.5 50 50 50
Arsenic 0.01 0.5 5 50 50
Selenium 0.02 0.2 2 2 2
* Not spiked due to high levels in sample

Evaluation Of Extraction And Cleanup Techniques For Semivolatile Organics
As a group, the analysis of semivolatile organics presents a significant
analytical challenge. The group contains the largest number of analytes.
The analytes include many analytes of concern to the petroleum industry
(Polynuclear Aromatics-PNAs and phenols). Interferences present in many
refinery samples (aliphatic hydrocarbons) can significantly impact the
method performance, and the methods allow for various sample preparation
and cleanup procedures. The method performance resulting from the use of
these various analytical options has not been thoroughly evaluated. The
work was therefore designed to evaluate some of these options. The
evaluation focused on the extraction techniques for soil, treated waste
and oily waste, plus cleanup techniques for the waste matrices.
Three approaches were evaluated for the extraction of the solid matrices.
The first approach was a sonic probe extraction of a 30 gram sample
aliquot using 1:1 methylene chloride:acetone as described in SW-846 Method
3550. The second approach was the same extraction with 100% methylene
chloride. Method 3550 states that the first approach provides a “more
rigorous extraction procedure.” However, it is well documented that the
acetone reacts with itself during the extraction process to form aldol
condensation products. The reaction products are manifested as a series
of early eluting interfering compounds. The third extraction approach was
from the “Handbook for the Analysis of Petroleum Refinery Residuals and
Waste” [ 4]. The procedure in this guidance manual is a simultaneous
extraction/acid-base partition preparation which includes the sonication
and phase separation of a sample with methylene chloride and aqueous
sodium hydroxide for “base/neutral” compounds followed by extraction of
the acidified aqueous layer for the “acid” compounds. Soxhiet extraction
using SW-846 Method 3540 was not evaluated. Comparisons of soxhiet versus
sonication extraction have been previously reported [ 5,6]. For the oily
waste samples, only the 100% methylene chloride and the refinery Handbook
extractions were evaluated. In each case, duplicate samples were analyzed
from each alternate extraction technique. The “best” technique was
selected; only the five remaining replicates from the “best” technique
were analyzed and evaluated.
SW-846 contains two cleanup procedures which are relevant to the analysis
of refinery wastes, Method 3640-Gel Permeation Cleanup and Method
3611-Alumina Column Cleanup and Separation of Petroleum Wastes. Method
3611 has been widely used to remove interferences from aliphatic
hydrocarbons in oily wastes for the determination of a select list of
semivolatile analytes. The application of this cleanup to a broader suite
of analytes (e.g., Appendix IX) has not been studied. Method 3640 was
developed in the CLP protocol to remove lipids and related materials. The
technique is in widespread use in the CLP for a limited suite of analytes,
and its use for Appendix VIII analytes has been discussed [ 7]. Its
utility for samples containing aliphatic hydrocarbons has not been
thoroughly evaluated. Furthermore, the improved reliability for using
either cleanup as opposed to no cleanup has also not been studied.

The first part of the cleanup evaluation consisted of subjecting a
standard solution containing the semivolatile analytes to both procedures.
This work eliminated bias due to sample interferences or extraction
efficiency. The second phase of the cleanup evaluation used data obtained
from the treated and oily waste matrices. The evaluation of extraction
techniques described previously and the generation of MDL data described
earlier assumes the use of Method 3611 for the treated and oily waste
matrices. The results from these analyses provide baseline overall method
performance. The improvement resulting from the GPC cleanup and the
impact of no cleanup was evaluated against this basic approach by
analyzing split extracts. For each matrix, four of the extracts were
analyzed by Method 8270 with no cleanup. Three extracts were subjected to
the GPC cleanup before analysis.
Recovery and precision performance for ICP (Method 6010) and GFAA (Methods
7041, 7060, 7740) are summarized in Tables 3 and 4. Method 3050 was used
for acid digestion of the solid samples. Average and standard deviation
are based on seven samples (eight for reagent water). The overall
standard deviation for all metals is the pooled standard deviation. As
seen in both tables, antimony was recovered poorly or not at all in the
solid matrices. This is noted in Method 7040, as there is currently not
an approved digestion procedure for the analysis of antimony in solids.
Antimony was evaluated in a (JSEPA study [ 8] of Method 3050; however,
antimony was not recommended for analysis by Method 3050.
Aside from antimony, recoveries were generally in expected ranges.
Recoveries higher than normal were seen with arsenic/ICP and GFAA in
reagent water and oily waste, nickel and selenium/ICP in treated waste,
and selenium/ICP in oily waste. Low recoveries were seen with
arsenic/GFAA, lead, and selenium/GFAA in oily waste; and arsenic/GFAA and
selenium/GFAA in clean soil. Chromium was not spiked in treated and oily
wastes because of high background levels of this metal; therefore,
recoveries could not be determined for these matrices.
There were a number of statistically significant difference in metal
recoveries between matrices. In general, a difference of 10 percentage
points between any of the recoveries shown in Tables 3 and 4 tested
Standard deviations (within-analyte precisions) for the ICP metals ranged
from 2% to 16%; GFAA performance was similar with a range of 3% to 21%.
Pooled estimates across all metals except antimony and chromium are shown
in the last column of each table. For the ICP metals included in these
estimates - arsenic, cobalt, lead, nickel, and selenium - the range was
4-9% for the matrices with a difference greater than two percentage points
between matrices being significant. The range for pooled estimates for
arsenic and selenium by GFAA were similar, 4-11%, with two percentage
points also considered significant.

Table 5 groups the metals analyzed by both methods (antimony, arsenic, and
selenium) for comparison. In overall recovery, GFAA performed better with
antimony and arsenic; ICP was better with selenium, but was spiked at much
higher levels. ICP had recoveries greater than 100% for both arsenic and
selenium. As noted above, ICP and GFAA had zero recovery with antimony in
clean soil; ICP also had nearly zero recovery in treated and oily wastes.
Although precision differences existed within individual matrices for
these two methods, the pooled precision estimates indicate that overall
performance was similar.
Metals Analyzed by ICP All
Antimony Arsenic Chromium Cobalt Lead Nickel Selenium Except
(Sb) (As) (Cr) (Co) (Pb) (Ni) (Se) Sb, Cr
Reagent Water
Average 89% 120% 96% 98% 93% 98% 103% 102%
Std. Deviation (7%) (12%) (4%) (2%) (7%) (3%) (3%) (7%)
TCLP Leachate
Average 108% 108% 107% 97% 98% 100% 107% 102%
Std. Deviation (3%) (8%) (4%) (4%) (5%) (4%) (6%) (6%)
Clean Soil
Average 0% 109% 109% 96% 98% 100% 111% 103%
Std. Deviation NC (4%) (6%) (3%) (7%) (2%) (2%) (4%)
Treated Waste
Average 8% 113% NA 110% 98% 126% 116% 113%
Std. Deviation (16%) (11%) NA (14%) (6%) (7%) (6%) (9%)
Oily Waste
Average 5% 125% NA 90% 58% 102% 118% 99%
Std. Deviation (3%) (7%) NA (4%) (5%) (5%) (4%) (5%)
NA = Not Applicable, Not spiked
Metals Analyzed by GFAA
Antimony (Sb)Arsenic (As) Selenium(Se) Excluding (Sb)
Reagent Water
Average 105% 120% 86% 103%
Standard Deviation (5%) (8%) (6%) (7%)
TCLP Leachate
Average 100% 98% 91% 94%
Standard Deviation (4%) (3%) (5%) (4%)
Clean Soil
Average 0% 76% 61% 68%
Standard Deviation NC (5%) (3%) (4%)
Treated Waste
Average 70% 93% 94% 94%
Standard Deviation (21%) (3%) (16%) (11%)
Oily Waste
Average 59% 82% 54% 68%
Standard Deviation (6%) (3%) (10%) (7%)

Recovery +1-. Standard Deviation(% )
Matrix Antimony Arsenic Selenium
Reagent Water
GFAA 105% + 5% 120% + 8% 86% + 6%
ICP 89% +7% 120% +12% 103%+ 3%
TCLP Leachate — — —
GFAA 100% + 4% 98% + 3% 91% + 5%
ICP 108% 1 3% 108 + 8% 107%+ 6%
Clean Soil
GFAA NC, 0% 76% + 5% 61% + 3%
ICP NC, 0% 109%+ 4% 111% ÷ 2%
Treated Waste
GFAA 70% + 21% 93% + 3% 94% + 16%
ICP 8% +16% 113% 11% 116%Th- 6%
Oily Waste
GFAA 59% + 6% 82% + 3% 54% + 10%
ICP 5% 3% 125%Th 7% 118%Th- 4%
Overall Average Recovery — —
GFAA 67% 94% 77%
ICP 42% 115% 111%
Pooled Standard Deviation
GFAA 10% 5% 9%
ICP 8% 9% 4%
NC = Not calculated, 0% recovery
MDLs calculated from the standard deviation of the replicate samples are
shown in Table 6 for both ICP and GFAA. For comparison with the aqueous
matrices, reagent water and TCLP leachate, detection limits published in
the analytical methods are given in the last column. The published
detection limits are for a clean, interference-free matrix and are
directly comparable to reagent water.
The MDLs for reagent water compared well with the published detection
limits. For ICP, MDLs for all the metals were lower, in some cases
significantly lower, than the published values. The MOLs for the three
metals analyzed by GFAA also compared well with values published in the
methods. Although arsenic and selenium MOLs in this study were twice the
published values, the absolute difference was not great (1-2 ugIL).
TCLP leachates were spiked 4-20 times higher than reagent water, and MDLs
were 4-20 times higher than reagent water. Clean soil MDLs were generally
lower than the waste matrices; however, the waste matrices contained
higher background levels of many of the analytes which elevates the MDL
value. In some cases, GFAA spiking levels were the same as spiking levels
used for ICP and MDL values were generally similar. However, GFAA is the
more sensitive method of the two.

Note different Reagent TCIP Clean Treated Oily Given in
units for liquids Water leachate Soil Waste Waste Method
and solids --> (ug/L) (ug/L) (mg/kg) (mg/kg) (mg/kg) (ug/L)
ICP Analysis
Spike level for 50 ug/L 500 ug/L 50 mg/kg 50 mg/kg 50 mg/kg
all metals except500 ugh (Se) 200 ug/L (Cr,Co) 20 mg/kg (Cr,Co) 20 mg/kg (Co) 20 mg/kg (Co)
as noted below. 2,000 ug/L (Se) 200 mg/kg (Se) 200 mg/kg (Se) 200 mg/kg (Se)
Cr not spiked Cr not spiked
Antimony 11 45 25 4 32
Arsenic 18 120 6 17 10 53
Chromium 6 23 4 62 86 7
Cobalt 3 28 2 9 2 7
Lead 10 82 11 10 8 42
Nickel 4 63 3 11 8 15
Selenium 43 365 16 35 27 75
GFM Analysis
Spike level for 20 ug/L 500 ugh 50 mg/kg (Sb) 50 mg/kg 50 mg/kg
all metals except 10 ug/L (As) 200 ug/L (Se) 5 mg/kg (As) 2 mg/kg (Se) 2 mg/kg (Se)
as noted. 2 mg/kg (Se)
Antimony 3 61 33 10 3
Arsenic 2 49 1 4 5
Selenium 4 33 0.2 1 1 2
# Zero for all replicates

Recovery (%)
Reagent TCLP Clean Treated Oily
Water Leachate Soil Waste Waste
Spike Level 5 ug/L 100 ug/L** 20 ug/kg 1,250 ug/kg 5,000 ug/kg
ng on column 25 100 100 50 50
Acetone 104 83 215 76 160
Benzene 110 96 97 93 105
2-butanone 104 119 137 70 48
Carbon disulfide 86 68 103 60 79
1,2-dibromoethane 98 105 88 79 92
1,4_dioxane*** 86 71 116 54 92
Ethylbenzene 99 109 104 111 124
Methylene chloride 108 100 138 307 109
Styrene 83 91 85 82 103
Toluene 107 95 102 94 106
m + p Xylenes 119 160 151 155 178
o-Xylene 97 122 109 115 128
Median 101 106 -- 105
* Most analytes in reagent water spiked at 5 ug/L, some at higher
levels (acetone and 2-butanone 25; 1,4-dioxane 500)
** TCLP leachate diluted 1:5 in reagent water
Except for Reagent Water, 1,4-dioxane spiked 20 X higher
Matrix Recovery +1- Standard Deviation (%)*
Reagent water 99% + 21%
TCLP leachate 99% + 11%
Clean soil 95% 6%
Treated waste 92% 3%
Oily waste 106%Th 8%
*Excluding outliers
Vol atiles
Seven samples of each of the five matrices were used for recovery and
precision evaluation of volatiles. Average recoveries are shown in
Table 7. Some problems were encountered with several of these analytes as
seen by the abnormally low and high recoveries in the table. Calibration
problems were experienced with the xylenes and blank contamination was
suspected with acetone and methylene chloride analyses. Subsequent
outlier analyses identified 2-butanone; carbon disulfide, and 1,4-dioxane
with recovery outside the normal performance for some matrices. (Grubbs

test for outliers [ 91 was used for this and subsequent outlier testing.)
The median was used to describe the overall performance for each matrix
prior to testing for outliers.
Matrix effects on recovery and precision for refinery volatiles were
analyzed only with the five analytes that did not have calibration, blank
or other problems. These five were benzene, 1,2-dibromoethane,
ethylbenzene, styrene, and toluene. A suninary of the performance of each
matrix is shown in Table 8 together with a pooled estimate of within-
analyte precision. Although the range in average recovery was 92%
(treated waste) to 106% (oily waste), no statistical difference was found
between matrices for these refinery volatiles. Note that the variable
recoveries for some outliers may be attributable to particular matrices.
For example, recovery of carbon disulfide is lower in TCLP leachate (68%)
and treated waste (60%)than the other matrices (79-103%). Similar results
for some matrices were observed with 2-butanone and 1,4-dioxane.
The tests for difference in precision showed that the precision for TCLP
leachate and oily waste were equivalent and that the precision of clean
soil and oily waste were equivalent; all other precision tests showed
significant differences. Treated waste had the best precision, followed
by clean soil and oily waste, then TCLP leachate, and finally reagent
water. This relatively poor performance with reagent water was not
expected because it was the simplest matrix. The poorer precision is
probably the result of spiking levels 2 to 4 times lower in reagent water
in comparison to the levels spiked in other matrices. Table 7 lists the
spike level and nanograms (ng) on column for a direct comparison of
spiking level by matrix. Given that precision often degrades as
concentration decreases and that the spiking levels for reagent water were
near the method detection limit, it is likely that this is the cause of
the apparently “poorer” precision performance for reagent water.
MDL5 for the volatile analytes are shown in Table 9. The MDL5 for
1,4-dioxane were much higher than for the other analytes; however, the
MDLs were consistent with the higher spike levels used for 1,4-dioxane.
Ranges in MDLs, excluding 1,4-dioxane, were:
Reagent water 2-9 ug/L
TCLP leachate 30-87 ug/L
Clean soil 3-13 ug/Kg
Treated waste 69-1600 ug/Kg
Oily waste 390-9600 ug/Kg
Semi vol atiles
The performance of the three sample preparation methods (initial
duplicates) for clean soil, treated waste, and oily waste for refinery
analytes is compared in Table 10. Overall, there was no significant
differences among the three methods. This is shown by the range of the
average recoveries across all matrices (73-79%) and the variability in
recoveries when looking at individual matrices and methods. An analysis
of variance on the recoveries showed no difference between methods or
between matrices.

Method Detection Limits (MDLs)
Reagent TCLP Clean Treated Oily
Water Leachate Soil Waste Waste
(ug/L) (ug/L) (ug/kg) (ag/kg) (ug/kg)
Acetone 7 87 8 700 3400
Benzene 2 33 5 110 830
2-butanone 9 51 12 1100 9500
Carbon disulfide 3 30 13 69 1800
1,2-dibromoethane 4 34 4 140 2500
1,4-dioxane 220 2600 460 14000 47000
Ethylbenzene 4 36 5 150 620
Methylene chloride 3 45 4 1600 2800
Styrene 3 34 3 89 850
Toluene 3 32 4 120 390
m + p Xylenes 4 54 8 230 830
o-Xylene 3 43 4 110 930
Median 3 40 5 140 1400
See Table 7 for spike level.
Extraction Method
3550 3550 Handbook
MC:AC MC Acid/Base
Recovery + Standard Deviation (%)
Clean soil 82% + 3% 74% + 10% 73% + 3%
Treated waste 71% 11% 80% i 5% 84% i 9%
Oily waste — 63% 4% 79% 18%
All Matrices 77% + 8% 73% + 7% 79% + 12%

The standard deviations in the table represent the pooled estimate of
precision for individual analytes (within -analyte precision). Precision
ranged from 3% to 18%. Statistically, there were some differences among
the matrices:methods, suggesting that the handbook method was less precise
than SW-846 Method 3550. However, there was no strong trend in the data
and given the small number of samples (2) for each combination, the
statistical differences may not be borne out in long-term performance.
In the comparison of extraction methods, five of the refinery analytes
were excluded. Acenaphthylene and benzenethiol were excluded because of
calibration problems with these analytes and three other analytes were
excluded because they were outliers: bis(2-ethylhexyl)phthalate; 7,12-
dimethylbenzanthracene; naphthalene. Naphthalene was an outlier because
the compound was not detected due to aldol condensate interference in the
methylene chloride:acetone (MC:AC) extractions.
Since no extraction was statistically different, the MC only Method 3550
extraction was selected for continued evaluation. The MC:AC procedure did
produce aldol condensate interference as expected. This is shown in
Figure 1 by comparing the mass spectra of naphthalene from the MC only
extract to the MC:acetone extract. The handbook technique was more labor
intensive and was not selected because there was not demonstrated method

2 S4.
.1 ii i ill
—T 1T1 1 1- I I I 1 L ‘
, j’L_ - j L
I •
I s
A. Methylene Chloride Only Extraction
B. Methylene Chloride:ACetOfle ExtractiOn/AldOl Condensate Interference
i _I
‘ l iz
I —
‘ -7

A. Standard Spectrum
N. S
Ii 11 IL
. i .. ..L. ii
ii. pr--r’i— • ‘ i . 1 r i rr IT
c — i ‘ . ‘ uI i ‘ ‘ i .
B. Alumina Column Cleaned
$114. ,S.S
U -L
‘LL LIiL tMI4 ::i:
C. GPC Column Cleaned
D. Uncleaned
‘4. .
.i ii .1
I I • • I I I
.11 I
I I . ,’
II. ’.
144 1.

Cleanup for semivolatile analysis using GPC and alumina columns were
compared for treated and oily wastes. One of the objectives of this
comparison was to see which analytes were not recovered by each column. A
second objective was to see what improvement in recovery, precision, and
detection limits was achieved at the cost of losing analytes during the
cleanup procedure.
Baseline performance of the GPC and alumina columns was first determined
with a standard solution. The average recovery for GPC and alumina was
83% and 78%, respectively; however, this was not a statistically
significant difference. The outliers and compounds with calibration
problems were excluded as were phenols. For the standard test only,
acid/base partitioning was not done prior to the alumina column cleanup.
Excluding the phenols, all compounds were recovered.
The second objective was evaluated using the treated waste and oily waste
data obtained from the seven replicates (analyzed after alumina column
cleanup) plus the splits analyzed after GPC (3 or 4) plus the splits
analyzed without cleanup (3 or 4). Results are summarized in Table 11.
Of the 28 analytes, 7 were omitted due to the outlier test or calibration
problems. Based on these results, the column cleanups did not have a
significant impact on the precision of the data. After cleanup,
single-analyte precision was either poorer or unchanged when compared to
no cleanup. Column cleanup did a poorer job of recovering four of the
PNAs from oily waste than no cleanup at all. These PNAs were 7,12-
dimethylbenzanthracene; 1-methyl naphthalene, phenanthrene, and pyrene.
In overall recovery, neither GPC nor alumina column cleanup showed a
statistical difference in overall recovery for treated and oily wastes
when compared to no cleanup (Table 11).
However, column cleanups do offer some advantages that are not easily
quantifiable. For example, the uncleaned samples were not analyzed at the
same concentration as samples which had been °cleaned” of interference
compounds. For example, the oily waste samples were analyzed at the
following sample weight/solvent volume ratios: uncleaned, 1 g/1O mL; GPC
cleaned, 1 g/2 mL; alumina cleanup, 1 gIl mL. As a result, the uncleaned
samples had higher PQLs than the column cleaned samples. In this study,
results for the uncleaned samples were reported below the PQL, which was
not necessary for the column cleaned samples. The spectra obtained below
the reporting limit are in general poor quality, and at times, technically
questionable (see Figure 2).

Cleanup Method
No Cleanup GPC Alumina All Methods
Recovery + Standard Deviation
Treated Waste 83% + 33% 87% ± 14% 78% + 14% 82% ± 22%
(number of samples) ( ) (4) (7) (14)
Oily Waste 61% + 16% 48% ± 22% 52% + 15% 53% + 18%
(number of samples) ( ) (3) (7) (14)
All Matrices 72% + 2691 68% + 18% 65% ± 15%
(number of samples) (7) (7) (14)
The GPC column cleanup removes high molecular weight interference, while
the alumina column cleanup removes interference from high concentrations
of aliphatic hydrocarbons. See Figure 3 for comparison of the cleanups
based on the total ion chromatograms and mass 57 (hydrocarbon)
chromatogram. Both cleanups are useful for their intended purpose, and
both may be necessary based on the materials in the sample. As discussed
in Method 3600 (cleanup), cleanups are designed to purify extracts to
prevent deterioration of column efficiency and loss of detector
sensitivity. Cleanups can also increase the lifetime of expensive
columns. While eliminating cleanups may apparently improve precision by
eliminating extra steps, the constant introduction of contaminants into
the measurement system will eventually degrade the performance of the
system and results in poorer detectability of analytes.
Recovery and Precision by Matrix
Seven samples were analyzed for each matrix with the exception of TCLP
leachate which had eight samples. The solid samples were extracted with
100% methylene chloride using the low level sonication Method 3550.
Following the extraction, the waste samples were cleaned using acid-base
partitioning and alumina column cleanup. Average recoveries for the
analytes are in Table 12. Semivolatile performance is shown in Table 13.
The standard deviations in Table 13 are pooled estimates of the within
analyte precision, excluding eight of the 28 analytes eliminated as
outliers or due to calibration problems. Statistical differences were
found between matrices in both recovery and precision. The recovery for
oily waste was lower than the other four matrices. The recoveries for
TCLP leachate, clean soil and treated waste were statistically the same.
In precision performance, TCLP leachate ranked first, followed by clean
soil. Reagent water precision tested equivalent to treated and oily
wastes. As noted for the volatiles, this poorer precision is related to
the lower spiking levels used for reagent water.
MDLs for the semivolatiles are shown in Table 14. Ranges in MDLs were:
Reagent Water 1-11 ug/L
TCLP Leachate 4-54 ug/L
Clean Soil 90-830 ug/Kg
Treated Waste 300-16000 ug/Kg
Oily waste 2800-59000 ug/Kg

v a
1 g/1O mL
1 g/2 mL
1 g/l mL
ma r

Reagent TCLP Clean Treated Oily
Water Leachate Soil Waste Waste
Spike Amount 5 100 1,000 5,000 25,000
ug/L ug/L ug/kg uglkg uglkg
Acenaphthene 96 73 83 83 45
Acenaphthylene 100 18 26 20 20
Anthracene 100 68 84 78 52
Benzenethiol ** 147 159 219 226
Benzo(a)anthracene 95 73 89 81 58
Benzo(b)fluoranthene 86 68 100 97 62
Benzo(k)fluoranthene 87 78 83 100 72
Benzo(g ,h,i)perylene 75 70 86 78 17
Benzo(a)pyrene 83 65 82 80 50
Bis(2-ethylhexyl)phthalate 291 87 90 78 73
Chrysene 107 83 91 89 49
Dibenzo(a,h)anthracene 73 73 87 76 67
Dibenzofuran 96 85 83 89 76
7,12-Dimethylbenzanthracene 77 60 69 45 43
2,4-Dimethyiphenol 81 57 79 81 50
Fluoranthene 97 75 87 81 55
Fluorene 93 75 83 85 40
lh-indene ** 6d 100 88 77
Indeno(1,2 ,3-cd)pyrene 67 73 86 71 60
3-Methylchloanthrene 53 46 61 51 36
1-Methylnaphthalene ** 59 82 39 24
2-Methylnaphthalene 87 78 83 77 56
3-&4-methylphenol 102 78 80 73 55
2-methyiphenol 86 76 77 76 57
Naphthalene 86 64 79 76 52
Phenanthrene 103 72 87 88 20
Phenol 95 80 77 62 47
Pyrene 104 74 94 91 38
Median 93 73 83 79 52
* Most analytes in reagent water spiked at 5 ug/L, 3-methylchloanthrene
spiked at 10 ug/L.
** Not spiked
Matrix Recovery ± Standard Deviation (%)
Reagent Water 91% ± 13%
TCLP Leachate 73% + 5%
Clean Soil 83% 8%
Treated Waste 79% 11%
Oily Waste 52% 15%

Reagent TCLP Clean Treated Oily
Water Leachate Soil Waste Waste
Note different
units for liquids
and solids --> (ugfL) (ug/L) (ug/kg) (ug/kg) (ug/kg)
Spike Amount
(in units shown) 5* 100 1,000 5,000 25,000
Acenaphthene 2 13 240 920 12000
Acenaphthylene 3 4 94 320 2800
Anthracene 2 12 230 980 10000
Benzenethiol ** 51 830 16000 35000
Benzo(a)anthracene 2 16 240 1300 10000
Benzo(b)fluoranthene 2 17 340 6400 9700
Benzo(k)fluoranthene 2 12 260 2800 13000
Benzo(g,h,i)perylene 1 11 210 4000 26000
Benzo(a)pyrene 1 14 240 1200 10000
Bis(2-ethylhexyl)phthalate 11 54 250 6200 59000
Chrysene 2 19 250 1600 15000
D’ibenzo(a,h)anthracene 2 22 250 2300 7800
Dibenzofuran 2 14 220 790 8600
7,12-.Dimethylbenzanthracene 1 13 310 2800 17000
2,4-Dimethylphenol 3 9 360 1400 7000
Fluoranthene 1 15 250 1400 11000
Fluorene 2 14 220 980 16000
lh-indene ** 17 360 1100 9100
Indeno(1,2,3-cd)pyrene 2 15 240 2300 7000
3-Methylchloanthrene 4 9 180 2300 6100
1-Methylnaphthalene ** 10 260 1200 17000
2-Methylnaphthalene 2 16 300 2800 10000
3-& 4-methyiphenol 2 19 260 1200 9600
2-methyiphenol 3 31 530 1900 21000
Naphthalene 2 15 290 1800 8600
Phenanthrene 2 15 260 1600 23000
Phenol 2 18 260 990 9500
Pyrene 2 13 250 2600 18000
Median 2 15 250 1600 10000
* Most analytes in reagent water spiked at 5 ugIL, 3-methylchloanthrene
spiked at 10 ug/L.
** Not spiked in reagent water

For metals, there were some statistically significant differences in
recovery and precision; however, in general, the performance on the
different matrices was similar. A notable exception was the low recovery
of antimony in solid matrices. For the five volatiles which did not have
calibration, blank, or other problems, there was no statistical
difference between recovery for the various matrices. Using the same
five volatiles, treated waste had the best precision, followed by clean
soil and oily waste, then TCLP leachate, and finally reagent water. The
poor performance for reagent water was apparently related to the absolute
spiking level; similar performance was seen for the semivolatiles. For
the semivolatile extraction, no clear differences in extraction
techniques were seen, but the MC only extraction was chosen as the
simplest and because it did not produce the aldol condensate products in
the MC:AC extraction. The evaluation of cleanup techniques indicated
that cleanup procedures do not improve precision. However, the cleanups
do allow the samples to be run efficiently with lower PQLs. The cleanups,
particularly the alumina cleanup for refinery wastes, do remove
interferences which eventually can degrade the performance of the GC/MS
system. For the semivolatile matrix evaluation, statistical differences
were found between matrices in both recovery and precision. The recovery
for oily waste was lower than the other four matrices. The recoveries
for TCLP leachate, clean soil and treated waste were statistically the
same. In precision, TCLP leachate ranked first, followed by clean soil.
Reagent water precision tested equivalent to treated and oily waste. As
noted above for volatiles, this poor precision for reagent water appeared
to be related to the spiking level.
Overall, the data indicates that, in some cases, precision and accuracy
values based on the analysis of water can be extended to other matrices
for the analytes studies. However, with particular analytes and
particular matrices, there may be significant differences in precision
and accuracy. These can generally only be determined by a detailed study
of the matrix and the analytes in question.

1. “Evaluation of Analytical Methods for Measuring Appendix IX
Constituents in Groundwater,” API Publication No. 4499, July 1989.
2. “Test Methods for Evaluating Solid Waste, Physical/Chemical Methods,”
USEPA, OSW/Washington, DC, SW846, 3rd Edition, November 1986.
3. 40 Code of Federal Regulations Part 136
4. “Handbook for the Analysis of Petroleum Refinery Residuals and
Wastes,” Radian Corporation for EPA/OSW, October 1984.
5. Hem, C.S. and A.B. Shartleff, “Extraction of Semivolatile Appendix IX
Compounds from Solid Samples,” USEPA Symposium on Waste Testing and
Quality Assurance, Vol. II, July 1988
6. Brilis, G. M. and P. J. Marsden, “Comparative Evaluation of Soxhlet
and Sonication Extraction in the Determination of Polynuclear Aromatic
Hydrocarbons in Soil,” Chemosphere, 1990, Vol. 21 1/2.
7. Marsden, R.J. and J. Longbottom, “Evaluation of Method 3640 (GPC
Cleanup) for Appendix VIII Analytes,” USEPA Symposium on Waste Testing
and Quality Assurance, Vol. II, July, 1987.
8. “USEPA Method Study 37, Sw-846 Method 3050 Acid Digestion of
Sediments, Sludges, and Soils,” EPA/600/54-89/012 July 1989, NTIS
Order No. PB 89-181 952/AS.
9. Grubbs, F. E., and G. Beck, “Extension of Sample Sizes and Percentage
Points for Significance Tests of Outlying Observations.”
Technometrics, TCMTA, 14 (No. 4) :847-54 (November 1972).

Lance A. Hines. Ph.D. . Chemistry Section Manager, Steven E. Mornssette, Senior Staff Geologist, and
Curtis J. Clowe, Chemist, Environmental Division, Woodward-Clyde Consultants, 101 South 108th
Avenue, Omaha, Nebraska 68154
Quality assurance issues associated with the identification and collection of representative samples for
treatability testing don’t always receive the appropriate attention. The ramifications can be monumental
in terms of lost time and expenditures and more importantly, the failure to properly remediate a site.
Treatability studies in general are very difficult to plan and perform, usually due to the number of
potential variables, an inadequate amount of funding needed to address all of the critical variables, and
extreme pressures from an already tight schedule. This predicament is exacerbated by the limited
understanding or the inadequate amount of time and resources used in the development of the data
quality objectives (DQOs) at the outset of the Rl/FS. Because of these factors, the importance of the
representativeness of the treatability samples that are collected becomes even more critical. The
treatability study discussed herein was performed under all of the aforementioned pressures and was
successful in part because of the recognition of the need for representativeness. The purpose of this
treatability study was initially to confirm a previous study replacing the former performance criteria of
the Extraction Procedure Toxicity (EPTOX) test with that of the Toxicity Characteristic Leaching
Procedure (TCLP). The study involved the development of a solidification/stabilization mix design
capable of p c.cing the TCLP test. The site was a former lead battery breaking and reclamation
operation located along the James River in VirginiL The lead concentration in the surface and near
surface soil reached concentrations of over 13 percent. Preliminary results of the test samples indicated
failures of the TCLP test of over 5000 percent including the mix originally purported as achieving
leaching concentrations of under 5 mg/L using EPTOX methodology. A review of the strict QA/QC
procedures used in sampling and analysis provided clues as to the cause of the failures. Samples for the
treatability tests were based on incorrect RI/FS data characterizing the concentrations of lead and the
surface and subsurface distribution of these concentrations. After a limited field investigation, a better
understanding of site conditions afforded the collection of samples which were more representative.
More appropriate mix preparations, coupled with a better understanding of the site chemistry, facilitated
the development of remedial mix designs capable of meeting TCLP performance criteria. The results
of the treatability study will be presented and the lessons learned will be discussed. This study was
performed under contract with the US Army Corps of Engineers for the USEPA Region Ill.
A limited bench-scale treatability study conducted in support of the remedial design for a Superfund site
in Virginia was intended as substantiative work to a previous stabilization/solidification study. The study
was designed to confirm the recommended stabilization mix components of Type II Portland cement,
soil, and either lime or sodium phosphate as presented in the EPA Record of Decision (ROD) for the
site. In accordance with the ROD, the contaminated material was to be stabilized and transported for
off-site disposal. Due to changes in RCRA regulatory requirements, i.e. the adherence to the TCLP
criteria in lieu of the EPTOX, it was important to refine and confirm the recommended treatability mix
During the excavation of the initial treatability test pit, a distinct underlying clay layer was encountered
at about 2 feet. This appeared to be inconsistent with existing site information. As it was desired to

obtain the most representative sampling and analysis of soils in selected areas across the site, this clay
layer was of concern. As such, additional activities including the installation of test pits and borings and
subsequent soil sampling and analysis were performed to better characterize site conditions which might
affect the overall treatment design and afford the evaluation of representativeness of treatability samples.
We present here the results of the treatability study.
Site Background
The site, located in EPA Region ifi, is approximately 11 acres in size and is located along the James
River in Virginia. The site is bordered on the west and north by dense woods, on the northeast by
sparser woods, and on the south and east by open fields and several buildings. The general area around
the site is used for industrial purposes.
The site is nearly level and is included in the Coastal Plain physiographic region of Virginia. The
Coastal Plain of Virginia is made up of various Pleistocene and Holocene Age sedimentary deposits
overlying older sedimentary deposits which overlie the granitic basement rock. In the region around the
site, and especially in areas near the James River, the granitic basement rock was expected to be
shallow. Borings and test pits indicated that the shallow subsurface profile included man-placed or
disturbed fill over alluvial deposits. The fill consisted of a mixture of sand, silt, and day with varying
amounts of crushed rock, cobbles, and battery casing fragments. The observed thickness of the fill
ranged from 2 to 4 feet. The alluvium is typically low to medium plastic sandy silty clay and extended
to the maximum depth of the test pits and borings 15 feet below ground surface. The actual extent was
not determined. This clay horizon appeared to be continuous under the site. The original RI/FS boring
logs indicated silty sand which was not consistent with observations made of treatability sample test pits.
The site was used from the early 1970s to 1985 for the recycling of used auto and truck batteries which
were delivered in bulk shipments. After cutting the batteries open and draining the acid into an on-site
retention pond, the lead-containing components of the batteries were removed and stored on site for
later processing. The empty battery casings were then shredded and also stored on site. Battery casing
fragments are currently present on the surface and in the shallow subsurface. Whole battery casings
were occasionally unearthed during test pit excavations.
Previous Stabilization/Solidification Treatability Studies
During the remedial investigation/feasibility study (RI/FS) an initial stabilization/solidification
treatability study was conducted at the site. This study investigated various combinations of soil, Type
I Portland cement, fly ash, lime, and a proprietary component. Based on preliminary results, it was
necessary to perform additional testing since results with Type I Portland cement and fly ash were
inconclusive. The additional stabilization mixes consisted of soil, Type II Portland cement, lime, sodium
silicate, and sodium phosphate in various combinations. The results of this study indicated that the two
mixes having the greatest reduction in lead mobility by the Extraction Procedure Toxicity (EPTOX) test
contained 36 and 48 percent Type II Portland cement combined with about 5 percent sodium silicate
and about 3 percent lime, respectively. Other mixes including one with sodium phosphate also were
effective with the level of Portland cement at 48 percent.
These studies seemed to indicate that the likelihood of achieving stabilization of lead was high using
Type II Portland cement combined with an additive such as lime to control the pH. It was determined
that fly ash was not a viable alternative.

Treatability Study Approach
The overall objective of the study was to better define the stabilization/solidification mix ratio of Type
II Portland cement, soil/sediment, and additives. A subsequent objective of the study was to provide
insight into the potential limitations of the design process by better characterizing subsurface soils with
respect to debris, contamination, and soil type. The general approach to this treatability study was
phased to provide the representative mix ratio recommendations under the given cost and time
constraints. The following paragraphs outline some of the various phases of the approach.
The first phase, which ran concurrently with the subsequent phases, involved the representative sampling
and analysis of soils in selected areas across the site. This phase included the screening and
characterization of soil components as well as the determination of levels of metals. This phase included
drilling and sampling of soil, water, and waste debris within the site and the nearby property. The study
also included physical and chemical laboratory testing of untreated soil, water, waste and
stabilized/solidified mixtures. Information concerning properties of the site soils/materials provided
data important for treatability, as well as the potential for volume increase of the treated material.
A second phase of the study involved the preparation of various stabilization/solidification mix ratios
of Type II Portland cement, hydrated lime, sodium phosphate and, subsequently, calcium phosphate
designed to focus the mix ratio recommendation based on the previous treatability study. The intent
during this phase, which included a series of physical and chemical tests, was to provide treatability
information on two areas of greatest concern with respect to chemical fixation as well as material
handling Based on preliminary test results, the treatability study design was modified to indude calcium
phosphate in lieu of sodium phosphate. Further physical and chemical tests of the potentially successful
stabilized/solidified mixes were used to assess feasibility in the treatability process.
The first of the two areas of concern centered on the drainage ditch where sediments posed treatability
and handling questions. The second area of concern addressed as part of this phase was in the vicinity
of some former acid ponds. Typical remedial concerns in this area were battery chip content (from
processed casings) , cobble content, soil pH, and high lead levels. This area was selected to provide a
representative sample indicative of the worst-case soils to be remediated. The recommendation of the
typical remedial mix ratio was to be based on results obtained on the representative sample taken from
this area.
Samples were obtained initially from two test pits (TP-l and TP-2), one from the reported acid pond
area and another from the drainage ditch in order to address possible differences in treatability based
on soil type, p11, and lead levels. In an attempt to obtain representative treatability samples, material
was composited from the face of the test pits in proportions consistent with reported contamination
levels as well as practical considerations with respect to excavation operations. As previously discussed,
additional test pits (TP-3 through TP-7 and TP-4A, TP-SA and TP-6A) were excavated to better
characterize any lateral variabilities which might affect treatability. Soil borings were drilled to
investigate the shallow subsurface conditions. Undisturbed samples of the soil directly beneath the
contaminated soils were obtained from the borings using 3-inch-diameter Shelby tubes to evaluate soil
The laboratory testing program of the untreated soil samples consisted of physical and chemical tests,
including tests of soils from both the upper disturbed soil horizon and the underlying clay. The physical

properties determined included the classification (Unified Soil Classification System), moisture content,
density; permeability, and shear strength. The chemical tests consisted of pH, total metals, TCLP
metals, TCLP alkalinity, and TCLP pH. Activities also included the screening on-site of untreated soils
to provide information on the potential extent of non-process materials requiring pretreatment such as
battery casing chips and cobbles. Testing of untreated materials was also important for assessing the
representativeness of materials selected for treatability testing.
Additional test pits TP-4A , TP-5A, and TP-6A were located near the previously excavated TP-4, TP-S,
and TP-6. These locations were selected based on their potential for “representativeness’. These test
pits were excavated to depths where lead contamination was equal to or just below the site action level,
based on information obtained from laboratory testing and previous investigations. Samples from the
additional test pits were separated into two groups: upper fill material (disturbed materials) and lower,
undisturbed, natural soils or clay. Two additional test pits, TP-9 and TP-1O, were excavated to
characterize the subsurface conditions in the reported acid and neutralization pond areas. The depths
of all the test pits ranged from about 4 to 6 feet. In addition to test pits, soil borings were drilled and
soil samples were taken to assess the “representativeness” of soils requiring remediation.
Soil borings were drilled immediately adjacent to test pits TP-4A, TP-5A, and TP-6A, respectively. The
purpose of these borings was to obtain undisturbed Shelby tube samples from the natural soil material
at these sites for laboratory testing to better characterize conditions which might affect the remedial
process. Laboratory tests consisted of physical and chemical tests of both untreated and treated
soils/waste. The physical tests included particle size, Atterberg limits, moisture content, permeability,
unconfmed compressive strength, density, and moisture-density relationships for compacted materials.
Physical tests were performed according to American Society of Testing Materials (ASTM) standard
procedures, if applicable. The chemical tests included total metals analysis, pH, TCLP metals
(predominantly lead), TCLP pH, and TCLP alkalinity.
Samples were shipped to the laboratory for physical tests, bench-scale treatability mix preparations and
subsequently for chemical analysis. The untreated material was subdivided into four types: fill material,
natural soil or clay, composited material, and water. The treated material consisted of soil-cement
mixtures with hydrated lime, sodium phosphate, or calcium phosphate.
The mixing procedures for the soil-cement-lime-sodium phosphate and soil-cement-lime-calcium
phosphate mixtures were relatively similar. After geophysical testing, soil samples were dried for a short
period of time and sieved through a No. 4 (4.76 mm) sieve. The appropriate proportions of the
additives were mixed with the sieved sample. The ratios of the mix were based on the dry weight of the
sieved soil sample. For the soil-cement-lime-sodium phosphate mixtures, all the materials were added
at the same time. For the soil-cement-lime-calcium phosphate mixtures, lime and/or calcium phosphate
and a small amount of water were added to the soil 1 hour before the addition of cement to enhance
the exposure of the soil to the stabilizing additives. Twelve treatability samples were prepared using the
soil-cement-lime-sodium phosphate mix and subsequently 23 samples were prepared using the soil-
cement-lime-calcium phosphate mix. The secondary mixes were prepared from disturbed soils
composited from TP-4A, TP-5A, and TP-6A. This combined sample, prepared exclusively from the
upper material, was identified as TP-456. It was determined from preliminary data that these soils
would be most ‘representative” of the soils requiring stabilization/solidification based on additional site
characterization data.

Analysla Of Waste Stream Characteristics
Physical tt frnfkmn wzd Mo cture Content
Physical ehs.cifications and moisture content tests were conducted on both disturbed, uupperu material
and natural, undisturbed soils. The results of the physical classification and moisture content tests are
listed in Table 1. The results indicate that the composited subsurface soils are generally low plastic
(lean) sandy clay or sandy silt with 1 percent or less gravel (the coarser materials were removed on site).
The particle size ranges from about 2 mm (medium-grained sand) to less than 0.001 mm (silt or clay
particles). The characteristics of the gradation curves (coefficient of uniformity >15 and a coefficient
of curvature near 1) and the broad particle size range show that the soil is generally well graded. With
these observed particle size characteristics, the on-site soils are likely to show favorable physical
characteristics after stabilization/solidification.
Measured moisture contents of the shallow soils depend on the amount of rainfall and can vary
throughout the year. Due to the low permeability of the underlying soils, saturation of the shallow soils
is frequent. Soils received for geophysical and treatability testing ranged in moisture content from about
13 to 23 percent. The variability of the site soil moisture indicates a requirement of moisture adjustment
during the remedial process.
f .Mconflned Comp’rssi e Stren gth (ASTM 2166-85)
The tests indicated that on-site soils exhibit a stiff to very stiff consistency, implying that this soil would
be expected to provide reasonable support to most construction equipment during removal of the
cont min ted soils.
Bu& Denthy
The measured dry unit weights and moisture contents are contained in Tables 2a and 2b along with the
corresponding total unit weight (bulk density). The dry unit weight of the test pit samples ranged from
about 94 to 110 pounds per cubic foot, with an average of 104 pounds per cubic foot. The dry unit
weight of the Shelby tube samples (average sample depth about 4 to 5 feet) varied from about 103 to
110 pounds per cubic foot, with an average of about 106 pounds per cubic foot.
Pe meabiliry (SJWJ(R))
Samples were taken at the soil borings (via Shelby tube) from the soils directly beneath the disturbed
upper layer at the site for laboratory permeability analysis. The measured permeabilities ranged from
7.1 x i0 to 2.0 x 10 cm/sec. Thus, the soil beneath the disturbed upper layer can be considered a
relatively impermeable plastic clay that will not allow free migration of water.
TCLP MetaLc Analysis cj Unfrented Soil Samples
Untreated soil samples were collected from test pits and soil borings and analyzed to determine the lead
concentrations present (Table 3). TCLP results for lead of the composite samples ranged from 56.4
mg/L at TP-1 to 345 mg/L at TP-4.

The pH of untreated soils at the site was typically about 6. A lower pH (p11<3) was obtained in the
drainage ditch (TP-1) and the deeper natural clays. Alkalinity values for TCLP leachates had a
considerable range of ND to 92.0 mg/L reflecting the heterogeneity of the surface soils.
Lead concentrations are presented in Table 3 for samples from upper fill material, samples from natural
soils at depths between 3108 feet, and composite samples. The samples were obtained from the soil
borings as well as test pits within the site. Lead contaminRtion at the site was limited to the upper fill
material that ranged in thickness between 2 to 4 feet. Lead contamination was also confined to the top
2 feet at off-site locations. None of the samples from the 3- to 5-foot depth have lead levels greater
than 120 mg/kg.
Analysis Of Treatability Study Data
Mcãfztre-De uity Rdatio &ships of C pacted & il-Cement Mà flu s (AS1M I) 558-82)
Compaction tests were performed on the soil-cement mixtures. Samples from TP-1 and TP-2 were
mixed with various amounts of cement, lime, and sodium phosphate. The results of compaction tests
of TP-1 and TP-2 samples show that increases in cement ratio produce minor changes in the optimum
moisture content and maximum dry density. However, increases in lime and sodium phosphate ratios
generally increase the optimum moisture content and decreases the maximum dry density. Similar
trends were observed for the TP-456 (exclusively upper material) sample. Maximum dry density for TP-
I and TP-2 mixtures ranged between 102 and 110 pounds per cubic foot, with optimum moisture content
between 17 and 22 percent. The maximum dry density of the TP-456 mixes varied from 108 to 122
pounds per cubic foot and the optimum water content between 13 and 20 percent. Because the TP-456
sandy silt and TP-1 and TP-2 samples are sandy days, the above difference in moisture-density values
are expected. TP-456 material was considered “representative” of the soils to be remediated at the site.
The TCLP tests and economic considerations indicated that the soil-cement mixture containing 15
percent cement and 10 percent calcium phosphate had a high potential for meeting remedial
requirements. The optimum moisture content and maximum dry density for this mixture were 14.3
percent and 112.7 pounds per cubic foot, respectively. An estimate of the volume of materials to be
handled during the site stabilization/solidification process was calculated based on the moisture-density
curve for the above sample and the average dry density of untreated samples. The average dry density
of the on-site soils was determined in the field. Based on the available data, the final volume of
solidified/stabilized mixture in the field is expected to be 10 to 20 percent larger than the volume of
untreated soils. This estimate assumes that the solidified/stabilized mixture will be compacted to at
least 95 percent of the maximum dry density obtained from the laboratory moisture-density tests using
the above ASTM D 558-82 test procedure. The moisture content is assumed to be within ±3 percent
of the optimum moisture determined from the above test. It was also assumed that the mix ratios are
based on the dry weight of soil.
Pezineabilily of Stabilized/Solidified Mirtwr (SW9100)
The permeability of the recommended soil-cement mixture (15 percent cement and 10 percent calcium
phosphate) was evaluated using a flexible wall permeability test. The measured coefficient of
permeability was approximately 1.4 x 10 cm/sec. Although this value is slightly higher than the
maximum permeability of 1.0 i0 5 cm/sec recommended by the EPA, the stabilized material must meet
TCLP criteria rendering it a noncharacteristic waste.

Unawrfrsed Cmqxessive Soength (Stabl(ized/So&ilfied Mir1w (ASTM I) 1633-PA)
Two unconfined compression tests were conducted to evaluate the strength of the recommended soil-
cement mixture (15 percent cement and 10 percent calcium phosphate). The average unconfined
compressive strength was about 340 pounds per square inch (psi), about 25 times higher than the
strength of untreated soil. The measured unconfined compressive strength is well above the minimum
recommended by EPA (about 50 psi). The stabilized/solidified material is expected to provide stable
support for construction equipment, cover material, overburden, and any other materials that might be
placed upon it.
TCLP j Soil-Cement Mixtures
The results of TCLP tests for stabilized/solidified soil-cement mixtures are presented in Tables 4 and
5. Only two samples of the soil-cement-lime-sodium phosphate (Table 4) resulted in TCLP lead values
less than the regulatory level (5 mg/L). The measured TCLP lead value for both mixtures was 33
mgfL and contained 40 percent cement with 5 percent lime or sodium phosphate, respectively. The
remaining mixtures exhibited significantly higher TCLP lead values (up to 316 mg/L).
Based on the results shown in Table 4, the mix design was re-evaluated and sodium phosphate was
replaced with calcium phosphate in the additional soil-cement mixtures. As shown in Table 5, the TCLP
lead values were below detection limit in the following samples: 50 percent cement; 15 percent cement
and 20 percent calcium phosphate; and 15 percent cement and 10 percent calcium phosphate. Several
other samples contained low levels of TCLP lead (samples with cement ratios between 15 and 20
percent and calcium phosphate between 10 and 20 percent). The mixture that contained 15 percent
cement and 10 percent calcium phosphate was considered the most suitable mixture for remedial design
because of the lower mix ratios and, therefore, lower cost. The mixture also had acceptable physical
characteristics such as strength, low permeability, and density. Also, the net increase in volume of
treated material would be lower for this mix ratio, thereby reducing costs of off-site disposal.
This limited bench-scale treatability study was intended as substantiative work to a previous
stabilization/solidification study and to confirm adherence to the TCLP criteria in lieu of the EPTOX.
Additional activities included the installation of test pits and borings and subsequent soil sampling and
analysis were performed to better characterize site conditions which might affect the overall treatment
design and afford the evaluation of representativeness of treatability samples. Test pits dug for this
purpose were designed to simulate practical excavation conditions based on the depth of lead
contamination above action limits as presented in the RI/FS. The collected site material was sieved
prior to shipping in order to estimate the composition and extent of battery casing fragments and
cobbles in contaminated soils throughout the site. Physical characteristics of the site soils were also
examined to assess effects of soil type on the stabilization of lead.
Initial mixes containing 20, 30, and 40 percent Type II Portland cement with respective additional
components of 1 and 5 percent lime or 5 percent sodium phosphate were prepared for soils from two
locations, the drainage ditch (Fest Pit 1) and in the vicinity of the former acid pond (Test Pit 2).
Preliminary results indicated that only two mixes from TP-1 soils containing 10,000 mg/kg lead passed
the TCLP criteria of 5 mg/L No mixes from TP-2 containing 24,000 mg/kg lead gave TCLP leachate
levels below SO mgfL The two mixes which passed, having identical TCLP leachate values of 33 mgfL ,
contained 40 percent Portland cement and either 5 percent lime or 5 percent sodium phosphate.

Based upon these results, two trends were further evaluated which indicated a response with respect to
TCLP criteria. One was the effect of the percentage of Portland cement, and the other was the
presence of phosphate. As sodium phosphate is not readily available, calcium phosphate, supplied as
a fertilizer known as triple superphosphate, was selected for additional testing. Triple superphosphate
is readily available and, according to the literature, calcium phosphate typically forms a more stable
mineral complex than sodium phosphate at a lower pH range.
Additional treatability mixes were then planned to examine a wider range of Portland cement content
(15. 50 percent) with respect to lime and calcium phosphate. Initially, three site areas were selected
as representative for site treatability. However, further characterization of the site surface soils and
underlying day horizon necessitated treatability testing of only the upper 2 to 4 feet of site soils.
Therefore, the additional treatability mixes were prepared with sieved soils composited from the upper
2 feet from Test Pits 4, 5, and 6 (TP-456). This material appeared to be more representative of
materials to be encountered during full-scale remediations.
Preliminary results from the additional treatability testing of the more contaminated soils indicated a
strong stabilization effect with increasing calcium phosphate levels at a lower Portland cement content
of 15 percent. Based on these results, which included one mix with 10 percent calcium phosphate
meeting TCLP criteria, a final set of 5 mix ratios was prepared with 15 to 20 percent Portland cement
and 10 to 20 percent calcium phosphate, respectively:
• Mix I - 60 percent soil, 15 percent Portland cement, 15 percent CaPO 4
• Mix 2 -65 percent soil, 15 percent Portland cement, 20 percent CaPO 4
• Mix 3- 70 percent soil, 20 percent Portland cement, 10 percent CaPO 4
• Mix 4 . 65 percent soil, 20 percent Portland cement, 15 percent CaPO 4
• Mix 5 - 60 percent soil, 20 percent Portland cement, 20 percent CaPO 4
Results of TCLP testing on these mix preparations following a seven-day cure indicated that all met
TCLP criteria with four showing nondetect (ND) for lead.
Based on these results and the results of the previous stabilization study, a mix of 15 percent Portland
cement and 10 percent calcium phosphate was selected for confirmatory testing given its potential for
remedial success at this battery site.
Further confirmatory physical and TCLP leachate testing was performed on this selected mix. The
TCLP leachate contained 83 mg/L of lead, which was higher than the regulatory criteria of 5 mg/L.
Even though this mix has shown TCLP results of ND and 83 mg/L, this particular mix and the other
five mixes containing calcium phosphate are presented below for informational purposes as having
potential remedial success. The slight failure (comparatively) can be attributed, based on results of
further site characterization, to the heterogeneity of the upper 2 to 4 feet of site soils with respect to
lead contamination. Test results have indicated lead levels across the site from low parts per million
to in excess of 13 percent (130,000 mg/kg). It is this heterogeneity of concentrations and the physical
form of the lead present, i.e. complexed and fme metallic particles, which most likely account for
differences seen in lead levels determined in TCLP leachate from these mixes. It is this key factor and
recognition of treatability sample representativeness which afforded a clearer picture of the potential
for remedial failure.

Mix Soil. Cemcnt:CaFO 4
TCLP Lead mg/L
ND, 8.7
ND, 0.6
65:20: IS
ND, 032
ND, 032
Bulk densities indicated an increase of 15 to 20 percent in material after stabilization, calculated on a
dry weight basis. The material or site soils used in this treatability study, although representative of soil
type, was found to contain vaiying amounts of debris including battery casing fragments of all sizes,
battery components, and stone cobbles, all of which were sieved out. Also, it was determined that
moisture played a significant role in the mixing process with respect to the availability of lead to
reactants. Therefore, it was determined that various material handling considerations must be addressed
prior to treatment such as particle sizing, drying, and the sequence and residence time in which reactants
are combined, i.e. better results were obtained when samples were dried and mixed thoroughly with
reactants such as calcium phosphate prior to the addition of Portland cement and finally water. One
final factor which was not addressed and may contribute significantly to the effectiveness of the mixes,
particularly the last six mixes containing calcium phosphate, was the curing time of the stabilized mix.
There was evidence during the study that curing beyond seven days appeared to improve the
containment efficiency (stabilization), but these results were not conclusive.
This treatability study began initially as a simple confirmatory effort. It was through the recognition of
the need for representativeness and a clear focus on the remedial objectives that variables were
uncovered which could have contributed to a faulty or unacceptable design. Additionally, this study has
alluded to the importance of the data quality objective process. The representativeness is, at best, a very
difficult concept to evaluate and without a dear understanding of the limitations involved, serious and
costly consequences can result.

GR-TP2-23 2
SB91-2 0
S891-2 1
TP-4A (at 03’ depth) 4
TP-4A (at 1.5’ depth) 4
TP-5A (at 03’ depth) 4
TP.5A (at 1.5’ depth) 4
IP-6A (at 03’ depth) 4
TP-6A (at 13’ depth) 4
TP-9 (at 03’ depth) 4
TP-9 (at 13’ depth) 4
TP-10 (at 03’ depth) 4
TP-10 (at 15’ depth) 4
Sandy lean clay, CL
Sandy lean clay, CL
Sandy lean clay, CL
Sandy lean clay, CL
Sandy lean clay, CL
Sandy lean clay, CL
Sandy silty clay, CL-ML
Sandy lean clay, CL
Sandy silt, ML
Sandy lean clay, CL
Sandy lean clay, CL
Sandy lean clay, CL
Sandy lean clay, CL
Sandy lean clay, CL
Sandy lean clay, CL
Sandy lean clay, CL
Sandy lean clay, CL
Sandy lean clay, CL
Sandy lean clay, CL
Sandy lean clay, CL
Sandy lean clay, CL
Sandy lean clay, CL
Sandy lean clay, CL
Sandy lean clay, CL
Sandy lean clay, CL
0 33 67 33
1 21 78 43
1 34 6.5 30
o 38 62 25
37 62 28
0 27 73 26
1 31 68 25
0 20 80 NA
1 47 52 23
0 18 82 28
0 40 60 32
0 36 64 41
0 25 75 36
0 34 66 33
0 32 68 27
0 22 78 NA
0 24 76 NA
0 26 74 NA
0 32 68 NA
0 19 81 NA
0 16 84 NA
0 25 75 32
o 28 72 35
0 23 77 37
0 28 72 32
NA Not analyzed
TP Teat pit
SR Soil boring
Test Pit,
Soil Boring
Particle Size
Content 1
Silt &
As reor ed
2 Grth nnçle o( undinuthed day layer at tl ’-2
43%,ilt,37% clay
Composite ean Ia taken from ditceete borizoos at deptha below the upper ditturbed layer

TP-i 1
Dry Density
Dry Density
(%) —
20% Cement
19.0 108.3
+ 1% lime
19.9 107.2
+ 5% lime
203 106.3
+ 5% NaPO4
22.2 103.8
30% Cement
173 109.6
+ 1% lime
193 107.8
+ 5% lime
20.7 105.7
+ 5% NaPO4
21.0 105.6
40% Cement
183 109.2
+ 1% lime
19.9 108.0
+ 5% lime
20.1 106.2
+ 5% NaPO4
21.2 1053
Notc Values determined by ASTh4 D 558-82
‘Drainage ditch

(SAMPLE TP-456)’
Calcium Maximum Optimum
Cement Lime Phosphate Dry Density Moisture Content
(%) (%) (%) (pcf) (%)
30 10 — 116.0 16.9
30 10 114.8 17.1
30 10 10 1093 17.3
30 5 116.7 15.8
30 — 122.0 14.0
40 5 111.7 17.7
50 120.9 15.1
50 10 — 117.1 15.8
50 — 10 114.9 16.9
50 10 10 111.3 17.8
15 120.3 13.3
15 10 — 1155 15.4
15 10 112.7 14.3
1.5 10 10 108.7 19.8
40 121.3 14.5
40 10 — 117.2 15.6
40 10 119.3 15.8
40 10 10 109.9 18.5
15 15 111.2 173
15 20 110.7 17.5
20 10 114.2 173
20 15 112.6 17.2
20 20 109.8 17.7
Note: Values determined by ASTM D 558-82
1 Compositcd surface material from Test Pits TP-4A, TP-SA, and TP-M.

Test Fit
Total Lead’
Total Lead
Upper 2
03’ Horizon 3
13’ Horizon 3
61.8 (753)
1’P- S
18.1 (143)
Lead levels in sieved and composited materials ()Duplicate sample analysis
2 Lead levels in composite of upper materials only
Lead levels in composite samples of discrete horizons in the underlying clay at depths below the disturbed upper material
Lead levels in composite of upper materials from TP-4A , 1’P-5A, and 1’P-6A

Type II
Cement (%)‘
Mix percentages calculated on a thy weight basis

Type II Calcium TCLP
Soil Portland Lime Phosphate TCLP Lead TCLP pH Alkalinity
(%)2 Cement (%)2 (%)2 (%)2 (mg/L) (mg/L)
85 15 0 0 57.4 11.3 2,450
75 15 10 0 46.4 113 2,760
75 15 0 10 ND (8.7) 93 (5.1) 1,940 (1,440)
65 15 10 10 32.2 10.9 2,380
70 15 0 15 ND (0.6) 5.7 (8.0) 2,360 (2,500)
65 15 0 20 ND (ND) 6.3 (6.9) 2,380 (2,550)
70 20 0 10 0.75 (5.7) 11.0 (6.3) 2,340 (2,290)
65 20 0 15 ND (032) 8.2 (9.6) 2,180 (2,580)
60 20 0 20 ND (032) 7.2 (8.3) 2,280 (2,620)
70’ 30 0 0 83 9 11.7 2,620
60 30 10 0 36.7 11 2,860
60 30 0 10 12.6 11.5 2,480
50 30 10 10 50.2 11.6 2,540
65 30 0 5 49.0 11.6 2,540
60 40 0 0 103 11.2 2,620
50 40 10 0 153 11.2 2,800
50 40 0 10 34.0 10.9 2,720
40 40 10 10 61.4 113 2,580
55 40 5 0 102 12 2,680
50 50 0 0 ND 6.4 4,400
40 50 10 0 33.7 11.6 2,820
40 50 0 10 47.0 113 2,660
30 50 10 10 55.3 113 2,640
Coinposited sample from TP-4A, TP-5A, and TP-6A ND = Not detected
2 Mix percentages calculated on a diy weight basis ()Parcntheses indicate reanalysis
Sample tested at moisture content 5% above optimum

Addressing Data Comparability and Defensibility
Richard G. Mealy , Quality Assurance Manager,
Warzyn Inc., 1 Science Court, Madison, Wisconsin 53711
In the wake of federal investigations regarding fraudulent environmental testing, modifications made to
published method by individual laboratories is a rapidly growing cause for concern. The ability to achieve
Data Quality Objectives (DQO) for comparability is compromised when these modifications impact inter-
laboratory precision and accuracy.
Although the rationale for these modifications is comprehensible in specific situations, there is a need for
more formal mechanisms to control this process. In the absence of regulatory guidance and enforcement on
this issue, the ability to make correct environmental decisions is endangered.
The issue of method modifications is highlighted in order to:
• understand the rationale behind modifications
• determine the impact of modified methods on environmental decision-making
• make the distinction between modifications and deviations
• review techniques used to validate modified methods
• guard against adverse impacts of invalid modifications on projects
The environmental testing business is a rapidly expanding market with total annual revenues estimated
from 0.5 to 1.4 billion dollars. This boom has resulted in a population explosion of smaller laboratories,
each attempting to cash in on a piece of this lucrative market. This has, in turn, placed an undue burden on
both federal and state agencies responsible for regulatory oversight of environmental testing operations.
In addition, a recession economy, such as that endured over the last 12-18 months, serves only to further
augment the problem as regulatory agencies tend to conserve budget dollars and laboratories are forced to
increasingly Cut operational costs. The result of the former is that all activities, including enforcement, are
limited, while that of the latter is to continually seek “short-cuts” to all stages of the analytical testing
In the absence of adequate enforcement, method modifications can evolve into distinct deviations, resulting
in substantial changes to method accuracy and precision limits published in referenced methods. Thus far,
the emphasis of Quality Assurance (QA) has been on the validity of laboratory data, while the
comparability of data generated by one laboratory versus another has been largely overlooked. Upon closer
examination of the method modification issue, it is apparent that there is a need for additional guidance
from the regulatory community as well as a need to better educate our clients on this and other relevant

Modifications are made to published methodology for a wide range of reasons invalid. It is important to
emphasize that there are cases in which modifications are necessary and/or appropriate, such as cases in
• published performance criteria cannot be duplicated using similar apparatus and operating
• the intent is to combine multiple, similar methods into a single method
• the modification is technically sound and yields equivalent performance data
• the modification arises from interpretation of nebulous issues
All too often, however, modifications represent shortcuts to obtain a competitive edge within the laboratoly
community through decreased labor and/or materials costs. This type of modification usually involves the
softening or complete elimination of method requirements.
The number of testing labs has grown to more than 1600, many of which are collecting multiple state
certifications in order to expand their revenue base. In general the laboratory community recognizes that
individual regulatory agencies are simply too understaffed to serve as an effective monitoring body to
screen out those laboratories who have crossed the line between modification and deviation. In many cases,
even the resources that are available to serve in an enforcement role lack the level of training necessary to
provide an accurate assessment of a laboratory’s ability to consistently generate quality data.
This issue is not nearly as black and white as one would prefer. Data Quality Objectives vary considerable
between projects, and there will always be situations in which the use of even grossly modified analytical
methods would be acceptable, based on the needs of the prqjecL
Comparability, one of the five DQOs identified by the EPA, refers to the confidence with which one data
set can be compared to another. The goal of this aspect of data quality is to procure samples which are
representative of the site during the field sampling activities, subsamples for analysis which are
representative of the entire sample collected, and consistent, standard protocols for sample analysis and
reporting the data. All of this, however, is dependent upon the ability to achieve equivalent precision and
accuracy between testing laboratories.
Defensibility refers to the extent to which the data are able to withstand the scrutiny of the court. The
general rule is that the most reliable data are those data that are generated following published analytical
methods, and are further supported by extensive documentation that provides a complete audit trail of the
analytical process.
For the client, perhaps the single greatest fear is the liability incurred when decisions are made that later
prove to be based on invalid data. Erroneous decisions span the range between spending more than
necessary for remedial activity to making inaccurate health risk assessments. Complicating this issue even
further, under the current level of enforcement, invalid data is typically not discovered for months, or even
years, after the data is generated.

The most obvious example of the effects of modifications occurs in cases where multiple laboratories are
involved in testing for the same parameters at different locations within a large-scale site investigation.
Frequently, samples will be “split” between two laboratories as a quality control check on the data being
generated. Discrepancies between split samples are often related to minor differences in method protocols.
Even the most seemingly obscure modification to a method can produce significantly different results from
those generated using the standard technique. This makes it difficult to evaluate data, and re-
sampling/analysis is often required. The question of who should bear the burden of these expenses
becomes another issue. If it cannot be proved which results are in error, the client inevitably absorbs the
In the event that analytical data are subpoenaed as part of the litigation process, the possibility exists that
any modification can be raised as a factor which reduces the reliability of the data. If the reliability of the
data is in question, the data may be rejected from inclusion in the record, and any subsequent decisions
made based on these data be summarily dismissed as well. The intensity associated with legal depositions
and expert testimony may well present one of the best mechanisms to reveal the detail associated with
deviations from published method protocols.
From a strict quality assurance perspective, even discussions about making modifications to methods is
cause for concern. The grim reality, however, is that many of the published methods in use today have not
been updated (save for minor typographical revisions) since their original promulgation many years ago.
This remains the case despite the degree of technological advancement that has occurred over the same
span of time. The laboratory community (and hopefully the regulatory community as well) recognizes the
fact that few, if any, laboratories adhere to method protocols exactly as published. Consequently, we must
learn to live with the fact that modifications are a necessary part of environmental testing, and focus on
making the distinction between modifications and true deviations.
For the purposes of this discussion, a deviation is distinguished from a modification by virtue of meeting
any one of the following qualifications:
• the modified method results in significantly different results as compared to data generated using
the published method
• there is insufficient documentation to support the contention that the first statement is false
• the modification is not deemed valid by the governing regulatory agency
The first of these qualifiers is designed to evaluate the overall method in terms of the results produced. The
term “significantly different results” can be interpreted as changes to precision and accuracy, the ability to
achieve specific detection limits, the differential susceptibility to interferences, or even the interpretation of
the resultant data. An example of a modification which fails to meet this criterion is the analysis of
base/neutral and acid extractables using a single, neutral extraction. The ability to recover the acid
components would be minimal at best 1 which in turn affects precision, accuracy, and detection limits. The
second qualifier focuses on the documentation associated with the modification.

Using the example above, the value of performing a method validation study is to demonstrate the ability
to effectively extract the compounds of interest. An inexperienced analyst, making this modification
without the necessary documentation, may be unaware of the impact of the modification. The final
qualifier is included because of the number of differences between state and federal as well as inter-state
regulations. For example, the state of Wisconsin has recently enacted a suite of unique methods associated
with the analysis of petroleum-related conU mination, while many states remain receptive only to EPA
method 418.1 (freon extraction/infra-red [ IR] spectrophotomeiry). While GC fingerprinting techniques are
valuable in that they provide qualitative identification of petroleum product types in addition to semi-
quantitative estimates of concentration, these techniques are useless when regulatory action criteria are
based on analysis using the IR technique.
Table 1 represents a summary of analytical method modifications, some of which are routinely employed,
as well as the impact that they can have on the reliability and interpretation of results. It is important to
emphasize the relevance of project DQOs in any discussion of the validity of specific modifications. If the
purpose of the analytical testing is merely to provide field screening of samples to identify grossly
contaminated areas, then those modifications that result in low bias and higher detection limits have no
significance to the nature of the operation. Consequently, these modifications would be valid
considerations for this level of DQOs.
As with most issues of similar stature, the line between modification and deviation is gray at best. Only
after careful consideration of a number of factors can the distinction be determined. Unfortunately, there is
no set of rules to adhere to, merely guidelines to follow.
If it is determined that modification to a specific method is either advantageous or necessary, the first step
is to outline a plan to validate the technique and document the results obtained. While this process may
seem labor intensive, (considering the intent of the modification is most likely to impart a measure of
efficiency to the process) the quality and depth of the validation process can be vital if the data are
eventually involved in litigation. It is also recommended that the laboratory or chemist attempt to obtain an
opinion or approval from the regulatory agency involved. Resources such as the RCRA Hotline (800-424-
9346) or the Method Information and Communication Exchange (703-821-4789) should also be consulted.
The validation program should be designed to evaluate precision, accuracy and detection limits both in the
presence and absence of known or suspected interferences. In the first method modffic ttion listed in Table
I, the silica gel cleanup step is designed to eliminate co-extracted interferences from oils of a mineral or
vegetable nature. Consequently, if the modification were only validated without evaluating the effects of
interferences, method precision accuracy and detection limits would be equivalent to those appearing in the
published method. If this same validation was performed in the presence of these interferences, significant
bias would be observed.

In this case, if the validation process were sufficiently detailed, it is possible that a more specific process of
filtration through silica gel could be identified with an interferent removal efficiency equivalent to that of
the referenced method technique. The published method indicates that stirring 3.0 grams of silica gel in
100 mLs of freon for 5 minutes will effectively remove 100 milligrams of interfering hydrocarbons. An
experiment could be designed which would determine the mass of silica gel which could consistently
remove 100 milligrams of interfering hydrocarbons using only gravity filtration.
The determination of detection limits should be performed according to the EPA protocol outlined in
Appendix B to 40 CFR Part 136. The determination of accuracy and precision is typically performed by
spiking the parameters of interest at a level of approximately 10 times the nominal detection limit, but at a
concentration that falls within the range of the calibration standards without dilution. For this process,
most methods require the analysis of a minimum of 4 replicates, but the more replicates analyzed, the
greater the confidence that the resultant accuracy and precision estimates are representative of the analysis.
Finally, one of the most important aspects of documentation is to include a summary of the modifications
made to the published method within the standard operating procedure (SOP) for the method. This level of
written documentation serves primarily to indicate the willingness to identify the modifications, and
prompts the reviewer to evaluate (or inquire) whether or not the modification poses any conflicts to the
Data Quality Objectives of interest. If a regulatory agency spokesperson or one of the resources mentioned
previously was consulted, the advice received could be noted in this section as well.
Safeguarding against the consequences of significant deviations from referenced methodology is of interest
to both the client and the testing laboratories. There is a broad spectrum of quality when the entire
population of testing labs is considered. While there is certainly a niche for those labs whose operations are
not finnly founded in quality control and quality assurance (QCIQA), there must be mechanisms which will
ensure that these laboratories are viewed based on their ability to meet the DQOs rather than on the price of
analytical services alone.
The enforcement branches of the various regulatory agencies have been actively pursuing litigation of Jabs
performing fraudulent testing, yet the level of enforcement simply cannot manage the sheer numbers of
laboratories. There are also a number of schemes in use which allow a laboratory to perform testing in a
state without having to undergo a site audit. As an example, the state of South Carolina currently does not
perform site audits outside the state. Consequently, a laboratory can obtain certification in a state such as
Wisconsin, apply for reciprocal certification in the state of South Carolina, and perform work in that state
without the trouble of an audit. This type of loophole would be eliminated by the initiation of a national
accreditation program, which is currently under discussion by the EPA and several independent lobby
groups. Such a program, however, would need to provide a means to distinguish between laboratories
capable of producing only data of field screening quality and those capable of producing legally defensible

From the perspective of our clients, the main concern is that the laboratory they choose to contract with is
capable of meeting the DQOs of interest. Most clients cureently rely on the federal and state agencies to
provide an evaluation of laboratory capability - in the form of state certification. More intensive
educational programs are required to ensure that the clients are aware of the lack of detail associated with
the audit process (in most cases) in order that they may consider a supplemental audit program of their own.
Our clients must learn that good results from the analysis of the WS and WP performance evaluation (PE)
programs administered by the EPA is not in itself a testament of laboratory quality. Clients should be
encouraged to request and review SOPs associated with analytical methods. They should further be
encouraged to audit the laboratory to verify that the analytical and QC/QA procedures are performed as
they are written in the SOPs and the QC manual. All too frequently, after an SOP is written, analytical
training is passed from analyst to analyst, each generation providing a more dilute version of the SOP.
The role of the regulatory agencies must be to provide clear guidance regarding protocols. Efforts must be
made to combine similar methods or identify equivalent techniques, in order that laboratories are
discouraged from independently creating third generation methods. In addition, there is a need for a forum
to identify and rapidly resolve ambiguities in the published methods. These ambiguities are the source of
many unintentionally “modified” methods as laboratories attempt to interpret the method guidance.
Finally, there must be mechanisms to ensure that method performance criteria are achievable using the
apparatus and conditions described in the method. A common complaint among laboratories is that method
performance criteria cannot be duplicated, only to find upon thorough investigation that one or more
aspects of the method was altered between the time it was written and the validation process.
The issue of method modifications has far reaching consequences. The possible outcomes from the use of
modified methods can ultimately include rejection of analytical data, should the modification(s) cross the
line from modification to deviation. Unfortunately, this line cannot be drawn in black and white, which
requires careful evaluation of any modification before incorporation into methods used for regulatory
Most modifications result from interpretations of ambiguous regulatory guidance, intent to incorporate
advancements in technology, or attempts to perform analytical testing more efficiently in order to be more
competitive in the price arena. The environmental community must be able to rely on the enforcement ann
of the regulatory agencies to provide independent assessment of the suitability of these modifications. In
order to achieve this, there is a need for more formal guidance from the regulatory side, more technical
training for our clients, and better documentation from the laboratory community.
The need for modifications to published methods can never be eliminated, but we can generate a common
set of guidelines to be followed in the event that modifications are required. Controls can also be
established to ensure that only successfully validated modifications be employed, and that modified
methods be consistent with the DQOs of concern.

Table 1: Some possible method modifications and resultant impact on the data generated.
EPA 418.1
Modify removal of non-
petroleum hydrocarbons by
filtering extract through silica gel.
Insufficient contact time for
removal of these interferences.
Results may be biased high.
EPA 601/602
Decrease purge time from 11
minutes to 4-5 minutes.
Purge efficiency will be
proportional to volatility. Uttle or
no effect on gases, or lighter
volatiles, but higher molecular
weight compounds
(dichlorobenzenes, xylenes) may
have lower purge efficiencies.
This will result in low bias, higher
detection limits, and increased
TCLP Organics
Reporting results when an
MSIMSD is performed rather
than a single MS. Recovery
from MS = 10%, MSD= 90%.
Some labs will bias correct the
data for 10%, others for 90%, and
yet others based on an average
of the two recoveries (50%).
Depending on the specific
analyte, some laboratories may
report results as failing TCLP
criteria, while others will not.
Test methods for
Reactive Cyanide
( and
Reactive Sulfide
( Chapter 7,
Modification of purge gas flow
rate to improve analyte
recovery/method performance.
These methods are recognized for
typical recoveries in the 0-10%
range. While modifying the
method to improve performance
would seem logical, quantitation is
based on a specific flow rate.
Making this modification will
increase the potential for a given
waste to either pass or fail
reactivity criteria.
Trace metals-
digestion techniques
Modifying initial sample size and
final digestate volume to meet
state-specific (e.g. Michigan)
detection limits,
Beyond certain limits, the sample
size to acid volume ratio may
impact the digestion efficiency,
leading to low bias. In addition,
this modification, while effectively
concentrating the analyte, also
concentrates any interferences,
rendering typical matrix modifier
organic analyses
Substitution of capillary or
megabore columns for packed
In theory, capillary columns
provide improved resolution and
precision, and thus should
produce data at least equivalent
to data generated using capillary
columns. Some states may not
EPA method 418.1
(method 9073)
Perform extraction from solid
samples using 5 minute
sonication vs. 4 hour Soxhlet
Eliminates labor intensive step,
reducing cost of analysis. Results
may be biased low.
Organic methods
Allowance of one or more
compounds to fail OC criteria in
matrix spikes, or continuing
calibration standards.
Data for analytes failing to meet
method criteria may be associated
with high or low bias. Ability to
detect the analytes at specific
levels may be in question.

Ian Philli a , Senior Environmental Chemist, GEl Consultants, Inc., 1021.
Main Street Winchester, Massachusetts 01890
On—site audits of environmental testing laboratories have been found to be
an effective tool to assess the level of quality that a laboratory
routinely produces as well as evaluate whether a laboratory can meet
project or program data quality objectives. This paper will present the
procedures and considerations that should be employed when defining the
scope, contents, methodologies, and conclusions of an on—site laboratory
On—site laboratory audits involve the review of the systems a laboratory
employs to control the quality of the data reported. Laboratories
frequently evaluate the quality of their data through the measurements of
accuracy and precision. However, accuracy and precision alone are not
adequate to evaluate the quality of the data generated. An audit allows
the laboratory user to evaluate the quality of the data from the
perspectives of reproducibility, comparability, traceability, and
authenticity. Reproducibility and comparability are, theoretically,
controlled by the employment of standardized methodologies. However,
internally, individual laboratory employees may be performing test
procedures in significantly different ways. In addition, the procedures
and data acceptability criteria used by individuals may differ not only
between employees but from published procedures and/or laboratory Standard
Operating Procedures (SOPs). Traceability of the complete testing
procedure employed, including the standards used, is critical to the
future defensibility of data. Without the proper documentation, data may
be rendered useless in litigation proceedings. Finally, an audit of a
laboratory will aid in the overall evaluation of whether a laboratory is
capable and cosunitted to the production of data of sufficient quality to
meet the data quality objectives of individual projects or programs.
Environmental testing laboratories provide critical data to industry,
regulators, and environmental consultants. These data are used to make
decisions relative to the nature and extent of contamination, regulatory
compliance, risk assessment, and remedial alternative selection. Without
proper quality controls by the data user, including the establishment of
data quality objectives, the data generated may be found to be inadequate
or unusable. The use of on—site audits is a very effective quality
control technique to help assure that data ar. of sufficient quality to
meet the data quality objectives of a project or overall program.
To many end users of chemical testing data, the laboratory is a black box.
Sample. are sent and, three weeks later, a stack of paper comes back with

numbers printed next to a list of compounds of interest. The end user
then compares these numbers to a Maximum Concentration Level or a clean-up
goal, writes a report, and sends it off to the regulators. The question
remains, however, “ lB the quality of the data adequate for its end use?”
The US Environmental Protection Agency’s document Data Quality Ob-iectives
for Remedial Response Activities (EPA, 1987) defines five levels of
analytical quality for remedial investigations/ feasibility studies
(RI/FS). Levels IV and V are data of sufficient quality and documentation
that the data may, without additional support, be used for litigation
support, risk assessment, and regulatory compliance. Contract Laboratory
Program (CLP) data are an example of Levels IV and V. The test methods
and quality control (QC) procedures used to produce Level III data are
generally similar to Levels IV and V without the supporting data that
accompany a CLP package. Level III data are frequently combined with
other data generated during a project to evaluate overall conditions and
remedial options at a site. Levels I and II are for data generated in the
field for screening or health and safety purposes.
The Data Quality Objectives (DQOS) of a project or a program should be
defined and an appropriate level of analytical quality should be selected.
However, as has been become increasingly apparent, simply requesting a
level of quality does not assure that a laboratory will produce, or is
even capable of producing, at that level. Without QC measures, such as
laboratory audits, the data may be found to be found to be inadequate,
unusable, or fraudulent.
This paper will present the procedures and considerations that should be
employed when defining the scope, contents, methodologies, and conclusions
of an on—site laboratory audit.
There are three components to designing an effective laboratory audit:
o Define the audit objectives;
o Review the laboratory’s Quality Assurance Program Plan (QAPP); and
o Prepare an audit checklist which incorporates the audit objectives and
evaluates the laboratory’s quality control procedures.
As stated previously, the audit’s objectives may be specific to a project
or general to a wide range of programs. The scope of the audit should
reflect those objectives. For project specific audits, the auditor must
carefully review the end use of the analytical data and what decisions
will rely upon that data. In addition, project specific audits may allow
the auditor to focus on a smaller portion of a laboratory’s operations.
For example, if a project objective is to determine the extent of volatile
organics in the ground water at a site, a detailed audit of the inorganics
laboratory would serve little purpose. Program audits are broader in

scope and are generally uBed to evaluate a laboratory’s level of data
Once the objectives and scope of the audit have been defined, it is
critical to review a laboratory’s QAPP. The information obtained from the
QAPP will, be the basis of some of the key questions posed to laboratory
personnel during the audit. The critical points to identify in the QAPP
are: 1) the quality control procedures employed, 2) the data tracking and
review procedures, and 3) the data acceptance criteria.
The QAPP should state the quality control (QC) measures that are employed
and documented for the routine testing procedures performed. The QC
measures include holding times, preservatives, blanks, matrix spikes,
duplicates, blank spikes, and/or surrogates. The QAPP should also state
the frequency for each QC measurement. The frequency and QC acceptance
criteria should be reviewed during the audit.
The procedures used to track and review the data should be clearly
presented in the QAPP. It is important to determine if a laboratory
documents the traceability of standards, instrument maintenance,
instrument performance, and employee training. This information may be
critical in later demonstrating that the personnel and equipment were
operating under the proper conditions. In addition, the QAPP should
identify how and by whom all data are reviewed. The levels, criteria, and
frequency of the reviews should also be presented.
The documentation retained by a laboratory should be evaluated for
completeness, both on the project level and the overall laboratory level.
On the laboratory level, a laboratory’s QAPP should state the criteria by
which data are judged to be acceptable for release to the client. Por
example, data validatore have cLP’s Data Validation Guidelines to
determine whether data are acceptable. A laboratory should have similar
criteria established for all of the tests they perform. If this
information is not presented in the QAPP, the SOPs for each test should be
reviewed either prior to or during the audit. The absence of this
criteria indicate. that data acceptance is subject to each analyst’s
interpretation, and therefore, data generated may not be comparable or
In addition, if it ii availabl, in the QAPP, the qualifications of the
laboratory personnel should be reviewed.
After defining the scop. of the audit and reviewing the laboratory’s QAPP,
a checklist should be prepared to reflect the issues that are critical to
meeting the audit objectives. The checklist should be composed of
question. that: 1) Verify the information presented in the QAPP,
2) Determine whether the laboratory personnel are following the QAPP and
the SOP ., and 3) Evaluate the qualifications of the personnel performing
the testing. In addition, the checklist should contain the questions that
are included on a CLP audit checklist.

The audit involves the physical visit to the laboratory. The visit may be
on an announced or an unannounced basis. Unannounced audits have the
advantage of seeing a laboratory “under real conditions.” However, “real
conditions” may mean that key personnel may not be present or available.
Generally, an announced audit is best for evaluating the overall quality
of a laboratory. Unannounced audits are best used on a project—specific
basis to verify that the data quality objectives of that project are being
met. Unannounced audits are also extremely helpful in verifying that
special modifications to methods, if required, are being performed.
The audit will verify the information provided by the laboratory and help
evaluate whether the laboratory routinely performs testing of adequate
quality to meet the objectives of a project or program. The contents of
the audit will likely include a combination of the following:
o Observation of the facilities and equipment;
o Review of the laboratory files;
o Evaluation of the Standard Operating Procedures; and
o Personnel interviews.
Depending on the objectives of the audit, the contents of the audit may
focus more heavily on one or more of the subjects presented above.
In observing the facilities and equipment of a laboratory, the auditor
should consider the amount of work space available, the duplication of
instrumentation, the maintenance records of the equipment, and the overall
cleanliness and organization of the laboratory areas. The available space
and the duplication of instrumentation are particularly important as they
control, along with the experience of the personnel, the volume and
efficiency of the laboratory.
The quality assurance (QA), individual laboratory area, and the project
files should be reviewed during an audit. The QA files should be reviewed
to verify that the quality control measures specified in the laboratory’s
QAPP are being performed and meeting the laboratory’s criteria. For
example, any control charts, method detection limit studies, corrective
actions and anomalies should be reviewed. In addition, many QAPPS state
that reports to management are made. The auditor should request to review
these reports as they will provide an internal picture of the laboratory’s
cos itment to its own QA program.
The files from each laboratory area should be reviewed in order to
evaluate the traceability of the data generated. Each laboratory area
should have, at a minimum, files tracing the preparation and sources of
all standards and spiking solutions, the daily temperature logs of all
sample storage refrigerators, and balance calibration logs.

The project files should be reviewed for their contents and completeness.
Project files should contain all of the information relevant to the
testing performed. The relevant information should include copies of the
chain of custody, the final data report, the sample preparation data, and
any instrument output. Project files may also contain calibration and
method blank information. If calibration i . not included in the file, the
location of this information should be referenced.
An effective method to evaluate the files is to select a final sample
extract and request all of the documentation used to generate the final
report for that sample. The laboratory should be able to generate the
information referenced above, the standards used to generate the
calibration, the stocks or neat material used to make those standards, and
any other information used by the laboratory to meet their data acceptance
criteria. All data should be traceable back to the original standards.
If the quality of the standarc s cannot be confirmed, the basis for all
quantitation ii undermined.
The details of the methods performed in each laboratory should be
documented and available for review. During the audit the presence of the
SOPs should be verified, and laboratory personnel ahould be questioned
about their contents. The use of SOPs is a critical component of a
laboratory’s ability to generate reproducible and comparable data. If
possible, more than one employee from each area should be asked similar
questions concerning the SOP for that area.
Throughout all of the steps of the audit, the auditor is interviewing
laboratory personnel in an attempt to evaluate their knowledge and
experience, their understanding of the SOP. and the QAP?, their
understanding of the methods, and their coamkitment to the quality of the
data. The nature of their responses gives a good indication of the DQO
level of th. data routinely generated.
As stated previously, the personnel from each laboratory area ahould be
asked about the contents of the SOP. In addition, where applicable,
personnel should be asked what makes the data acceptable. If different
answers are received from two employees in the same area, then the SOP
and/or QAPP is not being followed by at least one of the employees and
perhaps both.
Personnel should also be asked hypothetical questions du. to the data
interpretation probl.ma often presented by .nvironmental samples. While
no right answer may be available, the auditor will get an excellent
indication of how decisions are mad• and the level of experience of the
personnel. Thu type of question is particularly important for tests that
involve significant analyst interpretation, such as pesticide and PCSs as
well as many inorganic tests.

In evaluating the overall quality of the laboratory, an auditor should
review the contents of the completed checklist with the DQO of the project
or program in mind. Based on the audit, the auditor should be able to
answer the following questions:
o At what EPA DQO level is the laboratory routinely performing?
o Is the quality of the data traceable, well documented, and defensible?
o Do the personnel have the right types of experience for the project or
o Can they meet the data quality objectives of the project or program?
Laboratory audits can be a very powerful quality control tool to evaluate
whether the data are of sufficient quality to meet the objectives of a
project or program. Auditing can help prevent the data generated from
being found inadequate or unusable.
U.S. Environmental Protection Agency (1987). Data Quality Objectives for
Remedial Response Activities, EPA 540/G—87/003, March.
U.S. Environmental Protection Agency (1988). Laboratory Data Validation
Functional Guidelines for Evaluating Organics Analyses, February.
U.S. Environmental Protection Agency (1989). Preparing Perfect Project
Plans, EPA/600/9—89/087, October.

F. Joseph Unangst, Vice President, Galson Laboratories, A
Division of Galson Corporation, 6601 Kirkville Road, East
Syracuse, New York 13057
Abstract: Environmental service laboratories are under extreme
pressure to improve performance, but many forces militate
against such improvements. For example, the industry
requirements for holding times and report turnaround, marked
with excessive penalties, place managers in a position that
threatens them to compromise ethical practices. Seasonal
fluctuations in workload and increased competition provide an
unstable pricing scheme. Multiple certification programs and
regulatory requirements make standards variable and often
confusing. Knowing all this, environmental professionals face
special challenges in choosing a laboratory that invokes
One way to facilitate such a choice is to investigate a
laboratory whose principles include Total Quality Management.
Total quality begins with sound ethical practices. Ethics are
defined through guiding principles that can be used to
evaluate the laboratory’s performance. Next, management
personnel must commit to the Total Quality Management (TQM)
process. The “cost of quality,” as well as its returns, must
be demonstrated to top management to get this commitment.
A TQM program begins with organizing a quality improvement
team. This team works to design and implement the program. The
program typically consists of quality awareness training,
supervisor training, quality teams, and employee recognition.
The process requires a new behavior that stresses constant
improvement, ends the dependence on mass inspection,
continually improves service and production, emphasizes
training, replaces management with leadership, and elicits
total employee involvement.
The Total Quality Management system works in the environmental
laboratory industry just as it demonstrably has in numerous
other industries. It may be the only philosophy that will
assure a company of succeeding in this difficult business. The
stresses encountered become opportunities to improve. This
paper describes the application of a quality improvement
process to a service laboratory. The paper outlines how
changing a way of doing business is the only way to remain
competitive and cost—efficient.

The environmental laboratory industry is under extreme-
pressures to provide service to its customers. The industry is
subject to a significant number of competing demands that may
have caused a few laboratories to compromise their integrity.
The most pronounced pressures include:
• Sample Holding Time. This is undoubtedly the most
pressing time constraint, where failure to meet the
holding times results in suspect or lost data, with
potential resampling required.
• Report Turnaround Time. This is a frequently
neglected yet critical customer requirement.
• Regulatory Requirements. These include data report
formats that often change, variable QC requirements
that depend on which regulation the samples fall
under, and variable QA requirements that depend on
both regulations and specific customer
• Workload Fluctuations. Included in this are
decreases in work due to the winter in several
regions, the loss of work to competition, or the
termination of periodic contracts. Also, work may
increase dramatically because of any of the
following factors: project delays that later cause
project overlap, the use of sales persons who are
paid on a commission basis, management overbooking
work, and uncontrolled change of project scope.
• Competition. Increases in competition occur during
slow winter periods, when large laboratories
attempt to use price to maintain base workloads,
and during recessionary times.
• Staffing. The environmental laboratory business is
a young industry with typically younger management
and staff with less than 10 years’ experience. The
rapid growth of the business has caused a shortage
of people trained in performing the work. Younger
people are usually more mobile and more difficult
to retain.
• Penalties. The penalties associated with missed
holding times or late reports have the unfortunate
side effect of potentially undermining ethical

practices, which has been widely publicized in our
industry. The penalties may be tied to lateness of
the final report, which promotes poor report
These demands placed on the environmental laboratory make it
an excellent proving ground for Total Quality Management. The
reasons for implementing a TQM process are very clear. It is
very profitable due to the ability to demand higher prices for
quality services in a free market and because it is less
expensive to do things right the first time. It also creates
and preserves jobs through increasing business and making the
laboratory financially stronger. Lastly, it’s the right thing
to do — the customer deserves this as a covenant of ethical
business practice.
The TQM process begins with the highest level of management
being committed to sound operational principles, or guiding
Kanagement Commitment. The TQM process cannot be delegated
from the top. This is not something the CEO or President can
give to the Quality Assurance group to implement. If the
person at the top echelon in your organization is not involved
and committed to providing quality laboratory services and you
are not one of those two people, do not waste your time
attempting to implement total quality. It will not work. It is
capable of working in somewhat autonomous divisions or groups,
as long as the head of the group or division is committed and
not pressured to do otherwise by the next higher level of
management. Otherwise, OU must abandon any efforts until you
get the commitment from the top.
This commitment can often be attained by presenting to top
management the cost of not implementing a TQN program.
According to a 1989 American Society of Quality Control
survey, this cost averages 35 percent of gross sales lost in
the service industry.
Individuals in top management must understand this is a
business proposal with unequaled return on investment. Their
commitment must be visible, relentless, and educated. They
must know and accept that the quality process will involve a
change in corporate culture.
Quality Service Principles. Roy Disney once said “decisions
are easy when values are clear.” When every employee

understands the principles under which the laboratory
operates, they become capable of making decisions on their
own. In developing guiding principles, the people in the
laboratory gain the responsibility to perform according to the
principles. However, they must also be given the authority to
do so.
As an example of responsibility being equal to authority, in
a laboratory where the customer service policy states a
service objective of a satisfied customer at the end of every
transaction, management must allow the customer service person
to take any step necessary to solve customer problems, arrange
a quick satisfactory resolution, or a prompt refund or credit.
The principles must be clearly understood and accepted by
The principles should stress the basic fundamentals of total
quality service: focus on the customer, put quality first,
show continuous improvement and innovation, consider people
our greatest asset, encourage total employee participation,
foster training and education, coiamunicate clearly,
demonstrate integrity, employ teamwork, and be profitable.
These principles set the rules for becoming a world—class
quality operation. The principles should be reviewed
periodically to determine their continued relevance, allowing
for amending them, deleting some, or supplementing them, as
circumstances warrant.
The Mission. When the principles are in place, the laboratory
can embark on its mission. This is an area that requires a lot
of soul—searching to determine what you want to be as a
laboratory. For example, you may want to become the largest
laboratory network in the United States, you may want to be
most renowned for quality, you may want to be the most
expeditious, you may want to be the best RCRA laboratory. You
must define what you are in clear and precise terms so that
everyone in the laboratory unmistakably understands the
If the mission statement will not fit on a business card, it
is too long. Everyone must be able to recite and accept the
mission as his or her own.
The mission and principles can be developed through committees
made up of employees from all levels and divisions. The
principles and mission lay the groundwork for implementing a
Total Quality Management.

Implementing a TQN laboratory system is not as difficult, as
expensive, or as slow as many consultants or opponents may
lead you to believe. The initial step is to develop a Quality
Improvement Team (or any other name you may want to call it.)
The make-up of this committee must include high-level managers
from all divisions, or sections, who are committed to the
process. Again, this is not something that can be delegated.
This committee will have been involved in setting the
groundwork through its involvement in developing the
principles and mission.
These team members must fill themselves with knowledge of the
quality processes. They must educate themselves on the
available resources for accomplishing the goal of integrating
the new philosophy into the business culture. Much of the
knowledge can be learned quickly through networking with
others who have implemented similar systems. Consultants are
an expensive alternative that should be used only if the
initial level of understanding is minimal or if urgent
implementation is required.
The Quality Improvement Team determines how to implement the
quality concepts. The program has these basic components:
total participation (including top management), leadership
training, making authority equal to responsibility, quality
system structure, communication, training, measurement, and
recognition. More components may be necessary, depending on
the individual circumstances.
The implementation should not be a volunteer program; everyone
must be involved in a quality team. In order to get everyone
involved, a series of steps should be taken to allow rapid,
effective implementation. If the following approach is taken,
the benefits of implementation will be realized almost
Communication to all employees is vital during initial and all
subsequent stages of TQM implementation. There will be a lot
of ambitious employees who will want everything to be in place
immediately. The implementation team must manage the emotions
of the cultural change by keeping everyone informed. The team
must tell people the steps to be taken and the approximate
time to implement them (this time frame should be realistic,
allowing extra time so that people do not get discouraged).

The initial presentation to the company is critical and
demands special care and excellence. Explain the move toward
quality; prove that quality will preserve jobs through
improved profits and increased market share. It must be
stressed that a quality process is not being undertaken to fix
the errors the nonmanagement people have been causing in the
past. Management must take responsibility for past errors and
realize people have known how to do things right all along;
management just hasn’t allowed them to fix it. Finish with a
statement that describes the quality process as an opportunity
for the laboratory to improve together as a team.
Whatever you do, keep the plan simple. The entire
implementation can be accomplished using just four steps. The
concepts are simple and the approach is simple; keep it that
Step One: Provide Quality Awareness and Leadership Training to
Top Management and Supervisors. The people most affected by
the new cultural change will be the supervisors. The
supervisors will be required to change the way they have done
business to date. They will be asked to make decisions based
on input from everyone who works for them rather than being
the person who makes all the decisions. This is the most
difficult cultural change. Supervisors who have worked for
years as nonsupervisors under the old system of being told
what to do are now not allowed to tell their subordinates what
to do. In their eyes, they have lost out on a good thing. The
fact is it wasn’t a good thing when their supervisors did not
listen to them and they will now be part of the new culture of
participative management.
Because of this significant change in responsibility,
supervisors must be a vital part of the process of
implementation from the very beginning. We must remember that
the supervisors are usually the most experienced people, they
must be protected, and nurtured into the new system. A
laboratory cannot afford to lose these people. This radical
change may frighten supervisors; consequently, their
involvement will help to allay those fears.
The quality awareness training can be as little as four hours
or as many as twelve hours. In the environmental laboratory
business, most people are well educated and already familiar
with the scientific method, which is a significant portion of
the awareness training. The awareness training should touch on
the various tools of problem solving, the quality improvement
process, statistical process control, and the history and
success of TQM. Awareness should also stress that this is the

new way of doing business; everyone will be responsible for
making it happen.
The supervisors and top management will also require
leadership training. Since everyone in the company will be
required to be on a quality team, team leaders will be needed.
The team leader role is most easily and best filled by current
supervisors. In order to prepare the supervisors for their new
roles as team leaders, a training session on how to conduct
participatory meetings will be required. This can be done out
of house (check your local community college f or an economical
alternative to consultants) or developed as an in—house
The awareness training for supervisors will teach them in—
depth, as team leaders, to take their teams through the
systematic problem—solving process. This includes analysis of
problems, brainstorming, reaching consensus on solutions,
planning and implementing solutions, and evaluating results.
By teaching supervisors in-depth techniques for quality
improvement through teams, they will be able to disseminate
the process to all people in the lab.
The leadership training is extremely important for both top
management and supervisors to assure the long—term success of
the laboratory as well as their own personal success.
Leadership training shows the new team leaders they must move
from “management,” where people are controlled, to
“leadership,” where people are motivated and inspired to
participate. The managers must learn to respect their
employees, listen, solicit ideas, and realize everyone can
think as an adult. The managers must trust employees in order
to gain their trust.
Once the proper training for team leaders and top management
is in place, begin to train everyone in quality awareness.
Step Two: Quality Awareness Training for Everyone. If the
leadership training has been successful, nonsupervisory
employees will have realized some differences in their
managers and will be very eager to participate in the
training. Many will begin to orient themselves to the quality
process once they realize management is serious about
implementing it.
The quality awareness training for nonsupervisors can be as
little as four hours per employee if sufficient materials are
available for them to refer to when they begin functioning in
their team.

Awareness training will be similar to (but less detailed than)
the supervisors’ awareness training. This is possible if you
are using your previous supervisors as team leaders. The team
leaders will use team meetings to do more in—depth training on
problem solving with real laboratory situations.
The approach of in-depth problem-solving training by the team
leaders provides an excellent, cost—effective way to train a
large number of individuals while gaining benefit from their
Step Three: Assign Everyone to a Quality Team. The teams
should have eight to twelve people. Everyone is required to be
on a team, even the CEO. There may be more supervisors than
you need team leaders. That will be all right; use extra
supervisors as facilitators of slightly larger teams (maybe 13
or 14 people). Everyone on the team at first should have a
common interest in the assignment (e.g., for GC/MS volatiles,
make the whole group a team). This is the least disruptive way
to implement the process.
Teams will realize they need input from other areas to solve
their problems (e.g., sample management, sales or report
generation for the GC/NS volatiles team) and people will trade
positions on teams in order to get the diverse input they
need. The composition of teams will develop over time into
multidisciplinary, multilevel units.
The teams are now given the responsibility to improve the
operation of the group by use of the quality improvement
process. Since they are given the responsibility, the rules of
the quality system provide them the authority to make the
changes to improve the processes. Decisions of teams should be
presented to management if the team feels it is making a
recommendation to expend resources or money beyond its level
of responsibility. Typically, in this case, the financial
officer has already been brought into the team for cost
Keep in mind the entire quality improvement process and all of
its tools are not needed in every case. If someone suggests we
stop doing something that is obviously serving no purpose, and
nobody that still works here knows why we are doing this, then
just stop doing it. Typically, these things we were doing that
served no purpose will have no identifiable customer. There is
no need in these cases to use the six—step problem—solving
process; in some other cases, an abbreviated problem—solving
process may be used.

The quality team meetings should be totally focused on
quality. The team should have a top priority for everyone, and
the decisions that result from the teams must affect the
members. At the meetings, promote ideas that save time or
money or that reduce stress.
Benchmarking is an important tool for problem solving.
Benchmarking can be done elaborately, which will pinpoint
where the laboratory is with respect to a problem or service.
On the other hand, benchmarking can be done less formally, as
a communicative tool that obtains information from your peers
and competition on how to solve problems or improve
performance. People are often generous in sharing data (this
means you also must share some insight). This can be a very
quick and inexpensive problem—solving tool and should become
a mandatory part of quality improvement teams.
Btep Four: Bupport the Program. The ultimate support of the
program has to come from the top. The ways in which the
program is supported include recognition, maintaining a high-
level overseer, communication, measurement, and making it
permanent through a documented quality system. The highest
levels of management must be involved with the support of the
Recognition can take many forms, formally and informally,
publicly and privately. Forms of recognition can be simple
thank-you’s, buying having doughnuts, gift certificates,
providing plaques, parties, monetary rewards. The system of
recognition needs to be imaginative and sincere. The
recognition, reward, and gratitude must be ongoing and reach
the person in a way that he or she appreciates. Different
people have different ways in which they feel appreciated, so
it is best to use many different methods or to ask the
individual what they like. The recognition or reward should
not be an incentive; it should be a thank you.
The support of the team process should be overseen by a
quality chief, the highest level person possible. This person
is solely responsible for making the quality process work. The
person needs to understand the quality concepts and assist the
team leaders. The chief is required to monitor the progress of
the teams and report to the President or CEO.
Communication is one of the guiding principles and should be
used to support the quality team concept. All employees should
be aware of the progress of the teams’ and laboratory’s
performance based on some simple, consistent indicators. An
internal newsletter, a “quality” bulletin board, a “quality”

voice—mail box (for all to listen to), or periodic whole-lab
meetings can be used us avenues of TQM communication. These
allow the teams to share ideas and to be reinforced that the
process is working.
Measurement supports the process. But only measure things that
will help generate more ideas or show progress. Initially look
at the number of ideas generated and implemented and the
dollar savings of the implemented ideas. You may also look at
the time savings generated. The quality chief should be
ultimately responsible for communicating the measurement
results to everyone.
The most important step in maintaining the program is to
document the quality system. This documentation (the Quality
Manual) should state your policies with respect to quality and
customer service as well as your mission and principles. It
should clearly state the connections between the quality
program, the quality teams, the quality assurance functions,
the quality control functions, quality training, and
responsibilities for quality. This document becomes the
standard operating procedure of the TQM process. It is used to
audit your performance with respect to the quality process and
to orient new employees to the process.
A Final Note on Implementation: Keep it simple, keep it
exciting, keep reviewing the process for improvement, and keep
in mind you will all make mistakes in the transition. Say
you’re sorry when you make a mistake, accept other’s
apologies. If you’re upset with someone, allow them the
benefit of the doubt and believe they made a mistake. Ask
them. Communicate.
Total Quality Management will work to help control the
competing demands faced by the environmental laboratory
business. The answers to the problems will come from within,
from the people who perform the work day in and day out.
TQM will require a cultural change that must originate from a
highly committed and actively supportive top-level management
team. From this process will come guiding principles that
address all of the concepts required for a totally
, customer—focused, improvement—oriented, highly
motivated team of employees.
The implementation of TQM can be very simple and inexpensive,
wherein benefits are realized almost immediately. Getting

supervisors involved in the implementation is critical for
success. Assuring 100 percent participation throughout the
laboratory is also a vital component.
After initial planning, implementation involves these steps:
Step One: Provide Quality Awareness and Leadership
Training to Top Management and Supervisors.
Step Two: Quality Awareness Training for Everyone.
Step Three: Assign Everyone to a Quality Team.
Step Four: Support the Program.
TQM is a dynamic process that will remain a part of the
laboratory forever. As long as the process is communicated
well to everyone and the team leaders are held accountable for
the progress of their teams, the results will be phenomenal.
The new business arena is demanding total quality. To simply
survive in this demanding business will require the use of
TQM. To master the philosophy will earn the laboratory
competitive dominance.

David N. Speis , Vice President and Director, Quality Assurance and
Technology, ETC Corp. 284 Raritan Center Parkway, Edison, NJ 08818-7808;
Marilyn P. Hoyt, Principal Technical Consultant, ENSR Consulting and
Engineering, 35 Nagog Park, Acton, MA 01720
ABSTRACT : Characterizing non-homogeneous industrial scrap originating
from recycling operations for chemical compounds of environmental concern
typically poses many difficult field and laboratory challenges. Foremost
among these are defining the entity to be characterized, developing
representative sampling techniques, and selecting appropriate analytical
These challenges were the central issues of a recent dispute between the
United States Environmental Protection Agency and a scrap metal recycling
firm. The firm’s business activities involved the purchase of obsolete
electronic components and scrapped computers from the primary equipment
manufacturer. The purchased material was shredded, incinerated to remove
organic material, and smelted to reclaim gold and other precious metals.
An investigation by EPA into the environmental implications of the
operation indicated the presence of high concentrations of polychlorinated
biphenyls (PCB) - The company disputed the findings of the PCB
investigation, citing the sampling and analytical techniques used for the
investigation as non-representative and subject to false positives caused
by interferences. In the firm’s estimation, these procedures provided a
distorted view of the PCB content of the various industrial scrap piles.
At the request of the company and with the agreement of EPA, a unique
field sampling and analytical scheme was designed to address these issues.
Specific quality control procedures were employed during sampling which
enabled the investigators to determine field precision and accuracy.
Routine laboratory QC procedures enabled the investigators to assess
laboratory precision and accuracy. Overall method performance was
evaluated through the use of performance spikes that were also analyzed
by an EPA selected referee laboratory using traditional analytical
The use of these procedures provided data that enabled the company and the
government to obtain an accurate and representative assessment of the PCB
content of the scrap. The data was then used to make confident decisions
on the ultimate fate of the scrap material.

INTRODUCIION AND BACKGROUND : Appropriate field sampling and laboratory
procedures are key to providing representative analytical data on chemical
compounds of environmental concern in non-homogeneous solid matrices.
These procedures were the central issues in determining if polychiorinated
biphenyls (PcB) in electronic scrap posed a hazard to human health and
the environment as defined under terms of the Toxic Substances Control Act
A scrap metal recycling firm had purchased obsolete electronic components
and scrapped computers from primary manufacturers for precious metal
reclamation. This discarded material was transported to a shredding and
storage facility owned by the firm for processing.
The firms shredding and storage facility occupies a 6.5 acre site adjacent
to a tidal river. All scrap material arriving at the site was shredded
using an industrial automobile shredder and stored on site in discrete
piles, organized according to source, until its precious metal content
could be determined. Once an assay of the shred had been completed, the
material was incinerated to remove organic material. The resultant ash
was transported to a remote smelter for the reclamation of gold and other
precious metals.
In December 1986, the USEPA conducted a field inspection of the facility.
Samples were collected and analyzed using CLP methodology for pesticides
and PCBs. Non-representative grab sampling procedures were employed for
sample collection. The analysis performed by the USEPA indicated that PCB
concentrations in the scrap exceeded the 50 parts per million (PPM)
standard established by USEPA regulations. At the time the field
inspection was initiated, there were twelve (12) piles of shredded
material on the site ranging in size from 100 tons to 1500 tons.
The results of the USEPA analysis were not accepted by the metal recycling
firm which contended that the scrap was not representatively sampled and
the P B values were biased by positive interferences. A consent decree
regarding the dispute was eventually negotiated between the firm and the
USEPA. As part of the consent decree agreement, the firm agreed to
perform a remedial investigation which was designed to overcome the
sampling and analysis limitations of the initial USEPA study.
The overall objective of the investigation was to determine the PCB
content of the twelve (12) electronic scrap piles and the underlying
ground cover (consisting primarily of scrap pile residues and some soil)
at the site. The analytical objectives of the investigation were to 1)
develop and apply a scheme for representatively sampling the non-
homogeneous electronic scrap and 2) determine the PCB content of the scrap
using GC/MS methoda which would not be affected by positive interferences.
The data from the investigation was to provide the basis for determining
the remedial solution to be applied to the electronic scrap as dictated
by the terms of the consent agreement.

essential to the planning process for this investigation. The remedial
investigation objective focused on obtaining an accurate PCB value for
each scrap pile which was representative of its true PCB content. This
value would be used to determine if specific scrap piles exceeded the
5oppm TSCA limit. All field sampling, sample analysis, data collection,
reporting and validation activities were designed to be consistent with
this objective.
Several sampling and analysis concerns were identified by EPA and the
recycling firm which would impact the ability to achieve these objectives:
* The scrap piles ranged in weight from 100 to 1,500 tons.
Development of an adequate, representative sub-sampling approach
for this size scrap pile was required.
* The scrap material consisted of particle sizes which ranged from
dust sized organic polymers particles to grapefruit sized chunks
of steel. Procedures had to be identified which assured
homogeneity of laboratory sample aliquots.
* Effective cleanup procedures had to be identified and
incorporated into the sample preparation scheme which eliminated
interferences from polymeric material likely to be co-extracted
along with PCB5.
* Previous PCB analysis used pattern recognition techniques for
qualitative and quantitative identification. This may have
resulted in elevated concentrations because of interferences
calculated into the final results. Confirmatory techniques had
to be incorporated into the final analysis to ensure that false
positives did not occur.
Specific control parameters were incorporated into the project plan that
would enable EPA and the recycling firm to determine if the sampling and
analysis objectives had been achieved and the data was satisfactory for
addressing the objectives of the remedial investigation.
FIELD SAMPLING : The sampling procedures used for this investigation were
designed to obtain a representative sample of each scrap pile which could
be analyzed to determine the true PCB content and distribution. The
sample protocol used for the electronic scrap piles was based on a
hexagonal sampling design which has been developed by the USEPA for PCB
site cleanup 1 . The objective of this design is to obtain a representative
characterization of the scrap pile by collecting and analyzing a large
number of samples. A hexagonal grid is projected Onto a sampling circle
which is centered on the scrap pile, has a radius which extends to the
pile boundary and encompasses the entire pile. A site map (Figure 1)
depicts the position of each scrap pile.

For this project, each scrap pile was flattened to a depth of four (4)
feet. The diameter of the scrap pile was measured and a scaled diagram
of the pile was plotted on graph paper. The actual radius of the sampling
circle which enccxnpassed the scrap pile was measured. The number of
samples to be collected from the sampling grid was determined based on the
radius of the scrap pile as indicated in Table 1.
Table 1
Scrap Pile Radius, r(ft) Number of Samples
>4-li 19
>11 37
The radius of all flattened scrap piles exceeded eleven feet in this
investigation. A sample grid for the scrap pile was constructed. The
center point of the grid is the center of the sampling circle. The
distance between sampling points becomes a function of the circle radius.
The distance between sampling rows is also a function of the circle
radius. These distances were determined using Table 2.
Table 2
Distance Between Distance Between
Number of Samples Ad-i acent Points Adjacent Rows
7 O.87r O.75r
19 O.48r O.42r
37 O.30r O.26r
The sampling points were initially plotted on the scrap pile diagram
before being transferred onto the scrap pile. Each grid location was
marked with a wooden stake and labeled with a unique sample point code.
The sample points for each row were staggered midway between the points
of the adjacent row. Core samples were collected at each of the
designated points on the sampling grid using a power auger and composited
in pre-cleaned 55 gallon drums.
The power auger and shovel were steam cleaned prior to collecting any
samples and between different scrap piles. The condensate from the
equipment steam cleaning step was collected in separate sample bottles
and extracted for PcBs. The condensate extracts were stored at 4°C until
the PcB content of the previous sample and the next sample had been
determined. If the PCB value for the previous sample contained PCBs and
the next sample contained PcBs at concentrations >5Oppm, the condensate
extract was analyzed to determine if PCB cross contamination occurred
during the sample collection step.

SAMPLE PROCESSING AND HOMOGENIZATION : The composited sample borings from
each scrap pile were processed through a Hammer Mill and a Rotary Shear
for homogenization and particle size reduction using continuous f low-
trough processing procedures. This process was performed separately for
each pile to ensure that its integrity was maintained during this process.
The collected samples from all grid locations in a specific scrap pile
were composited and weighed. These composited samples were processed
through the Hammer Mill for further size reduction (approximately 2000 lbs
of material).
Occasionally, large un-millable metal chunks were found in the shredded
scrap sample. These chunks were manually removed from the granulated scrap
sample prior to Hammer Mill processing. The weight of this material and
its PCB content was accounted for in the final PCB concentration
The un-millable scrap was placed into a previously weighed, wide mouth,
2.5 gallon glass jars. The jars containing the scrap were re-weighed and
the weight recorded. These jars were transferred to the laboratory for
extraction and analysis.
The entire sample from the Hammer Mill was continuously transferred to a
Rotary Shear for final particle size reduction. A ten percent sub-sample
of the granulated material was withdrawn from the Rotary Shear using a
built-in continuous sampler. This sub-sample represented ten percent of
the original sample (approximately 200 lbs.). The scrap sample at this
point was reduced to small particles with a maximum particle size of 1/4
inch which had a consistency resembling potting soil.
The sub-sample was coned and split into two equal portions. Each sub-
section was coned and quartered until the remaining sample was sufficient
to fill two eight ounce jars. Duplicate sub-samples were collected to
determine the precision of the homogenization technique.
Residual contamination of the shredding and milling equipment was possible
during the sample homogenization process. Conventional decontamination
procedures could not be employed between samples for this equipment
because of equipment’s size and the difficulty accessing the contact
To decontaminate the milling and granulating equipment between processing
of each sample pile, all sample material was removed from the Hammer Mill
and approximately two mill volumes (7-8 cubic feet/volume) of previously
analyzed cleansing sand was processed through the machine. A ten percent
aliquot of the processed cleansing sand was collected using the continuous
sampler. Eight (8) ounce sampling jars were filled with aliquots of the
cleansing sand. Each sample aliquot was labeled and retained for PCB
analysis to verify the absence of cross contamination between processing
episodes if needed.

FIRLD QUALITY CONTROL : A field sampling quality control program was
employed to insure that all field samples collected for analysis
adequately represented each scrap pile. Table 3 defines each of the field
QC checks which was employed for this investigation. The duplicate
aliquot precision criteria is detailed in Table 4.
Table 3
Field Quality Control Checks
Field Blanks: P B free reagent water was provided to the field
technicians by the laboratory and was transferred to an additional clean
sample container while at the field sampling location. Field blanks were
used to evaluate environmental and procedural effects of the sampling
event and to determine if cross contamination occurred during sampling.
These blanks were analyzed for PCBs only if PCB cross contamination was
suspected to have occurred during the sampling event.
Equipment Rinse Condensates: Prior to collecting samples from any scrap
pile, the field sampling shovel and the auger were steam cleaned. After
sampling each scrap pile, the shovel and auger were steam cleaned again.
The condensate from the steam cleaning was collected and analyzed for PCBs
to ensure that cross contamination of equipment did not occur between
scrap pile samplings.
Field Duplicates: Duplicate samples were collected from all scrap piles
by rotating the 37-point sampling grid to permit sampling of points
adjacent to those points used for the first collection. The sample
collection process was repeated on the adjacent points. These samples were
homogenized, analyzed and evaluated to determine the precision of the
sampling procedure.
Duplicate Aliquots: Duplicate aliquots of each scrap pile sample were
collected in the field following the coning and quartering step. The
duplicate aliquots were analyzed as separate samples and evaluated to
determine the precision of the homogenization procedure.
Table 4
Precision Criteria for Field Quality Parameters
Dtçltcate Ahquot Precision:
Par neter Relative % Difference
Total X PCBs (Aroctors) >4oppm 50
Total X PCBs (Aroctors) 10 —4Oppm 1O0
Total X PCBs (Aroctors) <10 ppm Detected*
Field DLpticate Precision:
Parameter Relative % Difference
Total X PCBs (Aroctors) >4oppn 150
Total )( PCBs (Aroclors) 10 —4Oppn 100
Total X PCBs (Aroctors) <10 ppm Detected*
* Detected PCBs appLy to concentrations beLow the quantitation Limit that are capable of
being quaLitativeLy identified. Less accurate quantitative values can be assigned to these
PCBs based on extrapolation of the calibration curve. Detected PCBs nust be present in
each s ipLe of the i Licate pair.

ANALYTICAL METHODOLOGY : The analytical method used for this investigation
is based on USEPA Method 680.2 The method provides procedures for the
determination of polychiorinated biphenyls in electronic scrap samples by
gas chromatography/mass spectrometry (GC/MS) using full mass range
scanning or selected ion monitoring (SIN) for additional sensitivity. The
method is applicable to PCBs that occur as Aroclor mixtures or as single
congeners. Identified PCB5 were measured and reported as the commercial
Aroclor mixtures 1242 or 1254 when sample congener patterns match those
established for these two specific products. PCB concentrations were also
identified by isomer group depending upon the level of chlorination.
A 100 grain aliquot of homogenized electronic scrap was soxhlet extracted
using hexane:acetone. An aliquot of the solvent extract was concentrated
to lOml. The sample extract was washed repeatedly with Sulfuric Acid to
remove non-aromatic hydrocarbon interferences. PCB mixtures and other
extract components were separated by capillary column gas chromatography
and identified by low resolution, electron ionization, mass spectrometry
(GC/MS). The GC/MS was operated in the full scan mode for this
investigation. The GC/MS can also be operated in the selected ion
monitoring mode (SIN) if lower detection limits are desired.
Three surrogate compounds were added to the sample before sample
preparation to monitor method performance. Three internal standards were
added to each sample extract prior to GC/MS analysis and used to calibrate
MS response and serve as retention time markers.
PcBs were identified and measured as Aroclor mixes 1242 or 1254 when the
pattern observed matched that for standards of the commercial mix.
Selected peaks for individual congeners in the sample were compared to
those peaks in standards of the commercial mixes. The polychlorinated
biphenyl content of the samples was also calculated as isomer groups (i.e.
total dichlorobiphenyls, total trichiorobiphenyls, etc). PCB
concentrations were calculated for each level of chlorination and summed
together to obtain the total polychiorinated biphenyl concentration.
These values were compared to The Aroclor results data to determine
whether a concentration bias occurred at the detection limit and at the
upper concentration limit of the calibration range.
ANALYTICAL OUALITY CONTROL : Quality control systems were incorporated
into the analytical method to ensure that valid qualitative and
quantitative data was produced during the analysis of field samples.
Ongoing internal QC checks were routinely performed during sample analysis
to assure that method systems were in control.
Table 5 details the laboratory QC checks used for this project. The data
from these checks were used by the analyst to fine tune the analytical
process and take corrective action where required. They were also
employed by the QA staff to monitor data for systematic analytical
problems. The criteria for these QC parameters are detailed in Table 5.

Table S
Laboratory Quality Control Parameters
Method Blank: Method blanks were analyzed with each batch of samples (up
to a maximum of 20 samples/batch) to check “reagent” or “process”
introduced PCB contamination.
Laboratory Check Standard (LCS or Spiked Blank): A PCB free solid matrix
(fired sand) was spiked with a PCB congener standard and analyzed with
each batch of samples (up to a maximum of 20 samples/batch) to verify
method performance. Recovery criteria were established for the LCS based
on data generated during method validation.
Spiked Duplicate Samples: Duplicate aliquots of the same field sample
were spiked with a mixture of Pc3 congeners and analyzed to determine if
the accuracy criteria had been achieved and to assure that the method was
functioning properly for electronic scrap. The relative percent
difference (RPD) between the two values was used to check analytical
precision -
Internal Standards: Chrysene - d 12 , Phenanthrene - d 10 , and Perylene - d 12 were
added to the sample extract in known concentrations and used as references
to calculate concentrations of targeted compounds present in the sample.
The internal standards were also used to asses instrument sensitivity for
each sample analysis.
Surrogate: Tetrachloro-m-xylene, octachloronaphthalene and
decachlorobiphenyl were spiked into an aliquot of each sample prior to
extraction. The recoveries of surrogate compounds was used to monitor
method performance for each sample.
Single Blind Performance Evaluation Checks: A single blind performance
evaluation (PR) sample was prepared by EPA and introduced into the
laboratory with every batch of 20 samples. This PR sample consisted of
homogenized non-ferrous “white goods” (appliances) scrap that was
previously analyzed during the validation of the analytical method. The
PR was used to monitor the performance of the analytical system on a known
concentration sample.

Table 6
Precision and Accuracy Criteria: Method Quality Parameters
LCSIBtank Spike Accuracy:
Parameter Control Interval
2,4,4’ Trichiorobiphenyt 82 — 113%
2,2’,3,3’,4,4’ Hexachtorobiphenyt 87 — 119%
Matrix Spike Accuracy:
Parameter Control Interval
2,4,4’ Irichlorobiphenyl 61 — 134%
2,2’,3,3’,4,4’ Hexachtorobiphenyl. 63 — 143%
Matrix Spike Precision:
Parameter Relative % Difference
2,4,4’ Trichtorobiphenyt 30
2,2’ ,3 .3’ ,4,4’ Hexach Lorobi phenyt 30
Surrogate Accuracy:
__________________________________ Control Interval
Tetrachloro—m--xyLene 68 — 104%
Octachloronapthalene 62 — 99%
Decachlorobiphenyt 57 — 88%
Performance Evaluation Accuracy:
Parameter Recovery Criteria**
Total Aroctor (1242 + 1254) +1— 30%
** Determined by USEPA review of performance evaluation data.
Field Duplicates. Duplicate field aliquots from the non-ferrous bulk
(NFB) scrap pile (261005-01,02) were analyzed to establish the precision
of the field sampling technique. The precision criteria for the field
duplicates was achieved indicating representative field sampling.
Additional analysis of field duplicates from the remaining scrap piles was
therefore not required or performed by prior agreement.
Table 7
Field Duplicate Results
Field Saxr,le No. EA Lab plo. PCB (ppm) x PCB (ppm ) RPD
261005—O 1A 83670 121 129 9.8
261005—OlD 83673 137 — —
261005—02A 83699 107 117 16.3
261005—020 83702 126 — —
Criteria = <50%RPD for san Les containing >4OppmPCB
Duplicate Aliquots. Field duplicates of each sample were analyzed to
determine the effectiveness of the homogenization technique. The
precision data (RPD) of the data pair indicated the method in which the
PIB data was to be reported. The RPD criteria for duplicates was achieved
for all samples in the field study indicating that the homogenization
procedures were adequate for producing representative samples.

All scrap piles which contained an average P B content >65ppm for the
duplicate aliquots were reported as the average of the pair. Sample
261489-O1A, OlD and sample 26l491-O1A, OlD contained an average PCB
content between 35 and 50 ppm. The standard deviation of the replicate
pair for each of these samples is less than the average. Based on the data
reporting rules previously established (average greater than standard
deviation) the PCB value for this scrap pile was reported as the average
of the pair.
Table 8
Duplicate Aliquot Results
Sa ,Le No. SairiLe No. Aroctor
261005—O IA 83670 12
261005—OlD 83673 —
261 0 05—02A 83699 16
261005 —020 83702 —
261486—OlD 83686 4
261486—OlD 83689 —
26l47%—01P 83790 22
261478—010 83793 —
261454—OlA 83782 18
261454—OlD 83785 —
261492—O1A 84142 47
261492—OlD 84145 —
261510—O IA 84133 35
261510—010 84136 —
261491—O IA 84159 22
261491—O 1A 84162 —
261479—O IA 84185 34
261479—010 84188 —
261013—alA 841fl 17
261013—010 84175 —
261489—O IA 84247 19
261489—OlD 84250 —
261500— O IA 84239 30
261500—O1C 84241 —
261 005—03A 84259 22
261005—030 84262 —
200000—O 1A 84451 —
200000—010 84454 14
Criteria; <50% RN) for sanptes containing >4oppn PCBs.
<100% RPD for sanpLes containing 10 — 40ppn PCBs.
Laboratory Control Sample. The laboratory control sample (LCS) or spiked
blank was analyzed to ensure that the method was being properly executed
in the laboratory. The recovery of the LCS achieved the established
criteria for the majority of the spiked chlorinated biphenyls. A high
bias was observed for two LCS samples. Three spiked chlorinated biphenyls
in these two samples exceeded the established recovery criteria. The
surrogate recovery for these two samples parallelled the spiked
chlorinated biphenyl recovery.

Table 9
Laboratory Control Sample Results
Lab Control SanpLe No. % R C13 % R Ct6
LCS910076 * *
LCS9 1 0 095 110* 70*
LCS910099 83 88
LCS91O119 112 138
LCS9I O16O 127 134
LCS91 0093a 89 86
LCS9W O73a 96 95
a — Aqueous steam condensate san les
* — AnaLyzed as screen to verify recoveries. VaLues reported are estimates.
San ,Le subsequently confirmed during clean up attenpts.
TrichLorobiphenyL (CL3) — 82—113%
HexachLorobiphenyt (CL6) — 87—119%
Matrix Spike/Matrix Spike Duplicate (MS/MSD): The results of the MS/MSD
indicated that a general high recovery bias existed for spiked samples.
In general the high bias paralleled the surrogate recovery for these
samples. The recovery criteria was exceeded for trichlorobiphenyl on
three occasions. The criteria for hexachiorobiphenyl was achieved for all
MS/MSDa. The precision criteria was achieved for all spiked duplicates
with the exception of one hexachlorobiphenyl replicate pair.
Table 10
Matrix Spike/Matrix Spike Duplicate Results
Sm v(e No. ZR Ct3 RPD ZR Cj RPD
83699f4S 168 ii 129 26
83702!4SD 129 — 99 —
84185NS 131 20 96 36
84188MSt) 160 — 138 —
84239i4S e e
84241 14SD n
84451NS 128 7 72 7
84454MSD 137 — 77 —
e — Spiking error invalidated data.
n — High concentrations of native tr chLorobiphenyL in san le invalidated data.
Accuracy Criteria: Precision Criteria:
TrfthLorobiphenyt; 61 — 134% Trichiorobiphenyt; 30
Hexach torobiphenyt; 63 — 143% Hexachlorobiphenyt; 30
Field Sample Data. Results of the analyses indicated that two shred piles
contained polychiorinated biphenyls as .Aroclors and as congeners below 50
ppm. The remaining ten piles of electronic shred and the ground cover
contained PCBs above 50 ppm measured both as Aroclors and as individual

The Aroclor results demonstrated that the PCB content of most piles were
distributed between .Aroclor 1242 and 1254. Pile 261486-01 was
significantly different from the other piles; PCBs in the shred were over
95% Aroclor 1254. The Aroclor 1242 content of the shred pile was almost
entirely associated with the unmillable portion of the pile.
Concentrations as measured from congener analysis were generally lower (up
to 50%) than those determined from the Aroclor analysis. Most of the
analyses were performed at high dilution factors because of the
significant organic matrix present in the shred extracts. It is likely
that a significant portion of the individual congeners present fell below
the concentration level required for detection and ion ratio confirmation.
Congener/Aroclor agreement between analyses performed at lower dilution
factors was better than noted for analyses performed at high dilutions.
Significant weathering effects were not observed in the data for
electronic shred; the congener patterns observed in samples closely
resembled those of the Aroclor standards. Concentration results for each
sample were calculated separately using five individual peaks for each
Aroclor; these individual peak values, averaged for the reported sample
resultB, generally demonstrated low variance.
Significantly greater variances were evident in the observed patterns of
PcBs in the unmillable fraction than were evident in the Aroclor
standards. Variance among the Aroclor concentration of individual peaks
exceeded 100% in some cases, indicating selective losses of some isomers
from the mix.
The steam condensate samples did not contain PCBs above the program-
specified reporting limit of 10 ppm. Detectable Aroclors were, however,
present in the steam condensates associated with shred pile 261500-01.
With the exception of the first shred pile sample, 261005, Pile 261500-01
had the highest Aroclor content associated with the unmillable fraction
on a weight basis.
The PCB data for the scrap samples are listed in Table 12. PCB congener
data for a single sample from the duplicate aliquot pair of each scrap
pile and one PH sample have also been included in Table 12.

Table 12
Electronic Scrap - PCB Resulta
Scrap Field ArocLor Total PCB
Pile No. Sa!rple No. EA Lab No. Conc. PPM Congener PPM
1 261005—O 1A 83670 121 158
1 261005—010 83673 137 —
1 261005—02A 83699 107 159
1 261005—021 ) 83702 126
2 261486—010 83686 542 540
2 261486—010 83689 523 —
3 261478—O 1A 83790 198 189
3 261478—OlD 83793 158 —
4 261454—O 1A 83782 109 60
4 261454—OlD 83785 91 —
5 2614fl—O1A 84142 53 55
5 261492—OlD 84145 86 —
6 261510—OIA 84133 87 —
6 261510—OlD 84136 61 65
7 261491 —O 1A 84159 33 39
7 261491—OIA 84162 41 —
8 261479—O1A 84185 135 103
8 261479—OlD 84188 96 —
9 261013—O1A 84172 357 352
9 261013—010 84175 302 —
10 2614 -89—O 1A 84247 38 39
10 261489—OlD 84250 46 —
11 261500—O1A 84239 616 —
11 261500—O1C 84241 453 546
12 261005—03A 84259 125 90
12 261005—030 84262 156 —
13 200000—O 1A 84451 122 —
13 200000—010 84454 140 79
SU ARY : The PCB data obtained from the analysis of the electronic scrap
samples collected from the metal recycling firm’s scrap shredding facility
indicated that ten (10) of twelve (12) scrap piles and the ground cover
exceeded the consent decree limit of 50 ppm for PCBs. The sampling and
analysis objectives that had been established for the study were achieved.
Quality control deficiencies observed during sample analysis did not
impact data useability. Consequently, the data provided all accurate
depiction of the PCB content of each scrap pile.
The precision data for the analysis of duplicate field samples indicated
that the field sampling techniques used in this study were adequate for
obtaining representative samples of each scrap pile. The precision data
for the analysis of duplicate sample aliqouts indicated that the sample
homogenization techniques were also adequate for generating uniform sub-
samples from the scrap pile composites.
The criteria established for the quality control parameters in this study,
in general were achieved. High concentrations of background organic
compounds and high concentrations of PCB5 present in the scrap resulted
in multiple dilutions of the sample extracts. The dilution process
increased data variability.

The recovery of spiked analytes and surrogates were closely related which
demonstrated that the surrogate data could be used as an indicator of data
bias. In general, a low bias was observed in the P B data which may have
been caused by the substitution of a more rigorous extract cleanup
technique than the technique used during method validation.
1. USEPA Field Manual for Grid Sampling of PCB Spill Sites to Verify
Cleanup, EPA Contract No. 68-02-3938, G. Kelso, M Erickson, D. Cox,
USEPA, Office of Toxic Substances, Field Studies Branch (TS-798),
Washington, D.C. 20460
2. USEPA Method 680. Determination of Pesticides and PCBs in Water and
Soil/Sediment by Gas Chromatography/Mass Spectrometry, A. Stevens, T
Bellar, J. Eichelberger, W. Budde, Physical and Chemical Methods
Branch, Environmental Monitoring and Support Laboratory, Office of
Research and Development, USEPA, Cincinnati, Ohio 45268, November,
1. USEPA Contract Laboratory Program, Statement of Work for Organics
Analysis, Multi-Media, Multi-Concentration, 2/88,
2. SOP For Determination of Percent Moisture In Solid Samples, ETC Corp.,
Edison, New Jersey, December 11, 1989.
3. 40 CFR Part 136, I ppendix B, Definition and Procedure for the
Determination of the Method Detection Limit - Revision 1.11, Office of
the Federal Register, National Archives and Records Administration,
Washington, D.C., page 510.

Inventory List.
P• 1c Nt
Weights and Order of Sampling.
ID No.
r I- 1 r ,,
2 61478-01
261510- 01
261491- 01
261013 -01
261500- 01
261005- 81
200000- 00
478, 000
261, 760
290, 140
290, 140
(not to scale)
W , rTht - Lbg
Client A
Client B
Client C
Client D
Client S
Client C
Client D
Client F
Client C
Client S
NFB 1989
Product Layer (Ground)

for technical focus session — Design of Cost-effective Monitoring Programs
Dean Neptune. PhD. , Senior Environmental Chemist, Quality Assurance Management. U.S.
Environmental Protection Agency, 401 M St., SW, Washington, DC 20460
The US Environmental Protection Agency spends close to a billion dollars each year collecting
environmental data crucial to credible decision-making. Too often in these monitoring
operations, the data users (or decision makers) merely tell the data collectors to obtain data and
after collection, the data users decide on what data they require. As in any large program, there
are opportunities to do the job more effectively. In early 1983, Quality Assurance Mangement
group in EPA began a major effort in applying Total Quality Management (TQM) to the design
of expensive environmental monitoring programs. This TQM effort was labelled Data Quality
Objectives, which simply requires the data user and supporting technical staff to answer three
easy to state questions:
- What environmental data do you require to answer the question concerning the
data users?
- How will the environmental data be used to answer the question of concern?
- How good does the data need to be for this use?
Much progress has been made in developing the technical elements of the data quality objective
process. It has been applied to several Superfund cleanups with impressive results in time and
money savings. One very clear result has been to balance the need for sampling and analysis
with the cost of remediation. In some cases , less samples and analyses are required and in
others more samples are taki n to reduce the amount of media requiring remediauon.
Over the last year, there has been much effort on improving the DQO process . Although the
DQO process is a powerful tool for desigiiing cost-effective monitoring systems, it has required
significant commitment on the part of data users to establish their clear objectives early in
planning. This has not been easy to do. The new guidance issued in FY 1992, is simplifying
the demands required of the data users by relying more on the technical staff early in the
planning process. The technical staff is now presenting alternative scenarios of concern based
on problem priorities established by risk evaluation. This empowers the data users to select that
problem and solution scenario that reflects their real concerns and do it in an efficient manner.
This presentation will describe the customi7ation of the Agency’s programmatic DQO process
guidance for use by RPMltechnical staff in Superfund site remediation.

Gary L. Robertson , U.S. Environmental Protection Agency, EMSL—LV, Las Vegas, Nevada
89193 and Joyce Lee, Lockheed Engineering & Sciences Company, Las Vegas, Nevada 89119.
One of the more important parts of most quality assurance (QA) programs for
environmental data is the use of performance evaluation (PE) samples. The analytical
results from PE samples are used to make judgements regarding the useability of the
data from environmental samples. While PE samples provide important information on
the capabilities of the laboratory there are limits to the extrapolation of PE sample
results to environmental sample results. Factors which affect this are the quality
of the PE sample, the type of PE sample and how closely the matrix of the PE sample
approximates that of the field samples. The analytical result of the PE sample should
be used in conjunction with the other available quality control information
for both the PE sample and the field samples to evaluate the quality of the u.ita.
This presentation will, discuss these issues and provide a perspective on the
application of PE sample results to environmental data.

Keith A. Aleckson . Lockheed Engineering & Sciences Company, Las Vegas, Nevada
89119, and Edward J. Kantor, U.S. Environmental Protection Agency, nvironmentaI
Monitoring Systems Laboratory, Las Vegas, Nevada 89193
Inorganic data audits are perforrnedto assess the technical quality of analytical data
and to evaluate overall laboratory performance. Technical data quality is assessed
based on the total number of problems observed in each data package. The
processes used to identify problems in inorganic analytical data range from a check
of the quality control to a thorough investigation of the raw data submitted with the
case. In addition to providing the basis for determining technical data quality, the
number and type of problems provide a mechanism to track data quality for the
Contract Laboratory Program (CLP), or for an individual laboratory, over time. Long-
term tracking is accomplished by using a data base of standardized audit comments,
which explain common problems found within the data submitted by CLP
laboratories. Each comment represents an individual problem, and the frequency of
these comments is tabulated by the data base. Common problems observed during
the past year in CLP data packages include calibration errors, failure to submit
deliverables, instrument contamination, and the use of incorrect quality control
NQ Although the research described in this article has bean supported by the
United States Environmental Protection Agency tI ough contract 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.

Janice Armour. Viiaya Dandge. David Hewetson , Lockheed Engineering & Sciences
Company, Las Vegas, Nevada 89119, and Edward Kantor, U.S. EPA, Environmental
Monitoring Systems Laboratory, Las Vegas, Nevada 89119
Organic tape and data reviews are performed to assess the technical quality of the
data, adherence to relevant quality control criteria and overall laboratory performance.
The processes used to identify problems range from a thorough review of quality
control information submitted, to a review of the raw data and forms associated with
the case. The number and type of problems will be discussed, as well as how overall
laboratory performance is determined and subsequently tracked over a period of time.
A tool in charting overall laboratory performance, is the CLP Laboratory Performance
Scoring, which is used to assess laboratories based on Quarterly Blind scores, Audits
scores, CCS scores, and Timeliness scores. Long-term tracking is accomplished by
using the Organic Audit Data Base that contains all of the Standard Operating
Procedure (SOPS) and Quality Assurance Plan (QAPs) reviews, tape/data review
scores, On-site Recommendations, and tape submissions (both timeliness and
complete data file submissions). An additional long term tracking toot, and a
possible tool in forecasting, performed by the Laboratory Performance Data Base and
Scout, utilizes multivariate data to assess overall laboratory and program quality.
Data from the tracking tools, as well as problems observed during reviews of CLP
laboratory data over the past year will be presented.
Notice :
Although the research described in this article has been supported by the United
States Environmental Protection Agency through contract 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.

Lester J. Dupes , Quality Assurance Scientist, BCM Engineers Inc., One Plymouth Meeting. Plymouth
Meeting, Pennsylvania 19462
Although data validation guidelines for organic and inorganic parameters have been used since the
early 1980s to evaluate and determine data usability, few guidelines exist for the validation for
analytical data for polychiorinated dibenzo-p-dioxins/polychlorinated dibenzofurans (PCDDs/PCDFs).
The recent U.S. Environmental Protection Agency (EPA) Contract Laboratory Program (CLP)
Statement of Work for Analysis of PCDDs/PCDFs and the EPA Test Methods for Evaluating Solid
Waste (SW-846 Methods 8280 and 8290) provide consistent guidance for laboratory analyses but few
exist for data validation scientists. To remedy this situation, BCM has developed a company standard
operating procedures manual for data validation scientists for evaluating and validating PCDD/PCDF
data deliverables that is based on a review of multi-laboratory supplied analytical data reported under
EPA methods.
This paper provides general information on understanding laboratory analytical data deliverables and
their format. It covers quality assurance objectives, validation criteria, evaluation procedures, and
specific actions applied to the results. It provides specific information on reviewing PCDD/PCDF data
completeness, holding times, gas chromatography resolution, initial and continuing calibrations, blanks,
duplicate and matrix spike criteria, internal standards performance, interferences, identification criteria
for PCDDs/PCDFs, and toxicity equivalence calculations. These criteria and procedures will help the
data validation scientists ensure consistent validation of PCDD/PCDF data packages from
Data validation of polychiorinated dibenzo-p-dioxin and dibenzofuran data deliverables provides the
data user with data of known quality, and valid and legally defensible results and supporting data. The
validation process determines the validity and correctness of the analytical data provided by the
laboratory. The laboratory-reported positive results and detection limits are evaluated and qualified
based on the Quality Assurance/Quality Control (QA/QC) measures employed by the analytical
methodology. The exceedance of QA/QC limits may indicate the necessity for estimating or rejecting
results, based on the severity of the problem.
The following sections explain the major elements of data validation for PCDD/PCDF data
deliverables submitted under Contract Laboratory Program (CLP) 1 and Test Methods for Evaluating
Solid Waste (SW-846) Methods 8280 and 8290 methodologies. Quality Assurance objectives,
validation criteria, evaluation procedures, and actions applied to the results will be outlined in each
section. Technical details have been taken from the specific methodologies and guidance documents
listed as references.

For data validation scientists, the objective of the preliminary review is to confirm that the analytical
procedures followed by the laboratory and that the production of the deliverables required were
performed in compliance with the methodology requested. The data validator should first check to
determine if all raw sample data and QC data has been provided according to CLP or other
methodology. Any non-CLP method should include a level of data deliverables similar to the CLP.
Then, the validation scientist should review the laboratory case narrative and make note of any
problems mentioned and any resolutions made. If the data deliverables are incomplete, the validation
scientist should issue a written data validation inquiry to the laboratory requesting that the missing data
be supplied.
Samples collected during any sampling event may be used as evidence in litigation. Therefore,
possession of the samples must be traceable from the time each sample is collected until it is
introduced as evidence in legal proceedings. Chain-of-custody procedures are followed to document
proper custody of samples, to verify collection times and dates, and to identify the sample location,
matrix, and preservation techniques. Step-by-step details of custody procedures are included in the
project Quality Assurance Project Plan (QAPJP). If these are not available, the scientist should
reference the NEIC Polices and Procedures Manual 4 when reviewing the documentation for proper
Using professional judgment, the validation scientist should note and report all deviations from proper
custody procedures to determine any impact on the data. Severe problems such as broken custody
seals or loss of sample custody may invalidate all reported results.
Currently, the method-specific holding time for PCDDs/PCDFs indicates that all samples must be
extracted (date procedure started) within 30 days and completely analyzed within 45 days of sample
collection. The data validation and qualification procedures for review of analytical holding times for
volatiles, semivolatiles, and pesticides/PCBs have been well established in the EPA’s Laboratory Data
Validation Functional Guidelines for Evaluating Organics Analysis. 5 These procedures should be
applied by the validation scientist to dioxin/f uran analyses to provide for consistent validation
procedures. The validity of results is based on the time period from the date of sample collection to
the date of the laboratory’s preparation and analysis of each site sample. If samples exceed the
specified holding times, the validation scientist qualifies all positive results and sample detection limits
as estimated. Gross exceedance of holding times may indicate that the non-detect data are unusable.
The mass spectrometer calibration at the laboratory is conducted using the compound FC-43 prior to
performing any analysis using CLP 12/90 methodology. This calibration ensures that mass resolution
and identification are acceptable. it is similar in nature to bromofluorobenzene (BFB) and
decafluorotriphenyiphosphine (DFTPP) tunes for volatile and semivolatile analyses. Acceptable mass
spectrometer calibration is based on ion abundance ratios. The ion abundances of M/Z 414 and M/Z
502 should be 30 to 50 percent of M/Z 264 base peak.

SW-846 Methods 8280 and 8290 also mention mass spectrometer calibration to verify resolution.
Method 8290 requires use of perfinorokerosene (PFK) for calibration. Since mass spectrometer
calibration is a preliminary analytical method criteria and is only a recommended procedure, data
deliverables are not required and, therefore, no data should be qualified by the scientist.
The objective is to determine that adequate peak resolution exists for selected analytes prior to the
analysis of samples for PCDD/PCDF compounds. Resolution is performed for the primary and
confirmatory columns by evaluating resolution of 2,3,7,8-TCDD from other closely eluting isomers:
1,4,7,8-TCDD and the 1,2,3,7/1,2,3,8-TCDD pair. The peak resolution is measured from a column
performance mixture or calibration analysis. CLP 12/90 also requires that resolution for 1,2,3,4,7,8-
Hx.CDD and 1,2,3,6,7,8-HxCDD be evaluated. Acceptable resolution criteria for the TCDD isomers is
a valley between peaks of 25 percent and 50 percent for the HxCDD isomers. Laboratory analysis
should not proceed if GC resolution criteria are exceeded.
The validation scientist should reference the individual methodologies to determine that the proper
analytical frequency for measuring resolution criteria is maintained and evaluated. A validation review
of all initial and continuing calibrations and/or performance mixtures to ensure that continuing
resolution of the TCDD and HxCDD isomer is within the criteria must be completed. If the samples
are analyzed after exceeded resolution criteria, then poor resolution of 2,3,7,8-compounds present in
the samples may occur. Therefore, the validation scientist should qualify sample results for 2,3,7,8-
TCDD and/or 1,2,3,4,7,8 and 1,2,3,6,7,8-HxCDD as estimated.
The window defining mix is analyzed prior to the initial calibration and continuing calibration,
depending on the methodology employed. The laboratory uses this data to establish GC/MS switching
times for each ion monitored. This analysis creates a retention time window because the mix contains
the first and last eluting isomer for each dioxin and furan homologue. The established retention time
windo are used in evaluating peaks present in the samples as potential dioxins and furans. The peaks
outside each homologue retention time window can be eliminated as dioxin or furan isomers.
The window defining mix may also contain the chromatographic resolution compounds as a single
solution and both criteria can be evaluated during a single analysis. The window defining mix also
monitors potential retention time shifts by comparing the recovery standard retention times to the
retention times in the most recent continuing calibration. A maximum shift of±10 seconds is allowed.
The validation scientist must evaluate all sample peaks to determine if the retention time of the peaks
is within the retention time window established for each homologue. The peaks outside each
homologue retention time window can be eliminated as dioxin or furan isomers. Chromatographic
resolution is validated as previously discussed. The retention times of the recovery standards are
compared to the continuing calibration to verify a retention time shift of less than.± 10 seconds. If the
shift is greater than the criteria, the sample data must be closely evaluated to look for false positive or
negative results.

An initial calibration of five increasing concentration standards containing all native dioxins and furans
and related isotopically labeled internal standards are analyzed to demonstrate the GC/MS capability
of producing acceptable quantitative data. The initial calibration determines the linear range and
measures the individual response of all compounds. The continuing calibrations document instrument
stability, acceptable compound response, and deviation from curve linearity on a daily basis. All
methodologies have slightly different calibration procedures and specific methods must be used to
verify method compliance. However, the review of initial and continuing calibration data includes
validation of the following criteria: chromatographic resolution of TCDD and HXCDD isomers,
retention times, relative response factors, mass spectrometer (MS) sensitivity, and relative ion
abundance criteria for PCDD/PCDF peaks. Chromatographic resolution and retention times criteria
have been previously discussed. The relative response factors calculated for the 5-point initial
calibration must not exceed 15 to 30 percent relative standard deviation, depending on the analytical
method. Additionally, the percent difference between the initial and continuing calibration response
factors must not exceed 20-30 percent. MS sensitivity is laboratory-determined by measuring the
signal-to-noise ratio (S/N) of each ion. The S/N must be verified by the laboratory to be greater than
2.5:1 for the unlabeled PCDD/PCDF ions and greater than the 10:1 for the internal and recovery
standards to show acceptable detection of each compound.
The validation scientist must continue to evaluate and qualify data based on the chromatographic
resolution and retention times. The relative response factors, relative standard deviations, and percent
differences should be recalculated by the validation scientist to verify accuracy in calculation and
reporting. Positive results and detection limits associated with calibrations exceeding percent
difference criteria are estimated. All peaks for the unlabeled PCDD/PCDF ions and internal standard
compounds must meet the S/N criteria. A low S/N could result in a problem of actual detection of the
compounds and results or detection limits should be estimated.
The analysis of blanks and evaluation of the resultant data determines the potential existence and
magnitude of laboratory, field, or sample cross-contamination. Method and rinse blanks are used to
isolate the source of sample contamination. Method blanks monitor potential lab contamination and
are carried through all preparation and analytical procedures. A minimum of one method blank per
matrix must be analyzed with each sample analysis batch. According to CLP 12/90 criteria, method
blanks cannot contain any chemical interferences or electronic noise for the specific ion monitored
which is greater than five percent of the appropriate internal standard. Any peak positively identified
as a PCDD or PCDF present in the method blank must not be greater than two percent of the internal
standard used for quantitation. Rinse blanks of approved solvents are important in monitoring the
effectiveness of decontamination procedures during field sampling
Positive sample results for 2,3,7,8-substituted isomers should be qualitatively questioned by the data
validation scientist if the results are similar in concentration to the concentrations present in the rinse
or method blank(s). Since the total homologue concentrations reported in the blanks and samples may
consist of many isomers, the results should be validated on a peak-by-peak basis by requantitating the
individual non-2,3,7,8-peaks in the samples and blanks. These concentration can then be compared to
the blank results on an individual basis.

A selected site sample is divided into two portions: one is analyzed unspiked and the other portion is
spiked with several target compounds prior to being extracted and analyzed. Spike recoveries are
evaluated to determine sample matrix effects. The data provides information on the precision and
accuracy of the analytical preparation and analysis on various matrices. One spike sample is analyzed
for each matrix type (i.e., water, soil/sediment, chemical wastes, and ash) present in a sample analysis
The CLP requires a matrix spike analysis, Method 8280 does not require a matrix spike analysis, and
Method 8290 requires a matrix spike and a matrix spike duplicate analysis. CLP recoveries of each
spiked compound must be within 50 to 150 percent. Method 8290 specifies a 20 percent relative
difference between matrix spike/matrix spike results. The validation scientist must use professional
judgment to determine the impact of matrix spike recoveries outside criteria on the unspiked sample
results. Spike recoveries should be recalculated to determine accuracy in reporting.
Laboratory duplicates and field duplicates are evaluated as an indication of overall precision. Field
duplicate analyses measure both field and lab precision; therefore, the results may have more
variability than lab duplicates that measure only lab performance. it is also expected that soil or solid
matrix duplicate results generally have a greater variance than water matrices due to difficulties
associated with collecting identical, homogeneous field samples.
CLP criteria indicate that a duplicate of one sample for each matrix is to be analyzed. Most
environmental investigations require that field duplicates be analyzed at some specified frequency.
According to CLP 12/90, aj SO relative percent difference criteria (j 25 percent for Method 8290) is
used for laboratory duplicates. The validation scientist should estimated positive results for the original
and duplicate samples outside the 50 percent criteria. The field duplicates should be evaluated using
professional judgment. Concentrations near the detection limits should be carefully evaluated during
validation to determine if qualification is necessary.
Internal and recovery standards are spiked into all calibration standards, samples, and blanks. These
spikes consist of a number of isotopically labeled compounds added prior to sample extraction and
preparation, based on analytical method requirements. The internal standards recoveries measure
method efficiency and are used in quantitating native analyte concentrations present in the samples.
Recovery standards are added after sample preparation and extraction, but prior to instrument
injection. Recovery standards are used for calculating internal standard recoveries.
The recovery criteria for each method varies; however, recovery criteria range from 25 to 140 percent.
Since dioxin/furan methods use isotopic.dilution techniques, positive results and detection limits are
corrected for internal standards recoveries, only very low recoveries indicate a potential problem.
More importantly a minimum S/N of 10:1 is required to verify an acceptable level of instrument
detection; isotope ratios are used to identify the compounds and check for possible interferences.

The validation scientist must evaluate all internal standards recoveries to the method-specific recovery
criteria. Note in the validation report all recoveries outside criteria, but do not qualify data quantitated
using the internal standards, unless the S/N ratio or isotope ratio is not within criteria. Results
quantitated using an internal standard that exhibits a low S/N ratio or incorrect isotope ratio may be
estimated due to potential sample matrix effects.
The identification of a detected peak must meet all of the following criteria to be considered a PCDD
or PCDF: retention time, peak identification, and both signal-to-noise and isotope ratios.
If more than one of the following criteria is exceeded, the peak is not identified as a dioxin or furan.
The scientist must evaluate all peaks to verify the laboratory reported results. If the validation
determines that peaks have not met criteria, the result for the peak is qualified as rejected.
Retention Time Criteria
The positive identification of a 2,3,7,8-substituted dioxin or furan isomer is determined by comparing
the retention time of the peak to the isotopically labeled internal or recovery standard added during
extraction. The retention time must be within -1 to +3 seconds to be considered a 2,3,7,8-isomer. If
the labeled 2,3,7,8-standard is not present in the extract, a relative retention time (RRT) is calculated
from the analyte’s retention time and the corresponding internal standard. The RRT of the analyte
must be within 0.05 RRT units (0.005 units for Method 8290) of the RRT calculated from the
continuing calibration. Retention time windows defined during the window defining mix are used to
identify the non-2,3,7,8-isomers for each dioxin or furan homologue. Additionally, the recovery
standard previously discussed in the window defining mix must continue to meet the 10 second
retention time criteria from the associated continuing calibration to monitor potential retention time
The data validation scientist must evaluate all potential sample and blank peaks for the retention time
criteria listed above. The peak must be within the established retention time window to be considered
a non-2,3,7,8-isomer. The exact retention time or relative retention time must be met for the peak to
be considered a 2,3,7,8-isomer. The recovery standards should also be evaluated to determine if
retention time shifts are occurring in the samples. Peaks outside the retention time window or not
meeting specific criteria should be rejected or identification changed as necessary.

Peak Identification Criteria
The two characteristic mass ions monitored for each dioxin or furan homologue and the confirmation
mass ion (M- [ COCII +) for each compound at the specific 2,3,7,8-isomer retention time or detected
within the retention time windows must maximize simultaneously within ..± 2 seconds. The M-
[ COC1] + ion is monitored because the ion is a confirmation ion for dioxins and furans resulting from a
loss of 63 mass units from the dioxin or furan during mass spectrometer ionization.
Method 8290 does not monitor the M [ COCI1+ ion; however, a lock mass ion monitors the stability
and sensitivity changes during the entire GC/MS analysis. A positive or negative spike of the lock mass
ion at the retention time of a 2,3,7,8-isomer may be potentially enhance or supress the 2,3,7,8-isomer
integrated area.
ValidatIon of all sample peaks within the retention time windows, and evaluation of internal and
recovery standards must be performed by the scientist to verify simultaneous detection within 2
seconds. Peaks not meeting this criteria should not be considered dioxins/furans. The lock mass ion
must be evaluated for positive or negative spikes as part of Method 8290 criteria. Large spikes may
indicate that the affected positive results or detection limits should be estimated or rejected based on
the magnitude of the spike.
Signal-to-Noise Ratio Criteria
Each potential dioxm/furan peak must meet a minimum of a 2.5:1 signal-to-noise criteria. Internal
standards must meet a minimum of 10:1 criteria. These criteria must be met to be considered
acceptable signals for quantitation by the laboratory.
Verification by the validation scientist of all sample peaks within the retention tune window and
internal and recovery standards must meet the specified S/N criteria. if the sample peak does not
meet S/N criteria, but all other identification criteria are acceptable, the validation scientist must use
professional judgment to determine if the peak should be reported as a dioxin or furan.
Isotope Ratio Criteria
Isotope ratios are used to accurately determine the number of chlorine atoms present in a compound
by evaluating the Cl and Cl atoms present in a chlorinated compound. Isotope ratios are based on
the probability that a Cl and/or Cl atom is present in the chlorinated compound. These
probabilities change as the amount of chlorination increases or decreases. Evaluation of specific ratios
allows the identification of tetra through octa chlorination of dioxin and furans. The isotope ratio
corresponds to the integrated areas of peaks for the m, m + 2, and m + 4 mass ions. By division of these
areas (e.g., m/m+2), a ratio of the two ions is produced. These ratios are physical constants and both
slight errors in area integration or potential interferences may occur. Therefore, a + 15 percent
difference criteria from the theoretical ratio is used for evaluation.
If a compound does not meet an isotope ratio criteria, the compound is not reported by the laboratory
as a PCDD/PCDF. However, the peak is quantitated and reported as an Estimated Maximum
Possible Concentration (EMPC).

The validation scientist must evaluate all isotope ratios reported for positive sample results, internal
standards, and recovery standards by comparing the reported ratio with the list of isotope ratio ranges
(j 15 percent from the theoretical ratio) found in each method. Several ratios should be recalculated
to verify accuracy in laboratory reporting. The scientist must verify that peaks not meeting isotope
ratio criteria are calculated as EMPC by the laboratory.
The quantitation of positive results and detection limits provide the data user with values for each site
sample. Total concentrations are also calculated by summing the individual 2,3,7,8-and non-2,3,7,8-
isomer concentrations within the retention time window. The detection limit for a compound is
individually calculated to provide an estimated concentration of the analyte needed to produce a peak
at a S/N ratio of 2.5:1.
Validation is performed by recalculating the results to ensure accuracy. Each analytical method must
be referenced by the scientist to determine the correct quantitation method. Recalculation is
performed to verify that the correct sample peak area, internal standard area, response factor, sample
volume or weight, and quantity of the appropriate internal standard is used. Errors found in
quantitations should be checked by the laboratory through a data validation inquiry made by the
scientist, and affected forms and results should be resubmitted.
The objective of evaluating PCDPEs is to determine if these interferences are affecting PCDF
quantitations. A (M + 72) mass ion is monitored in conjunction with the compound-specific mass ions
for hexa through deca chlorinated diphenylether which interfere with dibenzo furans.
CLP criteria states that the positive identification of a PCDF cannot be made if a PCDPE peak of
greater than 2.5 S/N is detected at the same retention time as the corresponding furan. If a PCDPE is
detected, an EMPC is calculated and reported for the PCDF isomer. This quantitated result provides
a reference to the maximum concentration possible if the peak had met all of the identification criteria
as a furan.
The PCDPE mass ion must be evaluated by the scientist to identify peaks above a S/N of 2.5:1. The
retention time of the PCDPE is compared to the retention time of peaks present in the corresponding
furan. Furan peaks which are interfered with by PCDPE should be reported as EMPC by the
laboratory for CLP.
The methodology requires the calculation of the 2,3,7,8-TCDD toxicity equivalence according to the
procedures given in the EPA’s “Update of Toxicity Equivalency Factors (TEFs) for Estimating Risk
Associated with Exposure to Mixtures of Chlorinated Dibenzo-p-Dioxins and Dibenzofurans
(CDDs/CDFs)” March 1989 (EPA 625/3 89/O16).6 Of the possible chlorinated PCDDs/PCDFs, the
17 isomers that bear chlorine atoms in 2,3,7, and 8 positions are the greatest concern in the assessment
of risk to human health and the environment.
A factor is assigned to each of the seventeen 2,3,7,8-substituted PCDDs and PCDFs that relates the
toxicity of that isomer to a concentration of the most toxic isomer 2,3,7,8-TCDD. These factors are

called toxicity equivalence factors (TEFs). The concentrations of any of the 17 isomers that are
detected in an environmental sample can then be adjusted by the TEF and summed yielding a
concentration of 2,3,7,8-TCDD with an equivalent toxicity The “Total” concentrations are not
assigned TEF values in the March 1989 TEF procedure and, therefore, are not included in the toxicity
equivalence calculations.
If the toxicity equivalence exceeds CLP criteria, then analysis on a second column capable of resolving
2,3,7,8-TCDDJ2,3,7,8-TCDF and as many other isomers from coeluting peaks is required. Method
8290 requires a two-column analysis to provide a more accurate value for 2,3,7,8-TCDF due to
coelutions. All values except 2,3,7,8-TCDF are used from the original column analysis, while the
2,3,7,8-TCDF result from the second column is used in calculating TEFs. Since a quantitative and
qualitative uncertainty is associated with EMPC values, they are not included in the TEF calculation
performed in the methods according to CLP 12/90. Some EPA regional criteria indicates that EMPC
values for 2,3,7,8-isomer be used to obtain the worst-case scenario. 7
The validation scientist should verify that the correct TEF factors and sample concentrations are
reported by the laboratory. The scientist must also verify that the correct EPA regional or state criteria
is used and recalculation of TEF concentrations are accurate.
Data validation of deliverables produced from PCDD/PCDF analyses provides the end data user with
valid and legally defensible results and supporting data. The complexities of the various methodologies
require the data scientist to have a thorough understanding of the analytical techniques, quality control
procedures, and quality control criteria for these deliverables. The development of a Standard
Operating Procedure has provided BCM data validation scientists with guidance in validation. As with
the analytical methodologies that provide the analyst with a consistent protocol for performing the
analyses, the validation guidance provides for a consistent review of dioxin/furan data deliverables.

1. United States Environmental Protection Agency, Contract Laboratory Program
Statement of Work for Analysis of Polychiorinated Dibenzo-p-Dioxins (PCDD) and
Polychlorinated Dibenzofurans (PCDF), Document Number DFLMO1.0,
December 1990, revised September 1991.
2. United States Environmental Protection Agency, The Analysis of Polychiorinated
Dibenzo-p-Dioxins and Polychiorinated Dibenzofurans, Test Methods for
Evaluating Solid Waste, SW-846, Method 8280, September 1986.
3. United States Environmental Protection Agency, Polychiorinated Dibenzodioxms
(PCDDs) and Polychiorinated Dibenzofurans (PCDFs) by High Resolution Gas
Chromatography/High-Resolution Mass Spectrometry (HRGC/HRMS), Test
Methods for Evaluating Solid Waste, SW-846, Method 8290, November 1990.
4. United States Environmental Protection Agency, NEIC Policies and Procedures,
EPA-330/9-78-001-R, May 1978, revised May 1986.
5. United States Environmental Protection Agency, Laboratory Data Validation
Functional Guidelines for Evaluating Organics Analyses, February 1988.
6. EPA’s “Update of Toxicity Equivalency Factors (TEFs) for Estimating Risk
Associated with Exposure to Mixtures of Chlorinated Dibenzo-p-Dioxins and
Dibenzofurans (CDDs/CDFs) ” March 1989 (EPA 625/3-89/016).
7. EPA Region ifi QA Directives, Dioxin’s International Toxicity Equivalency Factors
(I-TEF/89), QAD 016, February 1990.

Mark Dymerski-Techflical Manager, Steven Malecha-Technical Manager, and Maureen
McDev itt-Technical Di ec, Enseco-Rocky Mountain Analytt cal Laboratory,
Arvada, Colorado 80002.
Because of limited resources available for investigation and remediation of
contaminated sites, there is a pressing need to produce quality data in a cost
effective manner. The development of the £nvlrenaental Protection Agency
Contract Laboratory Program (EPA-CLP) addressed project specifications by
requiring laboratories to follow to specific methodologies. While this
conformity was necessary ‘in the Infancy of the Environmental Testing Industry,
laboratories today can meet client needs with techniques and methodology that
differ from the strict guidelines of the CLP. The performance based process
allows the laboratory to tailor its techniques around the specific
requirements of a project arid site. The flexibility of this process means
that the laboratory is far more dynamic In Its analytical approach. Mew
techniques and Improvements In laboratory efficiencies can be utilized to
greatly reduce costs which benefit both the customer arid laboratory. The
performance based approach involves a process of shared information between
client and laboratory. This partnership between client and laboratory
instituted from a project’s beginning provides the necessary coniuunication to
meet data and budgetary objectives. With no official national laboratory
certification program, the CLI has functioned as such. A laboratory that
participates in the CIP Is given preferential treatment in the industry, even
if it does not follow the CLP protocol for Industrial clients. As a result of
this bias in industry, maintaining Cl) status is a cost of doing business. It
has become far more competitive to participate as winning bids for C I) sample
lots continue to decrease below most laboratories’ cost structure. The
problems between performance based methodology and CLI can be seen in recent
requests for proposal from the Department of Energy. Some parts 0 f proposals
state that the technical approach to clean up of these sites would utilizethe
technical experti so of lab personnel In solving analytical problems. Other
sections call for a program similar to the CLP. Both techniques cannot
successfully function in the se program. Other issues that affect the
Environmental Testing Industry pose additional problems. Laboratory capacity
Is a key issue that must be addressed in the DOE plan. The private sector
needs to know the true scope of the program, so that it can properly prepare
itself. Another issue that has risen Involves the clear discrepancy emerging
between the acceptable levels of radioactive materials a facility can accept.
Presently, there are two different regulatory definitions of radioactive
material licensing. The Nuclear Regulatory Coission (NRC) grants licenses
to laboratories in states that do not have their o i policies while
‘agreements states grant their own licenses based on their own criteria. The
discrepancies between state and federal regulations can vary greatly.
Definitions of safe levels greatly affect the functionality o laboratories
that analyze mixed wastes. These differences between a centrally regulated
industry and one that allows individual judgement create many questions about
how the Environmental Testing Industry functions.

Mary Fend and Marilyn Hoyt, ENSR, Acton, Massachusetts, 01720, and Joel Kamofsky,
Berkeley, California, 94702.
ENSR has recently developed an automated data validation (ADV) system for volatile,
semivolatile, pesticide/PCB, and inorganic data. This system is used for large data sets where
the analytical protocols specified follow the CLP protocol.
For selected projects, ENSR receives laboratory data deliverables both as hardcopy and on
diskette. The diskette data are read into the ADV system, while the basic screening-level checks
are performed. A variety of reports are available through the system; these reports are used to
highlight data problems and deficiencies.
Once the data has been validated, it is exported to ENSR’s Project Database Management
System. This system is used to manage both field and laboratory data, as well as facilitating
more graphical and statistical presentations of data.
This presentation will provide an overview of the system and present a detailed view of the ADV
reports used to detect data problems and deficiencies.

Waste Management’s Environmental Monitoring Laboratories (EML) receives over 200,000
pieces of field data information a year from groundwater wells at our solid waste and hazardous
waste facilities across the United States. The accurate transmittal of this data to our company’s
central environmental groundwater database at the EML is vital for regulatory reporting and
groundwater monitoring program analysis. Examination of the sampling and field data docu-
mentation process revealed significant time and accuracy advantages would be gained by in-
sernng handheld computers into the sampling procedures.
Handheld computers were programmed to accept all information on the EML Chain-of-Custody
and Field Information Forms. All data input was password protected to accept data for a par-
ticular sampling event only from authorized samplers who were trained and identified as the
designated samplers for the event. Barcoding capabilities were added to enable the samplers to
barcode well and sample bottle identifications. The system was developed to utilize a modem
to our centralized VAX computer system at Geneva, illinois. All field data and comments
were electronically loaded into the appropriate sections of our groundwater database. Data
range specifications were built into the system to prevent gross transcription errors such as
transposing pH and conductivity readings.
The system was programmed to print a copy of all field data information from a portable field
printer for final sampler quality checks, review and signatures prior to transmittal to the EML
database. Future application enhancements include obtaining direct readings from on-site or
“down the hole” instrumentation into the handheld computer unit for electronic transmital to the
Waste Management’s Environmental Monitoring Laboratories track sampler error rates on all
incoming samples to the Geneva facility. Because the lab is responsible for company wide
sampler training and sampler auditing, this information is useful to track sampling team

performance improvement and is part of the total quality improvement effort. Although the
error rate is low (0-3% ax most sites), every error has the potential to cause invalidation of the
data. There is also significant time and cost associated with the resampling of the event.
Several units were evaluated to automate as much of the sample collection process as possible.
Programming was completed on two units which were identified as most useful for our compa-
nies’ purposes. Pilot studies were completed at the EML and in the field. Error rate improve-
ment was particularly sought in the following areas:
Filtering Information
Correlation of Signatures/Dates/Sample Points/Bottles Sets
These categories related to the total error rate as follows:
Catagory % of Total Errors
Documentation 29.5%
Filtering Information 2.2%
Correlation of Signatures!
Dates/Sample Points! Bottle Set 20.6%
Miscellaneous 3.0%
Total Errors Affected 55.3%


Potential cost savings from elinmating these errors were as follows:
Resampling (184) events @1750 $322,000
Analytical Costs (184 events @2500) $460,000
Rescheduling/error tracking $ 38,825
Total Potential Savings $820,825
Several handheld computers were evaluated during the pilot testing phase; 1TRON T3000 hand-
held computer was selected as the unit that provided the most features that met list of require-
ments. The following is a list of requirements (both hardware and software):
1. Rugged unit - Unit required to withstand extreme weather conditions; water and dust
resistant, shock-absorbent and light in weight.
2. Screen Display - Unit required to display and collect data in a easy-to-read manner.
3. Bar code - Required to integrate with barcode reader.
4. Power Supply - Rechargeable hicad batteries or alkaline batteries.
5. Unit reliability - Unit required to have an excellent reliability and lifespan.
6. RAM Memory - 128K with expansion capablility to 1MB
1. Standard Operating System - MS DOS like operating system.
2. Programming capability - Ability to easily program a customized system.
Custom software for the 1TRON unit was developed to provide specific funtions.
These features include:

The user is required to enter a user name and PIN number before accessing any data on the unit
or functions.
Data can be uploaded or downloaded electronically. This allows sampling data to be verified
and checked for errors.
The unit supports a barcode reader to read labels on bottles and sampling points.
The unit has an internal clock that allows automatic date and time identification for all data en-
The unit collects the data entered by the user and performs calculations on the data entered (i.e.
purge times, volumes, flow rates)
A printer can be attached to the unit which provides the user the ability to print key forms or
The unit includes a built-in modem which can be used to transfer data to and from the unit.
Our pilot studies have pointed to the following advantages of incorporating handheld computers
into our groundwater sampling programs:
1. Documentation and Security -the validation of all data entries insures review by
sampling teams and prevents gross data entry errors. Only trained samplers are granted
access to the Field Data Collection System (FDCS)
2. Barcoding prevents bottle set mix-up and positively identifies sampling points prior
to actual sampling.
3. Calculations programmed into the FDCS system prevent human error on field infor-
mation forms.
4. Time Savings- field data entry time is cut in half, error tracking and subsequent
rescheduling of events is minimized.
5. Cost savings-the cost savings realized is substantial.
6. Data uploading and downloading-automated data transfer via modem prevents man-
ual data entry errors and saves significant time.

Pilot studies are complete and we are looking forward to the quality improvements and
complete company wide savings that will be realized as we phase in the FDCS system.

Marjorie Hummel , Lars Lindquist, Robert Dovi, Frank Dias,
Bruce Warden, WMI—Environmental Monitoring Laboratories, Inc.,
2100 Cleanwater Drive, Geneva, IL 60134
Reagent and standard preparation is necessary for almost every
technique used in the chemical laboratory. This is a laborious
task and little effort has been made to automate this process.
In 1989, our laboratory initiated a project to automate the
preparation of 27 standards and 22 reagents for EPA approved
methods. This automated system was developed using a custom—
designed robot built by Bohdan Automation(Mundelin,IL 60060).
The system provides the following benefits:
o Labor savings by unattended operation
o Better utilization of analyst time
o EffiCiency gains due to increased sample throughput and
improved turn—around time
o High quality of standards and reagents maintained while
reducing preparation and documentation time
o Gravimetric verification of all preparations
o An electronic audit trail
o Safer chemical handling.
This automated system is a PC menu—driven, 3—axis overhead
gantry arm, operational robot. The system prepares standards
and reagents gravimetrically using two different size pipets
and an electronic balance. Tests, to date, show the system is
equivalent to manual preparation using ASTM Class A glassware.
Standard and reagent preparation is necessary with most
methods and can be quite laborious and tedious. Accurately
prepared standards and reagents are essential in obtaining
the high quality of data needed. The safe handling of
chemicals by analysts is another consideration — handling
chemicals by an automated system is safer.
With these objectives in mind, our laboratory decided to
automate this process. The product of this project is a
custom—built laboratory automated system, known as the
Reagents and Standards Preparation Robot (RASP). The RASP is
capable of preparing up to 49 standards and reagents. It is
also capable of working with both liquid and dry chemicals.

Solutions and chemicals are weighed out manually and the
diluents and additional liquid chemicals are then added
proportionately by the RASP. Liquid chemicals are pipetted by
the RASP into the bottles and amounts are determined
gravimetrically. At the end of each reagent and/or standard
preparation, a label and a report are generated and the
pertinent data archived on the PC’s hard drive.
The RASP is 102 inches in height and sits on a 32x60 tabletop.
The interior of the system is enclosed behind Plexiglass
access doors. On the exterior is an air compressor to run the
pneumatics,a backup power supply for the robotic portion of
the system in the event of a power failure, and a PC with two
printers for label and report generation.
The overhead gantry arm moves throughout the whole work area
in the three axes (X,Y,z). The arm is equipped with a gripper
hand to pick up bottles or a lml pipet hand or a lOml pipet
hand. The arm is controlled by three SAC—560 Smart Axis
Controllers. These controllers are microstepper—motor
controllers which “ will interface to any step motor driver
requiring step and direction inputs”.’
The automated system also includes a top—load balance which
is used for all the gravimetric determinations.The robot
gravimetrically determines the weights of all standards and
reagents made. The balance is used to determine the fluid
levels in the bottles prior to pipeting.
The RASP has two polyurethane pipet heads which are connected
to automatic microstepper—controlled syringes. These heads
allow the use of either imi or lOmi pipets. Underneath the
heads is a drawer to receive spent pipet tips.
A pump dispenser is connected to four peristaltic pumps. These
pumps deliver four diluents which are used for both standards
and reagents.
Finally, the last component is a stir plate used to stir the
prepared solutions — reagents are stirred for five minutes
and standards for 30 seconds.
An equivalency study was performed to determine the precision
and accuracy of standards prepared by the robot against ASTN
Class A glassware.

Standards prepared by the RASP were used for Inductively
Coupled Plasma (IC ! ’) and Graphite Furnace Atomic Absorption
(GFAA) analyses.
Three separate calibration standards were prepared by the RASP
for each of the two analytical systems. Using the calibration
curves derived from these calibration standards, certified
standards were analyzed by the two techniques. Criteria for
equivalency corresponds to results within the acceptance
ranges of the certified standards.
To test for precision, ten—element samples and standards were
prepared ten times by the RASP and manually using ASTM Class
A glassware for ICP analysis. Ten replicate copper samples and
standards were prepared for GFAA. These were analyzed using
working standards previously prepared manually by the
analysts. Criteria for equivalence was for the averages of the
set of ten determinations to fall within the current
specifications for percent recovery (%R) ± 15% for GFAA and
± 10% for IC ?. See Tables 1, 2, and 3 for results. Only sodium
for the ten standards prepared for ICP analysis was out of
specification (87.8% recovery); recoveries for copper in the
robot and manually prepared samples ranged from 99.5% to
106. 3%.
Further testing was performed using the RASP by preparing six
calibration standards for six point calibration curves for a
combined chloride/sulfate analysis and for chemical oxygen
demand analysis. Certified quality control standards (three
for each analyte) were analyzed to test for accuracy of
standard preparation of the calibration standards. Acceptance
is based on the ± 2 sigma of the true value established by the
manufacturer of the certified quality control standard.The
results are summarized in Table 4 below:
Table 4. Recoveries for Standards
Concentration (ppm)
Analyte True Determined %Recovery
C1 63.9 63.9 100
Cl 35.4 38.4 108
C1 63.9 63.9 100
$04 24.3 22.4 92
SO 4 30.4 28.1 93
SO 4 24.3 21.7 89
COD 200 237 94
COD 200 228 90
COD 200 229 90

XC ? CCV (Robot)
Ag Ba Cd Cr Cu Mn Fe Ca Mg Na
(500) (500) (250) (500) (500) (500) (2500) (50000) (50000) (50000)
474 483 249 509 469 480 2499 47440 45540 46730
470 469 248 496 468 462 2342 45110 44170 45100
486 476 261 520 477 484 2510 48060 46400 47140
482 473 254 506 476 476 2464 49190 46640 46070
479 482 25]. 516 487 477 2476 47750 45290 46710
486 492 265 528 498 496 2557 49130 47010 47530
489 490 264 528 480 489 2555 49100 46710 4681.0
491 481 260 531 485 497 2556 49130 45610 46520
489 493 262 534 481 498 2568 48090 46730 47130
493 491 267 528 504 509 2598 49120 46900 46730
484 483 258 520 484 487 2513 48212 46100 46647
ave %R 96.8 96.6 103.2 103.9 96.7 97.4 100.3 96.4 92.2 93.3
ICP CCV (Manual)
Ag Ba Cd Cr Cu Mn Fe Ca Mg Na
(500) (500) (250) (500) (500) (500) (2500) (50000) (50000) (50000)
483 493 262 506 491 486 2450 48590 47340 48210
492 498 265 512 491 503 2504 47250 47170 47690
475 496 270 501 488 496 2487 46880 45780 48890
479 492 255 497 489 484 2432 48050 47250 48550
479 493 260 495 489 481 2420 47940 47510 49810
476 494 259 500 495 482 2417 48530 48150 49470
485 488 258 499 497 474 2396 48130 46900 48630
479 488 252 490 480 476 2399 47120 47060 48050
486 505 264 501 499 489 2442 47470 47380 47590
477 498 259 496 486 479 2447 46790 46620 47810
IC 481 495 260 500 491 485 2439 47675 47116 48470
ave %9. 96.2 98.9 104.2 99.9 98.1 97.0 97.6 95.4 94.2 96.9
— Conc in ppb.

Ag Ba Cd Cr Cu Mn Fe Ca Mg Na
(1000) (1000) ( 500) (1000) (1000) (1000) (5000) (100000) (100000) (100000)
990.2 948.3 559.0 1057 950.8 1047 5249 104000 96510 93620
918.7 851.2 509.4 956.9 865.6 956.9 4776.5 86970 80180 80925
926.5 878.3 527.8 985.2 895.4 986.2 4909 94880 86440 85105
936.6 886.1 521.1 975.0 892.2 989.8 4914.0 88910 84075 82570
970.4 956.2 517.1 955.2 988.7 965.3 4814.5 97120 96920 92110
935.3 915.8 508.1 942.5 934.6 966.2 4819.]. 96075 96190 88765
937.3 921.2 531.9 946.3 954.8 977.9 4939.8 98620 97055 89200
908.3 879.4 491.3 887.6 904.6 933.3 4651.5 86620 86445 82210
924.8 901.3 514.5 917.8 936.6 956.8 4763.0 93405 94145 86355
1036.5 1001.1 507.9 957.8 1046 989.5 4956.5 98510 95425 96670
x 948.5 913.9 518.8 958.1 936.9 976.9 4879.3 94516 91338 87753
Ave %R 94.8 91.3 103.8 95.8 93.7 97.7 97.6 94.5 91.3 87.8
IC? Standard (Manual)
Ag Ba Cd Cr Cu Mn Fe Ca Mg Na
(1000) (1000) (500) (1000) (1000) (1000) (5000) (100000) (100000) (100000)
964 985 490 958 947 944 4846 96910 93880 98180
964 994 492 969 952 955 4844 97090 94390 97340
964 966 490 958 943 956 4796 95540 93760 97970
971 978 487 970 946 942 4803 95460 94160 96650
953 981 491 955 954 955 474]. 97250 93900 95120
948 984 491 961 945 946 4805 96400 94260 95040
966 975 493 961 963 958 4835 96890 95710 98330
952 968 493 976 959 956 4854 96330 94640 97720
952 998 490 966 949 965 4851 96500 93240 97690
957 978 486 977 945 962 4859 95080 94200 94620
959 981 490 965 950 954 4823 96345 94214 96856
ave %R 95.9 98.1 98.1 96.5 95.0 95.4 96.5 96.3 94.2 96.9
( ) conc. in ppb.
AU. mans fell, within +10% recovery criteria.

MEAN 30.8 31.9 20.0 19.9
AVER %R 102.7 106.3 100.0 99.5
Standard Samples (30.0 ppb) CCV ppb)
Robotics Manual Robotics Manual

The RASP was considered validated for the preparation of
standards for these analytes.
The precision of the RASP was also checked by analyzing ten
standards each prepared at two concentration levels for the
three anaiytes(Cl,S0 4 , and COD).The relative standard
deviations for the tests are tabulated in the following Table
Table 5. Precision for C1, SO 4 , COD analytes
Concentration %RSD*
(mg/i) Manual RASP
Cl 2.5 4.6 12.2
C1 100 0.4 2.2
SO 4 40 0.9 2.1
SO 4 400 0.4 1.0
COD 20 18.9 24.9
COD 600 0.8 1.2
These results were considered acceptable even for the low
levels of C1 and COD. (Analytical error may have increased
the %RSD for these low levels.)

The RASP is now being used to prepare the following reagents
for the analytes listed in Table 6.
Table 6. RASP Prepared Reagents
eagent Ana].yte
Mercuric Thiocyanate Chloride
Ferric Nitrate Chloride
Barium Chloride Sulfate
Barium Methlythimalblue Color Sulfate
Automated Buffer Cyanide
Chloramine—T Cyanide
Pyridine Barbituric Acid Cyanide
Magnesium Chloride Cyanide
Buffer pH 3.1 Alkalinity Methyl Orange
Methyl Orange Alkalinity Methyl Orange
Sodium Persulfate Total Organic Carbon
Phosphoric Acid Total Organic Carbon
Our laboratory has automated the process of preparing
standards and reagents using a robotic system. Based on the
equivalency studies conducted comparing the automated process
verses a manual process, the validity of the RASP to prepare
standards and reagents has been demonstrated. The robot has
been preparing standards as needed on a daily basis for GFAA
and ICP and on a weekly basis for the COD, C1, and SO 4

l.SAC—560 Manual Software Version 2.7A; Motion Control Group,
American Precision Controls: Buffalo, New York; p.2.

Kim D. Johnson , Account Executive, Environmental Testing and Certification
Corp. (ETC)-Midwest, 3025 Hgwy. N78, Mt. Horeb, Wisconsin 53572.
In the environmental field, the design of monitoring programs is a daily
occurrence. As our field matures, the need for a cost-effective approach
to monitoring programs is gaining attention.
The Program Management group worked hand-in-hand with the engineering
group in the development of a monitoring program for an industrial client.
During the characterization phase of the project, we chose to perform all
analyses using the full list of parameters (i.e., EPTOX or TCLP) using
Level III Data Quality Objectives. Based on the test results, we met with
the engineering staff to discuss a recommendation to the state agency that
a reduced parameter list be considered. Agreement with the agency was
reached and we developed Level II data quality objectives for the next
phase of the project. For this investigation phase, Pb was identified as
a tracer analyte.
The purpose of the investigation phase was to determine the perimeter of
the contamination at this site. To cost-effectively support the project
needs, we developed a modified analytical procedure for Pb which consisted
of an abbreviated digestion procedure, limited QC, and shortened ICP run
time. The modification was an attempt to lower initial project costs,
and shorten the analytical turnaround time. After the analysis of a large
number of individual soil samples, we developed a correlation between the
EPTOX (Level III data) and the total analysis (Level II data). This
correlation data was presented to the state agencies as our basis to
justify clean-up levels at this site. Using the gite specific data, we
reached an agreement with the agency. The analytical results and the
support of the analytical laboratory senior staff played an intricate role
in the negotiation with the agency.
Many agencies take the position that clean-up must be to background
levels. Determining the background level for a specific site, however,
can be difficult: How are natural occurring elements factored into this
equation? What are potential alternate sources for non-natural occurring
elements? How do background levels compare to realistic detection limits?
Our strategy of working with the agency and justifying the background
levels based on site specific data has been very successful. With
concentrated efforts, we have developed a positive image with the agency;
based on our determination to be:
• Responsive
• Cooperative
• Responsible

The success of tying the analytical needs to the project needs was a true
test of teamwork. The success of the partnership has resulted in:
• Cost-effective program
• Saving of $100,000 for our client
• Accelerated project schedule
• Decreased remediation scope
The need for further cost-effective alternatives for monitoring program
development is increasing. In the future, analytical guidance must be
closely tied to the development of the project objectives and engineering
firms need to work closely with regulatory agencies to develop
“customized” site-specific solutions. The support from the analytical
laboratory staff is critical to assist clients in chemistry options which
can help in development of appropriate DQO. As a laboratory community,
we develop strong working relationships which can be very beneficial to
our clients. Also, as chemists and scientists, this evaluation of testing
options and interpretation of data are strengths. We are hopeful that
such actions will offer cost-effective savings to our clients and
ultimately, the agency programs.

Harish Mehra , Laboratory Director,Chemica l Waste Management, Inc., Western Region
Laboratory, P.O. Box 4249, Modesto, CA 95352
In this paper several methods of controlling the amount of hazardous waste generated
at environmental laboratories are discussed. Waste minimization is an important
element of current hazardous waste regulations, and developing a less is betterN
philosophy can help you meet your objectives in this area. In the body of this paper
we will outline how you can achieve waste reductions leading to cost of disposal
savings of up to 70%.
All approaches discussed in this report comply with existing health and safety
programs, QAIQC concerns, and prescribed EPA laboratory methods. Fourteen
specific guidelines are discussed concerning chemical inventory control, waste
segregation and treatment, standardized sample sizes, and more conservative
subsampling that increase overall efficiency and waste reduction, while ensuring high
standards of quality control and rapid turnaround time. Management practices are
suggested which can help to achieve goals of waste reduction. The obvious
advantages of a waste minimization program are outlined such as reducing costs,
saving time and protecting human health and the environment. Therefore, ensuring
that environmental laboratories are part of the solution and do not become part of the

Gloria L. Poling , Quality Assurance Manager, Robert A. ThomDson , WET
Chemistry Supervisor, Chemical Waste Management, Inc., Western Region
Laboratory, 1430 Carpenter Lane, Modesto California 95351
When analyzing hazardous waste samples, accurate analytical results are
critical in determining waste disposal decisions and complying with federal and
state regulations. The environmental and financial implications of inaccurate
analytical data are detrimental to companies dedicated to protecting the
environment and servicing customers. This poster session displays a pro-
3ctive, comprehensive Self-Audit program of data review and method
adherence which is designed to detect, in a timely manner, analytical problems
and compliance issues that could affect waste decisions.
The Western Region Laboratory self-audit program began with the development
of formalized checklists for the Lab Director, QA Manager, and supervisors to
complete over a prescribed time frame. These audits are performed on a daily,
weekly, monthly, quarterly, and semi-annual basis. The frequency is dependent
upon the severity of non-compliance for a particular issue. As the program has
matured, the frequency of self audits, the items covered, and the level of staff
participation have been modified to allow for new problems which occur from
time to time.
The Laboratory Director and QA Manager conduct self-audits for compliance
with controlled document policies; training performance, documentation, and
organization; MDL policies; QC procedures; proper documentation of
discrepancies; archiving procedures; facility adequacy; and a review of past
action items. In addition, the laboratory group self-audits and laboratory quality
assurance group activities are monitored.
Laboratory supervisors self-audits are extremely detailed and include review of
data calculations; frequency of QC samples and calculation of control limits;
proper documentation techniques; reagent tracking and compatibility of
storage; control chart trends; personnel adherence to analytical methods;
tracking of samples from receipt to report generation; proper sub-sampling
techniques; instrument maintenance and calibration; and effective trouble-
shooting for out-of-control situations. Sample processing coordinators perform
self-audits that include issues regarding sample storage, paperwork
discrepancies, submission of QC samples, and laboratory turn-around time.

The quality assurance group monitors documentation of corrective action for
out-of-control situations, errors in the monthly QC statistics report, develops
trend analysis charts for blind duplicate discrepancy resolution, and compares
results of other QC performance samples.
With the myriad of regulatory, compliance, and health and safety issues
currently facing environmental laboratories, a well-structured program to
monitor the laboratory’s compliance is highly beneficial. The Western Region
Laboratory’s Self-Audit Program is an extensive system of checklists and other
mechanisms designed to measure Quality Assurance, Health and Safety, and
Environmental compliance. The success of this program has been it’s
comprehensiveness, timeliness, and the involvement of personnel at all levels.
Daily, the Laboratory Director performs a laboratory safety/housekeeping
inspection. At the end of each month, the Laboratory Director awards points
that are applied to each group’s safety performance scoring. Likewise, the QA
Manager performs a similar inspection of administrative areas.
Monthly, quarterly, and semi-annually, self-audits are performed by the
Laboratory Director and QA Manager in accordance with a frequency table
located in the Laboratory Director/QA Manager Self-Audit Module. The topics
of these self-audits range from ensuring compliance with corporate policies,
procedures, and SOP’s to review of instrument maintenance and calibration and
parameter MDL studies. In addition, the supervisors’ self-audits are reviewed
with regards to comprehensiveness and seriousness of the discrepancies. The
Quality Assurance Manager utilizes a parameter checklist to monitor compliance
of Standard Reference Material submission, parallel data input, MDL
requirements, or any other activities which are parameter specific.
Upon completion of the self-audit during a given month, the Laboratory
Director/QA Manager generate a written summary of the audit results.
Corrective action is assigned to the appropriate supervisor for those items
which have discrepancies. Due dates for the corrective actions are designated.
Correction to action items are followed up by the Laboratory Director/QA
Manager on a monthly basis.
The Supervisor’s Self-Audit Module is designed to monitor compliance with the
Corporate Quality Control Policy and Procedures and with Good Laboratory
Practices. On a daily basis, the supervisor has the responsibility of monitoring
department productivity, ensuring that samples are analyzed by sample priority,
and reviewing raw data and benchsheets.

Using the Supervisor’s Quality Assurance Self-Audit Checklist, the supervisors
perform weekly, monthly, quarterly, and semi-annual self-audits. The
frequency of the items covered is based on the potential and severity of non-
compliance. At the end of each month, the results of the audit are summarized
with copies of the summary routed to the Laboratory Director and QA Manager,
The supervisor assigns responsibilities and due dates for any corrective action.
Monthly, the supervisor follows up the action items to ensure completion.
The supervisors’ self-audit checklists are extremely detailed. Weekly, at least
10% of the raw data is checked for correctness and completeness. Instrument
maintenance, calibration, and daily instrument performance checks are
reviewed for compliance. Laboratory contamination is monitored by means of
the method blank data. Quality control check sample data is scrutinized to
ensure that the QC check samples are plotted, in control, and documented. in
addition, duplicate and fortification date are checked for frequency, percent
error and recovery calculation, and proper corrective action to out of control
situations. Documentation techniques are then reviewed to ensure that the
data is defensible, comprehensive, and can “stand alone”. Reagents,
standards, and sample storage/compatibility are checked for compliance.
Monthly, supervisors ensure that personnel are sub-sampling according to
corporate SOP; instruments are calibrated and linearity criteria is acceptable and
documented; standards and reagents are prepared and disposed of when
necessary; and control limit trends and biases are being reviewed. Quarterly,
the supervisors are required to perform self-audits on all Standard Division
Practices (SDP’s). A new checklist is developed for any new or updated SDP.
in addition, sample tracking from receipt/Chain-of-Custody to raw data to report
generation is audited to ensure sample traceability. On a semi-annual basis
supervisors observe analysts performing procedures and document the
recertification of analysis.
Many of the self-audit tasks are performed by front-line chemists, technicians,
and administrative personnel. Laboratory chemists and technicians are
responsible for supply inventory, housekeeping, and ensuring that daily QC
samples are within the acceptable range. Immediate action and documentation
of out-of-control situations is expected. Also, on a monthly basis, the chemists
and technicians within a department cross review each other’s iogbooks for
discrepancies. A summary of this review is forwarded to the department
supervisor, who the assigns responsibility to the appropriate analyst as
Field Support Technicians conduct weekly self-audits of waste containers and
waste storage areas. This ensures that waste containers are properly labeled
and stored in secondary containment. The results are reviewed by the Waste

Approval Supervisor who delegates and distributes corrective actions and target
completion dates. As with all self-audit materials, once corrective action is
complete the documentation is filed.
Sample Receipt Technicians perform sample processing self-audits. When a
sample is processed, a pre-acceptance checklist is completed. If a serious
discrepancy is noted, the sample is put on “hard hold” and the information is
provided to the appropriate Customer Service Representative. No analysis can
be performed on these samples until the discrepancy is resolved. There are
other less serious discrepancies that must be resolved by a Customer Service
Representative, but the sample can be processed by the laboratory. In addition,
compliance for submission of blind duplicate and parallel samples is monitored
utilizing a parameter checklist. Sample Receiving Technicians perform a
secondary containment self-audit weekly to ensure that samples are stored by
compatible classes and are in secondary or tertiary containment.
Quality Assurance Administrative Support personnel perform self-audits to
ensure the laboratory’s quest for quality and compliance is meet. By using the
“Final Report QC Check” checklist, commercial date packages are reviewed
with regards to the customer’s request versus the quality of the final product.
Utilizing this process, discrepancies are resolved before submission of the data
to the customer. Monthly, at least 10% of all out-of-control QC data for a
particular lab group is evaluated for proper corrective action and
documentation. The results are forwarded to the Group Supervisor, Laboratory
Director, and QA Manager. In addition, the QA group graphs Standard
Reference Materials, Parallels. and Blind Duplicate results to monitor for trends
and frequent discrepancies. Quarterly, the QA group monitors performance of
fume hoods and local exhaust systems, documenting all results. Also quarterly,
working “Class P” weights are verified against NIST traceable “Class 5”
weights and the results and correction factors are documented. Semi-annually,
the temperature reading of all working thermometers is verified against a NIST
traceable thermometer and the results and correction factors recorded.
Waste disposal decisions are especially critical to the entire organization. Any
analytical error found during a self-audit is immediately reported to the Waste
Decision Group and appropriate corrective action taken. As a secondary check,
each waste disposal decision requires the consensus of two Approvals
Chemists who have reviewed the analytical date independently of the other.
Any errors that occur on the profile or at the point of disposal decision are
tracked by the Waste Approval Manager. Re-occurring errors require a
corrective action plan approved by the Waste Approval Manager.

The success of the self-audit program is evident in many instances. Calculation
errors that would have affected waste disposal decisions have been found and
corrected before any environmental damage or financial expense to the
company could occur. On several occasions method adherence issues were
noted and corrected to achieve method compliance and better comparability of
data. Most importantly, items which carry a higher severity for non-compliance
have been noted and are more frequently reviewed. As the self-audit program
has developed, the number of discrepancies has immensely decreased in
number and severity. Another benefit of the self-audit program has been the
involvement of all employees. It has created an “esprit de corps” among the
staff in that they are allowed and encouraged to locate discrepancies and to be
creative with new, innovative self-audit techniques. Compliance with company
QA/QC policies has consistently been achieved through the self-audit process.
To ensure the continued success of the program, each Self-Audit Module is
reviewed annually.

Ann Rosecrance, Core Laboratories, 10205 Westheimer, Houston, TX 77042
Abstract: The validation of analytical data is important in all environmental measurement activities to
assess the quality of the data generated and verify that method quality control requirements and
project data quality objectives are met. Data validation is the process of determining the compliance
of analytical data with established method criteria and project specifications. This paper provides
guidance for the validation of data from the most frequently used EPA organic and inorganic analysis
methods for ambient air, drinking water, wastewater, solid waste, and hazardous waste. Comparisons
of the quality control criteria for organic analysis methods and inorganic analysis methods are
presented in order to identify common elements and differences in the quality control requirements
between similar methods. A uniform approach to data validation is presented that can be used for
validating data from samples of different matrices analyzed by the various available analytical methods.
The U.S. EPA has developed numerous analytical methods for the determination of organic and
inorganic chemicals in a variety of sample matrices by GC, GC/MS, ICP, and AA techniques. Each
method defines specific requirements associated with application of the method; additional
requirements may be further defined in the associated quality assurance project plan. Laboratory
analysts and data reviewers need to be familiar with the requirements of all of the analytical methods
that are routinely used in order to ensure that the appropriate procedures are followed. With the large
number of analytical methods that are available, it is easy to become confused on the specific
requirements of each method. Further, since environmental sample data may be used as legal
evidence, and data can be rejected if not in compliance with the applicable requirements, it is critical
that sample analyses and associated data are in accordance with the method requirements and project
Data validation activities determine if analytical data are in compliance with the method requirements
and project specifications. Data validation procedures developed by the U.S. EPA and other state
agencies for specific programs are used as standards for data validation. 1 ’ 2 ’ 3 ’ 4 Previous studies have
provided method comparisons and data validation guidance for multiple organic analysis methods for
GC and GC/MS analysis of volatiles, semivolatiles, and pesticides/PCBs in drinking water, wastewater,
solid waste, and hazardous waste. 6 ’ 6 Additional method comparisons and data validation guidance
have been provided for organic methods of analysis for ambient air. 7 This paper provides an overview
of the quality control requirements for several EPA organic and inorganic analytical methods. Presented
are guidance for the validation of data from similar analytical methods and a uniform approach that can
be used to validate data from any organic or inorganic analysis method.
The EPA organic analysis methods for GC/MS analysis of volatiles included in this study are: 524 for
drinking water 8 , 624 for wastewater 9 , 8240 for solid waste 10 , the Contract Laboratory Program (CLP)
Statement of Work (SOW) OLMO1 .0 for hazardous waste 11 , and Method T014 for ambient air 12 . The
EPA organic analysis methods for GC/MS analysis of semivolatiles included in this study are: 525 for
drinking water 8 , 625 for wastewater 9 , 8270 for solid waste 10 , the CLP SOW OLMO1 .0 for hazardous
waste 11 , and Method T013 for ambient air 12 . The EPA organic analysis methods for GC analysis of
pesticides/PCBs included in this study are: 508 for drinking water 8 , 608 for wastewater 9 , 8080 for
solid waste 10 , and the CLP SOW OLMO1 .0 for hazardous waste 11 . The EPA inorganic analysis methods
included for metals analysis by ICP and AA are the following: Methods of Chemical Analysis of Water
and Wastewater 200 series methods 13 ; SW-846 Solid Waste Methods 6010 and 700010; and the EPA
CLP SOW for Inorganics Analysis, Multi-Media, Multi-Concentration 14 .

Data validation has been defined by EPA as a systematic process, consisting of data editing, screening,
checking, auditing, verification, certification, and review for comparing data to established criteria in
order to provide assurance that data are adequate for their intended use. 16 Data validation is
essentially a question and answer process to determine if data meet both the analytical method
requirements and the associated project specifications. The three major questions to assess in
validating data are the following: were the required quality control (OC) elements included, were they
included at the required frequency, and were the required acceptance criteria met.
The recommended approach for validating data from multiple (or single) analysis methods involves the
preparation and use of summaries of the required OC and other criteria for each similar method in use.
For example, comparisons of the quality control and other technical requirements of the volatile
methods are prepared for use in validating data for volatile organic analyses. The major requirements
of each of the analytical methods of interest are then readily available and are compared to other
similar methods in an easy to reference format. Revisions or additional methods can be included as
needed. This approach is straightforward because it is based on the common elements between
methods. The method documents are used for reference when needed to clarify specific requirements.
The data validation process proceeds by following closely to the sequence of the analysis and the
procedures established by EPA for data validation. Data generated from any of the methods are
compared for compliance with the applicable criteria. Using summary charts that provide the required
criteria and checklists that record compliance with the applicable criteria, the data validation process
can be performed effectively and efficiently for multiple analysis methods.
The quality control elements that are subjected to data validation are comprised of the following types
of analyses: method quality control, sample quality control, and other quality control. Method quality
control consists of the analyses necessary for setting up for the sample analyses and the analyses that
are common to the sample batch. This includes instrument tuning (for GCIMS analyses), calibration
standards, blanks, laboratory control standards, spikes, and duplicates. Sample quality control are the
criteria that are specific to each sample. For organic analyses, this includes internal standards,
surrogate spikes, and the identification and quantitation of target analytes and tentatively identified
(library search) compounds. Inorganic sample quality control includes ICP serial dilutions, furnace AA
duplicate injections and post-digestion spikes, and the identification and quantitation of target analytes.
Other quality control consists of additional analyses that are necessary to assess the field and
laboratory procedures and to utilize the data. This includes container certifications, field blanks, field
replicates, detection limit determinations, precision and accuracy determinations, and performance
evaluation sample analyses. A recommended sequence for data validation that addresses each of
these quality control elements is provided in Figure 1.
The following section summarizes the major OC requirements for several EPA GC/MS and GC methods
for the analysis of volatiles, semivolatiles, and pesticides/PCBs. This information is not a replacement
for the reference method documents or the EPA data validation procedures, but it is a management
tool to assist laboratory analysts and data reviewers in keeping track of the requirements of multiple
analysis methods.
Tuning. Tuning/instrument performance checks ensure that GC/MS mass assignments and relative ion
abundances are in accordance with the established method performance criteria. Tuning data are
evaluated for analysis of the correct compound, at the required concentration and frequency, and

within the required relative ion abundance criteria. A summary of tuning requirements for volatiles and
semivolatiles methods are provided in Tables 1 and 2, respectively. A comparison of the relative ion
abundance criteria for BFB and DFTPP for each of the methods are presented in Tables 3 and 4,
respectively. Tuning data that do not meet the required relative ion abundance criteria should be
evaluated to determine if the deviation is significant and would impact the sample results.
Initial Calibration. Initial calibration checks ensure that acceptable qualitative and quantitative data
could be generated at the initiation of the analysis. Initial calibration data are evaluated for analysis
of the required analytes, at the required number of levels and concentrations, at the required
frequency, and within the required response factor and linearity criteria. Pesticide/PCB data are also
evaluated for chromatographic resolution, the acceptability of retention time window determinations,
and DDT/endrin breakdown. A summary of initial calibration requirements for volatiles, semivolatiles,
and pesticides/PCB methods are provided in Tables 1, 2, and 5, respectively. Initial calibration data
that do not meet the required criteria should be evaluated to determine if the deviation is significant
and would impact the sample results, and if qualification of the data is needed.
Continuing Calibration. Continuing calibration checks ensure that the qualitative and quantitative
measurements established in the initial calibration could be met subsequent to the initial calibration.
Continuing calibration data are evaluated for analysis of the required analytes, at the required
concentrations, within the required frequency, and within the required response factor and precision
criteria. A summary of continuing calibration requirements for volatiles, semivolatiles and
pesticides/PCB methods are provided in Tables 1, 2, and 5, respectively. Further details on calibration
requirements are provided in the reference methods. Continuing calibration data that do not meet the
required criteria should be evaluated to determine if the deviation is significant and would impact the
sample results, and if qualification of the data is needed.
Method Blanks. Method blanks measure background contamination to ensure that contamination does
not interfere with the analysis and sample data. Method blank data are evaluated for analysis of the
correct source of material, at the required frequency, and within the acceptable background level for
target analytes. A summary of method blank requirements for volatiles, semivolatiles, and
pesticides/PCB methods are provided in Tables 1, 2, and 5, respectively. In general, the contamination
in the method blank should be no higher than the detection limit or reporting limit. If unacceptable
contamination exists in the method blank, then all associated sample data should be carefully evaluated
to determine if the sample data are affected by the background contamination. If affected, the sample
data should be qualified appropriately.
Spikes and Duplicates. Spikes and duplicates ensure that the analytical performance in specific
samples is within the accuracy and precision specifications that have been established for the method.
Blank spikes determine the recovery of analytes in blank matrices and matrix spikes assess the effect
of the sample matrix on the analytical results. Duplicates provide a measure of the precision of the
sample analysis. Spike data are evaluated for analysis of the required analytes in the required type of
spike, at the required concentrations, within the required frequency, and within the required
acceptance criteria for % recovery. Duplicate data are evaluated for analysis of the required type of
duplicate, within the required frequency, and within the required acceptance criteria for precision (RPD
or % RSD). A summary of spike and duplicate requirements for volatiles, semivolatiles, and
pesticides/PCB methods are provided in Tables 1, 2, and 5, respectively. Spike and duplicate data that
do not meet the required criteria should be evaluated for their impact on the sample results and the
data should be qualified appropriately.
Other Quality Control. Other quality control includes additional analyses associated with verification
of the acceptability of the sampling and analysis procedures. Sample collection devices should be
evaluated for the presence of background contaminants. Field quality control measures should include
field blanks and replicate samples to measure field contamination and sample precision. The analytical
system should be tested to establish that required method detection limits can be achieved and that

acceptable precision and accuracy data can be obtained. Additional laboratory quality control
measures include performance evaluation samples to determine analytical accuracy and method
performance. Data for each quality control analysis evaluated to determine if the applicable acceptance
criteria were met.
Sample Quality Control. Sample data are evaluated for adherence to a number of criteria in order to
determine the acceptability of the sample results. A summary of sample quality control requirements
for volatiles, semivolatiles, and pesticides/PCB methods are provided in Tables 1, 2, and 5,
respectively. Sample collection, extraction, and analysis times are evaluated to determine if technical
holding times were met. Internal standard areas and the retention times in the samples are compared
to those in the corresponding calibration standard to ensure that the instrument conditions were stable
between the analysis of the calibration standard and each sample. Surrogates are included in samples
to determine if the recoveries of non-target analytes were acceptable throughout the preparation and
analysis procedure. Surrogate data are evaluated for analysis of the required number and
concentrations of surrogates, within the required acceptance criteria for % recovery. A comparison
of the acceptance limits for surrogate recovery in volatile, semivolatile, and pesticides/PCB methods
are provided in Tables 6, 7, and 8, respectively.
Found target analytes in samples are evaluated against the corresponding analytes in the calibration
standard to ensure that the retention times or relative retention times are within the acceptance criteria
of the method. Mass spectra for found target analytes are evaluated against the standard mass
spectra to ensure that the major ions present in the standard are present in the sample mass spectra
within comparable relative ion abundances. Mass spectral library searches for tentatively identified
compounds are reviewed to ensure that identifications are acceptable. Quantitative results are checked
for correctness of calculations, the use of the appropriate units, and to ensure that found target
analytes concentrations are within the calibration range. The reported results are reviewed to ensure
that they fully agree with the raw data and that the appropriate quantitation or detection limits were
used for reporting the sample values. Sample data are reviewed for adherence to the associated
project specifications and reporting requirements. Sample data that do not meet any of the required
criteria should be qualified appropriately.
The following section provides a brief summary of the major OC requirements for several EPA ICP and
AA methods for metals analysis. Further details on method requirements are provided in the reference
methods and the EPA data validation procedures.
Initial Calibration. Initial calibration data are evaluated for analysis of the required analytes, at the
required number of levels and concentrations, at the required frequency, and within the required
acceptance criteria and recommended correlation coefficient (i.e., >0.995). A summary of initial
calibration requirements for metals by ICP and AA are provided in Tables 9 and 10, respectively.
Calibration Verification. Initial and continuing calibration verification data are evaluated for analysis
of the required analytes, at the required concentrations, within the required frequency, and within the
required acceptance criteria for % recovery. A summary of calibration verification requirements for
metals by ICP and AA are provided in Tables 9 and 10, respectively.
Blanks. Calibration blanks and preparation blanks are included with metals analyses to check the
background from the analysis and preparation procedures, respectively. Blank data are evaluated for
analysis of the correct source of material, at the required frequency, and within the acceptable
background levels for target analytes. A summary of blank requirements for metals by ICP and AA are
provided in Tables 9 and 10, respectively.

ICP Interference Check Sample (ICS). The ICS checks the interelement arid background correction
factors for the ICP instrument. ICS data are evaluated for analysis of the correct solutions, within the
required frequency, and within the required acceptance criteria (% recovery for known analytes and
values less than the instrument detection limit for not present analytes). A summary of ICS
requirements for metals by ICP analysis is provided in Table 9.
Laboratory Control Standard (LCS). The LCS checks the laboratory performance on the sample
preparation and analysis procedures. LCS data are evaluated for analysis of the correct type of LCS,
for each matrix and method, at the required frequency, and within the required acceptance criteria for
% recovery. A summary of LCS requirements for metals by ICP and AA are provided in Tables 9 and
10, respectively.
Matrix Spikes and Duplicate Samples. Matrix spike and duplicate sample data are evaluated for
analysis of the correct type of spike and duplicate, at the required frequency, with the required
analytes at the required concentration for spikes, and within the required acceptance criteria (%
recovery for spikes and RPD for duplicates). A summary of matrix spike and duplicate sample
requirements for metals by ICP and AA are provided in Tables 9 and 10, respectively.
Other Quality Control. Other quality control includes instrument detection limit determinations, linear
range analyses, and performance evaluation sample analyses. The EPA CLP SOW for inorganics
analysis requires that instrument detection limit determinations and linear range analyses be performed
quarterly. Performance evaluation sample analyses are performed at varying frequencies from quarterly
to each sample delivery group. Data for each quality control analysis are evaluated to determine if the
results were within the required acceptance criteria.
Sample Quality Control. Sample data are evaluated for adherence to sample specific quality control
criteria. Sample data are evaluated to ensure that preparation and analysis holding times were met
(analysis holding times for metals are generally 6 months from sample collection for samples that are
preserved at pH <2, except for mercury which is 28 days), that the target analytes are reported
properly, and that reported values are within the calibration range for AA and linear range for ICP. ICP
methods require that serial dilutions be performed to determine if there are major chemical or physical
interferences in the sample matrix. ICP data are evaluated for analysis of the required sample dilutions
and for the agreement of the original sample with the dilution. Furnace AA methods require that
duplicate injections and post digestion spikes be performed to measure the precision and accuracy of
each sample analysis. Furnace data are evaluated for analysis of the required duplicate injections and
post digestion spikes, for each sample, within the required acceptance criteria (% RSD for duplicates
and % recovery for spikes). A summary of ICP serial dilution and furnace QC requirements are
provided in Tables 9 and 10, respectively.
Data validation activities should be documented on standardized forms such as the data review
checklists provided in Figures 2 and 3 for organic analysis data. The forms should report the
adherence or lack of adherence to each of the required quality control criteria. The agreement of the
raw data and the data report should be noted. Any major deficiencies identified should be documented
in a data validation report describing each deficiency and its potential impact on the sample results.
Non-compliant or questionable data should be qualified with appropriate data qualifiers. Examples of
qualifiers used in EPA Data Validation Procedures are: (R), the results are rejected due to serious
deficiencies in quality control criteria; (J), the associated numerical value is an estimated quantity
because certain quality control criteria were not met; (N), presumptive evidence of presence of
material; (U) the material was analyzed for but not detected; and (UJ), a combination of U and J.’ 2 ’
Finally, the data validation report should include an overall evaluation of the data, in addition to any
recommendations for further action.

Data validation is an integral part of the environmental data generation process and in order to be
efficient and effective, the data validation process must be versatile and straightforward. With the
large number of analytical methods that are available for sample analyses, the requirements of each
method must be readily available. The information presented in this paper summarizes the QC
requirements for several EPA organic and inorganic analytical methods and compares those
requirements to similar methods. Data validation guidance for multiple methods is provided so that a
single approach can be used for validating data from similar analytical methods. This gives laboratory
analysts and data reviewers a management technique for addressing the specific requirements of each
method utilized and for ensuring that OC requirements of the applicable method are met. This
information is not intended as a replacement for the reference methods or EPA data validation
procedures, but is a reference on the quality control and data validation requirements of multiple
analysis methods.
1. Quality Assurance/Quality Control Guidance for Removal Activities; Sampling QA/QC Plan and Data Validation
Proceduree , EPA 5401G-90,004, U.S. EPA. Washington D.C.. April 1990.
2. Laboratory Data Validation Functional Guidelines for Evaluating Organics Analyses , U.S. EPA, 1988.
3. Laboratoçy Data Validation Functional Guidelines for Evaluating lnoraanics Ana!Ises , U.S.EPA, 1988.
4. Standard Operating Procedures for Quality Assurance Data Validation of Analytical Deliverables , New Jersey
Department of Environmental Protection and Energy, TCL-Organics SOP No. 5.A .13, October 1991 and TAL-lnorganics
SOP No. 5.A.02, February 1992.
5. A.E.Rosecrance, “Data Validation Guidance for Multiple Organic Analysis Methods,” in Proceedings of Water Pollution
Control Federation’s Specialty Conference on Analytical Compliance and Data Objectives , Durham, August 1991.
6. A.E.Rosecrance, “Data Verification Guidance for GC and CC/MS Environmental Analyses,” in Proceedings of the 1992
HazTech International Environmental Conference , Houston, February 1992.
7. A.E.Rosecrance, “Data Validation for Guidance for Ambient Air Methods”, presented at the Measurement of Toxics
and Related Air Pollutants Symposium, cosponsored by U.S. EPA and Air & Waste Management Society, Durham. May
8. Methods for the Determination of Organic Compounds in Drinking Water , EPA 600/4-R/039, U.S. EPA, Cincinnati,
December 1988, Methods 524, 525, and 508.
9. Methods for Organic Chemical Analysis of Municipal and Industrial Wastewater , U.S. EPA, Appendix A to 4OCFR Part
136, Vol. 49, No. 209, October 26, 1984, Methods 624, 625, and 608.
10. Teat Methods for Evaluating Solid Waste. Physical/Chemical Methods , SW-846, U.S. EPA, Washington D.C., Third
Edition, September 1986, Methods 8240, 8270, 8080, 6010, and 7000.
11. Statement of Work for Organics Analysis. Multi-Media. Multi-Concentration , U.S. EPA Contract Laboratory Program
OLMO1.O. August 1991 Revision.
12. W.T. Winberry, Jr., N.T.Murphy and R.M.Riggin, Compendium of Methods for the Determination of Toxic Organic
Compounds in Ambient Air , EPA 600/4-89/017. U.S. EPA, Research Triangle Park, June 1988, Methods 1013 and
13. Methods of Chemical Analysis of Water and Wastewater , EPA 60014-791020, U.S. EPA, Cincinnati, Revised March
1983, Section 200.0 and Method 200.7.
14. Statement of Work for Inorganics Analysis. Multi-Media Multi-Concentration , U.S. EPA Contract Laboratory Program,
ILMO1 .0.
15. Interim Guidelines and Specifications for Preparing Quality Assurance Proiect Plans , U.S. EPA, Washington D.C.,
QAMS-005-80, December 1980.

IC: Levels
Criteria (%RSD)
Minimum RRF
Criteria (%D)
IS Area
BLK: Frequency
SPIKES: Frequency
3 + blank
<5% RSD
8 hrs
± 30
CC or
% of last
±50% IC
12 hrs
RRT ±0.06
Ions >10%
12 hrs
In Control
12 hrs

50 ng DFTPP
5 ng DFTPP
50 ng DFTPP
50 ng DFTPP
50 ng DFTPP
8 hrs
12 hrs
12 hre
Table 4
Table 4
Table 4
Table 4
Table 4
IC: Levels
Criteria (%RSD)
<30% (13)
Min4mum RRF
0.050 (4)
0.01-Mm value
CC: Frequency
8 hrs
12 his
12 hrs
Criteria (%D
±30% (13)
± 25.0%
IS Area
± 30% of last
CC or ± 50% IC
-50 to +100% of
last CC
BLK: Frequency
1 fbatch
1/group ext.
<10 ng/cart.
Area -50 to
RI ±30sec
3 @ 5 ugIL
3 @ 100 ug/L
6 @ 100-200
8 @75-150 ug/L
Table 7
Table 7 —
RRT ±0.06
3 ions ±15-25%
RT ±30sec
3 ions ±20%
RI ±3Osec
3 ions ±20%
RRT ±0.06
Ions >10%
RRT ±0.06
Ions >10%
NS: Not Specified; IC: Initial Calibration; CC: Continuing Calibration;
C : Days from Collection; R : Days from Receipt; E’: Days from Extraction
Note: For more detailed information, refer to the corresponding method document.

50 % of mass 95
75 % of mass 95
96 % of mass 95
173 % of mass 174
174 % of mass 95
175 % of mass 174
176 % of mass 174
177 % of mass 176
51 % of mass 198
68 % of mass 69
70 % of mass 69
127 % of mass 198
197 % of mass 198
199 % of mass 198
275 % of mass 198
365 % of mass 198

IC: Levels
3 (1 for multi-
Criteria (%RSD)
<10.0-15.0% -
DDT/End ,in
CC: Frequency
Beg. and end
12 hrs
Criteria (RPD)
i 20%
± 25.0%
± 0.02 mm of
mean RI
BU(: Frequency
NS: Not Specified; IC: Initial Calibration; CC: Continuing Calibration;
C : Days from Collection; R : Days from Receipt; E : Days from Extraction
Note: For more detailed information, refer to the corresponding method document.

1,2-Dich loroethane -d4
To luened8
N itrob.nz.ne-d5
43-1 16
30-1 15
43-1 16
1,2Dich lorobenzene-d4
80-1 20
2.4,6 .Tribromoph.nol
4 .4-Dich lo robiphenyl
I.trachloro-m .xy lene
Lab Limits
Dthuty lcb$orendate
Lab Limits
NS Not Specified

METHOD 200.7
Initial Calibration
2: 1 standard and
a calib. blank
Calibrate and check with
2 stds. and blank
Calibrate with 1 standard
(mm) and a blank
• Frequency
Daily or every 24 hours
• Criteria
± 5% of true value
Calibration Verification
Mid-range standard
Mid-range standard
Mid-range standard
• Frequency
Beg., end and every 10
samples or every 2 hrs
Every 10 samples
and at end
Every 10 samples
• Criteria
90-110% Recovery
90-110% Recovery
95-105% Recovery
Other Standards
2x CRDL or IDL
Highest mixed std.
Highest mixed std.
• Frequency
Beg. and end of each sample
run or 2 per 8 hr shift
Before sample analyses
Before sample analyses
• Criteria
EPA QC limits
95-105% Recovery
95-105% Recovery
Calibration Blanks
• Frequency
Beg., end, and 10% of
samples or every 2 hours
Every 10 samples
and at end
Every 10 samples
• Criteria
±3 SD of mean value
±2 SD of mean value
Pr.paration Blanks
• Frequency
1 /SDG/digestion batch
• Criteria
Laboratory Control Standai’d
• Frequency
1 /SDG or dig.batch/matrix
Each IC and weekly
Each IC and weekly
• Criteria
80-120% Recovery
90-110% Recovery
95-105% Recovery
Matrix Spike Sample.
• Frequency
5% or 1 /SDG/matrix/level
5% or 1/batch
New sample matrix
• Criteria
75-125% Recovery
75-125% Recovery
90-110% Recovery
Duplicate Samples
• Frequency
5% or 1 fSDG/matrix/ )evel
5% or 1/batch
• Criteria
± 20% RPD
± 20% RPD for values
>lOx IDL
Interference Check Sample
• Frequency
Beg. and end of each run or
2 per 8 hr shift
Beg. and end of each run or
2 per 8 hr shift
Beg., end & periodic
• Criteria
80-120% Recovery
80-120% Recovery
±1 .5x SD of mean value
Serial D ution
• Frequency
• Criteria
1 /SDG/matrixilevel
5x Dil. within ± 10%
New sample matrix
4x DiI. within ± 10%
New sample matrix
Dilution within ±5%
NS Not Specified

Initial Calibration
4: blank and 3 standards
4: blank and 3 standards
4: blank and 3 standards
• Frequency
Daily or every 24 hours
• Criteria
±5% of true value
± 10% of true value
Calibration Verification
Mid-range standard
Mid-range standard
At or near MDL
• Frequency
Beg., end, and every 10
samples or every 2 hrs
Every 10 samples
Every 20 samples
• Criteria
90-110% Recovery
80-120% Recovery
90-110% Recovery
Other Standards
2x CRDL or IDL
• Frequency
Beg. of each sample run
• Criteria
EPA QC Limits
Calibration Blanks
• Frequency
Beg., end, and every 10
samples or every 2 hours
After each calibration
After each calibration
• Criteria
Preparation Blanks
• Frequency
1 /SDG or digestion batch
Each digestion batch
Each digestion batch
• Criteria
Laboratory Control Standard
• Frequency
1 /SDG/matrix
After each calibration
• Criteria
80-120% Recovery
90-110% Recovery
Matrix Spike Samples
• Frequency
5% or 1 /SDG/matrix/level
5% or 1 Thatch
• Criteria
75-125% Recovery
85-1 15% Recovery
Duplicate Samples
• Frequency
5% or 1/SDG/matrix/level
5% or 1/batch
10% or 1/batch
• Criteria
±20% RPD
±20% RPD
Furnace QC
• Frequency
• Criteria
Duplicate injections on all;
Post-digestion spikes on all
samples,_blanks and LCS
Duplicate injections:
±20% RSD or CV;
Spikes: 85-1 15% Recovery
NS Not Specified

GC/MS Tunmg
b ltial Calibration
Contlnukig Calibration
Method Blanks
Laboratory Control Standards
Contakier Certification
Field Blanks
Field Replicates
Detection Linit Determmation
Precision and Accuracy Demonstration
— Holding Tunes
— Internal Standards
— S irogates
— Target Analyte Identification
— Target Analyte Quantitation
— ICP Serial Dilution
— Furnace AA Duplicate Injection and
Post Digestion Spies

• R.q . /r.d_conworn.d?
• Cor . .oI h.qu.noy?
• Cinn.l. not?
• All .n.lyt.._pr.000t?
• R.qolr.d 11.1.?
• Cot .. .! Ir.quenoy?
• CdI.tl. or. !?
• C.rr.ot .ono.n(r .tion?
• All onllrt•• pno.nt?
• Cur,..! trequ.noy?
• CdI . .Ia nit?
• Corroct Sour..?
• Corr.ot fr.quoncy?
• Criterl. m.t?
• All .n.Iyt.._pr...nt?
• RoquIr.d conc.nt,.tIon?
• C.,,..! frequency?
• Critirlo met?
• R.qul,.d_typ.?
• Conic! fr•qu.ncy?
• Crlt.rl. mit?
• R.qolr.d Ir.quency?
• Ctit.!l. m.t?
• P ,ep.r.tlon holding tim.. mit?
• An.Iy .i . holding tin . .. root?
• Ar... wlklr ln lintit.?
• RRTo wltldn SntiM?
• R.q.dr.d p&oo lnoIod .d?
• R.oencrl.. witido Shot.?
• R.orwiy .u wltldn lintit.?
• RT.IRRT. wIthIn lintiti?
• MS oonrg.n.bue to otand.,d?
• ColorMtlons oort.ot?
• Conoentr dor* wIthin rang.?
• M . o, non-Igit .n.Iyt.
p.á.Id.n !* lhId?
• Id.nt lficolion . OK?
• C.IOtiI.t loni oan.ot?
• Ag , ...wlthrawddia?
• Appropd.t. QLinOLI?
0 ,
Figure 2. Data Validation Checktist for
Method Quality Control Review
Figure 3. Data Validation Checklist for
Sample Quality Control Review

Diann Sims Dwight , Peggy Zawodny, U.S. Environmental Protection Agency,
Region III, Central Regional Laboratory, 839 Bestgate Road, Annapolis,
Maryland 21401.
Abstract: Performance evaluation (FE) samples have been historically
employed to assess laboratory proficiency and to validate analytical
methods. Periodic use as part of a laboratory quality assurance program
provides indicators of analytical performance and analyst proficiency.
The FE sample is also used as a component of certification and
accreditation programs. Based on study results that are summarized in
this paper, results of matrix specific FE samples submitted and analyzed
with environmental samples can indicate systematic error that is not
apparent in routine precision and accuracy measurements. Example cases
will be presented to demonstrate the effectiveness of PE samples as an
external quality assessment tool. The cases will show that PE sample
results can be evaluated with respect to a specific sample batch and the
associated data quality objectives. Case study results confirm that the
PE sample data are effective when used to diagnose and verify the
analytical performance and capability demonstrated with a given sample
batch. This substantiates that the data quality achieved satisfies data
quality requirements.
Performance evaluation samples are defined as certified materials with
established limits of uncertainty. The composition and concentration of
constituents are unknown to the analyst. An environmental sample is
collected from an environmental source and may be of any material or
component. The sample serves to characterize or represent the
environmental condition of interest.
Historically, FE samples have been employed to assess laboratory
proficiency and to validate analytical methods. FE sample studies are
used as a component of certification and accreditation programs.
Generally, the sample identity is known to the participating
laboratories but the sample formulation is unknown. The results of FE
studies are used to estimate bias, and demonstrate lab performance and
analyst proficiency.
However, to be able to assess environmental data quality, it is
necessary to determine routine performance with respect to the ability
of the method to recover the analyte of interest within a sample batch.
An analytical sample batch is a group of samples processed together and
considered to be uniform in the properties upon which the measurement
system is based. The results of matrix specific FE samples submitted
and analyzed with environmental samples indicate systematic error that
is not apparent in routine precision and accuracy measurements.

The example cases demonstrate the effectiveness of the PE sample as an
external quality assessment tool. FE sample results can be evaluated
with respect to a specific sample batch and the associated data quality
objectives. Case study results confirm that the PE sample data are
effective when used to diagnose and verify the analytical performance
and capability demonstrated with a given sample batch. This
substantiates that the data quality achieved satisfies data quality
PE samples for this study were secured from the ICF Quality Assurance
Technical Support (QATS) Lab in Las Vegas, Nevada which is under
contract to EPA. The samples are provided to EPA requestors for
inclusion in batches of Superfund environmental samples. The QATS lab
maintains the inventory and ships samples to the designated laboratory
adhering to Agency procedures for shipping and Chain-of-Custody. Sample
composition is verified by using multi-laboratory studies and method
performance data.
Prior to scheduling of P.E. sample shipments to designated laboratories,
an EPA scientist reviews the site history, existing data summaries, and
the data quality requirements. With this information, the PE sample is
selected based on contaminants of concern and the known or suspected
concentration range. The sample is also matrix matched to the extent
possible. When data are generated in support of regulatory requirements
(e.g. SDWA), it is imperative that the compounds and concentration
ranges of the FE are targeted to the action levels.
Data from three laboratories performing analytical work in support of
Regional environmental investigations were used in the study. PE
samples were shipped directly to the laboratories from the QATS lab.
The shipments and accompanying documentation included method
specifications, sample preparation and analysis instructions reporting
requirements, and the analysis due date. The laboratory also received
instructions to analyze PEs with environmental samples from the
specified CERCLA site.
The FE sample results were initially assessed with PEACTOOLS, a software
program developed by the QATS lab. PEACTOOLS does not assign numerical
scores. The results are reported relative to whether the concentrations
fall within acceptance criteria. Acceptance criteria are based on the
calculated mean and measured variance from multi-laboratory studies.
The summary of results also indicates whether the laboratory failed to
report all components of the sample (Misses) and denotes compounds
reported by the lab that were not present in the sample (Contaminants).
The acceptance limits are based on three standard deviations around the
mean value generated from multi-laboratory studies.
Because the PE sample is selected based upon site specific data quality
requirements the greater focus of the overall performance assessment is

placed on the contaminants of concern and their degradation products.
If the initial assessment indicates significant error, a check of the FE
sample selection, shipping, and PEACTOOLs data entry is made. If no
errors are noted in any of the processes, the laboratory is asked to
perform a corrective actions study on the data and systems that
generated the data associated with the PE sample. If the corrective
actions study does not yield findings which account for noted errors,
other investigative tools may be used. These tools may include third
party data review, CC/MS tape audits, and on-site audits.
Case Study One:
Laboratory A., received FE samples for analysis of Volatile Organics,
Base Neutral Acids, and Inorganic fractions. The laboratory was
operating under a Quality Assurance Project Plan reviewed and approved
by several EPA regions.
The initial assessment of PE sample results showed that reported
concentrations for Volatile Organics and Base Neutral Acids were
acceptable. PE sample results for Inorganic Metals indicated that four
analytes were present but not reported and that of the nine analytes
detected, seven concentrations were outside of acceptance windows.
Corrective actions findings indicated that calibrations had been done
improperly due to inaccurate calibrants and that some values had been
transposed during reporting. Systematic errors were incurred with the
use of inaccurate calibrants. Because other quality control criteria
(e.g. Matrix Spikes, Duplicates, continuing calibration checks) were
within acceptance range, the error was unknown throughout the process.
Any data generated and calculated using an incorrect calibration are
questionable. Re-analysis of all affected samples was required.
Case Study 2
Laboratory A 2 received an aqueous PE sample for analysis by CLP Low
Concentration Inorganic Statement of Work (112101.0). The scored results
showed that all reported concentrations were outside acceptance range.
Six of nine reported analytes had results that were outside the lower
acceptance limit (low bias).
The laboratory response was that the corrective action study revealed no
problems and that all reported concentrations were based on verified
calibrations. An initial examination of the data package indicated that
all quality control criteria were met. However, further review of the
data revealed that the PE sample was not included on the digestion log.
The laboratory verified that the sample had not been digested.
Since all associated quality control data were acceptable, this
indicates that sample results generated at the time of analysis are

credible. The error was probably confined to the inability to follow
directions specified on the PE instruction sheet. The reliability of
the environmental sample results was confirmed by data from split
samples from an independent laboratory.
Case Study 3
Lab A 3 was selected to analyze samples for an environmental data
collection activity that was expected to last for approximately one
year. Three major sampling events were projected and PE samples were
scheduled to coincide with the arrival of each sample delivery group.
Data collection activities, including data validation, are expected to
be concluded by December of 1992.
The laboratory reported a significant number of compounds inaccurately
for the PE volatiles fraction and the metals on two occasions. The
reported results for the pesticides were inaccurate with a low bias
consistently for all PE samples. The laboratory conducted a corrective
actions investigation and noted problems with the instrumentation and
the source of the calibrants.
The laboratory has been required to correct the identified deficiencies
and analyze PE samples to determine whether the sources of error were
accurately defined. Pending successful re-analysis of the PE samples,
an assessment of all affected data generated through the life of the
project must be made. This assessment will aid the data users in
determining whether data quality requirements were met and if data use
was compromised.
In each of the case studies presented, data review of the environmental
sample results showed that associated routine quality control criteria
were met. However, the PE sample results, serving as external
performance checks, indicated systematic error including uncorrected
bias measurements. Study data illustrates that PE sample data are
effective in verifying analytical performance and capability associated
with individual sample batches as well as data generated during multiple
sampling event projects. PE samples are effective external performance
evaluation tools and can be beneficial in assessing whether the data
quality achieved during environmental data collection activities
satisfies data quality requirements.

L. M. Tomczak , Sr. Staff Scientist, J. C. Gore, Technologist I, and J. W.
Nixon, Technologist III, Radiological Environmental Monitoring, WEMCO* and
H. B. Spitz, PhD., J.J. Cardarelli, Lia Dai, Ning Liu, and Sue McGimpsey
Graduate Students, University of Cincinnati, Cincinnati, OH 45221-0072.
Westinghouse Environmental Management Company of Ohio* Fernald
Environmental Management Proj ect P.O. Box 398704, Cincinnati, Ohio 45239-8704.
ABSTRACT:The Radiological Environmental Monitoring (REM) group at the Fernald
Environmental Management Project is involved in two practices that will result
in the improved quality assurance of collected data (1) Quality Assurance
Program for Manually Entered Data, and (2) Electronic Data Transfer. The
first practice focuses on adopting strict quality assurance guidelines for
manual database entry. The second practice focuses on electronic data transfer
from the recording instrument in order to reduce the manpower normally required
for manual data entry.
The application of these two practices can enhance any data collection program
where instruments with electronic memories and a signal output are utilized.
Organizations employing either or both of these practices, as applicable,
can strengthen the quality and efficiency of their data collection program.
The use of these practices can assist in complying with Quality Assurance
requirements under ASME NQA-1, RCRA, CERC1A, and DOE Order activities.
Data are currently obtained from field installed Pylon AB-5 radon monitors,
configured to print data to a paper tape. The data are entered manually into
a PC database on a weekly basis. To ensure that data are correctly entered,
an oversight process has been developed. Basically, a two-person team is
assigned the task of ensuring correct data entry. One individual performs
the initial data entry and confirms the entered data. The second individual
is responsible for verifying that all of the data have been correctly entered.
Automated computer checks are also conducted for data validity between recorded
intervals due to light leaks etc. and the detection of outliers.
This portion of the paper will discuss the protocol that is followed to ensure
correct data entry, the process of resolving incorrect entries, the validity
of the data collected, and documentation of the data entry process. The
results of this protocol lead to a data management system that is capable
of successfully meeting the requirements of an external audit.

Fifteen continuous radon monitors have been installed at the DOE Fernald
Environmental Monitoring Project (FEMP) at both on-site and off-site locations
to measure what, if any, contribution to the natural radon background is
made by sources of radon located on the Fernald project. On-site outdoor
locations for each of the samplers have been selected to provide representative
measurements of radon that are either close to the sources of emission (K-65
silos that contain byproducts from the Manhattan Project) or near buildings
and areas which are occupied by the majority of the workforce. Several
monitors also measure radon at the FEMP fenceline. One background monitor
is located approximately 13 miles from the project in the direction of the
prevailing wind. An additional background monitor is also being considered.
Each monitor prints the results of hourly measurements and also stores the
data in the instrument computer memory. Each result consists of four fields
of data: sequence number, hour (24 hour clock), instrument response
(counts/hour), and radon concentration (pci/i). Data is retrieved from each
instrument by data collection personnel on a weekly basis.
Data collection personnel initial each data paper tape and reset the instrument
sequence number to initiate another run. An identification number, calibration
factors, and other data constants are printed on the data tape whenever the
instrument is reset. The instrument response is also conducted using a check
source on a weekly/monthly basis per established protocol.
The radon collection instruments are housed in environmental enclosures to
protect them from the direct adverse effects of the environment but are not
heated or sealed in any manner. They operate 24-hours a day and are expected
to function properly in all types of weather and generate 168 hourly readings
each week.
The instruments sample radon in the ambient air based upon the diffusion
principle and, thus, have no moving parts or pumps. Other than weekly
inspection by the data collection personnel to retrieve data, reset the
instrument, and insure that the instrument is still operating, no other
operator attention is provided or required. Instrument re-calibration is
scheduled on an annual basis contingent upon the satisfactory instrument
source check results.
A formal quality assurance program will improve the accuracy and reliability
of data, by establishing formal procedures and review processes to reduce
the occurrence of systematic interferences. The quality assurance process
will make it possible to identify problems and implement corrective actions
before errors are introduced into the data analysis process. Data accuracy
and reliability is essential because these monitoring results are associated
with radiological protection programs and governmental compliance requirements.

Any instance of a failure of data quality is not tolerable. Therefore,
formally documented quality assurance activities and responsibilities will
minimize the likeithood of an unknown interference which could sacrifice
data accuracy or reliability and permit propagating an unknown bias into
the data analysis.
The occurrence of random or intermittent electrical failures or other
systematic instrument-related problems are infrequent but non-negligible.
At times, data printed by the instrument is not completely legible or the
paper tape may jam or run out. Data collection personnel may also fail to
properly identify each data print-out or accurately reset the instrument.
In addition, a few hours of radon measurement data collected immediately
following instrument performance checks using the standard source is
censored/rej ected, since radon measurements cannot be performed immediately
following use of the standard source due to the persistence of the photo-
multiplier tube.
After collection, each data print-out must be independently previewed to
insure that all the data is complete, legible, and accurately labeled for
proper identification. The process of data preview is an important element
of the quality assurance program and should be as comprehensive as possible.
Following data preview, the validity of the data can be evaluated. Data
is invalid if the results fail to represent actual radon concentration
Electronic noise or other detector interferences can produce results which
can be differentiated from valid data by using statistical tests. These
tests are based upon the known radiological characteristics and the physical
behavior of radon. Comparisons are also made with other results using
instruments in similar locations. Activities associated with data preview
and tests to insure data validity are independent and provide two separate
checks of data quality. However, feedback between these activities is needed
to resolve outstanding data discrepancies and to identify systematic errors
which may not have been addressed in the overall quality assurance plan.
METHODOLOGY (Specific Elements of Data Oual .ity Assurance)
Figure 1 illustrates the elements of the radon monitoring project for which
a data quality assurance program was developed. Most of the activities and
tasks illustrated in Figure 1 are documented via procedures. These procedures
assign responsibilities and duties, and provide instructions for properly
accomplishing each of these tasks. Major emphasis of the data quality
assurance program is directed toward data validity activities since these
tasks involve quantitative numerical analyses using computer programs and
technically-based formal procedures.

Data collection personnel have procedures for performing instrument checks
and retrieving data from each of the instruments. They observe instrument
operation and determine that the data output tape appears normal. They also
initial the data tape and reset the instrument. Should the instrument appear
to be malfunctioning, they will notify management and initiate a service
request from the instrumentation group. Periodically, the data collection
personnel perform an instrument response check by removing the radon sampler
and attach a check source. Following the performance check, the data
collection personnel re-attach the radon sampler to the monitor and reset
the instrument.
Data Preview
The process of data preview is performed by a member of the technical staff
familiar with both the technical performance characteristics of the instruments
and the technical requirements of the radon monitoring program. Data preview
involves insuring that the data received from each of the fifteen radon
monitors is legible, properly identified, and complete. No evaluation of
the radon monitoring data is required at this point, other than to insure
that each instrument has provided the expected amount of data and that it
can be easily interpreted for data input. Whenever the data collection
personnel perform an instrument performance check, data preview will include
updating the response check control charts for each instrument and initiating
remedial action whenever an instrument fails to perform within specifications.
Data printed by the monitoring instrument is photocopied to improve legibility
and to provide a format that is more suitable for review and archival purposes.
Data preview is essential since computer input is performed by an independent
organization that may be unfamiliar with the application of the data or its
meaning. Provisions are made to encourage communications between the data
input group and the data previewer so that questions can be resolved by those
most technically knowledgeable about the radon monitoring program. Likewise,
data previewers have direct contact with data collection personnel should
questions arise about instrument performance.
When the data has been fully previewed it is submitted for data input where
it will be digitized and prepared for quantitative analysis. The individual
performing the data preview will provide a dated signature on the data sheet
to signify that the data has been previewed and provide a reference for future
Data InDut and Validation
The team of individuals that perform data input use a LOTUS 123 spreadsheet
which has been designed in a manner consistent with the format of the data
output produced by the radon monitoring instrument. A new LOTUS file having
a unique name is created every week for each instrument and contains all
appropriate data printed on the data tape from each instrument. The name
of the person who entered the data is also entered onto the spreadsheet.

At least two people are required to perform data input since the data is
entered in duplicate using two independent LOTUS spreadsheets. Data input
errors are identified using a LOTUS command to compare the duplicate files.
Original and discrepant data are compared to the original radon monitor data
records and corrections are made to the LOTUS file. If the error cannot
be resolved, the data in question is returned for additional data preview.
Correction of typographical errors represents the most frequently encountered
source of error that can be addressed by the quality assurance program.
Assuming that the data being entered into the LOTUS 123 file is legible and
easily deciphered, the mechanical process of data entry represents a
significant source of systematic error that can be eliminated using double
data entry.
After the data input errors have been corrected, a listing of the radon
measurement data for all the monitoring instruments is printed and the LOTUS
files are stored on the computer hard drive. Individual disks are also
prepared which contain all the radon measurement data for one week from all
the monitoring instruments in the program. In this manner, duplicate computer
records are maintained using more than one type of computer media so that
the likelihood of any permanent loss of data is minimized. Paper listing
of the radon monitoring data is also archived.
Data validation includes many other provisions for assuring the accuracy
and validity of the radon monitoring data. Results for measurement periods
immediately following performance tests are intentionally discarded since
the process of exchanging the radon detector for the standard source allows
light to enter the monitor and artificially increases the instrument background
for up to three hours after the performance tests. Data censoring associated
with the performance check may occur during data preview otherwise it will
occur during the data validation process.
Occasionally, the instrument will produce one or more spurious results which,
if not censored, would make a significant impact to the results of the radon
monitoring program. Incidents of spurious or questionable data can be
identified as a discontinuous increase or decrease in the data which is
physically not possible based upon the response characteristics of the detector
or the ambient meteorological conditions. Events of spurious results may
appear for one hour or extend for several hours, even for several days.
Most of these events are related to light leaks in the monitoring instrument
or extremes of ambient temperature or humidity which may produce an
intermittent failure in an electronic component.
Statistical tests have been developed to identify data which may be related
to spurious data. The first test for a questionable result involves
calculating whether any result differs from a neighboring result by 100 pCi/i.
It is very unlikely that the radon concentration would change by this
magnitude in one hour. This test is used to locate events in the data set
which require further evaluation. No data will be censored as a result of

this simple test without further review and documentation of the cause of
the event. The frequency of such events may lead to an investigation of
the instrument calibration or other possible remedial actions.
Another statistical test is based upon the characteristic time for the build-up
and decay of the short-lived radon decay products and is used to identify
a series of increasing or decreasing results which may not be wholly related
to the change in the radon concentration. This test is needed for a series
of suspect data having a magnitude which does not exceed 100 pCi/i. The
suspect data will be evaluated for validity by comparison with meteorological
and other radon monitoring data. No changes or censoring will occur without
proper documentation of the changes. This documentation will include at
a minimum: names of individuals involved in the correction of data, date
of the change, data corrected, and the reason for the correction of such
Data Reports
Data Reports and graphical analyses are important elements of the overall
quality assurance program. Graphs are useful for identifying trends and
determining whether the results are meeting expectations. Besides lists
of data and results of statistical analysis, individual reports of discrepant
data will be provided to specifically identify what, if any, data was censored.
Incorporation of the aforementioned activities into a formal quality assurance
program will improve the accuracy and reliability of data, by establishing
formal procedures and review processes to reduce the occurrence of systematic
interferences. The quality assurance process will make it possible to identify
problems and implement corrective actions before errors are introduced into
the data analysis process. Data accuracy and reliability is essential where
any instance of a failure of data quality is not tolerable. Therefore,
formally documented quality assurance activities and responsibilities will
minimize the likelihood of an unknown interference which could sacrifice
data accuracy or reliability and permit propagating an unknown bias into
the data analysis.
The REM group has also developed a process to electronically transfer stored
data. The data are transferred between each Pylon AB-5 field instrument and
a Hewlett Packard portable hand computer, model HP95LX. Later, all the data
is transferred to a PC database as an electronic file for analysis. The
advantage of this system is twofold: (1) Manual data entry errors are
eliminated and (2) considerable data entry time is eliminated.

This portion of the paper will discuss the interface and connector components
that allow this transfer of data from the Pylon to the PC to take place.
Collection of data, albeit an important function, comprises only half of
the environmental monitoring activity. Sometimes it can be just as cumbersome
to appropriately deal with the data that has been collected, as it can be
to collect the data. The Radiological Environmental Monitoring group at
the DOE FEMP facility has identified a solution, to at least take some cf
the dread and monotony out of the voluminous environmental data reduction
and reporting activity, without sacrificing quality.
Much of the preceding section dealt the quality assurance activities associated
with compiling data obtained from the data tape produced by the Pylon AZ-S
unit and preparing an electronic data file for use in accurately reporting
the data collected by the instrument. This section of the paper deals with
the process that is used to electronically transfer the data from the
instrument to an electronic data file that is used in generating reports.
There are essentially two steps in the electronic data transfer process.
The first step is the transfer of an electronic data file from the Pylon
AB-5 unit to a handheld computer in the field. The second step is the transfer
of field data to an electronic data file back at the office on the hard drive
of a PC.
This data is then transferred via disk along with the hard copy data tape
of the Pylon AZ-S data to a data previewer, and then to an independent
organization that produces the environmental reports. This activity should
eliminate data errors and data entry form.
The following is a list of equipment that is needed to complete the electronic
data transfer from the AB-5 Pylon to a personal IBM disk operating system
• Pylon AB-5 Radon Monitor
Pylon Model PPT-l Printer
Pylon Model CI-55 Computer Interface
CPRD (radon detector) or a 300A Lucas cell
DB15 Female A/B Switch Box
DB15 3 foot N.M. Transfer Cable
• DB15 3 foot M.F. Transfer Cable
DB25/9 N.M. Gender Changer
• DEIS N.M. Gender Changer

• HP95LX Computer
• HPF1001A Connectivity Pack
The steps that are needed to transfer the electronic data file from the Pylon
AB-5 unit to a personal IBM disk operating system computer and finally to
a 3.5 inch computer disk is presented in figure 2. Each type of transfer
will be discussed in detail in the appendix.
Once the data is transferred to a disk, the data can be previewed by the
individual assigned the task of quality assurance. A check of data will
then need to be conducted to confirm that the data was correctly transferred.
It will also be necessary to parse the data using this lotus function to
further manipulate the data to work with each of the columns.
By utilizing the electronic transfer of data, considerable time can be saved
in both the areas of manual data entry, and the checks that are required
to ensure that the data is correct. Activities that had previously taken
more than a day can be performed in much less than a day without sacrificing
data quality.
ACKNOWLEDGEMENT - The authors wish to thank Mr. D. W. Muller and Ms. J. A.
Peters for their assistance in the preparation of this paper.
HP 95LX User’s Guide, Hewlett-Packard Co., 1991, Corvallis, OR

Transferring Data From the Pylon Monitor to the HP9SLX Computer
1) Hook up the HPF100A1 Connectivity Pack to the HP95LX and the CI- 55
Computer Interface
Connect the CI-55 to a DB-l5 M.M. Transfer Cable.
2)Prepare the HP95LX for use
Turn on the HP95LX
Press:ON/OFF key
Press: BLUE FILER key
Select PRN (DIR)
Press: CURSOR DOWN key
Press: F3 key
Y key (to delete)
MENU key
COMM key
MENU key
Press S key for SETTINGS selection
Press: U key for USE selection
Press:Cursor Arrow - - -> key over to File (PYL.DCF)
Press: ENTER key
Check the following setup information:
Type Pulse
Press: Q key for QUIT selection
Hold CTRL key and press F-5 key simultaneously
Backspace to DAT and delete by using the c- - - (BREAK) key
Type PRN\filenanie . pm
Press: ENTER key
Delete all PRN files to start off the
day. This will ensure that sufficient
memory will be available on the unit
to collect the data.
Press: Q key to quit
3) Press:

Note: 8 characters max. name
Press: MENU key
Press: C key for CONNECT selection
(at this point set up the pylon to send data)
4) Transferring data from the pylon unit to the HP95LX:
Switch Box
Place the A/B switch to the B position
Pylon AB-5 unit
Press the RECALL key
Press the PROC STEP key, this will show the most current run.
If it is necessary to collect a different run use the PUMP key to change
the first digit and the START/STOP key to change the second digit.
After the RUN is selected press the PROC/STEP key three times.
Press and hold the STATUS key; the display will show the run, cycle
and interval.
While continuing to hold the STATUS key press the PROC/STEP to
start the data transfer.
Release the PROC/STEP and the STATUS key.
When the data transfer is complete the Pylon AB-5 display will show
Switch Box
Reposition the A/B switch to the A position
Disconnect the CI-55 Computer Interface cable from the transfer
cable on the Black Box.
HP95LX Computer
Press: MENU key to exit program
Press: Q key for QUIT selection
5) Checking data transfer
Press: FILER key
Select the PRN (dir) ENTER key
(check if file and data exists)
Press: MENU key
Press: Q key for QUIT selection
Turn off instrument
Press ON/OFF key
Note:Prepare for next station. At the next location it is not necessary
to repeat step #2, but Step #1, the hookup is still applicable.
Data Transfer from the HP95LX to the PC
1) Setting up the HP95LX for data transfer to the PC

Turn on the }IP95LX
Press: OFF/OX key
Press: LOTUS 123 key
Press: MENU key
Press: F key for FILE selection
I key for IMPORT selection
T key for ThAT selection
Arrow to - - -> _PRN% 1 enter
Search for file of interest
Press: ENTER key
2) Saving the file
Press: MENU key
F key for FILE selection
S key for SAVE selection
Type in file name
Press: ENTER key
Press: R key for REPLACE selection
Press ENTER key
Repeat this step to transfer the files to the LOTUS (dir)
3) Setting up the PC for Transfer
Hook transfer cable to the HP95LX and the PC
4) Transferring the data
On the PC select CPACK from the menu screen
Select FILER on PC
Press ALT 2 key
Turn on HP95LX
Press: ON/OFF key
Press: FILER key
Press: F6 on the PC keyboard (remote)
Press: F7 on the PC keyboard (split screen)
Highlight and tag F9 file or files to transfer
Press: F2 (copy) on PC keyboard
Screen will query where to copy files to,
type in c:\123
When complete exit on PC, press: CTRL and F]. to
exit to main screen

On the HP95LX press: CTRL, UP ARROW , and BREAK simultaneously to
return to normal operation
Unplug transfer line from HP95LX
Press: MENU key on HP95LX
Press: Q key for QUIT selection
Press: ALT 0 to exit PC
5) Transfer Data From C Drive To B Drive 3.5 inch Disc
Select LOTUS 123
MENU key *
F key for FILE selection
R. key for RETRIEVE selection
F3 key
Select desired radon file location
Press: ENTER key
MENU key
F key for FILE selection
S key for SAVE selection
Press: Esc three times
Type B: Location **
Press: ENTER key
Return to step * and repeat thru step ** until all locations are
Checking Data Transfer
1) Press MENU key
Press: F key for FILE selection
Press: B. key for RETRIEVE selection
2) Press ESC key two times
3) Type B:\
4) Press ENTER key
5) Scan Data
6) If you want to retrieve data, highlight, locations then press ENTER

Data Vali
Review of date
Read ability
r Maintenance ____
Data Collection
Data input
Statistical Analysis
Graphical Analysis
)C I Invalid data
Discrepency control

HP Connectwity
Pack _________
for data transfer
:: l éAiBSVv’ftChtÔ
B osition
. I idate AB 5
prtnt cyc’e
Prepare for next.
station .
Addition at
Stat ons
Data Transfer
To HP 95 LX
From AB-5 Unit
Data b ansfer
Return A/B switch
to A posit on

Transfer Data FIGURE 2 Con’td
From HP95 LX
Import file HP95LX
into LOTUS 1-2-3
Save the worksheet
in LOTUS 1-2-3
Prepare HP 95LX
and PC for transfer
Transfer file using
FILER option
‘I ,
from PC
* Option
Transfer data to 3.5”
Floppy disk
Review data

Copyright Statement
By acceptance of this paper, the publisher and or recipient acknowledges the U.
S. Government’s right to retain a non—exclusive royalty free license in and to
any copyright covering this paper.
This abstract/paper was prepared as an account of work sponsored by an agency of
the United States Government. Reference herein to any specific commercial
product, process, or service by trade name, trademark, manufacturer, or
otherwise, does not necessarily constitute or imply its endorsement,
recommendation, or favoring by the United States Government or any agency
thereof. The views and opinions of authors expressed herein do not necessarily
state or reflect those of the United States Government, Westinghouse Electric
Company, or any of its wholly owned subsidiaries, or any agency thereof.

Mary K Wolf and Darlene J. Elkins, Lockheed Engtneenng & Sciences Company, Las
Vegas, Nevada 89119 and Edward Kantor, USEPA Environmental Monitoring System
Laboratory, Las Vegas, Nevada 89193.
The GCIMS raw data audit, performed by Lockheed Engineering & Science Company
(LESC), is a quality assurance tool that is used by the USEPA to monitor the quality
assurance and technical performance of the laboratories in the Contract Laboratory
Program. EMSL-LV maintains a GC/MS raw data audit facility that has stand alone
data systems for all of the commonly used data systems in the Contract Laboratory
Program. The quality assurance evaluators, using the electronically stored data,
generate a new set of data using different identification files. This evaluator
generated data is then compared to the laboratory generated data to determine if
there are any discrepancies between the two sets of data. The procedures used in
the review of the laboratory GCIMS raw data will be discussed. A comparison of the
changes and frequency of previously found defects will be compared to defect trends
occurring over the fast twelve months.
Notice :
Although the research described in this article has been supported by the United
States Environmental Protection Agency through contract 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.


39 Paper in,available

Paul Marsden, Science Applications International Corporation,
10260 Campus Point Dr., San Diego, California 92121; Bruce
N. Colby, C. Lee Helms , Pacific Analytical, 6349 Paseo Del
Lago, Carlsbad, California 92009
Method 1311, the Toxicity Characteristic Leaching Procedure,
is used in part of the evaluation process for RCRA controlled
wastes. This procedure can be aplied readily to most solid
wastes and to aqueous wastes but it has proven difficult when
working with oily wastes, such as waste motor oil, in part
because the volatile TCLP analytes cannot be determined using
traditional purge and trap preparation technology. In order
to overcome the purge and trap problems, a study was
undertaken to evaluate headspace analysis as an alternative.
Because significant matrix effects were anticipated, isotope
dilution GC/MS was identified as the most likely means to
generate accurate analytical results. A simpler alternative
to the isotope dilution GC/MS method was also investigated.
It involves diluting waste oil 1:1 with heaxdecane followed by
syringe injection of a 2-uL aliquot into a GC/MS. This
approach requires a modern highly sensitive instrument in
order to achieve the necessary detection limits for analytes
such as vinyl chloride at 0.2ppm. Performance data for both
the headspace and syringe injection methods are presented to
demonstrate their effectiveness with waste motor oil samples.
The relative technical and logistic merits of the two methods
are also discussed.
Quantitative determination of volatile organic analytes (VOAS)
in waste oil presents a challenge to the analyst. The use of
conventional techniques for analysis of volatiles (i.e.,
purge—and-trap [ Method 5030]) in oil generally results in
severe contamination of analytical instrumentation.
Recoveries are poor; traps, transfer lines, and chromatography
columns may become contaminated. System contamination leads
to elevated baselines, hydrocarbon background in subsequent
analyses, and unacceptable blanks. Problems with the purge—
and-trap apparatus may be reduced by diluting a sample in
methanol (“Waste Dilution”, draft Method 3585). However, this
approach is not appropriate for petroleum products that have
poor solubility in methanol. Another disadvantage of waste
dilution with purge—and-trap is that it increases method
detection limits. Headspace and direct injection analysis are
potential alternate approaches to the analysis of oils.
However, headspace (Method 3810) is currently allowed only as

a screening procedure in SW—846. Guidance and performance
data for the use of direct injection is not currently included
in Method 8260.
In order to provide a suitable method in SW-846, the Methods
Section of the Office of Solid Waste (OSW) tasked Science
Applications International Corporation (SAIC) to provide
performance data for the determination of volatiles in oil.
SAIC engaged Pacific Analytical Laboratory in Carlsbad, CA, as
an analytical subcontractor for this project. Data were
collected for both headspace and direct injection analysis of
volatiles in oil. The results of this study indicate that
both headspace/isotope dilution and direct injection may have
application for the analysis of volatiles in oil. This report
provides performance data for both headspace/isotope dilution
GC/MS and direct injection GC/MS (using isotope dilution and
internal standard quantitation).
Headspace apparatus : Headspace analysis was accomplished
using a Hewlett Packard Headspace analyzer (Model 19395A), a
Hewlett Packard Model 5890 Series II gas chromatograph and a
VG Trio—l mass spectrometer. Oil samples analyzed using
headspace were heated overnight (80 oC) in lO-mL vials sealed
with Teflon-faced septa. After heating, a volume of headspace
was drawn automatically into the GC/MS system for analysis by
Method 8266. Carrier flow was diverted through the headspace
sample loop to introduce samples onto the GC column. Column
head pressure was maintained using the backpressure control on
the GC. Some adjustment of flows and pressures were required
to achieve reliable operation.
Direct injection apparatus : Direct injection analysis was
accomplished using a Hewlett Packard Model 5890 Seiies II gas
chromatograph and a VG Trio—i mass spectrometer. Oil samples
were diluted 1:1 (v/v) in
hexadecane and analyzed
using Method 8260. A 2-Ll
injection volume was used. .
The injection port liner
was modified by placing a
1-cm plug of pyrex wool
approximately 50-60 ThIn Figure 1 Modified Injector
down the length of the
liner (towards the oven).
An 053 mm id column was mounted 1 cm into the liner from the
oven side of the injection port, according to manufacturer’s

Gas chromatograi hy : Separation of samples introduced by
either headspace or direct injection was accomplished using a
75 in x 0.53 mm DB-624 column. The carrier gas was helium and
the flow rate was approximately 4 mi/mm. The oven
temperature was programmed from 40 to 260 oC at 8 0 C/minute
after an initial hold of 3 minutes. The GC effluent was
introduced into the mass spectrometer through an 0.25 mm id
uncoated restrictor column that was butt—sealed to the
analytical column with “press—tight” connectors. It was
necessary to “bake out” the oven at 260 oC for 75 minutes
following direct injection analysis to ensure that hexadecane
and high boiling interferences did not interfere with
subsequent analyses. A much shorter bake out was required for
headspace analysis.
Mass spectrometry : Mass spectra were collected using a VG
Trio—i tuned using p—bromofluorobenzene according to criteria
specified in Methods 8260 and 8266. The instrument was
calibrated using five solutions of standards to establish
linearity. The stability of the response was established by
daily comparison of the response factors for the mid-point
calibration solution with the initial calibration curve. The
MS source was maintained at 220 °C. Isotope dilution
calculations were used for 14 method analytes; internal
standard quantitation was used for the remainder. Some direct
injection data was reduced using both internal standard
technique described in Method 8260 and isotope dilution
calculations the described in Method 8266.
Significant mass interferences were observed for two internal
standards, a stable—labeled analog, and some target analytes.
Internal standards 2—bromo-l—chloropropane and 1,4-
dichlorobutane were not used for quantitation due to matrix
interference. Interferences of target analytes included:
iso-butanol (43), n—butanoi (41), 4-methyl-2—pentanone (43,
58), ethyl acetate (43, 45) and pyridine (79, 52, 51 and 50).
Major interferences were encountered for pyridine-d 5 (84, 56,
and 52).
Reagents and Chemicals : Isotopically labeled standards and
target analytes were obtained from Cambridge Isotope
Laboratories and Aldrich Chemicals. All solutions were
prepared from neat isotopically labeled standards prepared in
hexadecane at Pacific Analytical.
A total of 29 chemicals were evaluated as target analytes for
headspace and direct injection analysis (Table 1). Target
analytes included 25 of the 27 compounds listed in Revision 0
of the Toxicity Characteristic Leaching Procedure (TCLP,
Method 1311). Methanol fragments are all too small (m/z < 33)
which results in severe mass spectral interferences. 1,1,1—

ComDound Name CAS No. limit
Acetone 67-64-1
Benzene 71-43-2 0.5
Benzene—d 6 1076-43—3
n-Butanol 71-36-3
iso-Butanol 78-83-1
Carbon tetrachioride 56-23—5 0.5
Carbon tetrachloride- 13 C 32488-50-9
Carbon disulfide 75—15-0
Chlorobenzene 108-90-7 100
Ch lorobenzene—d 5 3114-55-4
Chloroform 67-66-3 6.0
Ch loroforin-d 1 835-49-6
1,4—Dich lorobenzene 106-46—7 7.5
1, 4-Dichlorobenzene-d 4 3855-82-1
1,2-Dichioroethane 107-06-2 0.5
1,1-Dich].oroethene 75-35-4 0.7
1, 1-Dichloroethene-d 2 22280-73-5
Diethyl ether 60-29-7
Ethyl acetate 141-78-6
Ethy lbenzene 100-41-1
Ethy lbenzene—d 10 25837-05-2
Hexachioroethane 67-72-1 3.0
Hexach loroethane- 13 C 93952-15—9
Methylene chloride 75-09-2
Methyl ethyl ketone 78-93-3 200
4-Methyl-2-pentanone (MIBK) 108-10-1
Nitrobenzene 98-95-3 2.0
Nitrobenzene-d 5 4165-60-0
Pyridine 110—86—1 5.0
Pyridine—d 5 7291—22—7
Tetrachioroethene 127-18-4 0.7
Trichiorofluoromethane 75-69-4
1, 1,2-Trichiorotrifluoroethane 76-13-1
Toluene 108-88-3
To luene-d 5 2037-26-5
Trichioroethene 79-01-6 0.5
Trichloroethene-d 1
Vinyl chloride 75-01-4 0.2
o-Xy lene 95-47-6
o—Xy lene—d 10 56004—61—6
m-Xy lene 108-38-3
p-Xy lene 106-42-3
p—Xy lene—d, 0 41051—88—1

Trichioroethane was not included because of its propensity to
dehydrochiorinate. Loss of hydrogen chloride from 1,1,1-
trichioroethane produces 1,1-dichioroethylene, which results
in spurious recoveries for both compounds.
Clean 30 weight motor oil (Citgo) was purchased from a local
7-11 store; residual volatiles were removed by heating a
volume of 200 ml overnight at 800 C. The resulting oil had a
small concentration of toluene (7 ppm). Benzene,
ethylbenzene, xylenes, pyridine and other target analytes were
not detected after heating (estimated detection limit 0.05
ppm). Used motor oil was collected during oil changes of
passenger automobiles. The contaminated oil was stored in a
garage for several months prior to the initiation of this
project. The used oil had 20-300 ppm of BTEX compounds and
Principles of headspace analysis : The concentration of
volatile constituents in solids and liquids can be measured by
analyzing the vapor phase (headspace) above a sample. As
headspace analysis involves a separation of the target
analytes from a condensed phase, it is well suited for the
analysis of highly complex samples like oil. Concentration of
target analytes in headspace can be related to their
concentration in a solid or liquid sample by this equation:
CH = Analyte concentration in the headspace
K Distribution Coefficient
Cs Analyte concentration in the sample
The distribution coefficient between sample and headspace is
extremely matrix-dependent.
IsotoPe dilution quantitatiofl As target analytes and their
labeled analogs have the same distribution coefficients,
isotope dilution seems well suited for headspace analysis.
Isotope dilution is a GC/MS technique in which the ratio of
the quantitation ions from environmentally incorporated target
analytes (1u 1 /z) and spiked isotopically labeled analogs (m 2 /z)
are compared in order to calculate analyte concentrations.
The relative response (RR) of each target analyte (m 1 /z) and
its analog (in 2 /z) is established during calibration by
analyzing different concentrations of target analytes.
Headspace isotope dilution analysis allows correction for
differences in distribution coefficients.

Method linearity : The linear range of the headspace method
was established by analyzing mixtures of standards prepared at
eight concentrations in hexadecane. Calibration standards of
target analytes were prepared at 0.2, 0.5, 2.0, 5.0, 25, 50,
100, and 200 Mg/mi, labeled analogs were present at 5.0 g/nhl
in each calibration standard. Isotope dilution response
factors were determined for 14 of the 15 targets for which
isotopically-labeled analogs were added. Mass spectral
interferences precluded the use of pyridine-d 5 for this
application. Internal standard response factors were
calculated for pyridine and the remaining 14 target analytes
for which suitable isotopically-labeled analogs were not
available. Table 2 provides the linear range, the average
response factors over the linear range, and the percent
relative standard deviation for the calibration curve of each
System Stability : Stability of the measurement system was
monitored by using a daily continuing calibration solution of
5 ppm. The continuing calibration solution was prepared with
all 29 target analytes and 14 isotopically labeled analogs in
hexadecane. Average percent recoveries for 14 of the target
analytes with isotopically-labeled analogs were calculated
using isotope dilution. Recoveries of pyridine, the 14
analytes without analogs and the 14 isotopically labeled
analogs were calculated using the internal standard technique.
The average recovery of most compounds were 80—120%. Average
recoveries outside of the 80—120% window were obtained for
1, 1-dichloroethene-d 2 , pyridine-d 5 , pyridine, 1,4-
dichlorobenzene—d 4 , nitrobenzene-d 5 , nitrobenzene, carbon
disulfide, tetrachioroethene, trichiorofluoromethane, iso—
butanol, and n—butanol. RSDs of greater than 20 % were
obtained for l,4—dichlorobenzene--d 4 , nitrobenzene—d 5 , and
nitrobenzene. As a result of the recovery correction inherent
in isotope dilution, RSDs for each of the 14 target analytes
determined using isotope dilution were smaller than the RSDs
than the isotopicaily—labeled analogs quantitated using the
internal standard technique.
Headspace isotope dilution performance : Performance of the
headspace isotope dilution technique was established by
analyzing two sets of seven replicate samples prepared by
spiking “new” and “used” oil. All TCLP target volatiles
except methanol were spiked at regulatory action levels, non—
TCLP alcohols and ketones were spiked at 200 ppm, chlorinated
and aromatic compounds were spiked at 5 ppm, and isotopically—
labeled analogs were spiked at 5 ppm. Compounds spiked at 200
ppm generally saturated the mass spectrometer. This resulted
in apparently poor recoveries for the butanols and relatively

TABLE 2 - Linear Range, Headspace/Isotope Dilution
Co ou d Name Low h1 1I
*Acet 5.0 100 12.9 4.620
Benzene 0.2 100 11.6 1.025
*n..Buta n et 0.5 200 18.1 2.916
*jso..But ( 0.5 100 13.3 4.767
Carbon tetrachtorjde 0.2 100 10.5 1.009
*Car disutfide 0.2 100 11.2 7.604
Chtorobenzene 0.2 200 9.5 0.826
ChLoroform 2.0 200 9.3 1.072
1,4-Dichtorobenzene 0.2 200 5.5 1.694
•12DichLor thane 0.2 200 16.2 3.001
1,1-Dichioroethene 0.5 50 5.7 1.981
*Diethyt ether 0.2 200 17.9 1.810
*Ethyt acetate 2.0 200 15.3 1.708
EthyLbenzene 0.2 200 7.1 0.356
Nexachtoroethane 2.0 200 9.9 0.870
*Nethytene chLoride 2.0 200 12.4 2.917
*Methyt ethyl ketone 0.2 25 5.9 1.398
*$ IBK 0.5 200 18.0 1.182
Nitrobenzene 0.5 200 9.0 0.700
*pyridl n e 5.0 200 15.1 1.759
*TetrachLoro th 0.2 200 5.4 0.426
*Trich(orOf [ uoromethane 0.2 200 10.2 4.006
*1,1 ,2.TrjchLorotrjfLLyJr o ethap 0.2 200 16.2 2.497
Toluene 0.2 200 5.1 0.889
Trich(oroethene 0.2 100 11.0 1.387
*Vinyl chLoride 0.2 25 18.3 3.226
o-Xytene 0.2 200 10.1 0.457
m/p-Xytene 0.2 200 9.4 0.472
* Internal standard quantitation

poor analytical precision for the alcohols and ketones.
Recoveries of greater than 150% which were obtained for 11 of
the 14 labeled analogs used in this study (i.e., 1,1-
dichloroethene, chloroform, carbon tetrachloride, benzene,
toluene, chlorobenzene, ethylbenzene, p—xylene, o—xylene,
hexachioroethane, and nitrobenzene).
Method accuracy and precision is improved when all of the
target analytes are spiked at 2 ppm rather than at the
regulatory limit. Even so, toluene (142%), pyridine (51%),
nitrobenzene (41%), methylene chloride (130%), methyl ethyl
ketone (137%), 1,2—dichioroethane (138%), MIBK (127%), diethyl
ether (147%), trichlorofluoromethane (163%), 1,1,2—
trjchlorotrifluoroethane (153%), iso-butanol (266%), n—butanol
(233%), and ethyl acetate (146%) were outside the 80-120%
acceptance windows. Toluene in the oil prior to spiking and
an interference for iso—butanol may have contributed to the
performance problems with those compounds. Data for the seven
replicates of new oil spiked at 2 ppm are provided in Table 3.
System Contamination : Instrument contamination will be a
problem with direct injection analysis of oil. Most of the
heavy molecular weight materials (e.g. asphaltenes) remain in
the quartz wool plug, but semi-volatiles and hexadecane are
volatilized onto the chromatography column. For this reason,
a bake out period is included in the oven temperature program.
Hexadecane and the oil hydrocarbons elute after the target
analytes. Conducting this study had a negative impact on
instrument performance; column resolution and instrument
sensitivity were lost. Examination of the system revealed
that some oily residue remained in the column and oil
contamination was evident on the source of the mass
It was recognized that the quartz wool plug would have a
limited capacity to hold up non-volatile contaminants.
Because laboratories are accustomed to a 12-hour shift for
analytical instruments, replacement of the injector liner and
the septa every 12 hours was selected as a reasonable
maintenance requirement. Ten replicate injections of used oil
spiked at 5 ppm were made over a 12 hour period in order to
demonstrate that analyte response factors could be stable in
a production environment. Response factors for labeled
benzene, toluene, ethylbenzene and pyridine obtained on
injections 1, 5 and 10 are provided in Table 4. The response
factors for the BTEX compounds are quite stable but the
response factor for labeled pyridine drifted downward through
the shift.

TABLE 3 - Headspace AnaLysis of New Olt at 2 ppn
Coa ound _______________ %RSD BLank ( m )
Acetone** 105 12.7 0.4 0.8
Benzene 103 8.2 0.1 0.5
Benzene-d 6 159 4.0 7.2
n.Buta L** 233 12.2 0.7 1.7
iso .Butanot* 1 ** 266 11.6 0.8 1.8
Carbon t trachLoride 107 7.6 0.0 0.5
CCL 4 - C 141 4.3 6.0
Carbon disutfide** 89 8.0 0.0 0.4
Chtorobenzene 117 7.2 0.0 0.5
ChLorobenzene-d 5 124 5.3 5.4
ChLoroform 88 7.9 0.0 0.4
Chtoroform-d 1 151 3.0 6.9
1,4-Dichtorobenzene 90 8.1 0.0 0.4
1,4-Dichtorobenzene-d 4 81 6.2 3.3
1,2 .Dichtoroethane** 138 7.5 0.0 0.6
1,1-Dichtoroethene 85 8.6 0.0 0.4
1,1-DichLoroethene-d 2 173 8.6 7.4
D ethyL ether** 147 10.6 0.0 0.9
EthyL acetate** 146 8.5 0.0 0.7
Ethy tbenzene 98 8.5 0.1 0.5
EthyLbenzene-d 10 166 6.7 7.1
Hexachtoroethane 104 8.0 0.0 0.5
HexachLoroethane 3 C 121 9.2 5.0
Methytene chtoride** 130 7.6 0.2 0.6
MethyL ethyL ketone** 137 8.3 0.1 0.7
4.p4ethyL.2 .p e ntanone** 127 8.9 0.0 0.7
Nitrobenzene 41 11.0 0.0 0.3
Nitrobenzene-d 5 123 12.7 3.6
Pyridine** 51 18.6 0.7 0.6
Pyridine-d 5 106 6.7 3.9
TetrachLoroethene** 90 9.3 0.0 0.5
TrichtorofLuoromethafle** 163 8.0 0.0 0.8
1,1,2 CL 3 F 3 ethane** 153 6.3 0.0 0.6
Toluene 117 6.5 0.6 0.6
Totuene-d 8 150 6.1 6.6
Trichtoroethene 113 7.6 0.0 0.5
Trichtoroethene-d 1 99 5.3 4.3
VinyL chtoride** 46 15.3 0.0 0.4
o-XyLene 108 8.3 0.2 0.5
o-Xytene-d 10 144 6.5 6.0
m-/p-Xytene 110 7.3 0.5 1.0
p-XyLene-d 10 157 6.4 6.5
Based on 7 measurements
*ALter te mass eaçtoyed
-- IS quantitation

TabLe 4 Stability of Response, Direct Injection
Et Benzene-
dl 0
Inj 1
Inj 5
1.3101 —
lnj 10
Method linearity : The linear range of the direct injection
method was established by analyzing mixtures of standards
prepared at eight concentrations in hexadecane. Calibration
standards of non-labeled target analytes were prepared at 0.2,
0.5, 2.0, 5.0, 25, 50, 100, and 200 ug/m1, labeled analogs
were present at 5.0 g/ml in each calibration standard.
Isotope dilution response factors were determined for 14 of
the 15 targets for which isotopically-labeled analogs were
added. Mass spectral interferences precluded the use of
pyridine—d 5 for this application. Internal standard response
factors were calculated for pyridine and the remaining 14
target analytes for which suitable isotopically-labeled
analogs were not available. Table 5 provides the linear range
for each compound, the average response factors over the
linear range, and the relative standard deviation for the
calibration curve of each compound.
System Stability : Stability of the measurement system was
monitored by using a daily continuing calibration solution
prepared at 5 ppm. The continuing calibration solution was
prepared with all 29 target analytes and 14 isotopically
labeled analogs in hexaecane. Average percent recoveries for
14 of the target analytes with isotopically-labeled analogs
were calculated using isotope dilution. Recoveries of
pyridine, the 14 analytes without analogs and the 14
isotopically labeled analogs were calculated using the
internal standard technique. The calibration for direct
injection appears more stable than with headspace. Average
recoveries were within an 80—120% acceptance window except for
13 C-carbon tetrachioride and pyridine. RSDs of greater than
20 % were obtained only for 13 c—carbon tetrachioride,
nitrobenzene and pyridine. As with headspace, RSDs were
smaller for each of the 14 target analytes determined using
isotope dilution than the corresponding isotopically-labeled
analog quantitated using the internal standard technique.
Performance of Direct Injection : Performance of the direct
injection isotope dilution technique was established by
analyzing two sets of seven replicate samples prepared using

TABLE 5 - Linear Range, Direct Injection with Isotope Dilution
Ccn ound Name r&d RRF
*Acet 5.0 200 11.0 2.488
Benzene 0.2 200 17.9 1.054
*n. 8uta, L 0.5 200 17.2 3.224
*iso.Buta L 0.5 200 15.4 4.507
Carbon tetrachtoride 0.2 200 14.9 1.025
*Carbon disuLfide 0.2 200 14.5 2.031
Ch lorobenzene 0.2 200 9.9 0.914
ChLoroform 0.2 200 9.6 1.116
1,4-Dich lorobenzene 0.2 200 6.4 1.462
*12 DichLoroethane 0.2 200 15.3 4.822
1,1-Oich(oroethene 0.2 100 13.1 1.612
*Diethyt ether 0.2 200 4.1 1.891
*Ethyt acetate 0.5 200 15.5 1.566
Ethytbenzene 0.2 200 11.6 0.337
HexachLoroethane 0.2 200 17.4 0.895
*Nethy(ene chloride 2.0 200 8.9 3.510
*MethyL ethyl ketone 0.2 200 5.3 2.480
*NIBK 0.2 200 6.9 4.057
Nitrobenzene 0.2 200 4.1 0.399
*pyridine 0.2 200 9.9 8.940
0.2 200 6.1 2.906
0.2 200 4.4 4.103
*112..TrjchLorotrif [ roethafle 0.2 200 11.5 1.938
Totuene 0.2 200 9.3 1.071
Trichtoroethene 0.2 100 12.3 1.661
*VinyL chLoride 0.2 200 18.0 3.229
o-Xy lene 0.2 200 10.3 0.488
mlp-XyLene 0.2 200 16.6 0.450
* InternaL standard quantitation
standards prepared in hexadecane

“new” and “used” spiked oil. All TCLP target analytes were
spiked at regulatory action levels, non-TCLP alcohols and
ketones were spiked at 200 ppm, chlorinated and aromatic
compounds were spiked at 5 ppm, and isotopically-labeled
analogs were spiked at 5 ppm.
These data demonstrate that direct injection can be used
for the analysis of volatiles in motor oil. Data for n—
butanol and iso—butanol appeared to show the greatest
improvement using direct injection rather than headspace
analysis. Recoveries of all analytes were comparable with
those obtained using headspace analysis. However, these
analyses caused noticeable contamination of the mass
spectrometer source which means that direct injection will
require more frequent instrument maintenance. Difficulties
with interferences required the use of four alternate
quantitation masses during the analysis of used oil.
As was the case for headspace, performance of direct
•injection analysis was evaluated over a narrower concentration
range than the three order of magnitude range (0.2 - 200 ppm)
required to satisfy the TCLP regulation. New oil was spiked
with 5 ppm of target analytes except where the regulatory
limit was less than 5 ppm. In those cases target analytes
were spiked at the regulatory limit (e.g., 0.5 ppm for
benzene). Each of the 14 isotopically-labeled analogs were
also spiked at 5 ppm. The recovery and RSDs of the target
analytes and isotopically-labeled analogs calculated using
isotope dilution and internal standard routines are presented
in Table 6. Spike levels are also given in the table.
Isotope dilution results demonstrate that method accuracy
using direct injection is comparable to headspace isotope
dilution analysis. However, method precision is significantly
worse using direct injection compared with headspace. Most
reported RSDS for the spiked new oil are greater than 20%.
RSDs for the analysis of used oil are generally 50—80%. This
lack of precision is probably due to the build up of oil
contamination in the instrument system during the conduct of
this method performance study. Table 7 presents data for new
oil spiked at low concentrations after recalculation using the
internal standard technique for the 14 analytes with useful
isotopically— labeled analogs.

TABLE 6 - Direct Injection AnaLysis of New Oil, at 5 ppm
CO lT o4Zld Recovery (% ) %RSD BLank (p Spike
Acetone** 91 14.8 1.9 5.0
Benzene 121 15.9 0.1 0.5
Benzene-d 6 71 7.3 3.2 5.0
n ButanoL*,** 107 27.8 0.5 5.0
iso .Butanot*,** 95 19.5 0.9 5.0
Carbon tetrachtoride 73 40.9 0.0 0.5
CCL 4 - 13 C 106 9.6 4.7 5.0
Carbon disu(fide** 53 22.3 0.0 5.0
Ch torobenzene 110 26.9 0.0 5.0
Chtorobenzene-d 5 74 8.0 3.2 5.0
ChLoroform 113 28.7 0.0 6.0
Chtoroform-d 1 66 7.3 3.1 5.0
1,4-Dich torobenzene 98 23.9 0.0 7.5
1,4-Dich lorobenzene-d 4 74 4.0 3.1 5.0
1,2 DichLoroethane** 101 23.1 0.0 0.5
1,1 DichLoroethene* 97 45.3 0.0 0.7
Diethyt ether** 76 24.3 0.0 5.0
EthyL acetate** 113 27.4 0.0 5.0
EthyLbenzene 105 26.6 0.2 5.0
EthyLbenzene-d 10 79 9.3 3.1 5.0
HexachtOrOethafle 107 33.2 0.fl 3.0
Hexachtoroethane 3 C 67 5.9 3.4 5.0
Nethytene chLoride*.** 98 45.3 0.0 5.0
MethyL ethyL ketone** 79 24.6 0.4 5.0
MIBK** 93 31.4 0.0 5.0
Nitrobenzene 100 26.3 0.0 2.0
Nitrobenzene-d 5 88 8.5 3.8 5.0
Pyricfine** 31 35.9 0.0 5.0
Pyridine-d 5 71 7.8 2.6 5.0
TetrachtorOethefle** 82 27.1 0.0 0.7
TrichLorofLuOrOmethafle** 76 27.6 0.0 5.0
1,1,2 CL 3 F 3 ethane** 69 29.2 0.0 5.0
Totuene 98 14.4 0.6 5.0
ToLuene-d 8 75 11.6 3.2 5.0
Trichtoroethefle 72 30.4 0.0 0.5
TrichlorOethefled 1 44 8.8 1.9 5.0
VinyL chtoride** 63 35.2 0.0 0.2
o-XyLene 101 25.4 0.4 5.0
o-XyLened 10 81 9.5 3.2 5.0
m/p-XyLene 107 25.9 0.6 10.0
p-XyLene-d 10 77 8.1 3.1 5.0
ALternate mass eppLoyed
* IS quantitation

TABLE 7 - Direct
Injection Analysis using Method
spiked at Low Concentrations
Coni,otrid Nani
IS Recovery
j ID Recovery
Carbon tetrechloride
Ch Lorobenzene
1 ,4-Dichlorobenzene
Hexach I oroethene
Ni trobenzene
Tot uene
Trich toroethene
‘Performed with older MS (actv)
‘Uses isotope dilution (dis)
•Requires new methods (dis)
‘Uses additional hardware (dis)
•Overnight equilibration (dis)
‘Uses larger saople size (adv)
‘More con lex procedure (dis)
•Less cross-contamination (ady)
‘Safety hazard (F.P.<90) (dis)
‘Fewer matrix probs (adv)
‘Poor partition coeff (dis)
‘Possible use for oily soil (adv)
The authors
The authors
provided by
Waste, U.S.
‘Requires sensitive MS (dis)
‘Uses IS calculations (adv)
‘Uses existing method (adv)
•No new hardware required
‘Quick turnaround (adv)
‘0.5-1 g saiTçle (dis)
‘Siir ler procedure (adv)
‘Instrunent contamination
‘Little explosion hazard
‘More matrix difficulties
‘Good recovery (adv)
‘Not suitable for soil (dis)
Headspace/isotope dilution/GC/MS and direct injection/GC/MS
can be used for the analysis of 28 of the 29 target analytes
for this study. Table 8 presents 95 percent confidence
intervals for measuring each of the TCLP target analytes using
headspace and direct injection analysis. Reliable analysis of
pyridine at regulatory limits may be difficult to achieve
without use of replicate analyses for individual samples.
Each technique can provide useable analytical data; the
particular advantages and disadvantages to each procedure are
listed below.
would like to
of Cairbridge
providing several
also gratefully
Mr. John Austin
Mr. Barry
of the
of Solid

TABLE 8 - 95% Confidence IntervaLs for TCLP VoLatites in New OiL
Benzene, 0.5 ppm
93-129 %
11-161 %
Carbon tetrachtoride,
0.5 ppm
112-190 %
0-173 %
0-204 S
100 ppm
73-111 5
44-176 5
8-154 5
6 ppm
68-106 5
43-184 5
12-156 5
7.5 ppm
70-112 5
39-156 5
41-133 5
0.4 ppm
143-193 5
45-158 5
0.7 ppm
68-126 5
58-208 5
113-193 5
3 ppm
81-117 5
26-188 5
0-146 5
MethyL ethyL ketone
200 ppm
100-158 5
75-116 5
0.8 ppm
6-102 5
35-164 5
16-162 5
5.0 ppm
3-107 5
0-119 5
0.7 ppm
107-145 5
16-149 5
0.5 ppm
95-129 %
0-147 5
0-141 5
VinyL chLoride,
0.2 ppm
94-238 5
0-149 5
ID - isotope diLution
IS - internaL standard

Paul E. I ester
Tekmar Company
P0 Box 429576
Cincinnati, OH 45242—9576
Currently, the methods used for analysis of volatile organic
compounds (VOCs) in soils matrices involve heated purge and trap.
In order to prevent contamination of the purge and trap
concentrator, and to ensure that analyte concentrations fall in the
linear range of the GC detector, samples are typically screened by
static headspace/GC/FID. With proper selection of analytical
parameters, static headspace can also be used effectively as an
alternate quantitative technique for the determination of VOC
concentrations in soil samples.
A non-polluting, matrix modifying solution is added to the soil
sample irt a headapace vial, along with surrogates and internal
standards. Samples are heated to 85° and mixed while in the heated
zone. With these parameters optimized, most VOC’s listed in Method
8260 can be determined with acceptable precision for a wide range
of soil types.
Precision, accuracy, linearity, and carryover of this method are
compared to existing methods. Sample collection, preservation, and
storage are also discussed.

David R. Yourtgman, Staff Scientist, Lockheed Engineering and
Sciences Company, Las Vegas, Nevada 89119 and Michael H. Hiatt,
Branch Chief, U.S. Environmental Protection Agency, EMSL-LV, Las
Vegas, Nevada 89119.
Vacuum distillation (VD) provides a low temperature alternative
to the analysis of conventional purge and trap analytes. This
technique offers the potential for analysis of large sample sizes
as well as different matrices and analytes for which there are
currently no analytical methods available.
The ‘ ID apparatus shown consists of a chilled condenser portion
where water interference is removed at temperature of —15 C 0 .
The analyte. of interest are trapped on a stainless steel loop at
liquid nitrogen temperatures and then desorbed and transferred to
a gas chromatography/mass spectrometer using a six port valve and
heated transfer line.
The authors will present the results of experiments conducted
during development of this technique. Items to be discussed
include development of the apparatus, precision, accuracy, and
recovery data. The results will be discussed as they apply to
current CERCLA VOA analytes, selected RCRA 8270 analytes and
organic amines. Matrices considered are water, soil and simulated
fish (cod liver oil).

Peter Del Mar, EM-9, Los Alamos National Laboratory, Los
Alamos, New Mexico 87545
A solid phase extraction technique is described in which the
inside walls of relatively large bore polymer tubing comprise
the sorbent surface rather than a packed bed. A practical ten
channel extractor is described. The open tubular approach is
compared to liquid/liquid extraction and to conventional solid
phase extraction cartridges. Parameters affecting recovery
are evaluated including tubing composition, inside diameter
and flow rate. Data illustrating recovery, precision and
detection limits for organo-chiorine pesticides and PCBs are
presented. Accuracy is discussed in conjunction with results
achieved on several EPA water studies. Preliminary data on the
extraction of more polar compounds using “hybrid column
materials is presented.
Liquid/liquid extraction of environmental water samples for
selected organic chemicals has been the standard for many
years. This approach requires an overnight extraction and
consumes large amounts of regulated solvent, typically
methylene chloride. Following extraction, a solvent reduction
step, (K-D evaporation), is required along with a solvent
exchange if an electron capture detector is being used.
Because the extraction is non-specific, one or more additional
cleanup steps are often required as well.
These time and solvent consuming steps have supplied much of
the impetus for the continuing development of solid phase
extraction. While modern SPE cartridges have successfully
addressed these issues, they also have some limitations. Among
these are a propensity to clog, (if the samples contain
suspended solids), water retention and the presence of various
interfering substances related either to manufacture and
packaging or the chemical degradation of the bonded silica
sorbents themselves (1).
In an attempt to overcome these limitations without giving up
the advantages of solid phase extraction, the open tubular
approach has been developed. Initially the work centered on
the use of polyethylene tubing with an internal diameter of 1
to 2 mm, and was evaluated for the extraction of certain

non-polar target compounds such as organo-chiorine pesticides
and PCBs. A prototype ten channel extractor has been built and
used in the continued development of this technology for the
extraction of actual field samples and in EPA Water Pollution
and Water Supply studies. In spite of it’s less efficient
geometry as compared to packed bed systems, the open tubular
system has a number of extremely practical advantages. The
“columns” will not clog, do not retain water and are reusable
indefinitely. We have also demonstrated that the system can be
configured to greatly reduce the risk of operator exposure in
the case of hazardous samples.
This paper will discuss the following parameters affecting
extraction efficiency: column diameter, flow rate, and column
composition. It should be noted that extraction efficiency is
actually the product of two factors; how much of a given
analyte is adsorbed onto the tubing walls, and then, how much
of this trapped material is actually eluted by the solvent.
This “elution efficiency” is also discussed. Though the column
length is also an important variable, it is not discussed
because of the limitations imposed by the commercially made
tubing available at the inception of this project. Even the
cleanest commercial tubing obtained contained additives, such
as phthalate plasticizers which could not be completely
removed even after prolonged solvent washing. The
concentrations of these interferents in the extracts naturally
increases in direct proportion to the length of the column;
therefore it was decided to postpone that part of the study
until “clean” tubing could be obtained. This work is now
All extracts were analyzed on an HP 5880 GC equipped with an
ECD and a 7673A autosampler. The injection volume was 1 ul.
The carrier gas was Helium with a linear velocity of 23 cm per
second at 200 deg. C and the make up gas was Nitrogen at a
flow rate of 40 ml per minute. Capillary columns were 30
meter, .32mm I.D., 0.25 urn film; both “Supelco SPB-5” and
“Restek Rtx-35” columns were used. The GC runs were programmed
in two steps, from 100 deg. C to a final temperature of 240
deg. C.
Stock solutions of organo-chiorine pesticides were made from
neat standards supplied by Chem Service. They were divided
into three groups:

Group A Group B Group C
Aldrin a-BHC Chiordane
4,4’-IJDD b-BHC Toxaphene
4,4’-DDE d-BHC
4,4’-DDT Endosulf an I
Die ldrin Endosulfan II
Endrin Endosulf an sulfate
Heptachior Endrin aldehyde
Hep. Epoxide
Lindane (g-BHC)
Methoxychi or
The parent stocks were made up in Iso-octane, and then taken
through parallel dilutions; one ending in Methanol for spiking
solutions and the other ending in Hexane for calibration
standards. The concentrations of spiking solutions were as
follows: single component pesticides - 0.5 to 2.0 ug/mi,
Chiordane - 1.6 ug/mi, and Toxaphene - 16 ug/ml.
The PCB studied was Arochior 1242. A spiking solution in
Acetone and a calibration standard in Hexane were derived from
a standard containing 500 ug/mi in Iso-octane. The
concentration of the spiking solution was 2.5 ug/ml.
One hundred ml samples were spiked with 100 ul of spiking
solution and the sample extracts were reduced to a final
volume of 1.0 ml prior to analysis. The calibration standards
were made up at concentrations 1/10 those of the spiking
An internal standard solution was made up in Iso-octane
2,6-Dibromobiphenyl 5.0 ug/ml
2,4,6-Tribromobiphenyl 3.5 ug/mi
2,2’ ,4,5’ ,6-Pentabromobiphenyl 4.2 ug/mi
Forty ul of internal standard mix was added to each ml of
extract or calibration standard.
Recoveries were calculated as follows: the peak areas were
first normalized using the nearest eluting internal standard
and then compared directly to a calibration standard made up
as explained above. It can be seen that the final extracts
would have the same concentrations as the calibration
standards assuming a recovery of 100%.

The sample matrix for recovery studies was tap water that had
been filtered through an 10 inch bed of 20-35 mesh activated
charcoal and the sample size was 100 ml unless otherwise
noted. All solvents used were “Baker Resi-Analyzed” grade.
The ten channel extractor, Figure 1, was built using of f-the-
shelf components including separatory funnels used as sample
holders, Hamilton HPLC valves and a Masterfiex ten channel
peristaltic pump with reversable flow and pumping rates of
0.36 to 36 ml per minute. Connections between components were
made with Teflon tubing.
The flow paths during separation and elution are shown
schematically in Figure 2. The plumbing is arranged to pass
the eluting solvent through the “columns” without coming into
contact with the tubing in the pump heads since it could be
damaged by the solvent. In operation, up to ten water samples
are placed in the extractor, spiked as required, and then
pumped through the columns. Following separation, the columns
are eluted with 10.0 ml of solvent, typically Hexane. The
sample extracts are blown down to a final volume of 1.0 ml
under dry Nitrogen, and internal standards are added prior to
The effect of column I.IJ. on extraction efficiency was studied
by pumping identically spiked aliquots through 2.5 meter
polyethylene tubes with internal diameters of 1.4 mm, 1.6 mm,
and 2.0 mm respectively. The flow rate was 1.0 ml per minute.
Flow rate was investigated by doing a series of extractions
using 1.5 meter lengths of 1.1 mm I.D. polyethylene tubing
while varying the pumping rate. The flow rates were measured
by weighing the output of each column over a fixed time
interval. Recovery of target compounds was then plotted versus
volumetric flow rate.
The efficiency of removal of the adsorbed analytes from the
tubing wall, (elution efficiency), was determined by analyzing
the extracts produced by subjecting a column to successive
elutions with 10 ml volumes of hexane.
Tubing with more polar composition was obtained from the
Phillips Petroleum Plastics Technical Center. The 1.0 mm I.D.
tubing made from a blend of high density polyethylene and
polyethylene terephthalate, (PET). The compositions ranged
from pure polyethylene to blends containing 3%, 5%, and 10%
PET respectively. The recoveries of Lindane and 2,4-
Dinitrotoluefle were then determined for various compositions.

- P
I . -
4 .$
j ;-’
A a
• .•
t 4

10 CC SYR.
10 CC SYR.
F )
FIGJRE 2. Flow Schematic.

Column diameter: The average recovery of each of the ten group
A pesticides was determined for each run and then normalized
to the recovery obtained from the 2.0 mm I.D. column. The
results are presented in Table I.
Table I. Recovery vs.
1.6 mm
1.4 mm
The average recovery for the 1.4 mm column was 11% higher than
that of the 2.0 mm column. This is not a very large gain, but
it should be remembered that at any given volumetric flow
rate, when the diameter decreases 30%, the linear velocity
doubles. Thus the shorter “residence time” of the water in the
column offsets, to some extent, the advantages of smaller
diameter. Although the smaller diameter results in somewhat
higher recoveries, there are two offsetting factors which
become increasingly important as the diameter decreases. The
first is that the optimum flow rate, at which good
reproducibility can be achieved, gets lower and thus
extraction times get longer. The other is that the surface
area available for adsorption is also decreased which reduces
the extraction capacity of the column.
Flow rate: Figure 3 shows the recovery of Dieldrin for a 1.5
meter column with an i.d. of 1.1 mm at various flow rates.
For packed bed systems, the flow rate can be varied over a
relatively wide range without adversely affecting
reproducibility, but this is clearly not true with the open
tubular configuration. This potential problem wa3 easily
overcome, however, since the ten channel pump used in the
extractor has a variable speed feature which makes it easy to
achieve and reproduce any desired flow rate. For a 1.6 mm
i.d. tube, the optimum flow rate was found to be approximately
.9 ml per minute.

Figure 3. Percent Recovery of Dieldrin related to flow rate.
Elution efficiency: This was determined by analyzing
successive 10 ml elutions from a given column. For organo-
chlorine pesticides and PCB5, it was found that a single 10 ml
Hexane elution removed 98% to 99% of the material trapped
during separation. This figure applies to columns made of high
density polyethylene; other polymers, notably low density
polyethylene, were not as efficient.It was found that the flow
rate during elution must also be controlled.
Recovery of target compounds: Table II gives recoveries and
detection limits for pesticides and PCBs. The extractions were
done with a 3 meter length of 1.6 mm i.d. commercial tubing.
The detection limits are based on the recoveries achieved with
the 3 meter length, extraction of 100 ml samples, a 1.0 ml
final volume, a 1 ul injection volume, and a signal to noise
ratio of approximately 10:1 for minimum detectable peaks..
0.2 0.4 0.6
Flow Rate ml/min
1.2 1.4

Samples were spiked at the following levels: single component
pesticides - 0.5 to 2.0 ug/L, Chiordane - 1.6 ug/L, Toxaphene
- 16 ug/L, and Arochior 1242 - 2.5 ug/L.
The recoveries vary over a fairly wide range and it has been
found that the water solubility of a compound does not always
correlate with it’s recovery. It has also been found that the
polarity of the tubing material is not the only factor
affecting recoveries. An experiment was done in which the
inner walls of the polyethylene tubing were coated with a thin
layer of Iso-octane prior to performing an extraction. It was
thought that the three dimensional hydrocarbon solvent layer
might result in a longer °residence time” for the arialytes
than the two dimensional hydrocarbon wall of the tubing thus
increasing recoveries. Instead, the recoveries of most of the
compounds studied dropped significantly, showing that polarity
alone is too general a property to accurately predict
Precision: Table III illustrates the precision possible with
this methodology. Ten replicates were spiked with Arochior
1242 at a level of 2.5 ug per liter. The internal standard was
2,4,6-Tribromobiphenyl. The extracts were then compared to a
standard prepared from the same stock. The recoveries
calculated for normalized peak areas are presented. It is
probable that one of the reasons for this precision is the
absence of the aggressive Kuderna-Danish solvent reduction
required with liquid/liquid extraction methods.
Accuracy: Perhaps the best indication of the accuracy possible
with this method is the record achieved in EPA Water Supply
and Water Pollution studies. The overall results of six EPA
studies are summarized in Table IV. Included in these studies
were the ten “Group A” pesticides previously mentioned,
Chlordane and Toxaphene.

Table II. Recovery and Detection Limits for
Selected Analytes
Analyte % Recovery LOD ug/L
Aldrin 55 0.03
a-BHC 53 0.02
b-BHC 23 0.10
d-BHC 26 0.12
Chlordane 74 0.25
4,4’-DDD 67 0.04
4,4’-DDE 52 0.04
4,4’—DDT 66 0.03
Die ldrin 79 0.04
Endosulfan I 60 0.05
Endosulf an II 61 0.08
Endosulfan sulfate 50 0.09
Endrin 90 0.03
Heptachlor 76 0.03
Heptachior epoxide 63 0.04
Lindane 51 0.03
Methoxychlor 85 0.08
Toxaphene 84 2.5
Arochlor 1242 65 0.30
Table III. Recovery of Arochlor 1242
% Recovery % Recovery
Replicate Peak 1 Peak 2
1 63 64
2 63 64
3 64 65
4 64 65
5 64 65
6 64 65
7 64 64
8 65 66
9 64 66
10 65 66

Table IV. Results of EPA Water Studies. --
Study # of Results Percentage Average
Number Reported “ Acceptable” Error
WP 022 14 100% 7%
WP 024 14 100% 4%
WP 025 12 100% 8%
WS 026 4 100% 4%
WS 027 5 100% 8%
WS 029 6 - 100% 4%
It must be noted that these results were averages based on
three replicates for each determination. To illustrate the
accuracy of this method more fully, the data for the single
component pesticides in studies WP 024, WS 026, and WS 029
were considered individually. The 22 reported values were
derived from 66 individual data points. The average recovery
for the 66 point set was 100% with a standard deviation of
Column composition: To be of practical value, and to preserve
the desirable properties of this methodology, any prospective
new column material must meet the following requirements:
1. Must be insoluble in the aqueous matrix.
2. Must be essentially impervious to the eluting
3. Must have an affinity for the target compounds under
4. Must be chemically stable in the pH range of the
5. Must have physical properties such that it can be
formed into tubing.
Polyethylene has these properties when used for certain non-
polar target compounds. The problem with the polyethlyene
tubing used initially was that commercial tubing always
contained additives, which become interferents in the
extracts. Tubing has recently been obtained made from pure
polyethylene resin which gives extracts virtually free of
interfering substances. The largest limitation remaining is
that polyethylene does not appear to extract some of the more
polar target compounds such as nitroaromatics and phenols
efficiently. To get a more polar column material, tubing was
obtained made from blends of polyethylene and polyethylene

terephthalate, (PET). The 1.0 mm i.d. tubing ranged in
composition from pure polyethylene to blends containing 3%, 5%
and 10% PET. Figure 4 shows the recovery of 2,4-Dinitrotoluene
plotted against the increasing percentage of PET.
Figure 4. Recovery of 2,4-Dinitrotoluene
This extraction was done on a 3 meter column and although the
recoveries are low, there is a clear correlation between
recovery and the increasing percentage of the PET polymer. In
a separate experiment, the 5% PET composition was found to
increase the recovery of Lindane by 33% as compared with a
pure polyethylene tube of the same deimensions. Future work
will focus on defining the possibilities and limits of
improved materials.
A solid phase extraction technique for organic pollutants in
water which employs open tubular “columns” has been described.
A ten channel prototype extractor has been built and validated
for organochlorine pesticides and PCBs. This method has the
low solvent consumption characteristic of solid phase systems
o 3.5
3 5 10
Percentage of PET

and is much faster than the liquid/liquid extractions now in
wide use. With the multi-channel device, ten samples can be
extracted, blown down and on the GC in less than 3 hours. The
extraction is quite specific and after some 200 field samples,
none has ever required any further cleanup. Several practical
advantages of the system have been pointed out; the open
columns will not clog, do not retain water and can be reused
indefinitely. The method is valuable from the waste
minimization point of view since no throw away trash is
generated. This is particularly desirable when the sample
matrices are radioactive, which is often the case for
envirdnmental work done in the DOE complex.. It should also be
underlined that this method does not add anything to the
sample during extraction. Liquid/liquid extraction
contaminates a sample with methylene chloride and some
commercial SPE cartridges contain phthalate esters which are
water soluble to some extent. Thus, a hazardous waste may be
created by the extraction process even if the original sample
was clean. Since it is a pumped system, it could also reduce
operator exposure to hazardous samples. A sample could be
pumped from a container or waste stream without actually being
handled by a technician. The possibility of the use of new
materials has been discussed. The method works well for non
polar compounds but more work will be required to determine if
it can be extended to more difficult target compounds such as
those specified in EPA’s Toxicity Characteristic Leaching
Procedure, a list containing compounds with appreciable water
Acknowledgement: The help of Mr. Joe Harder of the Phillips
Petroleum Plastics Technical Center in the fabrication of
specialty tubing used in this work is gratefully acknowledged.
Literature Cited
(1) Junk,G.A.;Avery,M.J.;Richard,J.J. Anal.Chem. 1988, 60,

Mark Bruce , Director of Research and Development, and Marvin W. Stephens, Technical
Director, Wadsworth/ALERT Laboratories, Division Of Enseco, Inc., 4101 Shuffel Dr.
N.W., North Canton, Ohio 44720.
James Carl, Product Development Manager, Bruce Killough, Product Development
Technician, Tony Zine, Senior Development Consultant, Science Products Division,
Corning Incorporated, Corning, New York 14831.
David Burkitt, President, Burkitt Consultants, Inc., 2 Captain’s Way, Exeter, New
Hampshire 03833.
Two years ago Corning introduced the One-Step apparatus which combined extraction
and concentration glassware into one unit. Labor requirements were reduced and most of
the solvent was recaptured during the concentration step. The new One-Step apparatus
maintains these advantages and reduces both extraction time and solvent volume.
Maintaining a solvent boiling/condensation rate of 15 mL/min allows the analytes to be
extracted from a 1 L water sample in less than 6 hours.
The solvent pooi at the bottom of a conventional continuous liquid-liquid extraction
chamber holds about 200 mL. The analytes must be transferred from the solvent pool to
the boiling flask. The new One-Step apparatus eliminates this step shortening the
extraction time while reducing the amount of solvent needed.
Solvent vapor is lost through many joints in the glassware. Careful re-design has reduced
the solvent vapor loss to 40 mL. Not only does this reduce the total solvent requirement,
but more importantly reduces analyst and environmental exposure to solvent vapor.
The liquid/liquid extraction time has been shortened from 18 hours to less than 6 hours.
Total solvent volume has been reduced from 500 mL to 100 mL.
Solvent reduction has become one of the battle cries of the Environmental Protection
Agency for the early 1990’s. Since the Montreal Protocol was signed there has been an
effort to reduce or eliminate the use of many organic solvents in laboratory sample
extractions. At the same time, pressure has increased to shorten sample turn-around time
to speed the environmental cleanup process. Corning Incorporated has developed a new
One-Step T M extraction/concentration apparatus which addresses both problems.
The largest volume of laboratory waste is from the organic solvents required by current
organic extraction methodologies. Both SW-846 methods 3510 (separatory funnel) and
3520 (continuous liquid-liquid) are used for aqueous samples and require large amounts
(400-500 mL) of organic solvents such as methylene chloride. The solvent pooi at the
bottom of a conventional liquid-liquid chamber holds about 200 mL. Transferring analytes
from the solvent pool to the boiling flask requires about 2 hours under typical conditions.

The boiling flask starts with 300 mL of solvent. Evaporative solvent losses through the
ground glass joints and the condenser may total 100 mL during the extraction. In addition,
continuous liquid-liquid extraction times are typically 18 hours plus setup and cleaning
The accelerated One-Step 1 ” extraction/concentration apparatus reduces solvent volume to
100 mL and extraction time to 6 hours. This new extractor was evaluated in several areas
to test its viability for routine organic extractions. Analyte, can the target compounds
covered by the Contract Lab Program be quantitatively extracted? Matrix, can all usual
water matrixes be extracted without mechanical problems from emulsions or particulate
plugging? Accuracy & precision, are method bias and reproducibility equal to or better
than conventional liquid-liquid extraction? Ruggedness, is the method and glassware
durable enough to tolerate misuse and still produce acceptable results? Is the extract “dry”
enough that drying with sodium sulfate is no longer needed?
In short, the goal was to develop an apparatus which would extract as well as conventional
liquid-liquid extraction for all common environmental water matrixes yet be safer and more
cost effective.
Accelerated One-Step liquid/liquid prototype (see Figure 1.)
Neslab refrigerated circulator, CFF-25
\‘WR heated circulator, 1130
Reagents and Standards
Methylene chloride
Sulfuric acid
Table 1. Representative Analytes
Surrogate Spike
Niirobenzene-d5 Teirachloro-m-xylene
2-Fluorobiphenyl Decachlorobiphenyl
Terphenyl-d 14
Pheno l-d5
Matrix Spike
Phenol gamma-BHC (Undane)
2-Chiorophenol Heptachlor
1 ,4-Dichlorobenzene Aldrin
N-Nitmso-di-n-propylamine Dieldrin
1 ,2,4-Trichlorobenzene Endrin
4-Chloro-3-methylphenol 4,4’-DDT

The accelerated One-Step apparatus differs from conventional continuous liquid-liquid
extraction (and the current One-Step extractor) in several key areas. 1) The solvent pool
at the bottom of the conventional extraction chamber has been eliminated. The solvent is
returned from the bottom of the extraction chamber to the distillation flask (or K-D) via
gravity feed rather than syphon action. A hydrophobic membrane is placed across the
bottom of the extraction chamber. Organic solvent is dripped through the sample in the
conventional manner. However the solvent passes through the membrane at the bottom
and runs back to the distillation flask. No pool of solvent is required at the bottom of the
extraction vessel for syphon purposes. Thus less solvent is required. The extraction time
is also shorter since it is not be necessary to transfer analytes from the solvent pool to the
distillation flask via the solvent pooi dilution process of a conventional liquid-liquid
extractor. Figure 1 shows the flow of solvent. 2) The hydrophobic membrane effectively
excludes water from the solvent thus eliminating the need for a sodium sulfate drying step.
3) The solvent volume in the boiling flask has also been reduced from 300 mL to 100 mL.
4) Careful design of the apparatus has decreased the places solvent can be lost. Thus,
evaporative solvent losses can be halved. Employee and environmental exposure to solvent
vapor is reduced. Also, initial solvent volume can be reduced, which saves on solvent
Solvent flow
indicated by
Extraction chamber
Boiling FlaskJ -
Concentrator Tube
4 Condenser
Water Sample
Figure 1. Accelerated One-Step Extractor / Concentrator

All extraction parameters are the same as described in SW-846 Method 3520 except for
glassware assembly and use, solvent volume and extraction time. The disposable
hydrophobic membrane is sealed in place above the valve. The valve is closed and 100 mL
of methylene chloride is added to the extraction chamber. The water sample is poured into
the extraction chamber. Surrogate and matrix spiking are performed normally, as well as
pH adjustment. Once the boiling flask is hot, open the valve allowing the solvent to run
through the membrane into the flask. Solvent cycles through the extraction system. The
solvent should boil off at a rate of 15 mllmin. When the extraction is complete in 5.5
hours close the valve to concentrate the extract to the desired volume.
Attainment of Goals
Analyte, the compounds listed in Table 1 were selected to represent the range of analytes
normally extracted from water matrixes. All surrogate and matrix spike compound percent
recoveries were well within CLP limits. Accuracy & precision, the average percent
recovery for the accelerated One-Step system was nearly the same as the average
recovery with conventional liquid-liquid extractors. Reproducibility was within CLP
Extraction time, the required length of the extraction was determined with a time study
that measured analyte recovery at 3 hour intervals. Figures 2 and 3 show the results for the
matrix spike and surrogate compounds. Most analyte recovery is achieved in the first 3
hours. Small amounts of a few analytes were recovered between 3 and 6 hours.
Hydrophobic Membrane One-Step® Extraction
CLP BNA Matrix Spike Compounds
— ---—-
—‘ ,
6 9 12 15 18
hours extracted
Figure 2. Extraction Time Study for BNA Matrix Spike Compounds
O 2-Chiorophenol
U 1 ,4-Dichloro benzene
D N-Nitroso-di-n-propylamine
1 ,2,4-Trithloro benzene
A 4-Chloro -3-methylphenOl
X Acenaphthene
X. 4-Nftrophenol
— 2,4-Dinitro toluene
• Pentachioro phenol
0 Pyrene

Hydrophobic Membrane One-Step® Extraction
CLP BNA Surrogate Compounds
R 70
e 60
C 50
v 40
e 30
r 20
Y 10
hours extracted
Figure 3. Extraction Time Study for BNA Surrogate Compounds
Matrix, the accelerated One-Step apparatus was challenged with various water matrixes
to ensure that the system would cycle properly with all typical samples. No plugging or
flow restriction problems were noted with samples that were 5% by weight insoluble
solids. Both organic “muck” and inorganic clays were tested. Soluble organic and
inorganic materials did not present a problem either. An organic loaded pond water and the
acetic acid/sodium acetate buffer from the toxicity characteristic leaching procedure were
tested. Small emulsions formed on top of the membrane but did not restrict flow.
Ruggedness, the glassware is as easy to assemble as a conventional liquid-liquid
extractor. The system has been designed for durability and is more compact than the
original One-Step apparatus. The membrane “dries” the solvent passing through it very
well. Drying tests performed at Corning indicate that a column packed with sodium
sulfate, the drying tube adaptor (available for the original One-Step system) and the
hydrophobic membrane all reduce the water content to 0.005% or less. This is the same
water content as brand new methylene chloride.
The accelerated One-Step ” extractor/concentrator apparatus quantitatively extracts
base/neutral/acid analytes (BNAs) and pesticides from 1 L water samples with 100 mL of
solvent in tess than 6 hours. The membrane is resistant to plugging by particulate loaded or
emulsion forming samples. The system is easy to assemble and use. Impact on the
environment is minimized because the initial solvent volume and solvent losses are less
than conventional liquid-liquid extractors. Analyst exposure to solvent vapors is also
reduced. This new One-Step system uses 1/ the solvent and 1/3 the time of a
conventional liquid-liquid extractor.
4. Nitrobenzene-d5
•0. 2-Fluorobiphenyl
U TerphenyI d1 4
‘ Pheno -d5
4 2-Fluorophenol
.& 2,4,6.Tribromophenol
0 3 6 9 12 15

ERCO, Nancy Rothman, Peter Kane
Wadsworth/ALERT, Don Kirstead, Brian Haueter, Chuck Jacobs, Dave Counts, Tom
Hula, Craig Hacket, Kim Davis
“The more I study nature, the more I stand amazed at the work of the Creator.”
Louis Pasteur

Michael G. Winslow , Manager, Organic Analytical Division, Bradley A. Weichert, Manager,
GC!HPLC Department, Robert Baker, HPLC Group Leader, and Preston F. Dumas, Manager,
Sample Preparation Department, Analytical Services, Environmental Science & Engineering,
Inc., P.O. Box 1703, Gainesville, Florida 32602.
Although extraction of environmental water samples with solid-phase extraction (SPE)
cartridges has gained increasing acceptance in recent years, specific applications of this sample
preparation technology are conspicuously absent from standard EPA methods, despite the cost-
effective and health and safety advantages that these cartridges potentially offer. For the past
five years, we have used SPE cartridges for the extraction of nitroaromatic and nitramine
compounds from ground and surface water samples collected on and near various military
installations. Last year at this conference, we presented a comparison of our SPE technique
with the salting-out solvent extraction technique proposed in Draft Method 8330 for the
determination of nitroaromatic and nitramine compounds in water samples using high
performance liquid chromatography (HPLC).2.6 The intent of that presentation was to
demonstrate numerous advantages offered by solid-phase extraction and to suggest that it be
seriously considered for adoption in SW846 Method 8330. This paper presents the results of
the analysis of over 450 water samples collected at military installations using SPE cartridges
and HPLC with UV detection. The following areas are discussed: (1) a detailed review of the
method; (2) results of the method detection limit (MDL) study; and (3) the target analyte and
QC data obtained from the analyses of environmental water samples. Results to date
demonstrate that the proposed method using SPE cartridges with HPLC/UV analysis for the
determination of explosives compounds in environmental water samples proves both rugged and
cost-effective in extensive real-world tests.
Nitroaromatic and nitramine compounds are the most widely used class of chemicals in
munitions components. They are produced in large quantities and, along with some of their
production impurities and degradation products, are subject to environmental regulation.
Throughout the 1990’s and beyond, there will be a need for an analytical method that can
heIve low limits of detection in water (< 1 jig/L) for as broad a scope of munitions compounds
as possible, especially in light of the projected survey and remediation needs on and near
numerous Department of Defense (DOD) and some Department of Energy (DOE) installlations.
Last year at this conference we presented the validation findings for a method that employs SPE
with HPLC/UV analysis for the determination of 14 nitroaromatic and nitramine compounds in
water. It was demonstrated that the method can achieve excellent precision and accuracy for all
target compounds, and at the same time minimize labor and materials costs, solvent use, and
waste disposal.
Over the past year, the method has been routinely applied to hundreds of surface and
Woundwater samples. Some of the data are presented and discussed below.

Sorbent Cleaning
Prior to sample extraction, approximately 100 g of 80-100 mesh Porapak R (N-Vinyl 2-
pyrrolidone divinylbenzene copolymer, normally used as a GC column packing and obtained from
Waters) is cleaned by serially extracting with acetone, methanol and acetonhtrile in a Soxhlet
continuous extractor. The sorbent is placed in an extraction thimble and extracted for at least 2
hours with each solvent at a rate of about 12 cycles/hr. After final extraction with
acetonitrile, the sothent is air-dried in a hood and stored in a dessicator.
Column Packing and Conditioning
An empty 6-mL Baker SPE filtration column with a 20-riM tnt at the base is packed with 0.5 g
of cleaned Porapak R. Another 20-riM fit is placed at the top of the sorbent bed to assist
packing and prevent channeling and disruption of the bed surface. The column is placed in
Visiprep SPE Vacuum Manifold (Supelco) and conditioned by eluting 15 mL of ACN followed by
30 mL of ASTM Type ll/HPLC water at a flow rate of 10 mI/mm. Both eluents are discarded. It
is important that the sorbent bed is not allowed to go dry before application of the water sample.
Sample Extraction/Extract Preparation
A 500-mL water sample is passed through the SPE column at 10 mI/mm. The sorbent bed is
then eluted with 3 mL of ACN at s 3 mI/mm. into a graduated centrifuge tube. The ACN eluern
is concentrated to 2 mL under a gentle stream of nitrogen. The eluent is diluted to a final volume
of 6 mL with ASTM Type Il/HPLC water prior to HPLC analysis.
HPLC Analysis
The ACN/H 2 0 sample extracts are analyzed with an HPLC equipped with a variable wavelength
ultraviolet absorbance (UV) detector set at 250 nanometers. ESE utilizes a Shimadzu model LC-
6A HPLC equipped with a Kratos 757 UV detector, a Shmmadzu SPD-6A autosampler, and a
Nelson 2700 Turbochrom sytem for data collection and quantitation. The target anaiytes are
separated on a 250 mm x 4.6 mm ID Phenomenex ODS reverse-phase column with a 5-jiM
particle size. Analyses are performed isocratically using a 55% methanoll45% H 2 0 (V/V)
mobile phase at an 0.8 mLlmin. flow rate. The injection volume is 500 uL. A chromatogram of
a calibration standard is shown in Figure 1. ConfIrmation analyses are performed on a 250 mm
x 4.6 mm ID Zorbax cyano column (5-jiM mesh) with a 50% Methanoh/50% H 2 0 (VN) mobile
phase at a flow rate of 1.0 mL/min.
Analytical standards are obtained from the U.S. Army Toxic and Hazardous Materials Agency
(USATHAMA), Aldrich Chemical Co., and the Naval Surface Weapons Center (NSWC).
Calibration standards are prepared in 30% acetonitnle (ACN)170% H 2 0 to approximate final
sample extract composition, 33% ACNI7O% H 2 0. The calibration standards are prepared at
five to eight concentration levels in the range of 1 - 200 j.ig/L. Sample extracts with target
analyte responses above the highest standard are diluted within the calibration range. Spiking

solutions are prepared in acetonitrile.
Qptmization of SPE Variables
The successful application of an SPE technique requires optimization of numerous process
variables, including sorbent type, sorbent mass, sample volume, sample flow rate, elution
solvent, elution solvent volume, and elution solvent rate. The variables will generally be
determined by the analyte set, analyte detection limit requirements, solvent extraction
efhciency, and the suitability of the elution solvent to the analytical system. Two of the critical
variables, sample volume (500 mL) and sample flow rate (10 mL/min.), were tested to ensure
that analyte breakthrough during sample extraction was not compromising the reliability of the
Two surface water samples which had been collected from a military instrallation were selected
for the first experiment. It had previously been determined that they contained high
concentrations of several munitions compounds. A composite sample was made by combining one
kter of each of these samples. Separate 500 mL and 1000 mL aliquots of this composite sample
were extracted by SPE and analyzed by HPLC!UV with the conditions described in the above
section. A direct injection analysis from the remainder of the composited sample was also
Table 1 summarizes the results of the above analyses. There was no significant difference
between the results obtained from direct injection analysis of the high level composite sample
and the results obtained by employing the SPE technique to extract a 500-mL sample at al 0
mlimin. flow rate. However, application of the SPE technique to the 1000 mL sample resulted
ii a greater than 20 percent loss of four of the eight target analytes found in the sample.
FIgure 2 plots the results of a second experiment to determine the optimum sample flow rate
through the extraction system when a 500 mL sample volume is employed. Five representative
target analytes were spiked at 40 - 50 jig/L into 500 mL aliquots of ASTM Type Il/H PLC water.
Duplicate samples were extracted at five different flow rates ranging from 2 to 50 mLimin.
Sample extracts were analyzed as described in the above section. The average percent
recoveries indicate that at flow rates above 10 mL/min. target analyte breakthrough becomes
Lower limits of detection for the target compounds were estimated by determining the method
detection limits (MDL5) as specified by the U.S. Environmental Protection Agency (U.S. EPA). 3
Seven 500-mL aliquots of HPLC grade water were equivalently spiked with an acetonitrile
sokalon containing the 14 target analytes. The target concentration level was about 5 times the
estimated MDL as determined by the analyst from instrumental responses and previous
experience. The seven spiked samples and an unspiked aliquot were analyzed by the method
described in the preceding section.
The MDL for each target compound is calculated by multiplying the standard deviation of the
seven replicate concentration measurements by the appropriate one-sided t-value
corresponding to n - 1 (6) degrees of freedom. The corresponding t-value for seven
measurements is 3.143. Table 3 presents the results of the MDL determinations, including the

mean percent recoveries for each of the target compounds. The mean percent recoveries of
tetryl (47.7%) and 1 ,3,5-TNB (68.2%) are low (<80%) compared to those of the other
target analytes. Both compounds are unstable in water at room temperature. This is very
noticeable at low concentrations. At concentrations greater than 1 jiglL, percent recovereles
for both analytes are generally greater than 80%.
481 surface and groundwater samples, collected at several military instalilations during the
late summer and fall of 1991, were analyzed for 14 nitroaromatic and nitramine compounds
using the SPE with HPLC/UV detection procedure described above. The analytical protocol that
was followed is described in detail in USATHAMA standard method UW32.1
One to eight of the 14 target compounds were confirmed in 82 of the 481 samples analyzed.
Figure 3 shows a chromatogram of a surface water sample containing eight of the target
analytes. The total target compound concentrations in the 82 samples ranged from 0.075 to
11,620 ig/L. Table 2 lists the distribution and concentration ranges of the target compounds
in the 82 samples with positive identifications.
Table 4 summarizes the surrogate recovery data for all samples analyzed, including laboratory
reagent blanks. With respect to method accuracy and precision as measured by the mean
percent recoveries and associated standard deviations of the surrogate compound, there is a
noticeable discrepancy between samples containing 2,4,6-TNT and those not containing 2,4,6-
TNT. This phenomenon can be explained by referring to the chromatograms in Figures 1 and 3.
The surrogate compound, 3,4-DNT, is not fully resolved chromatographically from 2,4,6-TNT,
which elutes immediately after it. Quantification of the surrogate recovery is consequently less
reliable than if baseline resolution were achieved. This is especially noticeable when the 2,4,6-
TNT concentration is significantly higher than the 5 jig/L surrogate concentration.
The findings to date confirm that the method is very reliable when applied to various water
matrices contaminated over a wide concentration range with the target nitroarornatic and
nitramine compounds. Currently our. laboratory is investigating the following: (1) extending
the method to include two nitroaliphatic compounds, PETN and nitroglycerine; (2) including a
wash step in the SPE process to help eliminate potential interfering contaminants; and (3)
including mass spectrometer (MS) and photo diode array (PDA) detector options.
An analytical method employing SPE with HPLC/UV analysis for the determination of 14
nitroaromatic and nitramine compounds in environmental water samples is reviewed in detail.
MDL determinations are also presented. Data obtained from several laboratory experiments and
the analysis of over 450 surface and groundwater samples collected at several military
installations are presented and discussed. Current method development work is mentioned.
We thank P. Durnas, D. Dabney, and S. McMillen for their laboratory assistance.

1. Environmental Science & Engineering, Inc., ‘Method for the analysis of Explosives in
Water by High Performance Liquid Chromatography Precertification/Certification
Report and Method Writeup”, May 1991, Contract No. DAAA15-90-D015, U.S. Army
Toxic and Hazardous Materials Agency, Aberdeen Proving Ground, Maryland.
2. Environmental Science & Engineering, Inc., Determination of Low-Level Explosive
Residues in Water by HPLC: Solid-Phase Extraction vs. Salting-Out Solvent
Extraction”, Proceedings of the Seventh Annual Waste Testing and Quality Assurance
Symposium, U.S. Environmental Protection Agency, July 1991.
3. Federal Register. , uDefinition and procedure for the determination of the method
detection limit,” Code of Federal Regulations , Part 136, appendix B, Oct 26 (1984).
4. 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.
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 90-3-, U.S.
Army Corps of Engineers, Cold Region Research & Engineering Laboratory, 1990.
6. SW846 Draft Method 8330, “Nitroaromatics and Nitramines by High Pressure Uquld
Chromatography (HPLC)”, Revison 1, December 1990, U.S. Environmental Protection
Agency, Office of Solid Waste and Emergency Response, Washington, D. C.

Table 1: Recovery Comparison: SPE vs. Direct Injection
Target Analytes Direct Inj. SPEI500 mL SPE/1000 mL
HMX 1390 1400 754
RDX 3120 2800 2020

Table 3: Method Detection Limit (MDL) Determinations
Target Mean Found Std. CRL Mean
Compound ( ug/L) ( ug.L) ( ug.L) %Rec .
HMX 0.221 0.180 0.008 0.024 81.5
F DX 0.183 0.180 0.010 0.031 98.4
1,3,5-TNB 0.096 0.065 0.004 0.014 68.2
1,3-DNB 0.072 0.069 0.004 0.013 96.4
Tetryl 0.123 0.059 0.014 0.043 47.7
0.090 0.087 0.007 0.022 96.5
2,4,6-TNT 0.101 0.085 0.008 0.025 83.9
4-Am-2,6-DNT 0.125 0.123 0.012 0.038 98.5
2-Am-4,6-DNT 0.102 0.095 0.009 0.027 93.5
2,6-DNT 0.101 0.095 0.007 0.023 94.9
2,4-DNT 0.062 0.059 0.002 0.007 96.5
2-NT 0.168 0.172 0.006 0.019 102
4-NT 0.217 0.190 0.008 0.025 87.4
3-NT 0.175 0.167 0.003 0.009 95.6
Note: n = 7, 1-value = 3.143
Table 4: Surrogate Recovery Results
Sample Type Number* Mean %Rec. Std. Dev. Range (%Rec.
Field 475 101 13 57 - 158
Field/Confirmations 76 98.4 1 9 57 - 158
Field/Confirmations 46 101 9.5 81 -106
(wlo 2,4,6-TNT)
Field/Confirmations 28 94,6 26 57 - 158
(with 2,4,6-TNT)
ReagentBlanks 35 100 8.4 81 -115
* outliers ommitted

c h y . . in__ t . . c i -
Date : 5Il6/ l 2:58 PM
Low Point -0.00 .0
Plot Scale: 100 .V
Page 1 ol 1
High Point 100.00 .Y
Retention Time min]
Figure 1: Chromatogram of a Calibration Standard
Fi leMa.e : D: 27O0\DATA\UTRC0l2.raw
Start Tue 0.00 sin tnd Tue : 30.00 •in
Scale Factor: 0 Plot Offset: 0 sV
t ) U
to tDN
o cq
C 0 )
tOO 0)
,a:tc L (or-
r ) CN
c..J -
C’.J (N (N
1,3,5-TN B
3,4-0 NT
2,6-0 NT
2,4-0 NT
floIm I

%Rec 85
-0 TNB
2,4 DNT
2 4 10 20 50
m L/m in

c t-i r— in __ ci .- in
11/21/91 10:4E r
lc.w Feint : —(‘.01 t
S ft: (;
Paae 1 of I
Hiab Pnint 199.99 .V
Figure 3: Chromatogram of a Surface Water Sample
D:\27 0 0\DAJA\WLl-019.raw
Start Tii 0. 0 ,in End Ti. 30.0( iir
Sc Ie Fdctc,r: ( F Ict 0f1snt C’ V
I ,3,5-TNB
3,4-DNT *
2 ,6-DNT
2 ,4-DNT
r j
Time [ mir J

Gabriel La run , Pat Rethwil]. and Jim Matteson, Pace INC., 1710
Douglas Drive N., Mi tneapo1is, 55422 Phone: (612) 525—3352,
FAX: (612) 525—3377
The analysis of water for explosives has traditionally been dons by
“dilute and shoot” aethods, such as EPA’s Method 8330, or by
methods which giv, some degree of concentration, such as solid
phas. extraction (STE) tithes or liquid/liquid extr ction (LLE)
techniques, using 47mm SPE discs containing polystyrene divinyl
benzene an alternativ, method was developed for explosives
analysis. Features of the method are 500-mi water samples, •lutiort
with approximately 10-mi solvent, concentration to 2-mi, followed
by HPLC determination of the 13 analytes. Good results were
obtained for the analytas, with extraction times as short as 5-
min.f 500-mi. This presentation will detail the extraction method
development and the results obtained.

Duane K. Root and Walter W. Li, Technology Development
Laboratory, IT Corporation, Knoxville, Tennessee 37923
Gregory E. Johnson, EA Engineering, Science, and Technology,
Sparks, Maryland 21152
Work was conducted to define procedures for the analysis of
soil, sediment and water samples for nionomethyl mercury
(NNHg) by gas chromatography and electron capture detection
(GC/ECD). First attempts were conducted using published
extraction and cleanup procedures emplç y ng a
thiosulfate/CuC1 2 treatment and back extraction. ’ This
procedure did not provide satisfactory results on our
sediment matrix spike samples and an alternate procedure was
modified and applied. The alternate procedure was adapted
from a published fish analysis method, which employed
sample pre—extraction to remove organic interferences.
Extraction of MMHg was then performed with benzene after
being released from the matrix as the chloride (MNHgC1) by
addition of HC1. This procedure was used with success on
the most difficult humic/organic sediment samples. The
method provided reproducible results with quantification
limits to 5 ppb Hg. Relative percent differences of matrix
spike pair analyses were within ±22%, and analyte recoveries
for the more organic sediments were in the range of 47—118%.
A trend of higher analyte recoveries were obtained for the
more humic/organic sediments; while a sandy/silicate soil
consistently gave much lower recoveries. Analysis of
surface water samples at these concentrations (3 ppb) did
not present many problems because of the significantly lower
amount of extractable interferences compared to soil or
sediment samples.
Specialized analytical procedures have proven to be
particularly useful for characterization of potential risk
to ecological receptors at hazardous waste sites. The
ability to characterize potential bioavailability of site
contaminants is further enhanced through the use of
analytical procedures that can provide more information than
is typically available using standard TCL/TAL analyses.
Procedures such as acid volatile sulfide with simultaneously

extracted metals (AVS/SEM) and direct quantification of
organometallics such as organolead, tributyltin and methyl
mercury have significantly benefitted the risk assessment
Several studies have noted significant correlations between
monomethy mercury (!INHg), total mercury and total organic
carbon. 4 Accurate knowledge of NNHg concentrations are
necessary to develop these relationships which assist in
prediction of methyl mercury exposure concentrations. In a
recent project application, direct quantification of MI4Hg
was used in conjunction with other analyses such as total
mercury, total organic carbon, pH, Eh and temperature to
characterize mercury contamination in sediment samples from
a southern Gulf Coast hardwood riverbottom swamp. The data
were used in conjunction with surface water analyses, biota
tissue analyses, and ecological risk modeling and risk
calculation to assess potential impacts to ecological
To quantify speciated forms of mercury in a sample, most
often the organic mercury component is extracted after acid
halide addition with an organic solvent. Organic mercury
can then be determined by analyzing the extract separately,
or it can be obtained as a difference from total mercury
analyses of a sample before and after extraction. MNHg
specificity can be achieved by GC/ECD analysis of the
organic extract, but in practice the analysis, particularly
for sediment samples, is riampered by matrix interferences
and constant GC column conditioning steps. The objective of
our work was to develop a reliable GC/ECD protocol for
analysis of environmental samples, i.e., soils, sediments
and waters to determine concentrations of MMHg in the low
ppb range.
GC/ECD methods of analysis specific for NMHg have been
develope and used for analysis of biological
samples, —3,7,8 but at the time of our work, procedures for
environmental samples had not been as successfully
developed. One widely used procedure for biological samples
isolates the MHHg from interferences in the organic extract
of a sample by employing a sodium thiosulfate (Na 2 S 2 O 3 )
solution extraction followed by Cud, or CuBr 2 a d t on and
an organic back extraction prior to C analysis. “° This
procedure was tested on organic extracts of sediment
samples, but gave unsatisfactory results for the sample
matrices we were analyzing due to low analyte recoveries.
An alternative procedure for biological samples is to
extract the sample three times with acetone and then once
with benzene to remove interferences. The sample is then
acidified with HC1 and re-extracted with benzene to extract

MMHg as the chloride (MMHgC1) . The extract is then
analyzed directly by GC/ECD. This procedure was adapted and
used for the analysis of !INHg in soil and sediment samples.
Surface water samples did not present difficulty for
analysis at 1-3 ppb concentrations by direct analysis of the
organic extract of an acidified sample.
Sediment and Water Analytical Procedures - The sediment
extraction procedure was adapted from a fish analysis method
from AOAC. The sediments and soils were extracted by
weighing a 4g aliquot into a 40 milliliter glass vial and
adding 30 mis of acetone. The vial was shaken by hand for
30 seconds and centrifuged at 1700—1800 rpm for 10 mins.
The acetone was decanted off and discarded and a second and
third pre—extraction with acetone was also decanted off and
discarded. Next, 30 mis of benzene was added to the vial
and shaken and centrifuged. The solvent was again discarded
and then 4 mis of 1:1 HC1 was added and mixed with the
sediment. Then, 5 mls of benzene was added to the vial.
The vial was shaken by hand for 2 minutes and centrifuged as
before with the benzene layer being pipetted off and
collected into a separate vial. The extraction was carried
out twice more with all the successive extracts combined
into one vial and the final volume was measured. The volume
needs to be determined because there will be some residual
benzene from the pre—extraction step. In some cases there
were emulsions present. These were broken up by using a
glass stirring rod, adding more 1:1 HC1, or centrifuging the
vial again at higher rpm. The benzene extract was then
dried over anhydrous sodium sulfate and analyzed directly.
Water samples were extracted by placing a 10 ml aliquot of
the sample into a 40 ml glass vial, adding 1 ml of 1:1 HC1
and mixing the solution. Then 3 mls of benzene were added
and the vial shaken for 2 minutes. After allowing the
layers to separate, the benzene layer was drawn off and
another 3 ml portion of benzene was added to the water. The
vial was again shaken and the benzene layer was removed and
combined with the first benzene extract for a total volume
of 6 mis. The extract was then dried over anhydrous sodium
sulfate and analyzed directly.
Matrix spike samples were prepared by spiking a solution of
MMHgC1 in benzene onto the matrix at a concentration of 2—3
times the anticipated sample concentration.
The analysis was performed on an Hewlett Packard 5890 gas
chromatograph equipped with an electron capture detector.
Data was collected on a Perkin Elmer/Nelson Turbochrome

software system. A Restek Rt -5 GC ColUmn (30M X 0.53
ID, 1.0 UN film thickness) was used to perform the
chromatography. The following GC temperatures were used;
injector temperature, 210°C; detector temperature, 270°C; GC
oven, 110°C for 5 minu..es then increased to 115°C at
10°C/mm. and held for 2 minutes. The helium carrier flow
was 7 mis/mm. The GC column was initially pretreated with
several 25 ul injections of 1000 ppm mercuric chloride in
benzene. Then at the beginning of each shift, two 10 ul
aliquots a few minutes apart were injected. Once the
baseline became stable, standards and samples could be run.
Concentrations were determined using a five point external
standard initial calibration of monomethyl mercuric chloride
(MMHgC1) and daily continuing calibrations (within 25%)
bracketing the analyses. Chromatographic conditions were
monitored by running a standard every few samples and noting
peak shape and retention time as well as response.
Thiosulfate/CUC12 Cleanup - This cleanup was used on
standard solutions and a few matrix spike samples to assess
performance, but was not used in the analysis of samples.
C1ean ip were performed as described below and in previous
work. ” One milliliter of benzene extract was placed in a
glass vial and shaken with 1 ml of .OlM or 0.lM Na 2 S 2 O 3 for
30 seconds. After allowing the layers to separate, the
benzene layer was removed and 0.5 ml of 0.5M Cud 2 solution
was added to the thiosulfate solution extract. A fresh 0.5
ml aliquot of benzene was then added to the vial and the
mixture was shaken for 30 seconds. The layers were allowed
to settle and the benzene layer was analyzed directly.
Chromatography - Much of the preceding work, which used GC
analysis for MNBg determination, utilized packed GC columns
(5% DEGS). 3 ’ 7 These columns need to be conditioned and
continually refreshed by injecting a solution of mercuric
chloride in benzene. We explored the use of megabore
columns seeking a column which might provide good resolution
with good durability and require little conditioning. Our
efforts centered on using the Restek Rt -5 column. The Rt -
5 GC column provided reasonable reso ution of MMHg with
retention times generally less than five minutes; however,
the system suffered the same need for conditioning with 1000
ppm HgCl 2 in benzene as others have described. 3 ’ 7 The
resolution and retention time of methyl mercuric chloride
was dependent on the condition of the system. The GC peak
broadened with increased tailing and the retention time
increased as the system condition degraded. Figures la and
lb show the chromatographic response of a 31.3 ng/m]. MMHgC1

standard on a fresh conditioned GC system, and after
degradation from sample analysis, respectively. When the
system degraded the column required reconditioning to
refresh performance. The rate of system degradation was
most dependent on the amount of co-extractant material that
was in the extracts being injected. Little degradation was
noted with analysis of standards or water sample extracts;
however, analysis of sediment extracts without cleanup steps
degraded the system rapidly. In some cases, reconditioning
was required after analysis of a single extract. Sample
cleanup procedures were essential to maintain
chromatographic performance. Using the sample cleanup
described, 20—30 samples could be analyzed between
In addition to column conditioning, we also had difficulty
maintaining good chromatographic peak shape if we were not
careful about maintaining a clean and inert injection port.
The chromatography suffered considerable tailing and what
appeared to be peak splitting unless the injection port and
liners were silanized prior to use and the liners were
exchanged regularly.
The performance of the GC analytical system was
characterized using our lowest calibration standard at 5
ng/ml NMHgC1. Analysis of the standard was performed five
times and the mean and RSD of the responses were calculated.
The instrument detection limit was calculated to be 0.65
ng/ml (3x Stand. Dev.) and the quantification limit was 1.1
ng/ml (5x Stand. Dev.).
Water Samples — The surface water samples presented little
difficulty for MMHg analysis in the low ppb concentration
range. The solids content of samples were well below 1%,
and little chemical interference was encountered in the
analysis. The estimated quantification limit for the water
samples was 3 ppb as MMHgC1. Matrix spike and spike
duplicate QC analyses were performed by spiking sample
aliquots prior to extraction. Table 1 details analyte
recovery and relative percent difference (RPD) results.
Precision of the analyses expressed as RPD was within 1.9 to
21%, and analyte recoveries ranged from 67% to 102%.
Thiosulfate/CUC12 Cleanup of Sediment Samples - Initial work
on sediments was conducted using the Na 7 S 2 O 3 /CUC1 2
extraction/back extraction cleanup procedure. The cleanup
steps were first performed on standard solutions of
monomethyl mercuric chloride (MMHgC1) in benzene.

Recoveries of NNHgC1 from this work ranged from 44% to 90%.
It was noted that some analyte recoveries were improved by a
second extraction of the thiosulfate/CuC1 2 solution with an
additional portion of benzene.
The cleanup steps were then performed on two different
sediment matrix spike extracts, and the results are detailed
in Table 2. The MNHgC1 analyte recoveries for both samples
were low, 13% to 46%, and the lower recoveries for sediment
#1 is consistent with what was judged to be a higher level
of interferences present in that sample. A subsequent
analysis of the extract from that sample, which had been
treated with the thiosulfate/CuCl , cleanup, revealed that a
significant amount of r4NHg was no € being extracted into the
thiosulfate solution in the first step of the cleanup. To
look at the effect of matrix interferences on cleanup
recovery, the extract from sediment #1 was subsequently
diluted 1:1 to reduce the concentration of interferences,
and then an aliquot was extracted with thiosulfate solution.
An analysis of the benzene layer showed that at least 74% of
the MMHgC1 still had not been extracted into the thiosulfate
solution. Further dilution of sample extracts to reduce the
concentration of interferences was not investigated because
analyte detection limits would have been compromised. No
further work was done with this cleanup procedure.
Pre—extraction Cleanup of Sediment Samples — Sample cleanup
steps are desired to remove chemical interferences, but it
is equally important to remove organic co—extractants from
the extracts being analyzed to provide stability to the
chromatographic system. The pre—extraction cleanup
procedure, which was applied in the analysis of sediment and
soil samples, offered improved performance over the
Na 2 S 2 O,/CuC1 2 cleanup for samples with humic/organic
charac€er. Sample concentrations were reported down to an
estimated maximum possible concentration of 5—10 ppb as
MMHgC 1. Figures 2a and 2b show the chromatographic
responses for an unspiked and MMHgC1 spiked sediment sample,
respectively; both analyzed without using cleanup steps.
Figure 2c shows the same MNHgC1 spiked sample analyzed using
the pre—extraction cleanup. The chromatogram displays a
strong early response, which may be due to acetone residual
in the extract; however, in the area of NNHgC1 elution, the
chromatogram is free of significant interferences. Table 3
details results of analyte recovery and relative percent
difference (RPD) from matrix spike and spike duplicate QC
samples of sediment and soil matrix types. Precision of the
analyses expressed as RPDS were within 2.0% to 22%. Analyte
recoveries ranged from 47% to 118% for organic matrices and
were significantly lower for one light color sandy matrix

soil. A duplicate set of MS/MSD samples were prepared and
analyzed for this matrix with similar recovery results as
the first set. MMHgC1 recoveries for this sample ranged
from 11% to 22%.
It is apparent that the more humic/organic samples have more
sites for MMHg to bind with the matrix, which allows the
pre—extraction steps to remove organic material and leave
MMHg behind. The addition of HC1 is an important step to
free MMHg from the matrix to allow the subsequent benzene
extraction. It would be expected that sandy samples have
fewer sites for MMHg to bind with the matrix so that the
addition of HC1 is probably not as important for the
extraction of MNHgC1. Lower recoveries of NMHgC1 for the
sandy soil most likely occurred because of losses in the
pre—extraction steps prior to HC1 addition.
• The procedures provided consistent results which met
performance objectives for the analysis of surface
waters and most soil and sediment types.
• MMHgC1 spike recoveries on soils and sediments were
higher for the more humic/organic samples.
• Low MMRgCL spike recoveries were obtained for a
sandy/silica type soil/sediment. This most likely
reflects the smaller MMHg binding capacity of this
matrix type, which allowed loss of MHHgC1 through the
pre—extraction steps of the sample cleanup.
• Quantification limits for MNBgC1 were in the low ppb
range for both water (3 ppb), and soil/sediment samples
(5—10 ppb).
• Negabore GC column requires conditioning with HgC1 2 ;
injection port components need to be silanized; and
injection port liners need to be exchanged regularly,
to obtain acceptable chromatography.
• GC performance and stability was maintained for a
longer period of time when the amount of analyte co—
extractants injected into the GC was minimized,
emphasizing the necessity for sample cleanup steps.
• A comprehensive extract cleanup procedure is needed
which can handle extracts from samples of all matrix

1. Filippelli, Marco, “Determination of Trace Amounts of
Organic and Inorganic Mercury in Biological Materials
by Graphite Furnace Atomic Absorption Spectrometery and
Organic Mercury Speciation by Gas Chromatography,”
Anal. Chem., 59, 116—118 (1987)
2. Baldi, Franco and Marco Filippelli, “New Method for
Detecting Methylmercury by Its Enzymatic Conversion to
Methane,” Environ. Sci. Technol., 25, 302—305 (1991).
3. Official Methods of Analysis , Association of Official
Analytical Chemists (AOAC), 25.146—25.152, “Mercury
(Methyl) in Fish and Shellfish Gas Chromatographic
Method,” Washington, DC (1984).
4. Bartlett, P. D. and P. J. Craig, “Total Mercury and
Methyl Mercury Levels in British Estuarine Sediments-
II,” Wat. Res., 15:37—47 (1981).
5. Rudd, 3. W. M., M. A. Turner, A. Furutani, A. L. Swick
and B. E. Swick, “The English-Wabigoun River System:
II, Can. J. Fish. Aquat. Sci., 40:2206—2217 (1983).
6. Winfrey, M. R. and 3. W. M. Rudd, “Environmental
Factors Affecting the Formation of Methylmercury in Low
pH Lakes,” Environ. Toxicol. Chem., 9:853—869 (1990).
7. Horvat, M., A. R. Byrne and K. May, “A Modified Method
For The Determination Of Methylmercury By Gas
Chromatography, ‘ Talanta, 37, 2, 207—2 12 (1990).
8. Cappon, C. 3. and 3. Crispin Smith, “Chromatographic
Determination of Inorganic Mercury and Organomercurials
in Biological Materials,” Anal. Chem., 49, 365 (1977).

Table 1
Quality Control Summary For Water Samples
Results of Matrix Spike And Spike Duplicate Analyses
(MHHgC 1)
Spike Level
% Recovery

Table 2
Results of Matrix Spike Analyses Using
Thiosulfate/CuC1 2 Cleanup
Spike Level
% Recovery
Sediment #1
1:1 Ext.
Sediment #2

Table 3
Quality Control Summary For Soil/Sediment Samples
Results of Matrix Spike And Spike Duplicate Analyses
Spike Level
% Recovery
sandy soil

Chrornatograms of MMHgC1 Standard
(31.3 ng/xnl)
Retention Time (mm.)

Chromatograins of Unspiked and NMHgC1 Spiked
Sediment Sample
Retention Time (mm.)

John Simon Jr. , Chemist, U.S. Environmental Protection
Agency, National Enforcement Investigations Center, Denver,
Colorado 80225, 3. Lowry, U.S. Environmental Protection
Agency, NEIC, Denver, Colorado 80225, and C. Ramsey, U.S.
Environmental Protection Agency, NEIC, Denver, Colorado
Purgeable organic halides (POX) are low molecular
weight, volatile organic compounds containing halogen atoms
(Cl, Br, I) which may be “purged” from water by sparging at
mildly elevated temperatures (e.g. 55 C). Examples include
trihalomethanes such as chloroform, and common solvents such
as methylene chloride and perchioroethylene. These POX
compounds are measured to obtain non—compound specific,
screening information about samples such as waste water and
drinking water. POX data are useful for independently
checking results obtained from more time consuming, more
expensive GC/MS analyses. These POX species are routinely
measured in process wastewaters by the pulp and paper
industry. POX are typically measured by
purging/combustion/coulometric titration methods. An
appropriately collected sample is introduced into a sparging
device, which is heated to 55 C. The sparged volatile
compounds are then swept into a pyrolysis tube using a
stream of oxygen. Hydrogen halides (HC1, etc.) are thereby
produced, collected in a solution of aqueous silver acetate
and are subsequently measured using a couloinetric titration.
Present implementations of the above POX technique, such as
EPA Method 9021, have been found by the authors and others
to be unacceptable for low concentration samples. When the
total POX concentration is less than about 1000 ug/l, poor
recoveries (i.e. less than 90 percent) are found for
representative analytes such as chloroform, trichioro—
ethene, chlorobenzene, etc. This limitation makes EPA 9021
unacceptable for routine analyses conducted at EPA—NEIC on
low concentration samples such as drinking waters. Research
at EPA-NEIC has identified the key factor responsible for
low recovery to be an insufficient flow of the oxygen used
for the combustion process. The
purging/combustion/couloifletriC titration processes was
systematically explored until this responsible component

was identified. Through these experiments it was determined
that increasing the flow rate of oxygen in both the sparging
device and the pyrolysis tube improved recoveries
dramatically. In order to achieve recoveries satisfactory
for drinking water analysis an oxygen flow rate of at least
800 mi/minute was required when using a 10 ml sample. This
flow rate was found to be incompatible with an off-the-shelf
instrument from one manufacturer but was compatible with
another instrument from another manufacturer. A simple
modification to the first type of instrument resulted in
acceptable recovery for POX compounds. With increased
oxygen flow, and depending on the POX compound(s) present,
recoveries of 99 ± 9.6 (One S.D.) percent can be obtained
with samples containing 11-20 ugh purgeable organic
chlorine. Because of the intrinsic low cost of POX
measurements, and with this enhanced ability to provide
reproducible results at low concentrations, we conclude that
POX measurements at drinking water levels are practical
using this approach. The technique allows for the screening
of a large number of POX samples easily. Appropriate
modifications to Method 9021 will be discussed.
Purgeabie Organic Halides (POX) are volatile organic
compounds containing halogen atoms (Cl, Br, I) that can be
purged at slightly elevated temperatures (i.e. 55° C). POX
species include solvents such as trichioroethene,
tetrachloroethene and other halogenated organics. Many of
these compounds are toxic and some are thought to be
carcinogenic; thus they are important from a regulatory
stand point. POX compounds are also a good indication of
anthroprogenic contamination in media such as ground water.
A common approach to POX analysis involves a oxidative
pyrolysis/coulometric titration technique. (1—2) This method
has been widely used in environmental analysis since the
early 1980’s due to the promulgation of regulations under
the Resource Conservation and Recovery Act (RCRA).
(Hazardous waste landfills are required under RCRA to
monitor ground water for Total Organic Halides (TOX).)
Diagram 1.1 shows the sequence of events involved in POX
analysis using the oxidative pyrolysis/coulometric titration
The above mentioned method of POX analysis can be broken
down into three major components: 1) Purging, 2)

combustion/dehydration, and 3) titration. The first
component of this sequence, purging, will be discussed next
with the other components following.
Pox analysis using the above method begins with the
introduction of a 5-10 ml sample into a heated sparging
chamber. The sparging chamber is heated to 23° C - 700 C to
facilitate “purging”. The sample is then sparged with
oxygen and the volatile halogenated compounds are purged
from the sample. These volatilized compounds are then
swept into a pyrolysis tube via a stream of oxygen.
The second step, combustion/dehydration, includes separate
combustion and a dehydration processes. The combustion
process occurs in a pyrolysis tube heated by an electric
furnace to approximately 900° C. After sparging the
volatilized POX compounds enter the heated pyrolysis tube
and combustion takes place in an oxygen enriched
environment. During the combustion process halide ions
disassociate from their parent compounds. These halide ions
then react with hydrogen or each other forming hydrogen
halide (HX) or X 2 . The HX is then swept via an oxygen
stream into a dehydrating tube which removes the water vapor
present from both the combustion and purging steps. The
water vapor is removed because water adversely effects the
titration process. The mixture of gases containing HX
enters the titration cell, again via an oxygen stream.
The third component of POX analysis described herein is a
coulometric titration. This titration can be summarized by
the following steps: Step 1) HX are introduced into an
electrolyte solution containing electrolytically generated
silver ions. Step 2) The silver ions and HX react to form
AgX. Step 3) The amount of AgX generated is measured
coulometrically and related to the POX concentration using
Faraday’s law.
The above mentioned technique provides the analyst with
quick, inexpensive, non—compound specific screening
information. This screening information can. then be used to
determine if more costly and time consuming GC/MS work must
be done.
Present methods utilizing the oxidative
pyrolysis/coulometric titration technique for POX analysis,
such as EPA method 9021, have been found by the authors and

others to be produce low recoveries for low concentration (<
100 ugh purgeable Cl) analysis.(3) Since these recoveries
are less than 90 %, this method is unsuitable for accurate
low level work. The present research has been conducted to
investigate why recoveries were low when using commercial
instrumentation for POX analysis. Also, of interest, was if
the commercial instrumentation could be easily modified, if
necessary, to obtain the desired recoveries at low POX
It was determined that a deficiency of oxygen in the
sparging and pyrolysis chambers was responsible for the low
rSecoveries. A simple modification made to a commercial
instrument allowing for an enhanced oxygen flow rate was
made and recoveries greater than 90 % at purgeable Cl
concentrations less than 1,000 ugh were obtained.
A series of experiments utilizing the increased flow rate
were then performed on a group of test compounds
(trichloroethene, tetrachioroethene, chloroform,
chlorobenzene) in distilled water. Results from these
experiments will show that reproducible recoveries greater
than 90% can be obtained at POX concentrations between 11
and 500 ugh purgeable Cl.
Materials and instrumentation: Deionized and distilled
water was used as the solvent for all solutions. Reagent
grade silver nitrate, laboratory grade acetic acid and high
purity oxygen was obtained from commercial sources. Stock
solutions of trichioroethene, tetrachloroethene,
chlorobenzene and chloroform were prepared from US—EPA
standards. Stock solutions were made from trichloroethene
obtained from commercial sources. All stock solutions were
prepared by diluting the appropriate volume of standard or
neat compound with methanol. A portion of the stock solution
was then spiked directly into a syringe containing 10 ml of
deionized water. The spiked solution was then transferred to
the sparger and sparged immediately thereafter.
A Mitsubishi total organic halogen analyzer model TOX-lO was
used throughout. The instrument parameters for all analyses
are in table 1.1 The TOX-lO was modified by attaching a
commercial flow meter to the oxygen inlet. (See diagram 1.1
for attachment)

Effect of oxygen flow upon POX recovery: A series of
experiments were conducted which measured the recovery of
trichloroethene (544 ugh purgeabie Cl) as a function of
oxygen flow. Graph 1.1 depicts the results of this
experiment. It is clear that recovery is greatly affected
by the oxygen flowrate and that this should be adjusted to
at least 800 mi/minute for quantitative (> 90 %) recovery.
It was also found that argon is not needed for any part of
the purging/pyrolysis/coulometric titration sequence.
The instrument manufacturer (Mitsubishi model TOX-lO)
recommends using an oxygen flowrate of 200 mi/mm. The
higher fiowrates used herein were achieved by incorporating
a ball-type flowineter into the oxygen stream.
Recoveries of POX species from distiled water using enhanced
02 flow: Experiments were performed on a group of test
compounds (trichloroethene, tetrachloroethene,
chlorobenzene, chloroform.) in distilled water. Table 2.1
illustrates that the recoveries of these test compounds are
favorable as the purgeable Cl concentration decreases from
500 ug/l to 11 ugh. Also, the relative standard deviations
in table 2.1 indicate that precisions at both the high and
low POX concentration levels are favorable even at low POX
It was described herein that increasing the oxygen flow rate
in both the pyrolysis and sparging chambers increases
recoveries to better than 90 % at low POX concentrations
(10-20 ug/l purgeable Cl). The precisions are also
favorable at the 10- 20 ugh purgeable Cl levels. To
achieve the flow rates needed for the improved recoveries a
simple modification of a commercial instrument (addition of
a flowmeter) is necessary and easily accomplished.
This study made use of number of test compounds in deionized
water. While recoveries were favorable under these
conditions, these conditions are not necessarily
representative of samples where a substantial matrix is

present, such as contaminated industrial wastewater.
Further study would be needed to determine if this modified
method is applicable to adverse matrices.

Electric Furnace

delay 1 10mm.
delay 2 0 ,0mm.
temp 1 850°C
temp 2 900°C
temp 3 55°C
gain 1 1.5—2.5 Q/mV
gain 2 4.5—5.5 Q/iuV
gain 3 9.5—10.5 Q/mV
end point 290-300 mV
sensitivity 1 mv
gas selection switch 02 mode
flow rate > 800 mi/mm.
Ar flow rate 0.0 mi/mum.

Oxygen Flowrate (mi/minute)
1 00
0.4 0.6 0.8 1

Compound (ug/1
Number of
Trich loroethene
Ch lorobenzene

(1) Harper, F. Ground Water Monitoring Rev. 1984, Winter,
(2) Takahashi, Y.; Joyce R. J. “Chemistry in Water Reuse”:
Ann Arbor Science: Ann Arbor MI. 1981; Vol. 2,
Chapter 7.
(3) Riggin, R.; Lucas, S.; Lathouse, J.; Jungclaus G.;
Wensky, A. “Development and Evaluation of Methods for
Total Organic Halide and Purgeable Organic Halide in
Wastewater”, EPA—60014-84—008, January 1984, Contract
Number 68—03—2984

Mike Zimmerman, ICF/QATS Laboratory, Las Vegas, Nevada, 89120
Determination of the CLP target pesticides in relatively clean samples
has been performed in our laboratory using an ITS4O ion trap mass
spectrometer in an effort to demonstrate and evaluate the performance of the
instrument in several configurations, including electron impact (El) and
chemical ionization (CI). Splitless and on-column injection techniques were
also compared. Results to date indicate that with El ionization the
instrument can detect and qualitatively identify most of the CLP target
pesticides at levels equivalent to or below the CLP contract required
quantitation limits (CRQLs) for electron capture (ECD) analysis. These El
experiments were conducted with full scan acquisitions and provided
unambiguous NIST library spectral matches for most of the target analytes
at the low ECD calibration level, 0.005 or 0.01 ug/mL. To date, chemical
ionization using methane and isobutane has not provided the sensitivity of
electron impact ionization. Three-point calibration relative response
factors using phenanthrene-dlO as the internal standard demonstrate that El
with on-column injection is the preferred configuration. Other data from
wide range pesticide calibration, pesticide water sample analysis, and some
semivolatile and polychiorinated dibenzo-p-dioxins and furans (PCDD/PCDF)
analyses are presented and compared to traditional quadrupole analyses.
Results of all of these experiments suggest that ion trap mass spectrolnetry
is a versatile and reliable analytical approach for many Contract Laboratory
Program (CL?) sample analysis requirements.
As a result of the exceptional sensitivity and qualitative
capabilities of ion trap mass spectrometry, this technique can be expected
to play an ever-increasing role in regulatory analysis for many
environmental sample types. The only current EPA method that we have
reviewed which describes the use of ion trap technology is Method 525 (Rev
2.1, 1988) which requires calibration for many pesticides to 0.1 ng/uL (100
pg injected) for water analysis. We have examined in detail the performance
of an ITS4O ion trap for analysis of CL? target pesticides, semivolatiles,
and PCDD/PCDF, in several configurations including electron impact (El),
chemical ionization (CI), and splitless and on-column injection.
The CLP method OLMO1 for pesticides targets twenty (20) compounds, not
including toxaphene or the Aroclors. The Contract Required Quantitation
Limits (CRQLs) require detection of 5 or 10 picograins when 1.0 uL injections
are performed (Methoxychior CRQL is 50 pg). The method describes use of
electron capture detection (ECD) with two complementary gas chromatography

columns with splitless or on-column injection. QC/QA requirements
associated with calibration, CC performance, and extraction/clean-up
techniques are included in the method. Several possible advantages of using
mass spectrometry instead of the traditional ECD include elimination of the
second confirmation CC column, more reliable target compound identification
resulting from mass spectral interpretation, and potential identification
of non-target compounds.
Most modern quadrupole mass spectrometers provide full scan detection
of approximately 1 nanogram of most of the chlorinated CLP target
pesticides. Sector instruments provide slightly better full scan detection
limits. Quadrupole single ion monitoring (SIM) will provide detection to
approximately 100 picograrns, while sacrificing the qualitative advantage of
full spectral acquisition. Again, single ion monitoring with sector mass
spectrometers provide improved detection limits to approximately 1-10
The ITS4O ion trap sensitivity specification is 10:1 signal-to-noise
at m/z 284 for full scan detection and library searchable spectra
acquisition of 10 picograms of hexachlorobenzene with El/splitless
injection. This specification was easily met confirming that this
instrument provides approximately 100-fold improved El full scan sensitivity
over quadrupole instruments. The overall full scan sensitivity of the ITS4O
for most target analytes is in the range of sector instruments operated in
single ion monitoring modes. Although the ITS4O can scan over narrow mass
ranges similar to quadrupole or sector SIN scan ranges, the fundamental
storage and scanning functions of the ion trap do not provide the same
improved detection limits with narrow SIM-like scan windows. Many
improvements and developments in the fundamental scanning functions of the
ion trap, including improved resolution, tandem MS/MS capabilities, and
greater sensitivity can be expected to be available to the average user in
the next few years.
Initial experiments to determine the best approach to low level
pesticide analysis involved calibration of the ITS4O for the pesticide
target compounds in four configurations at target analyte concentrations
defined for ECD analysis in OLMO1.0 (5-160 ng/mL). In each configuration
effort was made to optimize the ion trap performance for maximum sensitivity
and spectral integrity, which in the use of chemical ionization required
several efforts at optimizing reagent gas pressure. Table 1 lists the
average three-point relative response factors using phenanthrene-dlO as the
internal standard. Mass spectral base peaks were used for all calculations.
Missing values indicate that one or more of the calibration concentrations
were not detected with reliable library searchable spectra. Chemical
ionization spectra were compared to the Food and Drug Administration (FDA)
CI mass spectral compilation, which in many cases did not fit the ITS4O
As in all of the pesticide experiments a 30 m, 0.25 mm ID, 0.25 um
film DB-5ms column (J&W) was employed with helium carrier gas set at a flow
velocity of 38 cm/sec. This column was found to provide the lowest bleed
compared to the same dimensional DB-5 and DB-5.625 columns. On-column 1.2
uL injection was performed with a Varian SPI injector operated at 60 C
(heated to 280 C after 2 minutes) without use of a guard column. The CC
program was 80-280 C with a 2 minute initial hold, 8 C/minute ramp, and 10
minute final hold, allowing complete analysis through decachlorobiphenyl in
approximately 34 minutes.

TABLE 1. ITS4O Three-Point Average Relative Response Factor Summary
Target El El Isobutane CI Methane CI
Pesticides Splitless On-Column On-Column Cki-Coliiin
alpha-BHC 0.287 0.337
beta-BHC 0.235 0.287
delta-BHC 0.227 0.283
ganuna-BHC (Lindane) 0.284 0.289
Heptachior 0.039
Heptachlor Epoxide 0.142 0.149 0.783
Endosulfan I
Dieldrin 0.174
4,4’-DDE 0.359 0.337 0.328 0.042
Endrin 0.030
Endosulfan II
4,4’-ODD 0.421 0.336 0.564 0.290
Endosulfan Sulfate 0.068 0.100
4,4’-DDT 0.244 0.259 0.358 0.165
Methoxychlor 0.331 0.322 0.827 0.522
Endrin Ketone 0.065 0.102
Endrin Aldehyde 0.060
alpha-Ch lordane 0.175 0.354 0.387
gamma-Chlordane 0.176 0.312 0.433
Tetrachloro-m-xylene (Surr) 0.363 0.258
Decachlorobiphenyl (Surr) NA 0.406
14 of 21 target analytes were detected in both El splitless and El on-
column analysis. Three additional TCLs were detected with on-column El
analysis. From El splitless to El on-column we note 10 of 14 increasing
analyte response factors with increases ranging from 1.7 to 50.6%, averaging
Chemical ionization with isobutane and methane was largely
unsuccessful at detecting the low and mid-level calibration standards. In
the best case with methane, 7 of 22 target analytes were detected at all
calibration concentrations. Isobutane provided improved detection for those
analytes with 3-5 chlorines but sensitivity quickly degraded with those
analytes with 6 or more chlorines. Recent instrument upgrades in CI
hardware and scanning functions might provide improved sensitivity.
The next set of experiments involved triplicate 12-point calibrations
from O.5X to 80X CRQLS (SAS No. 7l47-HQ). These CRQL factors represent from
2.5 or 5.0 to 400 or 800 picograms injected for all analytes except
Methoxychior which was analyzed from 25 to 4000 picograms, respectively.
El/on-column analysis was employed for these analyses with ion trap AGC
(Automatic Gain Control) values at 10000, 18000, and 26100, respectively.
Table 2 lists the average minimum detection limit calculated from the three
lowest detectable concentrations for each compound. Figure 1 illustrates
the exceptional linearity for five characteristic analytes. Except for
those seven analytes discussed in the following paragraph the correlation

coefficient for all three 12-point calibrations was greater than 0.9. Three
compounds, 4,4’-DDD, 4,4’-DDE, and Methoxychior (*), were detected in all
three low level calibration concentrations suggesting that a lower detection
limit is possible.
(picograms on-column)
gamma-BHC (Lindane)...
Heptachior Epoxide.
Endosulfan I
Die ldrin
4,4’ -DDE
4,4’-DDE Corr Coeff: 1.000
Std Dev: 0.012
/ d/zzdz
Reptachior Corr Coeff: 0.999
Std Dev: 0.005
alpha-Chlordane Corr Coeff: 0.999
Std Dev: 0.016

25 *
delta-BUC Corr Coeff: 1.000
Std Dev: 0.007

Methoxychior Corr Coeff: 0.988
Std Dev: 0.135
Endosulfan II
4,4’ - DDD
Endosulfan Sulfate
4,4 - DDT
Endrin Ketone
Endrin Aldehyde
Tetrachloro-m-xylene (Surr)...
Decachiorobiphenyl (Surr)
‘ Is
u. s
5. 5
2. 5
Decachlorobtphen;lCoeff: 0.997
StdDev: 0.070

a le, la aIm

Seven target analytes were not reliably detected at the low
concentrations in any of the three calibrations. These included Endosulfan
I, Endosulfan II, Dieldrin, Aidrin, Endrin, Endrin aldehyde, and Endrin
ketone. These data suggest that although the CRQLs for ECD analysis cannot
be met for seven target compounds, the other thirteen compounds can be
quantified beyond the range of the 011(01.0 electron capture method.
The final set of pesticide experiments, to date, involved extraction
and analysis of two water performance evaluation samples (PES) routinely
used in the CLP. These PES have been analyzed in multi-laboratory
evaluations to establish advisory intervals for the 011(01.0 pesticide ECD
method. The results of these ITS4O analyses and the advisory limits for the
ECD method are presented in the following Table 3, using a single pesticide
16X standard and phenanthrene-dlO as the internal standard for the ITS4O
quantitation. With two exceptions, Endosulfan I and Endrin, the ITS4O
results are within the ECD intervals.
Units: ug/L
PES #1
Target Analytes ITS4O Result 99% Confidence Interval
alpha-BHC 0.18 0.11 - 0.25
beta-BHC 0.23 0.10 - 0.24
gamma-BHC 0.20 0.11 - 0.25
Heptachior 0.33 0.087 - 0.44
Aidrin 0.14 0.063 - 0.22
Heptachior Epoxide 0.29 0.14 - 0.54
PES #2
Target Analytes ITS4O Result 99% Confidence Interval
a lpha-BHC 0.19 0.11 - 0.25
beta-BHC 0.17 0.10 - 0.24
ganuna-BHC 0.19 0.11 - 0.25
Heptachlor 0.39 0.087 - 0.44
Aidrin 0.14 0.063 - 0.22
Heptachlor Epoxide 0.28 0.14 - 0.54
Endosulfan I 0.44 * 0.45 - 0.92
Endosulfan Sulfate 0.61 0.56 - 1.3
Endosulfan II 0.96 0.46 - 1.3
4 ,4’-DDT 1.2 0.46 - 1.4
Endrin nd * 0.38 - 1.0
* indicates that the ITS4O result is out of the CLP advisory action window
for ECD analysis.

The CLP seniivolatile quarterly blind QB292 was analyzed with the ITS4O
with 1:10 dilution of standards and extracts using El/splitless injection.
The results of this analysis, and the accompanying results of quadrupole
analysis and EPA performance windows, are presented in Table 4. The ITS4O
quantitation is based on a 5 ug/L standard. Owing to the nearly saturated
detection of many analytes in the 16 ug/mL standard, in future semivolatile
analyses we may dilute the standards and extracts 1:20, depending on trap
performance. The most significant difference in the ITS4O semivolatile
analyses is the characteristic ion trap DFTPP spectra which with default
scanning parameters usually results in a base peak at m/z 442. All of the
spectra of the target analytes with these tune conditions result in NIST
matchable spectra and the ITS4O quantitation is well within the windows
established by pooled quadrupole analyses.
ITS4O INCOS 50B QB Warning
Target Analytes Results Results Windows
Phenol 45 46 34 - 56
}lexachloroethane 28 26 22 - 43
2,4-Dimethylphenol 37 32 22 - 50
bis)2-Chloroethoxy)methane 17 17 15 - 23
l,2 ,4-Trichlorobenzene 27 24 22 - 37
Hexachlorocyclopentadiene 49 43 24 - 63
2,4,6-Trichlorophenol 43 28 27 - 43
Pyrene 93 126 80 - 140
Anthracene 3 (J) 4 (J) NU
The ITS4O was also calibrated and used for PCDD/PCDF analysis of a
severely contaminated soil sample using El/splitless techniques. This
analysis was conducted with full scan acquisition at the calibration
concentrations defined in DFLMO1.l (Sept 1991) for quadrupole MID analysis
(0.1 - 2.0 ng/uL for 2378-TCDD). The ITS4O full scan results and INCUS 50B
MID results are presented in the following Table 5. The advantages realized
in these ITS4O analyses include NIST library spectral identification for
native PCDD/PCDF which elute without l3C-labelled internal/recovery
standards, elimination of MID windows used in quadrupole analysis, and
identification of non-target compounds with appropriate sample clean-up.
In this case the extract was analyzed before and after carbon-column clean-
up resulting in unambiguous identification and estimated quantitation of 37
PCB isomers in the pre-carbon clean-up extract.
The 30m DB-5ms column was found to provide reasonable isomer
resolution. The INCOS 50B analysis was also performed with a 30 in column
(DB-5.625) of the same dimensions. With the program conditions employed in
the ITS4O analyses, (180-280 C at 15 C/mm), the total time for analysis
through OCDF was less than 32 minutes.

PCDD/PCDF ITS4O Full Scan Results INCOS 50 MID
(# of Isomers) (1*) (2*) (2*) RPD in 2*
2378-TCDD ND (0.3) (0.1) (67%)
Total TCDD (7) 31 40 34 15%
2378-TCDF 29 39 35 10%
Total TCDF (12) 120 140 130 7.1%
12378-PeCDD ND 1.0 (0.9) (10%)
Total PeCDD (8) 21 26 22 15%
12378-PeCDF 12 14 12 14%
23478-PeCDF 8.0 9.0 7.5 41%
Total PeCDF (13) 96 120 90 25%
123478-HxCDD 1.4 2.6 4.4 -69%
123678-HxCDD 3.5 4.2 2.9 31%
123789-HxCDD 2.5 2.8 3.1 -11%
Total HxCDD (6) 60 74 70 5.4%
l23478-HxCDF 27 110 140 -27%
l23678-HxCDF 14 37 37 0
234678-HxCDF 13 28 29 -3.6%
123789-HxCDF 7.1 14 19 -36%
Total HXCDF (13) 120 400 390 2.5%
l234678-HpCDD 11 9.3 7.6 18%
Total HpCDD (2) 24 17 17 0
1234678-HpCDF 73 54 55 -1.9%
1234789-HpCDF 15 18 20 -liZ
Total HpCDF (4) 120 110 110 0
OCDD 35 30 29 3.3%
OCDF 79 60 65 -8.3%
(1*) - GC/MS analysis performed after silica and alumina column clean-up,
before carbon column.
(2*) - CC/MS analysis performed after complete CU’ extract clean-up,
including carbon column.
Values in parentheses indicate quantitation below CRQL.
RPD — ((ITS4O 2* Conc) - (INCOS 50 2* Conc) / ITS4O 2* Conc) * 100

Polychiorinated Biphenyls , Estimated Concentrations, Units: ug/kg (ppb)
ITS4O Results Aroclor Chlorination Levels
% of Total Composition
Hutzinger, 1974
# Isomers Est Conc. XTotal 1248 1254 1260
Dichiorobiphenyls 0 nd 0 2 0.5 nd
Trichiorobiphenyls 6 47 4% 18 1 nd
Tetrachlorobiphenyls 9 220 17% 40 21 1
Pentachiorobiphenyls 10 680 53% 36 48 12
Hexachiorobiphenyls 6 260 21% 4 23 38
Heptachiorobiphenyls 4 50 4% nd 6 41
Octachiorobiphenyls 2 6 <1% nd nd 8
Nonachiorobiphenyls 0 nd 0 nd nd nd
Other Identification , Estimated Concentration, Units: ug/kg (ppb),
Anthracene or Phenanthrene 530
PCB and PAH estimated concentrations are all based on 50 ng of 1234-
TCDD-13C used as the internal standard. Unlike the CLP TIC approach,
characteristic ions were used for quantitation (e.g. pentachiorobiphenyls
iu/z 326) with assumed RRFs = 1.0. PCB and anthracene identifications are
based on NIST library fit values all >940. The chlorination levels
described for the Aroclors were taken from Hutzinger, et.al.,, 1974,
confirmed in our laboratory by analysis of Aroclor standards using the INCOS
50 instrument. Based on these relative chlorination amounts the PCBs found
in this soil sample are most likely Aroclor 1254.
The PCDD/PCDF RRFs obtained from the ITS4O calibration were
significantly higher than the RRFs from the quadrupole analysis, as
illustrated in Figure 2. These differences are a result of the high mass
sensitivity of the ITS4O as seen in the semivolatile DFTPP spectra under
default scanning parameters. For PCDD/PCDF analysis this high mass
sensitivity is a significant advantage over quadrupole analysis.

U) 2.1
O 2
I .-
0 1.6
U) 1.5
— 1.4
304—322 442—460
The ITS4O ion trap has proven to be a reliable and versatile
instrument for pesticide, semivolatile, and PCDD/PCDF analyses, largely due
to the approximately 100-fold increase in sensitivity over quadrupole full-
scan capabilities. Concern over interferences, which was know to severely
degrade the performance of previous trap designs and is a significant issue
for the CLP and RCRA programs, appears largely eliminated by current axial
modulation designs. Continuing experiments with our instrument will
include new approaches to chemical ionization for pesticide analysis which
have gained favor in other regulatory environments.
This work was entirely supported by the CLP National Program Office
under Contract Number 681)90041. Mention of trade names or commercial
products does not constitute endorsement or recommendation for use.
Although the research described in this article has been supported by the
United States Environmental Protection Agency, it has not been subjected to
Agency review and therefore does not necessarily reflect the views of the
Agency and no official endorsement should be inferred. We wish to thank
Finnigan MAT for their continuing invaluable assistance with ITS4O
operation, and J&W Scientific for early use of the low-bleed DB-5ms column.
340—358 374—392 408—426
Aoss Rong.

LINDA C. DOHERTY . Applications Chemist, Scientific Instruments
Division, Hewlett-Packard Company, 1601 California Avenue, Palo Alto,
California 94304
As Environmental Science enters the 1990s, more demands are being placed
on existing GO/MS hardware to meet lower detection levels. In order for
environmental laboratories to stay competitive, they must apply new
advances in technology to routine analyses. Techniques that increase the
sensitivity of existing GO/MS systems are discussed. Also, specifications for
new generation GC/MS systems are explored.
Among the novel applications that improve the sensitivity of existing GO/MS
systems, the ones presented are new GO injection techniques, various MS
ionization techniques and alternative GO inlets. Many of these approaches
are relatively low cost solutions and can be implemented simply. An example
of a new injection technique is electronic pressure controlled large volume
injections. Electronic control gives the chemist the ability to program the
pressure of the injection port to do high flow injections (burst) followed by
constant, low analytical flows. Programming column flow reduces run time,
improves resolution and increases retention time reproducibility. GO/MS
semi-volatile data acquired with electronic pressure control are shown.
Even with the improvement of existing systems, the need for high sensitivity
may necessitate the purchase of a new GO/MS instrument. This has made it
critical for instrument companies to develop even more sensitive GO/MS
systems. Insights about the key elements of a GO/MS that affect its
sensitivity- the GO inlet, MS ion source and detector- are outlined.
Finally, the common problems associated with ultra-trace organic analysis
are reviewed. These are the problems that eventually will limit “how low we
can go” because they affect the quality of data that is obtained.
Recent developments with the United States Environmental Protection
Agency (USEPA) have led to the formation of the Environmental Monitoring
Management Council (EMMO) [ 1]. This council’s objectives are to simplify
and consolidate methods that exist in the various branches of the EPA. The
consolidation of methods will:

• Simplify quality assurance/quality control differences that exist
between methods for similar analytes. This will make the
analyst job easier, with less confusion.
• eliminate outdated, duplicate methods
• have a permanent structure in place to eliminate duplication of
efforts between various branches
By combining methods the three most prominent outcomes are:
• More Analytes
• Lower Detection Limits
• Single Method Independent of Sample Matrix
The result of these outcomes will greatly impact the instrument companies
and their customers. Instrument companies will need to improve system
(GC/MS) performance with greater dynamic range, faster data review
capabilities, and better quantitation software. Also, customers will be
impacted. Consolidated methods mean more analytes per method. According
to Hites and Budde [ 2], the technical difficulty of a method increases as the
number of analytes does. This means higher technical expertise will be
needed at a time when we are observing the opposite. More analytes per
method, also means longer analysis times for good chrornatographic
resolution, which will lead to a decrease in billable samples per shift.
Lower detection limits will require a reduction in contamination and this
may mean clean rooms and stringent standard operating procedures. All and
all these changes will initially increase expenses. However, the final result
should be monetarily favorable to laboratories because QA “re-runs”,
(samples that have to be analyzed twice because of QA error), should be
drastically reduced.
Although it may be sometime before the consolidation of methods is finalized,
it is still important for laboratories to stay competitive. This is especially
true with the ever increasing importance of sensitivity in analytical
measurements. Environmental engineering firms are asking for lower
detection limits now . There are two ways to improve GQ/MS system
performance to meet the needs of the engineers. One is to optimize the
existing laboratory GC/MS with new components. These improvements are
typically less than $40,000. The second way is to buy a new GC,’MS system.
This will definitely cost more than $40,000. This paper addresses the
optimization of an existing system.

Figure 1 shows the six areas that will be discussed.
Figure 1. Six areas where newer technology can be applied to
existing GC/MS systems.
Column Technology
Today capillary columns are used widely by the environmental sector. Semi-
volatiles are normally analyzed with a 0.25 mm ID, 30 meter 5%-phenyl 95%-
methyl silicone column. Column manufacturers have come to the aid of
chromatographers by introducing this stationary phase with lower bleed.
These columns have a significant impact on the chemical noise in a system by
reducing column bleed. The signal-to-noise (S/N) ratio is greatly improved,
even at high temperature (3200 C).
Volatiles have been conventionally separated with packed columns. With the
advent of specially designed phases for volatile analysis, 0.53 mm ID
(Megabore ) columns being used routinely. The bonded 0.53 mm ID
columns greatly improve sensitivity over packed columns. The benefit of the
shorter, 30 meter columns is run time, but cryogen (CO 2 or N 2 ) is required to
retain the light gases. Longer columns (75 m and 105 m) are available and
have the advantage of not needing cryogen. For GC/MS purposes, a 75 m
column is the best choice. This length affords the resolution necessary for
the 3-D GO/MS data and only increases the run time by -4 minutes. Without
cryogen, operating costs are significantly reduced.

Injection Technique
The injection technique will be addressed for semi-volatiles only. Since the
purge and trap is such an important part of sensitivity in volatile analysis,
and is not a part of the GC/MS system, it will be excluded.
The two most common techniques for semi-volatile analysis are splitless and
on-column injection. Two areas where splitless injections can be improved
are: electronic pressure control (EPC) of the splitless inlet and proper
splitless liner choice. On-column injections are the best way to completely
transfer sample to the column, but until recently automation and the need
for retention gaps have made this technique less than desirable. Now
automated injectors for 0.25 and 0.32 mm ID columns are commercially
available. Even dirty samples can be run routinely, if a pre-colunm is used.
Why are splitless injections so hard to optimize? Figure 2 shows eight
reasons why! All of these parameters will affect the sensitivity, thermal
discrimination and thermal degradation of a sample.
(Why aplitless Injections are so hard to optimize.)
Inlet Temperature Sample Volume
Uner Design Solvent
Spfltless Valve Time InjeCtion Speed
Column Flow Sample Volatility
Figure 2. Various splitless parameters that must be considered.
The proper inlet temperature is needed to volatilize high boiling point
compounds without thermally degrading other compounds. Normally, the
inlet temperature is a compromise between these two factors. Compounds
that can be used to set the highest usable inlet temperature are endrin and
benzo [ g,h,i]perylene. Endrin’s breakdown to endrin aldehyde increases with

increasing inlet temperature and benzo [ g,h,i]perylene response decreases
with decreasing inlet temperature.
Liner design is one of the most difficult choices simply because of the variety
of liners available. The features that are most important are the volume of
the liner, whether it is deactivated or not and whether or not to use glass
wool. For the highest sensitivity possible, a 4 mm single tapered, deactivated
liner with no glass wool is recommended. For large volume injections >2 tl,
glass wool is necessary. For dirty samples, glass wool helps to keep the non-
volatiles from getting to the column, but too much glass wool can greatly
decrease sensitivity and increase adsorption of polar compounds.
Splitless valve time is critical. The time has to be long enough to assure that
all of the injected sample reaches the column. A textbook splitless injection
has the liner volume swept at least two times. A 4 mm liner has an
approximate volume of 1 ml. With a GC/MS flowrate of 1 mI/mm, a two
minute splitless injection would be necessary. This long splitless time is not
common because the initial linear velocity of the carrier gas is usually much
high than optimum. This is due to the design of the conventional splitless
inlets that are pressure regulated (constant pressure, regardless of oven
temperature) and not flow regulated (changing pressure with oven
temperature). So, a higher than optimal flow is set initially so that the flow
does not go to zero at high temperature. In contrast, electronic pressure
control (EPC) of the splitless inlet allows for high pressure initially, followed
by more typical GC/MS flowrates that are held constant as the oven
temperature increases because the pressure is programmed.
A high initial column head pressure during the injection is also favorable for
increased sample volumes. As the injected volume is increased, the required
expansion volume for the solvent greatly increases. With higher pressures
the volume is reduced (P 1 V 1 = P 2 V 2 ) and the entire injected volume moves to
the column. The higher pressure also decreases the likelihood of highly
volatile compounds from escaping out the top of the injection port through
the septum purge vent. Solvent choice affects expansion volume, however,
methylene chloride is normally used. Finally, it has been found that fast
injections tend to give the best results, universally.
The use of electronic pressure control to increase the injection volume size to
5 p1 was explored. Figure 3 shows the parameters used Figure 4 is a
graphical representation of those parameters. 5 .il injections of a 100 pg/pl
525 standard mix (AccuStandard) with 5 ngfpl of internal standards were
run. Notice, in Figure 3, that the initial oven temperature was 90°C.
Normally, to take advantage of the solvent effect, an initial oven temperature
of 10-20°C below the solvent boiling point is used. The solvent in this case
was acetone (BP 56°C). Thus, a typical initial oven temperature would be 45

30 PSI for 1.5 minutes
98 PSI/mm to 5.4 PSI then constant flow at 30 an/sec
PURGE VALVE ON at t5 minutes
90C for 2 minutes
35C/min to 13000
l2CJmln to 180C
9C/mln to 320C 4.35 minute hold
scan 45-450 amu. A/D = 2
start ac jistion at 4 mmutes
EM 500 ebove DFTPP tune
Figure 3.
The CC/MS parameters used in the large volume injection
experiments. An HP 5971A with an HP 5980 Series II
CC was used.
I —
5.4 PSI
5 10 15 20 25 30
1000 C
Figure 4. The pressure and temperature profiles for the EPC

Figure 5 shows what a normal constant pressure (flow decreases with oven
temperature) 5 p1 injection would look like at 45°C. Only 70% of the total
sample is transferred, and the chromatographic peak shape is unacceptable.
By raising the initial oven temperature to 90°C, the peak shape improves
remarkably, but now only 8O% of the sample is transferred. Finally, the
optimal temperature of 80°C gave a sample transfer of 9O%. These
percentages were compared to 100% transfer of sample with a 30 psi initial
injection port pressure and constant flow (EPC). Figure 6 shows the
optimum conditions for both constant flow and constant pressure. Note that
electronic pressure programming (EPP) decreased the run time and increased
the resolution. This is especially apparent in the late eluting peaks.
These data show that optimizing splitless injections can improve GO/MS
sensitivity with or without EPC, but EPC gives the best results. These data
confirm that larger volume injections are possible with optimization. So, at
the GO end, the proper choices of column, liner and pressure control can
greatly improve the amount of sample that is transferred to the mass
Once the sample is transferred to the mass spectrometer how can the MS be
optimized? There are three areas:
• Detectors
• Jet Separators
• Alternative Ionization Techniques and SIM
Detector manufacturers are improving the continuos dynode detectors to
decrease noise (shot noise, dark current, neutral noise) and improve the
dynamic and liner range. Many of these detectors fit into existing mass
spectrometers. High energy dynode (HED) detectors are also available.
These detectors improve sensitivity as mass-to-charge ratio (ni/z) of the ion
increases. The rule of thumb is: divide the m/z valve by 100 and that is the
improvement you will see. For example, if the ion has an mfz equal to 500,
the HED would improve the sensitivity by a factor of 5. The HED works by
increasing the energy of the ion as it reaches the detector. The energy is
directly proportional to its mass. The heavier the ion, the slower to is. By
increasing its velocity before it reaches the detector, it has a greater chance of
producing secondary electrons in the multiplier and being detected.
Jet Separators
Jet separators have been around since glass columns were interfaced to mass
spectrometers. The old glass jets were designed for packed column flows (30

54 1,eU 525 iuhc w/ 5si*4 JSlt)i V — constwt rni ve 45t 70%
54 )‘ J 5$ i x wi 5 I,ØU I$TT) 11 ata,.t .ssIt. 3 C
_kiL J _
Si ? t4 525 wJx qv 5 TDt fi c t i t p,. ,e, ?Ot
Time (mm.)
- ---I
Figure 5.
Total ion chromatograms of a 525 standard mix showing
the importance of initial oven temperature.
7.00 .406
t O 0 .406
0 . 00.+00
54 525 vth wI 5i U ISTO . 11p o tatw4 pma .r.. 70C
54 1i J
525W. w
I 5, Ai ISTO.
30 p bá ai
bust. 5 C 100%
i.ilji .ii.
Time (mini
Figure 6.
Two total ion chromatograms comparing constant
pressure and constant flow with EPC.
1 40?
9.00 . 406
t O O.+0S

mi/mm). As the 0.53 mm ID columns became more widely used modifications
were made to the glass jet (e.g., make-up gas) to provide 30 mI/mm through
the jet. This actually diluted the sample stream! Now jets have been
developed for lower flows. These new jet designs increase sensitivity over the
conventional glass jets.
Alternative Ionization Techniques and SIM
The third way to improve sensitivity is to change from electron impact (El) to
alternate ionization modes. Figure 7 outlines three very common chemical
ionization techniques. These three techniques have been especially useful for
increasing pesticide sensitivity. Positive chemical ionization (PCI) gives
primarily a protonated molecular ion. This combines all of the ion current
into a single mass, increasing signal. Drawbacks to PCI are that all
compounds have varying proton affinity and the sensitivity is class
dependent. For example chlorotriazines (simazine, atrazine) show no
improvement over El [ 3].
Electron capture negative ionization (ECNI) is very specific and sensitive. It
only works for electrophilic compounds and is best used with selected ion
monitoring (SIM) because all the useful ion current is found in the molecular
ion. A drawback of EON! is that many compounds are not electrophilic.
Also, a GO electron capture detector response for a certain compound may not
give an equivalent response by ECNI. By far, thebiggest problem with ECNI
is instrument variation. Differences in mass spectrometer design (different
manufacturer, different design) greatly influence spectra produced [ 4]. This
makes direct comparisons of results, lab to lab, impossible.
Another negative chemical ionization technique is negative ion chemical
ionization (NICI). This example is chloride attachment. Once again, it may
increase sensitivity but the variation of chloride adduct formed is compound
dependent [ 3].
In all cases, no standard libraries exist and methods developed would have to
be based on a certain compound class. This is contradicts the method
consolidation movement. The importance of scanning does not hold here
either. Since chemical ionization is a “soft ionization” technique, it produces
mostly molecular weight information. SIM with CI gives the highest
sensitivity. The movement toward these kinds of analyses will come with
lower detection requirements. These techniques will also require more
highly skilled chemists.
Although there are ways to improve the performance of existing GO/MS
systems, it is important to investigate new mass spectrometers. New mass
spectrometers yield improved system performance. The mass spectrometer

reduces fragmentation thus Increases sensitivity
+ + S
CH 4 +CH 4 -CR 5 + C l 3
+ +
CR 5 +M • M+R +CU 4
very sensitIve, hIghly specific
e+ M M
changes selectIvity, may Increase sensitivity
C1+M ir(M+CD
Figure 7. Alternative chemical ionization techniques that improve
sensitivity, especially for pesticides.
Figure 8. Limitations in obtaining the highest sensitivity possible.

may have a newly designed ion source that increases ionization efficiency.
The detector may have lower noise, greater collection efficiency and better
linear range. The data system may be faster, with better application
software and more advanced GC control. A different type of mass
spectrometer may be desired. An ion trap, magnetic sector (for dioxins) or
triple quadrupole are common examples. Also, a new mass spectrometer may
be needed if alternate ionization modes are of interest.
With all the possibilities for improving system performance and GC/MS
sensitivity, there are ultimate limitations. Figure 8 lists some. System
background (air leaks, pump oil) and column, trap bleed are difficult to
eliminate and are the background chemical noise. Better vacuum systems,
cleaner pump oils and lower bleed phases are solutions, but there is a limit to
how clean a system can be. Data system integration of low level, noisy data
needs to improve to speed up sample throughput. Manual integration of low
level samples is tedious. Sample adsorption that may lead to sample loss
(especially at low levels) and memory effects exist. Care in sample pre-
screening is one way to minimize memory effects. For low level volatiles,
laboratory air is a major concern and the volatiles should be kept separate
from the semi-volatiles. Phthalates are everywhere [ 5] and extra sample
preparation steps are necessary if phthalates are target analytes. Finally, as
solvent waste consumption is minimized, alternative sample preparation
with solid phase extraction cartridges has been encouraged. Unfortunately,
SPE cartridge quality has been poor. These cartridges tend to increase
contamination. Another alternative is the EMPORE disk from Varian
Associates. These Teflon disks with the solid phase bonded greatly reduce
Many solutions are available now to improve the performance of current
instrumentation. Investing time into learning how to use on-column
injections, instead of splitless, is becoming more realistic with the
introduction of automated injectors for 0.25 and 0.32 mmID columns. Larger
splitless injections are also possible with a small amount of method
development. It is also important to keep up with current column technology
to decrease the chemical noise introduced by high bleed columns.
If a new mass spectrometer is on the horizon, consider the total system
performance (GC, MS and data system). Also, since good automated
laboratory practices (GALPs) are becoming important, choose new systems
that help provide GALP compliance.
All instruments require people to run them. Remember, there is nothing
more important than proper training. Without it, all of the solutions will be
difficult or impossible to implement. Chemists must be trained so they can

comfortably compete. Better training leads to better quality work. Quality is
definitely the key to any lab s success in the 1990s.
[ 1] Friedman, David: Environmental Lab, 4, (1992), 20.
[ 2] Hites, Ronald and Budde, William: Environmental Science and
Technology, 25, (1991), 998.
[ 3] Barcelo, Damia: Trends in Analytic il Chemistry, 10, (1991), 323.
[ 4] Stemmier, Elizabeth et al.: Analytical Chemistry, 60, (1988), 781.
[ 5] Lopez-Avila, Viorica: Journal of the Association of Official Analytical
Chemists, 73, (1990), 709.

Daniel R. Doerge , Research Chemist, Division of Biochemical Toxicology, National
Center for Toxicological Research, Division of Biochemical Toxicology, Jefferson,
Arkansas 72079.
ABSTRACT . The introduction of particle beam (PB) interfaces has held great promise
for LCIMS analysis because typical El spectra can be obtained from thermally labile,
nonvolatile compounds. This paper will detail the use of PB-LC/MS for the analysis of
selected environmental toxicants.
Analysis of polycyclic aromatic hydrocarbons (PAH) and their metabolites in water and
sediments from the Alaska oil spill was performed by PB-LCIMS using a C18 silica
separator column. Spectra of PAH were essentially identical to those in the NIST library.
Detection limits (LOD) were highly dependent on PAH volatility: sub-nanogram LOD
were obtained for 5-ring PAH but the more volatile 3- and 4-ring PAH had higher detec-
tion limits due to losses in the particle beam momentum separator. The El fragmentation
patterns obtained from PB LCIMS were sufficient for structure elucidation of oxygenated
metabolites of PAH which often lack the volatility and thermal stability required for
analysis by (]CIMS. These studies suggest that PB LCIMS is complementary to GC/MS
for the analysis of PAH and metabolites.
Detection of pesticides in groundwater is a national priority reflected in the National
Pesticide Survey (NPS). The polar nature of these compounds makes them potential
groundwater contaminants and also more amenable to LC separation. A shortcoming of
most LC methods is the lack of an MS confirmation step. The use of PB LC/MS to
confirm and quantitate NPS analytes showed that Ca. one half of the compounds could be
sensitively detected (LOD Ca. 0.4-20 ng). This sensitivity is sufficient, in many cases,
for the confirmation of ppb levels of pesticides in ground water using conventional
cleanup and concentration procedures.
Ethylenethiourea (ETU) is a goitrogenic and carcinogenic metabolite of EBDC fungi-
cides which are used on ca. one third of all fruits and vegetables in the United States.
The low volatility of ETU has hampered GC analysis and LC methods have employed
electrochemical (EC) and UV detection. Analysis of ETU by PB LC/MS in several crops
commonly treated with EBDC’s gave LOD’s (5 ppb, 1.25 ng) that were comparable to
those obtained using LC/EC with the added specificity of MS cop irmation. The primary
quantitation method was external standard calibration but when C-ETh was used as an
internal standard, no evidence for coelution enhancement was observed.
The use of PB LCIMS as a tool to perform quantitative isotope dilution MS was investi-
gated. Although coelution of single-labelled internal standards with varying amounts of
native analytes showed large coleution enhancement, no enhancement was seen when
multiply labelled (3) internal standards were used. This suggested that chemical effects
(e.g., complex formation) did not affect particle formation and transmission efficiency.
Spectral overlap and detector nonlinearity were determined to be the cause of the ob-
served coelution enhancements. Isotope dilution PB LCIMS was used to quantitate the
concentration of caffeine in coffee and atrazine in spiked groundwater. These results
suggest that PB LC/MS has no inherent limitations for use in isotope dilution methods as
they have been previously performed by GC/MS.

Particle beam (PB)-LC/MS has proven useful for the analysis of many thermally labile or
involatile molecules, including environmental toxicants (1), primarily because of the
typical El spectra obtained. This facilitates comparisons with reference library compen-
dia (e.g., NIST) even though these spectra have been generated by GC or direct insertion
probe interfaces. Qualitative target compound identification is possible because of the
structural information revealed through fragmentation. In addition, the presence of
multiple ions in the mass spectra facilitate quantitative applications. This paper describes
the use of PB-LCIMS for both qualitative compound identification and quantification
using potential toxicants in water, soil and foods.
The VG Trio 2A mass spectrometer used in these studies was equipped with the LINC
PB interface and a Perkin Elmer series 10 isocratic LC pump or a Hewlett Packard 5890
GC. A source temperature of 200°C was used for all studies except PAH analysis where
250°C was used. The gas flow through the desolvation chamber was maintained at 30°C
and a Hildebrand double grid nebulizer (40 psi He head pressure). Typical operating
pressures are: 5, 0.8, 3x10 5 mbar at the first stage momentum separator, the second
stage momentum separator and the ion source housing, respectively. Positive ion spectra
were obtained using El conditions (70 eV) and full scan conditions (50-650 m/z in 1 sec).
The mass spectrometer was tuned and calibrated daily with PFTBA.
The use of PB-LCIMS for the analysis of PAH was investigated using a number of 3-6
ring compounds. These compounds displayed both linear and concave-upward calibra-
tion piots often seen for many analytes when using PB-LCIMS (2). This phenomenon is
likely related to the linearity of detector response as discussed below. Mass spectral
response sensitivity was dependent on source block temperature with a maximum in
sensitivity seen at 250°C. This higher source temperature requirement probably re-
flects the need for compound volatilization when the particle beam impinges upon the
heated block (3). The detection limits (LOD, S/N = 3/1) for a series of PAH were deter-
mined for five 3-5 ring PAH using flow injection analysis (FIA). LOD decreased with
ring number and correlated inversely with PAH volatility as measured by boiling point.
Figure 1 shows that subnanogram detection limits were obtained for the 5-ring PAH’s
but that LOD increased rapidly with increasing volatility to 25 ng for anthracene (3-
rings). LOD for more volatile PAH were in the jig range. This relationship probably
reflects the losses of volatile compounds in transmission through the momentum separa-
tors in the PB interface. These data suggest that PB/LC/MS can be an important adjunct

to GCIMS analysis of PAH, especially the higher molecular weight congeners. The low
volatility of high molecular weight compounds that could complicate GC analysis can
enhance the analytical sensitivity of PB-LC/MS methods.
PB-LC/MS was also used in the qualitative identification of oxygenated metabolites of
PAH based on fragmentation-derived structural information. Figure 2 shows the mass
spectrum of dibenzanthracene-7,14-quinone. Fragmentation was characterized by se-
quential loss of 28 daltons (-CO) from the molecular ion (308 m/z). This pattern was
also observed for quinone metabolites of benzanthracene and benzo [ a]pyrene. The
fragmentation of 7-hydroxyB [ aIP showed a prominent [ M-29] peak (-CHO). In addi-
tion, the relative intensities for the various 28 mass unit losses were distinct for regioi-
somers of benzanthracene. These data suggest that PB-LCIMS could be valuable in the
identification of unknown PAH metabolites. An important aspect of these findings is
that diagnostic mass spectra are obtained from underivatized samples that are often not
amenable to analysis by GC/MS (4).
PB-LCIMS was used to identify and quantitate PAH and potential oxygenated metabo-
lites in water and soil samples obtained in Prince William Sound, Alaska following the
Exxon Valdez oil spill. The combination of on-line UV (255 nm) and MS detection was
used to correlate UV-absorbing peaks with mass spectra of PAH. The spectra of PAH
consisted of a prominent molecular ion and were matched with NIST library spectra.
Table I shows the values obtained from oiled and non-oiled regions. More PAH were
found in samples from the oiled region (both soil and water) but the water sample from a
non-oiled region also showed PAH. No identifiable metabolites from these PAH were
found in these samples.
It is the polar nature of some pesticides that makes them potential groundwater contami-
nants. This property also can make LC separations more practical that GC. However, a
limitation on the use of multiresidue LC methods is the lack of LC/MS confirmation
procedures (5).
The use of PB-LCIMS as a qualitative confirmation tool and as a sensitive and selective
detection system was investigated using Ca. 100 analytes from the U.S. EPA National
Pesticide Survey (NPS) of pesticides in groundwater (6). The fragmentation produced
under El conditions was sufficient for matching with NIST library spectra in many cases.
However, in some cases the spectra differed from the library entries but had interpretable
differences. Some of the compounds were not included in the library. In all cases, user
defined libraries of target compounds could be compiled for use in identification of
unknowns. The existence of intense and characteristic ions in the spectra facilitated the

use of PB-LCIMS as a sensitive and selective detection method for analysis of pesticides
in groundwater. This technique was able to detect Ca. one half of the NPS analytes at
0.5-25 ng per injection (7). Pesticide classes giving the best response were the triazines,
phenylureas, carbamates/carbamoyloximes and organophosphates. Over the range of
volatilities encountered in selected NPS analytes, no association between detection limit
and volatility was observed. The techniques described were used to confirm the presence
of diuron and airazine in Hawaiian groundwater samples at ca. 1 ppb.
Environmental decomposition of ethylene-bis-dithiocarbamates fungicides (EBDC ‘ s)
leads to the presence of trace levels of ethylenethiourea (ETh) in many food crops (8).
ETU causes thyroid enlargement (goiter) and thyroid cancer when given to rodents in
large doses (8). The low volatility has made LC separations preferable and the required
sensitivity has made electrochemical detection most popular (9). While these methods
possess the required sensitivity, they cannot provide the selectivity and specificity of MS
LC was used to separate ETU from crop matrices when fortified into several fruit crops
at 10-20 ppb (2.5-5 ng injected) and the LOD was 5 ppb (1.25 ng) (10). The MS re-
sponse was linear from 0-10 ng ETU and the method of standard addition was used to
quantitate ETU residues in commercial papaya samples. Because of the interday and
intraday fluctuations in MS responses, the use of 3 C-ETU as an internal standard (IS)
was investigated. Under these conditions, no enhancement of IS signal (10 ng) upon
coelution with native analyte (10 ng) was observed. This was in contrast to previous re-
ports where such effects were observed (1,1 1). These workers concluded that enhance-
ment of transmission of IS occurred in the presence of increasing amounts of native
analyte because of complex formation leading to larger particle sizes.
The origin of the coelution effect was investigated with pairs of singly labeled IS/native
compounds (12). As previously described, significant enhancement of the IS signal was
observed upon coelution with increasing amounts of native analyte (see Figure 3). These
experiments were conducted as an isotope dilution experiment i.e., addition of a constant
amount of IS to varying amounts of native analyte. In these experiments no “self-C!”
effects were observed over the concentration range used. Other experimental parameters
(e.g., temperature of desolvatjon chamber, mobile phase composition, nebulizer type)
affected the magnitude of the coelution enhancement. However, when the data were
clotted in the form of an isotope dilution experiment (i.e. 12 C/ 13 C response ratio vs.
1 2 l 3 C concentration ratio), identical plots were obtained. This sugested that an exper-
imental artifact equally affecting native and IS signals was involved. This was tested
using GC/MS since no coelution enhancement has been reported for a capillary interface.

Figure 4 shows that for identical concentration ratios, large coelution enhancement
occurs with the PB Interface but none are observed for the GC.. However, the isotope
dilution plots are identical. This confirmed that transmission effects through an interface
were not involved. Moreover, these results suggest that there are no limitations on the
use of PB-LCIMS for isotope dilutions as they have been previously performed by
These hypotheses were further tested using 13 C 1 - and 13 C 3 -labeled IS compounds to
determine the effect of spectral overlap on the coelution phenomenon. Figure 5 shows
that under conditions where large enhancement of the 13 C 1 signal occurs, no effect on
13 C 3 signal occurred. This suggested that spectral overlap of native and IS signals was
required to observe the coelution effect and that detector nonlinearity was the cause.
This was confirmed by repeating the GCIMS experiment with a high split ratio in order
to reduce total signal output to the levels encountered in the PB experiment. Under these
conditions, large coelution effects were observed. These results show that detector
nonlinearity and spectral overlap are the cause of coelution enhancement under the
conditions of these experiments.
Isotope dilution MS analysis of caffeine (IS = 3 C 1 -IS, Table II) in coffee and atrazine
in fortified groundwater (d 5 -IS, Figure 6) was performed. The precision and accuracy of
these studies indicate that PB-LCIMS can be an important procedure for the quantitative
analysis of thermally labile or nonvolatile compounds.
1. Behymer, T.D., Bellar, T.A. and Budde, W.L. Anal. Chem. 62, 1686-1690 (1990).
2. Kim, LS., Sassinos, F.!., Stephens, R.D., Wang, J. and Brown, M.A. Anal. Chem. 63,
819-823 (1991).
3. Jones, G.G., Pauls, R.E. and Browner, R.C. ibid 63, 460-463 (1991).
4. Jacob, J., Karcher, W., Grimmer, 0., Schmoldt, A. and Hamann, M. in Polynuclear
Aromatic Hydrocarbons (M. Cooke and A.J. Dennis, Eds.), Battelle Press (1986), pp.
5. Miles, C.J. J. Chrom. 592, 283-296 (1992).
6. U.S. Environmental Protection Agency, in National Survey of Pesticides in Drink-
lug Water Wells, U.S. Government Printing Office, Washington, D.C. (1990) EPA-
7. Miles, C.J., Doerge, D.R. and Bajic, S. Arch. Env. Contam. Toxicol. 22, 247-251

8. U.S. Environmental Protection Agency, Federal Register 52, 27172-27177 (1987).
9. Doerge, D.R. and Yee, A.B.K. J. Chrom. 586, 158-160 (1991).
10. Doerge, D.R. and Miles, C.J. Anal. Chem. 63, 1999-2001 (1991).
11. Bellar, T.A., Behymer, T.D. and Budde, W.L. J. Am. Soc. Mass Spectrom. 1, 92-98
12. Doerge, D.R., Burger, M.W. and Bajic, S. Anal. Chem. (in press, 1992).

Table I. Quantitation of PAH in Soil and Water Samples from Alaska Oil Spill Sites.
BAy.Sa, F2b B [ e]P 20 .tg/m1 (0.5 ng)
Chrysene 2 j .tg/ml (5 ng)
Triphenylene 22 g/ml (5 ng)
BAY-W, F2 B [ e]P 0.3 .tg/m1 (0.5 ng)
Chrysene 3 pjg/mI (5 ng)
LIPS-W, F3 B [ e]P 0.1 .Lg/m1 (0.5 ng)
B [ a]P 0.2 j tg/m1 (0.5 ng)
aSamples designated BAY and LIPS were obtained in a heavily oiled waterway, and an
unaffected waterway, respectively (S = soil and W = water samples).
bSampies were processed by sequential extraction using hexane-benzene (F2) and
methylene chloride-ethyl acetate-methanol (F3). Quantitation was obtained from one
point calibrations.

Table II. Analysis of Caffeine Content in Coffee.
Sample Molecular Dataa (ng) Fragment j Data (ng)
Decaffeinated +40 ng 40.8 37.9
duplicate injection 41.0 38.3
Decaffeinated +40 ng 40.9 36.7
duplicate injection 40.4 37.5
AVERAGE 40.8 ± 0.3 37.6 ± 0.7
LC/UV 40.4±0.1
Regular 44.7 44.2
duplicate injection 45.1 46.8
Regular 48.4 48.4
duplicate injection 46.2 43.4
AVERAGE 46.1±1.7 45.7±2.3
LQUV 42.9 ± 0.6
aD were collected with the PB interface as described in Figure 4 using native and 3-
‘ 3 C 1 -caffeine. Peak areas for the molecular ions (194 and 195 m/z, respectively) or the
major fragment ion (109, 110 m/z, respectively).

Figure 1. PAH Detection Limits Depend on Volatility.
Detection limits were determined using FIA (75% acetonitrile/water) for anthracene,
pyrene, chrysene, benzo [ e]pyrene, benzo [ a]pyrene and dibenzanthracene in order of
decreasing LOD. PAH boiling points were obtained from Lange’s Handbook of Chemis-
Figure 2. El Mass Spectrum of Dibenzanthracene-7,14-quinone.
The spectrum was obtained using FIA (75% acetonitrile/water, 0.5 mI/mm) of a 100 ng
Figure 3. Coelution Enhancement for IS Signal by Native ETh.
The signal enhancement for 10 ng of 2- 13 C-ETU (peak area of 103 m/z) was determined
in the presence of varying amounts of coeluting native ETU using FIA (50%
acetonitrile/water). Enhancement factor = 1 - 3 C signal with coelution ÷ 13 C signal with-
out coelution.
Figure 4. Isotope Response Ratio Plots and Enhancement Factors for PB-LC/MS and
GCIMS Analysis of Caffeine.
Ratios and enhancement factors were determined for coeluting 3- caffeine in the
presence of varying amounts of native caffeine with sample introduction via PB or GC
Figure 5. Coelution Enhancement for 3- 13 C 1 - and 1,3,7- 13 C 3 -Caffeine.
The signal enhancement was determined as described in Figure 3 using 25 ng IS using
HA (20% acetonitrile/water).
Figure 6. Isotope Dilution Calibration Plot for Analysis of Atrazine in Groundwater.
Samples were chromatographed using a Perkin Elmer 0.46 x 3 cm C18 column (3 micron
particle size) with a flow rate of 0.5 mI/mm and mobile phase of 75% acetonitrile/water.
Response ratios were calculated using either the molecular ion or the major fragment ion
for native (215, 205 m/z) and d 5 (220, 205 m/z) atrazine and plotted vs. the amount of
native atrazine present with constant addition of 25 ng d 5 -atrazine (correlation coeffi-
cient = 0.995). Data points shown represent averages with standard deviations shown as
error bars.

300 350 400 450 500 550
PAH bp (°c)

UH EN’) MASS SPE 12/24/9 1
LINC EI+ 15:40
DBA1 2D (@335) COMBIME:(15 to 32)—(((1 to 12)+(41 to 65))*1.000)
3fl8 181248
3 9
7591 13 3W?
/ 154 248 253 281
99 14 I 224 / /
I Ii •i ili I , I ‘I I . 2 r . . liii .. I L .
15 25

0 100 200 300 400
12 CETU (ng)

0 100 200
(ng/1 O L)
C i )
r o 0
300 400

(I )
0 25 50 75
H 5

F— 9 °
12 C—CAFFEINE (ng)
200 400 600 800

52 The Determination of Semi—Volatile Organic Compounds in
Analytical Extracts Using Split Injection Technique with an
Ion Trap GC/MS.
Robert Brittain, Norman Kirshen, and Elizabeth Almasi,
Varian Chromatography Systems, 2700 Mitchell Drive, Walnut
Creek, CA 94598, 510—939—2400 / FAX 510—945—2335
The determination of semivolatile organic compounds in
analytical extracts of wast. water or hazardous waste
following EPA Method 8270 requires the screening of a large
number of target compounds at low PPB to PPM levels. This
translates to extract concentrations of approximately 0.1 to
1000 ng/mL. Hot splitless in ection has traditionally been
required to deliver sufficient analyte to the mass spectrometer for full-scan
con fix2nation and rluantitation. But, this in ection c ni ua ca. . .aui
to problems such as polar analyte losses. i nprecision, and discrimination. It
also causes the more rapid contaminaticr of injector inserts
and capillary columns which hav, to be replaced regularly.
When using th. split injection technique with the more
sensitive ion-trap MS several of these problems can be
reduced while obtaining much improved chromatography.
This study employed a 25:1 split injection. The
chromatography was superior to that obtained with other
injection techniques. Th. multi-ramp temperatur. program
used throughout the study was selected with the DryLab GC (R)
software from LC Resources, Inc. The following QC
acceptanc. criteria were investigated using the EnviroPro
data reporting package: vimum %RSD for initial
calibration, minimum average RRPs, continuing calibration
checks, internal standard area reproducibility, and
retention time precision. These criteria met both 8270 and
CLP guidelines. Finally, this approach was applied to the
analysis of two samples, one an NIST Standard Reference
Material (Coal Tar Extract, 1597) and th. other a TCLP
extract showing many phenolic analytes in a high level
hydrocarbon matrix. The results obtained from the NIST
Standard compared very well, with the published results.

DanielR.Doerge, Research Chemist, National CenterforToxicological Research, Division of
Biochemical Toxicology, Jefferson, AR 72079 and Steve Bajic, Development Department, VG
BioTech, Altrincham, WA14 5Z, U.K.
ABSTRACT . The inherent chemical differences between classes of analytes has been a limita-
tion for environmental LC/MS analysis because no single LC interface/ionization technique is
sufficient. For this reason, a more universally applicable LC/MS interface/ionization technique
possessing high sensitivity and ease of use has been sought. The atmospheric pressure ioniza-
tion (API)/MS technique of electrospray (ES) has proven invaluable in the analysis of large bio-
macromolecules ( 1.5 kD) and in some cases, smaller molecules. The major limitation for ES is
the low LC flow rate amenable to efficient ionization (2-5 .tL/niin or 20-100 .tLJmin for pneu-
matically assisted ES). Although microbore LC and the newer techniques of nanoscale chroma-
tography and capillary zone electrophoresis can be successfully coupled to ES,the wide-spread
use of conventional bore LC columns in environmental labs makes it desirable to perform
APIJMS under higher flow rate conditions (ca. 1 mllmin).
Interfaces which effect API using a coronadischarge haverecently been introduced for analysis
of thermally labile and nonvolatile compounds of environmental interest. Since this technique
involves the gas phase ionization of analyte molecules by reagent gas ions, it has been termed
atmospheric pressure chemical ionization (APcI). This paperdescribes the use ofAPcI/MS for
analytes in the EPA National Pesticide Survey (NPS) of groundwater contaminants.
Seventeen analytes from the NPS were selected from five different pesticide classes:
1) Triazines 2) Phenylureas 3) Carbamates 4) Organophosphates 5) Miscellaneous. Positive ion
spectra were obtained for all compounds using nitrogen as the source gas except p-nitrophenol
which was observed as negative ions produced using dioxygen. Varying flow rates and mobile
phase composition were used to separate analytes using a typical LC column (Perkin
Elmer 0.46 x 3 cm, C18, 3p. deactivated silica) or in flow injection analysis. Spectra
typically consisted of the protonated molecular ion (M+H+) or deprotonated negative ion
(M-H) but application of a potential difference in the intermediate pressure region
produced voltage-dependent fragmentation of all carbamates and alachlor. Detection
limits (LOD) from full scans ranged from 0.8-10 ng and these LOD were reduced to pg
levels by employing selected ion monitoring. The MS response was highly linear (cc 
0.99) over the entire range of concentrations tested (LOD to 100 ng) for all compounds
tested. Mobile phase composition had minimal effects on MS response for any class of
analyte over the range tested (50-100% acetonitrile). The MS response for diazinon was
not affected by flow rate between 0.25 and 1.0 mL/min but decreased ca. 50% at 1.25
m1 4 ’min.
These results show that APcIJMS is a versatile, practical, and highly sensitive technique. The
sensitivity is not affected by chemical differences over a wide range of analyte types suggesting
that this is a more universal detection system that other LCIMS techniques (e.g., thermospray,
particle beam). The low detection limits, high degree of linearity and the ability to produce
diagnostic fragments by use of cone voltage or collision-induced dissociation in an MS/MS
experiment are important elements required for the identification, confirmation and quantitation
of analytes in groundwater and other environmental matrices.

The chemical differences between classes of pesticides has limited the universal applica-
tion of a single LCIMS technique for multiresidue analysis. While thermospray (TSP)
and particle beam (PB) interfaces have been successfully applied to pesticide residue
analysis, significant gaps in sensitivity have emerged. For these reasons, the develop-
ment of a more universally applicable yet highly sensitive LCIMS technique would be a
valuable addition to pesticide analytical labs for the development of multiresidue meth-
ods that could employ a single ionizationlmtroduction mode.
Atmospheric pressure ionization (API) MS has been widely applied to analysis of large
biomolecules (e.g., proteins, peptides, nucleic acids) with molecular weights of> 1.5
kDa using electrospray (ES) ®. However, classical ES and pneumatically assisted ES
require LC flow rates of 2-5 LLJmin or 20-100 j.LL/min, respectively. This requirement
has limited the application of API/MS to pesticide residue analysis because such LC
methods have employed conventional-bore columns (4.6 mm) and flow rates of ca. 1.0
mI.Imin. Recent development of interfaces that effect API using a corona discharge have
facilitated analysis of small molecules (<1 kDa) that are thermally labile or nonvolatile
3. This technique produces ionization of analyte molecules via atmospheric pressure
gas-phase ion-molecule reactions with solvent reagent ions produced from the corona
discharge in the ion source volume. For this reason, the technique is called atmospheric
pressure chemical ionization (APcI). This form of ionization is extremely “soft” in
nature and yields primarily protonated molecular ions (M+H)+ or deprotonated molecu-
lar ions (M-H) in the positive ion and negative ion modes, respectively. In addition,
structural information can be obtained using cone voltages to effect collision-induced
dissociation of the sample ions in the intermediate pressure region (ca. 1 mbar) between
the API source and the MS source housing. Moreover, this technique is amenable to
flow rates between 0.2-2 mLlmin without the need for flow splitting.
This paper describes the use of APcI/MS for the analysis of multiple classes of com-
pounds included in the U.S. Environmental Protection Agency (USEPA) National Pesti-
cide Survey (NPS) of pesticides in groundwater . A major limitation of current LC
methods is the lack of MS confirmation procedures, especially those capable of identify-
ing multiresidue contaminants. The results of this study show that APcIIMS has high
sensitivity for all pesticide classes tested and much smaller differences in sensitivity
between classes when compared with results from TSP and PB interfaces.
Mass Spectrometer Conditions . The mass spectrometer used in these studies was a Trio
2000 single quadrupole instrument equipped with a VG BioTech dual ESI/APcI source
(Fisons VG BioTech, Altrincham, U.K.). Mass spectra were collected in full scan (m/z
100-400 in 1 see) or selected ion monitoring (SIM, dwell = 0.08 sec. span = 0.2 amu).
The source temperature was maintained at 130°C and the cone voltage was 34V unless
specified. Ions were generated using dinitrogen (positive ions) or dioxygen (negative
ions) as the bath gas (see below). Sensitivity was checked periodically by injecting a
reference compound whose responses remained consistent.
Interface Description . Referring to the schematic diagram in Figure 1, the entire LC
mobile phase flow enters the probe inlet and is transferred to the nebulization/desolvation
region via a 100 1m i.d. fused quartz capillary. The liquid flow is then pneumatically
formed into an aerosol by the action of a high velocity concentric dinitrogen gas flow

between the fused silica and an 0.5 mm i.d. stainless steel capillary. The nebulizer gas
flow and an additional concentric sheath gas flow carry the aerosol through a 45 mm
heated desolvation tube with an internal diameter of 2 mm. Heat from a 75W desolva-
tion heater is transferred to the aerosol droplets via the hot nitrogen gas flow which par-
tially converts the mobile phase (and hence sample) into the gas phase. Prior to ioniza-
tion, the resulting flow from the probe is carried towards the counterelectrode by a dini-
trogen carrier, or bath gas, at a flow of typically 300 1/hr. In order to prevent condensa-
tion in the source volume and corona voltage flash-over when high agueous content
mobile phases are employed, the source is typically maintained at 140’-’C. A “cone of
ionization” is formed between an axially-located nickel-plated carbon steel corona pin
and the 2 mm exit aperture of the counterelectrode by the application of a potential dif-
ference of ca. 2kV (at a fixed pin/aperture distance of 2 mm). For positive ions, the
corona pin and counterelectrode are typically maintained at voltages of +2.5kV and
+0.5kV, respectively. Primary ions created in the corona discharge rapidly react with
abundant mobile phase molecule in the gas phase to produce terminal (reagent) ions.
Sample molecules introduced into the source react with the reagent ions and then exit the
counterelectrode. At the adjustable off-axis sampling cone, the gas containing the
sample and reagent ions expands into an intermediate pressure region (ca. 1 mbar) which
is pumped by two 1 8m’/hr rotary pumps. Prior to this expansion, the atmospheric pres-
sure ions can only drift under the influence of the applied electric field, i.e. no significant
translational energy can be gained above thermal energy. However after expansion into
the intermediate pressure region, where the mean-free-path of an ion increases, ions can
gain an excess of energy between collisions. In practice, this energy is obtained by
applying a potential to the sampling cone (typically 10- 100V) with the skimmer held at
ground potential. From a practical viewpoint, the application of a cone voltage serves to
decluster solvated molecular ions and furthermore provides a method of obtaining struc-
tural information, i.e. by a process analogous to collision-induced dissociation (CID).
Finally, the ions that pass through a skimmer aperture are focused into the quadrupole
analyzer via a 2-lens source optics arrangement.
Conditions . Samples were introduced into the interface using an LKB isocratic pump
for flow injection analysis (FIA) with 100% acetonitrile as mobile phase or through a
Perkin Elmer C18 silica column (0.46 x 3 cm, 3 .t particle size). An 0.45 .t stainless steel
filter was installed between the LC column and the interface.
Pesticide Analytical Standards . Pesticides standards were obtained from the USEPA
repository (Research Triangle Park, NC) or from the respective manufacturer and were of
at least 97% purity.
The pesticides listed in Table I were selected from 5 classes: triazines, phenylureas,
carbamates, organophosphates and some other miscellaneous compounds on the NPS list
of pesticides in groundwater. This selection permitted intra- and inter-class comparisons
of sensitivity, ionization, fragmentation, etc. The results shown in Table I indicate that
APcJIMS gives high sensitivity detection of all compounds with a minimum of differ-
ences in detection limits (LOD) between classes. Low nanogram LOD’s were observed
for all compounds in scan mode and these were reduced to picogram levels by using SIM
mode. The LOD’s determined in scan mode varied Ca. 10-fold while those in SIM mode
varied Ca. 100 fold. These low LOD’s coupled with the limited vanability across these
diverse chemical classes suggests that APcI/MS may be a more umversally applicable
MS detection system for use in confirmation and detection of pesticides in environmental

samples. The high degree of linearity of response is shown for selected analytes in
Figure 2.
The ions shown in Table I were either (M+H)+ or (M-H) in positive or negative ion
modes, respectively. These results are consistent with reactions of reagent ions consist-
ing primarily of protonated mobile phase ions and clusters, in positive ion mode and
presumably 02 and solvent clusters in negative ion model.
Cone Voltage-Dependent Fragmentation . Only the carbamate pesticides and alachlor
showed fragment ions in the presence of higher cone voltages. The application of suc-
cessively higher voltages resulted in the progressive diminution of (M+H)+ ion intensity
with increases in the intensities of diagnostic fragment ions (see Table II). For example,
aldicarb sulfoxide showed three ions: 223, 166 and 148 m/z. These ions correspond to
(M+H)+, protonated oxime and protonated nitrile fragments, respectively. Table II
shows the effect of cone voltages on the relative intensities of these ions. In principle,
the voltage could be adjusted to produce any desired degree of fragmentation desired. In
this way, the desired amount of fragmentation required for confirmation and quantitation
is available from a single quadrupole mass spectrometer. Alternatively, this information
could be obtained in an MS/MS experiment with analytes that are recalcitrant to frag-
mentation via cone potentials.
Effect f Mobile Phase . The effect of mobile phase composition and flow rate on analyte
sensitivity was examined for ranges typically encountered in pesticide analysis. The
effects were monitored using 4 different classes of pesticides to determine if compound-
specific changes in sensitivity occurred. Figure 3 shows that the responses from 4 differ-
ent classes of pesticides were not affected below ca. 0.75 mL/min. However, when flow
rate was further increased, responses decreased to about 50% response at 1.25 mL/min.
These flow rates are readily compatible with those used in the many commercially avail-
able conventional bore LC columns 4. Little effect of mobile phase composition on MS
response was noted between 50-100% acetonitrile (data not shown). Although not stud-
ied in detail, it was observed that increasing the water content does cause some reduction
in analyte responses, especially with 100% aqueous solvents. The invariance of response
for the range of compound classes tested suggests that APcI/MS may be amenable to the
use of gradient LC separations. However, it was observed that some losses in sensitivity
occurred upon changing mobile phase flow rate or composition. The MS responses
could be restored to original levels by adjustment of the off-axis skimmer. These obser-
vations suggest that this is an important experimental parameter for optimizing the
sampling of gaseous ions as they expand into the MS analyzer.
APcI/MS Analysis of Triazine Herbicides in Water . A multiresidue LC method for the
analysis of 4 triazine herbicides often found in groundwater was developed. Figure 4
shows the results from a conventional bore LC separation using full scan conditions on
50 ng samples of each herbicide. Detection limits estimated using SIM were 1-2 ng on-
column and the major noise component was due to LC pump pulsations.
Comparison of APcI LOD’s with TSP and PB Interfaces
Table III shows the LOD for APcI (this work) and values obtained from the literature
from studies that used TSP and PB interfaces for the analysis of pesticides. The full scan
sensitivity for these selected pesticides varied over 1 qrde f magnitude for APcI, 50-
fold for TSP and >300 for PB. Thus, although TSP ’br’PB interfaces provide adequate
sensitivity for some analytes, APcI provides the optimal combination of high sensitivity

with broad specificity. These results suggest that APcI/MS is more suited to multiresidue
pesticide analysis than the more established TSP and PB methodologies for analysis of
groundwater contamination.

1. Smith, R.B., Loo, J.A., Edmonds, C.G., Barinaga, C.J. and Udseth, H.R. Anal. Chem.
62, 882-899 (1990).
2. Gilbert, J.D., Hand, E.L., Yuan, A.S., Olah, T.V. and Covey, T.R. Biol. Mass Spec-
rrom. j, 63-68 (1992).
3. U.S. Environmental Protection Agency in National Survey of Pesticides in Drinking
Water Wells; Phase 1 Report: 1990; Offices of Water and Pesticide Programs. EPA-
4. Miles, C.J. J. Chrom. 592, 283-296 (1992).
5. Miles, C.J., Doerge, D.R. and Bajic, S. Arch. Env. Contain. Toxicol. 22, 247-25 1
6. Voyksner, R.D. and Haney, C.A. Anal. Chem. 57, 991-996 (1985).
7. Voyksner, R.D., Bursey, J.T. and Pellizari, E.D. ibid 56, 1507-15 14 (1984).

Figure 1. Schematic Diagram of the APeI Interface.
Figure 2. Calibration Plots for Selected Pesticides.
The MS response was determined from average peak areas (n=3) of varying amounts of
the analytes using PTA (acetonitrile).
Figure 3. Effect of Mobile Phase Composition on APcI Response for Diazinon.
Diazinon (50 ng) was introduced using FIA at the indicated mobile phase compositions.
The data shown are means (n=3) with standard deviations as error bar.
Figure 4. Multiresidue Triazine Herbicide LC Analysis Using APcI Detection.
The indicated herbicides (50 ng each) were analyzed under full scan conditions as de-
scribed in the Exeprimental Section using 60% acetonitrile (isocratic) as mobile phase at
0.5 mLjmin.
Table I. APcI LC/MS Analysis of Pesticides, Figures of Merit.
Pesticide Ion LOD Scan LOD SIM Linearity 1
Atrazine 216 2.5 0.2 0.998
Ameiryn 228 1 0.2 0.999
Cyanazine 241 1 0.5 0.999
Hexazinone 253 4 1 0.999
Fluometuron 233 0.8 0.1 0.994
Diuron 233 4 0.2 0.960
Neburon 275 2.5 0.2 0.995
Linuron 249 4 0.1 0.999
Propanil 218 8 0.5 0.992
Aldicarb Sulfoxide 2232 7 0.5 0.998
Carbofuran 2222 2.5 0.15 0.998
Carbaryl 2022 5 0.5 0.999
Diazinon 305 1 0.01 0.999
Fenamiphos 304 8 0.2 0.999
Alachior 2702 8 0.2 0.996
p-Nitrophenol 138 10 0.2 0.996
Correlation coefficient of the calibration curve from the detection limit to 100 ng.
Cone voltage-dependent fragmentation observed
(M-H) Ion

Table II. Cone Voltage-Dependent Fragmentation of Aldicarb Sulfoxide.
Cone Voltage Ion Intensity
24 Volt 223 m/z 100%
166 0
148 0
34 223 100
166 40
148 40
44 223 40
166 70
148 100

Table ifi. Comparison of APcI, TSP and PB Interfaces for the Analysis of NPS Pesticides.
Pesticide LOD-APcI LQp TSP ma LOD-PB
Atrazine 2.5 20 19
Amelryn 1.0 20 602 11
Cyanazine 1.0 -- 2.0
Hexazinone 4.0 20 602 0.4
Fluometuron 0.8 -- 1.4
Diuron 4.0 50-80 0.4
Neburon 2.5 -- 3.4
Linuron 4.0 4-8 4.4
Propanil 8.0 0.6
Aldicarb Sulfoxide 7.0 -- 4.0
Carbofuran 2.5 1-4 9.0
Carbaryl 5.0 3 53 6.0
Diazinon 1.0
Fenamiphos 8.0 -- 4.2
Alachior 8.0 10-20 68
p-Nitrophenol 10.0 12
‘Reference 5
2 Reference 6
3 Reference 7

opT ics
0 0
pU p , pU

F )
I 200
0 25 50 75

075% ACN/WATER DIAZINON, m/z 305
• 50% ACN/ WATER
0100: .1.
0 I I
0.50 0.75 1.00 1.25
FLOW RATE (mL/min)

MS Data
1 - Cyanazine
2 - Metribuzin
3 - Atrazine
4 - Ametryn
Fl (MS,ES+),21 5+216+228+241
Miss I 0.150
I - - I - - I I - - I - - I - I I I
1.00 1.50 2.00 2.50 3.00 3.50

Joseoh M. Lew Manager of Technology and Customer Support. Suprex Corporation, 125 WUhlam Pitt Way,
Pittsburgh, PA 15238. Telephone: 412-826-5200, FAX:412-826-5215
Superchtical ifuid extraction (SFE) has a broad range of applicability, especially with regards to environmental
problems. SFE has achieved a significant amount of attention due to the benefits of eliminating toxic, hquid
solvent usage, reduction in sample preparation time and an increase i the overall analytical reliability of
determinations. On-line SFEIGC-MS Is a powerful technique to accurate analyze and quallitate environmental
analytes. In addition, the off-line transfer of SFE effluents to coUection vials adds a considerable amount of
fIex ildy in clwacterizkq complex matrices suice a lull corrçumwii i auialytical tools can be used (i.e. GC, LC,
IR, NMR and UV). Moreover, the advantages of SFE can be further augmented by the development of
automation for greater sample throughput which can be especially impoitant for environmental applications.
This paper will discuss the use of on-line and ott-line SFEFGC44S methodologies for the determination
polynuclear aromatic hydrocarbons (PAH) In soil. Details of method development will be presented demonstrating
t EPA method 8270 was followed sxce for the replacsmer* of Soxhiet sample preparation wllh SFE. This
ssion will also focus on the experinier*ah veilllcatlon of oçXlmlzed SFE variables to achieve efficient and
tftatlve extractions of the target anal tes in the soil. M example is shown In Table 1 where an off-line
SF EJGC comparison was made between the extraction of PAils from soil at different pressures, indicatIr’ that
higher oressures were necessary for the coIT lete recovery of the PAHa, especially the four and live ‘nc
The use of various pre-extractlon strategies (I.e. matrix mançulatlon, modifier addition , adsorbent use) for the
enhancement of extraction efficiencies will also be outlined.
Table I: 0ff-Line SFE/GC-MS of PAH Contaminated Sal: Pressure Variation $FE 65°C, 40 minutes, flow 0.9
ri/mm (compressed) GC: methyl 25 m x 0.2 rrvn 1.0., 60°C (2.0 min) to 280°C (30 Mn) at 7 0 C(mln.
Acce*arcs C oq q ratJon . .e eisjPPMI
Rar s $0 atm 350 atm 400 atm
242-40.6 23 23 25
fr.e n hthv l r in 14.7-23.5 20 • 22
£a n h na 527737 566 601 614
414-570 445 471 458
Phenanthrene 1270-1968 1682 1978 1911
Ar*hracene 373.471 357 439 400
Fbiorarthens 1060-1500 1 8 1459 1511
Pyrene 744-1322 703 1153 1259
Berizo(a)Anthracer i e 214-290 74 235 284
Chrysene 271-323 74 251 314
Benzo(bjc)Fluorar*hene 130-174 <1.0 107 155
Benzo(a)Pyrene 80.1-114.3 <1.0 64 89

Mark L. Bruce . Director of Research and Development, and Marvin W. Stephens,
Technical Director, Wadsworth/ALERT Laboratories, Division Of Enseco, Inc., 4101
Shuffel Dr. N.W., North Canton, Ohio 44720
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 almost a 100 fold reduction in the amount of solvent used as
compared to the present Soxhiet 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 solid phase adsorbent trap. The trap is rinsed with 3 mL of tetrachioroethene
which has replaced Freon-i 13® in this application. Extraction conditions are 400-500
atmospheres, 25-30 minutes dynamic step, 8 mL/min supercritical C02 flow and an oven
temperature of 150°C. Dry and wet soil samples have been successfully extracted with this
method. Many dry samples can be extracted in less than 20 minutes.
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 Soxhiet,
sonication, or Soxtec® using an organic solvent followed by gas chromatographic or
infrared analysis.
Freon-i 13® 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 with the Montreal Protocol on Substances that Deplete
the Ozone Layer and the Clean Air Act Amendments of 1990 (CAA), CFCs will be phased
out by the year 2000. Recent White House decisions have moved many solvent phase-out
deadlines to 1995.
Extraction of hydrocarbons from soil samples was one of the first applications of analytical
SFE with carbon dioxide. Numerous researchers have worked with these analytes in this
matrix. Petroleum hydrocarbons can be extracted from soil, that has been proven. The
question now is, “Can the extraction be performed in the typical environmental lab with all
its constraints?” Routine environmental work requires rugged methods that are easy to use
properly. The cost (economic, environmental and health) must be low. Yet the sample
throughput must be high and the extraction time short enough to meet client needs. These
additional nonanalytical constraints significantly narrow the SFE choices. Most SFE
parameters are directly or indirectly affected by these constraints such as; sample size,
solvent, pressure, temperature, flow rate and extraction time.

Numerous solvents were examined for use in the JR analysis. Freon-i 13® can be replaced
in this application by several solvents: hexafluorotetrachiorobutane, tetrachioroethene and
FC-77. Each of these solvents has advantages and disadvantages.
Superciitical Fluid Extractor
Suprex, PrepMaster, AccuTrap
5 mL extraction vessel
Restrictor, prototype VaiiFlow restrictor (adjustable from 0.1 to 8 mL/min)
Infrared Spectrophotometer
Perkin-Elmer, 710
Buck Scientific Oil in Water Analyzer
10 mm, 3 mL quartz cell
Reagents and Standards
Freon-i 13®, EM Science
Fluorinert® FC-77, 3M
Isooctane, Mallinckrodt
Xylenes, Mallinckrodt
Hexadecane, EM Science
Kaolin, Baker Analyzed
CO 2 . SFC grade with 1500 PSIA Helium headspace with dip tube, Scott Specialty Gases
Tetracholorohexafluorobutane, Horiba or Halocarbon
Tetrachiomethene, Aldrich
The development for this method was conducted in two areas, 1) the search for a suitable
collection/analysis solvent, and 2) the optimization of supercritical fluid extraction
parameters. The approaches and representative results are discussed below.
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 about 70 possible collection solvents. The desirable solvent
characteristics are listed in Table 1.
Table 1. Desired Solvent Characteristics
• no C-H bonds
• high hydrocarbon solubility
• high purity
• liquid at ambient temperature
• nonflammable
• nontoxic
• not an ozone depleter
• not a greenhouse gas
• not a long term health hazard target cost 45/sample

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 8 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 (2900-3 100 cm - 1 ). Even with background
correction, the concentration of hydrocarbon contamination of the possible collection
solvents was prohibitively high for all solvents except the Fluorinerts® FC-77 and FC-72,
teirachiorohexafluorobutane and tetrachioroethene.
Using Freon-il 3® 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-il 3® standard. High
molecular weight hydrocarbons such as motor oil have very low solubility in FC-77.
Tetrachlorohexafluorobutane (TCHFB) and tetrachioroethene (Perc) exhibited hydrocarbon
solubility characteristics similar to Freon-i 13.
In stage four, the EPA independently selected tetrachloroethene as the analysis solvent.
FC-77 did not have adequate solubility for all hydrocarbons. Tetrachiorohexafluorobutane
has excellent analytical and Health & Safety characteristics, however, it is a
chiorofluorocarbon that if released into the ozone layer could deplete it. Tetrachloroethene
does present some long term health concerns but special laboratory precautions and using
small solvent volumes should reduce the risk to analysts. Table 2 summarizes the solvent
Table 2. Solvent Short List
Characteristic Freon-i 13 FC-77 TCHFB Perc
TPH solubility yes diesel only yes yes
Carcinogenic no no no yes
Ozone depletion yes no ? no
Boilingpoint(°C) 48 97 134 121
Vapor pressure (mm Hg) 334 42 12 19
Optimization of Extraction Parameters
The goal at this stage of the method development was to optimize the extraction parameters
so that this method could be used with as broad a spectrum of environmental matrices and
hydrocarbon mixtures as practical. Due to the lack of standard reference materials (SRMs)
with known “native” TPH concentrations, spiked analyte and unknown “native” analyte
samples were used in the development of this method. Using the recovery of spiked
analytes to prove quantitative extraction of “native” analytes is tentative at best and very
misleading at worst. Spiked compound matrix interactions may be much weaker than
“native” analytes which have had months or years to adsorb on and into the matrix.
Kaolin, Fullers earth and montmorillonite clay types were used to produce worst case
matrix spike recovery data. Each is a highly absorptive, fine particle matrix with a high
surface area. Spiked clay samples were tumbled over night to homogenize the sample and
enhance the absorbance of the spiked compounds. Water was added to some samples prior
to extraction to produce wet clay samples.

The wet sample test is vital since SFE has well documented problems with water. Also
many environmental “soil” matrixes have large percentages of water. Figure 1 shows the
% moisture found in 650 “solid” samples. The typical sample was 20% moisture but a
significant number of samples had moisture content above 50%. A rugged extraction
method should handle at least 95% of the sample load. Figure 2 plots the cumulative
frequency of %moisture for the same sample set. To reach the 95th percentile the
extraction must handle up to 60% moisture.



0 j__
0 20 40 60 80 100
% Moisture
Figure 1. Percent Moisture Distribution
0 20 40 60 80 100
% Moisture
Figure 2. Cumulative Frequency of %Moisture
Under some extraction conditions the water is released from the sample and impinges upon
the restrictor. If a conventional fused silica capillary tube is used as the restrictor the liquid
water may drastically reduce flow, thus aborting the extraction. Heating the restrictor
improves its resistance to water plugging but this is not a rugged solution to the problem.
A variable restrictor was developed specifically for this problem. The VariFlow® restrictor
(from Suprex) is a simple, low cost and “water proof’ variable restrictor. Water plugging
tests have shown that several milliliters of water can be pushed through the restrictor
without degrading restrictor performance. The overall SFE system is shown in Figure 3.
Data Analysis Tools
Two primary data analysis tools were used, factorial design and extractograms. Factorial
design allows the experimenter to vary several extraction conditions during a systematic set
of experiments. Statistical calculations estimate the effect of each parameter (factor) and the
interactions between factors. Thus, each extraction parameter is optimized and its
importance is determined. Figure 4 plots the effect of varying 5 common SFE factors. The

native hydrocarbons in a 2g Kaolin sample were extracted using various combinations of
the conditions listed in Table 3. The TPH concentration was about 20 mg/kg. The
restrictor was a 60 cm fused silica capillaiy tube. Analytes were collected directly in 5 mL
of FC-77. Figure 4 shows that the high value for pressure, oven temperature, static step
time and dynamic step time yields greater TPH recovery than the low value. The low value
for the resirictor produces higher TPH recovery.
Filter VariFlow
[ SCO2 I
Flow Measurementi
Water Trap-
Sample Vial
Figure 3. SFE System Diagram
Table 3. Factorial Design Parameters
The extractogram is a profile of extracted components versus time. Extractograms are
generated by splitting off a small portion of the CO 2 stream and directing it to a flame
ionization detector which responds well to petroleum hydrocarbons.
As expected the largest quantity of analytes are extracted near the start of the dynamic step.
After peaking, the amount of extracted analyte exponentially declines. Eventually the FID
returns to the baseline indicating that analyte extraction has ceased.
Controlled Zone
Liquid CO
Flow Measurement
Factor Low Value High Value Best results with
Pressure 200 atm 400 atm high
Oven Temperature 60°C 120°C high
Static time 0 mm 10 miii high
Dynamic time 10 mm 20 mm high
Restrictor I.D. 32 jim 40 pm low

Dry Soil Matrixes
Supercritical fluid extraction of hydrocarbons from dry soil samples with unmodified
carbon dioxide has been reported often. Mild extraction conditions are usually sufficient to
achieve acceptable analyte recovery.
Extraction of environmental matrixes requires the use of large sample quantities to ensure
that the aliquot is representative of the entire sample. SFE works best with small samples.
Sample aliquots of 3 g have been chosen as a reasonable compromise between these two
Wadsworth/ALERT participated in a recent EPA interlaboratory study to evaluate draft
Method 3560. Petroleum hydrocarbons were extracted from relatively dry soils.
Extraction conditions were, 340 atm, 80°C, 1.2 mLjmin C02, 30 mm dynamic, analyte
trap - solvent vial with 3 mL of teirachloroethene. IR analysis was performed by Midwest
Research Institute, Mountain View, CA.
Table 4. Dry Sample Recovery Data from the EPA Interlaboratory Study
Soil Number % Recovery Analyte
1 75 614 mg/kg native TPH
2 92 2050 mg/kg native TPH
3 90 32600 mg/kg native TPH
4 79 10000 mg/kg spiked TPH
5 blanksoil
6 87 10000 mg/kg spiked TPH
7 97 10000 mgJkg spiked TPH
Factorial design experiments were used in an attempt to optimize the extraction conditions
and shorten the 30 minute extraction time required by draft EPA method 3560. It was
thought that higher pressure, temperature and flow rate would shorten the extraction time.
The first fractional factorial used 3 g of ERA TPH #1 Lot 91019 as the sample. This QC
sample consists of vacuum pump oil spiked onto a clean, dry soil. A VariFlow restrictor
was used for flow control and the analytes were trapped on a solid phase absorbent. The
absorbent was eluted with 3 mL of tetrachlorohexafluorobutane. Extraction time was
determined from the extractograms. The factors and levels tested are shown in Table 5
below. The effect of each factor is shown in Figure 6.
The most prominent effect is that the static step lengthened the extraction by 4 minutes for
this sample. Pressure, temperature and flow rate had small or negligible effects. The two
extractograms in Figure 7 illustrate the difference between extractions with and without the
5 minute static step.
The static step was eliminated and another factorial experiment was designed to examine the
effects of pressure, temperature and flow rate. The second factorial design also used 3 g of
ERA TPH #1 Lot 91019 as the sample. Flow control and analyte trapping were the same.
The total extraction time was the sum of oven warm-up time and dynamic step. The factors
and levels tested are shown in Table 6 below. The effect of each factor is shown in
Figure 8. The extractor oven needs about 4 minutes to heat from 20°C to 150°C. The
dynamic step did not begin until the temperature had reached the set point, thus a 4 minute
static step was essentially added to the extraction. Since the elevated temperature did not

shorten the required dynamic time, the net effect of elevated temperature was a longer
extraction time. Once again pressure and flow rate had no effect. The two extractograms
in Figure 9 illustrate the difference between extractions with the temperature at 80°C and
Static step
- I Oven Flow
Pressure Temp Rate
2 factor interactions
Figure 6. Dry Sample Factorial Design #1 Results
Table 5. Factorial Design #1 Parameters
Factor Low Value High Value Best results with
Static time 0 mm 5 mm low
Pressure 340 atm 450 atm no effect
Oven Temperature 80°C 140°C high*
Flow Rate 600 mL/min 2000 mL/min high
CO 2 gas CO 2 gas
* Oven warm-up time was not considered during this experiment.
0 mm static, 340 atm, 140°C,
2000 mLlmin CO 2 gas
5 mm static, 450 atm, 140°C,
2000 mL/rnin CO 2 gas
10 20
: 3
Figure 7. Extractogram Comparison - With and Without the Static Step
C l)

2 & 3 factor interactions
Figure 8. Dry Sample Factorial Design #2 Results
Table 6. Factorial Design #2 Parameters
Factor Low Value High Value Best results with
Pressure 340 atm 500 atm no effect
Oven Temperature 80°C 150°C low
Flow Rate 600 mL/min 3000 niL/mm no effect
CO 2 gas CO 2 gas
Figure 9. Extractogram Comparison -
80°C and 150°C
0 mm static, 500 atm, 80°C,
3000 mLlmin CO2 gas
0 mm static, 500 atm, 1
3000 mLlmin CO 2 gas

Wet Soil Matrixes
Wet matrixes present several problems for SFE. First, the moist soil particles stick
together reducing sample exposure to the supercritical fluid. This may be solved by mixing
an absorbent with the sample. Second, if the water is released from the sample during the
extraction, the restrictor may plug. Restrictor plugging has been reduced or eliminated by
the use of heated or variable restrictors. Lastly, the water may modify both the matrix and
the supercritical fluid to help or hinder analyte extraction. Currently the modifier
phenomenon is not as well understood.
Extractions of very wet soil samples (>30% moisture) at 340 atm, 80°C, 1.2 mLjmin for 30
minutes have produced poor and erratic analyte recovery with some samples. NIST round
robin samples were extracted wet and dry. Elevated temperature and flow were used to
drive off the water and allow the extraction to proceed normally afterward. One gram soil
and clay samples were extracted at 500 atm, 150°C, 10 mm static, 10 mm dynamic, 2000
mL/min C02 gas flow. The analyte trap was 0.5 g of Varian EnvirElut solid phase
adsorbent. SFE Diesel recoveries compared well with Soxhiet extraction.
Wet clay was studied with a factorial design experiment to determine the effects of
pressure, temperature and flow rate. Fuller’s earth was spiked with motor oil (10,000
mg/kg). Water was added to make it 50% moisture. Four grams were used for each
extraction. The dynamic extraction time was fixed at 25 minutes and percent recovery was
measured. The factors and levels tested are shown in Table 7. Up to 1 mL of water was
released from the wet clay and passed through the VariFlow restrictor. No plugging was
encountered. The water then passed through the solid phase absorbant trap and was
directed to a waste container. Since the trap was at ambient temperature, the water did not
cause any problems. The effect of each factor is shown in Figure 10. Oven temperature
and CO 2 flow rate had very dramatic effects. Higher temperature and flow increased
percent recovery by about 25% each. Other experiments show that a 30 minute dynamic
step is necessary to achieve 100% recovery with this sample. Those extractions that
recovered the most motor oil also released the most water from the sample. Apparently the
water insulates the sample from the supereritical fluid. Removing the water allows the
extraction to proceed normally.
2 & 3 factor interactions
Figure 10. Wet Sample Factorial Design Results

Table 7. Wet Sample Factorial Design Parameters
Factor Low Value High Value Best results with
Pressure 340 aIm 450 atm no effect
Oven Temperature 120°C 150°C high
Flow Rate 2000 mL/min 4000 niL/mm high
CO 2gas OO2gas
Tetrachioroethene and teirachiorohexafluorobutane are acceptable analytical alternative
solvents for Freon-113. The EPA has selected tetrachioroethene because it is not a
chiorofluorocarbon. Hydrocarbons can be extracted accurately and reproducibly from
many dry soil matrixes with mild extraction conditions. Extraction of many wet soil types
requires high temperature, high flow rate,waterproofrestrictor and a waterproof analyte
3M, John Ruffing
Horiba Jim Vance
Midwest Research Institute, Viorica Lopez-Avila, Richard Young
Motter & Son Inc., William Motter
Rocky Mountain Analytical Laboratory, Marshall Tilbury
Suprex, Doug Koebler, Phil Hunsucker, Joel Fontaine, Joe Levy, Ray Houck, Jerry
Wisser, Lan Dalatta, Athos Rosselli
Varian, Rex Hawley, Max Erwine
Wadsworth/ALERT, Edward Bruner, Katie Ritz, Leslie VanKuren, Chuck Jacobs, Bob
Scafate, Brian Haueter, Russ Sommer, Brad Custer, Doug Stimson, Connie
Schussler, Mike Paessun, Kim Davis
“Surely You have outdone Yourself!” I shouted to my partner in the lab.
George Washington Carver (complimenz ing his LonI on the many mysteries hidden in the peanut)
1) Industrial Safety & Hygiene News, 01/88.
2) NASA’s Innovators, NASA Tech Briefs Vol.15, No.4, 1991, pp.114.

Bruce N. Colby and C. Steve Parsons, Pacific Analytical, 6349 Paseo Del Lago,
Carlsbad, California 92009
Abstract: The vacuum centrifuge offers an alternative to Kuderna-Danish distillation,
rotary evaporation and nitrogen blowdown as a means for reducing the solvent volume of
environmental sample extracts. A vacuum centrifuge has been tested for reducing solvent
volume in the analysis of 49 organochiorine pesticides and for 2,3,7,8-TCDD and
2,3,7,8-TCDF analyzed according to Method 8290. It provides a particularly effective
mechanism for removing low volatility solvents such as toluene and acetonitrile, does not
emit large quantities of solvent into the atmosphere and is easy and efficient to operate.
Nearly all environmental methods for the determination of semivolatile organic
compounds involve some form of solvent extraction. The resulting extracts are typically
reduced in volume by about a factor of about 250 prior to sample cleanup and analysis.
In general solvent volume is reduced using some combination of Kuderna-Danish (K-D)
distillation, rotary evaporation and nitrogen blow down. Each of these techniques has
advantages and disadvantages with respect to how well it performs and under what
circumstances. K-D for example, is very effective with low boiling solvents such as
methylene chloride (bp 40°C) and hexane (bp 69°C) but it is impractical when used with
higher boiling solvents such as toluene (bp 111°C) or acetonitrile (bp 82°C).
When reducing the solvent volume of an environmental sample extract, several criteria
must be taken into account. These include the required time and skill level required of
the bench of the chemist, the potential for cross contamination of samples, the cost and
space requirements of the equipment, the types and quantities of waste generated and any
impact the process might have on the recovery of target analytes under investigation. As
part of an ongoing effort to improve laboratory operations, a vacuum centrifuge was
evaluated with respect to these considerations. Emphasis was given to the reduction of
low volatility solvents, in particular acetonitrile and toluene, because these have proven
the most challenging from a laboratory efficiency standpoint.
The vacuum centrifuge tested was a Labconco Centrivap. It utilizes a 5 position sample
head which rotates at 1800 rpm. Each position holds a 50 mL sample tube. It is
evacuated using an Edwards 5.0 mechanical pump which generates a vacuum of 750 mm

Hg. All solvent vapors removed from the extracts are cryogenically trapped so solvents
are neither lost into the laboratory air nor vented up a hood.
An acetonitrile solution containing 39 organochiorine pesticides (Table 1) at 0.1 ug/mL
was tested first. Five 10 mL aliquots were reduced in the Centrivap as follows. After
placing the sample vials in the centrifuge head, the vacuum pump and rotor were turned
on. After 15 minutes, the device was stopped and the solvent level was checked. If
solvent remained, another 15 minute cycle was initiated. This process was repeated until
the samples had been taken to dryness. The total time required was 55 minutes. After
solvent reduction, 10 mL of toluene was added to each sample tube, then the tube was
capped and votexed for 30 sec. A 1 mL aliquot was placed in an autoinjector vial for
analysis by GC-ECD (HP 5890). An aliquot of blank solvent was concentrated and
analyzed in the same manner. Percent recoveries were calculated for each analyte.
A toluene solution was spiked with 2,3,7,8-TCDD -and 2,3,7,8-TCDF at 0.1 ng/mL. Five
10 mL aliquots were reduced in the Centrivap as described for acetonitrile. After solvent
reduction, 1 mL of toluene was added to each sample tube, then the tube was capped and
votexed for 30 sec. A 10 uL aliquot was then transferred to a micro-autosampler and 10
uL of internal standard solution was added. The contents of the vial were analyzed by
HIRGCIHRMS (VU 7OVSE). An aliquot of blank solvent was concentrated and analyzed
in the same manner. Percent recoveries were calculated for both analytes.
Acetonitrile is particularly useful solvent for extracting pesticides from soils, sediments
and sludges and from tissues. However, because it boils at 82°C, it is not practical to
carry out solvent volume reduction via Kuderna-Danish (K-D) distillation using a water
bath as the heat source. It is possible to use a heating mantle in place of the water bath
but the higher temperatures causes some of the pesticides to decompose. Consequently,
acetonitrile has always been reduced in volume via nitrogen blow down or rotary
evaporator. Nitrogen blowdown is very time consuming and produces a waste stream
which contaminates the atmosphere. Rotary evaporation is less time consuming, does not
contaminate the atmosphere but is prone to cross contamination if the solvent flashes. It
is also expensive and space intensive if multiple units are required to achieve thruput.
When acetonitrile was reduced in volume using the Centrivap, the time requirement is
very similar to that of rotary evaporation but multiple samples can be processed at one
time. Also, because of the centrifugal forge created by spinning the sample, flashing is
not a problem. Further, because no heat is used, decomposition of the thermally labile
pesticides is not encountered.
The recoveries experienced for chlorinated pesticides (Table 1) indicated that six of the
more volatile analytes, dichiorbenil, alpha-BHC, PCNB, gamma-BHC, heptachlor and
aldrin, were being partially lost, most likely through vaporization. These six analytes also

generated precisions which were two to three times worse than those of the other
analytes. In an attempt keep the six problem compounds in the sample vial, a 10 uL
aliquot of a keeper solvent was added prior to reduction. Nonane and cyclohexanol were
tested as keepers. The average recovery values were little changed as a consequence of
adding the keeper solvents but precision improved for most of the more volatile analytes
(Table 2). No target analytes were detected in any of the blanks.
Experiments are currently underway to determine the effectiveness of less powerful
vacuum pump (higher pressure) as a means to control losses with the more volatile
The other solvent tested, toluene, is used for extracting chlorinated dioxins (PCDDs) and
furans (PCDFs) from solid media. It boils at 111°C so K-D using a water bath is out of
the question. Heating mantles could be used to generate the required temperature but,
even though the PCDDs and PCDFs do not thermally decompose to any noticeable
degree at this temperature, the potential for creating dioxin contaminated wastes as a
consequence of glassware washing keeps K-D from being desirable. Consequently, the
rotary evaporator has been the technique of choice.
When the vacuum centrifuge is used to remove toluene from dioxin containing extracts,
excellent recovery and precision were encountered (Table 3). No dioxins or furans were
detected in the blank and after six months of operation with dioxin sample preparations,
no cross contamination or blank levels have been noted.
The vacuum centrifuge is an effective and efficient means for removing high boiling
solvents such as toluene and acetonitrile from sample extracts. It is particularly useful for
dioxin analysis where waste generation and cross contamination are minimized. It is less
effective for applications involving more volatile analytes such as some of the early
eluted organochlorine pesticides.

Table 1 - Pesticide Percent Recoveries
BHC, alpha-
BHC, gamma-
BHC, delta-
Heptachior Epoxide
Endosulfan I
Chiordane, gamma-
Chiordane, alpha-
Chlorobenzi late
Endosulfane II
Endosulfan Sulfate
29 15.1
96 5.0
98 5.9
66 16.1
44 19.7
91 4.0
81 9.2
99 5.0
58 13.0
96 3.8
56 14.5
98 2.0
106 7.7
95 5.5
98 2.6
96 4.3
92 4.7
108 8.0
95 3.6
97 1.6
97 1.1
95 5.5
97 3.8
101 0.7
97 3.5
101 3.2
101 2.7
99 7.9
101 2.7
99 4.4
101 2.5
100 4.2
119 4.6
102 1.4
101 2.1
103 1.0
100 4.3
105 2.1
101 1.0
RI R2 R3 R4 R5
Mean %sd

Table 2- Precision (%sd) With and Without Keeper Solvents
Acetonitrile w/Nonane w/Cyclohexanol
59.0 45.2 9.9
5.8 4.9 11.8
6.6 9.9 5.1
BHC, alpha-
27.1 25.5 6.2
50.1 34.9 5.4
5.0 6.8 10.4
BHC, gamma-
12.8 9.7 7.4
5.7 2.3 6.3
25.1 13.8 7.7
4.4 3.6 5.5
28.7 14.1 5.7
2.3 6.3 8.8
8.1 7.5 5.2
6.4 6.3 8.8
BHC, delta-
2.9 2.0 4.7
16.7 12.8 7.3
Table 3 - Chlorinated Dioxin and Furan Recoveries
RI R2 R3 R4 R5
100 102 117 96 102
100 113 120 106 101

Werner F. Beckert , U.S. Environmental Protection Agency, EXSL—LV,
Las Vegas, Nevada 89119, and Viorica Lopez-Avila, Midwest
Research Institute, California Operations, Mountain View,
California 94043.
EPA SW—846 Method 9073 for the determination of total recoverable
petroleum hydrocarbons (TPH3) specifies infrared (IR) analysis of
extracts prepared from the samples using Freon—113. Soil samples
are subjected to Soxhiet extraction with Freon—113, according to
Method 9071A, followed by silica gel cleanup of the extract to
remove interferences. Since production and use of Freon-type
materials are being phased out, a new extraction method is
needed. The replacement solvent may not be of the Freon—type,
and it also must not contain any C-H bonds or other bonds that
would interfere with the IR determination of th. petroleum
hydrocarbons. Furthermore, sinc, the parameter of total
recoverable petroleum hydrocarbons is operationally defined by
the method (i.e., Method 9073), the new solvsnt should show the
sam. or very similar extraction efficiencies for TPHs.
We and others have reported that carbon dioxide, under
supercritical conditions(sup.rcritical fluid extraction - SFE),
is a good extraction medium for hydrocarbons from soil samples.
We have demonstrated for a limited number of solid matrices that
extraction of TPHs from soil samples with supercritical. carbon
dioxide gives recoveries that are similar to those obtained with
Freon-113 using Soxhiet extraction. The STE collection solvent
found to be most suitable as a replacement for Freon—113 was
tetrachioroethylene, which is transparent in the IR region of
This new method is now being subjected to a multilaboratory
evaluation study. Fifteen volunteer laboratories are
participating in this study. They will each extract nix ,, solid
samples, thre, of them in triplicate, and mail the fifteen
extracts to the lead laboratory, ) I - California operations, for
IR analysis. Both carbon dioxide and tetrachloro.thy]ene will be
provided to the participating laboratories to assure that all
laboratories are using solvents of the same quality.
The results of the studies will be evaluated according to
standard procedures, and the results will be reported at the

Mark L. Bruce . Director of Research and Development, Rita Tomayko, GC Specialist and
Marvin W. Stephens, Technical Director, Wadsworth/ALERT Laboratories, Division Of
Enseco, Inc., 4101 Shuffel Dr. N.W., North Canton, Ohio 44720
Azeotropic distillation has been used to effectively concentrate alcohols and other water
soluble volatile organic compounds from aqueous samples. This method has been
modified for use with solid samples. Water extraction of the solid sample is combined
with distillation of the volatile components from the aqueous extract. The distillation also
functions as a cleanup. Most nonvolatile and semivolatile sample components will not be
transferred to the distillate. This greatly reduces the contamination of the OC injection
port and chromatographic column. Total extraction and distillation time is less than 15
minutes. Concentration factors are typically one order of magnitude with a 5 g sample.
Currently methanol and other water soluble volatile organic compounds in solid matrixes
are extracted with water then analyzed by direct injection of the sample extract or purge-
and-trap, followed by gas chromatographic separation and detection. For many
compounds the method detection limits are not low enough to meet client needs. The
overall method response of the alcohols in particular is poor for direct aqueous extract
injection and purge-and-trap.
A microdistillation system was developed to address the shortcomings of direct aqueous
extract injection and purge-and-trap. The Wadsworth MicroVOC 3 was developed at
Wadsworth/ALERT Laboratories and currently is manufactured and sold by Shamrock
Glass of Seaford, Delaware. VOC 3 is an acronym for Volatile Organic Compound
Concentration and Cleanup. The microVOC 3 has been described in detail previously
(1,2). The system is shown in Figure 1.
Fractional factorial experimental designs have been used extensively through the
optimization process. Factorial design is a statistical procedure which estimates the effect
of each factor as well as interactions. EPA SW-846 Method 8015 (modified) was used
for analysis.
Gas Chromatograph/Data System
Hewlett Packard 5890 equipped with a flame ionization detector
Gas Chromatography Columns
Quantitation: RTX-Volatile, 30 m X 0.53 mm I.D.
Wadsworth MicroVOC 3 System®, Shamrock Glass (see Figure 1.)
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
Pipe insulation, polyurethane foam, 11/2” O.D., /8” I.D., 55 cm in length

Glass beads, 5 mm O.D.
Keck clamps, for 14120 ground glass jomt
Glass reducing union, 14/20 ground glass joint to 6 mmO.D. tube
Stainless steel reducing union, 1/16” to 1/4”
Air condenser, Teflon® tubing, 1/16” O.D., 1/32” Ii). (40 cm in length)
GC autosampler vials
Autosampler vial inserts, lOOp1, calibrated
Graduated cylinder, 50 ml
Support stand with rod, 1 meter
Three-finger clamp
Heating mantle, Glas-Col, 115 volts, 230 watts, S1’M 400
Temperature controller, Glas-Col PL1 15-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-1-propanol, Aldrich, purity 99.9%
1-Butanol, Aldrich, purity 99.8%
1,4 Dioxane, Aldrich
Acetonitrile, Aldrich, purity 99.9%
Propiomtrile, Aldrich, purity 99%
Acrolein, Aldrich, purity 97%
Acrylonitrile, iT Baker, purity 99%
Ethyl Acetate, Aldrich, purity 99%
Reagent water, deionized
The azeotropic microdistillation method is summarized in Figure 2. A 5 g aliquot of
sample is transferred to a round bottom flask. Internal standard(s) are added to the
sample followed by 40 niL of reagent water. 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 and at room temperature. The sample is heated to
the boiling point The fIrst 100 p1 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 less
than 50%. This calibration procedure is analogous to purge-and-trap calibration
The key parameters to optimize were sample weight, reagent water volume and heat input
(ie the reostat setting). Initial experiments evaluated the benefit of performing the initial
water extraction at a low heat input setting and followed by a quick distillation similar to
the method for aqueous samples.

Air Condenser
Collection Vial
Reducing Unions
Keck Clamp
I— Fractionation Column
4 Insulation
,Keck Clamp
Figure 1. MicroVOC 3 distillation system

The initial factorial design set of experiments with methanol
spiked loam and clay samples indicated that 20 mL of water was
better than 40 mL Also a hotter water extraction was better. The
second set of factorial experiments showed better performance
with 10 mL of water and a higher boiling rate. Methanol spiked
incinerator ash samples were used. The third factorial set
demonstrated that 10 mL of water was not enough for some very
absorptive matrixes such as dry incinerator sludge. The fourth
factorial design used “native” methanol in both incinerator sludge
and ash. The water volume effect was exactly the opposite of
what had been measured before with matrix spiked methanol.
More water (40 mL vs 15 mL) was much better. As expected
“native” analytes behaved differently than spiked analytes.
Apparently extracting the methanol from the solid matrix was
more difficult than distilling the methanol from the water extract.
Thus, using more water improved the extraction more-than it hurt
the distillation. Heating at the higher heat input setting was also
more efficient. Presumably the higher temperature improved the
water extraction step. Final optimized parameters will be
Data for the following compounds will be presented: methanol, 1-
propanol, 2-methyl-i -propanol, 1 -butanol, 1 ,4-dioxane,
acetonithie, propionitrile, acrolein, acrylonitrile and ethyl acetate.
Various solid matrixes have been studied. New internal standards
are being investigated since ethanol (the internal standard from the Figure 2.
aqueous version of the method) has been added to the analyte list. Method Summary
The microdistillation system (Wadsworth MicroVOC 3 ) has been used extensively for
aqueous matrixes. The water method has been modified for use with solid matrixes.
Small sample aliquots are required (5 g). Analyte concentration factors are about one
order of magnitude when a 5 g sample aliquot is used. The total extraction and
distillation time is less than 15 minutes.
1) Bruce, M.L., Lee, R.P., Stephens, M.W., Concentration of Water Soluble Volatile
Organic Compounds from Aqueous Samples by Azeotropic Microdistillation, Seventh
Annual Waste Testing and Quality Assurance Symposium Proceedings Vol II, p. 2, 1991
2) Bruce, M.L., Lee, R.P., Stephens, M.W., Concentration of Water Soluble Volatile
Organic Compounds from Aqueous Samples by Azeotropic Microdistillation, ES&T , 26
(l),pp. 160-163, 1991.

Paul H. Chen , William A. VanAusdale, W. Scott Keeran and Dwight F.
Roberts, Analytical Services Division, Environmental Science and
Engineering, Inc., P.O. Box 1703, Gainesville, Florida 32602-1703
In EPA Methods 625 and 8270 samples are generally analyzed for the target compound list
(TCL) components by an automated data system and for non-TCL components by a library
search. Non-TCL components identified and reported as tentatively identified compounds
(TIC) are rarely verified and sometimes incorrect information is reported. This incorrect
information includes incorrect compound identification, contamination in the lab or during
sampling, and artifacts produced during sample preparation. The purpose of this study
was to identify the artifacts formed during sample preparation using EPA Methods 625 and
8270. During our analysis of thousands of samples we have identified thirty-three artifacts
formed by the following reactions: I. Oxidation of phenolic surrogates and phenolic target
analytes: Three oxidation products of phenolic surrogates were reported by us two years
ago at this symposium. We report here the identification of oxidation products of some
phenolic target analytes. The compounds identified include benzoquinone, chlorobenzo-
quinone, 2,6-dichlorobenzoquinone, 2,6-dichiorohydroquinone, and tetrachiorohydro-
quinone. II. Halogenation or nitration of phenolic surrogates: The compounds identified
include 2- and 4-chlorophenols-d4, 2- and 4-bromophenols-d4, 2-nitrophenol-d4, 4-chlo-
ro-2-fluorophenol, 4-bromo-2-fluorophenol, and 4-nitro-2-fluorophenol. 111. Reaction of
preservative cyclohexene in methylene chloride with halogens to form halogenated cyclo-
hexanols, cyclohexanes, and cyclohexenes: Ten of these artifacts were identified, five of
them were major compounds and were reported elsewhere. Five minor compounds identi-
fied in this study are 3-chlorocylohexene, 3-bromocyclohexene, 2-chiorocyclohexanol,
1,2-dichlorocyclohexane, and 1-bromo-2-iodocyclohexane. IV. Autoxidation of cyclohex-
ene: Cyclohexene oxide (7-oxabicyclo [ 4.l.OlJheptane), 2-cyclohexen-l-ol, and 2-cyclo-
hexen-1-one are often present in the extract. 1-Formylcyclopentene and 1,2-cyclohexane-
diol are found sometimes. V. Reaction of cyclohexene with methylene chloride degrada-
tion products: 2-Chlorocyclohexanol was identified. VI. Aldol condensation of acetone:
4-Hydroxy-4-methyl-2-pentanone and 4-methyl-3-penten-2-one were found. If these
compounds are found in the sample, they should be labeled as artifacts and not considered
as contaminants in the site with a possible exception of 4-hydroxy-4-methyl-2-pentanone
and 4-methyl-3-penten-2-one which might be present in the sample before extraction.

EPA Methods 625 (1) and 8270 (2) are widely used in environmental laboratories for the
analysis of semivolatile organics by GCIMS. In these methods samples are generally
analyzed for the target compound list (JCL) components by an automated data system and
for non-TCL components by a forward library search of a published mass spectral data
base (3). The results of TCL components analysis are generally accurate. However, non-
TCL components identified and reported as tentatively identified compounds (TIC) are
rarely verified and sometimes incorrect information is reported. This incorrect information
includes incorrect compound identification, contamination in the lab or during sampling,
and artifacts produced during sample preparation. Though artifacts should be reported to
the client, it is important that they should be labeled as artifacts and not considered as
contaminants in the site. Artifacts formed during extraction of brine samples with cyclo-
hexene-inhibited methylene chloride were reported by Campbell et al. (4) and Fayad (5).
We reported earlier (6,7) artifacts formed by oxidation of phenolic surrogates during acid
extraction of water samples using EPA Method 625. The purpose of this study was to
identify the artifacts formed during sample preparation using EPA Methods 625 and 8270.
Sample Preparation . Water samples were extracted at pH>! 1 and then at pH<2 with
methylene chloride according to EPA Method 625 (1). Soil samples were extracted with
methylene chloride in Soxhiet extractors according to EPA SW-846 Method 3540/8270 (2).
Before extraction, each sample was spiked with 1.0 mL of surrogate spiking solution
which contains 100 pg/mL each of acid surrogates (2-fluorophenol, phenol-d5, and 2,4,6-
tribromophenol) and 50 jtg/mL each of base/neutral surrogates. Methylene chloride extract
was concentrated to 1 mL with Kudema-Danish concentrator and analyzed by GC/MS.
GCJMS Analysis . Samples were analyzed on a HP 5988 GCIMS system. The column
used was a 30m x 0.25 mm i.d. DB-5 (0.25p.m coating) fused silica capillary column
(J&W Scientific, Folsom, CA). The column temperature was held isothermal at 40°C for 4
minutes and then programmed at 10°C per minute to 280°C, and held isothermal at this fmal
temperature for 12 minutes. The mass spectrometer was scanned from 35 to 500 amu per
half second. The extract was spiked with a mixture of six internal standards before GC/MS
analysis according to CLP protocol (3). A forward library search was performed for non-
TCL compounds on a WiIeyINBS data base which contains 139000 different spectra (8).
Compounds were tentatively identified by library searches or by elucidation of the corn-

pound structure from its mass spectrum if no match was found in the library. The tenta-
lively identified compounds were confumed by the agreement of mass spectra and retention
times between the sample component and the authentic compound.
Three oxidation products of phenolic surrogates were reported by us two years ago at this
symposium (6). The mass spectra of these three compounds are reproduced in Figures 1A
to 1C. We report here the identification of oxidation products of some phenolic target ana-
lytes. The compounds identified include p-benzoquinone, 2-chloro-p-benzoquinone, and
2,3,5,6-tetrachiorohydroquinone which are oxidation products of phenol, 2-chlorophenol,
and pentachlorophenol, respectively. The mass spectra of these three oxidation products
are shown in Figures 1D to iF. These compounds were found in some matrix spike sam-
pies which were spiked with five phenolic matrix spike compounds according to CLP
protocol (3). When the samples which contained the oxidizing agents (7) were spiked with
the complete list of 14 phenolic target anaiytes (3), 2,6-dichloro-p-benzoquinone and 2,6-
dichiorohydroquinone were also found. The mass spectra of these two compounds are
displayed in Figures 1G and 111. These two oxidation products are probably from the oxi-
dation of 2,4,6-trichiorophenol.
The three acid surrogates used in the EPA methods for the semivolatile org anics analysis
are phenolic compounds. The phenolic group activates the aromatic ring toward eiec-
trophilic substitution at ortho and para positions of the ring. Therefore, it is not surprising
to find some artifacts formed during extraction by halogenation or nitration of the phenolic
surrogates. The artifacts found in the samples are 2-chlorophenol-d4, 4-chiorophenol-d4,
2-bromophenol-d4, 4-bromophenol-d4, 2-nitrophenol-d4, 4-chloro-2-fluorophenol, 4-
bromo-2-fluorophenol, and 4-nitro-2-fluorophenol. Their mass spectra are shown in
Figures 2A to 2H.
Methylene chloride is the solvent used in the extraction of samples for both EPA Methods
625 and 8270. Cyclohexene is added to methylene chloride by the manufacturer (Burdick
& Jackson) as a preservative and scavenger (4,5). The brand of methylene chloride we
used (B&J Brand ) is most widely used in the environmental labs. A number of artifacts
are formed by autoxidation (9) of cycloexene and by the reaction of halogens with cyclo-
hexene in the presence of halogens or halides and oxidizing agents in the samples. The
most common artifact formed by autoxidation during sample preparation using cyclohexene
-inhibited methylene chloride is cyclohexene oxide (7-oxabicyclo [ 4. 1.0]heptane).

2-Cyclohexene-1-ol and 2-cyclohexen-1-one, and to a lesser extent, 1-formylcyclopentene
and 1 ,2-cyclohexanediol are also often found in the extract. The mass spectra of the arti-
facts from autoxidation of cyclohexene are shown in Figures 3A to 3E.
A number of halogenated cyclohexanols, cyclohexanes, and cyclohexenes are found in
samples which contain halogens or halides and extracted with cyclohexene-inhibited
methylene chloride. These artifacts are usually formed during extraction at acidic condi-
tions (7). The most common artifact found is 2-iodocyclohexanol. In addition to 2-
iodocyclohexanol, 1-chloro-2-iodocyclohexane, and to a lesser extent, 2-bromocyclohex-
anol are sometimes found in the samples. In high salinity or brine samples, we found large
amounts of 2-bromocyclohexanol, 1 -bromo-2-chlorocyclohexane, 2-iodocyclohexanol,
1,2-dibromocyclohexane, and 1-chloro-2-iodocyclohexane (7) (see Figure 4 for the mass
spectra of these compounds). Similar fmdings were reported by Campbell et al. (4). A
much smaller amount of 3-chiorocyclohexene, 3-bromocyclohexene, 2-chiorocyclohex-
anol, 1,2-dichiorocyclohexane, and 1-bromo-2-iodocyclohexane was also found in the
brine sample. 2-Chiorocyclohexanol was sometimes formed by the reaction of cyclohex-
ene with the degradation products of methylene chloride. This is supported by the presence
of a substantial amount of 2-chiorocyclohexanol in some methylene chloride which has not
been used in the extraction of samples.
If acetone is present in the sample, the aldol condensation product of acetone, 4-hydroxy-4-
methyl-2-pentanone, and its dehydration product, 4-methyl-3-penten-2-one, are often
formed during extraction. These two compounds may also be present in samples before
extraction. The mass spectra of these two artifacts are shown in Figure 5. Extraction of
soil by sonication (SW-846 Method 3350) using acetone/methylene chloride as a mixed
solvent generates a number of other aldol condensation products in addition to the two arti-
facts mentioned above.
In summary, thirty-three artifacts formed during sample preparation were reported in this
study. We plan to put the mass spectra of these compounds in a library which can be
searched routinely. If these compounds are found in the sample extracted with cyclohex-
ene-inhibited methylene chloride, they will be labeled as artifacts and would not be consia-
ered as contaminants in the site with possible exception of 4-hydroxy-4-methyl-2-pen-
tanone and 4-methyl-3-penten-3-one which might be present in the sample before extrac-
tion. Among the thirty-three artifacts, the mass spectra of the following thirteen com-
pounds are not in the 1989 Wiley/NB S Registry of Mass Spectra Data (8): p-benzo-
quinone-d4, 2-fluoro-p-benzoquinone, 2,6-dibromohydroquinone, 2,6-dichiorohydro-
quinone, 4-chlorophenol-d4, 2-bromophenol-d4, 4-bromophenol-d4, 4-chloro-2-fluo-

rophenol, 4-bromo-2-fluorophenol, 4-nitro-2-fluorophenol, 2-iodocyclohexanol, 1 -bromo-
2-iodocyclohexane, and 1 -chloro-2-iodocyclohexane.
We thank P. Dumas and his department for extracting the samples and D. Schindler and C.
Diaz for their assistance in GCIMS analysis.
1. EPA 40 CFR Part 136 Fed. Regist., 1984 , 49 (No. 209).
2. USEPA Test Methods for Evaluating Solid Waste: Physical/Chemical Methods,
SW-846, 3rd Edition, 1986.
3. USEPA Contract Laboratory Program Statement of Work for Organic Analysis,
Multi-Media, Multi-Concentration, 1988.
4. Campbell, J. A.; LaPack, M. A.; Peters, T. L.; Smock, T. A. Environ. Sci.
Technol. 1987 , 21, 110.
5. Fayad, N. M. Environ. Sci. Technol. 1988 , 22, 1347.
6. Chen, P. H.; VanAusdale, W. A.; Roberts, D. F. Proceedings of Sixth Annual
Waste Testing and Quality Assurance Symposium, 1990, Vol. II, 142.
7. Chen, P. H.; VanAusdale, W. A.; Roberts, D. F. Environ. Sci. Technol., 1991 ,
25, 540.
8. McLafferty, F. W.; Stauffer, D. B. The Wilev/NBS Registry of Mass Spectral
Data , Wiley: New York, 1989.
9. Pryor, W. A. Free Radicals , McGraw-Hill: New York, 1966; p. 288.

100. E 2
80.54 82
60: ‘ 60 ‘
/ 114
80 120
T 1
I ’
1 JL lilt IL
- i•. .111
•• . - -- . .-
f-’-’-’ - ’ -i
Figure 1. Mass spectra of artifacts from oxidation of phenolic surrogates and target
I i .
100 120
I 0
80 ’ 1 •

2-ChIoropheno -d4
A ia
40. N
i . 1 1 pI .. ‘ri.. . .,

41 I 85
20 / 113
i ijIlti Iii I
I I .
40 60 80 100 120 140
40. N.
20 41 7 -‘ L
0 ‘i” . . I i .1-I .•. it t.(l II . .11.1 . I it
4-Chloro-2-fluoropheno t
20 2 16
Igt .._.Ijul ... .i .Ii .
40 60 80 100 120 140
40 80 120 160
60 69
2 0 .5k ‘
n it H
4-Bromo-2-fluoropheno l
40 57
9 N. 170
20 E uN.

.. - .1 II lilt
40 80 120 160 200
40 80 120 160 200
E) 176
40 97

4-Nitro-2-fluorophe not
60 83
.. . . . 1
40 80 120 • 160
40 80 120 160 200
Figure 2. Mass spectra of artifacts from halogenation or nitration of phenolic surrogates.

Cyclohexene oxide
60 41
67 - 97
0 ..•. I
i.Iii. Ii li,
35 40 45 50 55 60 65 70 75 80 85 90 95 100
2-Cyclohexen-1 -01
83 97
40. 55 - /
20 ‘ ‘ 53 67
II -
O..% I I’ii’. 1
35 40 45 50 55 60 65 70 75 80 85 90 95 100
2-Cyclohexen-1 -one
100 -
60 96
39 40
20- ‘-. — 53 67 81
- I — ‘
l%.tI i.i . i—
35 40 45 50 55 80 85 70 75 80 85 90 95 100
1 -Formyicyclopentene
80. -
- 95
40. 41
20. 53 68
a . . . . i. t .
35 40 45 50 55 60 65 70 75 80 85 90 95 100
I ,2-Cyclohexanediol
100. ° E
60. 5 ,7
40. 41 43
20 /
o . . . Iii I. 1’ 1 I .-.- -i . iii. .-i , 1. .
40 50 60 70 80 90 100 110 120
Figure 3. Mass spectra of aitifacts from autoxidation of cyclohexene present in the
methylene chloride used for extraction.

100: 57
Il 132 180
0 i_# r 4 1L,...,._i •,•• ,. .•,•. - .1•••,
40 80 120 160 200
40. 53
1 J 6
0 . l . IlI .%.(IIIII .11...., I
40 60 80 100 120 140
1 -Bromo-2-chlorocyclohexane
81 D

4° 117
20.41 i I

40 80 120 160 200
4° -
20. I 160
40 80 120 180
20. 55
127 155 226
0— -1—J .— . .-—l
40 80 120 160 200 240
. 80
40. 88
. 4
20. 134
. I I I 116 ,
0- thu i.i.h,)JI . 1 :i .j ....1...i...ui uf l
40 60 80 100 120 140
1 ,2-Dibromocyclohexane
100. -
20. 41 161
• .‘ 119 ‘ 242
k i’ . .’
( .1
1 ,2-Dichlorocyclohexane
100. —
41 75
20 ‘ 116
ii 119 152
0Jji. ..jlIJ ,Lfl.J .I .I+-•,•,.I.•—___
40 80 120 160
40 80
120 160 200 240
1 -Chloro-2-iodocyctohexane
100. -
- 117
(.•• .,
40 80 120 160 200 240
1 -Bromo-2-iodocyclohexane
100 —
50 100 150 200 250 300
Figure 4. Mass spectra of artifacts from reaction of halogens with cyclohexene present
in the methylene chloride used for extraction.

43 A
100 /
20. 101
- 39 56 98
n_. . tj 1 -lJ II....I.. 1.111 1 I . I.,
40 50 60 70 80 90 100 110
60 98
51 67 77
0 _ . 1 I.:.Il . J)J.J, J I I
40 50 60 70 80 90 100 110
Figure 5. Mass spectra of two common aldol condensation products of acetone.

Charles R. Hecht , Senior Analytical Chemist, Frank Thomas, Senior Scientist, John W.
Kolopanis, Director of Analytical Programs, Chemical Waste Management, Inc., 150 West
137th Street, Riverdale, I I. 60627
Reversed phase HPLC coupled with ultraviolet detection has been used for the analysis of
chloropheoxy herbicides for many years, and proven to be a sound analytical technique.
Numerous applications have been researched and well documented in the technical journals.
The use of HPLC and the consequent ability to analyze these herbicides in the free acid
form leads to a significant cost and time savings. Current methodology using SW-846
Method 8150 requires a series of lengthy ether extractions, followed by a K-D concentration
and caustic hydrolysis, followed by further ether extractions and K-D concentration.
Diazomethane esterification is then needed before gas chromatographic analysis can be
done. Typically 8 samples require 2 days by a single analyst for the complete extraction,
hydrolysis, derivatization and GC cycle. Additionally, there are major health and safety
concerns with the use of ether and diazomethane.
A streamlined analyses was developed for 2,4-D, 2,4,5-T, and silvex using a simple two hour
potassium hydroxide hydrolysis followed by acidification and a four minute reversed phase
HPLC analysis. This method was evaluated on a number of hazardous waste matrices
including wastewater, contaminated soils, oils, waste solvents, ash, sludges, and TCLP
extracts. Typically 16 samples or TCLP extracts can be prepared and analyzed by a single
analyst in 1 day.
Experimental results have been extremely good with less than 10% relative standard
deviation reported during method detection limit studies for all three compounds. Sample
preparation and analysis conditions, chromatography, UV spectra, precision and accuracy
data, and spike recovery data will be presented.

Liquid chromatography (LC) has been used in the pharmaceutical and food industries for
many years because of the ability to analyze many different types of compounds (eg. varied
compound classes that are thermally labile, high molecular weight or non-volatile.) More
recently liquid chromatography has gained interest in the environmental field. In addition
to an approved method for the analysis of polynuclear aromatic hydrocarbons (PNA’s), the
EPA has also devoted a significant amount of time and money to the evaluation of liquid
chromatography/mass spectrometry because of its promising use in the environmental field.
The chiorophenoxy herbicides are of major concern to the environmental community as well
as the regulatory agencies because of their toxic effects and widespread use. The monitoring
of these compounds is needed to prevent adverse environmental and health effects. The
analyses of these compounds is required under the RCRA program 40 CFR. Part 261.24
requires the TCLP extraction and analysis for the compounds when determining if a solid
waste displays a hazardous characteristic. In addition, 40 CFR Part 268 requires the
analysis of these compounds by total constituent analysis in waste extracts when determining
if a hazardous waste meets the Land Disposal Restriction treatment standards. Analyses
of hazardous waste and TCLP extracts for the presence of these herbicides is currently
performed using SW-846 Method 8150. Method 8150 is both time consuming and
potentially hazardous. The use of ethyl ether as the extraction solvent and diazomethane
as the methylating agent are of major concern because of their explosive and toxic
properties. The method is very labor intensive requiring multiple preparation steps before
analyses can be performed. It has proven difficult to consistently achieve quality results in
the complex hazardous waste matrices outlined in this paper. The HPLC method has
proven to be analytically sound over a wide range of matrices, and its use, with the
simplified preparation procedure can decrease analyses time four fold as well as eliminate
the use of ethyl ether and diazomethane.
The purpose of this work was to develop a streamlined method utilizing HPLC with
ultraviolet detection that can determine herbicides in hazardous waste matrices and TCLP
extracts. The simplification of the extraction and analyses will increase analyst safety,
reduce hazardous chemical use (eg. ethyl ether and diazomethane) and reduce the analytical
costs and sample turnaround times.

In liquid chromatography as in most chromatographic techniques, separation of the analytes
is achieved by partitioning the analytes between the mobile and stationary phases. In LC
the solubility of the analytes in a solvent based mobil phase versus their solubility in a liquid
stationary phase dictates the separation. The two basic types of partitioning are normal
phase (eg. the stationary phase is more polar than the mobil phase), and reverse phase (eg.
the mobil phase is more polar than the stationary phase). The availability of a wide range
of solvent types and stationary phases in addition to being an ambient technique makes LC
applicable to a wide range of compound types. The need to make volatile and thermally
stabile compounds needed for gas chromatographic (GC) analysis is eliminated. Because
of this the chiorophenoxy acid herbicides can be analyzed in their acid form instead of the
methyl ester form needed for GC. This eliminates the diazomethane derivatization step of
Method 8150.
The HPLC instrumentation used for this study was purchased from Millipore/Waters. The
instrument was configured as follows:
Model 600-MS controller and solvent delivery system
Model 991-MS Photodiode array detector
NEC 386/20 data system with PDA software
Nova Pak C18 Radial Compression Column SxlOOmm, 4 micron
Nova Pak C18 Guard Pak, 4 micron
Rheodyne Model 7125 manual injection valve with 500uL external sample loop

Chromatographic Conditions
The HPLCfUV chromatograph of the herbicide acids is shown in Figure 1. The peak
shapes are excellent, baseline resolved and of approximately equal intensity. The mobile
phase used was 75% methanol /25% water with 0.5% H 3 P0 4 at a flowrate of 2.5mLlmin.
A 500uL injection volume was delivered using an external sample loop. The photodiode
array detector was programmed to scan from 200 to 240 nm. 207nm was used as the
primary quantitationwavelengthwith 227nm and 235nm used as confirmatory and secondary
quantitation wavelengths. Standard spectra and contour plots are shown in Figures 2 and
Instrument Calibration
Instrument calibration was done at multiple levels to define the linearity and working range
of the photodiode array detector. Figure 4 shows the calibration curve plotted as the least
squares fit for nanograms on column versus peak height.
Hazardous Waste
Precision and Accuracy Study
Precision and accuracy studies were done on four different hazardous waste matrix types,
water/oil mix, contaminated soil, PCB oil, and waste solvents. Table 1 shows the data
collected. For a 200 ppm spike of various esters <10% relative standard deviation is
reported for all matrices with all but one in the 1-6% range.
Hazardous Waste Sample Preparation
Five grams of sample is accurately weighed into a lOOmL volumetric flask. (NOTE: If the
sample is a dense organic liquid eg. high PCB oils, chlorinated solvents etc., weigh 2.5g of
sample into the lOOmL volumetric then add 2.5g hexadecane, then mix well.) Fifty mL of
deionized water s then added to the volumetric flask. The sample is made alkaline by
adding lOmL of 37% KOH solution. Alkaline conditions are verified by checking the pH
and if needed additional KOH is added. The sample is then stirred at 70-80°C for 2 hours.
The extract is then brought to lOOmL volume with deionized water after drawing off any oil
layer or filtering off any solid material. The extract is then made acidic by adding
hydrochloric acid. The acidified extract is analyzed using reversed phase HPLC.

Hazardous Waste Sample Analyses
As a method ruggedness test a variety of hazardous waste samples were analyzed both as
received and after spiking to check for possible matrix effects or interferences. Samples were
spiked at both 200 and 1000 ppm. Sample types included PCB oils, dielectric fluids, oil
bottoms, waste oils, contaminated solvents, lab liquids, solids, sludges, incinerator ash, dry
scrubber solids, and others. Table 2 shows the data collected expressed as percent recovery.
Data Discussion
Tables 1 and 2 demonstrate the precision and accuracy of the HPLC/UV method when
analyzing hazardous waste. Hazardous waste matrices are extremely varied and complex
and typically require extensive cleanup procedures when using current SW-846
methodologies. The HPLC/UV method is robust enough to handle a wide variety of
matrices. Less than 10% RSD was seen when doing replicate analyses of spiked samples.
The four matrix types in Table 1 were chosen because they represent the common and
complex matrix types typically analyzed. Table 2 shows that spike recoveries of between 73
and 127% were seen during routine analyses. A variety of esters were used in the study.
TCLP Extract
Precision and Accuracy Study
Precision and accuracy studies were done on both TCLP extraction fluids #1 and #2. Table
3 shows the data collected. For a 1 ppm spike of the methyl esters, extraction fluid #1
shows <1% RSD, fluid #2 shows <2% RSD.
TCLP Extract Sample Preparation
Fifty mL of TCLP extract is added to a lOOmL volumetric flask. The extract is made
alkaline by adding 5mL of 37% KOH solution. Alkaline conditions are verified by checking
the pH and if needed additional KOH is added. The extract is stirred at 70-80°C for 2
hours. The extract is reacidified to pH 2 using hydrochloric acid, then brought to lOOmL
final volume with deionized water. The acidified extract is analyzed using reversed phase

TCLP Sample Analyses
As a method ruggedness test a variety of waste sample TCLP extracts were analyzed as
received and after spiking to check for possible matrix effects or interferences. Extracts were
spiked with methyl esters at approximately lppm. Sample types included sludges, carbon
fine dust, used fuel filters, contaminated soils, sands and debris, sludges and others. Table
4 shows the data collected.
Data Discussion
The data in Tables 3 and 4 demonstrates how precise and straightforward the analysis of
TCLP extracts by HPLC/UV is. Less than 2% RSD and average recoveries of between 90
and 100% at a lppm spike level are easily and consistently attainable. The minimal sample
preparation requires approximately 2.5 hours, and the entire procedure is carried out in a
single lOOrnL volumetric flask.
1. High pressure liquid chromatography with ultraviolet detection is an excellent
alternative analytical tool to GC/ECD for the analysis of the chiorophenoxy
2. The use of the simplified sample preparation procedure increases analyst safety and
significantly reduces analysis time. The use of ethyl ether and diazomethane are
eliminated, and easily four times the number of samples can be analyzed by a single
analyst as compared to Method 8150.
3. The combined use of HPLC/UV and the simplified sample preparation procedure
produces a robust, analytically sound method that can be used to analyze hazardous
waste and TCLP extracts.

LC724.DT3 05—06—1992 14:09:27 Sample name lppm std
Y-scale 1.1 AU’FS Paper speed 20 mm/mm
Sampling time 44 msec *lt)
Sense normal Column mm ID * mm
Resolution 1.4 nm Packing nc terial
Time range 1.2 --- 3.2 rain Mobile phase
Interval .82 sec Plow rate ml ’min
Baseline OFF Pressure
Smoothing 7 points Slope .5 AU/mm
Drift .002 AU’min Height .001 AU
Width .02 mm Mm. area .0001 AU*jajn
Time double 30 rain Minus peak OFF
___ .. .. ,
: ____

. . .
0 .2
Integrator LC724.DT3
14:09:27 Sample name
44 asec *16 Baseline
normal Resolution
1. 1 AU’FS Paper speed
3.2 mm Interval
7 points Slope
.002 AU’min Height
.02 mm Mm. area
30 mm Minus peak
u ID * mm Packing sateria
Flow rate
lppm std ::i
1.4 nm
20 mm/mm
.82 sec
.5 AU/mm
.001 AU
.0001 AU*min
— 1 ——— 1 AU

Waters 991
Samp .ing time
Y-sca le
Time range 1.2
Time double
Mobile phase
Report File LC724.DT3
mi/mi n
207 ma
No. Retention Height Left Right Area Area Mark
time (AUI time time [ AU*minl [ ZI
1 1.79 0.3512 1.69 1.98 0.024254 24.299 I
2 2.14 0.4453 1.98 2.39 0.039654 39.728 I
3 2.54 0.3295 2.39 3.01 0.035905 35.973 I

LC724. DT3
Saupling tiae
44 sec * 16
Y - Scale
.627 AU’PS
1.4 n
SaRple naue
ippi std
24—D 1.79 ain
2,45—T.1.5..J . .n.
8ILVR1 2.55 m i i i
Co I u•n
amID * am
Packing material
Mobile phase
Plow rate
al /mi n
W t€, 1 991
200 210 220 230 240
200 210 220 230 240
Wavelength 200 --- 240.6 na

W t rs 991 C ntai r pl t
LC724.DT3 05—06-1992 14:09:27 Sample name lppm std
Sampling time 44 msec *16 Paper speed 100 nim’mjn
Y—scale .01 --- 1 AU Baseline OFF
step .1 AU Column mm ID * mm
Wavelen th 200 --— 240.6 nm Packing material
Resolution 1.4 mm Mobile phase
Time range 1.2 —-- 3.2 mm Flow rate m I /mm
Interval 1 sec Pressure
210 . 230 2 0 210 200
2,4 -D ___________ ____
__ _
2 ,4,54 :
KI _
— i ....— .— . . .l .— . . .—
240 230 220 210 200
240. 6 ——— 200 rtm

x Silvex I
Amount (ng)
— 2,4-D A 2,4,5-T

Table 1
Waste Matrices
Spike level
Replicate 1
Replicate 2
Replicate 3
Replicate 4
Replicate 5
Replicate 6
Replicate 7
Replicate 8
186 140 195
191 137 218
211 136 228
194 143 222
198 144 221
189 139 224
191 142 224
204 143 228
203 145 228
199 185 198
200 208 191
212 216 205
208 214 200
205 212 200
195 209 193
205 205 193
192 200 185
201 209 193
Av % Rec
Std. Dev
198 141 224
106 101 115
7.78 3.36 3.72
3.94 2.38 1.66
202 209 195
102 113 98
6.63 5.08 6.30
3.28 2.43 3.23
Silvex 2,4-D
(ppm) (ppm)
Spike level
Replicate 1
Replicate 2
Replicate 3
Replicate 4
Replicate 5
Replicate 6
Replicate 7
Replicate 8
Average —
Note: All concentrations reported as acid equivalents of esters.
(a) Isopropyl ester used.
(b) Isooctyl ester used.
Methyl esters used for all others
(c) 2.5g sample + 2.5g hexadecane
2,4-D (a) 2,4,5-T (b) Silvex
(ppm) (ppm) (ppm)
2,4-D 2,4,5-T Silvex
(ppm) (ppm) (ppm)

Table 2
Waste Matrices
PCB contam. solvent
PCB contam. lab liquids
PCB contam. flamm liquid
PCB contam. water & oil
Isopropyl, isooctyl, methyl esters
Methyl esters
Methyl esters
Free acids
Ash composite
Brick/slag composite
Ethyl, n-butyl, methyl esters
Isopropyl, n-butyl, ethyl hexyl esters
Methyl esters
Dry scrubber solids
Dry scrubber solids
98 99
98 101
Isopropyl, n-butyl, ethyl hexyl esters
Ethyl, n-butyl, methyl esters
Isopropyl, n-butyl, ethyl hexyl esters
PCB oil
PCB oil
PCB transformer oil
Dielectric fluid/mineral oil
Oil bottoms/waste oil
Mineral oil with PCB
Methyl esters
Methyl esters
Methyl esters
Methyl esters
Methyl esters, 23ppm 2,4,5-T in samp
Free acids
Pheromone scrap
Electrostatic Iiq developer
Solv contam solid/sludge
Holding tank film
Methyl esters
Methyl esters
Methy’ eslers.2l2ppm 2,4.5-T in samp
Free acids
Free acids _________________
Note: Samples were spiked at approximately 200ppm

Table 3
Extraction Fluid # I Extraction Fluid # 2
Sample Amt = lOmL Sample Arnt = lOmL
2,4-D 2,4,5-1 Silvex 2,4-D 2,4,5-T Silvex
( ppm) (ppm) (ppm) ( ppm) (ppm) (ppm )
Spike level 0.95 0.94 0.98 0.95 0.94 0.98
Jnspilced ND ND ND ND ND ND
Replicate 1 0.96 0.99 1.05 0.98 0.99 1.03
Replicate 2 0.96 0.98 1.05 1.01 1.01 1.07
Replicate 3 0.95 0.99 1.05 0.97 0.99 1.05
Replicate 4 0.95 0.98 1.04 0.97 1.00 1.03
Replicate 5 0.97 0.99 1.05 0.98 1.01 1.04
Replicate 6 0.95 0.99 1.04 0.95 0.99 1.04
Replicate 7 0.95 0.99 1.04 0.96 0.98 1.03
Av % Rec
Std. Dev
Note: All concentrations reported as acid equivalents of esters.

Table 4
TCLP Sample Matrices
S2mple Type Percent Recovery . Comments
2,4- i) 24,5-1 ( 24,5 -TP
Drill cuttings 102 104 106
Sand and paint chips 102 106 107
Used fuel filters 85 108 105
Off spec lamp component 105 104 105
Shot gun blast 102 103 99
Waste sludge 103 106 107
Dock sludge 98 98 104
WWT sludge 100 M I 106 Original sample contamination
Paint sludge 99 107 104
Acetone contain, soil 100 106 107
Demolition debris 98 107 104
Contain. soil & concrete 95 114 104
Asphalt, dirt, and sand 96 108 100
Mercury contam. debris 91 103 104
Activated carbon 100 103 105
Furnace ash 92 102 103
MSD carbon 100 105 103
Carbon fine dust 93 96 106
Filter cake 79 MI 100 Matrix interference
Note: samples spiked with approximately lppm methyl esters

Taininv L. Jones , Research Chemist, U.S. Environmental Protection
Agency, EMSL-LV, Las Vegas, Nevada 89119, and Lopez-Avila V.,
MidWest Research Institute, California Operations, Mountain View,
California, 94043.
Substantial numbers of hazardous compounds are not currently
regulated by the U.S. Environmental Protection Agency (U.S. EPA)
simply because there are rio available or reliable methods to
quantitatively measure them. These compounds pose difficulty when
analysis is done by conventional analytical methods (e.g., gas
chromatgraphy) because they can be non-volatile (polar or high
molecular weight) or thermally labile. Carbamates are a class of
such compounds, and they include some of the most widely used
pesticides in agriculture. Methods for carbamate determination are
necessary for support of the Environmental Defense Fund decree (ED?
vs. Reilly, Civ. No. 89—0598), as supervised by the Office of Solid
Waste (OSW). The ED? consent decree mandates that EPA, under the
Resource Conservation and Recovery Act (RCRA), develop methods for
and list the carbamates found in industrial waste streams.
Therefore, it is important that reliable and rugged analytical
methods be developed for the analysis of this class of compounds.
A single-laboratory evaluation of a therinospray—liquid
chromatography/mass spectrometry (T LC/MS) method was previously
developed for the U.S. EPA. The int laboratory calibration study
reported here was undertaken as part of the support for the OSW
method development process and for support of the EDF consent
Nine carbamate pesticides (aldicarb, CAS# 116-06-3;
bendiocarb, CAS# 22781—23—3; carbaryl, CASt 63—25—2; carbendazim,
CAS# 10605—21—7; carbofuran, CASt 1563—66—2; diuron, CAS# 330—54—1;
linuron, CASt 330—55—2; methomyl, CASt 16752—77—5; and oxamyl,
CASt 23135—22—0) were selected for evaluation in this calibration
study. Seven of the analytes (aldicarb, bendiocarb, carbofuran,
oxauiyl, methomyl, carberidizam, and linuron) were chosen from the
target analyte list provided in the consent decree and the other
two carbainates are of importance to the State of California.
The main goal of this study was to obtain a preliminary
assessment of the ability of the TS-LC/MS to reliably detect and
quantitate carbamate pesticides. The purpose of this work will be
to collate and statistically evaluate the data collected during the
interlaboratory calibration study.

Norman A. Kirshen , Environmental Program Manager and Elizabeth B. Almasi, GC/MS Chemist,
Varian Chromatography Systems, 2700 Mitchell Drive, Walnut Creek, California 94598
Trace level Volatile Organic Compounds (VOCs) in ambient air are normally determined according to EPA
Method TO-14. This method describes the analysis in ambient air of 41 VOCs, ranging in boiling point
from -29 to 215°C. It covers a concentration range from 0.2 to 20 parts per billion, volume/volume (ppb),
and specifies sample enrichment with a 400 mL air sample on glass beads at .160°C. While this sample
volume provides sub-ppb levels of VOC detection for target analytes when using a quadrupole mass
spectrometer detector in SIM mode or when using CC detectors, the identification of non-target analytes
may only be done in full scan mode for higher concentration. Also with this sample volume a Nafion dryer
is needed for water removal thereby lowering the recovery of polar VOCs.
Because of the very high senstivity of the ion.trap MS, relatively small volumes (60 mL) are adequate to
obtain the required or lower detection levels. An integrated air analysis system based on a CC/Ion-trap
MS has been investigated and is described. This system has abuilt.in cryogenic trap and necessary valving,
internal standard gas sampling valve loop, and is controlled from the GC/MS workstation. The linearity,
precision, and method detection levels obtainable with this system when using small volumes are reported.
In addition, examples of the quantitative and qualitative analysis of ambient air samples are shown.
The determination of basic air pollutants in ambient air is of paramount importance as legislative acts,
such as the 1990 amendments to the Clean Air Act (CAA) of the United States, take effect. Federal, state
and local actions will ultimately reduce emissions from industrial and mobile sources to meet the
requirements of the CAA. The analytical techniques which are used to ensure that allowed emissions are
not exceeded must provide sensitive and definitive measurements of volatile organic compounds (VOCs)
in ambient air at the sub parts per billion volume/volume (ppb) level.
The United States was quick to initiate experimental guidelines for VOC analysis in air. The resulting
EPA method TO-14,’ 5 is the most commonly used method for VOC analysis worldwide and therefore it
has been used as a guideline for the following study.
Method TO-14 describes the analysis in ambient air of4l VOCs, ranging in boiling point from .29 to 215°C
(Table 1). It covers a concentration range from 0.2 to 20 ppb, specifies sample enrichment (400 mL) on
glass beads at -160°C, thermal desorption, separation on a capillary column, and detection with a mass
spectrometric detector. The first draft of the Contract Laboratory Program (CLP) method 6 was published
in February 1991. The samples to be analyzed by the CLP method are from known or suspected hazardous
waste sites, therefore the concentration range is from 2 to 100 ppb, higher than required for ambient air
Previous work with TO-14 systems based on CC detectors 7 has confirmed that volumes of approximately
400 mL are required to obtain sensitivities of 0.2 ppb. The same requirements apply to quadrupole mass
spectrometers. Because of the very high sensitivity of the ion-trap MS, relatively small air volumes (60 mL)
are required to obtain these or lower detection levels. An integrated air/soil gas analysis system based on
an CC/Ion-Trap MS has been investigated and is described here. This system has a built-in cryogenic trap,
internal standard gas sampling valve loop, sixteen sample automation and is controlled from the GC/MS
workstation. The linearity, precision, and method detection levels obtainable with this system when using
small volumes are reported. In addition, examples of the quantitative and qualitative analysis of ambient
air samples are shown.

Table 1 Quantitation Ions, Retention Times, %RSD and Method Detection Limits
for Analytes in Method TO-14.
Quan RT* %RSD** MDL
Compound Ion (mm) (area) (ppb)
Dichlorodifluoromethane 85 13:05 3.8 0.01
Chloromethane 50 14:11 8.5 0.03
1,2-Dichloro -1,1,2,2-tetrafluoroethane 85 15:11 3.5 0.01
Vinyl Chloride 62 15:30 6.0 0.02
Bromomethane 94 16:56 4.7 0.01
Chloroethane 49 17:36 9.0 0.03
Trichlorofluoromethane 101 19:23 3.2 0.01
1,1-Dichloroethylene 61 20:25 5.6 0.02
Dichloromethane 49 20:42 3.9 0.01
1,1,2-Trichloro-1,2,2-trifluoroethane 101 21:07 3.8 0.01
1,1 Dichloroethane 63 22:10 4.8 0.01
c-1,2-Dichloroethene 61 23:08 4.4 0.01
Chloroform 83 23:28 3.7 0.01
1,2-Dichloroethane 62 24:14 4.1 0.01
1,1,1-Trichioroethane 97 24:30 4.6 0.01
Benzene 78 24:59 3.4 0.01
Carbon Tetrachloride 117 25:08 3.4 0.01
1,2-Dichloropropane 63 25:50 2.9 0.01
Trichloroethene 130 26:05 3.8 0.01
c-1,3-Dichloropropene 75 26:59 4.7 0.01
t-1,3-Dichloropropene 75 27:32 5.7 0.02
1,1,2-Trich loroethane 97 27:43 3.9 0.01
Toluene 91 28:03 2.4 0.01
1,2-Dibromoethane 107 28:47 2.9 0.01
Tetrachloroethene 166 29:19 3.5 0.01
Chlorobenzene 112 30:06 3.8 0.01
Ethylbenzene 91 30:33 4.6 0.01
in,p-Xylene 91 30:47 2.9 0.01
Styrene 104 31:12 5.2 0.02
1,1,2,2-Tetrachloroethane 83 31:19 4.7 0.01
o-Xylene 91 31:21 5.0 0.02
4-Ethyltoluene 105 33:02 7.0 0.02
1,3,5-Trimethylbenzene 105 33:09 8.9 0.02
Benzylchloride 91 33:15 10.1 0.03
1,2,4-Trimethy1benzene 105 33:45 10.3 0.03
in-Dichlorobenzene 146 33:58 3.2 0.01
p-Dichlorobenzene 146 34:05 4.3 0.01
o-Dichlorobenzene 146 34:37 4.8 0.01
1,2,4-Trichlorobenzene 180 37:56 9.3 0.03
Hexach lorobutadiene 225 39:11 8.0 0.03
*RT includes the concentration step also, column DB-1
**% D calculated from area responses of 9 replicate runs.

System Description
The schematic of the GC/Ion-trap MS system is shown in Figure 1. The built-in trapping and
preconcentrating device, the Variable Temperature Adsorption Trap (VTAT, Figure 2) is capable of
trapping and preconcentating VOCs from air on glass beads at -160°C or on an adsorbent such as
Carbosieve ’/Carbotrap at ambient temperatures. In the present study only the subambient mode was
Samp’e 1 .... Samp’e 18
Coolant Input
Trap Column
Cryotrap: 2 in. glass beads
in stainless steel, 1/8 in.
j Heater Block
Figure 2. The variable temperature absorption trap (VTAT).
Flow Controfler
Ca r rier Gas
Sahzn GC S
Vent or Vacuum
Figure L Schematic of a GCIIon-trap MS System for VOCs.

Instrumentation and Conditions
Cryogenic concentrator:
• Variable Temperature Adsorption Trap (VTAT), 5 cm of 60/80 mesh
silanized glass beads
• Two automated valves, 4- and 10-port; capable of sample and internal standard
(1.5 ) introduction
• Electronic mass flow controller, 0-100 mL/min, with readout box
• Vacuum pump (metal diaphragm)
Air sample flow rate: 20 mL/min
Column flow rate: 1 mL/min He
Auxiliary flow rate: 20 mLlmin He
VTAT: -160°C for 4 mm, 180°C/mm to 120°C, hold
Valves: 160°C
Column: -50°C for 6 mm, 8°C/mm to 160°C, hold
1DB-i (J&W), 60m x 0.32 mml.D., 1 ji.m film or
DB-624 (J&W), 60m x 0.32 mm I.D., 1.8 im film
Ion-trap MS (Varian Saturn II):
Scan Range: 47-260 u
Scan Rate: 0.8 seeJscan (3 tscan/analytical scan)
RF storage Level: 210 DAC Steps; background Mass: 45 u
Segment Breaks: 70/78/150; Tune factors: 120/70/100/70
Automatic Gain Control (AGC) Target: 20000
Emission Current: 30 .i.A (Optimized parameters might vary instrument to instrument)
Alphagaz TO-14 standard, 41 component, 2 ppm
In Method TO-14 a critical part of the analysis is the preconcentration step. In the first stage of this
enrichment process the sample (generally VOCs present in low or sub ppb concentrations) is flushed
through the lines with a flow set by the electronic mass flow controller, while the loop (0.25 mL) is filled
by the internal standard (if required). After the initial column and VTAT temperatures are equilibrated,
the air sample and internal standard are directed to the -160°C VTAT and the VOCs are deposited onto
the glass beads.
The duration of this “trapping” time can be varied and the volume of the analyzed sample changed
accordingly. The sample flow during this step, usually 20 mL/min is held constant by the mass flow
controller. In this study the trapping time was 3 minutes resulting in a 60 mL sampled volume. After the
sample VOCs are deposited, the residual air is removed from the VTAT by the auxiliary flow. Then the
VTATis heated to 120°C and the analytes are backflushed to the capillary column where they are focussed,
separated, and detected. Later the VTAT is cooled down in preparation for the next analysis.

The main difference between the experimental parameters used in this work and those specified in the
TO-14 method is the sample size. The method specifies a sample volume of 400 rnL. This volume of air can
introduce a significant amount of water that might either plug the VTAT or capillary column. Elimination
of this residual water is possible with a semipermeable membrane dryer such as a Naflon dryer. The
removal of water with this type of dryer results in the loss of any trace polar organics that might be in the
sample. The sensitivity of the CC/Jon-trap MS allows trace level VOC detection by preconcentrating only
60 mL of sample. This small sample reduces the interference of water and eliminates the need for a Naflon
The linearity, precision, and method detection limits (MDL) were examined and real samples were
analyzed. Before analysis, blank runs were performed. Very often even good quality compressed air has
impurities. The Reconstructed Total Ion Current chromatogram (RTICC) of a blank and the accompanying
data file is shown in Figure 3. Only trace VOCs of approximately 0.2 ppb or less were found.
The standard and samples were introduced to the system from stainless steel SUMMA® polished canisters.
The standard used was a 41 component, 2 ppm VOC mixture (Alphagaz) diluted with air to the desired
concentrations. RTICCs of 2 ppb and 0.25 ppb v/v standards are shown in Figures 4 and 5. Gaussian peak
shapes are exhibited by all the compounds including the agases (the six most volatile compounds) as shown
by their mass chromatograms in the Figure 4 insert. For the quantitation of the gases a peak smoothing
algorithm was used, allowing precise quantitation of these components even at low concentrations.
The precision and MDL were determined by multiple ix jections of a 60 mL, 0.1 ppb standard. Standard
deviations of the single ion areas were calculated for nine runs and were between 2.9%, the average of the
41 compounds being 5%, Table 1.
Mae of Compaun&
Fit S/N
S Flee
Caic .t(A)
d ich Iaro&if luorceethae
l v i
0 • 183
Al 19 ichior Id.
Fr iclilurof luoroeet],an.
u N
EtJy1 . .r.
1. 1-DlcbloroeUiylene
O—DicJ iIorobe en.
1 ,3.5-Trie.tjlL .. .. . .... . .
— I
• I
Figure 3. RTICC and result ifie of a blank (pure air sampled) run.
NP indicates target compounds not found (below minimum spectral fit value).
Sorted via: Fit S’7I I

• .
62 J H 2 CC1H F
9 i t1 B 1 1.3
IØI 12: I32 14 4e 16 e i?
- I 1 - — I
680 1200 1080 2400 3080
16:08 24:86 32:86 46:86
Figure 4. RTICC of 41 VOC compounds, 60 mL, 2 ppb v/v
and mass chromatogram of the gases.
___ — J i
1260 1860 2480 3808
0:80 16: 138 24:68 32:66 48:66
Figure 5. RTICC of 41 VOC compounds, 60 mL, 0.25 ppb v/v

The MDL was calculated from integrated areas of single quantitation ions (nine replicate runs) with the
following formula:
MDL= s x t
where s is the standard deviation of the replicate analyses and t is the student’s t value appropriate for a
99% confidence level and a standard deviation estimate with n-i degrees of freedom. The calculated MDLs
were between 0.01-0.03 ppb.
Linearities of the quantitation ion responses versus concentration for the 41 components were examined
over the range required in the method, 0.1 to 20 ppb v/v. and were found to be very good. Representative
linearity plots are shown in Figures 6a and 6b.
In addition to identifying and quantitating target components in a sample it is often necessary to identify
and estimate the quantities of non-target analytes. For example, dibromochloromethane, a non-target
analyte is identified and its concentration estimated at 2 ppb in Figure 7a.
In ambient air some components are present at much higher concentrations than the VOCs. The two most
significant components which are concentrated together with the VOCs from the air are water (mentioned
above) and CO 2 . The reduced sample volumes used here suppress the problems caused by these
components. For example, to represent a very humid sample, an air sample was collected just above the
surface of a 60°C water bath. At this temperature the vapor pressure of water is 0.2 atmospheres. The
chromatogram and results shown in Figures 7a and 7b indicate that the preconcentration process was not
affected by the high level of moisture.
Carbon dioxide which is also present in air at high concentrations can be eliminated as an interference by
choosing the scanning range from 47 to 250 u and setting the background mass at 45 u. Then CO 2 (44 u)
is not stored or detected by the Saturn mass spectrometer and the detection of the early eluting VOCs is
Two sample applications are shown using the same conditions. The first sample shows a chromatogram
and the resu]ting report from ambient air collected in Walnut Creek, California on a rainy day in rush
hour traffic (Figures 8a and 8b). The aromatics which are the major components of exhaust gas emissions
found under these conditions are evident. The second sample was collected at an industrial site to screen
for several polar organics. The RTICC and the mass chromatograms at 31 and 45, characteristic mass ions
used to quantitate methanol and ethanol, respectively are shown in Figure 9.
An integrated air/soil gas analysis system based on a GCfIon-trap MS has been investigated and applied
to the analysis of VOCs following EPA Method TO-14. The very high sensitivity of the ion-trap MS allows
the use of relatively small air volumes (60 inL) to obtain both qualitative confirmation (full scan spectra)
and quantitative determination of sub ppb levels of VOCS. MDLs of 0.01-0.03 ppb have been calculated
from multiple runs at 0.1 ppb.
Since water interference is minimized using this small air volume, the use of Nafion dryers has been
eliminated allowing the determination of polar as well as non polar organic compounds.

(Peak Area of Sanpie) us (Anount of Sample Injected)
5.880 18.888
Figure Ga. Linearity of Bromomethane, 0.1-20 ppb.
(Peak Area of Sample) us (Ascent of Sanpie Injected)
5.880 18.088
Figure Gb. Linearity of I2,4-Triniethylbenzene, 0.1-20 ppb.

- - .
180 D ibroriocbloromethal)e
I library
I L 1 I • I • I p .
48 68 88 180 128 148 168 1.88
T OT Formula: C.H.Cl.8r2 FIt Rank 1 Index 1.56
ThItfi 1
888 1288 1680 2608 248* 1 2880
18:40 16:88 zi:28 26:48 32:60 37:28
Figure 7a. . RTICC of 60 inL air sample collected above the surface of a 60°C water bath
dibromochioromethane, a non-target analyte is identified (fit 90311000),
estimated concentration 2 ppb.
Sorted u 1a Ca lc Amount(A) 4 _____________________ ____________________
Cal Name o Compound Fit S.’N R Time Me Caic Amt(A) Units
23 Toluene 982 2755 ‘JR 5.22$ PPBA)
15 1,1.1—Trinklorceti ane 829 24:26 DV 1.861 PPB’V
36 Benz9ldlIoride 949 33:51 W 0.793 PPB,’V
2 CblorossetMne 821 13:59 BB 8.655 PPB/V
16 Benzene 983 24:54 IA ’ 8.685 PPBAJ
33 1,3.5—Trimethylbenzei1e 98? 33:39 IA’ 8.593 PPBAJ
34 12,4—TrimeThylbenzene 989 33:39 IA’ 0.587 PPB/V
39 1,2,4—TricMoob.iir.ime 937 37:51 RB 8.518 PPB V
9 D ic h loromethane 941 28:48 B’) 8.499 PPB/V
31 o—Xy lane 905 31:15 DV 0.458 PPB/U
29 St9rene 990 31:06 BB 8.362 PPBS1)
2? Et]sjThenzeoe 983 38:2? VB 8.347 PPEA
35 m—D IcMorobenzene 939 33:53 DV 8.384 PPBIU
28 rn p—X9 lena 90? 31:15 DV 0.252 PPB/U
7 Triclilorofluozomethane 994 19:21 DV 6.248 PPB”V
21 t—i,3—Dichloz -opropene 776 27:26 IA ’ 8.243 PPB.’V
S Bromomethane 984 16:53 RB 8.21? PPB’V
13 Chloroform 971 23:26 DV 6.194 PPBAI
16 1,1,Z—Tric]iloru-122— 979 21:66 BB 0.165 PPB”U
1 dichlorodifluorometban 93? 12:57 88 8.149 PPB.’V
26 C]ilorobenzene 877 29:59 RB 8.143 PPB/U
32 4—Etliyltoluene 985 33:19 DI) 8.143 PPB/U
25 Tetrachioroethene 764 29:2? BB 8.133 FFB/V
24 EDB 850 28:48 B’) 8.879 PPR/V
22 1,12—Trlckloroethane 794 27:19 RB 8.871 PPB/V
14 1,2-Dichloroethane 830 24:18 DV 8.643 PPB .’l)
Figure Th. Quantitation Report of the Sample Shown in Figure 7a.

Cal Marie of Compound F it Sili R Time ___________
23 Toluene 993 27:24 VB
20 m,p—Xylene 995 38:16 ‘A ’
31 c-Xylene 989 38:46 RU
33 1,3,5—Trimeth!jlbenzene 994 33:14 BR
32 4—Eth jitoluene 995 32:26 AJ
27 Ethy lbenzene 993 29:57 ‘JR
16 enzene 978 24:28 BE
36 Benzy ich lox’ ide 898 33:51 VU
29 Styrene 729 38:46 P 11
25 Tetrach loroethene 913 28 :4 P11
34 1,2,4—Trimethylbenzene 995 33:57 P11
22 1,1,2—Triclilornethane 787 28:63 P1 1
14 1, 2—D Ich loroethane 758 24:20 MM
Figure 8b. Quantitation Report of the Sample Shown in Figure 8a.
. j t _J
Figure 8a. RTICC of 60 mL air sample collected in Walnut Creek, California
on a rainy day in heavy traffic.
Caic ( t(A> Units
5.927 PPB/V
3.165 PPB/V
2.25? PFB/V
2.151 PPB/U
1.584 PPBAI
1.455 PPB/I)
6.981 PPBIIJ
6.689 PPB/V
8.356 PPBAJ
8.288 PPBAI
6.222 PPB”U
8.175 PPB’U
6.152 PPBAI

1. Compendium Method TO-14, The Determination of Volatile Organic Compounds (VOCs) in
Ambient Air Using STJMMA Passivated Canister Sampling and Gas Chromatography Analysis,
U.S. Environmental Protection Agency, Research Triangle Park, North Carolina 27711, May 1988.
2. K.D. Oliver, J.D. Pleil, and Wit. McClenny, Sample Integrity of Trace Level Volatile Organic
Compounds in Ambient Air Stored in SUMMA Polished Canisters, Atmospheric Environ. 20:1403,
3. M.W. Iloidren and D i. Smith, Stability of Volatile Organic Compounds While Stored in SUMMA
Polished Steel Canisters, Final Report, EPA Contract No. 68-02-4127, Research Triangle Park,
North Carolina, 1983.
4. WA McClenny, J.D. Pleil, J.W. Holdren, and R.N. Smith, Automated Cryogenic Preconcentration
and Gas Chromatograph Determination of Volatile Organic Compounds, Anal. Chem. 56:2947,
5. Wit. McClenny, et. aL, Canister-based Method for Monitoring Toxic VOCs in Ambient Air, J. Air
Waste Management Association, 41, No. 10, 1308, 1991.
6. Analytical Method for the Determination of Volatile Organic Compounds (VOCs) in Air Collected in
Canisters and Analyzed by Gas Chromatography! Mass Spectrometry (GC/MS), Exhibit D, Chapter
1, Part IA, Contract Laboratory Program, February 1991.
7. Elizabeth Almasi and Norman Kirshen, The Analysis of Volatile Organic Compounds in Air,
Variable Volume System, Varian GC Application Notes 19 and 32, 1989 and 1990, respectively.
za:ee 32:08
Figure 9. RTICC and characteristic ions for methanol and ethanol from a
60 mL air sample collected at an industrial site.
Courtesy of Air Toxics, Ltd., Rancho Cordova, California.

Viorica Lopez-Avila , N. S. Dodhiwala, and J. Benedicto, Midwest Research Institute— California
Operations, Mountain View, California 94043, and W. F. Beckert, U.S. Environmental
Protection Agency, EMSL-LV, Las Vegas, Nevada 89119.
A rapid and efficient method was developed for extracting chlorophenoxy acid herbicides from
soils using supercritical carbon dioxide, tetrabutylammonium hydroxide, and methyl iodide. The
extraction is carried out at 400 atmI8O°C/15 mm static, followed by 15 mm dynamic at a carbon
dioxide flowrate of approximately 3 mLlmin. The use of other derivatization agents
(trimethyiphenylammonium hydroxide, benzyltrimethylammonium chloride, and
benzyltriethylammonium chloride) proved to be less effective than the tetrabutylammonium
hydroxide/methyl iodide combination. Attempts made to extract other compounds currently listed
in Method 8151 using supercritical carbon dioxide and tetrabutylammonium hydroxide/methyl
iodide were unsuccessful, either because these compounds did not derivatize (dalapon, dinoseb,
and 4-nitrophenol), or they were found to decompose (DCPA diacid, pentachiorophenol, and
picloram), apparently in the injection port of the gas chromatograph, despite the fact that they
were reported to be amenable to gas chromatography.
Although supercritical fluid extraction (SFE) is becoming more and more recognized as an
efficient and rapid alternative to Soxhiet extraction, the reports published so far that deal with
extraction of samples by SFE are addressing mostly nonpolar organic compounds such as
polynuclear aromatic hydrocarbons, polychlorinated biphenyls, polychiorinated dibenzodioxins,
etc. Recently, Hawthorne and coworkers (1) reported the extraction and methylation of 2,4-D
and dicamba from stream sediment samples using supercritical fluid derivatization/extraction with
trimethyiphenylammonium hydroxide (TMPA) or BF 3 /methanol as derivatization agents. While
the derivatization with TMPA is known to require high temperatures, Hawthorne and coworkers
claimed that the derivatization reaction of 2,4-D takes place during SFE with carbon dioxide at
400 atm/80°C. Their conclusion was based on the fact that increasing the concentration of
TMPA yielded better recoveries of 2,4-D, while adding more TMPA to the extracts after SFE
did not. The derivatization with BF 3 fmethanol under supercritical conditions was demonstrated
in the case of 2,4-D; however, dicamba was not methylated under these conditions (1).
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.

The purpose of our study was to develop an SFE method for extracting chlorophenoxy acid
herbicides (Table 1) from soils using supercritical carbon dioxide. It was desirable that the
extraction and the derivatization reaction be conducted simultaneously since the chlorophenoxy
acid herbicides need to be converted to their corresponding methyl esters anyway in order to be
analyzed by gas chromatography.
Several derivatization agents, e.g., trimethylphenylammonium hydroxide (TMPA),
benzyltrimethylammonium chloride (BTMAC), benzyltriethylaminonium chloride (BTEAC), and
tetrabutylammonium hydroxide/methyl iodide (TBA/MI) were investigated. The extractions were
carried out with supercritical carbon dioxide alone and with supercritical carbon dioxide modified
with 10 percent methanol. Attempts were also made to extract other compounds currently listed
in Method 8151 using supercritical carbon dioxide and tetrabutylammonium hydroxide (25
percent in methanol)/methyl iodide.
The nine chlorophenoxy acid herbicides (Table 1), except dicamba, were purchased from Aldrich
Chemical (Milwaukee, WI). Dicamba was purchased from Chem Service (West Chester, PA).
The corresponding methyl esters were purchased from Crescent Chemical (Hauppauge, NY).
All compounds were used as received without further purification (their purities were stated to
be at least 98 percent, except MCPA at 95 percent and 2,4-DB at 97 percent). Stock solutions
of the individual acids or esters were prepared in methanol at 10 mg/mL and kept at 4°C in the
dark. A spiking solution of the chiorophenoxy acids was made by combining the individual stock
solutions and diluting them to 1 mg/mL.
Derivatizing agents were purchased as solutions in methanol or as neat materials. TMPA
(1.5 percent in methanol) was purchased from Eastman Kodak Company (Rochester, NY) and
Fluka Chemical (Ronkonkoma, NY). To prepare TMPA in methanol, ethanol, and isopropanol,
we evaporated 20 mL of 1.5-percent TMPA in methanol to dryness under a gentle stream of
nitrogen, and then dissolved the residue in 3 mL methanol, ethanol, or isopropanol. BTMAC
and BTEAC (purity 98 percent) were purchased as neat materials from Fluka Chemical. Forty
grams of each were dissolved separately in 100 mL methanol for use in the experiments. TBA
(25-percent solution in methanol) and MI (neat) were purchased from Fluka Chemical.
The extractions were performed using an Isco (Lincoln, NE) SFE System 1200 with an Isco
Model 260D syringe and an Isco Model SFX 2-10 extraction module. The extraction module has
shutoff valves at the inlet and outlet of the extraction vessel, allowing static and dynamic
extractions to be performed. The SFE system was equipped with two fused-silica restrictors
(37-cm length x 50-nm ID), which resulted in flowrates of approximately 2.5 mL/min (as liquid
carbon dioxide). All extractions were carried out at 400 atm; the extraction temperature was
varied (80°C, 100°C), and the static and dynamic times were also varied (from 5 to 15 mm), as
indicated in the tables with results. The extracted analytes were collected in 2 mL methanol.
Initially, some of the extracts were concentrated to 1 mL by blowdown evaporation using a
gentle stream of nitrogen. This, however, resulted in poor recoveries of the test analytes, and
the blowdown evaporation was therefore not used in subsequent experiments.

All extracts were analyzed by GC/MS using the operating conditions given in Table 2. Attempts
were made to analyze the extracts by GC with electron capture detection (ECD); however, the
presence of excess derivatization reagent (MI) in the extracts interfered with the GCIECD
determination. Quantification of the analytes was performed using internal standard calibration.
The two internal standards used were acenaphthene-d 10 and phenanthrene-d 10 (Table 3).
Extraction without derivatizing agent
Preliminary experiments to recover the underivatized target compounds from sand samples
were carried Out with supercritical carbon dioxide, but the recoveries were low and difficult to
reproduce. The recoveries ranged from 0 to 65 percent, when we extracted two 2-g samples,
spiked with the chlorophenoxy acid herbicides at 2,500 j g/g in parallel, at 150 atm for
10 minutes, 200 atm for another 10 minutes, and 250 atm for an additional 10 minutes (the
temperature of the extraction was maintained at 70°C, and the extraction was performed
dynamically using a 30-cm length x 50-jtm ID fused-silica restrictor heated at 100°C). In
addition, the recoveries were not reproducible. For example, the recovery of 2,4-DB was
37 percent in one experiment, but only 7.1 percent in the other. In the case of 2,4,5-TP, the
recovery was 54 percent in one experiment, but only 6.7 percent in the other. Other researchers
(3) have reported similar results. Thus, we focused our study on finding conditions under which
these chlorophenoxy acid herbicides can be extracted from a solid matrix and, at the same time,
are derivatized to their corresponding methyl esters.
Extraction in the presence of selected derivatizin2 agents
Solutions of BTMAC and BTEAC in methanol were selected as ion-pair...methylating reagents
based on results by Chiang and coworkers (3,4), who reported that the chlorophenoxy acid
herbicides can be extracted from soil samples and esterified in situ to their corresponding methyl
esters using BTMAC. The authors further claimed that the pH of the BTMAC solution (4.8)
appeared to be beneficial to the formation and stabilization of the methyl esters.
TMPA in methanol was selected as ion-pair methylating reagent based on results by
Hawthorne and coworkers (1) who reported that 2,4-D and dicamba were extracted from a soil
sample as methyl esters (the extraction was carried Out at 400 atm/80°C/5 to 45 minutes static,
followed by 5 to 15 minutes dynamic, and the amount of reagent per 2-g samples spiked at
20-ppm levels varied from 50 to 1,000 DL).
Our data (Table 4) indicate low recoveries of the derivatized chiorophenoxy acid herbicides
from sand when BTMAC and BTEAC were used as derivatization reagents, regardless of the
extraction temperature, and mostly good recoveries with TMPA. When we used BTMAC or
BTEAC, we detected not only the methyl esters but also the benzyl esters and ethyl esters,
Additional experiments were performed with TMPA dissolved in methanol, ethanol, and
isopropanol to determine whether the methylation reagent is TMPA or the alcohol in which

TMPA is dissolved. The recoveries are given in Table 5. In all cases, we found the methyl
esters and only traces of ethyl or isopropyl esters.
Extraction in the presence of TBAIMI
Tetrabutylammonium hydroxide and MI were selected as derivatizing agents because they were
successfully used by Hopper (5) to derivatize five chiorophenoxy acid herbicides (2,4-D; 2,4,5-T;
2,4-DB; 2,4,5-TB; and 2,3,6-TBA) and pentachiorophenol in foods. After preliminary work with
the TBA/MI reagent, we concluded that TBA acts as an ion-pair reagent which assists in the
transfer of the acid from the solid matrix into the stream of carbon dioxide. TBA and MI are
dissolved by the carbon dioxide and transported into the collection solvent, which in our case was
methanol. Upon injection of the extract into the GC/MS system, the acids are obviously
converted to their corresponding methyl esters. The minimum injector temperature at which the
methylation reaction occurs is 100°C. To determine the reproducibility of the methylation
reaction, we prepared 10 working standards of nine chlorophenoxy acids in methanol, added TBA
and MI (nominal concentrations of the test compounds were 100 ngI L), and injected these
solutions into our GCIMS system (injector temperature was held at 250°C). The derivatization
yields (Table 6) were determined by comparing the instrument responses for these 10 injections
with those obtained by analyzing a composite standard containing the nine methyl esters. The
reproducibility of the methylation reaction was 8.6 percent or better, except for dicamba at
16.6 percent, and most compounds had percent RSDs between 7 and 8 percent. The absolute
derivatization yields ranged from 46.4 to 93.6 percent, with two compounds exhibiting yields
lower than 70 percent (2,4,5-T at 46.4 percent and dicamba at 57.3 percent).
After these preliminary experiments, we proceeded with the determination of the actual
extraction recoveries by SFEITBAIMI. All experiments had to be performed with spiked
materials due to the lack of certified standard reference materials. Three different matrices (sand,
clay soil, and topsoil) were spiked at two concentrations with the chiorophenoxy acids (only seven
of them were available at the time these experiments were performed) and extracted immediately.
Table 7 summarizes the individual recoveries for each of the seven compounds. Figures 3 and
4 present the average recoveries for each spike level and each matrix across the seven
compounds. Overall, the recoveries were quantitative at the two spike levels.
Attempts to extract other conçounds c rreru4y listed in Method 8151 using SFE combined
with TBA/MI were unsuccessful, either b ause the compounds did not derivatize to the methyl
or butyl ester (dalapon, dinoseb, and 4-nitrophenol), or the derivatization product decomposed,
as in the case of DCPA diacid, pentachlorophenol, and picloram. We also found that the analysis
of the extracts of some of these compounds, when injected into the GC/MS system, exhibited
large peaks that did not match those of the corresponding methyl esters. These peaks were
identified as the butyl esters. For two of the compounds, we found that the concentrations of the
butyl esters were 4 and 16 times as high as those of the methyl esters. The experimentally
determined ratios of the butyl esters to methyl esters are shown in Table 8.
A draft protocol of this method wa jwepared and is currently being reviewed. A round-robin
study has tentatively been scheduled for later this year.

1. Hawthorne, S. B., Miller, D. J., Nivens, D. E., and White, D. C., “Supercritical Fluid
Extraction of Polar Analytes Using In Situ Chemical Derivatization,” Anal. Chem . 64, 405-
412 (1992).
2. Richards, M., Dow Chemical, personal communication (1991).
3. Chiang, L., Magee, R. J., and James, B. D., “Chromatographic Determination of
Chlorinated Phenoxyacetic Acids by Simultaneous Extraction-Methylation and Application
to Soil Analysis,” Anal. Chim. Acta . 255, 187-196 (1991).
4. Chiang, L., Markovec, L. M., Magee, R. J., and James, B. D., “A Convenient Method
Using MethanolfBenzyltrialkylammonium Reagents for Simultaneous Extraction and
Methylation of 2,4-Dichlorophenoxyacetic Acid in Soil, with Subsequent Analysis via Gas
Chromatography,” J. A ric. Food Chem . 39, 1110-1112 (1991).
5. Hopper, M. L., “Methylation of Chiorophenoxy Acid Herbicides and Pentachiorophenol
Residues in Foods Using Ion-Pair Alkylation,” J. Agric. Food Chem . 35, 265-269 (1987).

Compound Compound
no. name CAS no. Chemical name
1 Dicaniba 1918-00-9 3,6-Dichloro-2-methoxy benzoic acid
2 MCPP 93-65-2 2-(4-Chloro-o-tolyloxy) propionic acid
3 MCPA 94-74-6 4-Chloro-o-tolyloxy acetic acid
4 Dichlorprop 120-36-5 2-(2,4-Dichlorophenoxy) propionic acid
5 2,4-D 94-75-7 2,4-Dichlorophenoxy acetic acid
6 2,4,5-TP 93-72-1 2-(2,4,5-Trichlorophenoxy) propiomc acid
7 2,4,5-T 93-76-5 2,4,5-Trichlorophenoxy acetic acid
8 MCPB 94-81-5 4-{4-Chloro-o-tolyloxy) butync acid
9 2,4-DB 94-82-6 4-(2,4-Dichlorophenoxy) butyric acid
Parameter Value
GC instrument Hewlett-Packard GC/MS Model 5971A or
Column PTE-5
30-m length x 0.25-mm ID or equivalent,
0.25-j&m film thickness
Column supplier Supelco, Inc., Bellefonte, PA
Carrier gas flowrate 0.87 mL/min (helium)
Temperature program 80°C to 140°C at 20°C/mm, 140°C to 250°C
(5-mm hold) at 4°C/mm
Injector temperature 250°C
Detector temperature 250°C
Injection volume I L
Solvent Methanol
Type of injector Split/splitless; splitless mode; 60 sec
Detector type MSD or equivalent
Scanning mass range 40 to 600 amu
Scan rate  1 sec

Compound name
IS used for
Acenaphthene-d 10
Phenanthrene-d 10
a The concentration of the internal standard is 20 ng/ L.
b The quantitation ions recommended in this method are actually for the corresponding methyl
esters of the chiorophenoxy acid herbicides.

no. Compound name
Temperature —80° C
1 Dicamba
5 2,4-D
7 2,4,5-T
9 2,4-DB
Temperature — 150° C
1 Dicamba
5 2,4-D
7 2,4,5-T
9 2,4-DB
a The extractions were performed at 400 atm/80°C/15 mm static, followed by 15 mm dynamic,
with the reagents added to the sample in the extraction vessel. Extraction vessel volume:
10 mL for BTMAC and BTEAC experiments and 2.5 mL for TMPA experiments. The sample
size was 2 g and the spike level was 250 g/g per compound. Compounds 4 and 6 were not
available for spiking. The extracted material was collected in 2 mL methanol and adjusted to
5 mL for GC/MS analysis. Single determinations.
b The volume of reagent was 3 mL BTMAC solution (40 percent in methanol).
C The volume of reagent was 3 mL BTEAC solution (40 percent in methanol).
d The volume of reagent was 1 mL TMPA solution (10 percent in methanol).
e ND - not detected.

Compound name
a The extractions were performed at 400 atm/80°C/15 mm static, followed by 15 miii dynamic, with
1 mL TMPA solution added to the sample in the extraction vessel (10 percent in methanol,
ethanol, or isopropanol). Extraction vessel volume: 2.5 mL. The sample size was 2 g and the
spike level was 250 zgIg. Compounds 4 and 6 were not available for spiking. The extracted
material was collected in 5 mL solvent (methanol, ethanol, or isopropanol). Single determinations.
1, Not detected. Approximate detection limit was 10 ng/ L.
no. Compound name
1 Dicamba 57.3 16.6
2 MCPP 73.1 7.2
3 MCPA 93.6 7.5
4 Dichlorprop 93.3 7.5
5 2,4-D 87.1 7.4
6 2,4,5-TP 84.6 7.2
7 2,4,5-T 46.4 8.6
8 MCPB 87.1 3.1
9 2,4-DB 86.6 5.9
a The number of determinations was 10. The derivatization yield was determined at a concentration
of 100 ng/jiL in solution. The target analytefTBA/MI solution was injected into the GC/MS system
at an injector temperature of 250°C. The GC/MS operating conditions are given in Table 2.

Clay soil
87.5 103
79.0 76.0
100 110
85.0 103
77.0 116
98.4 82.6
83.4 57.4
102 73.5
91.7 68.8
86.3 67.5
94.0 108
92.5 99.0
121 122
105 97.0
106 120
90.9 85.4
72.7 70.6
84.5 80.5
77.2 70.4
75.2 69.0
91.0 111
80.5 75.5
90.5 96.5
80.0 85.0
83.5 97.5
107 108
114 61.8
107 104
91.6 96.4
111 106
124 124
126 70.4
128 141
102 105
123 120
a The extractions were performed at 400 atmI8O°CI 15 mm static, followed by 15 mm dynamic, with the reagents (0.5 mL TBA 25 percent in methanol
and 0.5 mL MI) added to sample in the extraction vessel. Extraction vessel volume: 10 mL. The sample size was 2 g and the spike level was
50 ppm or 250 ppm per compound. Compounds 4 and 6 were not available for spiking. The extracted material was collected in 2 mL methanol.
Duplicate determinations, except for the topsoil at 250-ppm spike level.

Ratio of
butyl ester to
Compound name methyl ester
Acifluorfen 4.03
Bentazon 0.05
Chioramben 0.75
2,4-D 0
2,4-DB 0
Dalapon a
DCPA diacid b
Dicamba 6.4
3,5-Dichlorobenzoic acid 16.0
Dich lorprop 0.25
Dinoseb a
5-Hydroxydicamba c
MCPA 0.09
MCPB 0.13
MCPP 0.14
4-Nitrophenol a
Picloram e
2,4,5-T 0.13
2,4,5-TP 0.30
a Not able to detect formation of esters.
b The derivatization product appears to decompose.
This compound was not available for testing.
d The derivatization product was found to be tetrachiorophenol.
e The only compound detected in this reaction was the decarboxylated

thundance ( ) TIC: 0801009.D
1200000 (j)
( : )
800000 -
0 — r J
Time -> 10.00 15.00 20.00 25.00 30.00
Figure 1. GC/MS chromatogram of the composite standard containing the methyl esters
of the chiorophenoxy acid herbicides. The GC/MS operating conditions are
given in Table 2.

bundanc e
1200000 -
900000 -
700000 -
500000 -1
L ime
TIC: 1010020.D
Figure 2. GCIMS chromatogram of the chiorophenoxy acid herbicides derivatized to their
corresponding methyl esters upon injection into a GC/MS system (injector
temperature 250°C). The GC/MS operating conditions are given in Table 2.
10.00 15.00 20.00 25.00 30.00

Spike Level - 50 ppm
Compound Number
m Sand
o Clay
• Topsoil
Figure 3. Average recoveries of the seven chiorophenoxy acid herbicides extracted from
sand, clay soil, and topsoil samples spiked at 50 ppm. The extractiohs were
performed at 400 atm/80° C/iS mm static, followed by 15 mm dynamic, with the
TBA/MI reagent added to the samples in the extraction vessels.
0 )
2 3 5 7 8 9

Spike Level - 250 ppm
Compound Number
Figure 4. Average recoveries of the seven chlorophenoxy acid herbicides extracted from
sand, clay soil, and topsoil samples spiked at 250 ppm. The extractions were
performed at 400 atm/80°C/15 mm static, followed by 15 mm dynamic, with the
TBA/MI reagent added to the samples in the extraction vessels.
o Clay
100 -
60 -
40 -
1 2 3 5 7 8 9

Gale 0. S itton. Laboratory Dtrector, G&Lson C rporauoa,
660 1 X .zr vd1e Road, East SyYltuse, ew York 13057
(315) 432-0506
The benefits of nderstandtag Laboratory analytical methods for volatile organic compounds
(VOCs) ihould not be underesumated. Differences of detection limns, quality control and method
limitations can re u1t tn significant vantuona in anilytical results. Knowledge of analytical
methods can Limit unnecessary laboratory costs and decrease laboratory turnaround witho
sacrificing the quaLity or usability of the results.
Volatile analyses cf water and soil samples consists of: a sample concentration techntquc. the
separation of anaI,ftes from the ma ix (as well as from other snalytes) and finally detection an
quantüatioo. The foilowing types of equipment serve these purposes: an in .ctor, a purge and
trap liquid sample concentrator, chromatography cotuiii , and detectors (F1D, ECD, HaWELCD,
ND, and mass specuometer)
Each regulatory pi gr*m has deiig r ’ed its own methods and each method has dis nc: adv&ntages.
There are certain advantages some methods have ov others. Drink.ing water methods such a.s
502.2, 503.1 and 524.2 irs sometimes used for Low level grouzadwaser analysis. EPA methods 601.
602 and 624 ir s pan of the 1’JPDESISPDLS (Clean Waxer Act) program. RCRA (SW84 ) divides
methods into sample preparation aid concentration methodi (i.e. 5030-purge and trap) and
analysis methods (i.e. $010, 8020, 8021, 8015, and 8240). Finilly, CERCLA and NYS DEC ASP
have developed thair own variation of RCIA volatile method *240.
Alone with the an&yZicai differences concerning calibration, detection and quanritation, the
quaLity- control requirements iN alio sp.csflc to each method. Quality control checks, matrix
spikes, duplicates, and deliverables may be specified or (as in the cue of ECRA methods) may be
Loosely defined.
In conclusion, the choice of method can affect t usability of the data. The best method for a
project will be the one that provides the fairest turnaround, assures the required detection limits
and yields ucce*.sful validation performance.

! d Shirkhan , Fluid Management Systems, 125 Walnut Street, Watertown, MA
02172, and T. 3. Tiernan, P. J. Wage]., G. F. VanNess, 3. G. Soich and J. H.
Garrett, DepaLtrJ ent of Chemistry, Wright State University, Dayton, 011 45435.
previous reports from our laboratory have described the development and testing
of two computor—controlled low—pressure liquid chromatographic systems for
isolating polychiorinated dibenzo—p—dioxins (PCDD) and polychiorinated
dibeezofurans (PCDfl from complex sample extracts containing these. The puIn
and multi-port Teflon solenoid valves of the automated system were connected to
glass chromatographic columns and were controlled by a microprocessor which was
interfaced to n computer. The tirst automated system sequentially directed as
nany as five sample extracts through separate multi—layer silica gel and
activated alumina columns and then through a common carbon column. Experiments
conducted with this sequential system demonstrated that the silica and alumina
oluan eluates did not need to be concentrated prior to adding each eluate to
the next column, and that the sequentially processed samples were not cross—
contaminated, even when samples containing high and low concentrations of
PCDD/PCDF were processed in series. The concentrations of PCDD,’PCDF which were
seasured in incinerator flyash, paper sludge and wastewater extracts, which were
fractionated with the automated system, were in excellent agreement with the
concentrations neasured in identical extracts which were cleaned up by using the
normal manual procedures. ln addition, the extracts processed with the
automated system contained lower levels of background interferences than the
eanually prepared samples. Apprrncimately 16 hours are required for the
sequential system to process five sample extracts. To reduce the time required
to prepare a set of samples, a modular system was constructed in which each
sodule contained all of the hardware component5 which are required to proparo a
single sample extract. A computer was interfaced to eight of these modules and
instructions were downloaded to the concurrently operatin.g nodules. Each nodule
of this parallel—processing liquid chromuatographic system processed an extract
for GC-MS aualysis in two hours and these chromatographic separations were
highly ‘eprodueibl . The g1as columns used with the ntnmat d system in these
previous studies required careful manual cleaning and repacking between each set
of samples. In order to further reduce the effort required to process a set of
samples with the automated system, prepacked silica, alumina and carbon columns
Vere obtained from a supplier and their use was evaluated in the present study.
Use of these new disposable Teflon columns can result in saving approximately
three hours of labor between each set of samples processed, and the prepacked
columns are easily attached to the automated system. The chromatographic
properties of the prepacked columns were found to be quite similar to those of
the previously used columns. Results of experiments evaluating the
reproducibility and stability of the prepacked columns will be presented. The
concentrations of PCDD/PCD and the levels of background interferences in
samples prepared with the prepacked columns will also be compared to results
obtained with laboratory—prepared columns during the earlier studies.


Lois Jassie , Elaine Hasty, Robert Revesz, CEM Corporation, 3100 Smith Farm Road,
Matthews, North Carolina 28105; H. M. Kingston, Department of Chemistry, Mellon
Hall, Duquesne University, Pittsburgh, Pennsylvania 15282
The primary environmental concerns regarding hazardous wastes in the geosphere are the
possible contamination of groundwater aquifers by waste leachates and leakage from the
wastes. Through complex interactions of a combination of landfills, streams, wells and
aquifers, these hazardous wastes, in particular those containing toxic metals that find their
way into the geosphere, have the potential to effect biota and human health. Processes that
influence the availability of toxic, heavy metal mobility include sorption by the soil, ion
exchange and interaction with soil bacteria, etc. These activities are highly dependent on a
number of chemical parameters that include pH, temperature, and soil type.
Properties of metals dissolved in water depend largely on the metal species, thus speciation
of metals in aqueous systems plays a crucial role in characterization of ground water and
wastewaters. Hydrated metal complexes, organometallics, metals incorporated into
crystalline minerals and adsorbed forms can have vastly different solubilities and transport
properties. It is difficult to know the amount of these potentially harmful pollutants that
make their way into potemtiahly potable water sources. Through a complex interplay of
natural and man-made phenomena many of these materials become available to the dynamic
and ever-changing environment. Leaching, especially by acids, tends to mobilize
otherwise precipitated hydrated metal oxides and at reduced pH retains many metal ions in
solution. Biologically available pollutants are also available to man through incorporation
into the food chain. Shellfish tend to concentrate a number of pollutants, including heavy
metals and finding them at elevated levels in this tissue is generally considered an indicator
of contamination. Analysis of solid wastes, soils, contaminated sludges, biological tissue
or waste water for action levels of regulated toxic metals is in essence a sample preparation
exercise that must parallel, to some extent, the bioavailability of these toxic metals. This
paper will examine the application of microwave-assisted leaching schemes to the analysis
of heavy metals in samples of environmental interest, in particular, sediments, soils,
biological tissues, and water matrices.
History of the Development of EPA Microwave Methods-A number of
individuals and groups have interacted with various Environmrntal Protection Agency
(EPA) programs over the last five years to produce the three principal microwave test
methods for trace elements currently pending approval. A microwave method for
preparation of industrial furnace feedstreams for trace metals is also in the system.
Nation-wide approval was sought by CEM Corporation for the closed vessel microwave
digestion of wastewater samples for metals determination. Under the Clean Water Act, we
have applied for a variance to use an alternate test procedure based on microwave heating
for the analysis of (metal) pollutants. A proposed rule to amend 40 CFR part 136 appeared
in the October 25, 1991 Federal Register approves this method as an alternate (additional)
procedure for the determination of thirteen metals in domestic and industhal wastewater at
specified concentrations by inductively coupled plasma optical emission spectrometry
(ICP-OES). These metals and their limits of quantitation are shown in Table I. The

method is also approved for the determination of 11 metals by direct aspiration AAS, and
for 10 metals by DCP/AES analysis.
Table I. Elements and their Limits of Quantitation Established by the EPA for Domestic
and Industrial Wastewater Samples Prepared by Microwave Digestion for ICP/AES
Element Maximum Concentration. .tg/mL
Al 50
Sn 50
As 50
Ba 0.5
Cd 0.5
Cr 10.0
Cu 6.0
Fe 50.0
Pb 0.4
Mn 5.0
Ni 40.0
Se 80
Zn 10.0
Concurrently, efforts were undertaken in conjunction with the Office of Solid Waste
(OSW) for the RCRA program and the Office of Emergency and Remedial Response for
the CERCLA (Superfund) program to produce microwave dissolution methods for the
analysis of metals in solids and aqueous samples, respectively. These methods seek to
bridge the gap between older prescription based methods and performance based methods.
For example, Method 3015 for aqueous samples other than drinking water contains the
prescription for conducting the test in a “follow me “type procedure. However, it is
dependent on calibration of the microwave unit for correct transference of the critical
parameters of the test in order to accurately reproduce the requisite conditions under which
decomposition occurs. Both methods also contain the performance criteria for correct
implementation of the procedure. That is, they stipulate the important temperatures at
which dissolution must be conducted in order to achieve conditions for reproducibly
leaching metals.
Chemistry and Methodology-The empirical basis of the tests is derived from
studies on the decomposition temperatures for the principal components of biological and
botanical matrices (Kingston & Jassie, 1988). Specific procedures were developed to
prepare a variety of standard reference materials (SRMs) for elemental analysis. Tailoring
the decomposition of a specific matrix means that samples like wheat flour, which are
largely carbohydrate with small amounts of protein and no fatty component, are easily
digested in nitric acid at 145-150 °C. Decomposition at the appropriate target temperature
does not however insure total recoveiy of any given analyte from solution; the chemistry is
also important. Organic and inorganic matrices as well, require HF to release elements that
may be bound up in the siliceous phase.
Comparison of Methods 3015 and 3051-This target temperature concept has
been developed into the performance based criteria that are features of the two new
proposed U. S. EPA SW-846 (microwave) Methods 3051 and 3015 for the sample

preparation of solid wastes and waters. Shown below is a table comparing the empirically
derived parameters for both tests. The performance criteria for Method 3015 include
160 ± 4 °C for 10 minutes followed by 160-170 °C for 10 minutes. Similarly for Method
3051, samples must reach 175 °C in <5.5 minutes and remain between 170-180 °C for the
balance of the 10 minute heating period.
Table II. Comparison of Parameters in EPA Methods 3015 and 3051
Vessel # of Sample HNO3 Time Temp. Program
vessels size niL m i i i °C Power. W
Method3Ol5 PFA 5 45 mL 5 10 1)160±4 545
(aqueous) 10 2)160-170 344
LDV 5 45mL 5 10 1)160±4 473
10 2)160-170 237
Method3O5l PFA 6 0.5 g 10 5.5 1)175 574
(solids) 4.5 2)170-180
IDV 6 0.5 g 10 5.5 1)175 445
4.5 2)170-180 256
note: PFA-Al1 PFA Teflon Vessel
LDV-Lined Digestion Vessel
Current Status and Activities of EPA vis a vis Microwave Methods-
Final approvals of the NPDES method await adjudication of comments from the field.
Both SW-846 Methods 3015 and 3051 were released prematurely. The long-awaited 2nd
revision of the third update of the Statement of Work is complete. Its appearance in the
Federal Register however, has been delayed pending some internal conflict resolution.
Role of Standards in Method Validation- Reference Materials (RMs) are
widely used in the the analytical community to evaluate analytical methods and laboratory
performance. Proper use of such materials is imperative to insure that information derived
from them is scientifically accurate and defensible. Criteria for their use include a
reasonable match with the client matrix and elemental concentrations of roughly the same
order of magnitude (Becker et al, 1991). It is also helpful to select an element with small
uncertainties, the more easily to detect bias. Herein lies a dilemma. How well can these
RMs be used to validate something for which they were not certified?
Among the most widely used reference materials are those from the National Institute of
Standards and Technology (NIST), the National Research Council (NRC) of Canada, and
the International Atomic Energy Agency (IAEA). The U. S. Geologic Survey has several
water standards, and U. S. EPA labs use consensus standards. EPA consensus materials
generally have been extensively analyzed and, for certain elements, there are
concentrations, occasionally, with confidence intervals. According to the laws of
statistics, large numbers of analyses of the same elements will tend to cluster around a
certain value. However, problems of method bias, systematic, or random errors that may
be present, can skew these expected normal distributions. Even if all laboratories
performed the same analysis the same way, all of the results could be similarly biased.
Some of these problems have been addressed in the validation study of the proposed

microwave solids method and in the bias study for the proposed microwave method for
aqueous samples.
In 1989, the deputy administrator of EPA established the Environmental Monitoring
Management Council (EMMC) whose charge was to develop a solution to monitoring
problems within the Agency. The first issue addressed by EMMC was the integration of
the Agency’s environmental monitoring methods and quality control techniques. A multi-
part program consisted of simultaneous efforts to integrate five existing methodologies,
devise a system of coordinating the future development of methods across the Agency, and
develop a system for the speedy approval and publication of new methods (Freidman,
D.,1992). Among the five laboratory techniques selected for their broad utility and
suitability for demonstrating the feasibility of methods integration are 1) conventional and
2) microwave-assisted strong acid extraction of elemental species from liquids and solids.
In the preparation of draft methods, efforts have been made to develop a single acid
mixture composed of nitric and hydrochloric acids and the accompanying experimental
conditions that can be used in both conventional and microwave extractions to give
comparable results. Such conditions for the mixed acids have been derived empirically for
the microwave technique as one of the research goals of an interagency agreement between
NIST and EPA Region III Laboratory. Temperature profiles of these conditions will be
In this section examples of extraction and leaching in various environmental applications
will be presented. In most cases, information was derived from the analyses of samples
prepared by microwave dissolution in hot nitric acid. Examples of microwave assisted
leaching of soil samples using aquaregia also will be discussed, and one sequential
fractionation will be cuecL
Sequential Leaching-To mimic environmental conditions that might be
encountered in nature, Mahan sequentially extracted Ca, Fe, Cr, Mn, Pb, and Zn from
sediments using microwave leaching techniques (Mahan et al, 1986). They felt sequential
fractionation was particularly useful since anthropogenically produced metals normally
show up in the labile fractions rather than in the residual sediment. Nitric and hydrochloric
acid (4:1) leaching played a prominent role in assessing the residual metals and, when HF
was added, total metals could also be determined. Selected metal recoveries for SRM 1645
River Sediment in the fractionation scheme are shown below. Because water waste and
leachates inhabit various environments that are dynamic and continuously changing,
perhaps we should be trying to compare one-step acid leaching procedures to such
fractionation schemes.
Table ifi. Metal Recovery by Bonding Fraction in Microwave Extraction of NBS
SRMI645 River Sediment Determined by FLAAS
Metal Exch+Carb Anoxic Organic Residual Total Certificate
Fe,% 0.17 2.5 0.85 3.5 7.0 11.3± 0.6
Cr,% 0.08 2.6 0.76 0.11 3.6 2.96± 0.14
Mn,j ig/g 103 279 55 124 561 785±48
Pb,p g/g 160 434 163 nd 757 714± 14
Zn,pg/g 358 1215 280 168 2021 1720±85

Iron was tied up in the silica fraction because subsequent decompositions with HF
recovered the remaining iron from the residual fractions and in a total digest of starting
material as well. For both SRM1645 and SRM1646 Estaurine Sediment, summing the
fractions for the other metals gave consistently high results when compared to the
certificate value. This led them to suspect a systematic error either in the extraction
procedure or in the determination of the metals.
One-step Leach Procedures-Under contract from OSW, Research Triangle
Institute (RTI) conducted a collaborative study validating the microwave-based acid
dissolution SW-846 Method 3051 for soils, sediments, sludges, and oily wastes (Binstock
et al, 1989). Using reference materials, and a simulated oily waste composite of two
reference materials, the analysis of precision showed an order of magnitude variation in
repeatability of elemental determinations, e.g. 8.5% for Mn to 85% for Mo in the
sediment. The study also showed that 1) if an extra limit of error, ± 20%, of the total value
is acceptable, then the microwave method may be satisfactory to estimate the concentration
of Zn, Pb, Ni, Mn, Cu, and Cd in sediment, 2) estimates of concentration can be done
with about the same precision as open beaker digestions for the same metals, and 3) for all
of the soil, sediment and oily matrices, with somewhat better precision than classical open
vessel techniques. When the uncertainty bounds of the SRM are statistically enlarged, the
extra limit of error often allows a larger number of experimentally determined mean values
to be acceptable. It would be of interest to establish which sample preparation parameters
(temperature, time, sample size, etc.) are most critical to reducing the large relative
standard deviations that were cited for analysis of the sediment? What affect have these
parameters on the recovery of metals from the SRMs or from the synthetic mixture?
Because the wear metals in oil (SRM 1085) is a relatively uncomplicated matrix, dissolution
in hot nitric acid should have achieved essentially complete digestion. Thus it is the only
material in which it was conceivable to examine bias for the microwave method. If the
criteria of “no method bias” means recovery of 90-110% of the certified value, then the
following table suggests that 7 of the nine metals can be reliably estimated.
Table IV. Recovery and Bias Data for SRM 1085 Wear Metals in Oil Using Method 3051
Element Mean ± SD %RSD Certificate %Bias
Ag 234± 35.9 15 (296) -20
Al 295 ±31 .1 10 296±4 0
Cr 293± 26.6 9 298±5 -2
Cu 289±23.8 8 295± 10 -2
Fe 311± 34.7 Ii 300±4 +4
Mg 270 ± 29.1 11 297±3 -9
Mo 238± 30.3 13 292± 11 -18
Ni 293 ±25.0 8 303 ±7 -3
Pb 279 ± 22.1 8 305 ± 8 -8
note: Elements determined by ICP-AES; Silver value for information only;
Concentration in tg/g; SRM obtained from NIST.
Except for iron, all values are below their certificate values. Simple estimates of bias, the
difference in means divided by the certificate value, were provided at the time of the
study’s completion. Recalculation of the absolute value of the bias estimate at the 0.05
significance level, suggests that 27 tg/g for Mo is larger than the critical value of

18.7 .tgfg. This is almost certainly the case for Mg and Pb as well. Even the confidence
interval for bias in these three cases does not span zero (Becker, et al, 1988).
Fortunately, the absence of an approved microwave method for the preparation of soils
and sediments has not kept analysts from using the method for routine screening.
Laboratories often use reference materials as benchmarks to gage the laboratory
performance of methods. What is the validity of using materials with certified levels of
specific elements that have been analyzed for total metal concentrations? Have they
become inter alia quality control standards, or performance standards for a chemical
extraction process which by its very nature is incomplete? To improve the quality of
analytical data coming from testing labs, reference materials are now routinely included in a
batch of samples. Wide spread adoption of such practises brings with it problems of
misuse and misunderstandings about the infonnation associated with the RM certificate. In
the context of quality assurance, the proper role for standard reference materials may well
be as a primary reference standard against which one or more secondary laboratory
reference standards are “calibrated”. These secondary standards may be suitable QC
samples in the batch concept; the SRM is only called upon infrequently for recalibration of
the laboratory standards. Suppliers of certified reference materials may be hard pressed to
fill the demand of all the busy laboratories requesting RMs, not to mention the high cost
associated with such indiscriminate use. Does the analytical method shut down when no
more of the RM is available? In collaboration with Jean Kane from the materials
certification program in the Office of Standard Reference Materials at MST we have begun
to collect and assess data from members of the analytical community using such materials
with microwave methods.
During the process of developing the microwave dissolution procedure for water samples,
a number of reference materials were identified in an effort to construct a valid analysis of
the bias that might be present in such a leaching technique. Members of EPA Regional Lab
ifi (Annapolis, MD) examined a suite of water standards that were used to assess the bias
of SW-846 Method 3015. Elements of interest were leached in the presence of hot nitric
acid at temperatures where the organic components present in the matrix are completely
oxidized. Preliminary data from the Method 3015 bias study showed that certain metals
could be estimated without bias.
In her study comparing different decomposition procedures, Kruschevska showed that
zinc can be determined in milk matrices that were simply digested in nitric acid. Recovery
approaches 100% in nearly all cases, despite the presence of relatively large amounts of
residual carbon (Kruschevska, et al, 1992). Measurable carbon residues chronicle the
incompleteness of the digestion process, especially in protein-rich, cholesteric, or high-fat
matrices. Additions of peroxide or sulfuric acid in the microwave preparation and
temperatures of 300 °C in a high pressure asher were effective in reducing the residual
carbon content of the milk samples, however, the recovery of zinc was not substantially
improved. It is clear that nitric acid has been able to release the analyte of interest despite
the presence of undigested matenal.
As the addition of peroxide to biological matrices improves the recoveries of many
elements, complete mineralization using appropriate reagents such as hydrofluoric acid is
needed to determine total metal content in silica-based soils and sediments. Certain metals,
like cadmium and aluminum, can be fully recovered when digested in aquaregia. For
example, in citrus juices, microwave digestion was comparable to a muffle furnace method
for the determination of certain elements. Because the authors suspected that the products

were contaminated with tin from the metal storage container aquaregia was used for the
digestions (Rezaaiyan et al,1990). The improved recoveries over nitric acid alone provide
a cogent argument for the inclusion of aquaregia into both microwave Methods 3015 and
One of the most often used reference materials in environmental laboratories is NISTs
SRM2704 Buffalo River Sediment. Data from selected analyses performed by microwave
hot acid leaching are shown in the following table along with other reference materials that
have been used routinely for quality control.
Table V. Determination of Selected Toxic Metals in Inorganic Reference Materials Prepared
by Microwave Leaching in Nitric Acid
concentration, j .tg/g
As Cu Pb
Reference Mat’l Cert. Found Cert Found Found
Buffalo River 23.4±0.8 23.5±0.32 98.6±5.0 96.3±2.14 161±17 168±4 Army
Sediment 25.2±2.1 102±1 178±1 CEM
SRM2704 31.7±6.3 93.5±4.3 125.4±11.8 ManLab
Marine Sediment 211±11 180±1 452±1 452±8 404±20 399±9 CEM
Peruvian Soil (90) 95±9 (80) 78±9 (100) 143±2 CEM
SRM 4355 53±6 120±8 RTI
note: numbers in parentheses are information values only
Table VI. Determination of Selected Toxic Metals in Biological Reference Materials
Prepared by Microwave Leaching in Nitric Acid
concentration, 1g/g
As Cu Pb
Reference Mat’l Cert. Found C Foim Found
Oyster Tissue 13.4±1.9 — 63.0±3.5 61.0 0.48±0.04 3.5 EPA-S
SRM 1566
11.8±2.5 60.5±4.2 3.5±0.4 3.1±0.2 EPA-7
SRM1566a 14.0±1.2 13.1±0.8 — Ybanez
Bovine Liver 0.047±.006 — 158±7 149±15 0.44±0.06 0.40±0.03 Lyon
SRM1577a 169±1 Sah
It is apparent from these tables that a judicious selection of elements can give good answers
to the correct questions. Data from a large number of replicates may give insight into
decomposition problems (matrix). Or, as in the case of high arsenic values, they may
suggest a measurement problem, such as spectral interference. Measurement errors must
be separated from leaching errors. Leaching efficiency for all elements is not the same, as

was shown in the RTI study and not all 26 elements for the soils or 23 elements for the
waters, can be expected to behave as well as these few.
There are many ways to assess the performance of an analytical method. In (3FAAS for
instance, spike recoveries, standard additions, and matrix matching of blanks and
standards all can provide important information on the validity of data acquired. If
reference materials are required, then they should be as similar as possible to the sample
and have the same concentration range for the elements under scrutiny. They must be
prepared by the same sample preparation method and the chemical measurement process
must be under control. A window of 20% of a certificate value for an analysis may be
problematic if the acid leaching efficiency is unknown. The limit of elror must be defined,
otherwise such windows are arbitrary and may be meaningless in light of low efficiency
extractions. For example, even in aquaregia, gold-bearing ores have low levels of
extractability. Thus, accepting or mandating leach values that are 80-120% of total values
may be unscientific. How do we know which partial leach comes closest to that which
occurs in nature?
The importance of submitting data on method performance during the comment period
cannot be overstated. It is in fact critical to assessing the robustness and durability of the
methods. When the SW-846 methods finally appear, there will be a golden opportunity to
participate in the method validation process. Analysts with studies on comparability of
3051 and 3050 are urged to speak up. And, in the absence of a collaborative study,
performance data for Method 3015 will be essential.
For those analysts and laboratories with time on their hands, a microwave leachability
study of reference materials under various simulated environmental conditions is
guaranteed publication. RMs must be well characterized when used as performance
standards so it is important to know what parameters affect recovery when microwave
assisted leaching is performed. For example, acid combinations like aquaregia produce
nitrosyl chloride and can elevated the pressures in the digestion vessels. Such a system
will almost certainly behave differently from just HNO 3 . Aluminum and cadmium
recovery will be improved but does aquaregia increase the extraction of other metals? If
the temperature of miric acid is kept at 175±5 °C, as suggested in Method3OSl, does the
recovery improve with longer extraction times? How are metal recoveries affected by
temperature? What influence does the presence of organic material have on the extraction
recovery of the elements of interest from inorganic or aqueous matrices? Answers to these
questions will help the analytical community to properly assess the environmental impact
of hazardous wastes.
Becker, D., Christiansen, R., Currie, L., Diamondstone, B., Eberhardt, K., Gills, T.,
Hertz, H., Kiouda, G., Moody, I., Steel, E., Taylor, J., Watters, R., Zeisler, R. NIST
Special Publication 829, U.S. Department of Commerce, January, 1992.
Binstock, D. A., Grohse, P. M., Gaskill, A., Sellers, C., Kingston, H. M., Jassie, L. B.
I. Assoc. Off Anal. Chem. 4(2), 1991, 360-366.
Friedman, D., Environinenwi Lab, April/May, 1992. 18-23.
Hewitt, A. D., Reynolds, C. M. Atomic Specrroscopy, fl(S), 1990, 187-192.
Kammin, W., Manchester Laboratories, private communication, 1992.

Kingston, H. M.; Jassie, L. B., J. Res. Nat’l. Bur. Srds, (3), 1988, 269-274.
Kruschevska, A, Barnes, R. M., Amarasiriwaradena, C., Foner, H., Martines, L., J.
Anal. Atomic Spectrometry, 1992 (in press).
Lyon, T. D. B., Fell, G. S., MacKay, K., Scott, R. D., J. Anal. Atomic Spectrometry, ,
1991, 559-564.
Mahan, K. I., Foderaro, T. A., Garza, T. L., Martinez, R. M., Maroney, 0. A.,
Trivisonno, M. R., Wiiging, E. M. Anal. Chem. , 1987, 938-945.
Paus, R., USEPA Region VII, private communication, 1992.
Rezaaiyan, R.; Nikdel, S. J. Food Sci. (5), 1990, 1359-1360.
Sah, R. N., Miller, R. 0. Anal. Chem. , 1992, 230-233.
Ybanez, N., Cervera, M. L., Montoro, R., de la Guardia, M. J. Anal Atomic
Spectrometry , 1991, 379-384.

67 A new Device for High Pressure Microwave Digestion
Dr. ft KrAmer
Berg hof Maassen Laborgeräte GmbH
Harretstr. 1
W-7412 Eningen
Microwave heated digestion today is a widely known method of sample
preparation in analytical chemistry.
The main advantage of this method compared with traditional ones is a
very quick heating of the sample, even in the range of high temperatures
(T> 200C, since the ‘temperature gradient’ between the sample and the
microwave field is extremely large.
The possibility of quick heating reduces the total time of digestion
considerably, especially at higher digestion temperatures (1> 250’C).
The pressures to be handled at such high temperatures can be
1400 psi or more. Therefore we constructed a pressure vessel made of
steel (p 2845 psi) which can be used as a microwave oven. The
microwave energy is directly coupled into the steel vessel. In the same way
as with traditional digestion bombs P 1 FE or PFA liners and, of course,
quartz glass can be used in combination with the steel vessel.
The microwave heating is controlled by a pressure sensor, which allows to
measure the pressure inside the bomb during the digestion procedure.
As an additional safety precaution, rupture discs are used which will burst
at a well defined pressure of 2100 psi without any damage at the
microwave digester.
To obtain reproduceable results it seems necessary to measure the
digestion temperature. This problem was solved by Installing a Infrared
emission thermometer, which allows to measure the temperature inside the
Teflon liner (or the quartz glasses).
The microwave heated high pressure digestion system MDA II can be used
for the same applications as with traditional digestion bombs.

Zhaoguang Yang, Ph.D., Group Leader, and G. Lynn Schwendinian,
Laboratory Manager, Aptus Environmental Service—Westinghouse
Electrical Corporation, P.O. Box 27448, Salt Lake City, UT
The microwave assisted acid digestion method is intended to
provide a rapid multielement acid leach digestion prior to
analysis by ICP and AA. Recently, closed teflon vessel
microwave digestion is receiving much attention as the new
“high tech” metals digestion (1-5). The primary objective of
this study is to investigate the relation of element recoveries
between the digestion time and pressure control. Various
digestion techniques are used in this comparative study: open
vessel hot plate digestion and closed vessel teflon microwave
The USEPA SW—846 method 3050, 3051 and CEM digestion methods
were used to prepare samples. CEM Microwave Digestion System—
205 with pressure controller and ARL 3410+ICP have been used in
the experiments. The results of this study indicate that
element recoveries are the functions of microwave temperature,
digestion time, and pressure. Microwave digestion can be done
much faster than the comparable hot plate digestion.
Microwave digestion of incinerator ash offers a good
alternative to conventional hot plate digestion for use in the
determination of metals. The results of this study were precise
and compared favorable with certified reference materials. A
comparison between microwave and conventional hot plate
digestion showed that the two methods give close results.
Microwave—assisted environmental sample digestion has received
considerable attention and has become well accepted in the
analytical laboratory. Sample preparation is important. It is
also a much neglected area in the determination of metals by
atomic absorption and inductively coupled plasma atomic
emission spectroxuetry. The problems with the conventional
digestion procedures include elemental loss by volatilization,
atmospheric contamination and longer digestion time
The purpose of this study was to use this technique to
determine the required digestion time and pressure setting to
obtain acceptable results for incinerator samples.

The regent grade HNO 3 , HC1 and solid waste laboratory control
sample were purchased from Fisher Scientific Co., Chemical
Division, Fair Lawn, N.J. Standards were prepared by diluting
1000 mg/i certified ICP stock solutions (SPEX Industries, Inc.,
Edison, N.J.). The stock solutions were also used for spikes.
Milii-Q deionized water (Millipore Corporation, Bedford, MA)
was used for sample digestion and all analysis works.
A CEM microwave digestion system, Model MDS-205, was used for
all microwave digestion procedures. It consisted of an
operator selectable output of 0-900 w ± 90 w in 1% increments,
a fluorocarbon coated microwave cavity, a three—speed cavity
exhaust fan, a digital computer programmable in nine separate
stages, an alternating turntable device system to rotate
samples within the microwave field and a pressure control
system. The closed vessel system used consists of a vessel
body, safety pressure relief valve, vessel cap, venting nut and
tubing, all of which were made of teflon.
CEM digestion method (6): transfer 1 g of sample into a vessel
and add 20 ml of 1:1 HNO 3 :H 2 0 solution. Program the Pressure
Controller at 100 psi and set the time 30 minutes.
The effectiveness of the microwave digestion technique is
dependent on three primary variables: temperature, pressure and
digestion time. Figure 1 and Figure 2 illustrate the affect of
pressure on the recovery of various metals. According to SW846
Method 3051, 0.5g sample in 10 ml concentrated nitric acid and
programed 10 minutes. It is demonstrated that the internal
pressure in the digestion vessel should be above 70 psi (except
Ba) for maximum recovery. In Figure 2, Barium shows a
significant dependance on the pressure. Maximum recovery for
Ba is only achieved at a pressure over 80—100 psi.
Digestion methods require time. The microwave digestion
technique requires significantly less time then the
conventional methods. CEM digestion method (1.0 g sample .n 20
ml }1N0 3 :H 2 0) was used in this group experiment except extention
the time. The results illustrate that there is a variation in
metal recoveries as a function of time. Figure 3 illustrates
that time is significant in the recovery of Barium. Barium
requires twice as much digestion time as Mn and Cd. At high
concentrations, the concentration of the alkali metals and zinc
gradually increases in solution, as illustrated in Figure 4.
Figure 5 illustrates that Cu, As, and Cr are less dependant on
time for efficient recovery in incinerator ash. The recovery
of Sb and Ti vary as much as 25 percent for a digestion time
between 15 and 25 minutes of digestion time as demonstrated in
Figure 6. Figure 7 shows that many metals require at least 15

minutes for best recovery efficiencies.
For many years the hot plate digestion methods (SW846 3005,
3010, 3050 etc.) have been the standard techniques. Table 1
(SW846 Method 3051) is a comparison of metal recoveries in an
certified incinerator ash using both the hot plate and the
microwave digestion techniques. The results show comparable
recoveries for each method. Table 2 (CEM Method) is a
comparison of two analyses for Aptus incinerator ash. The
results are within a standard deviation of each other.
There is no temperature measurement equipment in CEM MDS—205
system. According to CEM information (2,4,5,7), once the
pressure achieves approximately 60—100 psi, the temperature
will reached 175°C. Based on the experimental results, 85—110
psi is preferred.
Microwave digestion offers a good alternative to conventional
hot plate digestion in incinerator samples. The results
obtained after microwave digestion in reference materials
showed close agreement with the stated values. Also a
comparison between microwave and conventional hot plate
digestion showed that the two methods give close results. These
results also indicate that longer digestion times may be
required for some elements.
Advantages of Closed Vessel Microwave Digestion
1. Digestion work can be done much faster than other
2. Sample contamination can be minimized because the teflon
vessel is one of the best materials to use in
trace element analysis.
3. Microwave staged program can often be left unattended to
complete the digestion.
Disadvantages of Microwave Digestion
1. Reduced sample size, allows for less representativeness.
2. Expensive digestion system and parts.
3. Additional time required to assemble and clean the
digestion containers and pressure controller.

20 -
10 -
0 30 60 90 150
Figure le Concentration of Sb, Ti and Se as a function of
digestion pressure.
• Ba
A Mn
500 Cu
o As
El Cr
0 I
0 160 200
Figure 2. Concentration of Ba, Mn, Cu, As and Cr as a function
of digestion pressure.
40 80 120
• Sb
1 20

0 10 20 30 40 50 60
Figure 3. Concentration of Ba, Mn and Cd as a function of
digestion time.
k .1.
0 10 20 30 40
• K
U Na
o Mg
+ Pb
Figure 4. Concentration of K, Na, Zn, Mg and Pb as a function
of digestion tin e.
— 400
I .-
U i
50 60

10 20 30
• Cu
• As
t Cr
40 50 60
Figure 5. concentration of Cu, As, and Cr as a function of
digestion time.
I - - I
10 20 30
Figure 6. Concentration of Sb, Ti and Se as a function of
digestion time.
I .-
I —

• Ni
30 iii iiiii Ag
0 I I
0 10 20 30 40 50 60
Figure 7. Concentration of Ni, V and Ag as a function of
digestion time.

Table 1 Analysis Results for Incinerator Ash
by ICP
Element Hot Piat Microwave Certified Value
ppm ppm ppm
Cd 426 425 422 ± 37
Cu 302 30]. 276 ± 26
Mg 6420 6151 6318 ± 750
Ni 19.8 25.0 22.4 ± 3.2
Pb 4538 4206 4531 ± 325
Zn 20659 20123 21421 ±2094
Table 2 IC? Analysis Results for Incinerator Ash
Element Microwave Hot Plate
ppm ppm
Ag 2.75 2.28
As 268 266
Ba 493 515
K 40541 43709
Mn 472 470
Na 46019 43709
Se 36.6 32.6
Ti 38.2 40.4

1. J. Nieuwenhuize and etc., “Comparison of Microwave and
Conventional Extraction Techniques for the Determination
of Metals in Soil, Sediment and Sludge Samples by Atomic
Spectrometry,” ANALYST.Vol. 116 , (1991), 347.
2. E.T. Hasty and S. Littau, “Microwave Sample Preparation
with Temperature Feedback Control,” Presented at the 1991
Eastern Analytical Symposium, Somerset, N.J.
3. V.L. Verma and T.M. McKee, “Comparison of Procedures for
TCLP Extract Digestion; Conventional vs Microwave,”
Present at the USEPA Seventh Annual Waste Testing and
Quality Assurance Symposium, 1991, Washington, D.C.
4. E.T. Hasty, S.E. Littau and R. Revesz, “Optimization of
Microwave sample Preparation Methods Using Pressure
Feedback Control,” Presented at the 1991 Pittsburgh
Conference and Exhibition, New Orleans, LA.
5. R. Revesz and E.T. Hasty “Microwave Digestion of Soils,
Sediments, and Waste Water for Analysis of Environmentally
Significant Elements,” Presented at the 1990 International
Conference on Metals in Soils, Waters, Plants, and
Animals, Orlando, FL.
6. “Microwave Application Note for Acid Extraction,”
Application Note EW-9 Revision, CEM Corporation, 1989.
7. Letter from G.N. LeBlanc, CEM Corp. to Z. Yang, Aptus,
Inc., March 30, 1992.

Joseoh P. Romano . applications chemist, Millipore Corporation,
Waters Chromatography Division, 34 Maple Street, Milford,
Massachusetts 01757
Capillary Ion Analysis (CIA) has recently been introduced as a new
separations technique for the analysis of inorganic and organic ions.
CIA is a branch of capillary electrophoresis (CE) which is optimized
for the rapid analysis of low molecular weight anions and cations.
It separates ions according to their mobility in electrolytic
solutions. This paper will review the characteristics of CIA that
are especially significant for the analysis of anions in ground water.
Among the many attributes of CIA are rapid, highly efficient
separations with different selectivities (compared to ion
:-:—:::;‘aohy), simplicity, and economy. ie instrumentation ‘ as
few moving parts and uses a low cost, easily replaceable hollow
capillary instead of a packed chromatography column. Sample
preparation is minimal because there is no chromatography column
to oe protected from extraneous materials in the sample. Analyses
are completed in less than four minutes.
The scale of CIA electrolyte consumption is at least an order of
magnitude smaller than liquid or ion chromatography eluent
consumption. High sensitivity is achieved while analyzing only
nariotiters of sample, using only a few microliters of electrolyte in
the capillary. This ability to analyze complex samples without
producing any significant volume of additional waste has caught the
attention of analysts who are becoming involved in the mixed waste

Laurence W. Strattan , Ph. D., and Marcela Saio, U.S.
Environmental Protection Agency, National Enforcement
Investigations Center, Box 25227, Denver, Colorado 80225.
Instrument manufacturers tout capillary electrophoresis (CE)
as a versatile, fast, and sensitive alternative to ion
chromatography techniques for the analysis of anions. We
recently used CE for the determination of inorganic acid
anions and C 2 -C 4 aliphatic carboxylates in groundwater at a
municipal landfill that is now a Superfund site. This paper
discusses the results of that project, and presents
suggestions for the use of CE for environmental samples.
The objective of the project was to track highly water
soluble parameters which would migrate via groundwater, but
not through interstitial vapor spaces in the soil.
We found that with the use of an internal standard as a
retention time reference, and with the dilution of samples
to an appropriate concentration range, bromide, chloride,
sulfate, nitrite, and nitrate were straightforward to
determine using CE. Determination of the carboxylates was
also straightforward, except when much higher concentrations
of inorganic anions were present, and could be performed
without changing instrumental conditions. The dirtier
samples required rinsing the column with KOH between samples
to provide reproducible conditions. Fluoride and phosphate
proved more difficult to determine due to the closeness in
their retention times in chromate electrolyte, to a
sensitivity to pH, and to the presence of bicarbonate in the
In order to maintain narrow peaks, and thus confidence in
qualitative identification, the optimum concentration for
analysis was in the range of 0.5 to 10 ug/mL (ppm) for
carboxylates, and 0.5 to 15 ppm for the “straightforward”
strong acid anions. Precision, as relative standard
deviation, and bias and were both approximately 10% or less
for bromide, chloride, sulfate, and C 2 —C 4 carboxylates.
The Operating Industries superfund site in the Los Angeles
area operated from the 1940s until 1984, accepting both

industrial and municipal wastes at various times. The
190—acre site is nearly a mile long and over two hundred
feet high, having filled in the saddle point between two low
hills. Temperatures within the landfill can exceed 150 0 F.
At one time a commercial operation recovered over one
million cubic feet of methane per day from the landfill.
The site is currently the subject of cleanup settlements
totaling over $200 million.
Because of the temperatures and the considerable volume of
gas emitted from the landfill, there is the potential that
volatile organic pollutants observed in groundwater off—site
may have moved in part through the gas phase rather than by
direct movement of groundwater. The mechanism of migration
is significant to selecting appropriate remediation
We undertook this investigation, which included water
miscible solvents as well as ionic species, in hopes of
being able to trace the movement of hydrophilic species in
groundwater. Previous analyses had identified these
components at concentrations of thousands of ppm in on—site
samples, and it was hoped that even with dilution the
components would be detectable for some distance.
Our laboratory had recently purchased a CE instrument
because the technique appeared to be both versatile and
simple to use, and would provide an alternative method to
ion chromatography. It seemed ideal for this project since
we could determine both organic acid anions and other common
anions such as chloride and sulfate without changing
analytical conditions.
The CE instrument used was a Waters Quanta 4000. The basic
components of the system are a 30—ky power supply, two
electrolyte reservoirs containing electrodes connected to
the power supply, a UV detector, and the column. The column
is uncoated fused silica capillary, typically of 50 to 75
micrometers inside diameter and 60 cm length.
A simplistic description of the instrument’s operation is
that an applied voltage separates ions in solution according
to their mobility in an electrolyte. Ionic mobilities are
related to basic physical properties such as charge and
size. The migration of ions is the sum of two components,
first the bulk or osmotic flow of liquid caused by the
applied voltage, and second the migration of ions toward the
electrode of opposite charge. In most CE applications such
as the separation of biological molecules, these components

are in opposite directions. Waters has developed a system
in which an osmotic flow modifier is added to the
electrolyte. This modifier changes the sign of the
effective charge on the column wall from negative to
positive, reversing the direction of osmotic flow so that
the two components are in the same direction. The pertinent
point about the osmotic flow modifier for this discussion is
that sample matrix can change the reproducibility of its
effects, as will be discussed later.
The order in which species would reach the detector for
these analyses is negative ions, neutrals, and positive
ions. In routine operation, the analysis was terminated
after detection of the negative ions of interest, and the
column flushed with electrolyte to remove the remainder of
the sample and to reequilibrate wall conditions.
We used the generic conditions suggested by Waters for the
separation of inorganic anions, namely 5 mM chroluate
electrolyte containing the osmotic flow modifier. Most
analyses employed a 60 cm, 75 urn I.D. column and a voltage
of 20 Ky. A few analyses used a 60 cm, 50 um I.D. column at
30 Ky. Anions were detected indirectly by UV absorbance at
254 nrn. Chromate absorbs at this wavelength, but the
targeted anions do not. The anion packets dilute the
chromate concentration causing changes in tJV intensity as
they pass the detector.
Parameters which need to be controlled to achieve stable
migration times are ionic strength of the electrolyte, gas
content of the electrolyte, particulates, and organic
content of samples. Ionic strength of the electrolyte can
be affected by absorption of carbonate from the atmosphere;
preparation of fresh electrolyte is the best practical way
to maintain stability from day to day. Gas in the
electrolyte typically causes an unstable baseline, and in
extreme cases bubbles may break the electrical connection.
Particulate in the column obstructs migration causing
instability. The lifetime of a column is usually determined
by how long it remains free of particulate. The organic
content of samples is not controllable, but it must be
considered, and may require modification of instrumental
The procedure for analyzing samples, starting with the
installation of a new column, was to clean the inside wall
by rinsing for several minutes with 0.5 N potassium
hydroxide, followed by distilled water, and then electrolyte
for approximately ten minutes. One or two blank runs at the
beginning of each series of analyses helped to stabilize
conditions. Electrolyte was prepared fresh daily, degassed

in a vacuum desiccator for 30 minutes, and filtered through
a 0.45 urn filter before use. The only sample preparation
was filtration, possibly dilution, and the addition of
internal standards at 10 to 40 ppm.
As a check of the CE results, we also determined the organic
acids by gas chromatography/mass spectrometry (GC/MS) using
direct aqueous injection on a polar phase fused silica
capillary column. This analysis used perdeuterated analogs
of acetic, butanoic, and heptanoic acids as internal
standards to correct for injection port discrimination
effects. Figure 1 lists instrumental conditions for the
GC/MS analyses.
Figure 2 shows the electropherogram of a standard containing
the one— to four-carbon carboxylates and seven common
inorganic anions. Except for for]nate, the carboxylates are
well separated from the common inorganic anions. This
allowed the determination of carboxylates in samples
containing much higher concentrations of chloride or
carbonate, which elutes just after phosphate. (At the pH of
these analyses these species are both in their monohydrogen
forms, bicarbonate and monohydrogen phosphate.) The
migration time of tungstate is midway between two groups of
comrnonly—occuring inorganic anions.
Figure 3 shows an electropherogram for the same inorganic
anions without the carboxylates. The poor separation of
sulfate and nitrite is due to the relatively high
concentration of sulfate, 15 ppm for sulfate vs. 2.5 ppm for
nitrite. In order to maintain narrower peaks, and thus
greater confidence in identification, the maximum
concentration of individual anions should be limited to 10
to 15 ppm. Note also the time separation between peaks is
as low as 0.03 minutes. This is of the same magnitude as
the variation in migration times we observed for single
components from run to run. We found that a migration time
reference and the use of relative migration times was
necessary to make qualitative identifications. We chose
tungstate as the primary reference peak because it is
relatively uncommon and it appears at a convenient migration
time. We also used valerate in addition to tungstate as a
reference peak for the carboxylates.
Finally, note the separation of fluoride and phosphate in
Figure 3. These two parameters and formate were not
determined in our analyses for a variety of reasons,
including lack of interest for the immediate project.
Quantitation of these parameters was not straightforward due

to inconsistent separation of fluoride and phosphate, and to
the presence of relatively large amounts of carbonate in
samples. The determination of these parameters could very
well be possible for more experienced users of CE, and/or in
samples containing less carbonate. Figure 4 shows the
electropherogram of a sample containing a relatively large
carbonate peak which might mask the presence of fluoride and
phosphate even in the absence of other difficulties.
We also did not determine nitrate and nitrite in our
analyses, solely due to the fact that holding times had long
since expired by the time of our analysis. We saw no reason
why they could not be determined by CE.
Figure 5 contains the electropherogram of a different sample
analyzed at a marginally useful dilution. It is marginally
useful because the sample matrix is so affecting peak shapes
and migration times that analysis of a more concentrated
sample would be useless. Even though the tungstate peak is
distorted, the peak area is consistent with previous
We found it necessary to change the usual instrument cycle
by adding a potassium hydroxide rinse after each sample when
analyzing samples such as those shown in Figures 3 and 4.
The XOH rinse removes the wall coating and any remaining
sample matrix organics from the column, ensuring
reproducible conditions at the start of each analysis.
For samples such as those in Figures 4 and 5, GC/MS gave
lower limits of detection than CE. GC/MS would always give
greater certainty of identification. However, the GC/MS
method was tedious to perform because of carryover of the
acids from run to run. The acids appeared to partition
between salt and acid forms upon each injection, with the
salt form remaining in the injection port to be partitioned
again at the next injection. The use of perdeuterated acids
as internal standards for quantitation helped to offset
these effects, but frequent recalibration and analysis of
blanks were necessary to obtain accurate results.
We found that the need to maintain narrow peaks in order to
be confident of identifications based on relative migration
times limited the calibration range on CE for these
analyses. The useful calibration range extended from 0.5
ppm up to about 10 ppm for the carboxylates and up to 10-15
ppm for the strong acid anions.
Figure 6 contains precision and bias data for CE analysis of
groundwater. Precision as indicated by the relative
standard deviation of analytical triplicates is less than

10%. Bias as indicated by the percent recovery of matrix
spikes was usually within the range of 90-110%. Limits of
detection reported with the sample results corresponded to
0.5 ppm injected for the strong acid anions, and 1 ppm for
the carboxylates. The higher figure for carboxylates is due
more to sample matrix than to instrument sensitivity. The
low point of calibration curves was routinely 0.5 ppm, and
was sometimes 0.25 ppm.
The usefulness of our results for the Operating Industries
site is uncertain. The carboxylates were detected in too
few samples to be useful to trace groundwater migration.
However, chloride and sulfate may ultimately prove to be
CE appears to be a useful tool for environmental analyses if
precautions are observed to maintain confidence in
identifications. The use of a migration time reference peak
seems mandatory to us. Keeping peaks narrow by avoiding
concentrations above 10 or 15 ppm may be slightly less
important, but is recommended. In a monitoring situation
where the sample matrix is known and consistent, a
calibration range up to 50 ppm or higher may be possible.
CE is certainly a nice complimentary technique to ion
chromatography. The separations are based on different
mechanisms, and the two could be used as alternate
techniques to approach the identification of unknowns.
We wish to thank Jon Beihoffer of our laboratory and Lynn
Lutz of ICF Technology Inc. for performing GC/MS analyses of
carboxylic acids.

Figure 1. Gas Chromatograpic Conditions for GC/MS Analysis
of Carboxylates.
Hewlett Packard 5890 gas
chromatograph/Finnigan MAT 8200 mass
30 in x 0.32 mm I. D. fused silica
0.25 urn DB—FFAP
Helium at 25 cm/sec
80°C for 2 mm., 7°C/mm. to 220°C,
hold 2 mm.
22 0°C
Injector temperature:
Injection volume: 1 uL
Injection mode: splitless for 0.4 mm.
Transfer line temperature: 220 0 C
Mass range: 35-270 daltons
Cycle time: 0. sec
Ionization mode: electron impact at 70 eV
Liquid phase:
Carrier gas:
Temperature program:


Q 4 4J4.)
o 0 r4 Q.,
In . 4 0

w o
r4 () Q
t a
a) a)
OQ. a)
U .4 )
‘ r4Ø
I 4 J
i rli O
‘T .-49-l
i (1 ) 0 a)
ow we-i
•rl 4. )
4J U)
Z 0
00 )O
) 4 .— 1-
04. . ) C
tU Q40
Q4 4 U
o b u
) 4 .r4 c
o .r4
C4 U) I4
• a) O4 .
e• 4)
0 .4)

£... U-JL .P t:
1144 . )
4 -..
0 .
C ..
4 ’
4 -’
.4- 5
4- .
4 - ’
I - .
- 4
. 4 -
. 4
S ..
‘ -5
4 .
4 .4
4- ’
- . 4
. 4.
.4 .
4- .
‘ -5
: !..
3 oo
:3 50
i flutes

. 4 - I
4 . 1
4- I
04 . 1
• .0
w0 4

2.130 3. 1 30
r 4 a )
ri i.rbI.It:es
C -’
• -l
C -.
( —I
I ‘
:3.20 3 .40

p— I
- - I
. —‘ .-. . .
4 .’
4.’ ‘B
• ‘ I a. a,
• B\ .. a
A .. B-.
C .’
•, ...
• ..i. ‘. —4
a) d . ‘-4
I 4J U)
4Jr-I 4)Ifl
00 m
,-4•Q 0
Q4 )44JI-
a) 0
‘-4 0 0
t I QI
4JO 0
1 )•r4 4)
4J 4) )4
.a -.- a
q . 1
o 0.c E -’
I 4 !fl
b 4 )4 4
4) f $4
. v1
o .
$4 -4 C 0
4.’ 0
u0 4 .4-4.
a) 004.’ 0
4$44J 4.’
a) “-4
• o’4J
a) “-4 I.) C)
.44J r4. -4
•r4 ‘rI 0 U) CA
d .i-4
ri t ru J tE? S


-I 4.1
r-1 U1
•1 I )4
4.1 0 •
0, 0’
r-4 0
p 4 w 0
94 p
‘o u
4U U)
094 IhI
00)4 )
.r4r 4 0,
e4 0.d 0
Iii 494
4 - I
350 4.00
ri inute

Figure 6. Precision and Bias Results for Capillary
Electrophoresis Determination of Carboxylates and
Strong Acid Anions. All units are mg/L or percent.
Acetate Propionate Butyrate
Sample 1
Average Concentration: 697 254 324
Percent RSDa: 3.10 3.61 7.93
Matrix Spike Added: 500 500 500
Percent Recovery: 93.7 96.2 93.0
Sample 2
Average Concentration: 49 36 NDb
Percent RSD: 4.45 3.53 —
Matrix Spike Added: 250 250 250
Percent Recovery: 94.4 103 96.8
Bromide Chloride Sulfate
Sample 3
Average Concentration: ND 4073 ND
Percent RSD: - 3.92 —
Matrix Spike Added: 15000 15000 15000
Percent Recovery: 104.7 105.0 102.9
Sample 4
Average Concentration: ND 95 .9 227.9
Percent RSD: — 3.60 2.53
Matrix Spike Added: 250 375 375
Percent Recovery: 104.2 105.5 104.6
Sample 5
Average Concentration: ND 251.8 94.8
Percent RSD: — 0.89 1.00
Matrix Spike Added: 250 375 375
Percent Recovery: 111.4 113.1 110.4
RSD = Relative Standard Deviation
ND = Not Detected

71 The D -Ju iom of Hg aad Other Trace Elements in Soil Using Neutron A iv ion Analysis
L. Robinson’, F. F. Dyer’, D. W. Comba J. W. Wade’, N. A. Teasley 1 ,
I. E. Canton 3 , A. L. Ondracek and J. P.. Stokoly’
In the early 1950’s, a U.S. government facility in Oak Ridge, Tennessee, u. ed a process that
reqilred large quantities of k g. It was disclosed to the public In 1983 that about 2.4 million pounds
of Hg had boen released into the ecosystem. The primary route of nwcury into the areas outside
of the plaia was a stream, East Fork Poplar Creek, whose headwaters are in the vicinity of the plant.
Aspart ofasludyto determine the distribution ofHgaswellasAs,Cr,Sb,Se U,and Zn along the
flood plains of East Fork Poplar Creek, a procedure using neutron activation analysis was developed.
In this talk, the procedure will be described, typical sample analysis results presented, axxl quality
asnirance axl control dana willbv eiainined in detail.
1 Oak Ridge National Laboratory
2 Science Applications International Corporation
Oak Ridge Research institute
4 Great Tafr c Colleges Association Participant, Albion College
Research sponsored by the Office of Energy Research, U. S. Department of Energy, under contract
I)E-AC05440R2 1400 with Martin Marietta Energy Systems, Inc.
u wd h bi
a Is . US
sm r cs N CC-
A 6 14OC AC. aI . Is. US
G w ii.* l. t
Is. s& th.d Iuias o ià
. w o*s.is iu. 1a LLS 6 s..ai

Richard E. Phil1ips’ , Raymond J. Bath, Ph.D. , Pamela D. Greenlaw
Joseph Hudek Richard D. Spear, Ph.D. “
The field analytical service (FAS) program of the Region II U.S. Environmental Protection
Agency (EPA), Environmental Services Division, is used to support various Superfund field
programs. The FAS program provides high quality transportable analytical instrumentation for
on site project management. A thermal chromatograph/mass spectrograph (TC/MS) is used for
organic analysis in soils; an open path Fourier-transform infrared/ultraviolet (FTIRJUV )
spectrograph system is used for air monitoring; and until recently, energy dispersive x-ray
fluorescence spectrometry (EDXRF) was used exclusively for metals analysis in soils. This
presentation outlines a procedure whereby selected metals at the sub-ppm level can be
determined in ground water, surface water and soil leachates using EDXRF.
A transportable EDXRF spectrometer equipped with secondary target excitation was employed
to provide optimum sensitivity for trace metal analysis. A pre-concentration step involving the
utilization of selective chelating ion exchange resins was investigated. The resin was presented
in bulk using both column and batch techniques. The use of ion exchange resin impregnated
paper was also investigated. Variations caused by resin packing densities were corrected for by
ratioing to the appropriate target backscatter intensities.
The presentation contrasts the analytical results obtained through these various methods and
compares them to those obtained by fixed-base laboratory atomic absorption instrumentation.
‘United States Environmental Protection Agency, Region II, Edison, NJ
2 Lockheed Engineering & Sciences, 2890 Woodbridge Ave., Building 209, Edison, NJ

W11ATTVI IVALU&TI 3N 1 laXI p _g v p gz , u*z.n rj
C 0 1 C I I T I £ T X 0 I I I V I 1 0 I N I I T £ 1. 5 £ N P 1. 1 5
‘ - ‘ IC
Clifton Jones , 5.nior Inorganic Qiusist, ICY Technology, Qual.it.y Assurance Technical. Support LAboratory, Las
Vegas, Nevada 89120 WiiLi 1. 5.wb.rry III • EPAD L; Xavi.r Suares, IC ? T.chnoLo ’; Dave Dobb, LESC.
Tb. U. S. Envirorental Protection Agency (EPA) Contract LAboratory Progr (CLP) protocol. specifies direct
usbulization inductively coupl.d plasma optical emission sp.ctroscopy (ICP-O!S). hydride inductively coupled
pl a (WIICP). grap iit. furnace aton_ic absorption (GPAA) • and cold vapor stonic absorption (CVAA)
analytical techniques b. as.d for the determination of m.t*l analytes in high concentration envirormental.
s l.s. Tb. Ei$h Concentration Inorganic Statement of Work 9/90 ( I5 9/90 S ) s le pr.parsticn
procedure is a potassius hydroxide ( ) fusion method (Method 200.62-A-CLP). Solid s ies (soil.
sediments, and sludge), non-aqu.ous liquids end aqueous s Les ist mdergo the fusion smeple
preparation procedure.
Several. problems have been observed and reported with the use of the fusion method.
e Anatytes that are typicaU.y d.terminad using GFAA (Ss, Sb, Pb, Cd, As, and Ti) are difficult to
d.t.rmin. using the GFAA procedures recomended in the ICI! 6/90 S I, because of high background
effects. High background •ff.cta result iron high solids content (2.0 to 2.251) which is inherent.
in the KOB fusion method. Consequently, HYIC? has been Loy.d for to. analyses of Se • Sb, and As.
• Pb end Ti are not easily d.t.rsined using EYICP. Direct n.buLisation ICP-OES has been used for to.
analysis of Pb and TI. with decr.esed sensitivity. Tb. contract-required quantitation Limits ( )
for Pb and Ti er. 100 pg/I .. Estimated detection Limit, for Pb range from 200 to 500 ‘g/L for
s les prepared using the K fusion method snd analyzed by direct nebulisation ICP-OES; s rang. 2
to 5 tim.s greater than the CRQLs that are specified in the IJ 6/90 S J. Estimated detection
limits for Ti by direct nebuliastion ICP—OES rang. from 1000 to 3000 ig/L; a range 10 to 30 times
greater than the SCA’I—specified QL.
• Alternate s L. preparation methods that reduc. ths high back.gro nid effects so that the more
sensitive GFAA may be eeployed for to. determination of Pb or TI. are not specified in the HCIN 6/90
S 4.
Alternat, procedures were proposed for preparation of s Les analyzed under the ICIN 6/90 S I. The
alternate preparation methods selected for Phase I of the .v*luation include:
• EPA J-356 methods 3051 3013 microwave digestion for water, soils, oils, and sediments
• EPA I1101. open beaker digestion methods for water, soils and se tm—ts
• EPA 1 12101. proposed bI.ock digester method for waters, soil.a • — sediments
e Proposed microwave digestion method for total metals using hydroftuoric acid (HF)
• Proposed lithius metaborste fLux fusion method
The ICY-Quality Assurance Technical. Support (OATS) Laboratory undertook the evaluation of the 5C 5 6/90 X
fusion method, EPA methods 3051 and 3013, and 112101. open beaker digestion methods. The other three methods
were evaluated by the D L-LV/LESC group end the results of the cc thed effort will be discussed. rsport.
A c ariscn of the proposed s 1. preparation methods for possible inclusion in the ICIN S .4 as alternate
methods of s Ie preparation are presented with pre1I 4 ary conclusions. Criteria used in conducting these
evaluations include * statistical c aZison of precision end accuracy. Labor effort, and .quipoent cost.

Alan J. Spilkin , Dean Rood. J&W Scientific, 91 Blue Ravine
Rd., Folsom, California 95630 (916)985—7888
Abstract: Sample cleanup by solid phase extraction (SPE)
usually involves the retention of desired solutes followed
by their elution from the phase. Ion—exchange sorbents are
dependent on four major variables. These four variables are
the pKA difference between the sorbent and solute, solvent
pH, ionic strength and the presence of counterions. This
paper will explore the relationship of these four variables
on solute retention and elution for organic anions
(primarily phenoxyacid herbicides) on a strong anion
exchange (SAX) phase. The impact and influence of ionic
strength and counterion identity will be determined for a
series of anions with different pKa’s relative to the
sorbent. These factors will be applied to make method
development more efficient and reliable for ion exchange
solid phase extraction.
The purpose of this paper is to demonstrate a general
approach to method development for ion exchange SPE
applications. A strong anion exchange cartridge (SAX) and
phenoxyacid herbicides were used as an example.
Experiment: Three parameters, sample pH, counterion
identity and ionic strength all can play important roles in
sample retention and elution. These parameters must be
taken into consideration when developing a method for ion
exchange. Although these parameters are often
interdependent of one another, their affect on sample
retention and elution were considered separately.
The structure of J&W SAX phase is a silica based trimethyl
amino propyl substituent without endcapped silanols. In
addition to the primary ion exchange mechanism, hydrogen
bonding with surface silanols and reverse phase interactions
with bonded alkyl groups can occur.
Sample pH adjustment is often employed to ionize the sample
and phase. For ion exchange to occur both the analytes and
the sorbent must have opposite charges. Ion exchange occurs
when the pKa values for the analytes differ by at least 4
pKa units from the phase. Theoretically, 99% of a species
is ionized at 2 pH above its pKa value if it is anionic or 2
pH units below its pKa value if it is cationic.

For simplicity, ion exchange can be expressed by the
R— + PX+ = PR+ + X- Anion Exchange
R+ + PX- = PR- + X+ Cation Exchange
where R is the analyte, P is the phase, and X is the
associated ion that is exchanged for R.
Counterions are those ionic species in solution that can
compete with the analytes for binding sites on the phase.
Some counterions have stronger interactions than others.
The third parameter, ionic strength, is simply the
concentration of counterions in solution. One would expect
as ionic strength increases, the competition between the
analytes and counterions increases.
Results: Sample pH was the first parameter investigated.
To examine the effects of pH on the retention of phenoxyacid
herbicides, 1.6 g of four phenoxyacids (2,4-D, 2,4 DB,
DichiorprOp and 2,4,5 TP) were loaded onto the cartridge in
water at a pH range from 3 to 9. Since the phenoxyacids
have pKa values between 2.6 and 4.8, one would expect that
at the lower pH, most of the acid would remain neutral.
Since the pKa of the phase is about 10-11, one would expect
that the phase would become uncharged at the higher pH
range. It is important to adjust the pH of the solution
with weak counterion acids and bases to minimize the
counterion effects on retention. In this case, HC1 and NaOH
were used to pH adjust the water. Surprisingly, all
phenoxyacids retained under the varied pH conditions. The
high retention indicates that the phenoxyacids are not
solely bound by ion exchange but also by hydrogen bonding
and/or reverse phase interaction.
The other parameters affecting analyte retention are the
counterions and ionic strength. Three counterions were
examined in increasing attraction towards the phase,
acetate, chloride and citrate, at ionic strengths of 0.00114,
0.0114, 0.lM and 1.OM concentrations. The total
concentration of phenoxyacids in solution was 0.00114. Four
mLs of the phenoxyacids in the above solutions were loaded
into the cartridge and collected. The phenoxyacids with the
weak counterions, acetate and chloride, showed breakthrough
occurring above ionic strengths of 0.0114. As expected,
breakthrough in the citrate solution occurs at lower ionic
strengths of nearly 0.OO1M. Interestingly enough, the
phenoxyacids that showed highest retention, have the lowest
pKa values.

The next step in the experiment was to determine at what
point breakthrough occurs on the rinse step. Ideally, the
strongest possible solvent should be used without incurring
any breakthrough of the analytes. NaOH, a weak counterion,
in water was chosen for the rinse step. At 2 mL volumes of
NaOH exceeding 0.01 M, breakthrough of the acids was
observed. Thus, a 2 mL solution of 0.01 M NaOH was chosen.
The elution solvent for ion exchange should contain a strong
counterion to elute the analytes from the phase. In this
experiment, 2 mL citrate solutions in increments of 1% from
1—5% (w/w) citric acid in water were chosen to elute the
acids. At concentrations greater than 2%, there was not an
appreciable difference in recovery. Surprisingly, the
recovery of the 2,4D was very low, only 40% at 5% citrate.
The low recovery suggests that the 60% of the 2,4D left on
the phase is bound by silanol and/or reverse phase
Next, 100% methanol was tried as an elution solvent. Only
the 2,4DB eluted with methanol. The next step to take was
to combine the 5% citrate solution with methanol 50/50.
This combination eluted nearly 100% of all the phenoxyacids.
These results suggest that the phenoxyacids are bound by
both ionic attraction and reverse phase and/or hydrogen
The real test was whether or not this experimental method
worked on a real sample. A lake water sample was spiked
with the same amount of phenoxyacids used in the method
development procedure. The final method was as follows:
1. Condition: 3 mL of NeOH followed by 3 mL of water.
2. Load: Adjust sample to pH7 with HC1 or NaOH.
3. Rinse: 3 mL of 0.01 M NaOH.
4. ELTJTE: 1 mL of 50/50 5% citric acid (aq)/ MeOH.
The results of the deionized water compared to the lake
water sample are as follows: RECOVERIES
c vater
6 5
!4 DB
n=3 all recoveries In %
These results are as good or better than can be expected by
liquid/liquid extraction techniques.

The systematic adjustment and examination of three
variables, pH, counterions, and ionic strength, resulted in
high and repeatable analyte recovery for ion exchange solid
phase extraction. By following this logical approach,
method development for any ion exchange SPE application may
be achieved. It is important to understand that other
factors such as reverse phase and hydrogen bonding can
influence retention and elution but may not always play a

Douglas Kendall, National Enforcement Investigations Center,
U.S. Environmental Protection Agency, Denver Federal Center,
Denver, Colorado 80225.
X—Ray fluorescence (XRF) spectrometry offers many advantages
for the analysis of hazardous materials, both for field work
and for laboratory work. Two of the principal advantages
are ease of sample preparation and new, universal
calibration methods. Three examples which illustrate areas
of application of XRP spectrometry to laboratory analyses
will be presented.
Often the advantage of easy sample preparation for XRF
measurements is offset by the difficulty of calibrating the
instrument for a specific matrix. Even if the preparation
of matrix matched standards is attempted, it is impossible
to guarantee that some of the actual samples will not have
concentrations outside the range of the calibration
standards or high concentrations of elements not in the
standards. Constant progress has been made in the
calibration of spectrometers using universal standards, and
compensating for matrix effects using mathematical
corrections (fundamental parameters methods). Results from
a particular program (UNIQUANT, Philips Electronic
Instruments) will be used to show the utility of this method
for the quantitative and semiquantitative analysis of
complete unknowns.
The determination of toxic metals in soils, slags and
similar materials is a common laboratory analysis. XRF
determinations are superior in terms of sample preparation
and speed of analysis. Data will be presented to show the
comparability of XR7 to ICP and Ak determinations. XRF
determinations can be used to show whether acid digeE’tiOfls
are partial or total extractions, and to confirm ICP results
by an independent, alternate method.
The determination of sulfur and chlorine in waste oil is
important when the oil is to be burned. The XRF
determinations of sulfur and chlorine compare well with the
results from oxygen bomb combustion with an ion
chromatography or ICP finish. Sediment and other problems
with XRP determinations can be largely overcome with a thin
layer method.


Ann E. Johnson , Hydrogeologist, Richard Cariston, Programmer/Analyst, Science
Applications International Corporation, 7600-A Leesburg Pike, Falls Church, Virginia
22043; 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.
Appendix IX is the list of ground-water monitoring constituents for permitted hazardous
waste treatment, storage, and disposal facilities that are implementing the ground-water
monitoring requirements of 40 CFR Part 264 Subpart F. The Appendix IX Handbook
and Database provide specific chemical and physical properties and SW-846 analytical
methods for each of the 222 constituents listed in Appendix IX, and 19 constituents that
are being proposed or considered for addition to Appendix IX. For each constituent
listed in Appendix IX, the Handbook and Database provide the following information,
as available: Appendix IX Name; CAS Name; CAS Number; empirical formula;
maximum contaminant level (MCL); molecular weight; boiling point; melting point;
specific gravity; solubility; vapor pressure; Henry’s Law Constant; log K ; possible SW-
846 methods and their Estimated Quantitation Limits (EQLs); and chemical structure.
The Appendix IX Handbook is a pocket-size manual that contains physical and chemical
data for each of the Appendix IX constituents on a separate page. The Appendix IX
Database is a Clipper-based data management system for PCs that is function-key driven
and requires no additional software or programming ability to operate. EPA believes the
Appendix IX Handbook and Database will be useful to both regulators and the regulated
community in designing and evaluating ground-water monitoring programs,
hydrogeologic characterizations, and corrective measures, including: determining the fate
and transport of contaminants in ground water; choosing an analytical method for ground-
water analysis; evaluating corrective actions; selecting treatment methods; reviewing
RCRA Corrective Measures Studies; evaluating the potential for LNAPLs or DNAPLs
in ground water; and contaminant transport modeling.
The Handbook and Database of RCRA Ground-Water Monitoring Constituents contain
physical and chemical properties of constituents listed in 40 CFR Part 264, Appendix IX.
Appendix IX is the list of ground-water monitoring constituents for permitted hazardous

waste treatment, storage, and disposal facilities that are subject to the ground-water
monitoring requirements of 40 CFR Part 264 Subpart F.
Appendix IX was intended to be a “living” list -- to be updated and revised as analytical
methods are developed and standardized, and as research on the subsurface fate and
transport of specific contaminants advances. The constituents contained in the Handbook
and Database are those currently listed in Appendix IX, however the Handbook and
Database also contain and denote constituents that are being considered for addition to
or deletion from Appendix IX. The Appendix IX Handbook and Database provide
specific chemical and physical properties and SW-846 analytical methods for each of the
222 constituents listed in Appendix IX, and 19 constituents that are being proposed or
considered for addition to Appendix IX.
Purpose of the Handbook and Database
The Appendix IX Handbook and Database are designed to provide a concise and readily
accessible resource for ground-water professionals who are designing and evaluating
ground-water monitoring programs, hydrogeologic characterizations, and corrective
measures, including:
• Evaluating the fate and transport of contaminants in ground water;
• Choosing an analytical method for ground-water analysis;
• Evaluating corrective actions, selecting treatment methods, or reviewing
Corrective Measures Studies;
• Evaluating the potential for LNAPLs or DNAPLs in ground water;
• Contaminant transport modeling.
Computer and Software Requirements for the Database
The Appendix IX Database is a Clipper-based data management system for PCs that
requires no additional software or programming ability to operate. Hardware
requirements for operating the system are as follows:
• 4 Megabytes of hard drive space;
• 64OKbofRAM;

• DOS 3.X or higher;
• Optional color monitor.
Contents of the Handbook and Database
Figures 1 presents an example of the format of the Appendix IX Handbook. Figure 2
is an example of a display screen from the Appendix IX Database. For each constituent
listed in Appendix IX, the Handbook and Database provide the following information,
as available:
1. Appendix IX Name: The name of the constituent as it appears on the list of
ground-water monitoring constituents listed in 40 CFR Part
264 Appendix IX. The Handbook is organized
alphabetically by Appendix IX name.
2. CAS Name: The name of the constituent as it appears in the Chemical
Abstracts Service (CAS) Registry.
3. CAS Number: The CAS registry number.
4. Empirical Formula: The chemical formula that provides the number of each
type of atom in the molecule. For metals, the Handbook
and Database provide the chemical symbol.
5. Maximum The maximum permissible level of the constituent in water
Contaminant which is delivered to the free flowing outlet of the ultimate
Level: user of a public water system. If there does not exist a
maximum contaminant level for the constituent, the
notation “NA” is presented.
6. Molecular Weight: The molecular or formula weight of the constituent in
7. Boiling Point: The temperature in degrees celsius at which the vapor
pressure of the constituent in aqueous form is equal to
atmospheric pressure.

CAS Name:
CAS Number
Empirical Formula:
Molecular Weight:
Melting Point:
Boiling Point:
Vapor Pressure:
Specific Gravity:
Henry’s Law Constant:
CAS Name:
CAS Number
Empirical Formula:
Molecular Weight:
Melting Point:
Boiling Point:
Vapor Pressure:
Specific Gravity:
C H 12
0.0002 mgIL
252.32 g/moi
179-1 79.3C
5x10 mm Hgat2tTC
1.351 [ UT)
3.8x10 mgILat25°C
Henry’s Law Constant: 4.9 x 1007 atm • m 3 /moi at 25°C
Possibie SW-M6 Analytical Methods
Muthod EOL4igIL)
Benzo [ g, h,i]perylene
Benzo [ g,h,i)peryleno
191 -24-2
C 22 H 12
Benzo(a)pyrene _______________
276.34 g/mol
222 C
> 500 C
1 x10 10 mm Hgat2OC
2.6x 1004 mg/Lat25°C
1.4 x 10 .01 atm • m’ 3 /mol at 25°C
Possibie SW-M6 Analytical Methods
Method EOL(pgIL)

Molecular Weight
Boiling Point
Melting Point
Specific Gravity
@ 25C
E 03
Vapor Pressure
Henry’s Law Constant
atnvm ’3/mol
Log Kow
Appendix IX Name 2.4,5-Trichiorophenol
CAS Name Phenol, 2 ,4,5-trichloro-
CAS Number 95-95-4
Empirical Formula C6—H3—C13—0
Possible SW-846 Methods
Help I
NA mg/L
Method: 8250 EQL: NA ug/L
a: Ret
erence List
I F. ; vie

8. Melting Point: The temperature in degrees celsius at which the constituent
in solid phase is in equilibrium with the liquid phase at
atmospheric pressure.
9. Specific Gravity: The ratio of the density of the constituent at a specified
temperature relative to the density of water at a specified
10. Solubility: The concentration of the constituent in milligrams per Liter
that is required to form a saturated solution in water at a
given reference temperature.
11. Vapor Pressure: The pressure in millimeters of mercury of the vapor phase
of the constituent that is in equilibrium with its liquid or
solid phase.
12. Henry’s Law The ratio of the equilibrium concentration in atmospheres
Constant: of the constituent in air relative to its concentration in
moles/cubic meter in water.
13. Log K : The log of the ratio of the equilibrium concentration of the
constituent in octanol relative to its concentration in water.
14. Possible SW-846 The analytical methods presented in the USEPA publication
Methods: “Test Methods for Evaluating Solid Waste” (SW-846),
Third Edition (as amended by Updates I and H), that may
be used to determine the concentration of the constituent in
ground water. For each method, an Estimated Quantitation
Limit (EQL) is given if an EQL is available. When
different EQLs exist for 5 and 25 mL sample sizes, the
EQL for the 5 mL sample size is provided in the Handbook
and Database. When the Handbook and Database present
two EQLs for one analytical method, the two EQLs
represent the EQLs obtained using two different detector
15. Chemical Structure: The chemical formula written to show the relative
arrangements of the atoms. For metals, the Handbook and
Database provide the chemical symbol.

The information contained in the Handbook and Database is from a wide variety of
recently available and standard references, including the following:
• USEPA. 1992. RREL Treatability Data Base , Version 4.0. USEPA,
Glenn Shaul.
• Weast, Robert C., ed. 1986. CRC Handbook of Chemistry and Physics ,
67th Edition. CRC Press, Inc., Boca Raton, Florida, 2406 pp.
• Budavari, Susan, ed. 1989. The Merck Index , Eleventh Edition. Merck
& Co., Inc., Rahway, New Jersey, 1606 pp.
• Verschueren, Karel. 1983. Handbook of Environmental Data on Organic
Chemicals , Second Edition. Van Nostrand Reinhold Company Inc., New
York, New York, 1310 pp.
• Aldrich Chemical Company. 1990. Aldrich Catalog Handbook of Fine
Chemicals . Aldrich Chemical Company, Inc., Milwaukee, Wisconsin,
2150 pp.
• Montgomery, John H. and Linda M. Welkom. 1990. Groundwater
Chemicals Desk Reference . Lewis Publishers, Chelsea, Michigan, 640
• Howard, Philip H., ed. 1991. Handbook of Environmental Fate and
Exposure Data , Volumes I and Ill. Lewis Publishers, Chelsea, Michigan.
• Yaws, Carl, Haur-Chung Yang, and Xiang Pan. November 1991.
Henry’s Law Constants for 362 Organic Compounds in Water. Chemical
Engineering, vol. 98, no. 11, pp. 179-185.
Operating the Database
The Appendix IX Database is completely function-key driven, and requires no
programming ability to operate. When the Database is accessed by typing “APX9” at
the DOS prompt, the first screen that appears is a brief description of the Database. The
user can read the introductory screen using the cursor or by paging down, or access the
main menu by pressing the escape key . The main menu of the Appendix IX
Database is the list of 241 constituents in the Database, listed alphabetically by Appendix
LX name. With the  key, the user can sort the constituents alphabetically!
numerically by toggling between Appendix IX name, CAS number, or CAS name (see

Figures 3 a, b, and c). To view data for a particular constituent, the user highlights the
constituent either by using the cursor or the  and  keys to
move through the list of constituents, or by completely or partially typing the Appendix
IX name, CAS name or CAS number of a constituent. Data for the highlighted
constituent is retrieved by hitting .
Figure 2 shows a typical data screen for an Appendix IX constituent. The Database is
capable of displaying multiple values for specific gravity, solubility, Henry’s law
constant, log K , and possible SW-846 analytical methods and EQLs. Multiple values
are accessed by highlighting the displayed value using  or 
and moving the cursor up or down for additional values, as indicated by an arrow (see
Figure 4). The  key takes the user back to the main list of Appendix IX
The Database has several other functions that may be accessed from the data screen. To
view the constituent’s chemical structure, the user presses the  key while in the
data screen (Figure 5). References for chemical and physical properties are indicated in
the Database by number. The user may view specific citations by pressing the 
key for the Reference list (Figure 6a) while in the data display screen for a specific
constituent. When a specific reference is highlighted, the user presses  for the
full citation (see Figure 6b). The user may obtain help in using the database (e.g.,
explanations of movement keys, descriptions of function keys, and abbreviations) by
hitting  while in the data screen or main menu.
The Appendix IX Handbook and Database were developed to provide ground-water
professionals easy access to commonly used physical and chemical properties of the
RCRA ground-water monitoring constituents. The Appendix IX Database provides
chemical and physical properties of the Appendix IX constituents in a user-friendly
system designed to operate on minimally equipped PC systems.
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.

05/29/92 Appendix IX Chemical Constituents Database System 09:06:09
Version 1.0
2,3 • 4, 6—Tetrachl orophenol
2,3. 7, 8-TCDD
2,4,5-T; 2,4,5-Trichlorophenoxyacetic acid
2,4, 6-Tn c l ii orophenol
2.4-D; 2,4-Dichlorophenoxyacetic acid
2, 4-Di clii orophenol
2,4-Dimethyiphenol; m-Xylenol
2, 4-Di ni trophenol
2, 4-Di ni trotol uene
120-83 —2
5 1-28—5
Appendix IX Chemical Constituents Database System
Version 1.0
2,2’—methylenebis [ 3,4,6-tric h loro—
2,3,4,6-Tetrac hloro-
2, 4-di chi oro-
2, 6-di chl oro—
Appendix IX Name
CAS Number
v : Help F2: Reindex
Toggle Sort I Enter:
,t Item I
CAS Registry Name
CAS Number
T ’ He1pI :Reindex f f3:T -- I

05/29/92 Appendix IX Cheiiical Constituents Database Syst n 09:06:21
Version 1.0
ii Molecular Weight
Boiling Point
Melting Point
Specific 6ravlty
Vapor Pressure
Henry ’s Law Constant
Log Kow
1.678 25C
1.19 E 03
2.9 E—05
1.76 E-07
Method: I
CAS Ntaither
Appendix IX Name
2-Chl orophenol
1 ,2,4-Trimethylbenzene *
1,2,4, 5-Tetrachlorobenzene
2,4, 5—TricMo upheno1
Endosul fan I
1, 2-01 bramo-3-chl oropropane; D8CP
I ,2 ,3-Tri chioropropane
Ethyl methacrylate
tert-Butyl benzene *
Isopropylbenzene *
rt: iii.,.. t.2: Rein_
Appendix IX N
CAS Na,e
2,4, 5—Trichlorophenol
Phenol, 2,4,5-trichloro-
CAS Nt er
E ir$cal For*ila
C6-H3-Cl 3-0
Possible SW-846 Methods
F... .. ...ice List
NAmg/t. --
10 ug/L --
- n I/I ) Esc



77 Rationale for the Design of Cost-Effective Groundwater Monitoring Systems in
Limestone and Dolomite Terranes: Cost-Effective as Conceived is Not
Cost-Effective as Built if the System Design and Sampling Frequency
Inadequately Consider Site Hydrogeology
James F. Quinlan, Principal Hydrogeologist, Quinlan & Associates, Inc., Box 110539, Nashville,
Tennessee, 37222; Gareth J. Davies, Hydrogeologist, Ozark Underground Laboratory, Protem,
Missouri 65733; and Stephen R. H. Worthington, Instructor, Department of Geography,
McMaster University, Hamilton, Ontario, L8S 4K1, Canada
Abstract: If one assumes that the purpose of a monitoring well is to detect the presence or absence
of contaminants draining from a facility, if one has the goal of doing so reliably and successfully,
and if one must accomplish this goal in carbonate rocks, then it cannot be accomplished by the
conventional technique of drilling several down-gradient wells — except by improbably good luck.
Where there is carbonate rock, in most settings there is also karst and a karst aquifer. The only
way to design a reliable monitoring system for a facility in a karst terrane is to determine where it
drains to. Such drainage can only be established by well-designed, properly performed, and
correctly interpreted tracer tests to springs and cave streams to which the facility drains and to wells
which intercept their flow lines. [ Such interception is difficult to accomplish and must be proven
by tracing.] In a karst, any so-called monitoring point, until and unless it has been shown by
tracing to function as such, is a monitoring point in name only. The springs to which a karst
aquifer (and a facility) drains may have significant storm-related variation in discharge and water
quality. Sampling of these springs and other monitoring points must be done at a frequency that
reliably senses these variations. Regular annual, semi-annual, or quarterly sampling is incapable of
doing so — again, except by improbably good luck.
The cost-effectiveness and reliability of a monitoring system in a karst aquifer is directly
proportional to the extent to which its hydrogeology and flow dynamics are understood. The
hydrogeology must be studied on a regional scale. One of the aquifer parameters useful for the
design of a monitoring system in any type of aquifer is estimated flow velocity, a parameter that can
be used to estimate hydraulic conductivity at at a regional scale. In carbonate aquifers the
measurement of hydraulic conductivity is scale-dependent over many orders of magnitude and
directly proportional to the scale of the investigation, as shown in Figure 1. When the
representative elementary volume [ REV] is at the scale of a pumping test, such well tests can be
excellent for evaluating water supply potential of an aquifer. But when concern is with contaminant
movement. then the whole catchment is the appropriate scale of investigation. Only a tracer test can
represent aquifer-scale worse-case contaminant velocities.
Current RCRA [ Resource Conservation and Recovery Act] and CERCLA [ Comprehensive
Environmental Response, Conservation, and Liability Act (Superfund)] regulations require the
installation of wells for groundwater monitoring of all sites — regardless of their rock types. The
following assumptions are implicitly made by designers of monitoring-well systems:
1. The aquifers are isotropic and homogeneous.
2. Flowisinaplumetobeintersectedbythewell.
3. Flow is dispersive, as a result of advection, hydrodynamic dispersion, and molecular
These assumptions are generally valid in granular aquifers and in many fmely fractured aqui-
fers. In limestone and dolomite (carbonate) terranes— nearly all of which are karat terranes and
have karst aquifers — these three assumptions are either presumed to be valid or an alternative
assumption is made:

4. Flow is to a cave stream (often more than one) that:
a. Drains the site to be monitored.
b. Is intercepted by the well.
In most karst terranes, the first two assumptions are always wrong. Assumption 3 is correct.
Dispersion may occur in the X,Y, and Z directions and will be predominant in one of them;
diffusion into the rock matrix is usually trivial, but there are exceptions.* Assumption 4a is usually
right., but 4b is almost always wrong.
It is appropriate to define the teims karst and karst aqu fer
Karst: A landscape and its subsurface characterized by surface water flow and
groundwater flow through caves (or other dissolutionally enlarged cavities) and a variable
suite of distinctive surface landforms and hydrologic features. These include the following
nine types of features, any one of which is diagnostic of karst:
1. Sinkholes (any closed depressions, with or without a discrete opening at their
bottom, formed by dissolution and/or collapse of bedrock, with flushing and/or
collapse of soil into a subjacent cavity), and internal drainage to them
2. Dry valleys (in humid climates)
3. Spiir gs (draining carbonate, sulfate, or halide rocks)
4. Caves (open to the surface or encountered by drilling)
5. Sinking streams (that sink at a hole known as a swallet)
6. Dissolutionally enlarged joints and/or bedding planes (as seen in cores or outcrops)
7. Grikes (soil-filled, dissolutionally enlarged joints or grooves; also known as
cutters or soil karren)
8. Karren (dissolutionally, subaerially, water-carved grooves on rock; commonly
9. Significant loss of fluid circulation during drilling of wells
Karst aquifer: An aquifer in which flow of water is or can be appreciable through one or
more of the following: joints, faults, bedding planes, pores, cavities, and caves — any or
all of which have been enlarged by dissolution of bedrock.
Most karsts are developed in carbonate rocks, but they also develop in gypsum, salt, carbonate-
cemented sandstone, and (to a trivial extent) in other rocks.
A few examples of karat features 1 through 5 may be shown on a published topographic map,
but most of them and features 6 through 9 can be found only by field work. If one looks in the
field, and does so diligently, some of these nine features, perhaps all of them, will be found.
Experience has repeatedly shown that if there is carbonate rock, there is almost certainly some type
ofkarstandakarstaquifer (Quinlanetal., 1992).
The apparent lack of some or all of the above nine karst-diagnostic features in a carbonate ter-
rane, or their non-recognition, does not mean that a karat is not present. It commonly means that
more field work or drilling is needed to find them. Similarly, the presence of just one of the fea-
tures in areas where there are few outcrops is diagnostic of a karat because that one is usually
representative of a much larger population. This is true because the scale of sampling is so coarse
relative to the actual distribution of the karat features. The statistics of searching for (and finding)
* If the mean flow time is less than a month, the tracer has had insufficient time to penetrate deeply
into the matrix. Nevertheless, diffusion cannot be ignored in the analysis of dye recovery
(Maloszewski and Zuber, 1990, 1985). Correct interpretation of diffusion in moderate- to long-
term tests is complicated, however, by the effects of bacterial decomposition on tracer dyes (Aley
et a!. 1993).

many of the more subtle and sparsely distributed features diagnostic of karst favor failure to find
Useful introductions to karst hydrogeology are the texts by Milanovic (1981), White (1988),
Ford and Williams (1989), and Kresic (1993).
The consequences of accidental spills or of intentional or permitted discharges of harmful
materials into karst aquifers are in many ways more severe than into other aquifers. Travel time is
usually extremely rapid; attenuation is usually minimal (Field, 1988). Flow velocities in many karst
terranes vary as much as 10 to 1500 ft per hour (- 0.001 to 0.1 m/s) between the same two points
— the latter, in response to storms — and are tens of thousands to several million times faster than
those characteristic of many granular aquifers. The flow velocities are thousands of times more
rapid than slug or pumping tests in wells in carbonate aquifers would indicate (Kiraly, 1975;
Sauter, 1992; Smart et at., 1992). Consequently, one can study karat aquifers more readily and
interpret real data rather than synthetic data which are honorably (but sometimes erroneously)
produced by computer simulations of flow in karst aquifers. Consequently also, there are
groundwater monitoring problems and welihead delineation problems unique to wells and springs
in karst terranes. For monitoring, these problems include the following:
1. Where to monitor for pollutants
2. Where to monitor for background
3. When to monitor for pollutants and background
4. How to reliably and economically determine the answers to problems 1,2, and 3.
We do not know why there is so little recognition of the unique hydrologic nature of karst in
texts concerning hydrology, hydrogeology, groundwater monitoring, hazardous-waste site
management, and remediation (as discussed by Quinlan et a!., 1992, p. 582-583), but the fact
remains that karat is usually ignored by most engineers and hydrogeologists, be they regulators or
consultants — in the U.S., Canada, and much of Europe. Accordingly, most of these
professionals, whatever their education, lack formal training in karat hydrogeology and rarely have
the opportunity to learn about it. Many are unaware of the significance of karst. Too many deny its
existence or importance at a given site. Yet approximately 40% of the U.S. east of Tulsa is
underlain by carbonate rocks; most of this 40% is karstic.
Our experience has repeatedly shown that unless special approaches to the study of facilities in
carbonate terranes are taken, their monitoring programs yield data that are irrelevant to the purpose
of monitoring — the denial by professionals of the need for such special approaches
notwithstanding. This paper and publications by Quinlan (1989, 1990), Aley (1986), Ewers
(1992), Alexander et a!. (1992), Edwards and Smart (1989), Smart (1985), Field (1988), and
Quinlan and Ewers (1985) describe these approaches. They have been written because there is
overwhelming proof that most carbonate rocks are karstified. This karstification affects the validity
of monitoring efforts in them and it requires special approaches.
The lack of reliable monitoring in karst aquifers has many origins. These include:
1. Many of the regulators who wrote the RCRA and CERCLA regulations were unfamiliar
with the complexities of karst.
2. Sometimes there is an unthinking literal interpretation of the regulations, both by regulators
and by consultants.
3. Some attorneys (and others) tend to become uncomfortable about the implications of off-
site monitoring that is sometimes all that is economically feasible.
4. Some consultants, fearful of losing a client’s future business, report only that which they
think the client wants to hear. They intentionally suppress potentially embarrassing facts
and/or ignore clues that may lead to them.

Porosity is the percentage of the total volume of rock or sediment that is void space. In
limestones and dolomites, the initial or primary porosity can be high or low, depending on the
depositional environment in which the rock formed, and the subsequent history of burial and of
exposure to the atmosphere and in-sins dissolution and recrystallization (Scholle et al., 1983;
Tucker and Wright. 1990). Changes in depositional environment during sedimentation, such as
interruptions caused by alterations in sea level, or temporaiy fluctuations in water depth or energy,
cause differences in rock texture, generally as partings or distinct bedding planes in what became
the rock.
As these rocks are eventually uplifted and as overlying rocks are eroded, stresses are released
and vertical fractures known as joints develop, commonly normal to bedding planes. Faults
develop where there was movement of rocks relative to each other. Fractures and faults comprise
secondary porosity.
Carbonate rocks are soluble in meteoric waters, most of which are slightly acidic, and most
carbonate rocks contain sulfide or sulfate minerals that oxidize to produce sulfuric acid in situ, thus
enhancing dissolution of the rock. As the rocks are subaerially exposed, water infiltrates and
dissolution along joints, faults, and bedding planes creates tertiary porosity. Such dissolution
creates flow pathways for groundwater, thus producing a karst aquifer (White, 1988; Ford and
Williams, 1989). Exposed bedding planes and joints are readily exploited by chemically aggressive
meteoric waters that produce underground flow paths with a length that may be many tens of miles.
They form an integrated network of shafts, tubes, fractures, and conduits (Palmer, 1991; White,
1988; Ford and Williams, 1989; Worthington, 1991).
If a unit volume of rock is assumed, the efficiency of the movement of groundwater through the
void spaces in it is a function of the degree of interconnection of the void spaces. A measure of this
efficiency, the hydraulic conductivity, is defined as the volume of fluid (usually water) that will
move through a medium in a unit of tune under a unit hydraulic gradient through a unit area
measured perpendicular to the flow.
Laminar flow in a granular rock such as a sandstone with pores and narrow fissures can be
described by Daity’s Law:
where Q is the discharge, K is the hydraulic conductivity, i is the hydraulic gradient, and A is the
cross sectional area.
Laminar flow through fissures or na11 tubes is described by the Hagen-Poiseuille equation:
Q it pg r 4 i/(8 11)
where p is the fluid density, g is the acceleration due to gravity, r is the radius, and r is the dynamic
viscosity of the fluid. Flow in most carbonate aquifers is often in pipe-full conduits and is
turbulent. Accordingly, it can be described by the Darcy-Weisbach equation:
Q (8gRiA 2 / 05
where R is the hydraulic radius of the conduit (area + wetted perimeter), A is the passage cross-
sectional area, and f is the friction factor. Fissure or conduit width has the greatest influence on
discharge in both the Hagen-Poiscuille and the Darcy-Weisbach equations. Note that K is not
relevant to flow described by either of tbese equations.

Older carbonate rocks (pre-Cenozoic) have been subjected to long periods of diagenesis which
has generally reduced their porosity to a few percent at most. In these rocks, flow is most likely to
be concentrated along Joints, faults, and bedding planes. Solutional enlargement produces conduits
which may have cross sectional areas large enough to pennit turbulent flow.
Conversely, karstification is less likely in younger carbonates where lack of burial has allowed
high porosities (l0%+) to be maintained. For instance, the limestones of Florida and Puerto Rico
are Cenozoic, have high porosities, and have considerable flow through their pores. Nevertheless,
aquifer characterization, using Darcy’s Law, has only been successfully accomplished in the
unconfmed parts of the Floridan aquifer by using “coarse-mesh” digital models on a regional scale
(Johnston and Bush, 1988). In those areas in central and northern Florida, flow is to a series of
major spnngs. Scuba divers have penetrated some of these springs, following water-filled conduits
for distances of up to 3 km. sometimes at depths in excess of 100 m below the water table (Farr,
1991). One system has 16 km of mapped underwater passage accessible from a total of 19
entrances and to a depth of 75 m (Irving et a!., 1992). Other mapped systems are equally
impressive (Wilson and Shies, 1988). Such daring exploits are compelling evidence that conduit
flow is widespread in the unconfined parts of the Flondan aquifer.
The successful application of Darcy’s Law to well-based groundwater monitoring in sand and
gravel aquifers, and application of it to computer modelling of flow, does not mean that the same
techniques can be applied to fractured-rock and karat aquifers. Most certainly, there is
overwhelming evidence that computer models cannot be applied successfully to most karst aquifers
(Palmer, 1992; Teutsch and Sauter, 1992; Sauter, 1992). By the time one has enough data to
meaningfully model a karat aquifer, one no longer needs the model. For example, the work by
Teutsch and Sauter (cited above) is based, in part, on 12,000 data values acquired during 10 years
of geologic, hydrologic, and tracing studies.
An aquifer in sand or gravel can be reasonably assumed to have hydrologic characteristics that
can be described using Darcy’s Law. Although nature never produces laboratory-like conditions
similar to the sand-filled vessels used by Darcy, an assumption can usually be made with some
confidence that flow is laminar and that conditions throughout the aquifer are approximately
isotropic and homogeneous. A representative elementary volume (REV) of aquifer material would
be relatively small because it could be reasonably assumed that, at the small scale of a 6-inch
diameter monitoring well, hydrologic conditions in the aquifer would be adequately described, or
the continuum approach would work (Domenico and Schwartz, 1990, p. 84). Therefore, in most
sand and gravel aquifers, monitoring wells are generally reliable — if they intercept the plume and
if they are screened in the proper interval.
The only void spaces in most fractured-rock aquifers which are significant for flow are the
fractures. [ For simplicity we deliberately ignore here the important concept of dual porosity
(Maloszewski and Zuber, 1990; Wang, 1991; Wheatcraft and Cushman, 1991).] The density of
these fractures would dictate whether a 6-inch monitoring well would adequately describe
hydrologic conditions assuming, for instance, that the well had a 5-ft screened interval. The
problem would be how many fractures (in three dimensions) would the screened interval in the well
intersect? Also, would all of these fractures produce water? To have confidence in the efficiency of
the well, the REV would obviously have to be much larger than in a sand and gravel aquifer
because, at the same scale of a 6-inch monitoring well, it is very likely that far fewer fractures than
equivalent pore spaces in a sand and gravel aquifer would be intersected.
It should be stressed that the hydraulic conductivity of the rock matrix of most fractured-rock
aquifers is essentially trivial when compared to flow in a single fracture. Both Darcy flow
conditions and Hagen-Poiseuille conditions would exist. In certain situations, flow could also be

turbulent. Wells are often less reliable for groundwater monitoring in fractured-rock aquifers than
in granular aquifers because their distribution (their density or number per unit area) is sometimes
insufficient and because their screened intervals are long relative to the total width of fractures they
intersect. However, if a sufficiently large-scale approach is adopted, it can be reasonably assumed
that the aquifer approximates a porous medium, and hydrologic parameters can be determined as for
a granular aquifer.
A significant difference between fractured-rock and karat aquifers is the development of tertiaiy
porosity in the latter, as already described. Not only are fractures utilized by groundwater flow, but
specific sets of them are dissolutionally enlarged by it. In fact, there is constarn dissolutional
modification happening in a flow-field. Thus, over time, any one part of the aquifer may evolve
from flow through pores (Darcy’s Law) to laminar flow through fissures (Hagen-Poiseuille
equation) to turbulent flow through conduits (Darcy-Weisbach equation) (Worthington, 1991).
In these circumstances, the REV is not on the macroscopic scales (centimeters to meters) which
are valid in granular aquifers where computer modeling, using Darcy’s Law, is reliable. Indeed,
the REV of karat aquifers is equivalent to the entire groundwater basin (Ford and Williams, 1989,
p. 21 1). This can be demonstrated by testing hydraulic conductivity at different scales.
In a granular aquifer, hydraulic conductivity [ K] is independent of the scale of measurement.
Thus data from rock cores taken from wells and from in situ slug or pumping tests will all yield
approximately the same value of K, even though K may range over several orders of magnitude
within the same aquifer (Dykaar and Kitanidis, 1992). However, variations in K with scale of
investigation do occur in fractured rocks (Domenico and Schwartz, 1990, p. 84-87; Clauser,
1992). Such variation is even more pronounced in karst aquifers, as has been convincingly
demonstrated by studies in Switzerland (Kiraly, 1975), Germany (Sauter, 1992; Teutsch and
Sauter, 1992), and in Great Britain (Smart a aL, 1992; Edwards a a!., 1992). Averages from
these and other data are shown in Figure 1. The hydraulic conductivity as measured by pumping
tests is also dependent upon their duration (Sireitsova, 1988, p. 366; Sauter, 1992).
The smallest scale shown in Figure 1 is that of a rock core taken from water wells (Kiraly,
1975; Sawer, 1992) or from petroleum wells (Archie, 1952; Murray, 1960). Mean K values of
about iO- 7 to io- rn/s in carbonates are similar to those for sandstone aquifers and reflect flow
through pores.
Double packer tests measure the flow and pressure in a small interval of a well isolated by
means of seals (packers). Slug tests involve the recovery of a well from an induced sudden change
in its water level. Both of these methods yield K values which represent flow over a distance of
some meters and in karst yield values that are routinely 100 to 1000 times greater than K values
from cores in the same aquifers (Sawer, 1992; and Figure 1). These higher values reflect flow in
small fissures.
Pumping tests involve removal of water from a well and produce K values wh’ch represent
flow over tens to hundreds of meters. At this larger scale, larger fissures and small conduits are
often encountered, so K is typically ten times larger than from slug tests in the same wells (Smart et
a!., 1992; and Figure 1).
The largest scale of investigation is that of a complete catchment (typically 1 to 100 km in
length). Water movement at this scale in karst aquifers is measured by tracer tests. Such tests yield
velocities rather than hydraulic conductivines. From the Darcy equation it can be shown that these
two parameters are related by:
v = K I / Sy
where v is water velocity, i is the hydraulic gradient, and Sy is specific yield, the proportional
volume of the rock from which water will drain freely in an unconfined aquifer (Worthington,

1992). Typical values in karst aquifers aie I = 0.001 to 0.02 and S , = 0.005 to 0.02, so v and K
are approximately equivalent (Worthington, 1991; Quinlan and Ray, 1989; Smart et al., 1992).
Nevertheless, derivation of K-values from tracer-test velocities is not valid because conduit
velocities are almost always beyond the range of validity of Darcy’s Law (Ford and Williams,
1989, p. 144; Worthington, 1992). Hundreds of tracer tests in karst have been conducted over
distances >10 km, with dye injection into sinking streams and dye recovery at springs. The
majority of these tests have yielded velocities in the range 0.OOlto 0.1 m/s (Ford and Williams,
1989; Worthington, 1991; Quinlan and Ewers, 1989). These tracer velocities are, on average,
some 10 to 1000 times faster than pumping tests in these aquifers would indicate (Figure 1).
100 • Predominant Range 100 .
• in the Same Aquifer E
0 I Range Reported
102 5
E 102 I in the Literature
__________ z
>- __________
Dye Tests
iO- in Conduits

C) C D I-.
10.8 A Core (Lab) Tests 10-8
I I B B Double Packer Tests
C Slug Tests
10.10 0
A D Pumping Tests .
E Dye Tests
1012 I 1012
0.01 01 1 10 100 1000 10,000 100,000
Figure 1. Range in hydraulic conductivities of carbonate aquifers as a function of scale of
investigation. REV = representative elementary volume. The data shown by the heavy bars are
from a Jurassic karst aquifer in the Swabian Aib of Germany, as described by Sauter (1992) and
Teutsch and Sauter (1992). Other data are from references cited within text. The mied rectangle
shows the range of velocities occurring in dye tests from sinking streams to springs and is based on
1405 values from 25 countries (Worthington, 1992). This figure shows that the hydraulic
conductivity of an unconfined karst aquifer is extremely dependent upon the scale of measurement.
Also, tracer-test velocities, other than those which are from well-to-well, can be used as an
estimator for worst-case aquifer-scale contaminant velocities. [ Modified after Kiraly (1975) and
Sauter (1992)]

The conclusion to be drawn from these results is that a hydraulic conductivity or velocity test in
karst yields values that are only valid for the scale of the test, and that extrapolation to a larger scale
will underestimate the mie values by several orders of magnitude. These results are semi-consistent
with those of Clauser (1992) for crystalline rocks, but whereas our Figure 1 shows the regional
data on a continuation of the same trend, his Figure 1 shows the regional data on a plateau that is
not a continuation of the initial trend. We interpret this plateau to be an artifact of the data and
where they are subjectively plotted. There is no proof that any of his values are truly regional in
To the best of our knowledge, no unconfined carbonate aquifer has ever been demonstrated to
have similar range in K values at the core, packer test, slug test, pumping test, and tracer test
The conceptual model that can be built up from the above tests and from the thousands of tracer
tests that have been performed is that flow in carbonate rocks forms a tiibutaiy system. Percolation
into the soil and down through the pores and narrow fissures in the uppermost bedrock proceeds at
low velocities (10-8 to 10-2 mIs). Once the flow reaches a conduit, it then proceeds rapidly (10-sto
10’ mis) to one or more springs. These dual ranges in velocity describe why the residence time of
some of the dye (and pollutants) within an epikarst may be 10 to 100 or more times longer than
when transit through an aquifer is solely from a sinking stream to a spring (Quinlan and Ray,
1992). Many, or possibly most, karat conduits terminate in distributary systems, so that flow from
one conduit may reappear at two or more springs (Quinlan and Ray, 1989; Quinlan and Ewers,
1989; Worthington, 1991).
The low-velocity component of a carbonate aquifer system is short-circuited where there are
sinkholes or sinking streams. In such cases, flow (and pollutants) have direct access to the conduit
system, so that a toxic spill into a sinkhole may penetrate several miles into an aquifer within a
period of hours. A graphic and serious example of high-velocity flow occurred at West Plains,
Missouri, in 1978. The collapse of sewage lagoons there on May 28 released 94,000 m 3 of raw
and partially treated sewage into the groundwater system. Some of the dye injected into the
disappearing effluent reappeared twelve days later at Mammoth Spring, Arkansas, which is 23
miles away from the spill site. At least 700 illnesses were reported by residents drinking
groundwater contaminated by this spill (Vineyard, 1981).
Wells are typically very unreliable monitoring devices in karst aquifers (Quinlan and Ewers,
1985; Quinlan, 1989, 1990; Ewers, 1992). At the scale of wells, features such as conduits are very
difficult to find, even if they are quite large; Benson and La Fountain (1984) state that statistically it
takes 1000 3-cm drill boles per acre to intercept a 2.5-rn diameter cavity with 95% probability. The
number of randomly or grid-fixed drill holes necessary to intercept a conduit with the same diameter
would be smaller but would still be prohibitively expensive.
The only realistic axiom regarding conceptualization of a karst aquifer should be: Few, if any,
assumptions about aquifer mechanics will be valid prior to measurement and tracing. It therefore is
equally realistic to assume that conceptualization will be an evolutionary process and that many
parameters have to be measured and field observations made before an accurate conceptualization is
achieved. Indeed, many years of karat research have revealed that collection of data unfettered by
assumptions about specific hypotheses is one of the more dependable tools in karst aquifer studies.
if karat groundwater is to be understood, tracer tests are — without exception — necessary. Their
use requires no assumptions about the aquifer to be valid; the tracer goes where it will. The only
assumptions that have to be made relate to the tracing technique, test design, test execution, and test

It is not appropriate to assume that a karst aquifer does not exist in a carbonate rock because
wells or cores do not indicate dissolution, or because Darcy conditions seem to be indicated, or
because karst landforms are seemingly not present or not significant. Data from many countries of
the world supply overwhelming evidence that where there is karst, there is also a karst aquifer
(Quinlan et al., 1992).
Recharge in karst is either directly onto a carbonate rock terrane (autogenic recharge) or from
nearby non-carbonate rock terrane (allogenic recharge). Most allogenic recharge also eventually
becomes concentrated at joint intersections where swallets (the sinking points of sinking streams)
form; alternatively, autogenic recharge is distributed through many joints (Ford and Williams,
1989). A highly corroded zone of the upper bedrock (usually below residual soil), known as the
epikarst, is also very important. Maximum fracture utilization is achieved to a depth of about 10
m, but the most efficient vertical drains below are formed only at intersections of master joints
(Williams, 1983). Recharge effects are veiy important because they are a major cause of water-
quality variation at springs (Newson, 1971).
Storage occurs in soil, seasonally flooded conduits and fractures, and in perennially saturated
fractures, and is especially important in the epikarst. Storage effects in karst aquifers are difficult to
quantify, although attempts have been made using spring recession and aquifer water-level analysis
(Atkinson, 1977), and time-series (spectral) analysis (Mangin, 1984). Storage can very severely
affect water quality for an entire groundwater basin because contaminants can be temporarily
trapped in many fractures, cavities, and overflow conduits which, when they flood, reintroduce
fresh pulses of contaminants to the springs or wells. In any karst aquifer, storage can prevent
remediation because near-total removal of contaminants by pumped-wells is rarely, if ever,
achievable (Quinlan and Ray, 1992; Travis and Doty, 1990; Freeze and Cherry, 1989).
The hydraulic gradients within karst aquifers commonly change one to two orders of magnitude
with changes in flow. They decrease exponentially in a downstream direction (Worthington,
1991). Initial flow along bedding planes is down the hydraulic gradient but might not necessarily
be down the stratal dip. Hydraulic gradients ensure that some flow must reach the lowest outcrop
point of an aquifer and therefore could under-flow several springs or river valleys for considerable
distances (Worthington, 1991).
Springs are typically parts of an underground distributary formed by falling external base level
and yielding a vertical hierarchy (Ewers et al., 1978; Quinlan et al., 1978; Smart, 1983). This
includes overflow springs that flow only during flood conditions and underflow springs that drain
perennially. Full-flow springs that are the sole drain of an aquifer are quite rare. Underfiow
springs are difficult to find because they are commonly aggraded but, because they drain
preferentially and perennially, they are very important for monitoring for pollutants. It is suggested
by Worthington (1991) that all overflow springs have complementary underfiow springs and that
all streams in karst terranes gain or lose water from and to the subsurface.
Many karst springs have extremely large catchments. An example is Big Spring, Missouri,
where the groundwater basin extends at least 43 miles, but travel time for dye, along the regional
strike of the strata, takes only about two weeks (Aley, 1975).
Groundwater velocities in karst conduits are both substantial and highly variable. Low-flow
velocities are typically 0.001-0.003 rn/s and high-flow velocities usually reach about 0.1 rn/s
(Worthington, 1991; Quinlan, 1989). Repeated tracings between the same sinking stream and
spring have shown that flow velocity in conduits is directly proportional to discharge (Stanton and
Smart, 1981; Worthington, 1991, p. 34-35). These numbers emphasize, however, the unique
nature of karst hydrogeology. The significance of a contaminant spill affecting a major spring and
possibly drinking water supplies 40 miles away in two weeks, as described above, is the kind of

fact often ignored or unrecognized by consultants, regulators, and administrators unfamiliar with
karst terranes.
Many contaminants that axe spilled in karst terranes infiltrate rapidly through thin residual soils
(Quinlan and Aley, 1987) and enter the groundwater system. The epikarst has only inefficient
vertical connection with the conduit system below (Williams, 1983). If contaminants enter the
epikarst, it is likely that they will be adsorbed onto soil particles and that they will also smear over
rock surfaces. Even if the majority of contaminants travel in a dissolved phase through the
groundwater system, a constant source of them tends to remain in the epikarst zone. Quinlan and
Ray (1992) show that the concentrations of tracer dyes used as surrogates for pollutants in
groundwater do not decline rapidly in springs draining the epikarst or in pumped wells that
penetrate it, even after many years.
Ewers eta!. (1992) show that if light non-aqueous phase liquids (LNAPL’s) such as gasoline
are involved, free-phase product can be carried to the water table, where it could float almost
indefinitely. Water flowing through the floating product will produce a dissolved plume for as long
as the floating product exists, but it might be sufficiently diluted so as to be undetectable at a nearby
spring to which flow occurs. Alternatively, if the contaminant is a dense non-aqueous phase liquid
(DNAPL), it will descend to the base of any fracture or conduit it enters. Any pathway that is made
available, such as a newly drilled ll, will typically allow the DNAPL to migrate deeper. It can be
difficult and expensive to monitor for pesticides and herbicides, but Sabatini and Austin (1991)
have shown that careful choice of two fluorescent dyes can effectively bracket the sorption and
retardation properties of two of the more mobile herbicides. Various contaminants, including
metals, are commonly sorbed onto clays and are typically transported with the suspended sediment
load, which varies directly with discharge velocity. Sediments can be deposited on conduit floors
and walls and can, therefore, produce long-term problems when they are periodically and
repeatedly eroded and redeposited.
Measurement of variation of specific conductance, as a possible surrogate contaminant, can be
extremely useful as an indicator in many karst aquifers of their transport characteristics in relation to
aquikr mechanics (Quinlan eta!., 1992).
Traditional studies of karat springs have usually included plots of variation of discharge against
time (hydrographs) and plots of variation of a chemical or physical parameter versus time
(chemographs). The parameters have commonly included: hardness, stable isotopes, specific
conductance, temperature, turbidity, and dissolved oxygen (Ford and Williams, 1989).
Worthington (1991) showed that anions such as HCO 3 - and SO 4 - , as well as the traditional
cations such as Ca and Mg , should also be measured routinely because they are useful
indicators of aquifer mechanics.
Variation in the form of spring hydrographs in karat aquifers reflects not only variations in
storm impulses but also differences in aquifers having predominantly vadose or phreatic flow, and
autogenic or allogenic recharge. The shapes of spring hydrographs are numerous, with some being
markedly oscillatory, peaked, or relatively flat and broad (Foni and Williams, 1989). Smart (1983)
showed that, in the Canadian Rockies, different forms of hydmgraphs would occur for full-flow,
overflow, and underfiow regimes, and that the truncated segments of the hydrograph, when
combined, could produce the complete hydrograph because there was a hierarchy of springs at
different elevations, with many overflow springs and underfiow springs. Worthington (1991) used
tbe hydiograpb recession exponent to estimate different aquifer boundary conditions between full-

flow, overflow, and underflow springs.
A frequency distribution for specific conductance of waters from different springs in France has
shown that for porous aquifers, the distribution is unimodal with a relatively high specific
conductance; for fissured aquifers, distribution is again unimodal but with relatively low
conductance; and for karat aquifers, conductance is polymodal, with a wide range of values
(Bakalowicz and Mangin, 1980). This is further proof that many different processes are operative
in karat aquifers.
Monitoring wells in karst terranes generally do not work as such for the same reason one does
not win a state lottery with every ticket: The odds do not favor success. The odds for a monitoring
well being successful are explained in terms of scale, however, not economics — although in both
situations, one cannot afford to continue drilling (or playing) until success is attained. The
extremely heterogeneous organization of groundwater flow in caves and dissolutionally enlarged
fractures of a karst aquifer, an organization that is commonly dendiitic or trellised and similar to that
of tributaries of a surface river (Quinlan and Ray, 1989), is not adequately sensed by the number
and size of wells drilled — except by improbably good luck. These facts compromise the
presumed relevance and effectiveness of a conventional monitoring system based solely on wells.
The wells yield water and data, but the samples are unlikely to sense drainage from the facility in
question. Accordingly, the samples from such wells are most probably irrelevant. Certainly the
cost of their analysis does not justify the gamble on their relevance.
The probability of a randomly located well intercepting a conduit draining from a site is similar
to that of a dart thrown at a 30-foot wide wall map of the U.S. hitting the Mississippi River or one
of its major tributaries. It could happen — and sometimes does — but don’t bet on success.
Neither you nor your employer can afford to do so.
The consequences of installing an ineffective monitoring well, missing pollutants that are going
off-site, and missing the fact that a drinking water supply is adversely affected are serious.
However, it is typically argued that such monitoring results indicate a “clean well” (i.e., no
contamination) or, if no cavities were intercepted or there was no loss of drilling fluid, non-
existence of karat! Randomly located wells in karat terranes are known to be unreliable for
monitoring, as explained above. In contrast, springs, wells, and cave streams, all of which have
been confirmed by properly conducted tracing to drain from a facility, are known to be reliable
(Quinlan and Ewers, 1985; Quinlan, 1989, 1990; Ewers, 1992). Accordingly, the motive for
monitoring only via wells, and not on the basis of tracing to them or to springs, is commonly based
on the fact that tracing raises issues beyond the interest and property boundaries of some of the
parties involved.
We make the assumption that the person or agency requiring a reliable system for monitoring
groundwater in a karat terrane really wants to know what is happening. If this goal is sought, a
series of well-designed, properly executed, and correctly interpreted tracer tests offers the best
means of attaining it. This is true because their effectiveness is unconstrained by the numerous,
often ill-founded assumptions made when using a well for monitoring (that the plume is intercepted
by the well and at the correct depth, that the flow is unidirectional, etc.). The beauty of a tracer test
lies in the minimal number of assumptions that must be made. The dye travels with the water and
goes where it goes. [ Yes, there are complications when one is trying to make predictions about
DNAPL’s and LNAPL’s, and there are different retardation factors for different dyes, but these

factors, even if inaccurately represented, do not compromise the results of a tracer test. No other
flow-prediction method is as reliable and as readily subject to empirical verification as is tracing.]
The major assumptions made are that the tracer recovered is that of the test executor, that the field
work and test design have been thorough, and that test execution has been done by reliable
personnel. Other important assumptions are discussed by Quinlan (1989, p. 40-58; 1990).
The field and laboratory procedures for execution of a tracer test are relatively simple and
straightforward (Aley et al., 1993; Quinlan and Alexander, 1992). Nevertheless, as with most
techniques, there are major elements of skill and experience that enhance the probability of success.
Dye-tracing, like neurosurgery, can be done by anyone. But when either is needed, it is judicious
and most cost-efficient to have it done by experienced professionals, those who have trained under
the tutelage of an expert or those who have already made the numerous mistakes associated with
learning and who have learned to avoid the procedural errors that could have economically and
physically adverse consequences.
There are three types of tracer tests with dyes: qualitative (using either visual observation of the
dye-plume or visual observation of dye eluted from passive detectors consisting of activated
charcoal, semi-quantitative (using elutant from passive detectors and instrumental analysis with
either a filter fluorometer or a scanning spectrofluorophotometer), and quantitative (using
instrumental analysis of water samples, as discussed by Quinlan (1989. p. 32-34), Quinlan and
Alexander (1992), Aley et a!. (1993), Duley (1986), and Behrens (1988). Each type of tracer test
can be best under different conditions.
No matter how superbly and efficiently conventionally located wells may be able to detect the
migration of pollutants from a facility — if, by mere chance, they succeed in doing so. and
assuming they would function so reliably — them is no way, other than by tracing, to identify
correctly and conclusively on-site and off-site the places to which pollutants would flow. Stated
another way, there is no way, other than by tracing or by regular frequent monitoring of numerous
wells and springs both on-site and off-site, to discover the numerous consequences of leakage from
a facility. This point is convincingly illustrated by the work of Aley (1986).
The uniqueness and relevance of tracer data for prediction of flow velocities is fflustrated by the
following case study. A detailed investigation, using geophysical logs, borehole video logs,
lithologic core analysis, and unspecified aquifer tests was made of the dissolution porosity and
permeability at a Florida site (Robinson and Hutchinson, 1