The Twelfth Annual
PROCEEDINGS
Waste Testing
& Quality
Assurance
Symposium
July 23-26, 1996
The Washington Hilton
Hotel & Towers
Washington, DC
-------
-------
Contents
SAMPLING
&
FIELD
Paper Page
Number Number
1m Sampling Strategies: Impact of Heterogeneity on Waste Characterization. J. Maney. 1
2m Data Quality Objectives and the Optimization for Design. M. Cahill, M. Miller. 15
3m Compositing and Subsampling of Media Related to Waste Management Activities.
S. Gagner,A. Crockett. 22
4m Sample Representativeness: A Necessary Element in Explosives Site Characterization.
T. Jenkins, C. Grant, G. Brar, P. Thome, P. Schumacher, T. Ranney. 30
,*
5m Guidance for Characterizing Explosives Contaminated Soils: Sampling and Selecting
On-site Analytical Methods. A. Crockett, H. Craig, T. Jenkins, W. Sisk. 37
6m Performance of a New Disposable Sampling and Storage Device for Soil VOCs.
D. Turriff, C. Reitmeyer, L. Jacobs, N. Melberg. 45
7m Field Screening of Soils Contaminated with Explosives Using Ion Mobility
Spectrometry. A Crockett, D. Atkinson, T. Jenkins. 46
8m Identification and Quantitation of Petroleum Substances in Environmental
Samples Using Friedel-Crafts/Hanby Spectrophotometry with Chemometrics.
J. Hanby, J. Miller. 47
9* Site Characterization Technology Demonstrations.
G. Robertson, S. Billets, E. Koglin 48
10m A Field Demonstration of Portable Mass Spectrometers. G. Robertson, S. Bender 49
11m Comparison of Soil Gas, Heated Headspace, and Methanol Extraction
Techniques for Soil Volatile Organic Compound Quantification.
B. Schumacher, M. Minnich 50
12m Evaluation of a Standard Test Method for Screening Fuels in Soils.
S. Sorini,J. Schabron 51
13m Detection of Pesticides and PCBs in the Vapor Phase for Site Screening
J. Whetzel,Jr. 59
continued on next page
-------
continued
RADIOLOGICAL
Paper Page
Number Number
14m Monitoring and Methodology of Radionuclides in Domestic and Imported Foods
in the U.S. E. Baratta. 67
15m Analysis of Actinide Elements in Soils and Sediments.
W. Burnett, D. Corbett, M. Schultz, M. Fern 77
16. Quality Control in the Radioanalytical Laboratory: What Do We Really
Need to Do? R. Litman 87
17m ANSI/ANS Standard for Radioanalytical Data Validation. T. Rucker, S. Salaymeh,
J. Griggs, C. Liu, D. McCurdy, A. Rosecrance, D. Vance, R. Wells, R. Holloway. 88
18m Environmental Radiation Monitoring in Venezuela. L. Sajo-Bohus, E.Greaves 89
19m Analytical Methodology for Non-normal Distributions of Environmental Data.
K. Inn, L. Zhichao, J. Filliben 90
INORGANIC
2Om Method Development Strategies for ICP-MS. R. Wolf, Z. Grosser. 91
21 m Rapid Fieldable Analysis for Mercury. D. Foust, J. Gui. 96
22m EPA Method 3052: Development, Chemistries and Validation.
P. Walter, H. Kingston, D. Link. 104
23m The Accurate Determination of Species by Speciated Isotope Dilution Mass
Spectrometry: Exemplified by the Evaluation of Chromium (VI) in Soil.
H. Kingston, D. Huo, S. Chalk, P. Walter. 112
On-site X-Ray Fluorescence Spectroscopy for the Determination of Metals and
Predicting Acid Rock Drainage. M. Higgins, C. Einberger, A. Burgess, J. Scheuering 120
'. Extraction of Anions from Solid Phase Samples for Capillary Ion
Electrophoresis. ft. Smith, D. Roth, J. Krol, J. Romano 128
Choosing Between Closed Vessel and Atmospheric Pressure Microwave Sample
Preparation for Environmental Analysis. L. Collins, P. Shymanski, K. Kelly. 139
-------
Paper Page
Number Number
27m The Extension of EPA Microwave Leach Methods 3015 and 3051.
D. Link, P. Walter, H. Kingston 140
28m Application of Automated Microwave Digestion for Environmental Sample Analysis.
M. Moses, S. Leikin, B. Moshiii. 143
29m Heterogeneous Reduction of Nitrate Nitrogen in Water and Wastewater for
Compliance Monitoring. B. Pate/, M. Harris 144
QUALITY
ASSURANCE
30m The Development of NIST Standard Reference Materials (SRMs) for Methylmercury
Content in Complex Matrix Environmental Samples. M. Behlke, M. Schantz, S. Wise 145
31 m Variability Among Toxaphene Reference Standards. F. Car/in, Jr., J. Hoffman 148
32m GC/MS Automation Techniques for the Environmental Chemist. P. Cocuzza,
S. Bucher, R. Phillips 156
33* Incorporation of Pollution Prevention Principles in Environmental Methods.
M. Erickson, J. Alvarado, C. Lu, D. Peterson, J. Silzer. 163
34. Contract Laboratory Program (CLP) 2000 - A Model for Providing Superfund
Analytical Services into the Next Millennia. S. Kolb, H. Fribush, R. Thacker. 171
35m Handling Nonlinear Calibration Plots. D. Harris, N. Klueh, R. Noyes 172
36m Comparability of Commercial Environmental Reference Standards.
D. Henderson, K. Herwehe, J. Criscio. 179
37m Improper Use of Significant Figures and Number Rounding by LIMS
Jeopardizes Chemical Data Usability. G. Kassakhian 180
38m How to Use U.S. Environmental Protection Agency World Wide Web and Listserv Internet
Resources to Increase Environmental Compliance and Reduce Environmental
Compliance Costs. L. Lazarus, P. Savoia, P. Flax 187
39m Comparison of the Region 2 QA Program to the Air Force Center for
Environmental Excellence: CERCLA Outreach in Region 2. A. Jackson 188
40m "QA in Cyberspace": The Region 2 Quality Assurance Outreach Program
for the 90's. P. Sheridan, P. Savoia 189
continued on next page
-------
continued
Paper Page
Number Number
41m Non-Traditional Use of Environmental Site Assessments: A Tool for
Determining Potential Environmental Impairment/Compliance.
S. McCone, R. Najjar. 190
42m A Report on Sample Precisions in Hanford Tank Waste Samples.
H. Meznarich, D. Dodd. 196
43m Comparison of EPA and API Laboratory Results as Part of the 1992-1996
Petroleum Refinery Listing Study. J. Parr, J. Austin, Jr., C. Carter, R. Claff. 197
44m Comparability of Measurements of Selected PCB Congeners, PAHs, and
Chlorinated Pesticides in the Marine Environment - Results of a
Performance-Based Quality Assurance Program. R. Parris, M. Schantz, S. Wise 198
45m The Effectiveness of the Procedure for Estimating Instrument Detection
Limits for Inorganic Analyses in the Contract Laboratory Program.
G. Robertson, G. Laing, F. Gamer. 204
46m Automated Data Validation: What are the Limitations? K. Storne. 205
47, Automation of ICP Data Validation fpf||r^fli0Eftvnalysis. M.Tatro 209
4Sm New Quality Systems at the U.S. Environmental Protection Agency. N. Wentworth 210
49m Achieving Flexibility Through Effective QA Practices. J. Warren 211
50m Data Quality Management for Emergency Response Cleanup Services.
J. Fields, G. Schupp. 212
51m Documentation and Record Keeping Guidelines. A. Rosecrance, L Kibler. 220
52m NIST SRMs for Organic Environmental Analyses-Current Availability and
Approach to Certification. R. Parris, D. Poster, L Sander, M. Schantz, S. Wise 228
53m Data Usability: The Next Step to Data Validation for Site Remediation. F. Fairbanks. 235
54, NELAC Update. R. Trovato. 236
55m The SHELL Requirements for Chemistry Data Generated for USACE Projects.
J.Solsky 237
-------
Paper Page
Number Number
56. Integration of the DQO Planning Process in the Closure of Mixed Waste
Storage Units. T. Hosaka, M. Schlender, D. Lutter, M. Miller. 245
57m Proper Selection of Performance Evaluation Samples for Chemical Data
Acquisition Based on Vendor Acceptance Limit Criteria. G. Kassakhian 252
58m Performance-Based Methodology for the Transuranic Waste Characterization Program.
M. Maskarinec, W. Griest, J. Keller. 258
59m Streamlining the Data Quality Process Without Compromising Defensibility.
P.CIine, A. Verstuyft. 267
ORGANIC
6Om Comparison of Inertness Properties of Tubing in Gas Chromatography.
S. Adams, J. Walsh, W. Cooke 269
61, Easier and Faster GC/ECD Analyses of Pesticides and PCB's.
S. Brillante, P. Wiley, T. Stark. 276
62m Cost and Quality Improvements Using a Mass Spectrometer as the Detector
for Methods 8010, 8020, and 8021. R. Burrows, A. Quick, Jr., F. Feyerherm 277
63m Comparison of Extraction Methods for Heavy Petroleum Hydrocarbon
Measurement in Soils. P. Calcavecchio, B. Kelley, A. Felix, G. Van Gaalen, E. Drake 283
64, Open-Vessel Microwave Extraction of Soil and Sediment Samples.
K. Kelly, D. Stalling, L Collins 291
65m Bias Factors in Recovery of Spiked Organic Compounds from Aqueous Samples.
K. Kelly, D. Stalling, R. McMillin, M. Daggett, L. Palmer. 293
66m PAH Extraction from River Water Using New Novo-Clean C18 Extraction Membranes.
G. Nixon, T. West. 301
67m Screening for Silvex by a Magnetic Particle Enzyme Immunoassay in TCLP
Extracts from Soil and Wastewater. F. Rubio, T. Lawruk, A. Gueco, D. Herzog 307
68m Evaluation of the Accelerated Solvent Extraction System for the
Extraction of Environmental Matrix Reference Materials. M. Schantz, S. Wise 317
69m Methods of Preparing Soil Samples for Headspace Analysis of Volatile
Organic Compounds: Emphasis on Salting Out. A. Hewitt. 322
continued on next page
-------
O I"! "t O •"! "ft S continued
Paper Page
Number Number
70m Detection of Cyclodiene Insecticides in Water and Soil Using a Magnetic
Particle-Based Immunoassay. M. Selisker, D. Herzog, W. Day, J. Itak. 330
71, The Determination of Specific Polychlorinated Biphenyl Congeners in
Fish Samples by Gas Chromatography and Electron Capture Detection. J. Snyder. 338
72m New and Specially Developed Bonded Silicone Phases for Pesticides Trace
Analysis with Capillary GC. J. deZeeuw, P. Heynsdijk, C. Duvekot. 339
73m A New HPLC Stationary Phase for the Analysis of Polycyclic Aromatic
Hydrocarbons (PAH's). J. deZeeuw, N. Lammers, W. Verstraeten, J. Marinissen 34O
74m Overview of the Current Status of the RCRA Organic Methods Development
Program. B. Lesnik. 341
75. Freon Alternatives for the Environmental Lab. M. Bruce, J. Hall. 342
76m Evaluation of Solid Phase Extraction Followed by Supercritical Fluid Elution of
Semi-Volatile Organic Compounds in Toxicity Characteristic Leachates. W. Corl, III. 348
77m Evaluation of the Microdistillation Method (Method 5031) for Measuring Volatile
Water-Soluble Compounds in Pulp Mill Treatment Influents and Effluents.
A. Gholson, D. Cook, D. Hoy. 360
78m Extraction of Polychlorinated Dibenzo-p-Dioxins and Polychlorinated Dibenzo-Furans
from Environmental Samples Using Accelerated Solvent Extraction (ASE).
B. Richter, J. Ezzell, D. Knowles, F. Hoefler, A. Mattulat, M. Scheutwinkel,
D. Waddell, T. Gobran, V. Khurana 370
79m A Confirmatory Holding Time Study for Purgeable VOCs in Water Samples.
D. Bottrell, O. West, C. Bayne, R. Siegrist, W. Holden 371
80m How to Modify Existing Methods for New Applications. B. Lesnik. 378
81m What to Expect from Immunoassay Methods and How to Obtain Their Benefits.
S. Friedman 379
82m Method 4670: A Quantitative Immunoassay Method for the Determination of
Triazine Herbicides as Atrazine. H. McCarty, B. Bathija, B. Lesnik. 386
83. Method 8085: Pesticides by GC-AED. R. Araki, R. Cummings, R. Rieck,
N.Olson, R.Carrell. 387
-------
Paper Page
Number Number
64m Volatiles Productivity Enhancement with an XSD Detector.
M. Bruce, T. Lavey, R. Burrows 394
85m Fast GC Capillary Columns Applied to Environmental Separations and Analysis -
Screening Methods. D. Gere, W. Snyder, V. Giarroco, R. Kolloff, B. Rothweiler. 396
GENERAL7AIR
&
GROUNDWATER
86m The Application of High Speed Gas Chromatography to Air Analysis.
N. Kirshen, Y. Bao, C. Hodges, D. Coe 405
87m A Rapid, Cost-Effective Method for Detection of Trihalomethanes in
Drinking Water. K. McKenzie, W. Studabaker, K. Carter. 413
88m Overview of Air Sampling and Analysis Methods. G. Button 417
89m Determinations of N-Nitrosodimethylamine (NDMA) at Part-Per-Trillion(ng/L) Concentrations
in Contaminated Groundwaters and Drinking Waters Featuring Carbon-Based Membrane
Extraction Disks. B. Tomkins, W. Griest, G. Connolly, H. Hayes 422
90. Limited Life Cycle Analysis. T. Barber. 432
91m Reducing Waste Generation and Radiation Exposure by Analytical Method
Modification. A. Ekechukwu. 440
-------
SAMPLING
&
FIELD
'< 4
-------
Sampling Strategies: Impact of Heterogeneity on Waste
Characterization
John P. Maney. Ph.D. President, Environmental Measurements Assessment, 5 Whipple
Road, South Hamilton, MA 01982-1517
ABSTRACT
The US Environmental Protection Agency (EPA) Office of Solid Waste is in the process
of revising Chapter 9 of SW-8461. To support this effort, the Office of Solid Waste is
working with the American Society for Testing and Materials (ASTM) to accelerate the
development of ASTM standards needed for these revisions. Under this mandate, Task
Group D34.01.12 was assigned the task of authoring guidance on sampling strategies for
heterogeneous wastes. The resulting standard has passed sub-committee and main
committee and final changes have been submitted for main committee balloting.
The standard is a practical non-mathematical discussion and guide for heterogeneous
waste sampling which is consistent with paniculate material sampling theory as well as
inferential statistics and can serve as an introduction to the statistical treatment of
sampling issues. The standard approaches the sampling of heterogeneous wastes in a
generic sense while leaving project specific details to a planning process such as the Data
Quality Objective process.
The annex to the standard contains an introductory discussion of homogeneity, practical
homogeneity, random heterogeneity, non-random heterogeneity, stratification and the
relationship of samples to populations. The main body of the standard focuses on the
sampling of highly stratified wastes, which are usually very difficult to characterize. An
appendix applies the guidance to a storage area containing 4000 drums.
INTRODUCTION
The first edition of EPA Publication SW-846, Test Methods for Evaluating Solid Waste
Physical/Chemical Methods was announced in May of 1980. Around 500 to 1000 copies
of the first edition were distributed. The first edition included a limited discussion of
sampling issues. The second edition of SW-846 which included an expanded discussion of
statistical issues was announced in September of 1982. Approximately 5000 copies of this
edition were distributed. The third edition of SW-846 was announced in the Federal
Register in March of 1987. Chapter Nine of the third edition which is entitled, "Sampling
Plan" incorporated the earlier version of the statistical discussion along with an expanded
discussion of sampling plan implementation. Chapter Nine has remained unchanged since
it was authored in 1986.
EPA recognized that Chapter Nine must be updated to reflect the increased understanding
and complexity of waste sampling issues. As part of its strategy to update Chapter Nine,
Test Methods for Evaluating Solid Waste Physical/Chemical Methods, Third Edition.
-------
EPA has worked within the ASTM consensus standard process. EPA sponsored several
accelerated standards on issues pertinent to waste sampling, including a standard guide on
Sampling Strategies for Heterogeneous Waste, which is the subject of this paper.
In January of 1994, a task group defined a scope for the accelerated standard guide on
heterogeneous waste. In response to EPA's request, the task group agreed that the
standard would address the more extreme cases of heterogeneity that can be encountered
during waste characterization. Since it was an accelerated standard, the task group met on
a quarterly basis for a period of 2 years. A number of drafts were completed and balloting
has progressed such that a Society approved standard is expected this year.
The focus of the Standard Guide on Sampling Strategies for Heterogeneous Waste and of
this paper is four fold: 1) to describe the causes and types of heterogeneity; 2) to discuss
the relationship between samples and populations; 3) to discuss the impact of
heterogeneity on the sampling process and 4) to present a guide for sampling
heterogeneous wastes. This paper addresses issues presented in the standard as well as
related issues that the author believes further develops aspects of heterogeneity and their
impact on sampling.
HETEROGENEITY
The following subsections discuss sampling in terms of populations, samples and
characteristics. The following definitions should facilitate this discussion.
Characteristic— a property of items, a sample or population which can be measured,
counted, or otherwise observed. A characteristic of interest may be the cadmium
concentration or ignitability of a population.
Heterogeneity— the condition of the population under which all items of the
population are not identical with respect to the characteristic of interest.
Homogeneity— the condition of the population under which all items of the
population are identical with respect to the characteristic of interest.
Item— a distinct part of a population, (e.g. microscopic particles, macroscopic
particles and twenty-foot long steel beams).
Population— the totality of items or units under consideration.
Practical Homogeneity— the condition of the population under which all items of the
population are not identical. However, for the characteristic of interest the differences
between individual physical samples are not measurable or significant. That is, for
practical purposes the population is homogeneous.
Random—lack of order or patterns in a population whose items have an equal
probability of occurring.
-------
Sample— a portion of a population which is taken for testing or for record purposes.
According to the prior definitions, sampling of an abandoned property would be a simple
task if it was a homogeneous population. Being homogeneous, all items of the abandoned
property would be identical for the characteristic of interest. Thus sampling and measuring
any item from the property would allow one to evaluate the entire property.
Unfortunately, homogeneity is a theoretical state and actual populations don't consist
solely of identical items.
Since heterogeneity is the norm, the items of the abandoned property would be dissimilar
to some degree. The degree of heterogeneity can range from that of a population whose
items are so similar that for practical purposes the population can be considered
homogeneous (Practical Homogeneity) to a population whose items are all dissimilar.
Usually, the greater the degree of heterogeneity, the greater the sampling difficulty.
Due to the identical nature of items, no distributional differences can be detected within a
theoretical homogeneous population. However, the different items of heterogeneous
populations can be distributed differently to create distinctly different types of
heterogeneity.
Random heterogeneity— occurs when dissimilar items are randomly distributed
throughout the population.
Non-random heterogeneity — occurs when dissimilar items are non-randomly
distributed resulting in the generation of strata.
Strata— are subgroups of the population which are separated in space or time from
the remainder of the population and are internally consistent with respect to the
characteristic of interest and different from adjacent portions of the population.
Figure 1 is a graphical depiction of homogeneous, randomly heterogeneous and non-
randomly heterogeneous populations. The drum-like populations portray different types of
spatial distributions while the populations being discharged through the waste pipes
represent the different types of temporal distributions. No spatial or temporal distribution
is identifiable between the identical items of a homogeneous population. Identifiable
spatial and temporal distributions are present in the heterogeneous populations, with the
existence of strata in the non-randomly heterogeneous populations. It is important to note
that the perception of a population's homogeneity or heterogeneity is relative to analytical
sample size, population size, sampling objectives and at times the analytical methods.
For example, pure silver nitrate, some of which is powder and some of which is in the
form of large crystals would be considered heterogeneous by the analyst concerned with
particle size while the analyst concerned with leachable silver content would find the
material to be homogeneous.
-------
POPULATION
Differences in
Spatial/Temporal Distribution
Heterogeneity
Type
I Time 2
El X
~i
Time 1
3£Hi
«*%
WASTE DISCHARGE
NO
Homogeneous
c-'-;i
Tinw 2
WASTE DISCHARGE
1 I I X I I I -. ,
-1«v '1
: I ) i i f
WASTE DISCHARGE
-------
Figure 2. HETEROGENEITY IS RELATIVE TO
METHOD DETECTION LIMITS
(MDLs)
FURNACE AA
1
a
4
MDL
—
n
M
SAMPLE ID
HETEROGENEOUS
X-RAY FIELD
SCREENING
MDL
n
~i
SAMPLE ID
1
APPARENTLY
HOMOGENEOUS
homogeneous matrix can be distributed in such a stratified manner that it creates an
extremely heterogeneous waste2. A common factor for these extremely heterogeneous
wastes is their large number of strata. Thus these wastes which are very difficult to
characterize will be referred to as "Highly Stratified" wastes. Highly-stratified wastes
which are a type of non-random heterogeneous wastes have so many strata that they
become difficult to sample and characterize. Classifying a waste according to its level
of stratification is a relative issue. It is relative to the persons planning and performing
the sampling, their experience, available equipment, budgets, the characteristic of
interest and sampling objectives. Highly-stratified wastes are such that it is not
practical or effective to employ conventional sampling approaches to generate a
representative database. Nor would the mean concentration of a highly-stratified waste
be a useful predictor (i.e. the level of uncertainty is too great) for an individual subset
of the waste that may be subjected to evaluation, handling, storage, treatment or
disposal.
2 S. Feenstra. Soil Sampling for Environmental for Environmental Studies: Matching Methods with
Objectives . ASTM Symposium on Sampling Environmental Media. Denver, CO., April 5, 1995
-------
Figure 3. Highly Stratified Waste
Poly Bag Liner (s)
Inner Bags Tied Shut
Taped Metal
50 ml Liquid
(•.g., wlpas, tongs,
packing materials
cans.bottl«s...)
Gallon Patnt Can
(LU on • bottto
ln«M« containing
unknown liquid)
Liquid i-,\-
Vermicuifte
Additionally may contain:
asbestos gloves, respirator cartridges,
photographic materials
Characterizing Heterogeneous Wastes
EPA 600/R-92-033
Strata are different portions of a population separated in time or space with each
portion having internally similar concentrations or properties which are different from
adjacent portions of the population
• A landfill may display spatially separated strata, since old cells may contain
different wastes than new cells (stratification over space).
• A wastepipe may discharge temporally separated strata, if night-shift production
varies from the day shift (stratification over time)
-------
• Lead-acid batteries will constitute a strata separate from commingled soil, if lead is
the characteristic of interest (stratification by component).
• Drums from an inorganic process may constitute a different strata from those co-
disposed drums generated by an organic process (a sub-type of stratification by
component referred to as stratification by source).
Different strata are often generated by different processes or a significant variant of the
same process. The different origins of the strata usually result in a different
concentration or property distribution and mean concentration or properties.
While stratification over time or space is widely understood stratification by component
is less commonly employed. Some populations lack obvious spatial or temporal
stratification yet display extreme levels of heterogeneity. If these populations contain
easily identifiable components such as large crystals, rods, blocks, gloves, pieces of
wood or concrete then it may be advantageous to consider the population as consisting
of a number of component strata. An advantage of component stratification is that it
can simplify the sampling and analytical process and allow a mechanism for making
inferences to a highly stratified population. Component stratification shares many
similarities with the gender or age stratification applied to demographic data by
pollsters.
POPULATIONS & SAMPLES
This section discusses the interrelation of populations and samples, how samples are used
to make inferences about populations and the resulting implications on sampling activities
Populations can be homogeneous, randomly heterogeneous or non-randomly
heterogeneous. Populations possess a mean, a mode and median concentrations as well as
a distribution of the characteristics of interest. Environmental professionals often need to
know these population attributes to determine if contamination has occurred, the
magnitude and breadth of contamination, the suitability of waste for incineration or
stabilization, or the health/environmental risks associated with a certain population. To
meet these informational needs, the environmental professional will study the population.
Ideally, he/she would study the entire population. However, this is often not possible and
almost always is not practical, cost or time effective. Thus, in most instances the
environmental professional will study portions of the population (samples) and use
information from these samples to make inferences about population attributes. These
samples are the windows through which the environmental professional views the
population.
The population and its boundaries will be a function of sampling objectives. Examples of
populations that may be subjected to sampling are a 1-gallon bottle, a waste drum, a
parking lot, a pond and a city block. Definition of the population's physical and temporal
boundaries are key to defensible decision making. Population attributes can change as the
boundaries of the population are changed. For example, the middle population in Figure 3
is clearly randomly heterogeneous. However, if the population was to be re-defined such
-------
that it contained only the four items in the top left-hand corner of the population, the new
population would be non-randomly heterogeneous with a layer of black items on top of a
layer of white items. Although, changing population boundaries can change a perspective,
the re-defining of a population into smaller more manageable exposure or remediation-
sized units may be advantageous from a cost savings or waste management perspective.
Samples may be representative of a population, or may not be representative of a
population. Samples may be similar to other samples or display significant intersample
differences for the characteristics of interest (intersample variance). The differences in the
measured characteristics of samples may not be correlated with space or time (i.e.
randomly heterogeneous) or the differences in the measured characteristics may be
correlated with time or space (i.e. a non-random heterogeneous waste stratified over time
or space). Table 1 summarizes how samples can be used to identify a population's
heterogeneity type.
Table 1. Population and Sample Attributes
POPULATION
ATTRIBUTE
SAMPLE
DESCRIPTION
SAMPLE ATTRIBUTE
INFERENCE
Homogeneous
(Theoretical
Homogeneity)
All samples contain only
identical items
No Significant
intersample variance. No
correlation of
concentration or
properties with time,
space, component or
waste source.
Samples are representative
of a homogeneous
population
Practical Homogeneity
All samples contain
dissimilar items, but
each sample contains
same proportions
No Significant
intersample variance. No
correlation of
concentration or
properties with space,
time, component or waste
source.
Samples are representative
of a homogeneous
population
Random Heterogeneous
All samples contain
dissimilar items, each
sample has different
proportions, but these
proportions are not
correlated with time,
space or components
Significant intersample
variance. No correlation
of concentration or
properties with space,
time, component, or
waste source
Samples are representative
of a random
heterogeneous population
Non-random
Heterogeneous
(Stratified)
All samples contain
dissimilar items, each
sample has different
proportions and these
proportions are
correlated with time,
space, components or
source
Significant intersample
variance. Correlation of
concentration or
properties with space,
time, component, or
waste source
Samples are representative
of a non-random
heterogeneous population
8
-------
INFERENCES
The accuracy of inferences regarding a population will be a function of how well the
samples represent the characteristic of interest. The Resource Conservation and Recovery
Act (RCRA) defines a representative sample as a sample of a population which can be
expected to exhibit the average properties of the population. The use of the singular form
of the word "sample" can be misleading to the non-statistician. The varying characteristics
of a heterogeneous material are usually better reflected in more than one sample, while
measurements of mode, median and variance require the collection and analysis or more
than one sample.
It is better to think of representative sample(s) as those sample(s) that are collected in
such a manner that they accurately reflect the population characteristic of interest. Figure
4 depicts how samples collected in the field as well as subsamples generated in the
laboratory, as a group, must physically and chemically reflect population characteristics for
the resulting data to accurately represent the population.
ANALYTICAL
SUBSAMPLES
PHYSICAL SAMPLES
n
a o o
a a
Subsamples in total
physically represent
population
DATABASE
Collect Field Samples
,1 r\
Samples in total physically
represent population
Illlllllll
Database represents
the measured population
characteristic
POPULATION
Figure 4 Samples Representing a
Population Characteristic
The representativeness of samples and the associated inferences are dependent upon
sample collection considerations such as sampling strategy, sampling locations, sample
mass, number of samples, sampling equipment and sample handling. Figure 5 shows how
the mass/volume of a sample and sampling locations can affect the representativeness of
samples. The inferences listed in the middle and right-hand columns are based on sets of
different samples masses One sample of each set was taken from the top of the population
while the second sample was taken from the bottom of the population.
-------
As expected, sampling of a homogeneous population (top population in the left hand
column) is simplified by the fact that any item accurately represents the population.
However, sampling of the randomly heterogeneous population depicted in the center of
the far left column is more complicated and different inferences can be made depending on
the sample size and the luck of the draw. The large mass samples give a consistent picture
of the population, correctly infer that the population is randomly heterogeneous and would
yield a representative measure of population characteristics Different and incorrect
inferences regarding population characteristics can be made when collecting the smaller
mass samples. If the randomly heterogeneous population consisted of a 50/50 mix of white
and black items and only two of the small samples were collected, there would be a 50%
chance of inferring that the population was randomly heterogeneous, a 25% probability of
inferring that the population was homogeneous and consisted only of white items and a
25% probability of inferring that the population was homogeneous and consisted only of
black items
I l
Q
0)
o
C
re
JZ
O
•o
C
re
< !(")< )C
iooc
QQOOQs
Population Is Non-Randomty
Helerogtneous
•
Population Is Non-Randomly
Heterogeneous
Figure 5 Population Inferenes
The inferences for the non-randomly heterogeneous population depicted as the bottom
population in Figure 5 are consistent regardless of sample size. However, this
10
-------
similarity of inferences for different sample masses would decrease if the samples were
collected from random locations within the population, as opposed to the top and bottom
locations specified by the figure.
The homogeneous and heterogeneous populations depicted in Figure 5 are greatly
simplified, however the lessons learned from studying these simplified populations can be
extrapolated to actual populations. For example, in the above example, inferences were
based on two samples. After studying the figure, it becomes apparent that as the number
of samples increases, the accuracy of inferences can be expected to improve. Also,
regardless as to whether the depicted populations represented a drum or a landfill, if the
sample consisted of a vertical core from top to bottom, there would be increased
confidence that any existing horizontal strata would be represented in the collected
sample.
The relative proportions of sample mass to population mass are much smaller in actual
sampling situations. The actual percentage of population sampled is usually infinitesimal.
All other conditions being the same, the fewer the samples and the smaller the sample
mass, the greater the likelihood that the characteristics of a population will not be properly
represented.
Equally important is the relationship of item size to sample size. As item size becomes
smaller with respect to sample size and the number of items per sample increase the
greater the likelihood that a representative sample will be collected.
As the relative size of items increases, the fewer items in a sample and the greater the
likelihood of collecting less representative samples. The worse case scenario occurs when
item sizes are too large to be accommodated by sampling equipment. Referring back to
Figure 5, assume that black items are twice as large as the white items and that the chosen
sampling device was only large enough to accommodate the smaller white marble. This
sampling scenario could lead to substantial errors if the characteristic of interest differed
from white to black items. Considering a more concrete example, 500 g samples collected
with a properly designed scoop may be appropriate for collecting granular material.
However a 500 g sample mass would not be appropriate if the designated population
consisting of a mixture of fine sand and chromium-plated automobile bumpers was to be
investigated for chromium. Sampling theory as described by Pitard, discusses methods for
using the maximum item size to calculate optimum sample masses that will accommodate
all items.
CHARACTERIZATION OF HETEROGENEOUS WASTES
The standard offers strategies for sampling and characterizing different types of highly
stratified wastes. Wastes can be heterogeneous in particle size and/or in composition,
allowing for the existence of;
• Strata of different sized items of similar composition
• Strata of similar sized items of different composition
11
-------
• Strata of different sized items and different composition
The last two highly stratified waste categories include the wastes that are usually the most
difficult to characterize. Figure 6 describes an approach that is applicable to the
characterization of highly-stratified wastes. The standard guide suggests modifications to
conventional waste characterization to accommodate the complexities presented by highly
stratified wastes. Prior to implementing these modifications the following information
should be collected as part of a structured planning process (e.g. ASTM D5792-95 or
EPA QA/G-4);
• Is the characteristic of interest correlated with time, space, item size, component or
source
• Does any waste component not contain the characteristic of interest be
• Do small items represent a stratum as well as large cumbersome items
• Is the characteristic of interest innate or absorbed to surfaces
This information regarding the distribution of the characteristic of interest can be used to
modify sampling and laboratory subsampling, All assumptions used to modify sampling
must be justified and well documented. When possible the validity of assumptions should
be periodically confirmed by sampling and analysts.
Suggested modifications to the sample preparation step include;
• Increasing the analytical sample mass/volume
• Generation of homogeneous extracts and digestates
• Component elimination and surrogates
• Surface wiping
• Particle Size Reduction
Alternate analytical methods that may allow for the analysis of larger sample
mass/volumes and more samples per budget due to decreased analytical costs are
summarized.
If modification of sampling, sample preparation or analytical methods do not allow the
objectives to be met, the standard suggests review of the reasoning behind the original
plan and possible modification of waste management or even of the objectives themselves.
CONCLUSION
The ASTM "Standard Guide on Sampling Strategies for Heterogeneous Wastes" and the
cited references should give the experienced scientist a review and the less experienced
scientist a background in the significance of heterogeneity when employing samples to
make inferences regarding populations of interest.
12
-------
FIGURE 6. Approach for the Characterization of Heterogeneous
Wastes
Is waste highly stratified?
No
Yes
Can Sampling be
Modified?
Not •+•
Yes
No
Can Sample Preparation
be Modified?
NO
I Yes
Can Analysis be Modified?
Change Handling,
Treatment, Disposal of
Waste or Target Parameter
— Change Objectives
Yes
No
Use Conventional
Sampling & Analysis
Approach
Will Modified
Approach
Allow Waste
to be
Cost-effectively
Characterized
and Allow Objectives
to be met?
' Yes
13
-------
REFERENCES
Generation of Environmental Data Related to Waste Management Activities:
Development of Data Quality Objectives. ASTM D5792-95
Characterizing Heterogeneous Wastes. EPA 600/R-92/033
Compilation of ASTM Standard Definitions 1990. 7th Edition, American Society for
Testing and Materials, 1990
Environmental Monitoring Issues. EPA /600/R-93/033
Guidance for the Data Quality Objectives Process. EPA QA/G-4, September, 1994.
Pitard, F.F., Pierre Gy's Sampling Theory and Sampling Practice. 2nd Edition, CRC
Press
S. Feenstra, Soil Sampling for Environmental for Environmental Studies: Matching
Methods with Objectives , ASTM Symposium on Sampling Environmental Media,
Denver, CO., April 5, 1995
Maney, J.P. and Wait, A.D., The Importance of Measurement Integrity. Environmental
Laboratory. October/November 1991.
Taylor, J.K., Statistical Techniques for Data Analysis. Lewis, Chelsea MI, 1990
Test Methods for Evaluating Solid Waste Physical/Chemical Methods. EPA SW-846.
14
-------
DATA QUALITY OBJECTIVES AND THE OPTIMIZATION FOR DESIGN
Marty Cahill, EnChem, Inc., 802 Deming Way, Madison, WI53717; Mitzi Miller, EQM, 10801 Fox Park,
Knoxville, TN 37931
ABSTRACT
The objective of the paper is to provide an overview of the latest standard practices developed by EPA and
ASTM under a joint effort. These standard practices have been developed to provide guidance focused to
project managers using the Data Quality Objective Process and Optimizing Sampling Designs supporting
waste management activities.
Environmental data are often required for making waste management and regulator decisions. Both the
government and public sector decision makers of the environmental community, have continued to wrestle
with I) mechanisms to define the purpose of an environmental waste management activity and 2) the criteria
by which to judge whether or not that activity has been successful.
The development of Data Quality Objectives (DQOs) is the first step in data generation and includes, the
development of the criteria on which to base the success of the environmental activity. The other two aspects
are 1) implementation of the sampling and analysis strategies and 2) data quality assessment. The seven steps
of the DQO process are as follows:
1. Stating the problem
2. Identifying possible decision
3. Identifying inputs to decisions
4. Defining boundaries
5. Developing decision rules
6. Specifying limits on decision uncertainty, and
7. Optimization of sampling and analysis design
After establishing the acceptance criteria via steps four through six, for the problem and decisions, a
sampling and analysis design is selected, implemented and ultimately evaluated against the criteria. The
sampling design and output of the DQO process, is the blue print by which the data collection activities are
implemented. The focus of the Design Optimization strategies presented includes practical and
implementable designs, recommendations for evaluation of design and methods for presenting designs to
decision makers.
ASTM documents must be balloted and extensively reviewed prior to acceptance as a standard practice.
ASTM in conjunction with US EPA have completed development of the DQO document. The Optimization
of Design Standard Practice is under development.
INTRODUCTION
The EPA published initial guidance regarding the Data Quality Objective Process (DQO) in 1987 and
followed with EPA Guidance for Planning for Data Collection in Support of Environmental Decision Making
Using the Data Quality Objective Process (EPA QA/G-4, September 1994). The goal of these documents
is to provide a planning method which balances risk including health, ecological and risk of making an
incorrect decision with resources including time and budget. A secondary goal was to identify the quality
and quantity of data needed to make decisions.
15
-------
One of the implementation problems in the DQO process is providing sufficient guidance in sampling design.
Some organizations believe that defensible statistical designs result in collection of too many samples while
other organizations never use a statistical approach. EPA recognized this problem and is working with
ASTM D34.01, Waste Management Committee, Subcommittee-Sampling and Monitoring to generate a
document for the non-statistician that explains the elements of design and the optimization methods. The
optimization allows one to choose logical statistical designs which balance data needs with risk and resources.
This standard guide is one in a three part series of ASTM documents focusing on DQOs (ASTM D5792),
quality assurance (ASTM D5283) and design optimization which is currently in the ballot process. The ballot
process should be completed early next year.
The DQO process is briefly described with emphasis on the information generated from the process which
is needed for design optimization. The last section of this paper describes three major sections of the ASTM
document sampling design criteria, design selection, and optimization.
DQO Process
Step 1 - Problem
Problem statements are clearly written after historical knowledge and data are compiled and understood by
the stakeholders. Historical data are compiled and summarized and given to the decision makers. The
decision makers agree on information which is known, not known and situations that result in problems.
Prior to beginning the process, the task lead identifies the team of technical support staff and the appropriate
decision makers from DOE, EPA, and Ecology who will perform the planning.
Step 2 - Decisions
For one or more critical problems, decisions are identified. These decisions result in actions which will
correct the problem. The focus of the DQO process is on decisions requiring collection or evaluation of
data.
Step 3 - Inputs
The inputs are the data and information necessary to make the project decisions. Prior to completing the
process the quality, and quantity of data required are agreed upon by the stakeholders. A description of how
the data will be used in the decision is agreed upon by the stakeholders. This assures that no unnecessary
data are requested.
Step 4 - Boundaries
The spatial and time boundaries associated with the decisions are specified. The boundaries may be physical,
practical, or have agreed upon characteristics. Large areas should be subdivided depending on
characteristics. For example, each cylinder of grouted sediment will be tested and decisions made from each
test. An example of a time boundary is the time required for grouted material to solidify prior to testing or
storage. Sampling must be done after the grouted waste is solidified.
Step 5 - Decision Rules/Logic
Decision rules are the criteria upon which decisions will be based. These may be stated as "If..then"
statements or presented in logic diagrams. The essential requirements are that the situation be identified
16
-------
followed by the associated action which should occur. For example, "If the average of three measurements
of lead in the grout exceeds 5 mg/1, then the waste will be disposed as a hazardous waste." If measurements
are made, die logic should contain the parameter, the mode of comparison (i.e. average) and the resulting
action. Logic should be complete by describing actions which occur if die limits are met or are exceeded.
Step 6 - Limits on Decision Uncertainty
The decision makers must agree on the acceptable probability of making an incorrect decision. In order to
assess the acceptable probability, the consequences of making an incorrect decision at various concentrations
should be understood. Decision error probabilities are based on the:
• number of samples,
• variance or standard deviation,
• level of false positives and negatives, and
• action, risk or regulatory level
Step 7 - Optimization of Sampling and Analysis Design
The data collection activity can be optimized using the above information. Using this information, the
technical team develop and document the sampling and analysis plan. Multiple designs are typically
presented to decision makers and the decision makers agree upon the resulting plan.
The presentation to the decision makers must include the trade offs in risk and resources. This balance is
the greatest factor in optimization.
Table 1.1 - Inputs to Optimization of Design
DQO Step # and Title
Information Used in Optimization of Design (Step 7)
1
Problem
Technical staff and decision makers, resources, conceptual site and
risk models, historical data
Decision
Decision, actions resulting from the decision, consequences of actions,
priority of decisions, questions which allow one to make decisions
Inputs
Lists of data and information needed to make each decision, whether
the data exists or must be obtained, data generation activities, use of
the data, contaminants of potential concern, potential sampling and
analysis methods, standard deviations or variance of historical data,
typical detection limits by matrix, sample size/volume, holding times,
turn around times, in-situ methods, traditional analysis methods
Boundaries
Spatial areas for sampling, division of areas by matrix, division of
areas by potential level of contamination, division of area based on
process knowledge, future use of land or facility, receptors of risk,
risk area or exposure unit, time of sampling and project/decision
schedule
17
-------
Table 1.1 - Inputs to Optimization of Design
DQO Step # and Title
Information Used in Optimization of Design (Step 7)
Decision Rule
Action or regulatory limits, consequences of each action, statistic to be
used (e.g. mean, upper limit, etc.), logic diagram or If..then...
statements
Decision
Uncertainty
Number of samples, variance or standard deviation based on past data
or similar data by matrix, level of false positive and negatives, action
level, expected level of contaminant, risk or regulatory level, area
where consequences of a wrong decision has less consequence,
consequence of a wrong decision by concentration level
OPTIMIZATION OF SAMPLING DESIGN
The development of alternate sampling designs requires a knowledge of
• the information generated in the DQO process,
• sampling design selection criteria,
• commonly used and alternative sampling designs.
With the development of sampling plan alternatives, the process of optimization begins. The resultant
sampling design should strive to be: practical, objective and technically defensible. Figure 1.1- Process
for Optimization of Sampling Design provides a two dimensional view of the process outlined in the ASTM
Guide, which includes descriptions of the inputs needed to select an appropriate sampling design and the
subsequent process for optimization.
Decision makers need to consider all of the information available. The following general trends are a good
rule of thumb.
• Usually the greater the number of samples, the shorter the confidence interval on the
environmental parameter of interest.
• More sophisticated analytical chemistry methods can achieve lower detection limits.
However, this may mean higher costs per sample and an increase in sample size
mass/volume.
• Composite samples may not identify point source contaminant locations or hot spots.
• If something is known about the expected heterogeneity of a waste site, the non-random
sampling designs, such as stratified sampling or a combination design may be the most
effective.
• For containerized waste, both the within container and between container sampling error
should be considered.
• Usually sampling errors are larger than the analytical errors. However, it may be more
difficult to quantify sampling errors.
18
-------
Optimization is an iterative process of assessing the design alternatives and determining the most resource-
effective design which satisfies all of the project objectives or DQOs. For a small project, the entire selection
and optimization process may be conducted at the same time. The lack of a formal assessment does not
necessarily indicate that an assessment has not taken place. For a large Corrective Action, the process may
bear some resemblance to the order the flow chart, however, the "steps" for the assessment and optimization
can be and are followed in the most practical order.
It is important to assess the sample design(s) selected with respect to the project objectives. If a resource-
effective design is not included in the sampling design alternatives, it may be necessary to modify the closest
design alternative or revise the project objectives, returning to the Planning Process and a review of the
Project Objectives.
Once the initial set of sampling designs are established, the design alternatives and resources required for
each are reviewed by the decision makers. Careful consideration of each alternative allows for an
understanding of the benefits and resource commitments needed for each sampling design. The design
alternatives may include both directed and statistically based approaches.
Each design is re-evaluated against the project objectives, with respect to the following design constraints,
which can be grouped into general categories. Some of these overlap or fall into more than one category and
as previously stated, criteria may be applied in any order:
Technical
• Identification of the sampling unit(s)
• Optimum number of samples
• Regulatory or Action Levels
Practical, such as
• equipment limitations
• site considerations - cross contamination potential
• field and laboratory resources
• the experience of the field sampling team
• special analytical needs (low level analyses, dioxin)
• special analytical concerns (interferences, multiple phases, incompatibility)
• safety considerations
• resistant matrices (solidified material)
• investigation derived waste (IDW) generation
• limits on access to sampling locations (buildings, refusals)
• transitory events (start-up, shut-downs)
• special site concerns (unexploded ordnance)
Cost Estimates
Statistical
• Sensitivity analysis
• Statistical Power Curve
• Statistical hypothesis test
• Measurement error
The assessment step of each design constraint will result in a reduced set of designs. Following the
completion of the design constraints, final selection of the most resource-effective design occurs. However,
19
-------
if no design meets the requirements of the project objectives, then either the sampling design or the project
objectives will need to be modified.
Design modifications include increasing the number of samples, using design tools such as compositing, or
satisfying practical limitations. If significant changes are made, then the "new" alternate designs would
require a repeat of the optimization process with a re-calculation of the number of samples, re-development
of the cost estimate, and a final review of the potential practical constraints.
Modifications to the project objectives may include a change in the study boundaries, an increase in the
tolerable design errors, a relaxation of project constraints, or modifications in the initial hypotheses. A
change in the project objectives would necessitate a complete cycling through the optimization process. The
efforts to achieve an optimized sampling and analysis design ensure a defensible sampling design and
subsequently a successful project.
20
-------
Figure 1.1-Process for Optimization of Sampling Design
Problem j
Gianni
r
1
Project Objectives
Optimization
I
Initial set of feasible designs
# of samples, false + or -,
action levels,
schedule cost, etc.
it of samples, false + or -,
action levels,
schedule cost, etc,
1
t
Practical considerations, etc.
Reduced set of designs
# of samples, sample
mass, cost, etc.
# of samples, sample
mass, cost, etc.
t
t
t
Process
Iterations
Design
Iterations
Statistical considerations
1
Yes
r
Final selection criteria
of more than 1
T
t
Possilbe design modifications
• Sample mass
• Compositing
• Total variation
etc.
STOP
Possible DQO modifications
• change false +, -
• change boundaries
•etc.
21
I
-------
COMPOSITING AND SUBSAMPLING OF MEDIA RELATED TO WASTE
MANAGEMENT ACTIVITIES
Authors: Susan D. Gagner, Sampling Group Supervisor, Hazardous Waste
Management, Lawrence Livermore National Laboratory, 7000 East Ave.,
L-311, Livermore, CA 94550, Alan B. Crockett, Consulting Scientist/Engineer,
Lockheed Martin Idaho Technologies, Idaho National Engineering Laboratory, P.O.
Box 1625, Idaho Falls, ID 83415-2213
ABSTRACT
The American Society for Testing and Materials (ASTM) has been working with the
U.S. Environmental Protection Agency (EPA) to develop methods for use in the
Environmental RCRA/CERCLA program. Two standards being developed are guides
to field compositing and laboratory subsampling.
Correctly performed compositing and subsampling are critical in the chain of
sampling and analytical events. They must be accomplished in compliance with
project objectives and instructions to assure that the resulting data is representative.
In a site characterization effort, the collection of composite samples may be used to
estimate the mean concentration of a waste analyte in contaminated media. Other
reasons to composite include reducing costs, efficiently determining the absence or
possible presence of a hot spot, and, when coupled with retesting schemes, locating
hot spots. If composite samples are collected, it is necessary to ensure that a
representative sample is obtained for analysis. This may mean that samples must be
mixed and subsampled using procedures that could include pan mixing and
quartering, a mixing square, kneading, sieving and mixing, and particle size
reduction. Field subsampling procedures include use of a rectangular scoop, an
alternate scoop technique, and the slab-cake methods.
A significant source of analytical error exists in obtaining a representative subsample
of a sample delivered to a laboratory. Techniques for obtaining representative
subsamples include homogenization, layer analysis, grinding and sieving, mixing,
transversal subsampling, cone and quartering, riffling, use of a mixing square, use of
mechanical mixers, and particle size reduction. Other techniques include changing
the physical state such as digesting, drying, melting or freezing. These methods can
be applied to a variety of matrices including solid wastes, single-phased liquids,
sludges, and multilayered samples. These techniques are dependent on sample
matrix, the type of analysis performed, the characteristic of interest, and the project
specific instructions. These subsampling methods are designed to reduce the three
major sources of subsampling errors; fundamental errors, grouping and segregation
errors, and materialization errors.
Both of these guides are still in the developmental stages at ASTM.. They should
soon be available for general use.
INTRODUCTION
Data Quality Objectives are generally dependent on getting data that is
representative of the site. The more heterogeneous the site matrix, the more difficult
this is to achieve. Two areas that influence the ability to achieve a representative
sample are compositing in the field, and subsampling in the laboratory. Due to the
22
-------
lack of regulatory guidance in these areas, the EPA asked ASTM to write standards
addressing compositing and subsampling as part of a cooperative agreement between
these agencies.
COMPOSITING
Composite sampling (compositing) is the combining of two or more samples into one.
The principle assumption in the decision to composite is that the analytical costs are
high relative to the sampling costs. The reasons to composite include:
• improving the precision mean estimation
• reducing costs associated with analysis
• efficiently determining the absence or presence of a hot spot
• when coupled with retesting schemes, locating hot spots
• providing a degree of anonymity where population statistics are necessary.
When estimating the mean concentration, a set of composite samples from a
heterogeneous population always provides a more precise estimate of the mean than
a comparable number of discreet samples. This is because of the physical process of
averaging that occurs which provides a greater level of statistical confidence. The
composite samples tend to be more normally distributed than the individual samples.
This is an advantage for compositing, since the calculation of the mean, standard
deviation, and confidence levels generally assumes the data is normally distributed.
Spatial design of the compositing scheme is also important, since compositing can
either help determine the spatial variability of a site, or be used to improve the
precision of the parameter of interest across the whole site.
Due to the increased precision of compositing, the number of composite samples
required to achieve a specified precision is smaller than that required for individual
samples. The higher the analytical costs relative to the sampling and compositing
costs, the greater the savings to the project budget.
Samples can be composited to determine whether an individual sample exceeds a
specified limit as long as the action limit is relatively high compared with the actual
detection limit and the average sample concentration. For example, if a site was
being analyzed for PCB, the known analytical detection limit is <5 mg/kg of soil, and
the action level is 50 mg/kg, then up to 10 samples could be composited to determine
if any of the samples are a "hot spot." Depending on the difficulty and probability of
having to resample, it may be desirable to retain a split of the discreet samples for
possible re-analysis to find the specific spot. This type of composite sampling can
only be used as a cost saving measure if the possibility of finding a hot spot is
relatively low (<40%).
The principle limitation of compositing is the loss of discreet information achieved
from a single sample. However, the following situations may not lend themselves to
cost-effective compositing either:
• when the integrity of the individual sample values change due to compositing
(e.g., chemical reactions occur between constituents in the samples being
combined, or volatiles lost during mixing)
• where the composite sample cannot be properly mixed and subsampled in the
field or the whole composite cannot be analyzed
23
-------
• when the goal is to detect hot spots and a large proportion of the samples are
expected to test positive for the characteristic of interest
• when analytical costs are low relative to sampling costs (e.g. field testing
equipment)
• when regulations specify that a grab sample must be collected.
In general, the individual samples to be composited should be of the same mass.
However, proportional sampling may be appropriate in some cases, depending upon
the objective. For example, if the objective is to determine the average contaminant
concentration of the waste contained in a group of drums, the volume of each sample
to be composited should be proportional to the amount of waste in the drum. Another
example would be in estimating the contaminants concentration of soil overlaying an
impermeable layer. Soil cores should be collected and composited from the surface to
the impermeable layer, regardless of the core length
Frequently, it is necessary to mix an individual or composite sample and obtain a
representative subsample(s) for transport to an analytical lab. Even when the entire
sample collected is to be submitted to the laboratory, it may be desirable to
thoroughly mix the sample to help assure that the laboratory analyzes or extracts a
representative aliquot. The thoroughness of mixing needed for subsampling is a
function of the proportional size of the subsample. The principle problem is that the
smaller the aliquot, the less representative the aliquot may be, unless it is thoroughly
mixed, homogenized and subsampled. Compositing samples without adequate
mixing can nullify the potential benefits of compositing. Prior to mixing, project
specific instructions (e.g., the sampling and analysis plan) should be followed.
Methods that may be applicable to field mixing are dependent on matrix. This
includes field mixing and quartering in a pan, sieving, kneading, grinding, particle
size reduction, and other mixing equipment (e.g., riffle splitters, cone and quartering,
etc.). While it is not always possible to determine if a sample is adequately mixed,
following standard operating procedures and observing sample texture, color, and
particle size distribution are practical methods. If samples are sieved or large
materials are removed, it may be necessary to record the mass of materials removed
for later estimation of contaminant concentrations in the original sample.
Once the composite sample is mixed, it may need to be subsampled. The reasons for
this include a composite sample that is larger than the sample container, or the need
to split the composite into multiple sample bottles. If mixing procedures could assure
a homogenous sample, subsampling in the field would be simple. However, particles
may segregate according to size during mixing, and improper subsampling would
introduce bias. Since homogeneity cannot be assumed, appropriate subsampling
procedures in the field should be used by field personnel to achieve representative
samples. Methods for field subsampling include riffle splitters, cone and quartering,
rectangular scoop, alternate scoop, and slab cake techniques.
LABORATORY SUHSAMPT JT^H
Sources of uncertainty arise from sample heterogeneity and from the actual
subsampling procedure used. There are three general sources of error that pertain to
the subsampling of environmental samples. These must be understood before
choosing a subsampling method. These include:
24
-------
• fundamental error
• grouping and segregation error
• materialization error.
Fundamental error occurs when the contaminants of interest exist as particles or are
attached to particles that are randomly dispersed throughout the matrix while the
matrix itself is not contaminated or is contaminated with much lower concentrations
of the contaminant of interest. The measured concentration of contaminant in such a
waste will depend upon the number of these particles that are in the specimen
subjected to analysis. For a given concentration, the greater the particle size, the
fewer contaminated particles there are in the subsample, hence the smaller the
chance is that the contaminant will be appropriately represented in the specimen.
Grouping and segregation error is a function of the distribution of unlike particles
(i.e., size, shape, density) within a sample. Differences in size, shape and density can
affect heterogeneity further by the development of different strata upon agitation or
vibration of the sample.
Materialization errors are those that result from using incorrect subsampling devices
or from the incorrect use of subsampling devices.
Successful implementation of laboratory subsampling depends on effective
communication between the data user and the laboratory staff. The selection of
optimal procedures by the laboratory depends on the intended use of the data. The
data user should submit appropriate instructions with all samples and, when
necessary, the laboratory staff should contact the data user for confirmation or
further clarification of these instructions. Options should be discussed before
initiating any subsampling procedure. These options include:
• removal of artifacts, such as rocks and twigs, from the sample prior to
subsampling
• digesting or extracting the contaminant from the outside of the large particles
instead of breaking down the entire particle
• digesting or extracting particle sizes separately
• using or forming an emulsion layer so that the material may be treated as a
single layer
• separation of layers
• drying the sample
• changing the physical state, such as freezing the material so that it may be
treated as a solid, or melting the material so that it may be treated as a liquid
• analysis of only one layer of multilayered samples, such as analyzing only the
oil portion of an oil/water mixture for PCBs
• compositing portions of a sample for volatile analysis directly in a purge unit
vs individual analysis of these portions
• choice of standard method for solids, such as grinding, mixing, cone and
quartering, riffling, sieving, or particle size reduction.
Sampling theory suggests a minimum sample size that increases as the size of the
largest particle increases. Often, these minimum sample sizes are larger than the
normal sample sizes subjected to environmental analysis. Sometimes, the analytical
subsample can be increased, or the size of the subsample subjected to digestion or
extraction can be increased and a smaller subsample of the well mixed digestate or
25
-------
extract can be submitted for analysis. Another approach involves dividing a sample
of the requisite size into smaller subsamples, which can be subject to preparation.
The resulting digestates or extracts can be analyzed individually or proportionally re-
composited and a composite sample analyzed.
Subsampling techniques are different when analyzing for volatile compounds than
non-volatile compounds. The differences are included for each sample matrix.
I. Solid Matrices
If the maximum particle size is too large for the specimen used in the laboratory,
sampling theory would suggest that either particle size reduction is used, or a larger
sample size is digested. If something is known about the scale of the "Contaminant
Unit" and the mechanism of contamination, the use of particle size reduction may be
avoidable. If the contaminant unit is on the atomic or molecular scale or is a particle
much smaller than the maximum particle size of the sample, then alternative
approaches can be employed with acceptable bias.
For volatile analyses, the Table 1 sums the alternative subsampling methods, since
particle size reduction is not possible.
Method
A
Method
B
Method
C
Particle Size of
Sample
< Maximum
Particle Size
Allowed
> Maximum
Particle Size
Allowed
> Maximum
Particle Size
Allowed
Contaminant Unit
in Sample
< Maximum
Particle Size
Allowed
< Maximum
Particle Size
Allowed
> Maximum
Particle Size
Allowed
Subsampling Method
Mix with minimum
disturbance, weigh
specimen, place in
appropriate solvent.
If data user can
accommodate false
high concentration,
subsample only
smaller particles.
Another method is to
extract the sample
with the appropriate
solvent, and analyze
the solvent as a
single phase.
An alternative to
particle size
reduction is to use
larger sample sizes,
dividing the total
amount into
manageable sizes,
then combine the
extracts into the
same purge chamber.
Table 1. Subsampling Solids for Volatile Analysis
26
-------
For non-volatile analysis, Table 2 sums the subsampling methods. Gelatinous
samples, such as asphalt, may be cooled with dry ice or liquid nitrogen. This makes
it possible to grind gelatins and other semi-solids. It is also important to be
knowledgeable about the limitations on each of these subsampling methods.
Method
D
Method
E
Method
F
Particle Size of
Sample
< Maximum
Particle Size
Allowed
> Maximum
Particle Size
Allowed
> Maximum
Particle Size
Allowed
Contaminant Size
in Sample
< Maximum
Particle Size
Allowed
< Maximum
Particle Size
Allowed
> Maximum
Particle Size
Allowed
Subsampling Methods
Transversal
Subsampling
Cone and Quartering
Riffle Splitter
Grinding and Sieving
Mechanical Methods
(e.g., spiral mixer,
cement mixer, twin-
shell V blender, or
mills)
If the data user can
accommodate false
high concentrations,
large particles can be
removed, and the
methods listed in
Method D may be
used.
Particle Size
Reduction, using such
devices as cutting
mills, micro-mills,
grinding mills, or jar
mills.
Another method is to
use larger sample sizes
by extracting or
digesting several
aliquots. The extracts
or digestates may be
analyzed separately
and the results can be
weight-averaged, or
the extracts or
digestates may be
combined prior to
analysis.
Table 2. Subsampling Solids for Non-volatile Analyses
27
-------
II. Single-phased Liquids
Aqueous samples for non-volatile compounds may contain settleable solids. If the
settlcable materials are to be considered part of the sample, they must remain in
suspension, or be able to be re-suspended and remain so during the subsampling
operation. These samples are then treated as a single-phase liquid. These samples
should be gently swirled for 15 seconds or slowly inverted 6 times to assure
homogenization. If the settleable materials are not considered to be part of the
sample, the liquid subsample or specimen may be obtained by filtering, centrifuging,
or decanting the liquid portion from the solid portion. The liquid portion can also be
pipetted directly into the analytical vessel.
For volatile compounds, samples should not be mixed prior to subsampling. For low
concentration materials, the specimen is pipetted straight into the syringe barrel.
For high levels of contaminants, use a microliter syringe to collect the specimen
directly, or transfer the subsample into a solvent appropriate for analysis in an air-
tight container (dilution process).
III. Sludge
Laboratory samples with significant solids cannot be mixed vigorously enough to
suspend a solid phase without the potential of losing volatile components. One
method of handling sludge is to freeze it, and then handle it as a solid. Generally, the
most practical method is to separate them out into their component phases.
For non-volatile compounds, the liquid subsample can be obtained by filtering,
centrifuging, decanting, or by pipetting out only the liquid phase from a solution that
has settled. Prior to starting, determine the percentage of the phases either by
volume or by weight. After separating the liquid from the solid, analyze separately,
and weight average the results.
For volatile compounds, gentle centrifuging, or settling can be used to separate the
phases. After the liquid subsample is removed, using the methods described in
single-phased liquids, the liquid is decanted into a separate container. The solid
phase can then be weighed and added directly on the purge unit as quickly as
possible.
IV. Multilayered Samples
Multilayered samples are those with two or more distinct visual layers of material.
These layers may be the result of differences in density, such as liquid/liquid layers
(e.g., chlorinated solvents and water, or water and oil), liquid/solid layers (e.g.,
sludge), solid/solid layers (e.g., small rocks and large rocks), or combinations of these
layers (e.g., water, oil, and dirt). These layers may also be the result of depositional
layering, such as green clay and sandy loam from a coring sample.
The methods used in the discussion on sludges will also work for multilayered
samples. This includes the use of dry ice or liquid nitrogen to freeze the sample, or
separating these samples into their component phases.
For liquid/liquid layers, a separatory funnel may be helpful in separating the phases,
if the analysis is for non-volatile compounds. These phases can then be
28
-------
proportionally recombined in the analytical vessel. For volatile compounds, they can
be recombined directly on the purge unit to form a representative sample.
One or more of the previous methods mentioned in this paper can be utilized to
subsample a multilayered sample, depending on how many layers are encountered in
the sample.
SUMMARY
In a sampling plan, the collection of composite samples may be used to improve the
estimation of the mean contaminant levels, reduce analytical costs, determine the
absence or presence of a hot spot, and, when coupled with retesting schemes, locate
hot spots. While discreet sample information is lost in compositing, collection of
multiple composite samples from a grid cell permits estimation of within cell
sampling error. There are several mixing and field subsampling techniques available
to ensure that the sample sent to the laboratory is representative of what is in the
field..
After the field sample arrives at the laboratory, proper subsampling is a key to
ensuring that the data meets the data quality objectives. Subsampling must be
completed only after a review of the appropriate instructions, or after discussing the
options with the data user. Subsampling should then proceed in a manner dependent
on the sample matrix, the type of analysis required, and the characteristic(s) of
interest.
ACKNQWT.KnrtF.MENTS
Work by Susan Gagner performed under the auspices of the US Department of
Energy by Lawrence Livermore National Laboratory under Contract W-7405-Eng-48.
Work by Alan Crockett was funded under U.S. Department of Energy Idaho
Operations Office Contract DE-AC07-94ID13223.
REFERENCES
ASTM Draft Standard, subcommittee D34.01, New Standard Guide for Composite
Sampling and Field Subsampling for Environmental Waste Management Activities
ASTM Draft Standard, subcommittee D34.01, Standard Guide for Laboratory
Subsampling of Media Related to Waste Management Activities
29
-------
SAMPLE REPRESENTATIVENESS:
A NECESSARY ELEMENT EV EXPLOSIVES SITE CHARACTERIZATION
T.F. Jenkins*, C.L. Grant, G.S. Brar, P.O. Thome and P.W. Schumacher,
U.S. Army Cold Regions Research and Engineering Laboratory
Hanover, New Hampshire 03755;
T.A. Ranney,
Science and Technology Corporation, Hanover, New Hampshire 03755-1290
ABSTRACT
Explosives-contaminated sites are generally characterized by collecting discrete grab
samples of surface soil and shipping them to off-site laboratories for analysis. Decisions
as to whether or not site remediation is needed are made based on the results of these
analyses, assuming they represent site conditions over fairly large grids. This study was
conducted to assess the degree of short-range heterogeneity in analyte concentrations
present at explosives-contaminated sites. This information is essential if sampling
methods are to be established that provide representative samples on which informed
decisions can be based.
Soil was sampled at nine locations on three installations. At each location seven discrete
grab samples were collected in a wheel pattern of radius 61 cm (one sample from the
center and six equally spaced around the perimeter). Each of the seven samples was
homogenized in the field and duplicates were analyzed by an on-site colorimetric method
(SW846 Method 8515) as well as being sent for off-site analysis using SW846 Method
8330. Portions of the seven discrete samples were also used to form a composite sample
which was analyzed on site using the colorimetric method as well as being analyzed off
site using Method 8330.
On-site results from seven of the nine sampling locations indicated that TNT was the
contaminant present at highest concentration. This was confirmed by results from
Method 8330. TNT concentrations varied substantially at all seven sampling locations;
TNT concentrations obtained from both types of analysis varied by as much as 2 1/2
orders of magnitude within a sampling wheel. Partitioning of overall variances for data
indicated that sampling error dominated over analytical error. Therefore, the probability
of collecting discrete samples that represent average analyte concentrations is very
unlikely with these levels of heterogeneity. In order to represent this case using discrete
sample collection and analysis, many more samples would have to be used. Results from
analysis of composite samples, however, had mean concentrations representative of the
average condition, as well as low RSDs, indicating that field homogenization was adequate
for characterization. On-site analyses using the EnSys colorimetric method gave results
essentially equivalent to those from off-site analysis with Method 8330. Similar results
were obtained for the other two sampling locations, where 2,4-DNT and ammonium
picrate were the contaminants present at highest concentration. A combination of
composite sampling, on-site homogenization, and on-site colorimetric analysis provided
an inexpensive and rapid site-characterization procedure that was accurate and precise,
and provided results that were representative of site conditions.
30
-------
INTRODUCTION
Historically, the location and quantification of contamination sources at hazardous waste
sites has been conducted by collecting a series of discrete soil samples followed by
shipment of these materials to off-site analytical laboratories for characterization.
Potentially contaminated areas were often divided into grids of various dimensions, with
the dimensions of the grids determined by the per sample cost and the size of the
analytical budget allocated for that project. An implicit assumption in this approach is
that the distribution of contaminants is sufficiently homogeneous that results from the
analysis of these discrete samples can be used to represent average site conditions within
the grid. In general, the number of discrete samples collected is insufficient to address the
short-range distribution of contaminants and practically speaking all samples are
automatically assumed to be representative of average site conditions within the grid from
which they were collected.
Explosives are solids at ambient temperature, dissolve slowly and sparingly in aqueous
solution and have low vapor pressures. These properties limit modes of mobility
compared to other contaminants such as fuels or solvents. Thus the areas of high
concentrations which serve as sources for contamination of ground water remain at or near
the surface where deposited, unless the soils themselves are moved. Thus characterizing
the contamination distribution for explosives will often be possible using samples of near
surface soils.
Inexpensive, on-site, analytical methods for the most common explosives in soil have
been developed and are now in common use. These procedures appear to be sufficiently
accurate and precise to enable their use in mapping locations of contamination and, if a
sufficient number of samples are analyzed, provide estimates of short-range contaminant
heterogeneity. Intensive sampling combined with on-site analysis is an attractive means of
evaluating the degree of representativeness that can be attained using collection and
analysis of discrete grab samples to represent site contamination within grids at
explosives contaminated sites.
OBJECTIVES
The major objective of this work was to characterize the short-range heterogeneity of
contaminants at explosives contaminated sites. This was done by conducting field
sampling and analysis studies at a number of explosives contaminated sites that varied in
the type of explosives analytes present, mode of contamination, soil type, and
geohydrology. Statistical analyses of the results were conducted to determine the
following:
1. analytical error which was estimated from the pooled variances from duplicate
analyses of seven grab samples collected within a localized area. Short-range
sampling error was estimated from the variance computed from the differences of
mean values of the seven grab (soil) samples.
31
-------
2. the degree to which some form of composite sampling could be used to reduce
sampling error.
3. whether an inexpensive, colorimetric field screening method (SW846 Method
8515) could be used to provide an accurate description of contaminant distribution
and a reliable estimate of sampling error.
EXPERIMENTAL
Sampling sites
Sampling studies were conducted at three explosives-contaminated sites. The Monite site
is a small, former industrial site near Sparks, Nevada, that has about 1.5 acres of land
contaminated with TNT and DNT. The second installation sampled was Hawthorne
Army Ammunition Plant (AAP), which is located in Hawthorne, Nevada, and was a load,
assemble and pack facility operated by the Navy until 1977 when it was transferred to
Army control. The third installation sampled was Volunteer AAP near Chattanooga,
Tennessee. This installation is a TNT and DNT production facility, although it has not
actively produced these munition compounds since 1977. Three sampling locations were
selected at each facility which varied the major contaminant present, concentration, and
mode of contamination as much as possible.
Soil sampling procedure
A common sampling pattern was used for soil sampling at all three installations. A plastic
template was placed on the ground with the center at the selected sampling location, and
seven samples were collected in a wheel pattern, with sample number 1 in the center. The
radius of the wheel was 61 cm and samples arranged around the wheel were separated by
61 cm.
All seven soil samples were collected at the surface from 0 to 15 cm using a manual 5.6-
cm-diameter stainless steel hand auger. Vegetation when present was removed. Cores
were transferred to plastic zip-lock bags and taken to a processing area.
Soil processing and analysis
Soil samples from the Monite site and Hawthorne AAP were dry and mostly consisted of
a mixture of sands and gravels. Soils from Volunteer had a higher moisture content and
were composed of a higher percentage of finer-grained material. Individual samples from
all three sites were homogenized in the field and subsamples for field and laboratory
analysis removed prior to preparing composites for each location. Details on the
processing and subsampling techniques used are presented elsewhere (Jenkins et al.
1996).
Soil samples from all sampling locations were extracted with acetone as described in detail
elsewhere (Jenkins and Walsh 1992) and analyzed on site using the EnSys colorimetric
TNT method (SW846 Method 8515, EPA 1993). The major deviations from the protocol
specified by EnSys were as follows: (a) 20 g of field-moist soil was extracted with 100
32
-------
mL of acetone rather than 10 g of dried soil with 50 mL of acetone, and (b) concentration
estimates were calculated by subtracting twice the initial absorbance at 540 nm (rather
than four times the initial absorbance) from the final absorbance at 540 nm.
Soil samples for laboratory analysis were returned to the laboratory in coolers by
overnight carrier. Upon receipt they were maintained at 4°C until processed. Samples
were air dried and processed as described in detail elsewhere (Jenkins et al. 1996).
Duplicate 2.00-g subsamples from each discrete soil sample and seven replicate 2.00-g
subsamples from composites were extracted and analyzed using SW846 Method 8330
(EPA 1995).
SUMMARY AND DISCUSSION
Because of the large volume of data collected and the similarity of results from site to site,
only the individual sample results for the field screening and laboratory analyses for one
sampling location at the Monite site are presented here (Table 1). TNT was the major
analyte present at this location with concentrations varying from sample to sample over 2
1/2 orders of magnitude. Acetone extracts for field analysis were highly colored even
before reaction with the EnSys reagent. Reaction of the acetone extracts with the EnSys
reagent resulted in the development of red-colored solutions indicative of the presence of
TNT. Substantial dilutions (as high as 1:2000) were required to obtain absorbances after
reaction with the EnSys reagent in the linear range (absorbance less than 1.0 A.U. at 540
nm).
Duplicate field analyses for a given soil at sampling location 1 were in excellent agreement
(mean RSD was 3.9%), indicating that field sample homogenization was adequate.
Duplicate laboratory analyses varied to a greater extent than field analyses (mean RSD
was 11.2%), probably due to the smaller sample size used for lab analysis (2 g versus 20
g).
TNT concentrations varied by a large amount from sample to sample and were not
normally distributed. In addition, absolute variances were not homogeneous. Thus, in
order to perform analysis of variance (ANOVA), data were transformed by taking the
logarithm of individual values for both the field and laboratory results. ANOVA was
performed on both sets of log transformed data. For the field analyses, the F ratio was
233 indicating that a significant difference was detected among the seven discrete samples
at greater than the 99.5% confidence level. Results of a least significant difference test
indicated that all seven discrete samples were significantly different from each other.
Nearly identical results were obtained when ANOVA was conducted on the laboratory
results. For sampling location 1, therefore, identical conclusions were reached regarding
the magnitude of TNT concentrations and the nature of the analyte distribution using
either the results of field screening or results of laboratory analyses.
Because the mean concentrations and absolute analytical variances for various samples
from site 1 differ so drastically, we were not able to compare the uncertainties introduced
by sampling with those from analysis by partitioning variances of untransformed data
using normal distribution statistics. This was also true for four other sampling locations.
Another simple way to compare sampling and analytical uncertainties is to compare the
33
-------
ratios of mean concentrations obtained for various samples with those for duplicate
analyses from the same sample. For location 1, the ratio of highest mean concentration to
lowest mean concentration was 243 for the field analyses and 304 for the laboratory
analyses. The highest ratios for duplicates were 1.16 Tor the field analyses and 1.47 for
the laboratory analyses. Thus for this location, and the other four where distributions
were extreme, sampling error contributes many times more uncertainty than analytical
error for either field or laboratory analysis.
Results of the analyses of the composite samples at sampling location 1 were also quite
interesting. The mean of the seven field screening analyses for the composite was 13100 ±
532 ng/g in comparison to the mean of the seven discrete samples, which was 13500 ng/g
(Table 2). Clearly, analysis of the composite provides a good estimate for the mean
concentration of the area sampled.
ANOVA was conducted to compare the laboratory and field screening results for the
composite samples. The F ratio of 3.05 indicated that the results of the laboratory and
field analyses for this sampling location were not significantly different at the 95%
confidence level. This is true even with the good precision (RSDs of 4.1% and 10.1% for
field and laboratory) obtained for the analyses of these composite samples. Therefore, a
good indication of the degree of contamination could be obtained using a combination of
composite sampling and colorimetric field screening analysis.
Results from the other eight sampling locations were very similar to those presented for
sampling location 1. A summary of the ranges of concentrations for these data is
presented in Table 2. For four of these locations the analyte distributions were
sufficiently normal to enable fractionation of variances to get a direct comparison of
analytical and sampling variances. In all four cases, sampling error overwhelmed analytical
error whether on-site or laboratory results were used.
Results from analysis of composite samples for sampling locations 2-9 were quite similar
to those from location 1 (Table 2). In all cases composite results were a good
representation of the mean of the discrete samples making up that composite, and the
relative standard deviations were small indicating that on-site homogenization was
adequate for characterization. Thus a combination of composite sampling, on-site
homogenization, and on-site colorimetric analysis provided an effective, inexpensive and
rapid site-characterization strategy.
ACKNOWLEDGMENTS
Funding for this research was provided by the U.S. Army Environmental Center,
Aberdeen Proving Ground, MD, Martin H. Stutz, Project Monitor. This publication
reflects the personal views of the authors and does not suggest or reflect the policy,
practices, programs, or doctrine of the U.S. Army or Government of the United States.
The authors acknowledge Louise Parker and Marianne Walsh for providing useful review
comments on this manuscript.
34
-------
REFERENCES
Jenkins, T.F., C.L. Grant, G.S. Brar, P.O. Thorne, P.W. Schumacher and T.A. Ranney
(1996) Assessment of sampling error associated with collection and analysis of soil
samples at explosives-contaminated sites. U.S. Army Cold Regions Research and
Engineering Laboratory Special Report 96-15, Hanover, New Hampshire.
Jenkins, T.F. and M.E. Walsh (1992) Development of field screening methods for TNT,
2,4-DNT, and RDX in soil. Talanta, 39: 419-428.
EPA (1995) Nitroaromatics and Nitramines by HPLC. Second Update SW846 Method
8330.
EPA (1993) Soil Screening for Trinitrotoluene (TNT). Third Update SW846 Method
8515 (Draft), July 1993.
35
-------
Table 1. Analytical results for TNT at Monite site, sampling location
On-site Analysis Laboratory Analysis (Ug/g)
EnSys (EPA 1993) RP-HPLC (EPA 1995)
Sample usfe
Discrete Samples
la
Ib
2a
2b
3a
3b
4a
4b
5a
5b
6a
6b
7a
7b
Mean
Composites
Cl
C2
C3
C4
C5
C6
C7
Mean
StDev
42700
36900
492
507
174
154
28000
27600
24400
24400
1240
1310
327
334
13500
12900
12900
13300
14200
13000
13200
12500
13100
532
TNB
107
104
30
30
12
11
97
—
—
—
42
33
23
17
—
—
—
—
—
—
TNT
37500
45000
390
382
113
116
44400
41200
33000
22400
1170
1200
305
227
11800
13400
13600
15200
13900
15000
16100
24DNT Total
70 37700
45100
420
412
20 145
127
44500
41200
33000
22400
1210
1230
328
244
16300
11800
13400
13600
15200
13900
15000
16100
14100
1420
Table 2. Summary of on-site analytical results from sampling locations 1-9
Concentrations (|ig/g)
Location
1
2
3
4
5
6
7
8
9
Major Analyte
TNT
DNT
TNT
TNT
TNT
Ammon. Picrate
TNT
TNT
TNT
Maximum
39800
37000
115
5880
373
4200
82000
30800
34.4
Minimum
164
3490
2.3
85.1
12.9
6.1
21700
580
4.2
Discrete
Mean
13500
16100
19.8
1970
156
869
57500
9920
13.7
Composite
Mean
13100
23800
12.6
1760
139
970
55200
11300
16.6
36
-------
GUIDANCE FOR CHARACTERIZING EXPLOSIVES CONTAMINATED SOILS:
SAMPLING AND SELECTING ON-SITE ANALYTICAL METHODS1
Alan B. Crockett. Idaho National Engineering Laboratory, Lockheed Martin Idaho Technologies, P.O. Box
1625, Idaho Falls, Idaho 83415-2213; Hany D. Craig, U.S. Environmental Protection Agency, Oregon
Operations Office, 811 S.W. Sixth Ave, Portland, Oregon 97204; Dr. Thomas F. Jenkins, U.S. Army Cold
Regions Research and Engineering Laboratory, 72 Lyme Rd., Hanover, New Hampshire 03755-1290;
Wayne E. Sisk, U.S. Army Environmental Center, SFIM-AEC-ETD, Bldg. E-4430, Aberdeen Proving
Grounds, Maryland 21010-5401
ABSTRACT
A large number of defense-related sites are contaminated with elevated levels of secondary explosives.
Levels of contamination range from barely detectable to levels above 10% that need special handling due
to the detonation potential. Characterization of explosives-contaminated sites is particularly difficult due
to the very heterogeneous distribution of contamination in the environment and within samples. To improve
site characterization, several options exist including collecting more samples, providing on-site analytical
data to help direct the investigation, compositing samples, improving homogenization of samples, and
extracting larger samples. On-site analytical methods are essential to more economical and improved
characterization. On-site methods might suffer in terms of precision and accuracy, but this is more than
offset by the increased number of samples that can be run. While verification using a standard analytical
procedure should be part of any quality assurance program, reducing the number of samples analyzed by
the more expensive methods can result in significantly reduced costs. Often 70 to 90% of the soil samples
analyzed during an explosives site investigation do not contain detectable levels of contamination. Two
basic types of on-site analytical methods are in wide use for explosives in soil, colorimetric and
irmnunoassay. Colorimetric methods generally detect broad classes of compounds such as nitroaromatics
or nitramines, while Lmmunoassay methods are more compound specific. Since TNT or RDX is usually
present in explosive-contaminated soils, the use of procedures designed to detect only these or similar
compounds can be very effective. Selection of an on-site analytical method involves evaluation of many
factors including the specific objectives of the study, compounds of interest and other explosives present
at the site, number of samples to be run, sample analysis rate, interferences/cross-reactivity of the method,
skill required, analytical costs per sample, and the need for and availability of support facilities/services.
Other factors to be considered are the precision and accuracy of the on-site analytical method, but it should
be remembered that the analytical error is generally small compared to field error and that the precision and
accuracy of a method are dependent on the site (compounds present and relative concentration) and the
specific objectives. Modifications to on-site methods may improve method performance including
extracting a larger soil sample to improve the representativeness of the analytical sample, ensuring that the
shaking/extraction phase of all methods lasts at least three minutes, and evaluating the rate of extraction for
heavy soils by conducting a simple kinetic study. With appropriate use, on-site analytical methods are
valuable tools for characterization of soils at hazardous waste sites and monitoring soil remediation
operations. This paper summarizes an Issue Paper prepared for the EPA's Federal Facility Forum through
the EPA's Technology Support Center for Monitoring and Site Characterization, Characterization Research
Division, Las Vegas.
1 The VS. Environmental Protection Agency (EPA), through its Office of Research and Development, funded this research and
approved the abstract as a basis for an oral presentation. The actual presentation has not been peer reviewed by EPA. Mention of
trade names or commercial products does not constitute endorsement or recommendation for use.
37
-------
INTRODUCTION
Historical disposal practices from manufacturing, spills, ordnance demilitarization, lagoon disposal of
explosives contaminated wastewater, and open bum/open detonation (OB/OD) of explosive sludges, waste
explosives, excess propellants, and unexploded ordnance, often results in soil contamination. Facilities that
may be contaminated with explosives include active and former manufacturing plants, ordnance works,
Army ammunition plants, Naval ordnance plants, Army depots, Naval ammunition depots, Army and Naval
proving grounds, burning grounds, artillery impact ranges, explosive ordnance disposal sites, bombing
ranges, firing ranges, ordnance test and evaluation facilities, etc. A number of these facilities have high
levels of soil and groundwater contamination, although waste disposal was discontinued 20 to SO years ago.
Because of such extensive contamination, the EPA's Federal Facilities Forum determined that remedial
project managers need guidance regarding field sampling and on-site analytical methods for detecting and
quantifying secondary explosive compounds (Table 1) in soils.
Under ambient environmental conditions, explosives are highly persistent in soils and groundwater,
exhibiting a resistance to naturally occurring volatilization, biodegradation, and hydrolysis. Site
investigations indicate that TNT is the least mobile of the explosives and most frequently occurring soil
contamination problem. RDX and HMX are the most mobile explosives and present the largest
groundwater contamination problem. TNB, DNTs, and tetryl are of intermediate mobility and frequently
occur as co-contaminants in soil and groundwater.
The frequency of occurrence of specific explosives in soils was assessed (Walsh et al. 1993) using
analytical data on soils collected from 44 Army ammunition plants, arsenals, and depots, and two explosive
ordnance disposal sites. Of the 1,155 samples, a total of 319 samples (28%) contained detectable levels
of explosives. The frequency of occurrence and the maximum concentrations detected are shown in Table
2. TNT was detected in 66% of the samples and 80% of the samples if the two explosive ordnance disposal
sites are excluded. Overall, TNT or RDX or both were detected in 72% of the samples containing explosive
residues, and 94% if the ordnance sites are excluded. Thus, by screening for TNT and RDX at these
facilities, 94% of the contaminated areas could be identified (80% if only TNT was determined). This
demonstrates the feasibility of screening for one or two compounds.
OVERVIEW OF SAMPLING AND ANALYSIS FOR EXPLOSIVES IN SOIL
The environmental characteristics of munitions compounds in soil indicate that they are extremely
heterogenous in spatial distribution. Concentrations range from non-detectable levels (<0.5 ppm) to percent
levels (> 10,000 ppm) for samples collected within several feet of each other, hi addition, the waste disposal
practices at these sites, such as OB/OD, exacerbate the problem and may result in conditions ranging from
no soil contamination up to solid "chunks" of bulk high explosives, such as TNT or RDX. Secondary
explosives concentrations above 10% (> 100,000 ppm) in soil are of concern because of potential reactivity
that affects sampling and materials handling processes during remediation.
Reliance on laboratory analyses only for site investigations may result in poor characterization and high
analytical costs ($250-350/sample), especially when a large percentage of the samples (up to 80%) contain
non-detectable levels of explosives. Because of the extremely heterogeneous distribution of explosives in
soils, on-site analytical methods are a valuable, cost-effective tool to assess the nature and extent of
contamination. Since costs per sample are lower, more samples can be analyzed, the availability of near-
38
-------
real-time results permit redesign of the sampling scheme while in the field and on-site screening also
facilitates more effective use of off-site laboratories.
DATA QUALITY OBJECTIVES
EPA's Data Quality Objectives process is designed to facilitate the planning of environmental data collection
activities by specifying the intended use of the data (what decision is to be made), the decision criteria
(action level), and the tolerable error rates. Integrated use of on-site and laboratory methods for explosives
in soil facilitate achieving such objectives as determining the horizontal and vertical extent of contamination,
obtaining data to conduct a risk assessment (EPA 1992), identifying candidate wastes for treatability
studies, identifying the volume of soil to be remediated, determining whether soil presents a potential
detonation hazard (reactive according to Resource Conservation and Recovery Act regulations), determining
when sufficient excavation has occurred, and determining whether remediation activities have met the
cleanup criteria (typically 10 to 100 ppm).
UNIQUE SAMPLING DESIGN CONSIDERATIONS FOR EXPLOSIVES
Heterogeneity Problems and Solutions - In a recent study, Jenkins et al. (1996) collected and analyzed seven
soil cores within a radius of 2 ft from nine locations. Results showed extreme variation in concentration in
five of the nine locations and in all cases, only a small fraction of the total error was due to analytical error;
field sampling error dominated total error. To improve site characterization, the major effort should be to
increase sampling densities and composite samples to reduce sampling error. There are several practical
approaches to reducing overall error during characterization of soils contaminated with explosives, including
increasing the number of samples or sampling density, collecting composite samples, using a stratified
sampling design, and reducing within sample heterogeneity.
One simple way to improve spatial resolution is by collecting more samples on a finer sampling grid such
as a 5-m instead of a 10-m spacing. This approach has been rejected in the past because of the higher costs
but when inexpensive on-site analytical methods are used, this approach becomes feasible.
Samples are always taken to make inferences to a larger volume of material, and a set of composite samples
provides a more precise estimate of the mean than a comparable number of discrete samples. This occurs
because compositing is a "physical process of averaging." Decisions based on a set of composite samples
provides greater statistical confidence than for a comparable set of individual samples (Gagner and Crockett
1996). In Jenkins' study, composite samples were much more representative of each plot than the
individual samples that made up the composites. Using a composite sampling, it is possible to reduce costs
and the total number of samples collected while improving characterization.
Stratified sampling can also be effective in reducing field and subsampling errors. Using historical data and
site knowledge or results from preliminary field screening, it may be possible to identify areas where
contaminant concentrations are expected to be moderately heterogeneous (pond bottom) or extremely
heterogeneous (open detonation sites). Different compositing and sampling strategies can be used to
characterize different areas that can result in a more efficient characterization. Another means of
stratification is based on particle size. Since explosive residues often exist in a wide range of particle sizes
(crystals to chunks), it is possible to sieve samples into various size fractions which may reduce
heterogeneity.
39
-------
Within sample heterogeneity is frequently observed during field screening when duplicate subsamples are
analyzed and results differ by an order of magnitude. To reduce within sample heterogeneity and obtain a
representative analytical sample, two methods can be employed: homogenization and extraction or analysis
of a larger sample. The smaller the subsample, the more homogeneous the sample should be and this may
require sample drying, grinding, and riffle splitting (Gagner and Crockett 1996).
While sample-mixing procedures such as sieving to disaggregate particles, mixing in plastic bags, etc.
should be used to prepare a sample, extracting a larger sample is perhaps the easiest method of improving
representativeness. CRREL recommends extraction of 20 g of soil, and the same approach can easily be
used to improve results with most field screening methods.
Sample Holding Times and Preservation Procedures - Based on spiking clean soils with explosives in
acetonitrile, Maskarinec et al. (1991) recommended the following holding times and conditions: TNT-
immediate freezing and 233 days at -20°C; DNT-107 days at 4°C; RDX-107 days at 4°C; and HMX-52
days at 4°C. Grant et al. (1993,1995) spiked soils with explosives dissolved in water to eliminate any
acetonitrile effects and also used a field-contaminated soil. The results on spiked soils showed that RDX
and HMX are stable for at least 8 weeks when refrigerated (2°C) or frozen (-15°C). When nitroaromatics
are of interest, spiked samples should be immediately frozen as some results showed significant TNT and
TNB degradation within 2 hours. However, both compounds and 2,4-DNT can be adequately preserved for
8 weeks or longer by freezing. The results on field-contaminated soils did not show the rapid degradation
of TNT and TNB observed in the spiked soils and refrigeration appeared satisfactory. Presumably, the
explosives still present in the Geld soil after many years of exposure are less biologically available than in
the spiked soils. Explosives in air-dried soils are stable at room temperature if kept in the dark.
Acetonitrile extracts of soil samples are expected to be stable for at least 6 months under refrigeration.
Acetone extracts are also thought to be stable if stored in the dark under refrigeration.
Explosion Hazards and Shipping Limitations - EPA regions and the U.S. Army Environmental Center
consider soils containing more than 10% secondary explosives (i.e., TNT, RDX, HMX, DNT, TNB, and
DNB) by weight to be susceptible to initiation and propagation (EPA 1993). If on-site analyses indicate
that soil samples contain less man 10% total secondary explosives by weight, they can be shipped to off-site
laboratories as environmental samples. Samples with over 10% explosives must be shipped to an
explosives-capable laboratory for analysis, and they must be packaged and shipped in accordance with
applicable Department of Transportation and EPA regulations for reactive hazardous wastes and Class A
explosives (AEC 1994). For sampling at sites with unknown or greater than 10% by weight of secondary
explosives contamination, special sampling procedures must be followed (AEC 1994).
SUMMARY OF ON-SITE ANALYTICAL METHODS FOR EXPLOSIVES IN SOIL
Ideally, screening methods would provide high-quality data on a near-real-time basis at low cost and of
sufficient quality to meet all intended uses including risk assessments and final site clearances without the
need for more rigorous procedures. While the currently available screening procedures may not be ideal (not
capable of providing compound specific concentrations of multiple compounds simultaneously), they have
proven very valuable during the characterization and remediation of numerous sites. Currently available
field methods that have been evaluated against standard analytical methods and demonstrated in the field
include colorimetric and immunoassay methods (Table 1). Each method has relative advantages and
disadvantages, so one method may not be optimal for all applications. To assist in the selection of one or
more screening methods for various users needs, a fairly comprehensive table (Table 3 in Crockett et al.
40
-------
19%) has been developed comparing the available colorimetric and immunoassay screening tests for
detecting explosives in soil. The selection criteria presented include: method type, analytes determined,
detection limit and range, sample preparation and extraction procedure, analytical production rate,
interferences and cross-reactivities, recommended QA/QC, suggested storage conditions and shelf life, skill
required, availability of training, cost per sample, references to tests evaluating the method against Method
8330, and additional method selection considerations. The table is accompanied by explanatory text.
Interferences/Cross-Reactivitv - A major difference among the field methods is with interferences for
colorimetric methods and cross-reactivity for immunoassay methods. The colorimetric methods for TNT
and RDX are broadly class sensitive, that is, they respond to many other similar compounds (nitroaromatics
and nitramines/nitrate esters, respectively). Immunoassay methods are relatively specific for the primary
target analytes. The cross-reactive secondary target analytes for TNT are mainly other nitroaromatics but
this varies considerably among the four TNT immunoassay test kits. Depending upon the sampling
objectives, broad sensitivity or specificity can be an advantage or disadvantage. If the objective is to
determine whether any explosive residues are present in soil, broad sensitivity is an advantage. For the
CRREL and EnSys colorimetric methods, the color development of the extracts can give the operator an
indication of what type of compounds are present in soil. An advantage of some colorimetric methods is
they can be used to detect compounds other than the primary target analyte. For example, the colorimetric
RDX methods can be used to screen for HMX when RDX levels are relatively low, and for NQ, NC, NG,
and PETN in the absence of RDX and HMX.
For colorimetric methods, interference is defined as the positive response of the method to secondary target
analytes or co-contaminants similar to the primary target anah/te. For TNT methods, the primary target
anah/te is TNT, and the secondary target analytes are other nitroaromatics TNB, DNB, 2,4-DNT, 2,6-DNT,
and tetryl. For RDX methods, the primary target anah/te is RDX, and the secondary target analytes are
nhramines (HMX, NQ), and nitrate esters (NC, NG, and PETN). If the primary target anah/te is the only
compound present in soil, the colorimetric methods measure the concentration of that compound If multiple
analytes are present in soil, the field methods measure the primary target analyte plus the secondary target
analytes, nitroaromatics for the TNT test kit, and nitramines plus nitrate esters for the RDX test kits. In
addition, the response of colorimetric methods to the secondary target analytes are equivalent to that of the
primary target analyte, and remain constant throughout the concentration range of the methods.
For immunoassay methods, cross-reactivity is defined as the positive response of the method to secondary
target analytes or co-contaminants similar to the primary target analyte. For TNT methods, the primary
target anatyte is TNT, and the secondary target analytes are nitroaromatics TNB, DNTs, Am-DNTs, and
tetryl. For RDX methods, the primary target analyte is RDX, and there is little cross-reactivity, 3% with
HMX. If the primary target analyte is the only compound present in soil, the immunoassay methods
measure the concentration of that compound. If multiple analytes are present in soil, the immunoassay
methods measure the primary target analyte plus some percentage of the cross-reactive secondary target.
Both colorimetric and immunoassay methods may be subject to positive matrix interference from humic
substances in soils. For colorimetric methods, this typically occurs below 10 ppm, and is indicated by
yellow extracts. These interferences can be reduced by careful visual analysis prior to colorimetric analysis.
Nitrate and nitrite, common plant nutrients in soil, are potential interfercnts with the CRREL and EnSys
colorimetric procedures for RDX. An extra processing step can be used to remove these interferents in soils
that are rich in organic matter or that may have been recently fertilized.
41
-------
Comparisons to Laboratory Method. SW-846 Method 8330 - Precision and accuracy of the screening
methods are most appropriately assessed by comparison to established laboratory methods such as EPA
Method 8330. Methods of comparison that have been used include relative percent difference (RPD), linear
regression, correlation, percent false positive and false negative results, analysis of variance and paired
Wests. It should also be remembered that analytical accuracy is generally quite small compared to total error
(field error is the major contributor).
Three studies have evaluated multiple methods under sh'ghtly different field conditions. Readers consult the
original studies for more details; however, some summary conclusions from the three cited studies follow.
An EPA (1995) study compared the CRREL, EnSys (colorimetric), D Tech, Quantix, and Ohmicron
methods for TNT and concluded that, overall, "no single method significantly out-perfonned other rnethods"
and that accuracies for all the field screening methods were comparable, with CRREL, EnSys, and Ohmicron
being more accurate in the greater-than-30-mg/Kg TNT ranges, and D Tech being more accurate in the less-
than-30-mg/Kg range. The same study compared CRREL, EnSys, and D Tech methods for RDX in soil and
concluded mat they were slightly less accurate than the corresponding TNT methods. Haas and Simmons
(1995) evaluated immunoassay kits for TNT (EM Science/D Tech, EnSys/EnviroGard Tube and Plate,
Idetek/Quantix, and Ohmicron/RaPID Assay. They concluded that for semi-quantitative screening, all kits
have the potential to accurately screen soil samples for contamination at risk-based levels. For quantitative
analyses, "several of the assays had significant positive bias compared with HPLC results below 1 ppm,"
"measurements near the detection limit are often problematic," and "above 1 ppm, the correlation between
[immunoassay kits] and HPLC was generally good." Myers etaL( 1994) evaluated and compared the EnSys
(cokximetric) and D Tech methods for TNT in soil versus EPA Method 8330. "EnSys demonstrated a good
one-to-one linear correlation with RP-HPLC that can be attributed to the procedure for extraction, i.e., a
large sample size of dried homogenized soil." For the D Tech kit, comparison was more difficult due to the
concentration range type data and because "one-to-one linear correlation with RP-HPLC was poorer.M The
EnSys kit was well suited for studies requiring good quantitative agreement with the standard laboratory
method and mat the D Tech kit was "better suited for quick, onsite screening in situations where all samples
above a certain range will be sent forward to a laboratory for confirmation by the standard method."
Emerging Methods and Other Literature Reviewed - Other screening procedures are being used but limited
information is available on them. Emerging procedures include: an antibody-based continuous-flow
immunosensor for TNT and RDX and a fiberoptic biosensor for TNT are being evaluated by the Navy for
use in soil, the U.S. Army is developing a cone penetrometer for in situ detection of explosives, ion mobility
Spnll'UinHiy is bang evaluated by y"t"l rrymi-ratims * modified Method 833Q has been used in a mnhile
trailer, thermal desorption followed by gas chromatography/mass spectrometry analysis has been reported,
and work is underway within CRREL to investigate the use of a simple thin-layer chromatographic method
for use as a confirmation test following colorimetric-based procedures.
SUMMARY
The heterogeneity of explosive in soils poses significant problems to site characterization. Several options
exist including collecting more samples, providing on-site analytical data to help direct the investigation,
sample compositing, improving homogenization of samples, and extracting larger samples. On-site
analytical methods are essential to more economical and improved characterization. What the on-site
methods lack in terms of precision and accuracy in simultaneously identifying specific multiple compounds,
they more than make up for in the increased number of samples that can be run.
42
-------
Modifications to on-site methods may be able to improve method performance. In most cases, a larger soil
sample can be extracted to improve the representativeness of the analytical sample. Also, with heavy soils
or soils with high organic matter content, it may be useful to conduct a short-term kinetic study to determine
whether a 3-minute extraction period is adequate. It is recommended that the shaking/extraction phase of
all methods last at least 3 minutes, hi all cases, it is recommended that a portion of the on-site analytical
results be confirmed using a standard laboratory method.
ACKNOWLEDGMENT
Work partly performed under the auspices of the U.S. Department of Energy, Office Contract No. DE-AC07-
94ID13223, through Inleragency Agreement No. DW89937192-01-2 with the U.S. Environmental Protection
Agency.
REFERENCES
AEC. 1994. Standard Comments for Health and Safety Document Review, Memorandum for Record, SFIM-AEC-
TSS, July 18,1994, U.S. Army Environmental Center., Aberdeen Proving Ground, MD, 9 p.
Crockett, A.B., H.D. Craig, T.F. Jenkins, and W.E. Sisk. 19% draft. On-Site Analytical Methods and Field
Sampling for Explosives in Soil, EPA Federal Facilities Forum Issue, EPA, CRD, Las Vegas, NY*.
EPA. 1992. Guidance for Data Useability in Risk Assessment (Part A). Final Report, OSWER Directive 9285.7-
09A, Office of Emergency and Remedial Response, U.S. EPA, Washington, D.C.
EPA. 1993. Handbook: Approaches for the Remediation of Federal Facility Sites Contaminated with Explosive or
Radioactive Wastes, EPA/625/R-93/013, OR&D, EPA, Washington, D.C., 116 pp.
EPA. 1995. Field Screening Technologies Umatilla Explosive Washout Lagoon Soils, prepared by Black &
Veatch Waste Science, Inc., Tacoma, WA for U.S. EPA Region 10, Seattle, WA, draft.
Gagner, S.D. and A.B. Crockett. 1996. Compositing and Subsampling of Media Related to Waste Management
Activities, hi Proceedings Twelfth Annual Waste Testing and Quality Assurance Symposium, ACS,
Washington D.C.
Grant, C.L., T.F. Jenkins, and S.M. Golden. 1993. Experimental Assessment of Analytical Holding Times for
Nitroaromatic and Nitramine Explosives in Soil, Rept 93-11, USAGE., 18 pp.
Grant, C.L. et al.. 1995. Holding-time Estimates for Soils Containing Explosives Residues: Comparison of
Fortification vs. Field Contamination., Environ. Tox. and Chem. 14(11): 1865-1874.
Haas, R.A. and B.P Simmons. 1995. Measurement of TNT and RDX in Soil by Enzyme Immunoassays and High
Performance Liquid Chromatography (EPA Method 8330), California EPA, Hazardous Materials Lab.
Jenkins, T.F., C.L. Grant, G.S. Brar, P.G. Thome, and T.A. Ranney. 19%. Assessment of Sampling Error
Associated with Collection and Analysis of Soil Samples at Explosive Contaminated Sites (Phase 1):
Status Report, Special Report 96-15, CRREL.
Maskarinec, M.P., C.K. Bayne, L.H. Johnson, S.K. Holladay, R.A. Jenkins, and B.A. Tomkins. 1991.
Stability of Explosives in Environmental Water and Soil Samples, ORNI/TM-11770, ORNL, Oak Ridge,
TN,98pp.
Myers, K.F., E.F. McCormick, A.B. Strong, P.G. Thome, and T.F. Jenkins. 1994. Comparison of Commercial
Colorimetric and Enzyme Immunoassay Field Screening Methods for TNT in Soil, TR-IRRP-94-4,
USACE-WES, Vicksburg, MS.
Thome, P.G. and T.F. Jenkins. 1995. Development of a Field Method for Ammonium Picrate/Picric Acid in Soil
and Water, Special Report 95-20, U.S. Army Cold Regions Research and Engineering Laboratory, Hanover,
NH,22pp.
Walsh, M.E. 1989. Analytical Methods for Determining Nitroguanidine in Soil and Water, Special Report 89-35,
U.S. Army Corps of Engineers, Cold Regions Research and Engineering Laboratory.
Walsh, ME., T.F. Jenkins, P.S. Schnitker, J.W. Elwell, and M.H. Stutz. 1993. Evaluation of SW-846 Method
8330 for Characterization of Sites Contaminated with Residues of High Explosives, Special Report 93-5,
Army Cold Regions Research and Engineering Laboratory, Hanover, NH, 17 pp.
43
-------
Table 1. Analytical Methods for Commonly Occurring Explosives, Propellants, and Impurities/Degradation Products.
Acronym Compound name
Field
method
Developer/test kit
Nltnmromatics
TNT 2,4,6-trinitrotoluene
TNB 1,3,5-trinitrobenzene
DNB 1,3-dinitrobenzene
2,4-DNT 2,4-dinitrotoluene
2,6-DNT 2,6-dinitrotoluene
Tetryl Methyl-2,4(6-trinitrophcnylnitramine
2AmDNT 2-amino-4,6-dinitrotoluene
4AmDNT 4-amino-2,6-dinitrotoluene
NT nitrotoluenc (3 isomers)
NB nitrobenzene
Nttramlnes
RDX Hexahydro-],3,5-trinitro-l,3,5-triazine
HMX Octahydro-I3,S,7-tetnuiitro-134,7-tetrazocine
NQ Nitroguanidine
Nitrate Esters
NC Nitrocellulose
NG Nitroglycerin
PETN Pentaerythritoltctranitrate
Ammonium Plcnte/Pkrk Acid
AP/PA Ammonium 2,4,6-trinitrophenoxide/
2,4,6-trinitrophenol
Cs CRREL, EnSys/Ris«®
Cp CRREL, EnSys/Rii1®
Cp USAGE
Ip EMScience/D-TECH™
Ip Idetek/Quantix™
Ip Ohmicrcm/RaPID Assay*
Ip EnSys/EnviroGard™
Cs CRREL, EnSys/Ris5®
Is Ohrnrcron/RaPID Assay®
Cs CRREL, EnSys/Ris5®
Cp,Cs CRREL
Cs, Is CRREL, EnSys/EnviroGard™
Cs CRREL
Is EnSys/EnviroGard™
Cs CRREL, EnSys/Ris1®
Cp CRREL, EnSys/Ris5®
Ip EMScfence/D-TECH™
Cs CRREL, EnSys/Ris5®
Cs CRREL
Cs CRREL
Cs CRREL
Cs CRREL
Cs CRREL
CRREL
Cp CRREL,
Is EM Science/D-TECH™?
Lab.
method
N
N
N
N
N
N
N
N
N
N
N
N
N
N
0
L
P
P
A ' Ammonium Picrate/Pichc Acid (Thome and Jenkins 1995).
Cp = Colorimdric field method, primary Urget analyte(i).
C« = Colorimetric field method, secondary target analyte(i).
O - Nhroguanidine (Walih 1989).
Ip - Immuno«5My field method, primary target analyte.
Is »Immunoauiy field method, secondary target analyte(i).
L - Nhrocelluloee (Walxh unpublished CRREL method).
N - EPA SW-846, Nhroaromatics and Nitramines by HPLC,
Method 8330.
P - PETN and NO (Walsh unpublished CRREL method).
Table 2. Occurrence of Analytes Detected in Soil
Contaminated with Explosives.
Nitroarotnatics
TNT
TNB
DNB
2,4-DNT
2,6-DNT
2-AmDNT
4-AmDNT
Tetryl
Nitramines
RDX
HMX
TNT/RDX
•/.Sample with
Analyte Present
66
34
17
45
7
17
7
9
27
12
72
Maximum
Level
(*
-------
Performance of a new disposable sampling and storage device for soil VOCs.
David Turriff, Ph.D., Chris Reitmeyer, Lloyd Jacobs and Nils Melberg
En Chem, Inc., 1241 Bellevue St, Green Bay, WI 54302
A stainless steel volumetric sampling device(the EnCore sampler) has previously been
demonstrated to be a very effective device for collecting target volumes of soil and for
storing VOCs with limited loss. These sampling devices overcome the limitations that
exist with other methods. Handling of solvents in the field is eliminated, subsampling of
the soil is not required, and the sample can be easily and quantitatively transferred to
standard 40 ml vials in the laboratory.
We report here on the development and validation of a disposable sampling tool that can
be used equivalently to the stainless steel sampler. Data will be presented on the inertness
of the composite material and its leaching potential when exposed to solvents. Data will
be presented to validate the performance of the sampler for short term storage of soil
VOCs and on extension of the time of storage stability under freezing conditions(<-15
degrees C) to fourteen days.
45
-------
FIELD SCREENING OF SOILS CONTAMINATED WITH EXPLOSIVES
USING ION MOBILITY SPECTROMETRY
Alan B. Crockett. David A. Atkinson, Idaho National Engineering Laboratory, Lockheed Martin
Idaho Technologies, Idaho Falls, ID 83415-2213; and Thomas F. Jenkins, U.S. Army Cold Regions
Research and Engineering Laboratory, Hanover, NH 03755-1290
ABSTRACT
Soils contaminated with explosives constitute a high priority problem at some Department of Energy
focflifes and many Army installations. Because explosives in soil are often heterogeneously distributed,
field screening is essential to characterize sites more quickly, economically and accurately. Current
immunoassay and colorimetric field screening procedures have proven useful, but have significant per
sample costs and limited throughput. Often, only a single analyte or anaryte group determination is
possible per sample. At present, several field screening procedures are available for TNT in soil, three
procedures for RDX and one procedure for 2,4-DNT and ammonium picrate/picric acid (AP/PA). Ion
mobility spectrometry has been used for several years in law enforcement work to detect explosives in
air at ppt levels, but very little work has been done to apply the technique to detecting explosives in soils.
IMS offers great potential since many compounds (TNT, RDX, PETN, DNT, TNB, NG, etc.) can be
quantified simultaneously in an acetone extract within a few seconds and at a cost of under a
dollar/sample. This study involved the comparison of IMS screening with EPA's standard method for
explosives, Method 8330. The U.S. Army provided a large number of soil samples that had been
collected from three locations at each of three explosive contaminated installations. The samples, had
been dried, ground, homogenized and analyzed in duplicate by Method 8330. Duplicate two gram
aliquots of these samples were extracted with 10 mL of acetone by shaking for three minutes, allowed
to settle, and then analyzed by IMS for Method 8330 compounds. Half of the extracts were also
analyzed in duplicate by QMS. The results of the statistical comparison between IMS and Method 8330
data will be presented Based on these results, the intention is to provide adequate validation data to
EPA for inclusion of the method as a screening procedure in SW-846. This comparison test was run on
specially homogenized samples to reduce subsampling errors and improve the quality of the method
comparison. For field application, the extraction procedure should be changed to extract 20 g of moist
soil with 100 mL of acetone. The chief limitation of the IMS method is the initial cost for
instrumentation (~$50 K).
46
-------
8
IDENTIFICATION AND QUANTITATION OF
PETROLEUM SUBSTANCES IN ENVIRONMENTAL
SAMPLES USING FRIEDEL-CRAFTS/HANBY
SPECTROPHOTOMETRY WITH CHEMOMETRICS
John D. Hanby, Hanby Environmental Laboratory Procedures, Inc., 501 Sandy Point
Road, Wimberley, TX78676
John Michael Hiller, Locheed-Martin Energy Systems, Oak Ridge National Laboratories,
P.O.Box 2009, Oak Ridge TN 37831
Visible (400-650 nm) spectral signatures of aromatic compounds provided by vigourous
Friedel-Crafts reactions of extracts of soil and water samples have been utilized to provide
qualitative (identification) analyses as well as quantitative information. Extract solutions
were prepared with a broad range of mono- and polynuclear aromatic compounds utilizing
a heptane/carbon tetrachloride solvent. Friedel-Crafts reactions were performed on the
extracts utilizing stoichiometrically large amounts of A1C13 (Hanby Method). Precipitates
were scanned utilizing a small, portable, reflectance spectrophotometer developed by
H.E.L.P., Inc. Spectral information obtained across the range of 400-650 nm utilizing
SpectraScope software was then subjected to interpretation by the Mathworks(TM)
Matlab program and the Chemometrics and Neural Network toolboxes. Quantitation over
a range of 1-1000 mg/Kg was achieved and the neural network software was able to
provide identification of the various compounds based on their spectral signature over the
400-650 nm range. Approximately 20 crude and refined oils and fuels as well as the
individual compounds have been successfully analyzed with the method in similar
contamination ranges (1-1000 mg/Kg). Also, liquid/liquid extracts of aromatic compounds
and fuels,etc. have been prepared and analyzed with similar success in the .05 to 50 mg/L
range. Portable (notebook) computers have been loaded with the appropriate software
providing on-site use of the procedure. Spectral signature information is amenable to
interpretation of source and date of contamination with appropriate library data.
47
-------
Abstract
Site Characterization Technology Demonstrations
Gary Robertson, Stephen Billets, Eric Koglin
U.S..Environmental Protection Agency
National Exposure Research Laboratory
Characterization Research Division
Las Vegas, Nevada 89119
The Consortium for Site Characterization Technology (CSCT), which is coordinated by
the U.S. EPA Characterization Research Division in Las Vegas, has developed a Guidance
Manual for the Preparation of Site Characterization Technology Demonstration Plans. The
CSCT is a partnership between the EPA, the Departments of Defense and Energy (DoD and
DOE) and the private sector, all of whom share the goal of increasing the use of innovative
characterization technologies in assessing contaminated sites. The CSCT conducts
demonstrations to show that new and innovative field-deployable measurement technologies
perform as claimed by the developers and to provide a verified data set that confirms the
performance of the technology. These demonstrations are intended to give potential users the
information needed to determine the applicability of the technology for their uses. This manual
was used as guidance for three technology demonstrations during 1995. The three
demonstrations, Cone Penetrometer-Laser Induced Fluorescence, Volatile Organic Compounds
by Field Gas Chromatography/Mass Spectrometry, and Metals by Laser Induced Breakdown
Spectroscopy, will be briefly described. Lessons learned about planning and performing
technology demonstrations will be presented. Two major areas of concern include developing
specific developers'claims for the technology to be demonstrated and assuring the quality of the
data produced by the reference laboratories.
48
-------
10
ABSTRACT
A Field Demonstration of Portable Mass Spectrometers
Gary Robertson, U.S. Environmental Protection Agency
National Exposure Research Laboratory
Characterization Research Division
Las Vegas, Nevada 89119
Susan Bender, Sandia National Laboratory
Albuquerque, NM 87185
A field demonstration of portable mass spectrometers was held at two hazardous waste
sites. These demonstrations were sponsored by the Consortium for Site Characterization
Technology (CSCT). The CSCT is a partnership between the EPA, the Departments of Defense
and Energy (DoD and DOE) and the private sector, all of whom share the goal of increasing the
use of innovative characterization technologies in assessing contaminated sites. The CSCT
conducts demonstrations to show that new and innovative field-deployable measurement
technologies perform as claimed by the developers and to provide a verified data set that
confirms the performance of the technology. These demonstrations are intended to give potential
users the information needed to determine the applicability of the technology for their uses. The
demonstrations were held at the DOE Savannah River Site, Aiken, South Carolina in July 1995
and at Wurtsmith Air Force Base in Oscoda, Michigan in September 1995. Instruments
demonstrated included the Bruker-Franzen Analytical EM640 mobile gas chromatograph/mass
spectrometer, the Teledyne Electronic Technologies 3DQ Discovery ion trap mass spectrometer
with a direct injection inlet system (developed by Martin Marietta Energy Systems at Oak Ridge
National Laboratory), and the Viking Instruments Corporation SpectraTrak 672 transportable gas
chromatograph/mass spectrometer. The demonstration was designed to evaluate whether the
instruments met the developers' performance claims for a variety of media and analytes under
field conditions. At each location, samples of soil gas, ground water, and soil were analyzed for
volatile organic compounds. The instruments were operated by developer personnel. The
samples were analyzed by developer-selected methods. Split samples were analyzed by standard
analytical methods at commercial laboratories to provide a reference for verifying the
performance of the instruments. The study design, execution, and results will be presented.
49
-------
11
Title: Comparison of Soil Gas, Heated Headspace, and Methanol Extraction Techniques for Soil
Volatile Organic Compound Quantification
Authors:
M. M. Minnich
Lockheed Environmental Systems & Technologies, Inc.
980 Kelly Johnson Drive
Las Vegas, NV 89119
B. A. Schumacher
U.S. Environmental Protection Agency
National Exposure Research Laboratory
Characterization Research Division - Las Vegas
P.O. Box 93478
Las Vegas, NV 89193-3478
Abstract:
Three soil volatile organic compound (VOC) measurement techniques were compared at
two hazardous waste sites located within different climatic and geologic regions. Dissimilar soil
properties found at the sites also allowed for the examination of the influence of the differing soil
properties on VOC quantification. A total of 41 soil gas, 52 heated headspace, and 51 methanol
extraction/purge-and-trap measurements were obtained on collocated samples. Contaminants
present at both sites included the following chlorinated hydrocarbons (CHCs): cis-1,2-
dichloroethene, 1,1,1-trichloroethane, trichloroethene, and tetrachloroethene. Heated headspace
offered the highest sensitivity as indicated by the greatest percentage of detections per number of
analyses. The statistical regression between headspace concentrations and methanol extraction
concentrations was highly significant (p<0.001) with a r2 = 0.53. Headspace concentrations were
approximately 20% to 30% of the methanol extraction concentrations indicating that the methanol
was able to extract significantly more of the CHCs than the headspace extraction, even in soils
with relatively low organic carbon contents (< 0.25%). None of the soil properties, viz.,
gravimetric moisture content, organic carbon content, percent sand, or percent clay, significantly
improved the regression fit. The soil gas responses were unlike either headspace or methanol
extraction data. CHC measurement by vapor extraction/soil gas could not be used to predict soil
CHC concentrations at these sites.
Notice: The U.S. Environmental Protection Agency, through its Office of Research and
Development (ORD), prepared this abstract for a proposed poster presentation. It does not
necessarily reflect the views of the EPA or ORD.
50
-------
12
EVALUATION OF A STANDARD TEST METHOD FOR
SCREENING FUELS IN SOILS
Susan S. Sorini, Principal Scientist, and John F. Schabron, Manager Chemical Monitoring
Division, Western Research Institute, 365 N. 9th Street, Laramie, Wyoming 82070
ABSTRACT
A new screening method for fuel contamination in soils was recently developed as American
Society for Testing and Materials (ASTM) Method D-5831-95, Standard Test Method for
Screening Fuels in Soils. This method uses low-toxicity chemicals and can be used to screen
organic-rich soils, as well as being fast, easy, and inexpensive to perform. Fuels containing
aromatic compounds, such as diesel fuel and gasoline, as well as other aromatic-containing
hydrocarbon materials, such as motor oil, crude oil, and coal oil, can be determined.
The screening method for fuels in soils was evaluated by conducting a collaborative study on
the method. In the collaborative study, a sand and an organic soil spiked with various
concentrations of diesel fuel were tested. Data from the collaborative study were used to
determine the reproducibility (between participants) and repeatability (within participant)
precision of the method for screening the test materials. The collaborative study data also
provide information on the performance of portable field equipment (patent pending) versus
laboratory equipment for performing the screening method and a comparison of diesel
concentration values determined using the screening method versus a laboratory method.
INTRODUCTION
A field method for screening fuel contamination in soils was developed within American
Society for Testing and Materials (ASTM) Main Committee D-34 on Waste Management (1).
This test method is ASTM Method D-5831-95, Standard Test Method for Screening Fuels
in Soils (2). Unlike many of the existing methods for screening fuel contamination in soils,
the ASTM method provides a fast, easy, and inexpensive procedure that uses low-toxicity
chemicals and can be used to screen organic-rich soils.
The method calls for extracting a soil sample with isopropyl alcohol, filtering the extract, and
measuring the absorbance of the extract at 254 run (3). Calcium oxide is added to the soil as
a conditioning agent to minimize interferences from organic materials. If the contaminant fuel
is available for calibration, the approximate concentration of the fuel in the soil can be
calculated; if the fuel type is known, but a sample of the contaminant fuel is not available for
calibration, an estimate of the contaminant fuel concentration can be calculated using an
average response factor; and if the nature of the contaminant fuel is not known, the
absorbance value is used to indicate the presence or absence of fuel contamination. Fuels
containing aromatic compounds, such as diesel fuel and gasoline, as well as other aromatic-
containing hydrocarbon materials, such as motor oil, crude oil, and coal oil can be determined.
51
-------
A collaborative study was conducted to determine the reproducibility (between participants)
and repeatability (within participant) precision of the method when applied to two different
soil types spiked with various levels of diesel fuel (1). Data generated in the collaborative
study also provide information on the performance of portable field equipment versus
laboratory equipment for performing the screening method and a comparison of diesel
concentration values determined using the screening method versus a laboratory method. The
purpose of this paper is to further evaluate ASTM Method D-5831-95 using these data.
DESCRIPTION OF THE WORK
Method Development and Collaborative Study Design
Development of ASTM Method D-5831-95 and the collaborative study design have been
previously described (1). As a result, these are briefly summarized below.
The screening method became an ASTM standard test method in September 1995. The
method that was approved by ASTM is the same method used by the eight participants in the
collaborative study. Because the screening method can be performed in the laboratory using
laboratory equipment or in the field using portable equipment, three of the participants used
laboratory equipment; three participants used field equipment; and two used a combination
of both for their testing. The laboratory equipment included various models of a laboratory
stir plate, balance, and spectrophotometer. The field equipment consisted of a soil test kit
supplied by In-Situ, Inc. (patent pending). The soil test kit contains a portable, mechanical
stirrer, portable balance, and portable photometer that measures ultraviolet absorbance at 254
nm.
Each participant tested seven materials in triplicate. The test materials were a sand spiked
with three different concentrations of diesel fuel (test materials A, B, and C), an unspiked
sand (test material D), an organic soil spiked with two different concentrations of diesel fuel
(test materials E and F), and an unspiked organic sofl (test material G). The participants used
the absorbance values they recorded for the test materials to calculate both approximate and
estimated diesel fuel concentrations in the materials. Calculations to correct those values for
concentrations reported in the blank materials (test materials D and G) and statistical
evaluation of the collaborative study data were performed by Western Research Institute.
ASTM Practice D-2777-86, Standard Practice for Determination of Precision and Bias of
Applicable Methods of Committee D-19 on Water (6), and ASTM Practice E-691-87,
Standard Practice for Conducting an Interlaboratory Study to Determine the Precision of a
Test Method (7), were used for guidance in statistically evaluating the collaborative study
data. Calculations outlined in these standards were used to determine the mean concentration,
reproducibility standard deviation, and repeatability standard deviation for the approximate
and estimated concentration data determined using both laboratory equipment and portable
field equipment.
52
-------
The collaborative study materials were tested to make sure they met a specified homogeneity
criterion prior to being sent to the participants (1). Homogeneity testing involved mixing the
bulk materials and analyzing subsamples of them for their diesel concentrations. Analysis was
by gas chromatography with flame ionization detection (GC-FID) of methylene chloride
extracts using modified EPA Method 8015 (4). These data were used to establish a 95%
confidence interval for the concentration of diesel fuel in each test material (5). The bulk
materials were then taken through an additional mixing procedure. After additional mixing,
two subsamples were withdrawn from each of the bulk materials and analyzed. The criterion
for determining homogeneity was if the concentrations of diesel fuel determined in the two
subsamples fen within the 95% confidence interval, expanded on both sides by 10%, then the
bulk material was homogeneous. The 95% confidence interval was expanded by 10% on both
sides to allow for error in the GC-FID method due to extraction, concentration, calibration,
GC sample injection, and diesel pattern interpretation.
RESULTS AND DISCUSSION
Cornparison of Diesel Fuel Concentrations Determined Using ASTM Method P-5831-95
Versus Diesel Concentrations Determined Using Modified EPA Method 8015
A total of 24 approximate and 24 estimated concentration values were generated for each test
material by the eight participants in the collaborative study. In the statistical evaluation of
these data, the mean approximate concentration of diesel fuel in each test material and the
mean estimated concentration of diesel fuel in each test material were calculated (1). These
values can be compared with the concentration values determined in the test materials during
homogeneity testing using the laboratory GC-FID method. This comparison is shown in
Table 1.
Table 1. Comparison of Collaborative Study Data Versus GC-FID Data, mg/Kg
Material GC-FID Method Screening Method Percent Difference
Mean Concentration
A
B
C
E
F
x = 122
95% C.I. = 103-143'
x = 384
95% C.I. = 329-443
x = 841
95% CJ. = 719-972
x=156
95%CJ. = 133-180
x = 826
95% CI.» 704-957
Approximate = 156
Estimated = 179
Approximate = 3S2
Estimated = 459
Approximate = 802
Estimated = 972
Approximate = 103
Estimated = 125
Approximate =618
Estimated = 737
28%"
47%'
0.5%
19%
5%
16%
-34%
-20%
-25%
-11%
*95% confidence interval for the concentration of diesel fuel in the test material expanded by 10% on each side
"Percent difference between screening method mean approximate concentration and GC-FID x value
'Percent difference between screening method mean estimated concentration and GC-FID x value
53
-------
In Table 1, higher absolute percent difference values between the screening method mean
concentrations and the GC-FID mean values (20 to 47%) are generally shown for the lower
diesel concentrations at approximately 100 to 180 mg/Kg. At diesel concentrations in the
spiked sand and organic soil of approximately 380 to 970 mg/Kg, the absolute percent
differences between the screening method mean concentrations and the GC-FID mean values
range from 0.5% to 25%. Comparison of the screening method mean concentration values
to the expanded 95% confidence intervals for the GC-FID analyses shows that seven of the
ten mean concentration values determined using the screening method fall within the
corresponding expanded 95% confidence interval or are just outside the interval by less than
20 mg/Kg. If the absolute values of the percent differences listed in Table 1 are averaged, the
result is 20%. This value can be used to give a general indication of how the results from the
screening method and laboratory method may vary.
The concentrations determined using the screening method to test the diesel-spiked organic
sofl (materials E and F) are lower than the corresponding GC-FID values. This may be due
to the spiked-organic soil adhering to the sides of the glass vials in which the material was
shipped to the collaborative study participants. During addition of this material to the vials
and during testing of the material using the screening method, the spiked organic soil adhered
to the sides of the glass vials, and even with significant shaking, not all of the material could
be loosened from the glass. It is believed that this may have resulted in lower concentrations
of diesel fuel in the spiked organic soil that was removed from the glass vials by the
participants for testing. Despite this problem and considering that the ASTM method is a
screening method and the EPA method is a laboratory procedure, the variation between the
values determined using the two methods would be acceptable in most cases.
Performance of Portable Field Equipment Versus Laboratory Equipment for Performing
ASTM Method D-5831-95
As mentioned, three of the collaborative study participants used portable field equipment to
perform their testing, and three of the participants used laboratory equipment. The mean
concentrations of diesel fuel determined to be present in the test materials using laboratory
equipment and field equipment can be compared. This comparison is shown in Table 2.
The maximum absolute percent difference between the mean concentration values determined
using laboratory equipment and those determined using the soil test kit is 12%, and for six of
the comparisons shown in Table 2, the absolute percent difference is 5% or less. This shows
very good agreement between the results of the method when laboratory equipment is used
and when portable field equipment is used. From the data shown in Table 2, it appears that
at lower diesel concentrations in the spiked sand and organic soil, results from using the
method with laboratory equipment and field equipment may vary slightly more than at higher
diesel concentrations.
54
-------
Table 2. Diesel Concentration Comparison: Portable Field Equipment Mean Concentration Values
Versus Laboratory Equipment Mean Concentration values, ing/Kg
Approximate Concentration Determinations
Material* Soil Test Kit Mean Value Laboratory Mean Value % Difference
A 143 162 -12%
B 373 383 -3%
C 780 812 -4%
E 95 108 -12%
F 579 646 -10%
Rcrimnted Concentration Deteiminatjnnfi
Material* Soil Test Kit Mean Value Laboratory Mean Value % Difference
A
B
C
F
170
475
968
123
722
193
452
962
127
759
-12%
5%
1%
-3%
-5%
* Materials A, B, and C are a diesel-spiked sand, and materials E and F are a diesel-spiked organic soil.
The reproducibility and repeatability standard deviation values were used to express the
precision of the screening method. Information given in ASTM Practice E-177-90, Standard
Practice for Use of the Terms Precision and Bias in ASTM Test Methods (8), was used for
guidance in expressing the precision of the screening method. The index used for expressing
reproducibility and repeatability of the test method is the 95% limit on the difference between
two test results. The 95% limit means that approximately 95% of all pairs of test results from
users similar to the participants in the collaborative study can be expected to differ in absolute
value by less than 2.8 s (standard deviation) or 2.8 CV% (percent coefficient of variation) (8).
The 95% repeatability and reproducibility limits, expressed as 2.8 CV%, for screening the
diesel-spiked sand and diesel-spiked organic soil using laboratory equipment and the soil test
kit are listed in Tables 3 and 4, respectively.
Table 3. 95% Repeatability Limits' for Testing Diesel-Spiked Sand and Organic Soil Using the Screening
MeoWfor Fuels in Soils
Materials: Diesel-Spiked Sand and Diesel-Spiked Organic Soil
Equipment: Laboratory
Test Range. mg/Kg 95% Repeatability Limit (% of the test result^
108-962 (approximate or 15% (9 to 22%)
estimated)
Materials: Diesel-Spiked Sand and Diesel-Spiked Organic Soil
Equipment: Soil Tesf Kit
Test Range, Tng/Kg 95% Repeatability IJmit (% of the test result)
95-968 (approximate or 18% (11 to 26%)
estimated)
'Within participant
55
-------
Table 4. 95% Reproducibility Limits* for Testing Diesel-Spiked Sand and Organic Soil Using the
Screening Method for Fuels in Soils
Material: Diesel-Spiked Sand
Equipment: Laboratory
Test Range. mg/Kg
162-962 (approximate or
estimated)
Material: Diesel-Spiked Sand
Equipment: Soil Test Kit
143 (approximate)
170 (estimated)
373 (approximate)
780 (approximate)
475-968 (estimated)
Material: Diesel-Spiked Organic Soil
Equipment: Laboratory
Test Range. mg/Kg
108 (approximate)
127 (estimated)
646 (approximate)
759 (estimated)
Material: Diesel-Spiked Organic Soil
Equipment: SoU Test Kit
Test Range. mg/Kg
95-579 (approximate)
123-722 (estimated)
95% Reproducibilitv Limit (% of the test result!
15% (10 to 22%)
95% Reproducifrility Limit f % of the test result>
47%
58%
12%
28%
33% (32%, 34%)
95% Reproducibility Limit (% of the test result)
65%
46%
35%
28%
95% Reproducibility Limit (% of the test result)
14% (15%, 13%)
33% (32%, 33%)
'Between participants
The 95% repeatability and reproducibility limits listed in Tables 3 and 4 are specific to the test
materials used in the collaborative study. For other soil types and fuel contaminants, these
data may not apply. However, using these data to evaluate the precision of ASTM Method
D-5831-95 using laboratory equipment versus portable field equipment shows the following.
• There is very good agreement between the repeatability precision of the screening method
for testing the diesel-spiked sand and organic soil using laboratory equipment (15% of the
test result) and the repeatability precision of the screening method for testing the two
materials using the soil test kit (18% of the test result). In terms of repeatability
precision, these data show comparable performance of the method using both types of
equipment.
• There is variation in the reproducibility precision of the method using laboratory
equipment and the soil test kit to screen the diesel-spiked sand and organic soil. For
screening the diesel-spiked sand, the reproducibility precision of the method using
laboratory equipment is better than when the soil test kit is used. However, for screening
the diesel-spiked organic soil, the overall reproducibility precision of the method using the
56
-------
soil test kit is better than when laboratory equipment is used.
• The reproducibility precision of ASTM Method D-5831 -95 using laboratory equipment
to screen the diesel-spiked sand is very good. The 95% reproducibility limit equals 15%
of the test result for both approximate and estimated concentration determinations.
« For approximate diesel concentrations in the organic soil ranging from 95 to 579 mg/Kg,
the 95% reprodudbfliry limit of the screening method using the soil test kit is 14% of the
test result; and for estimated diesel concentrations in the organic soil ranging from 123
to 722 mg/Kg, the 95% reproducibility limit is 33% of the test result.
• The 95% reproducibility precision of the method for screening the diesel-spiked sand
using the soil test kit and for screening the diesel-spiked organic soil using laboratory
equipment varies with concentration for both approximate and estimated concentration
determinations. In both cases, at lower concentrations, approximately 100 to 150 mg/Kg,
the reproducibility precision of the method is poor at approximately 55% and at higher
concentrations, the reproducibility precision of the method is a little better at
approximately 33%.
• For the case where laboratory equipment and the soil test kit give better reproducibility
precision and for the case where laboratory equipment and the soil test kit give lower
reproducibility precision, the 95% reproducibility limits of the method are similar. As a
result, neither type of equipment can be judged more suitable for performing the method
in terms of reproducibility precision.
CONCLUSIONS
The following conclusions can be made concerning the performance of ASTM Method D-
5831-95 for screening the diesel-spiked sand and diesel-spiked organic soil used in the
collaborative study.
• The average absolute percent difference between the approximate and estimated diesel
concentrations determined using ASTM Method D-5831-95 and the diesel concentrations
determined using modified EPA Method 8015 to test the diesel-spiked sand and organic
sofl is 20%. This value can be used to give a general indication of how results from using
the screening method and laboratory method may vary. This variation would be
acceptable in most cases.
• Average diesel concentrations determined using laboratory equipment and the soil test kit
to screen the diesel-spiked sand and diesel-spiked organic soil using ASTM Method D-
5831-95 are very comparable. In all cases, they vary by 12% or less. In terms of test
results, this shows comparable performance of the method using both types of equipment.
57
-------
• There is good agreement between the repeatability precision of ASTM Method D-583 1-
95 for testing the diesel-spiked sand and organic soil using laboratory equipment (15%
of the test result) and the repeatability precision of the screening method for testing the
two materials using the soil test kit (18% of the test result). In terms of repeatability
precision, these data show comparable performance of the method using both types of
equipment.
• There is variation in the reproducibility precision of ASTM Method D-583 1-95 using
laboratory equipment and the soil test kit to screen the diesel-spiked sand and organic
soil. However, for the case where laboratory equipment and the soil test kit give better
reproducibility precision and for the case where laboratory equipment and the soil test kit
give lower reproducibility precision, the 95% reproducibility limits of the method are
similar. As a result, in terms of reproducibility precision of the method, neither type of
equipment can be judged more suitable for performing ASTM Method D-5 83 1 -95 .
ACKNOWLEDGMENTS
Rinding for this study was provided by the U.S. Department of Energy under cooperative agreement DE-FC21-
93MC30127 and by In-Situ. Inc., Laramie, WY.
Mention of specific brand names or models of equipment is for information only and does not imply endorsement
of any particular brand.
REFERENCES
1. Sorini, S.S. and J.F. Schabron, 1996, Development and Validation of a Standard Test Method
for Screening Fuels in Soils. Journal of Testing and F.valuation. JTEVA, In press.
2. American Society for Testing and Materials, 1996, ASTM Method D-583 1-95, Standard Test Method for
Screening Fuels in Soils. Annual Book of ASTM Standards. In press.
3. Schabron, J.F., ND. Niss, BJC. Hart, and S.S. Sorini, 1995, Remote Chemical Sensor De, ycloppy-pt: A New
Field Screening Method for Soil Fuel frtntaminatinn. Laramie, WY, WRI Report WRI-95-R016.
4. U.S. EPA, 1986, Method 8015: Nonhalogenated Volatile Organics. Test Methods for Evaluating Solid Waste:
Physical/Chemical Methods (SW-846). Vol. IB, 3rd Ed.
5. Gunman, L, S.S. Wilks, and J.S. Hunter, 1971, Introductory Knyineqring Statistics. J. Wiley and Sons: New
York, NY.
6. American Society for Testing and Materials, 1991, ASTM Practice D-2777-86, Standard Practice for
Determination of Precision and Bias of Applicable Methods of Committee D-19 on Water. Annual Book of
ASTM Standards 11.01: 31-44.
7. American Society for Testing and Materials, 1990, ASTM Practice E-691-87, Standard Practice for
Conducting an Interlaboratory Study to Determine the Precision of a Test Method. Annual Book of ASTM
Standards, 14-02: 430-449.
8. American Society for Testing and Materials, 1990, ASTM Practice E- 177-90, Standard Practice for Use of
the Terms Precision and Bias in ASTM Test Methods. Annual Book of ASTM Standards. 14.02: 90-101.
58
-------
13
DETECTION OF PESTICIDES AND PCBs IN THE VAPOR PHASE FOR SITE SCREENING
James E. Whetzel. Jr.. Analytical Chemist, W. L. Gore and Associates, 101 Lewisville Road, Elkton,
Maryland 21922
ABSTRACT
Pesticides and polychlorinated biphenyls (PCBs) are well known as hazardous environmental
contaminants. Assessing sites for the presence of these and other organic pollutants typically involves
collecting soil and/or ground water samples and analyzing them by one or more analytical methods. A cost
effective alternate approach to traditional intrusive subsurface site characterization techniques is the
analysis of organic compounds in the vapor phase (soil gas). However, detection of pesticides, PCBs, and
other semi-volatile organic contaminants using traditional soil gas techniques (active) has been difficult
due to low vapor pressure of compounds, tendency of compounds to adsorb onto organic and mineral
components in soil, and the short sampling intervals associated with active soil gas methods.
Detection of pesticides and PCBs utilizing a unique soil gas screening technology known as
GORE-SORBER® Screening Surveys has been examined. This technology has been successfully used
over the past four years by W. L. Gore and Associates, Inc. to screen for a variety of organic contaminants
in the field. The collector is constructed of a patented, hollow insertion/retrieval cord made of
GORE-TEX® membrane. This membrane, which is made of expanded polytetrafluoroethylene (ePTFE),
is chemically inert, microporous, and hydrophobic. The node and fibril structure of ePTFE allows
unimpeded vapor transfer, while preventing liquid water and soil particles from impacting sample integrity.
Inside the insertion/retrieval cord are smaller ePTFE tubes sealed at both ends and filled with sorbent
material. Analysis of the sorbent filled ePTFE tubes is performed by thermal desorption/ gas
chromatography/ mass spectrometry.
The purpose of this study was to determine which pesticides and PCBs from EPA SW846 method 8080
could be successfully detected using the GORE-SORBER® technology. The study included both bench
scale experiments and field work. The pesticides that were detected included aldrin, BHC isomers
(including lindane), DDD, DDE, dieldrin, endosulfan I, endrin, heptachlor, and heptachlor epoxide.
Individual PCB congeners that were detected ranged from monochlorinated biphenyls to pentachlorinated
biphenyls.
INTRODUCTION
Success of soil gas screening depends on the physical and chemical characteristics of the compound being
monitored and the availability of soil pore space. For compounds to be detected, sufficient amounts of the
compound must be available in the soil pore space during the collection time.
Properties such as vapor pressure, solubility in water, and adsorption onto soil organic matter and minerals
directly affect the amount of the compound that will exist in the soil gas. The tendency of a compound to
adsorb onto the soil organic matter can be expressed using the organic distribution coefficient (K^) or
estimated using the octanol water partition coefficient (KoW). K^ is used to relate the ratio of the amount
of compound adsorbed onto the soil organic matter to the amount in the soil solution. Kow is the ratio of
the amount of material that will preferentially dissolve into octanol verses water. Table 1 "'2' shows the
vapor pressure (VP), solubility, and Kow or K.QC for selected organic compounds. Data for Kow '2 and
KQC 3 are noted in their logarithmic form in the references. To better illustrate differences, data from
references have been converted from their logarithmic form to the actual number for use in this table. All
vapor pressures are noted for substances in their liquid states.
59
-------
Table 1.
Constants for a Selection of Common Organic Contaminants
Compound
. StfHte - ^ " : ; H
1 ^ | iTVtcfe1i3tTM*f httttff- ' ' * ^
4 % J ^ 1 y E rxvllUjXcuJpflipItft?- ^ "• "" •
Tillage - 1
T«*rachJ<»oeifeane ;
Undecane
Monochlorinated biphenyls
Trichlorinated biphenyls
Pentachlorinated biphenyls
Hexachlorinated biphenyls
Lindane (gamma-BHC)
Dieldrin
DDT
VP.Pa
~$3$$&, :; * .
\mm " ; -
tv<*vV ,
3^S ' ' •> I
2415
52.2
0.9-2.5
0.003 - 0.22
0.0023 - 0.05 1
0.0007-0.012
0.00839
0.000659
0.0000134
Solubility, g/m*
I'm. . . .
f jjfccrc *f
XTT^rU
515 ;
IS8 • ,
0.04
1.21 -5.50
0.015-0.40
0.004 - 0.020
0.0004 - 0.0007
7.3
0.25
0.0012
Jf / ' Y \
OH' \ ()i'J
,155 .
"• '^ftQ' ^ '
•£\fjf f ^ • •• '
4^NJ
^59^
8710000
20000 - 40000
320000 - 790000
1600000-3200000
5000000 - 20000000
(1290)
(12000)
(245000)
The darker shaded region of the table indicates compounds which can typically be detected using active
soil gas methods. Undecane, which is lightly shaded, can be detected by active soil gas methods, but not as
readily as the others. The unshaded region contains pesticides and PCBs, compounds which have
significantly lower vapor pressures, significantly greater affinity for organic material and are not readily
detectable by active soil gas methods.
To increase the chances of detecting pesticides and PCBs the collection time must be increased over that of
active sampling. Passive sampling, using a sorbent based soil gas collector, allows time integrated
sampling. The collector can be placed in the soil subsurface or saturated zone for an extended period of
time allowing the compounds of low concentration to concentrate on the sorbent material.
To maximize chances for detection, the collector design must not inhibit exposure of adsorbent to the soil
gas and the collector must protect the adsorbent from soil particulates and water. Also, the analytical
technique chosen must be highly selective and sensitive.
COLLECTOR DESIGN
Figure 1 shows the design and placement in the soil subsurface of the passive soil gas collector utilized in
this work. The insertion/retrieval cord and the protective housing of the sorbent packets are made
completely of expanded polytetrafluoroethylene (ePTFE). The membrane is chemically inert,
microporous, and hydrophobic. The node and fibril structure of the membrane allows unimpeded vapor
transfer onto the sorbent material, while preventing liquid water and soil particles from contacting the
sorbent.
The collector design enables the use of differing sorbent materials within the same module, allowing for
selective adsorption of a broad range of organic compounds. For this study a porous polymer sorbent
material was used in the collector.
The design also allows for easy installation. Installation is performed by making a narrow pilot hole with
simple hand tools and inserting the module into the hole using an insertion rod.
60
-------
Figure 1.
GORE-SORBER® Soil Gas Module
I
f
ePTFE
membrane
Vapors
Water(liquid)/ Soil
particles
Sorbenl
BENCH SCALE EXPERIMENTS
Bench scale experiments were designed to determine which pesticides and PCBs could possibly be
detected using the passive soil gas collector (Figure 2). Compounds chosen for examination were from
EPA SW846 method 8080 and were obtained from Chem Service, Inc., West Chester, PA. Collectors were
placed in jars containing neat pesticide and PCB materials. A control sample was also prepared by placing
a collector in an empty jar. The collectors were protected from direct contact with pesticides and PCBs by-
using small open vials to hold the neat materials. The outer jars were capped and placed in a fume hood
for a two week time period (manufacturer's recommended time period for exposure in the field). After the
exposure period, one sorbent packet was removed from each collector and analyzed using thermal
desorption/ gas chromatography/mass spectrometry (TD/GC/MS).
Figure 2.
Setup for Bench Scale Experiment
Jar with cap
Module
Open vial
Neat pesticide/
PCB
61
-------
Pesticide Results
Analysis results for pesticides are found in Table 2. Pesticides are listed in GC retention time order. The
'ug detected' column shows the amount of the pesticide detected on the sorbent packet analyzed. Results
are meant to show relative amounts detected and are not intended to indicate absolute amount of compound
present in the vapor space.
Typical method detection limits for this technology are between 0.02 and 0.2ug. Using 0.2ug as a practical
cut off for determining which compounds are suitable candidates for this technology, assignments can be
made for the probability of detection. The darker shaded portion of the table indicates compounds which
have the greatest probability of being detected in the field (compounds with detected amounts greater than
0.2ug). The lightly shaded compounds are those which are considered to have some possibility of being
detected in the field (compounds with detected levels near 0.2ug). The unshaded region of the table
contains compounds which are probably not detectable using passive soil gas sampling techniques
(compounds not detected or detected well below 0.2ug). Since GC retention times give a rough estimate of
relative boiling points and vapor pressures, a possible vapor pressure cut off for readily detectable
pesticides can be seen in the region of heptachlor epoxide and endosulfan 1 and a possible vapor pressure
cut off for potential detection can be seen in the region of endrin and 4,4'ODD.
Table 2.
Results of Bench Scale Experiment for Pesticides
Pesticide (amount)
Retention Time (min.)
fig Detected
Heptachlor Epoxide (50mg)
&ttdn£ft$&ft l/lflOmfi'l * "
4,4''DDE(100mg)
Dieldrin (SOmg)
Endrin (Ig)
4,4' ODD (Ig)
Endosulfan 11 (lOOmg)
Endrin Aldehyde (lOg)
4,4' DDT (Ig)
Endosulfan Sulfate (SOmg)
Methoxychlor(lg)
10.88
^ J 11 J8 ( '-
1.30
1.38
.56
.63
.62
.76
.91
.94
12.31
0.23
i£3
0.19
0.19
0.24
0.05
0.00
0.00
0.00
0.00
0.00
(B-BHC was investigated but results are suspect and under investigation)
PCB Results
Collectors exposed to PCBs were not analyzed along with calibration standards, therefore no quantitative
data is available. However, chromatograms from the analysis of the exposed collectors show a very strong
response relative to the pesticide standard examined. Figure 3 shows chromatograms from the analysis of a
lOug pesticide standard and collectors exposed to 1 gram of Aroclor 1016 and SOmg Aroclor 1260.
62
-------
Collectors were exposed to Aroclors 1242, 1254, 1221, 1232, 1248, 1260, and 1016. Individual congeners
up to pentachlorobiphenyl were detected in the exposed modules.
In addition to experiments using neat Aroclor mixtures, a similar experiment was setup using
50g of soil spiked with 39.4mg/kg of Aroclor 1242 (approximately 2mg of aroclor 1242). A soil gas
collector was placed in a glass jar containing the spiked soil and an empty jar (control sample). The jars
were then capped and placed in a fume hood for 2 weeks. After the 2 week exposure period, one sorbent
packet was removed from the module and analyzed by TD/GC/MS. The resulting chromatogram can be
seen in Figure 4. Individual congeners up to tetrachlorobiphenyl were detected.
Figure 3.
Example Chromatograms from a lOug Pesticide Standard and
GORE-SORBER® Modules Exposed to I g Aroclor 1016 and 50 mg Aroclor 1260
Abundance 10ug Pesticide Standard
3000000
2800000
2600000
2400000
2200000
2000000
1800000
1600000
1400000
1200000
1000000
800000
600000
400000
200000 :
Tim e-->0 00 8.50
Abundance
1 800000
1600000
1400000
1 200000
1 000000
800000
600000
400000
200000
'
Tim e-->0 00 750
Abundance
280000
260000
240000
220000
200000 |
180000
160000
140000
120000
100000
80000
60000
40000 i
20000 :
Time-->9 .00 7.50
9.00
800
i
A foclo r 101
1C IB
.
8 50
2C
,,r |
9.
1 I1
!i
; ;,
1 ;
[] Jj Jl UA I".
on 10.50 11.00 11 50
s
IB
I I »
I ,']4CIB
I. 3 CIB
LAflJl'l "'. i -ill '\i 'u\hl ''- '
00 9.50 1000 1 0.50
Aroclor 1 260
"'••V -••'•-.
8.00
(
'
1
•--,.
8.50
2CI3
CIB
i
f
III
9
• 2CIB 3CIB I
'
3C IB
5CIB
j'^Uj\ , ...-._•-. JLJlru- - k .- JL
00 9.50 10.00 10.50
Key: I CIB = monochlorobiphenyl; 2CIB = dichtorobiphenyl; 3CIB = trichlorobiphenyl;
4CIB = tetrachlorobiphenyl; 5CIB = pentachlorobiphenyl.
63
-------
Figure 4.
Chromatogram of Module Exposed to Soil Spiked with 39.4mg/kg Aroclor 1242
Abundance S o il S piked w ith A roclor 1242 2CIB
1800000
1600000
1400000
1200000
1000000
600000
600000
400000
200000
T im e-->°
1 C IB
i
2C IB
3C IB
3C IB
:' 3C IB I , r IR
L -AU.'-_'!_«£'P..
750 8.00 8.50 9.00 9.50 10 00
Key: 1CIB = monochlorobiphenyl; 2CIB = dichlorobiphenyl; 3CIB = trichlorobiphenyl;
4CIB = tetrachlorobiphenyl.
FIELD RESULTS
1 0.50
Bench scale experimentation gives a good approximation of what one can expect to be able to detect in the
field. However, only field results can give final proof as to whether pesticides and PCBs can be detected in
soil gas for site screening, due to the impact of soil permeability and other site specific conditions.
A site belonging to a large pesticide manufacturer was screened for volatile organic contaminants and
pesticides in the region of a drum washing area. Sixteen soil gas collectors were installed on a grid at a
depth of approximately 2 to 3 feet. Collectors were retrieved after two weeks and returned to the laboratory
for analysis..
Aromatic and chlorinated hydrocarbons were detected in the vapor phase as well as several chlorinated
pesticides. Pesticides were detected as tentatively identified compounds since calibration standards for
pesticides were not used. Pesticides detected included aldrin, heptachlor, endosulfan 1, and dieldrin (Figure
5 is an example chromatogram). These results demonstrate the capability of detecting volatile organic
contaminants and semi-volatile pesticides in one analytical run using this technology.
Modules returned from another site showed PCB contamination in the presence of other volatile and semi-
volatile organic contaminants. Figure 6 is an example chromatogram showing individual PCB congeners
detected.
Figure 5.
Example Chromatogram from Volatile Organic and Pesticide Contaminated Site
Abundance
m /p-X yl
1 .1 .1 -
TC A
700000
650000
600000
550000
500000
450000
400000
350000
300000
250000
200000
1 50000
1 00000
50000
I
I
J
T im e - - > 0 2.00
3
PC E
E Ibe nz
I
E n do I
o-X yl
3 00 4.00 5.00 6 00 7 00 8 00 9.00
Key: I.IJ-TCA = 1,1.1-trichloroethane; PCE = tetrachloroethene; Etben: = ethylbemene;
m'p-Xyl = m/p-xylene; o-Xyl - o-xylene; Hept = heptachlor; Aid = aldrin; Endo I = Endosulfan I
64
-------
Figure 6. .
Example Chromatogram From PCB Contaminated Site
Abundance 1 C IB
160000
150000
140000
130000
120000
110000
100000
90000
80000
70000
60000
50000 TMB
40000
30000
20000
TOO 00 ,v I; j%(. v !
Time .-» 0 5 50 6.00 6.50
b ip
M e N a p h
M aph
I '
II
!!
|j '
700 7.50 8
n
,
2 C IB
|
1 C IB
,! I ! 2 C IB
: ,udjJ.L? il l'l ' '
.00 8.50 900 950
Key: TMB = trimelhylbenzene; Naph = naphthalene; MeNaph = methylnaphthalenes; Biph = biphenyl;
1CIB = monochlorobiphenyl; 2C1B = dichlorobiphenyl.
CONCLUSION
This investigation has shown that certain pesticides and individual PCB congeners can be detected in the
vapor phase using the GORE-SORBER® soil gas collector and TD/GS/MS analysis. Bench scale
experiments indicate that using this technology alpha-BHC, lindane, delta-BHC, heptachlor, aldrin, and
endosulfan I can be readily detected in the vapor phase and that heptachlor epoxide, 4,4'DDE, dieldrin,
and endrin have the potential of being detected in the vapor phase. Analysis of field samples confirms that
heptachlor, aldrin, endosulfan I, and dieldrin can be detected in the vapor phase in the soil.
Individual PCB congeners detected in the vapor phase using this technology include mono, di, tri, tetra.
and pentachloro biphenyls in bench scale experiments and both monochloro and dichloro biphenyls in field
samples.
In addition to the determination of which PCBs and pesticides that can be detected, it was also found that
this technology can be used to detect both volatile organic compounds and semi-volatile compounds
including pesticides and PCBs in a single analytical run.
GORE-SORBER Screening Survey, GORE-SORBER Module, and GORE-TEX are registered trademarks
of W.L. Gore & Associates, Inc.
References
1. Mackay, D., Shiu, W.Y., and Ma, K.C., Illustrated Handbook of Physical-Chemical Properties and
Environmental Fate for Organic Chemicals. Vol. I. Lewis Publishers, Ann Arbor, 1992, pp.141, 600.
2. Mackay, D., Shiu, W.Y., and Ma, K.C., Illustrated Handbook of Physical-Chemical Properties and
Environmental Fate for Organic Chemicals, Vol. Ill, Lewis Publishers, Ann Arbor, 1992, pp.151, 288,
621-622.
3. Devitt, D., Evans, R., Jury, W., and Starks, T., Soil Gas Sensing for Detection and Mapping of Volatile
Organics, National Ground Water Association, Dublin, pp. 42-43, 113.
65
-------
RADIOLOGICAL
-------
14
MONITORING AND METHODOLOGY OF RADIONUCLIDES
IN DOMESTIC AND IMPORTED FOODS IN THE U.S.
Edmond J. Baratta U.S. Food and Drug Administration, Winchester Engineering and Analytical Center, Winchester,
MA 01890
ABSTRACT
The U.S. Food and Drug Administration (FDA) is responsible for the wholesomeness of the nation's food supply.
The FDA modified its food-monitoring program in January, 1973, to include radioactive isotopes. In 1975, it
expanded this program to include selective foods originating in the vicinity of nuclear power stations. It was
fortunate that this program was set in place, since over the years it has responded to several nuclear incidents. These
include the Three Mile Island incident in Pennsylvania and the Chernobyl accident in the Ukraine of the then Soviet
Union. The program at the time of the latter included food products originating from Europe or adjacent areas. In
addition, it has responded to concerns over former low-level radioactive ocean waste dump sites off the coast of
California and New Jersey. Also, the Massachusetts Bay dump has been surveyed. These were done in cooperation
with other federal agencies such as the EPA and NOAA. Recently, a program was set in place to monitor fish from
the Arctic Ocean when it was learned that the then Soviet Union had disposed of nuclear submarines in the Aral
Sea.
The methodology used to perform analyses on these food products are taken from the standard setting societies such
as the AOAC, International, ASTM and APHA/A WWA/WEF. In addition, methods not tested by these societies are
taken from the literature or from DOE manuals such as the HASL and also the EPA manuals. These include the
methods for long-lived radionuclides such as strontium-90 and cesium-137. The short-lived radionuclides such as
iodine-131, the radio cesium and ruthenium.
This paper will show the data from the monitoring of the U.S. food supply, imported foods and some of the special
surveys. The interpretation of the results will be discussed in light of the Federal Radiation Council Guidelines and
the "FDA Levels of Concern."
67
-------
INTRODUCTION
The U.S. Food and Drug Administration (FDA) is responsible for the wholesomeness of the nation's food
supply. This has been ongoing since 1961, (1-6) when contamination from above ground weapons testing
was prevalent. It was suspended in 1969 to avoid duplication of effort. Also, there was a downward trend
in fallout following the Test Ban Treaty of 1963. In 1973, the FDA resumed its radionuclide food-
monitoring program because of concern of the possibility of food contamination from selective foods
originating in the vicinity of nuclear power plants and other sources. It also decided it should maintain a
radiochemical capability to detect any upward trend in the radioactive contamination of foods.
In 1975, it expanded this program to include selective foods originating in the vicinity of nuclear power
stations. It was fortunate that this program was set in place, since over the years it has responded to several
nuclear incidents. These include the Three Mile Island incident in Pennsylvania and the Chernobyl accident
in the Ukraine of the then Soviet Union. The program at the time of the latter included food products
originating from Europe or adjacent areas, hi addition, it responded to concerns over former low-level
radioactive ocean waste dump sites off the coast of California and New Jersey. Also, the Massachusetts
Bay dump was surveyed. These were done in cooperation with other federal agencies such as the EPA and
NOAA. Recently, a program was set in place to monitor fish from the Arctic Ocean when it was learned
that the then Soviet Union had disposed of nuclear powered submarines in the Aral Sea.
EXPERIMENTAL SAMPLES
From the fiscal years(FY) 1987 (7) through 1995 (October 1986 through September 1995) 3 types of food
collections were examined: Total Diet Study (TDS), reactor survey, and imported. TDS samples
represented the general U.S. food supply and were obtained through FDA's TDS program. Portions of all
foods comprising one TDS collection were analyzed each fiscal year. During FY87 through FY91, each
TDS collection consisted of 234 individual foods selected according to the protocol reported in 1982. The
FY92 through FY95 collection included 265 individual foods selected according to a revised protocol (8).
Approximately 300 reactor-survey food test portions, including raw vegetables, food crops (primarily
fruits), fish, and milk, were collected in the vicinities of 33 nuclear power reactors (representing 71% of the
states having nuclear power reactors). Reactor sites were selected to include many geographical regions
and varied from year to year. Reactor-survey food test portions were analyzed for 3H. Half the foods were
also analyzed for '"Cs, ml, "*Ru, and *Sr.
Imported foods have not been monitored continually since the program began in 1961. Only during the
period from 1972-1982 were they previously monitored. However, extensive collection and analysis efforts
were initiated in May 1986 to monitor for contamination resulting from the Chernobyl nuclear accident
Samples of food were collected at points of entry during routine import inspections.
Selection of shipments to sample was based on country of origin and food type. All samples were analyzed
for |]1Cs and l34Cs, which were used as the main indicators of contamination. This monitoring effort is
continuing, but on a limited basis, depending upon the product being imported. This is based on the past
history of the products.
Other surveys mat were undertaken recently were a repeat of the low-level waste dump sites fish survey in
that included the Farallon Island area near San Francisco, the coast of New Jersey and Massachusetts Bay.
Then again in 1990, the Farallon Island area was sampled. Also, in 1992, the Massachusetts Bay was
sampled. The EPA and NOAA took part in this survey.
68
-------
When, in 1993, it was learned that the then Soviet Union had discarded nuclear powered submarines in the
Arctic Ocean, another fishery survey was undertaken. This lasted for two years.
Analytical/Methodology
The methodology used to perform analyses on these food products are taken from the standard setting
societies such as the AOAC, International, ASTM and APHA/AWWA/WEF (9-11). In addition, methods
not tested by these societies are taken from the literature or from DOE manuals such as the HASL and also
the EPA and PHS manuals (12-16). These include the methods for long-lived radionuclides such as
strontium-90 and cesium-137. The short-lived radionuclides such as iodine-131, the radio cesium and
ruthenium.
Radionuclide analyses were performed at FDA's Winchester Engineering and Analytical Center in
Winchester, MA. The basic analytical methods used were described previously (17-20). The activity
concentrations of l37Cs, ml, and l06Ru were determined by using gamma-ray spectrometry with germanium,
lithium-drifted germanium, and sodium iodide gamma-ray detectors. The method for the isotopic analysis
of plutonium in fish was based on a method for soil and was modified for food ash. The isotopic plutonium
content is determined using an alpha spectroscopy system, after the sample is plated.
Limits of detection (LODs) were defined as three times the one sigma analytical uncertainties at low
concentrations near the LODs. Uncertainties near the LODs were primarily due to counting statistics. Large
variations in food density and, in some cases, the quantities available for analysis, led to wide ranges of
LODs. For most foods, LODs were ca 2 Bq/kg for "'I, U7Cs, and 106Ru; 0.1 Bq/kg for "Sr; and 10 bq/kg
for3H.
More recently the equipment (1992) has been upgraded to include three intrinsic germanium detectors with
efficiencies of 40%, 80% and 90% (rated against Nal(Tl)). A Nal(Tf) detector is still functioning but is not
being used. The older 6700 ND system was also replaced with the 9900 ND (Canberra) system. The system
handles, in addition to the gamma detectors, two alpha spectrometers. A newer gamma system is being
explored.
The FDA/WEAC takes part in the EPA Intercomparison Quality Assurance Program at Las Vegas, NV. It
also receives split samples from a DOE contract laboratory. In addition, as part of another quality assurance
program, it receives "blind" sources from the NIST.
RESULTS AND DISCUSSION
The Federal Radiation Council (FRC) (21) recommendations are used by the FDA in the interpretation of
the analytical results for domestic foods. The FRC gave Radiation Protection Guides (RPGsXTable 1) and
corresponding action ranges for intake rates. Application to FDA's ongoing monitoring was made under the
conservative assumption that the daily average consumption of any individual contaminated food
(including beverages) would be 1 kg/day. This assumption led to derived-concentration action ranges, also
given in Table I.
No monitoring or actions (other than normal quality assurance exercises) are needed when average activity
concentrations are within Range I. Monitoring only (i.e., no control actions) is recommended when average
activity concentrations are within Range II or only transient in Range HI. Control actions are recommended
when activity concentration averages are in Range II but rising toward Range III or are in Range III.
Control actions are to be strong and prompt when activity concentrations exhibit a rising trend in Range III.
An illustration that detecting capabilities were more than adequate for monitoring the food supply is
presented in Figure 1. Relative LODs are given based on the transitions to Range HI. where control actions
69
-------
are recommended, set at 100%. Presented in this way, Range II, which corresponds with monitoring only,
extends from 10 to 100%. Range I is below 10%. The radioisotopes presented in Figure I are the main
long-range contamination-indicator isotopes that have half-lives of several years and are indicators for the
most likely nuclear accident scenarios. The relative LODs are very low. Therefore, contamination from a
nuclear event would be very clearly monitored, even if radioactivity concentrations were too low to pose a
health concern.
For imported foods, FDA uses the levels of concern (LOCs) recommended by the FDA/USDA Food Safety
and Inspection Service Task Force (23), which was set up in May 1986 by the Director of FDA's CFSAN.
The radioactivity LOCs in Table 2 were derived from the Preventative Action Guides (24) and use m I and
the combination of 1MCs and I37Cs as indicators of contamination. Shipments of food containing
radioactivity concentrations exceeding the LOCs are not allowed into the United States.
Table 2. Radioactivity levels of concern recommended by the FDA/FSIS Task Force*
Infant Foods, Bq/kg Other foods, Bq/kg
Radionuclide (pCi/kg) (pCi/kg)
I31I 56(1500) 300(8000)
IMCs+'"Cs 370(10000) 370(10000)
'Reported inref. 15
TOTAL DIET STUDY
The TDS findings in this paper focus on data from 9 fiscal years, FY87-95 when approximately 2100 food
test portions were analyzed for'"Cs, "'I, "*Ru, and ""Sr. I3II and "*Ru were below the LODs in all test
portions. l37Cs activity concentrations were below the LODs in all but ten test portions; the IJ7Cs activity
concentrations in eight of these ranged up to 4.4 Bq/kg, which is very low in Range I and not unusual (the
Range I-Range II transition is 54 Bq/kg).
"°Sr findings were consistent with those for previous years. Radioactivity concentrations were below the
LODs in most foods (70%), measured and in RPG Range I for 29% of the foods, and low in Range II for
1% of the foods. There was no noticeable increase in "°Sr contamination after the Chernobyl accident
Dietary intake rates for **Sr were calculated and are plotted in Figure 2 for all TDS data obtained since the
Radionuclides in Foods program began in 1961. Intake rates for 1961-1982 were derived from the TDS
protocol for mat period and relate to the 16- to 19-year old male age/sex group. For consistency, intake
rates for 1982-1995 were derived from the 14- to 16- year old male age/sex group, which, under the
protocol, is the closet match to the earlier protocol. In general, *°Sr activity concentrations in the U.S. food
supply continued to decline from the high that occurred in FY64 (1.1 Bq/day, very low in Range II), during
a period of atmosphere testing of nuclear weapons, to sO.05 Bq/day after FY87.
IMPORTED FOODS
Since April 1986, imported foods have been monitored for radionuclide contamination resulting from the
Chernobyl accident Extensive sample collection and analysis efforts were begun immediately after the
accident and are still in effect. Samples were collected at points of entry during routine import inspections
for more than 2600 shipments originating from a total of 48 countries. Food types most likely to have
70
-------
contamination were given preference during collection, and the practice of routing foods through countries
other than those of origin was taken into consideration.
Entry of shipments into the United States was allowed or disallowed on the basis of the LOCs for I3'I and
Cs (sum of IMCs and 137Cs) activity concentrations as indicators of contamination. Although other
radionuclides also may have been present in contaminated food, the I and Cs isotopes were predicted to
pose the greatest health risk resulting from an accident of this type. Because its half life is only 8 days, "'I
was useful as a contamination indicator for only a few weeks after the accident. No appreciable 13II activity
was found in samples collected after June 1986. Approximately one year after the accident, FDA
reaffirmed that the Cs isotopes continued to be the most effective indicators of radionuclide contamination
from this accident.
Contamination was found in 705 test portions, most of which were collected within the first two years after
the Chernobyl accident. The ratio of'" Cs to IMCs activity concentrations (decay corrected to die mid-1986
commencement of this monitoring initiative) was approximately 1.5 and was consistent with that predicted
for Chernobyl-related contamination. During FY86 and FY87, contamination was found in approximately
40% of the samples collected and indicated that FDA inspectors were successfully targeting contaminated
shipments. Since the accident, a total of 23 shipments were denied entry into the United States because
their contamination concentrations were above FDA's LOCs. The foods involved were mainly cheeses,
mushrooms, pasta, reindeer/elk meat, and spices. Of all the samples collected, only two, cheeses collected
shortly after the accident, had ml contamination concentrations above the LOC.
REACTOR SURVEY
FDA's reactor-survey findings for FY-87-95 were similar to those from earlier years (1-6). Out of a total
of approximately 200 test portions analyzed, I3II and l06Ru activity concentrations were below LODs in all
test portions. "7Cs activity concentrations were below LODs in all but 7 test portions: 5 fish, a honey and
one maple syrup that originated from the vicinities of 4 reactor sites. The '"Cs activity concentrations
ranged up to 26 Bq/kg (low in Range I). No definitive control measures are recommended below 540
Bq/kg.
Tritium activity concentrations were below the LODs for all but about 4% of the reactor-survey test
portions. For these portions, activity concentrations ranged up to 70 Bq/kg (very low in Range I), primarily
in fish and vegetables from the vicinities of 4 reactor sites (Note: These are not the same 4 reactor sites
indicated previously for l37Cs). No specific releases of radioactive materials were identified, but the
detected tritium probably resulted from routine low-level releases. Ninety-four percent of the reactor-
survey test portions had "Sr activity concentrations in Range I, most (64%) had activity concentrations low
in Range II. "Sr was detected in all types of foods, in foods from the vicinities of virtually all reactor sites,
and at activity concentrations up to 5.8 Bq/kg (mid-Range II).
The highest activity concentration was in a maple syrup test portion. Because maple syrup results from a
preparation procedure that reduces the matrix volume by a factor of about 40, common contaminants such
as those from the 1960s atmospheric testing would be predicted to be concentrated and, therefore, at higher
concentrations than found in other foods. No additional surveillance was suggested.
SUMMARY
Analysis of TDS foods showed that the radionuclide content of the domestic food supply during the last
nine years was very low and that no control measures were indicated. ml and l06Ru activity concentrations
were below the LODs, and "7Cs activity concentrations were below LOD or at detectable concentrations in
Range I in all TDS foods. The highest activity concentration was in an apple juice test portion. The ratio of
71
-------
"7Cs to IMCs supported the assumption that the contamination came from the Chernobyl accident. A
gradual decrease in die ""Sr intake rate has continued since the mid-1960s and has dropped to <0.05
Bq/day. "Sr activity concentrations were below the LODs for most TDS foods. No other long-range trend
was noted for domestic foods.
The radionuclide content of the reactor-survey foods was low and no control measures were indicated. m I
and l06Ru activity concentrations were below LODs in all test portions, and '"Cs activity concentrations
were below the LODs in all but 6 test portions, which contained up to 26 Bq/kg. This activity concentration
range is very low and, according to FRC recommendations, no control actions would be suggested unless
me average activity concentrations approach 540 Bq/kg. "Sr activity concentrations were comparable to
those found in the TDS foods. The activity concentrations were in Range I or Range II of the surveillance
and control recommendations of the FRC. Tritium was low in Range I in all reactor-survey foods.
Out of approximately 7 million imported food shipments received by the United States since the Chernobyl
accident, approximately 2600 shipments were targeted to be most likely to have Chernobyl contamination
and were sampled for analysis. Findings indicated some of the samples had small amounts of
contamination, but for the vast majority of foods, contamination concentrations were below FDA's LOC.
Only 23 shipments have been detained since the accident occurred. The average contamination
concentration and its incidence have declined and this trend is predicted to continue. However, because
foods contaminated at levels above the LOCs have been found as recently as FY91, the chances of
importing foods with contamination have not yet become negligible. Systematic surveillance for
radionuclide contaminants in imported foods will continue, along with the domestic food monitoring of
TDS and reactor-survey foods.
REFERENCES:
1.) Simpson, R.E., Baratta, E.J., & Jelinek, C.F. (1977) J. Assoc. Off. Anal. Chem. 60,1364-1368
2.) Simpson,R.E., Shuman, F.G.D., Baratta, E.J., & Tanner, J.T. (1981) Health Phys. 40,529-534
3.) Stroube, W.B., Jr., Jelinek, C.F., & Baratta, EJ. (1985) Health Phys. 49,731-735
4.) Lombardo, P. (1986) in Environmental Epidemiology, F.C. Kopfler & G.F. Craun (Eds), Lewis
Publishers, Inc., Chelsea, MI, pp. 141-148
5.) Cunningham, W.C., Stroube, W.B., Jr. & Baratta, E.J. (1989) J. Assoc. Off. Anal. Chem. 72,15-
18
6.) Cunningham, W.C., Anderson, D.L., & Baratta, E.J. (1994) J. Assoc. Off. Anal. Chem. 77,1422-
27
7.) Pennington, J.A. (1983) J. Am. Diet Assoc. 82,166-173
8.) Pennington, JA. (1992) J. Nutr.Educ. 24,173-178
9.) STANDARD METHODS FOR THE EXAMINATION OF WATER AND WASTEWATER,
18th Ed. (Washington, D.C. American Public Health Association, (1995)
10.) AMERICAN SOCIETY FOR TESTING MATERIALS, 1989 ANNUAL BOOK
OF ASTM STANDARDS, WATER AND ATMOSPHERIC ANALYSIS, Vol. 11.02,
Philadelphia, PA (1989) [ASTM, 1989]
11.) OFFICIAL METHOD OF ANALYSIS, 16th Ed, Vol. 1, Association of Official
Analytical Chemists, Washington, D.C. (1995)
12.) RADIOASSAY PROCEDURES FOR ENVIRONMENTAL SAMPLES, U.S. Public Health
Service Publication No. 999-RH-27 (Superintendent of Documents, Washington, D.C.) (1967)
13.) ENVIRONMENTAL MEASUREMENTS LABORATORY: Procedures Manual (U.S.
Department of Energy - New York, NY) HASL-300 (Revised 1992)
14.) DOE METHODS FOR EVALUATING ENVIRONMENTAL AND WASTE
MANAGEMENT SAMPLES (NITS, U.S. Department of Commerce, Springfield, VA) DOE/EM-
0089T(1993)
72
-------
15.) RADIOCHEMICAL ANALYTICAL PROCEDURES FOR ANALYSIS OF
ENVIRONMENTAL SAMPLES, (U.S. Environmental Protection Agency-Las Vegas, NV)
EMSL-LV-0539-17 (1979)
16.) RADIOCHEMICAL PROCEDURES MANUAL, (U.S. Environmental Protection Agency,
Montgomery, AL) EPA 520/5-84-006 (1984)
17.) Baratta, E.J., & Reavy, T.C. (1969) J. Agric. Food Chem. 17, 1337-1339
18.) Lieberman, R., & Moghissi, A. (1970) Int. J. Appl. Radiat. hot. 21,319-327
19.) Measurements of Radionuclides in Foods and the Environment (1991) IAEA Technical Report
Series Pub. No. ICD295, available from UNIPUB 4611-F, Assembly, Dr. Lanham, MD 20706-
4391
20.) Shuman, F.G.D., Easterly, D.G., & Baratta, E.J. (1982) J. Assoc. Off. Anal. Chem. 65, 1039-1043
21.) Federal Radiation Council (1961) Background Material for the Development of
Radiation Protection Standards Report 2, U.S. Government Printing Office,Washington, D.C.
22.) National Council on Radiation Protection (1963) NCRP Report 22 (National Bureau of
Standards Handbook 69), U.S. Department of Commerce, Washington, D.C.
23.) FDA/FSIS Task Force Report (1986) U.S. Food and Drug Administration,Washington, D.C.
24.) Federal Register (1982) 47,43073-47084
25.) Exposure of the Population in the United States and Canada from Natural Background Radiation
(9187) NCRP Report No. 94, available from National Council on Radiation Protection and
Measurements, Bethesda, MD.
73
-------
Table 1. Radiation Protection Guides (RPGs) and derived-concentration action ranges for selected radionuclides, as recommended for the
general population average
Derived concentration action ranges, "Bq/kgb
Radionuclide
mj
u'Cs'
"Sr
3H<
Target
Thyroid
Whole Body
Bone
Whole Body
RPG dose, mSv/year
5
1.7
5
1.7
I
0-0.37
0-54
0-0.74
0-7400
II
0.37-3.7
54-540
0.74-7.4
7400-74000
III
3.7-37
540-5400
7.4-74
74000-740000
1 Range I requires no specific action; Range II requires surveillance and routine control of upward trends toward Range III;
Range HI requires surveillance and controls to reduce exposure to Range II (1); the Range II-Range HI transition corresponds with
the RPG dose; 1 Bq- ca 27 pCi.
b Derived concentrations were calculated on the basis of an average contaminated food intake of 1 kg/day (includes water and other
beverages).
c IJ7Cs and 3H were not considered by the FRC. The ranges were derived by using the radionuclide concentrations in water tabulated by
the National Committee on Radiation Protection (22) for occupational exposure, x 1/30 to apply to the average of the general
population.
-------
LOD
Relative
120
100
80
60
40
20
0
(1.1%)
Range III
Range II
(0.37%)
Range I (below 10%)
-4
(0.01%)
Sr-90 Cs-137
Radioisotopes
H-3
Figure 1. Surveillance LCDs for long-term contamina-
tion indicators, expressed relative to FRC Action Range
IWII cutoff.
-------
O)
(Bq/d)
1.6
1.4
1.2
I
0.8
0.6
0.4
0.2
0
Figure 2
J ' '
\
Range III
(7.7 Bq/d)
" f
Range 11
Range I
1960 1965 1970 1975 i960 1985 1990 1995
Year
U.S. daily dietary intake of 90Sr, 1961-1995
-------
15
ANALYSIS OF ACTINIDE ELEMENTS IN SOILS AND SEDIMENTS
William C. Burnett, D. Reide Corbett, and Michael Schultz, Environmental Radioactivity
Measurement Facility, Department of Oceanography, Florida State University,
Tallahassee, FL 32306-3048; Tel: 904-644-6703, email: bumett@ocean.fsu.edu
Michael Fern, Eichrom Industries Inc., 8205 South Cass Avenue, Darien, IL 60561; Tel:
800-422-6693, email: mfernl409@aol.com
Abstract
We have investigated soil dissolution techniques and group separation of actinide
elements using a new type of resin in an attempt to make the process simpler and more
predictable. We have found that a NaOH fusion gives excellent dissolution results and,
with a programmable furnace, can be performed with minimal analyst time. Our approach
to avoid troublesome "matrix effects" is to preconcentrate actinides into a common form
that would then behave uniformly and predictably during a subsequent separation
scheme. A new extraction chromatographic resin based on diphosphonate chemistry
called Actinide Resin, exhibits extremely high affinity for actinide elements even in the
presence of high concentrations of other, potentially interfering ions. We have
documented near-quantitative recoveries of U, Pu, and Am through the preconcentration
portion of the procedure by use of natural matrix standards and multiple isotopic tracing
techniques. Final recoveries of uranium, americium, and plutonium have been good
although thorium yields have been consistently lower for an, as yet, unexplained reason.
The quality of the separations themselves have been excellent as judged by peak
resolution of the alpha spectra and the complete absence of interfering energies.
Actinide Resin
Recently, a new ion exchange resin based on diphosphonate chemistry was developed at
Argonne National Laboratory by E.P. Horwitz and his co-workers (Horwitz et al., 1993;
Horwitz et al, 1994). This resin, commercialized as Diphonix® Ion Exchange Resin, and a
derivative extraction chromatographic resin called "Actinide Resin" by Eichrom Industries
(8205 S. Cass Avenue, Suite 107, Darien, IL), exhibit extremely high affinity for actinide
elements even hi the presence of macro concentrations of many commonly occurring soil
constituents. While Diphonix has been applied to a variety of industrial processes,
Actinide Resin has characteristics that make it very attractive for analytical work. Both
resins are relatively insensitive to complexing agents such as oxalates, phosphates, and
sulfates. Horwitz has shown that for a 1M HC1 system, Actinide Resin has KD'S on the
order of 106 - 107 for tri-, tetra-, and hexa-valent actinides with little sensitivity to the
presence of HF up to approximately 1M. Tri-valent rare-earth elements (REE) also bind
77
-------
strongly as does Ti(IV) which could cause a capacity problem should much Ti be present
in the sample. Fortunately, the addition of 1M HF to a 1-2M HC1 sample load reduces
the KD for Ti about 4 orders of magnitude while actinides are still strongly bound.
We present below a discussion of some our recent results using Actinide Resin for the
analysis of U, Th, Am, and Pu in soil samples up to 5.0 grams in size. Because much of
the difficulty in obtaining good results for soil samples depends upon the dissolution
steps, we first present some background information concerning this issue and discuss
how our laboratory handles this problem.
Dissolution, Speciation, and Analytical Problems
The analysis of soils and sediments for environmental levels of uranium, plutonium, and
other actinide elements is often hampered by sample-dependent problems involving
composition and/or mineralogy. For example, the presence of a refractory actinide-rich
yet minor soil component would result in a low bias if this phase is not fully digested
together with the remainder of the soil constituents. Clearly, if the actinide-bearing phase
is not dissolved, the correct result cannot be obtained. Other problems encountered in
soil analysis are often more inexplicable yet very common and tend to be lumped together
into so-called "matrix effects." This ill-defined yet persistent type of problem can result
in low recoveries or incomplete separations on some samples, while others go through the
analysis sequence virtually trouble-free. There is thus some justification for the
interpretation that the dilemma has something to do with the specific sample
composition.
Many laboratories have adapted a practice of using a strong acid leach for extraction of
anthropogenic radionuclides such as transuranic species (Pu, Am, etc.) or fission products
(l37Cs, ^Sr, etc.). Generally, a total dissolution technique is applied if analysis of
naturally-occurring uranium or thorium is required. While acid leaching is thought to be
suitable for recovery of fallout-derived nuclides, it may not be suitable in other cases. For
example, Oughton et al. (1993) showed that the common laboratory practice of applying
a 6M HC1 treatment could only recover about 25% of the ^Sr from some soils, even after
about 24 hours of leaching. These soils, collected from near Sellafield and Chernobyl,
were contaminated with small particles of irradiated uranium oxide from the high-
temperature accidents at the two locations. Similar observations have been made for
actinide elements over at least the last two decades. Sill et al. (1979), for example,
discussed problems dealing with "highly-fired" refractory U/Pu oxides. Some, perhaps
many, soils have actinides bound in more than one form. The NIST Rocky Flats standard
soil, for example, contains about 90% Pu which is acid leachable and approximately 10%
that requires HF treatment to remove (K. Inn, pers. comm.). Soils impacted by the
Chernobyl accident show variable amounts of extractable fission products, apparently
78
-------
reflecting the variable weathering of fuel particles (Salbu et al.,. 1994). In a recent review
of sample dissolution techniques, Sill and Sill (1995) state that "...any procedure that fails
to obtain complete sample dissolution for whatever reasons of economy, speed, sample
load, or other expediency is untrustworthy at best...". We agree with this assessment and
conclude that unless one has reliable prior information concerning the speciation of the
analytes in a sample (a rare occurrence), its hard to argue with the approach that
dissolving everything before the analysis should be the goal.
As an example of the type of difficulties that may arise, we present frequency
distribution plots (Fig. 1) for ^'U in four soil performance standards issued by DOE's
Environmental Monitoring Laboratory (EML) to a few dozen contractor laboratories
throughout the country. EML's Quality Assurance Program specifies that all samples
will be run blind by whatever methods individual laboratories choose. EML employs a
total dissolution via a pyrosulfate fusion in a platinum crucible for U in soil. The data
presented in these histograms were taken from the reports issued by EML (eg. Sanderson
and Greenlaw, 1996) and are shown as a ratio of the individual lab result to the EML
result. Thus, a ratio of 1.0 would be the expected result, assuming that the sample is
homogeneous with respect to U and that both the EML and contractor lab have obtained
the "correct" result. Note that while the concentration of 238U in the samples varies in a
relatively narrow range (25.5 - 33.0 Bq-kg"1), the overall quality appears to be much lower
hi some samples (as 9309) compared to others. While there is some suggestion that the
results may be improving, the best mean result (9509) of those presented is still 12%
lower than the EML result. There is clearly a low bias in the majority of these U results.
We have no way of knowing for certain what caused these differences. For the most part,
the same laboratories were participating, most likely with the same methods. We feel that
the most likely explanation for the low bias is incomplete dissolution of refractory
uranium-bearing phases hi these soils. The differences between samples, therefore, must
be related to the speciation of U in these soils as well as any improvement in the
procedures of the participating laboratories. Interestingly, when the results for Am and
Pu in the same soils are examined, there is no indication of bias and there is a definite
pattern of improved results at higher concentrations, not surprising as blank corrections
become increasingly more important at low concentrations. This is consistent with the
interpretation that these contaminants have a soil speciation which, at least in these
samples, allowed easier dissolution of Am and Pu then U.
An Approach to the Problem
Soil Dissolution and Preconditioning
We present below a summary of our soil dissolution - actinide preconcentration method
as it is currently being applied. The method is still under development and a more
79
-------
12
8-
4-
9309
25.5±1.0 Bq-kg'1
0.65±0.31
n=39
in
12.
8-
4-
9409
33.0±2.4 Bq-kg'1
0.83±0.13
n=38
cr
CD
12
8-
4-
2340
12
9503
0.85±0.17
n=51
8-
4-
9509
29.5±2.3 Bq-k
0.88±0.16
n=44
Lab Result/EML
Figure 1. Histograms of reported ^'U activities in EML QAP soils. Each histogram also
contains the EML reported activity, mean ratio and standard deviation, and
number of participating laboratories.
detailed protocol will be published elsewhere and may be requested from the authors. We
have tried a number of different soil dissolution techniques in our laboratory. Although
the potassium fluoride - pyrosulfate fusion technique is known to be an excellent method
for total dissolution of difficult materials like soils, and the fusion can be done fairly
quickly (less than one hour for a 10-g sample reported by Sill and Sill, 1995), it requires
platinum crucibles and is performed over an open flame which could be hazardous if
employed by less experienced personnel. We prefer an approach that is simpler and more
amenable to batch work, i.e., can be automated to some degree. A series of tests involving
80
-------
a few different dissolution techniques on EML soil-9309, a very difficult soil judging by
the results discussed above (see Fig. 1), shows that the NaOH fusion gave excellent
results for U (Fig. 2). We use a NaOH-to-sample ratio of 5:1 in an alumina crucible in a
programmable muffle furnace. The program is set to ramp the temperature slowly to
160°C and hold for 4 hours (this prevents creep of the flux up the crucible), and then
ramp to 550°C where the actual fusion occurs. We have been holding the furnace at
550°C overnight but shorter periods should work equally as well.
Strong acid leach - ::•:•:::::•::::::
Microwave bomb 1 -
Microwave bomb 2 -
SS acid digestion bomb 1
SS acid digestion bomb 2 -
SS acid digestion bomb 3 - wij&j:.
NaOH fusion, alumina crucible - :*:£*$*§
25 50 75
% EML U Value
100
Figure 2. Results for total U, relative to the EML value, for soil 9309 based on several
different dissolution techniques. The different runs on the microwave and
stainless steel digestion bombs reflect various treatments of the insoluble
residues.
After cooling, the resulting cake is dissolved in deionized water, which solubilizes many
of the constituent ions while insoluble hydoxides of Fe and other metals remain and
should effectively scavenge actinide elements. We add a small amount (~200 mg) of SFS
(sodium formaldehyde sulfoxylate; commonly known as "Rongalite") as a reducing agent
to ensure no soluble hexavalent species are present which would be lost during the water
dissolution and ensuing rinses. After two rinses with a strongly alkaline solution (0.5 -
5M NaOH) to remove some silica and aluminum, the scavenge is rinsed once with
deionized water and the insoluble fraction is dissolved in HC1 to a final concentration of
about 3M. At this point, one may either remove the remaining Si or add HF to the HC1
solution to hold the Si in solution while it goes over the column. In the work reported
here, we have chosen to remove Si beforehand because of concern that addition of HF
would result in losses of actinide elements by co-precipitation with REE fluorides. We
thus treated all samples by either fuming the sample in the presence of HF to volatilize Si
81
-------
as silicon tetrafluoride or by addition of polyethylene glycol (PEG-2000) directly to the
3M HC1 sample solution to cause the Si to floe. In the case of the PEG floe, the Si is then
centrifuged off, rinsed with 2M HC1 - 0.1M PEG, and either discarded or added to the
Actinide Resin column later in 2M HC1 - 1M HF. We have found that the separated Si
floe may contain significant amounts of Th and Am which apparently sorb to the Si
surfaces (Corbett et al., 1995). Repeated rinsing with 2M HC1 - 0.1M PEG does help,
but we consider it safer to either fume off the Si or run the Si fraction through the column
as well.
Group Separation via Actinide Resin
After the preconditioning of the hydroxide scavenge and Si removal, we prepare the
solution for loading onto the Actinide Resin column by addition of ascorbic acid (make
solution about 0.05M AA) to reduce Fe(III) which can bind strongly to the resin. After
preconditioning the resin with a few column volumes (~5 mL) of 2M HC1, we load the
sample (hi about 100 mL 2M HCI) onto the resin bed (0.7g Actinide Resin loaded into 6 x
0.9 cm plastic columns manufactured by Isolab, Inc., Akron, OH). We next rinse the
resin with 5 mL 2M HCI followed by another rinse with 2M HCI - 0.5M HF which
removes Ti and some Si which often seems to persist through some of the Si removal
steps and appears to be partially retained on the column. In some cases we have
observed a significant drop off in the flow rate of the column which we believe is due to
precipitation of Si gel within the resin bed — the HC1/HF rinse usually clears this up. A
rinse with 5 mL of 2M HCI follows to remove the HF from the previous step. The
sample load and all rinses through this part of the procedure are collected in a waste
beaker and discarded. The actinides may now be eluted off the column as a group. We
have been doing this by adding 15 mL of isopropanol which solubilizes the Dipex™
extractant which is collected in a clean alumina crucible. Finally, 5 mL of 1M HNO3 are
added and also collected directly in the crucible. This procedure was found to increase
recoveries — presumably because the Amberlite bead supports for the Dipex™ have
some ion exchange capacity which sorb charged actinide species (A. Rollins, pers.
comm.).
The Dipex™ extractant must now be oxidized in order that the actinide elements will be
in a form suitable for chemical separation. Fortunately, the oxidation product of the
extractant is phosphoric acid, which upon addition of Ca2+, is a useful matrix for
precipitation of basic calcium phosphate (CaHPO4), an excellent scavenger of actinide
elements. Although other approaches are possible (eg. wet oxidation, direct high
temperature combustion), we have found it convenient to simply add some Ca carrier as
Ca(NO3)2 directly into the crucible, a small amount of NaOH, and then go through
another fusion cycle. When that is complete, the CaHPO4 may be readily dissolved in
HNO3 or other acid and then proceed with a chemical separation by whatever approach is
82
-------
convenient. At this point, the complex soil matrix has been converted to a simple form,
CaHPO4, which should make the ensuing separations simpler and much more predictable.
While many schemes are possible, we used a tandem set-up of 2 columns packed with
Eichrom's UTEVA (U and Th separation) and TRU (Am and Pu) resins for the work
reported here. An overview of the entire approach reviewed here is shown in the
accompanying flow chart (Fig. 3).
.
Dissolution
DDW, SFS
Conditioning of Fe Scavenge
i
NaOH Fusion
•5:1 ratio
•550°C
•wash 0.5-5M NaOH, 2x
•DDW rinse
•dissolve HC1, ascorbic acid
•fume or floe Si
Oxidation of Extractant
I
•evap isopropanol
•add Ca(NO3)2
•NaOH fusion, oxidize Dipex™
•ppt CaHPO4
•dissolve in load solution
Figure 3. General flow chart of the method for dissolution, group separation and final
separation of actinide elements from soils.
Some Results
We described in an earlier report (Corbett et al., 1995) how we were able to verify that
actinide recovery through the Actinide Resin part of the procedure was almost
quantitative using a standard sediment and multiple tracing techniques. At that time,
83
-------
difficulties in the conversion process of the Dipex™ extractant to a form suitable for
column separation resulted in lower recoveries then desired. Those series of experiments
resulted in the change from a simple high-temperature combustion of Dipex™ to the
NaOH fusion now employed. In order to evaluate the entire system as now designed, we
analyzed a batch of 5 natural matrix soil and sediment standards, each run at 2.5 and 5.0
grams (Table 1). These standards were provided by the International Atomic Energy
Agency (IAEA) and EML and all have recommended values for U, Am, and Pu while the
2 IAEA standards also have values for Th.
Table 1. Results for U, Pu, and Th isotopes and 241 Am from 5 different EML soil and
IAEA sediment standards run at two different sample weights.
IAEA 135
2.5g 5.0g IAEA
23«UBq-kg-'
U Yield (%)
24lAm Bq-kg'1
Am Yield (%)
239Pu Bq-kg1
238Pu Bq-kg'1
Pu Yield (%)
232Th Bq-kg'1
230Th Bq-kg-1
Th Yield (%)
29.1 31.3 30.0
28.2 29.7 28.3
81.7 73.1 -
353 377 318
98.6 50.8 -
204 202 213
40.3 39.8 43.0
75.3 50.0 -
40.0 44.8 38.2
74.9 81.4 69.1
24 21 -
IAEA 300
2.5g 5.0g IAEA
60.960.9 64.7
64.665.4 69.0
60.2 66.2 -
1.1 1.3 1.4
51.222.1 -
3.5 3.3 3.6
0.2 0.1 0.2
67.3 47.3 -
86.294.7 72.4
116 122 88.0
14 6 -
EML 9309
2.5g 5.0g EML
26.4 25.5 25.5
26.7 25.4 24.8
62.1 66.1 -
0.2 0.3 0.3
92.3 51.8 -
1.7 1.6 1.5
0.2 0.2 -
91.3 49.0 ~
32.2 32.7 -
25.8 25.0 24.8*
32 25 -
EML 9409
2.5g 5.0g EML
29.2 30.0 33.0
29.2 30.4 32.6
73.4 61.4 -
2.2 1.3 1.7
59.3 34.0 -
8.6 7.2 7.8
0.2 0.2 0.3
59.5 30.6 -
35.8 33.0 -
30.2 28.8 32.6s
19 20 -
EML 9509
2.5g 5.0g EML
29.1 30.4 29.5
30.6 30.1 30.4
65.8 84.6 -
I.S 1.6 1.8
87.3 80.3 -
5.3 4.9 5.2
18.1 15.6 17.5
74.5 44.9 -
39.3 24.7 -
34.2 23.5 30.4'
11 6
'Assumed secular equilibrium between ^U and 2JUTh.
Our results show that there is excellent agreement between our reported U isotope
activities and the recommended values. The recoveries for U are good and quite constant
with none of the 10 runs outside the range of 60 - 85% yield. The 24lAm activities are
generally in good agreement with the standard activities, even at very low concentrations.
Since we did not include a step to separate REE from the Am fraction (this can be done
fairly easily using TEVA Resin), the resolution will deteriorate should there be high REE
in the sample. This was the case for IAEA-135 which is why our values are too high.
The Am recoveries were generally good although more variable (22 - 99%) than U. We
84
-------
also note that the Am yield is lower in every case for the 5.0-g sample compared to the
2.5-g run. The Pu isotopic analyses display excellent agreement with recommended
values for all samples. The Pu yields were good although somewhat variable (31 - 91%)
and also displayed consistently lower recoveries for the larger sample size. An
independent experiment was run with 2.5 g of IAEA-300 using exactly the same
procedure but with 2 g of the Actinide Resin instead of the usual 0.7 grams. The results
showed significantly higher yields for Am and Pu, implying that there may have been
breakthrough of these ions in the previous runs. The U recovery remained in the same
range as all the other experiments. Unfortunately, we do not have many recommended
values for Th isotopes in these standards. Our calculated results agree reasonable well
with the recommended values for IAEA-135 but they do not for IAEA-300 which also
had very low Th recoveries. While the EML does not include Th in its program, we can
assume secular equilibrium between 234U (which is "known") and 230Th to estimate its
concentration. If this assumption is correct, a comparison indicates that our 230Th results
are in good agreement with that present in these soils. Unfortunately, Th continues to
display disappointing recoveries for reasons which we have not been able to discern. A
few tracer studies using 234Th indicate that Th behaves as predicted on all the columns
used through the procedure. It may be that a large fraction of Th is lost before loading
onto the Actinide Resin, either with the silica fraction when forced down with PEG (one
tracer test indicated over 50% of the Th was lost in this fraction) or coprecipitation with
micro amounts of insoluble sulfates, fluorides, or other compounds. Additional
experiments are planned to resolve this problem.
In general, use of the Actinide Resin appears to offer promise as a way to greatly
simplify the complex matrices present in natural soils in order to obtain more consistent
and predictable results.
Acknowledgments
Although the application of the Actinide Resin to soil analysis should prove useful to the
radioanalytical community, the truly innovative work associated with this development
was the initial synthesis of the Dipex™ extractant by Phil Horwitz and his group at
Argonne National Laboratories. We thank Phil as well as Andy Rollins (Eichrom
Industries, now with Motorola) for their helpful advice on our efforts. Additional
insights were also gamed through numerous communications with Bill McCabe (Institute
of Geological and Nuclear Sciences, New Zealand) and Barry Stewart (Environmental
Physics, Inc.). Partial financial support for this project was provided by the National
Science Foundation (OCE-9214493).
85
-------
References
Corbett, D.R., M. Schultz, and W. Burnett, 1995. Preliminary results for actinide
separations using Actinide Resin. Eichrom User's Workshop, 4ls( Conference on
Bioassay, Analytical, and Environmental Radiochemistry; Boston, MA, Nov. 13-17.
Horwitz, E.P. et al., 1993. Uptake of metal ions by a chelating ion exchange resin: acid
dependencies of actinide ions (Part I). Solvent Extraction and Ion Exchange, 11,943.
Horwitz, E.P., et al., 1994. Uptake of metal ions by a new chelating ion exchange resin:
the effect of matrices (Part V). Solvent Extraction and Ion Exchange, 12, 831-845.
Oughton, D.H., B. Salbu, T. L. Brand, J. P. Day, and A. Asker, 1993. Under-
determination of strontium-90 in soils containing particles of irradiated uranium oxide
fuel. Analyst, 118, 1101-1105.
Salbu, B., D.H. Oughton, A.V. Ramikov, T.L. Zhigareva, S.V. Kruglov, K.V. Petrov, N.V.
Grebenshakikova, S.K. Firsakova, N.P. Astasheva, N.A. Loshchilov, K. Hove, and P.
Strand, 1994. The mobility of l37Cs and ""Sr in agricultural soils in the Ukraine,
Belarus, and Russia, 1991. Health Phys., 67,518-528.
Sanderson, C.G. and P. Greenlaw, 1996. Semi-Annual Report of the Department of
Energy Office of Environmental Management Quality Assessment Program,
Environmental Measurements Laboratory, New York, 277 p.
Sill, C.W., F.D. Hindman, and J.I. Anderson, 1979. Simultaneous determination of alpha-
emitting nuclides of radium through californium in large environmental and biological
samples. Anal. Chem., 51,1307-1314.
Sill, C.W. and D.S. Sill, 1995. Sample dissolution. Radioactivity & Radiochemistry, 6, 8-
14.
86
-------
16
Quality Control in the Radioanalytical Laboratory; What Do We
Really Need to Do? Robert Litman, Chemistry Support Supervisor,
Seabrook Station, NAESCO, Seabrook NH, 03874
Environmental radioanalytical laboratories are being pressed to analyze more samples at a
higher frequency, and lower detection levels than ever before. Critical to this mission is
the performance of the instruments, how we evaluate their performance and when we
decide that remedial servicing of our process is required. The process includes not only
the instruments themselves but the samples and their preparation for analysis.
Routine quality control analyses will tell us important information about the instruments
we are using on a daily basis. However, there are other independent quality assessments
which are important to the entire sample analysis process. Some of these include
background determinations, independent blind analysis, sample splits and spikes.
Important to these assessment samples(whether solely instrumental or chemical
separation followd by instrumental analysis) is the activity level of the nuclide, the matrix
that it is presented in, and the level of interferences present.
The frequency and quantities of these assessments is an analytical decision based on
many factors. Do we appropriately challenge the analyst, instrument and method? A
discussion of personal experiences in this process of decision making will be presented,
and common sense approach to what we really need to do will be discussed.
87
-------
17
ANSi/ANS Standard for Radioanalytieal Data Validation
Thomas L. Rucker, Ph.D. - Science Applications International Corporation; Saleem R. Salaymeh,
Ph.D. - Westinghouse Savannah River Company; John Griggs - U. S. Environmental Protection
Agency; Chung King Liu, Ph.D. - U. S. Department of Energy; David E. McCurdy, Ph.D. - Yankee
Atomic Electric Company; Ann Rosecrance - Core Laboratories; Diane E. Vance, Ph.D. - Lockheed
Martin Energy Systems; Richard Wells - Lockheed Idaho Technologies Company; Robert W.
Holloway, Ph.D. - Nevada Technical Associates.
In past decades, radioanalytical data was usually accepted from the laboratory and assumed to be
correct without questioning its validity, accuracy, or precision. More recently, due to standard
protocols put in place by the U. S. Environmental Protection Agency (EPA), laboratory data has
undergone increased scrutiny before being used for critical decision making by a process known as
data validation. However, no standard EPA protocols have been developed for the validation of
radioanalytical data. Therefore, many different protocols with varying criteria have been developed
and used. This has led to inconsistency in approach and often to the unnecessary rejection of a
large percentage of data for use in decision making.
The American National Standards Institute/American nuclear Society (ANSI/ANS) Radioactive Waste
Management Committee is responsible for the development of radioactive waste management and
environmental remediation standards that address the generation, monitoring, characterization,
treatment, storage, and ultimate disposal of all categories of radioactive waste, including mixed
waste, and criteria and operations required for the environmental remediation of nuclear facility sites
that have become contaminated. The Subcommittee on Environmental Remediation of Radiation
Contaminated Sites manages the development and maintenance of standards that address the
cleanup of radioactive materials and radioactivity mixed with hazardous substances. This
subcommittee has authorized a writing group to develop a new ANSI/ANS Standard, 41.5, for
validation of data from radiological analysis supportive of environmental remediation.
This standard sets forth criteria and processes for demonstrating the validity of radioanalytical
results. It addresses the specific measurements and data required to permit the use of the results
for environmental remediation activities. This standard will be a consensus standard specifying the
essential requirements for accepting radioanalytical data as input to process control, site
characterization, waste acceptance criteria, waste certification, litigation, and other applications as
deemed necessary. This standard will provide a minimum set of checks and test that will ensure
a consistent approach for validation of data produced by any radioanalytical laboratory for waste
management, environmental remediation and process control.
The standard is being developed from the assumption that a proper data quality objective (DQO)
process has been used to define the quality of data needed for the decision process. Therefore, set
limits for the quality control parameters are not being recommended in the standard, but rather
qualification of data is to be tied to the DQOs. This allows the qualification to be based on how
much the error, bias, lack of precision, lack sensitivity, or lack of selectivity affects the decision that
is being made from the data. Furthermore, the standard will provide general guidance for when and
how much of the data should be validated.
It is expected that a draft of the standard will be completed this year. This paper will describe the
current status of the standard and the philosophies and approaches that the writing committee is
currently considering.
88
-------
18
Environmental Radiation Monitoring in Venezuela
Sajo Bohus L. and Greaves E.D.
Universidad Simon Bolivar -Physics Dep. Caracas - Venezuela
Abstract. During the past years, the human population increased
considerably in Venezuela. The natural resources suffered deeper the
negative consequences of an unplanned explotation contributing to the
planet global changes. These included also changes in the
radionuclides concentration in the environment. With the purpose to
establish the extent of the interference of the human activity with
environment and the changes in the radioactivity levels, a monitoring
program started in 1990. As an oil producing country, we are aware,
that the subsoil material enter the human environment contributing to
increase also the radiation level due to the redistribution of natural
radioactive substances. In order to determine the environmental
radiation changes and to assess the possible human risk the
concentration of alpha emitters were measured in potable water and
springs including thermal spas. Food and environmental samples (soil,
air, sediments and industrial waste) undergo radiation measurements
to determine the concentration of both gamma and alpha emitters. Since
several recicling programs exist, the ashes produced by oil burning
power plants were also studied determining that they are not suitable
as an additive material in the construction. Used tecnetium
generators were also studied establishing that environmental
regulation should be enforced. Results so far obtained are resumed as
follows: in the food it was determined the presence of cesium 137 in a
low concentration (<10 Bq/Kg), some had a high natural
concentration. The radon concentration in most of the water sources is
less than 11 Bq/1 however more than 5% are well above. Some thermal
wells did present around SOkBq/1; that is very unusuall and may be
considered as an exception. Gross alpha concentration in drinking
water for the 50% of the wells analyzed is less than 0.150 Bq/1 and
only the 8% has three time this value. The mean out door radon
concentration is 36 Bq/m3; In door values have a wide range being the
maximum of 400Bq/m3. The natural gas for domestic use contributes to
increase the in-door radon concentration by 20% in most of the
cases. In some caves we measured radon concentration that exceeds
20kBq/m3. The construction materials other than the natural
potassium-40 contained cadmium-109 in the range between 44 and 390
Bq/kg (this isotope has a short lifetime and its identification is
uncertain). Oil ashes did show that their content of radioactive
substances has a concentration ten times higher than the avarage
surface soil. In the south-east region where the human activity is
still very limited and restricted by a protected area law, the
berilium-7 was measured in tropical plants and fruits.
89
-------
19
Analytical Methodology for Non-normal Distributions of Environmental Data
Lin Zhichao, James J. Filliben, Kenneth G.W. Inn
National Institute of Standards and Technology
Gaithersburg, MD 20899
Over the past fifteen years, the Radioactivity Group at the National Institute of
Standards and Technology (NIST) has certified environmental levels of radionuclide
concentrations in River Sediment (SRM 4350B), Human Lung (SRM 4351), Human
Liver (SRM 4352), Rocky Flats Soil - 1 (SRM 4353), Freshwater Lake Sediment (SRM
4354), Peruvian Soil (SRM 4355), and Ocean Sediment (SRM 4357). Invariably, the
distribution of data for a considerable fraction of the certified nuclides are non-normal.
The methodology for the evaluation of the interlaboratory data has evolved over the
history of the program and now is fully capable of characterizing the non-normal
distributions. The results of the data analysis yields information on: 1) identification of
the best choice of statistical distribution; 2) robustness of the statistical distribution; 3)
mean values and standard deviation of the means; 4) tolerance limits and their
uncertainties at any desired level of confidence. The discussion of the methodology
will focus on the analysis of the Ocean Sediment intercomparison data. This
methodology could be applied to the analysis of field measurements and to estimate
the degree of false positive and negative measurements.
90
-------
INORGANIC
-------
20
Method Development Strategics for ICP-MS
Ruth Wolf. Senior ICP-MS Product Specialist and Zoe A. Grosser
The Perkin-Elmer Corporation, 50 Danbury Road, Wilton, CT 06897-0219
Abstract
Since its introduction as a commercial technique in 1983, Inductively Coupled Plasma Mass Spectrometry
(ICP-MS) has developed from a research instrument to a routine environmental production tool.
Increasing numbers of laboratories have invested in the technology for environmental analyses with much
success. However, many laboratories are still concerned about implementation of a performance-based
method and the choices that must be made in optimizing method performance.
In this talk we will describe the development of analytical parameters for implementation of RCRA SW-
846 Method 6020. The thought processes used in decision making will be outlined to give the user
guidance in general method development and injudicious choice of operating parameters for the most
rugged performance. Other topics that will be covered include:
• Development of internal standards to give the best performance in complicated matrices. The choice
of internal standards will be discussed and the tests performed to optimize the choice described.
• The dissolved solids-handling capability of the nebulizer will be discussed for different types of
nebulizers. Certainly, RCRA sample types can be more complicated than those encountered in the
analysis of drinking water. The problems with matrices will be discussed and strategies outlined.
• The performance of other elements not included in the current list of method 6020 analytes will be
highlighted. Information on several additional elements, such as Se, can be very useful. Instrument
and method performance will be evaluated for several types of sample matrices, including soil digests
and TCLP extracts.
The final method parameters chosen and the performance on several samples will illustrate the completed
development and conclude the discussion.
Introduction
As many laboratories contemplate purchasing an ICP-MS, questions and concerns are raised about the
implementation of ICP-MS in their laboratory. Indeed the "comfort factor" with how reliable, easy to use,
and the amount of method development time required are usually an important considerations in the
ultimate decision. The development of "turnkey" methods for commonly used methodologies such as
EPA Method 6020 and Method 200.8 by vendor application specialists has gone a long way to alleviate
the new technology anxiety that has been associated with ICP-MS. In this talk, we will describe the
development of an instrumental method for the implementation of RCRA SW-846 Method 6020 and
discuss the thought processes and decision-making involved.
The first decision to be made in developing a method is what regulatory programs will be supported by the
resulting ICP-MS instrumental method. This will help determine which path method development
strategies will take. If RCRA SW-846 Method 6020 is the primary method for implementation, then
appropriate sample preparation procedures must be selected to go along with that method. In some cases,
91
-------
only one sample preparation method (e.g., microwave digestion, SW-846 Method 3051) will be used and
the overall development scenario is greatly simplified. However, if a laboratory doesn't have the capability
to perform Method 30S1, more than one sample preparation procedure (e.g.. Methods 3010, 3020, or
3050) may need to be investigated. Since most interferences are caused by either the sample matrix or the
digestate matrix, some consideration should be given to the impact on analytes of interest. For example,
if As and Se are important analytes you may wish to avoid sample preparation methods (such as the TCP
digestion in Method 3050) that introduce chloride to the sample so that ArCI interferences on As and Se
can be minimized. If avoiding hydrochloric acid in the sample preparation procedure is not possible,
isobaric corrections may be used; however, their impact on the data quality objectives (DQO) of the
analysis should be evaluated. This can be done by performing preliminary background scans on the
sample preparation blanks and evaluating the differences in the overall background between different
digestion matrices. This information can then be used in all stages of the method development to help
determine what interference corrections, dilution levels, and Method Detection Limits (MDLs) might be
possible with each different sample preparation scheme.
Once the sample preparation methods have been decided upon, the real work of method development for
the ICP-MS can begin. A confusing point for most ICP-MS novices is setting up the measurement timing
on the quadrupole mass spectrometer. Although the measurement time can be an important factor in
some applications, for most environmental work, the default dwell and sweep times are generally
adequate. For elements where better detection limits may be necessary to achieve the DQOs (such as a
particular MDL level) the dwell times can be doubled or tripled to spend more time acquiring data on a
particular element. An example of this case is arsenic, which doesn't ionize well in an argon plasma. It
may be necessary to measure 3-4 times as long to get count rates similar to those achieved for the more
easily ionized elements such as copper.
Other considerations in setting up data acquisition times include sample volume and productivity. In the
cases of limited sample volume, the total acquisition time may be determined by the amount of sample. In
a production-type environment the DQOs may need to be balanced by productivity concerns. For
example, the profitability of an ICP-MS purchase might be based on the need to analyze 250 samples in
an 8-hour shift, limiting the data acquisition time for each sample to about two minutes including sample
uptake and washout. However, if the lowest possible detection limits are the goal and productivity is not a
real concern, then longer data acquisition times can be used to help achieve the necessary detection limits.
In conjunction with measurement time the appropriate choice of data collection mode (i.e., peak hopping
or full mass scanning) must be made. This decision is also affected by DQOs and productivity concerns.
For example, if you scan 20 points per peak with the same dwell time per amu, you will be spending less
time at the point of highest signal to noise than compared to counting on the peak maximum (peak
hopped) for the entire dwell time. This difference will ultimately affect your detection limits and the
number of samples that can be analyzed in a given time period. It is a simple matter to set up several test
data acquisition scenarios and test the differences that might be obtained in detection limits by running a
10-replicate IDL. The use of the data export features of the instrument software can make these
calculations automatic by importing the data into an IDL/MDL calculation spreadsheet.
The dissolved solids content of the digested matrix will influence the washout time of the system. Long
washout times will prevent build-up and ensure that the system is clean for the next sample. However,
long washout times will reduce productivity and may be unnecessary. Several short studies can
characterize the system for a representative matrix and determine the optimum washout time. A sample
digest, diluted to a representative concentration can be monitored over time for sample build-up by
measuring the internal standard recoveries until they deteriorate below the acceptable level. Washout
studies will determine how soon the next sample can safely be introduced. Table I shows a representative
list of elements introduced at high levels and the washout observed for an instrumental method based on
92
-------
EPA 200.8. Most elements were rinsed out at or below 10 times the MDL after one analysis cycle,
indicating that a 35-second washout was sufficient for this method.
Table 1
Estimated Rinse Times (Selected Elements)
Analyte
v
Zn
Se
Ba
Ife
Tested Cone
(Hg/L)
5000
20,000
"20,000
5000
20
Measured Concentration (ng/L)
Cycle 1 j Cycle 2 ! Cycle 3
0.113 10.048 | 0.041
0.360 j 0.189 0.184
0.182 j -0.330 -0.169
MDL (ng/L)
0.03
0.03
0.05
0.110 j 0.039 0.025 j 0.04
0.125 | 0.090 0.071 j 0.02
10(MDL)
0.3
0.3
0.5
0.4
0.2
Analysis Cycle Time
Rinse time: 35 sec @ 48 rpm
Uptake time: 25 sec @ 48 rpm
Stabilize time: 10 sec @ 24 rpm
Analysis time: 2:06 min @ 24 rpm
Total time: 3: 16 min
Next the choice of which isotopes to monitor for each element must be addressed. If you are modifying a
"turnkey" method, this becomes a less cumbersome task than starting from scratch. In the beginning of
any method development, the "more is generally better" rule can be used. It is always better to collect too
much data and not have to use part of it than to have not acquired enough and find you need it later.
Elements with multiple isotopes can be monitored at several masses and the results compared.
Interferences in ICP-MS are additive, so the lowest answer is probably more accurate in a situation where
isotope results differ. Internal standard elements must be selected after screening typical samples using
semiquantitative analysis to determine the presence of any commonly used internal standard elements. A
general guideline in choosing specific elements is to keep internal standard elements fairly close in mass
(approximately within 50 amu) to the analyte elements and use internal standards that exhibit the same
properties in an argon plasma. Sometimes a different internal standard can be used to correct a matrix
related problem. For example, during the development of Method 6020 it was observed that suppressed
recoveries were obtained for elements with fairly high ionization potentials (As, Se, and Zn) high levels
of the alkali metals (e.g., Ca, Na, K, Mg, or Fe) were present. This appeared to be an ionization
suppression related interference. A quick study was performed to verify this and then the use of an
alternate internal standard with a first ionization potential (see Table II) closer to those of As, Se, and Zn
was used. It turned out that by simply switching internal standards to Ge the problem was solved.
Table II
Ionization Potentials
Element
In
'Y ~
"sic
Rli
Ge
Zn
Se
As
I 1st Ionization Potential (v)
f5;76
j5".30 ~ "
6'.54
7.46
7.90
9.34
9.75
9.81
93
-------
Once the best isotopes are determined for a given element the others can be taken out of the method to
save time (if this is an issue) and make the method simpler. The best way to test a method for accuracy
and isotope selection is the use of certified reference materials in a matrix similar to the samples.
Knowing the right answer is a great help in determining which isotopes may have interferences. Table III
demonstrates the method performance of RCRA 6020 for an estuarian sediment certified reference
material. Other ways to test accuracy include performing spike recovery tests and matrix dilution tests on
representative samples. Finally, once method development is completed you can run comparability studies
on your own samples and compare results obtained between ICP-MS and other techniques used in the lab
such as graphite furnace atomic absorption or optical emission ICP. It is a very important to remember to
compare the same sample preparation method for each analytical procedure. It is widely known that each
of the EPA weak acid leaching digestions induces method bias into the measured concentrations, so it is
important to use the same exact digest for each analytical method you wish to compare. For example,
most analyses for lead in soils will yield higher results for method 3050 using the ICP finish (with
hydrochloric acid) than if the same preparation method were used with the GFAA finish (no hydrochloric
acid).
Once the comparability testing is complete, you can report numbers to your customers that you have
confidence in and be able to explain any differences they might observe from past analyses using alternate
techniques and sample preparation methods. Indeed, performing quick comparability studies and
submitting the data to your clients can go a long way in gaining their acceptance of a new technique that
can usually give them more information about their sample with fewer analyses and problems.
Table III
QC Checks - Method 6020
High Purity Estuarian Sediment
Element
At
Sb
As
Ba
Be
Cd
Cr
Co
Cu
Pb
Mn
Ni
A6
Tl
Zn
Sample Cone.
(mg/kg)
548
0.004
0.11
ND
0.02
0.0004
0.78
0.12
0.20
0.31
3.64
0.31
ND
, IP . _
1.52
%Rec. of Cert.
Value
84.2
98.9
112
-
-
104
97.4
116
99.6
104
104
104
-
.
109
Spike 1 % Rec. of Spike
_ dffi/L) 1— _.
NA I NA
150 ! 103
150 I 98.4
150 j 98.9
150 ! 117
150 j 93.6
150 1 97.0
150 | 97.4
150 ! 96.0
150 j 102
150 | 90.2
150 ! 96.3
150 | 94.9
150 1 99.6
150 i 98.2
% Rec. Limits
75-125
75-125
75-125
75-125
75-125
75-125
75-125
75-125
75-125
75-125
75-125
75-125
75-125
75-125
75-125
Conclusion
There are many considerations involved in the method development process. Data Quality Objectives
must be balanced with laboratory productivity needs in order to obtain the most cost effective
94
-------
measurement scheme. Method development must ultimately begfn with the sample preparation step,
which can greatly affect the overall method development in both time and effort. Once a draft method has
been developed, testing procedures should be implemented to test its overall accuracy for the range of
sample matrices expected to be analyzed using the method. In addition, the use of comparison studies can
be performed to help overcome any client concerns regarding comparability issues with past data and to
give them an overall confidence about their sample results using the new analytical technique.
95
-------
21
RAPID FIELDABLE ANALYSIS FOR MERCURY
Donald F. Foust and John Y. Gui, Corporate Research and Development, General
Electric Company, P. O. Box 8, Schenectady, NY 12301
ABSTRACT
The cost for cleaning-up contaminated sites is often high due to the large number of
samples required for site characterization. This process is slow since analyses are
performed off-site. A method for rapidly determining mercury concentrations in
environmental samples has been developed and tested at the General Electric Company.
The analysis uses Anodic Stripping Voltammetry (ASV) to determine mercury
concentrations in solution. Mercury is transferred to solution from environmental
samples by extraction with an aqueous mixture of potassium iodide and iodine. Analysis
requires less than 10 minutes for a 1 ppm detection limit and extraction needs less than 30
minutes per sample. The reproducibility of the electrode surface is enhanced by using an
iodine atomically modified electrode. The method can be made fieldable and requires no
corrosive acids. The method is also independent of the type of contaminated media and
the species of mercury. Recently, this method was part of a screening evaluation for
mercury determination technologies conducted by the Energy & Environmental Research
Center (EERC) at the University of North Dakota in Grand Forks, North Dakota. A
comparison of the data derived from the ASV Method versus the standard EPA Method
7471 laboratory methods for determining mercury content in environmental soils showed
an excellent correlation from less than 1 ppm to greater than 10,000 ppm mercury.
INTRODUCTION
Mercury has caused considerable concern when improperly released into the environment
[1-2]. The costs associated with contaminated sites are often substantial. A significant
portion of the cost is associated with site characterization and assessment. Long delays
between acquisition of samples and the analysis of samples leads to long idle periods for
personnel and equipment. The development of fieldable analyses has resulted in order to
contain costs [3].
Several methods for the determination of mercury in environmental samples while in the
field are in various stages of development. Mercury determination is accomplished by
the use of x-ray fluorescence [4], immunoassay [5], or vapor analysis 16-7]. While each
method is capable of determining the presence of mercury, several serious deficiencies
exist including a lack of sensitivity, matrix interferences, limited range, limitations due to
the form of mercury, and/or the use of hazardous reagents.
An alternate fieldable method for rapidly determining mercury in environmental samples
has been developed at the General Electric Company [8]. The technique utilizes a rapid,
selective extraction of mercury from the solid sample. Once mercury is solubilized, it is
detected by electrochemical means [9]. The method has been found to have an excellent
correlation for total mercury content with the standard laboratory method of sample
leaching with strong, hot acid followed by atomic absorption analysis of the leachant
[10]. The extraction solution may also be analyzed by chemical reduction of the mercury
followed by vapor analysis of the resulting elemental mercury [ 111.
96
-------
In addition to site characterization, rapid field analysis can be used in the remediation of
contaminated sites [12]. On-site monitoring of the output from decontamination
processes can aid in the control of the operation, helping to contain the operating costs.
In this study, the electrochemical methodology was applied to a large number of
environmental soil samples obtained from natural gas metering stations. Mercury spills
occurred along the natural gas pipeline due to spills and vandalism of manometers
containing mercury [13]. Cost containment during site characterization is critical for
these sites because of the large number of metering sites and their relatively small size.
Thousands of metering sites require characterization. Those sites containing mercury
average only 1-2 cubic meters of contaminated soil [14].
Rapid field analysis has been identified as a necessary technical need for these sites [15].
In this study, a comparison of the results produced by the ASV Method and the standard
laboratory method for determining total mercury content (EPA 7471) was undertaken.
EXPERIMENTAL
Soil samples were obtained by EERC from a variety of natural gas metering stations in
which mercury spillage from manometers had occurred. Three ranges of mercury
concentrations in soil were desired; those containing >1000 ppm mercury, those
containing 100-1000 ppm mercury, and those containing <1 ppm mercury. Often,
samples from the high mercury-containing soils had visible beads of elemental mercury
present. Oversized rock and debris were physically removed and the soil blended in a
stainless steel mixer. The homogenized samples were then split into 12 sub-samples, two
of which were analyzed by EPA method 7471. Samples were submitted as duplicates in
a double-blind experiment. Standard soils were obtained from NIST (SRM 2709, 2710
and 2711).
A 1.0 g soil sample was weighed directly into a four-inch threaded Ace pressure tube. To
the tube was added 10.0 mL of extracting solution containing 1.0 M KI and 0.5 M \2- A
rubber gasket and Teflon threaded cap were used to secure the vessel. The mixture was
shaken, then placed in a boiling water bath for 30 minutes. Next, the sample was
removed, shaken, and allowed to cool to room temperature. The mixture was then
filtered through a 0.45 micron syringeless filter made of PTFE. The filtrate was saved.
A portion of the filtrate (0.10 mL) was pipetted into a 25 mL glass vial containing 10 mL
electrolyte solution (0.25 M K2SO4 and 0.1 M KI). The vial was then capped and shaken
for solution homogenization. Next, the vial cap was replaced with a special Teflon
electrode assembly that was constructed in such a way that once tighten onto the vial, all
electrodes were immersed into the electrolyte solution. The assembly consisted of a
working electrode (Au disk, 1 mm in diameter), an auxiliary electrode (Pt wire) and a
reference electrode (Ag/AgI, 0.1 M KI). Subsequently, the electrodes were connected to
a computer controlled potentiostat (Cypress Systems) and a square-wave anodic stripping
voltammetric (SW-ASV) experiment was performed. The resulting stripping current was
then converted into mercury concentration by comparison with the standards. Deposition
time was adjusted to produce a 0.5 ppm detection limit for mercury.
97
-------
The gold working electrode was initially polished on a polishing pad with alumina slurry,
then electrochemically cleaned by the in-situ oxidation-reduction cycling between 1 850
mV and -150 mV in 1 M H2SO4 solution. The clean gold electrode was then stored in
the 1 M H2SO4 solution until needed. The electrode was cleaned by in-situ
electrochemical cycling once its surface was contaminated based on the decreased
stripping current of a standard solution. No additional electrode polishing was required
for the remainder of the experiment.
RESULTS
The mercury content of 37 different soil samples is listed in Table 1. The samples were
analyzed in duplicate by both General Electric's Anodic Stripping Voltammetry (GE
AS V) Method and EPA's Standard Method 747 1 .
DISCUSSION
A mixture of potassium iodide and iodine dissolved in water serves as the extracting
agent. This mixture can thoroughly remove a wide concentration of mercury species
from contaminated solids due to the high affinity of iodine for mercury. The combination
of an oxidant (iodine) and a solubilizing agent (potassium iodide) give rise to the
mixture's exceptional ability to remove a wide variety of mercury species (equations 1-3).
The combination of elevated temperature and increased concentration enable through
mercury removal in a minimum of time. This extraction mixture has been employed as a
means of decontaminating solids containing mercury [16].
Hg + \2 + 2KI -» K2HgU (equation 1)
HgO + H2O+ 4 KI -> K2HgI4 + 2 KOH (equation 2)
HgS + 12 + 2 KI -> K2HgI4 + S (equation 3)
As seen in equations 1-3, no acids or bases are required for the removal of mercury from
the soil. Therefore, the extraction process can be conducted without the use of hazardous,
corrosive acids.
A measure of precision for the method is illustrated in Figure 1. The amount of scatter in
the results between EPA Method 7471 and GE ASV is comparable. The data from GE
ASV produced a line with a better linear regression coefficient (R=0.97) than that from
the standard method (0.92). A line with a slope of 1 is present to show an ideal
correlation of the data. Scatter in the data at high mercury levels in the soil is most likely
due sampling. At levels greater than 1000 ppm mercury, elemental mercury was visible.
The presence or absence of a bead of mercury can have a dramatic effect on the mercury
content of a small sample. Scatter in the data at lower mercury levels may be a better
measure of the reproducibility of the methods.
Voltammetric methods often suffer from a lack of reproducibility due to fouling of the
electrode surfaces. It is known that both iodine and iodide strongly absorb onto gold
surfaces as iodine atoms. This monolayer of iodine protects the electrode from
contamination. In the GE ASV Method, gold serves as the working electrode and both
98
-------
iodine and iodide are present in the electrolyte solution. This results in a reproducible
electrode surface. As seen in Figure 1, the reproducibility of the data is comparable to, if
not better than, the standard technique.
Table 1
Data Comparison
GE ASV Mei
(PP
Set#l
13500
5850
5200
1950
1860
1840
1320
950
730
350
340
320
257
142
119
113
47.8
46
38
21.5
18
16
13
7
4.3
2.5
1.8
0.5
0.5
<0.5
<0.5
<0.5
<0.5
<0.5
<0.5
<0.5
<0.5
•cury Content
m)
Set #2
9400
6200
4000
1730
1350
1748
3100
950
1150
220
460
243
170
140
234
69
49.5
70
54.3
26.5
22.5
16
9
5.5
6.9
3.6
2.1
<1
<1
2
1
<1
<1
<1
<1
<0.5
<0.5
EPA Method 7471
(PP
Set#l
14100
2330
6950
726
624
8670
6390
654
"""""850
143
... _____ „
218
106
96.3
102
30.1
34.2^
27.5
25.6
12.4
10.3
14.4
11.4
5.76
0.972
2.4
0.643
0.103
" 0:286 ""
0.234
0.466
0.262
0.056
0.216
0.076
0.561
0.088
Mercury Content
m)
Set #2
8830
3420
4780
1370
300
561
3670
536
1036
110
445
408
253
125
120
29.7
60
45.7
30
17.2
10.9
16.7
9.43
7.1
3.29
1.99
0.707
0.112
0.352
0.27
0.206
0.202
0.057
0.165
0.275
0.438
0.137
99
-------
Figure 1
Comparison of Data Set #1 with Data Set #2 for GE ASV and EPA Method 7471
100,000
1 10,000
3 1,000
=
<3
t
100
1 0
0.1
0.1
10 100 1,000 10.000 100,000
Mercury Contenl from Dili Set »1 (ppm)
The data in Figure 2 show a comparison of the average mercury content in the soil
samples measured by GE ASV and EPA Method 7471. The error bars in the x direction
reflect the scatter in the data from EPA Method 7471 while the error bars in the y
direction reflect the scatter in the data produced by the GE ASV Method. The line with a
slope of 1 indicates perfect correlation between the two data sets. An excellent
correlation between the two data sets exists from 1 ppm to 10,000 ppm mercury content
in the soil samples. As seen in Figure 2, the mercury content measured for a majority of
the GE ASV data is slightly greater than that measured by EPA Method 7471.
Another way to view the GE ASV Method is as a screening technique rather than a
quantitative one. The data in Table 2 show that the GE ASV Method can be used as a
screen for mercury in soil. Setting the screen for a content of 1, 10, 100, or 1000 ppm
mercury produced no false negative results. The only false results were false positive
readings, and these were infrequently encountered.
100
-------
Table 2
Screening Analysis of 37 Duplicate Soil Samples Using GE ASV*
Screening Level
(ppm)
1000
100
10
1
Number of
Samples Above
Screening Level
7
13
23
28
Number of
Samples Below
Screening Level
30
24
14
9
Number of
False Positive
Results
1
0
0
2
Number of
False Negative
Results
0
0
0
0
*Comparison to Average Mercury Content Determined by EPA Method 7471
Figure 2
Comparison of Average Mercury Content Measured by GE ASV Versus EPA
Method 7471
100.000
I 10,000
u
I-
u
t
t
1,000 r
100 r
1C
0.1
0.1
10 100 1,000 10,000 100.000
Average Mercury Content by EPA Method 7471 (ppm)
SUMMARY
The method of determining mercury content in soils by first solubilizing the mercury with
a mixture of iodine and potassium iodide followed by analysis of the resulting liquid with
Anodic Stripping Voltammetry was successfully tested on a wide variety of samples from
natural gas pipeline sites. Not only is this new method an excellent screen for mercury, it
can yield quantitative data rivaling the standard method of acid digestion followed by
101
-------
analysis of the extract by atomic absorption. The method yields quantitative results from
1 ppm to 10,000 ppm mercury in soil samples in less than 1 hour. The method is also
amenable to field situations.
ACKNOWLEDGEMENTS
The authors would like to acknowledge the assistance of Shannon Arrowood of En-Sys
Inc. in the acquisition of the data. In addition, thanks are extended to John Harju of the
University of North Dakota for the invitation to participate in the round-robin study.
Financial support for EERC's program from the Gas Research Institute, the U. S.
Department of Energy, and Nova Corporation of Canada is acknowledged.
REFERENCES
[1] C. M. Caruana, "Mercury Pollution: Seeking a Quicksilver Lining", Chemical
Engineering Progress, January 1996,10-16.
[2] D. R. Sasseville, M. Barg, and S. R. Clough, "The Reemergence of Mercury as a
National Concern", Journal of Environmental Regulation, Autumn 1995, 42-49.
[3] J. Glanz, "Field Kits Challenge Fixed Labs in Environmental Testing", R&D
Magazine, March 1993,12-15.
[4] N. A. Wogman, "Evaluation of an In-Situ X-Ray Fluorescence Analyzer for Inorganic
Pollutants in Sediments and Water Columns", PNL-3168, September 1979.
[5] C. Schweitzer, L. Carlson, B. Holmquist, M. Riddell, and D. Wylie, "Enzyme-Linked
Immunoassay (ELISA) for the Detection of Mercury in Environmental Matrices", 10th
Waste Testing & Quality Assurance Symposium, Arlington, VA, July 11-15, 1994.
[6] A. Roffman, E. J. Verbanic, and R. P. Shervill, "A Dynamic Methodology for
Estimating Cleanup Efforts for Mercury in Soil at Gas Utility Gate Stations",
Remediation, Autumn 1993,413-424.
[7] A. A. Kriger and A. A. Turner, "Technique for Rapid Field Analysis of Mercury in
Water and Soils", Extended Abstracts, I&EC Special Symposium, American Chemical
Society, Atlanta, GA, September 19-21,1994,1294-1297.
[8] J. Y. Gui and D. F. Foust, "Detection and Measurement of Heavy Metals", U. S.
Patent 5,391,270 (February 21,1995).
[9] J. Y. Gui and D. F. Foust, "Rapid Electroanalysis of Mercury in Environmental
Wastes", 3rd International Symposium on Field Screening Methods for Hazardous
Wastes and Toxic Chemicals, Las Vegas, NV, February 24-26, 1993.
[10] "Assessing and Remediating Mercury Contamination", The Hazardous Waste
Consultant, July/August 1993,1.23-1.25.
102
-------
[11] N. L. Ayala, R. R. Turner, and D. F. Foust, "Field Headspace Method for the
Determination of Mercury in Soil Samples Using an Iodide-Based Extractant", Pittcon'96,
Abstract 965, March 3-8, 1996.
[12] S. H. Peterson, E. L Lahoda, D. C. Grant, E. F. Sverdrup, T. V. Congedo, J. Bartko,
R. E. Witkowski, A. L. Wolfe, W. D. Partlow, and M. C. Skriba, "System and Method for
On-Line Montoring and Control of Heavy Metal Contamination in Soil Washing
Process", U. S. Patent 5,133,901 (July 28,1992).
[13] D. S. Charlton, C. R. Schmit, D. J. Stepan, F. W. Beaver, and J. M. Evans, "Research
Program Dealing with Mercury in Soil at Natural Gas Industry Sites", Arsenic and
Mercury Workshop on Removal, Recovery, Treatment, and Disposal, EPA/600/R-
92/105, August 1992, 70-72.
[14] D. S. Charlton, J. A. Harju, D. J. Stepan, V. Kuhnel, C. R. Schmit, R. D. Butler, K.
R. Henke, F. W. Beaver, and J. M. Evans, "Natural Gas Industry Sites Contaminated with
Elemental Mercury: An Interdisciplinary Research Approach", in Mercury Pollution
Integration and Synthesis. C. J. Watras and J. W. Huckabee, Eds., Lewis Publishers, Boca
Raton, FL, 1994, 595-600.
[15] K. R. Henke, V. Kuhnel, D. J. Stepan, R. H. Fraley, C. M. Robinson, D. S. Charltcn,
H. M. Gust, and N. S. Bloom, "Critical Review of Mercury Contamination Issues
Relevant to Manometers at Natural Gas Industry Sites", GRI-93/0117, August 1993.
[16] D. F. Foust, "Extraction of Mercury and Mercury Compounds from Contaminated
Material and Solutions", U. S. Patent 5,226, 545 (July 13, 1993).
103
-------
22
EPA METHOD 3052: DEVELOPMENT,
CHEMISTRIES AND VALIDATION
Peter T. Walter. H.M. 'Skip' Kingston, Dirk D. Link
Duquesne University, Department of Chemistry and Biochemistry
Pittsburgh, PA 15282-1503
ABSTRACT
EPA Method 3052 has been proposed to meet the EPA's need for analyses
requiring total decomposition. This performance based method has the
flexibility to allow the analyst to modify the acid digestion chemistry for the
particular matrix, analyte, and analytical technique. Due to the methods
inherent flexibility, the analyst is encouraged to understand more of the
fundamental chemistry of the target analytes and matrix. This paper
discusses some of the parameters and unique aspects of this new method.
INTRODUCTION
Due to a recent court decision, acid leaching methods are not sufficient
decomposition techniques for ash and other incinerator wastes and fuels.
Method 3052 was created to answer the need for a general method for the total
analysis of inorganic constituents in a matrix. Total sample decomposition is
interpreted to mean the sufficient destruction of the matrix to permit the total
analysis to be performed for the elements of interest. The proposed EPA
Method 3052, "Microwave Assisted Acid Digestion of Siliceous and
Organically Based Matrices" is applicable to many solid matrices such as ash,
oil, oil contaminated soil, sediment, sludge, soil, and biological tissue. It
provides a total digestion of the above matrices for analysis of the 26 RCRA
regulated elements. The performance based method involves different
combinations of HNO3, HF, and HC1 to decompose the matrix. Standard
Reference Materials (SRMs) of each matrix type were used to test the proposed
method. Method 3052 is a performance-based sample preparation method
that has extensive alternatives and more capabilities for adaptation to varied
media and matrix applications than previously promulgated methods. It
represents the first of a new generation of methods aimed at producing
accurate analyses of total analytical composition through performance-based
criteria. While it handles all 26 RCRA required elements, it also provides the
opportunity to extend sample preparation to most elements in the periodic
table. It permits optimization for specific elements and matrices, and for the
application of various detection techniques.
104
-------
Method 3052 can be used to evaluate elemental concentration as a screening
method for method 1311, the "TCLP" method. Section 1.2, of Method 1311,
states:
Method 1311: Toxicity Characteristic Leaching Procedure
1.2 If a total analysis of the waste demonstrates that individual
analytes are not present in the waste, or that they are present but at
such low concentrations that the appropriate regulatory levels could
not possibly be exceeded, the TCLP need not be run.
Accordingly, EPA Method 3052 could be used to determine the potential
elemental contamination. If the results of the analysis using Method 3052
were below the defined elemental contamination limits, performing TCLP
Method 1311 would not be required.
The original closed vessel microwave protocol that became EPA method 3052
was developed for elemental certification of Standard Reference Materials
(SRMs) at the National Institute of Standard and Technology (NIST) by Dr.
Kingston and his research staff (1-3). This technology was developed for the
sample preparation procedure for elemental analysis by many instrumental
techniques, including isotope dilution mass spectrometry (IDMS), inductively
coupled plasma mass spectrometry (ICP-MS), inductively coupled plasma
optical emission spectroscopy (ICP-OES), flame atomic > absorption
spectroscopy (FAAS), graphite furnace atomic absorption spectrometry (GF-
AAS), neutron activation analysis (NAA), chelation ion chromatography
(ChIC), polarography, and laser enhanced ionization (LEI). Due to the
method's flexibility of using different chemistries, matrices, analytes, and
analytical techniques, this method is one of the most robust performance
based methods ever proposed to EPA.
Validation of this method has been underway for over eight years, starting at
NIST and continuing at Duquesne University and at other collaborating sites.
A description of some of the analytical aspects is presented.
EXPERIMENTAL
A 0.25 to 2.0 g sample, depending on the reactivity and the potential for the
production of gaseous by-products during digestion, of the NIST SRMs were
transferred into a fluoropolymer digestion vessel liner (Milestone
Corporation or CEM Corporation). 9 mL of sub-boiled distilled concentrated
nitric acid and various quantities of sub-boiled distilled concentrated
hydrofluoric acid and sub-boiled distilled concentrated hydrochloric acid
were added to each vessel. The choice of digestion reagents is dependent on
105
-------
many factors including the matrix, analytes of interest, and detection
technique. The vessels were capped and sealed according to manufacturer's
directions. The samples were simultaneously digested by heating the samples
to 180 ± 5°C within 5.5 minutes and maintaining the temperature at 180 ± 5°C
for the remainder of the 15 minute heating program. This heating profile
may be altered for reactive matrices or excessively slow decomposing
components and shall be viewed as the general heating profile. A typical
digestion temperature and pressure profile of a soil and an oil are illustrated
in Figures 1 and 2. Once the vessels were cool, the vessels seal integrity was
evaluated to determine whether the vessel had vented during dissolution.
The sample was filtered through a 0.2 urn Teflon filter and then diluted to 100
mL with 18 MQ-cm water (Barnstead NANOpure). All operation steps
involving an uncapped microwave vessel as well as all post-digestion
procedures were performed in a class 10 clean environment.
200
0 5 10 15
Time (min)
Figure 1: The typical temperature and pressure profiles for the digestion of
a soil. Approximately 0.25 g soil digested with 9 mL HNO3 and 4 mL HF.
RESULTS and DISCUSSIONS
EPA Method 3052 is a total digestion method designed to achieve total
digestion of a wide variety of matrices. Data will be presented for elemental
analysis of NIST SRMs including biological, botanical, geological, and
metallurgical matrices.
EPA Method 3052 is a performance based method, designed to achieve or
approach total decomposition of the sample through achieving specific
106
-------
reaction conditions. The temperature of each sample should rise to 180 ± 5°C
in approximately 5.5 minutes and remain at 180 ± 5°C for 9.5 minutes.
200
10 15 20
Time (min)
Figure 2: The temperature and pressure profiles for the digestion of an oil.
0.306 g motor oil digested with 9 mL HNO3.
Reactive Matrices
If the matrix is composed of reactive materials, the digestion profile may be
altered for safety purposes. The digestion of motor oil is an example of a
matrix that will decompose rapidly and produce considerable quantities of
gas. For safety reasons, the matrix should be decomposed by slowly heating
the sample to 180 ± 5°C, thus allowing the easily decomposable components
to react in a controllable manner. The decomposition should be maintained
at a temperature of 180 ± 5°C for 9.5 minutes to complete the digestion. The
safe and controlled decomposition of motor oil is illustrated in Figure 2,
of Digestion
The method's robustness and new higher pressure microwave vessels enable
the sample size to be scaled up to larger sample sizes, see section 7.3.12 of
Method 3052 (4).
The scale-up of the digestion of NIST SRM 2710 (soil) is illustrated in Figure
3. The linear relationship between the analyzed concentration of cadmium to
the mass of the soil digested indicated that as the sample size was increased,
the digestion chemistry and reaction conditions were sufficient to completely
107
-------
7.3.12 Sample size may be scaled-up from 0.1, 0.25, or 0.5 g to 1.0 g
through a series of 0.2 g sample size increments. Scale-up can produce
different reaction conditions and/or produce increasing gaseous
reaction products. Increases in sample size may not require alteration
of the acid quantity or combination, but other reagents may be added to
permit a more complete decomposition and oxidation of organic and
other sample constituents where necessary (such as increasing the HF
for the complete destruction of silicates). Each step of the scale-up must
demonstrate safe operation before continuing.
decompose the matrix. However, this does not imply that any matrix can be
scaled-up to 2 g. The limits of scale-up are dependent on a number of factors
including the pressure produced from the digestion of organic components,
the stoichiometric limitations or the consumption of digestion reagents, and
the stability of the analytes in the digest solution.
X)
a,
*
U
<44
o
c
2
S
u
5
50
40
30
20
10
0
y = -0.616 + 2^.41 x R= ^0.992
0 0.5 1 1.5 2 2.5
Mass of NIST SRM 2710 Digested
Figure 3: The linear range of cadmium analyzed from 0.25 to 2g of NIST
SRM 2710. The soil was digested with 9 mL HNO3 and 3 mL HF.
Reagent Reactions and Analyte Stability (Robustness)
Method 3052 allows the analyst to select the specific digestion reagents for
specific matrices and analytes of interest. The method suggests the acid
digestion solution should be 9 mL nitric acid and 3 mL hydrofluoric acid.
108
-------
However, hydrochloric acid, hydrogen peroxide, water, and many other
reagents could be added for several reasons including aiding the completeness
of the reaction and the stabilization of a particular analyte(s). This latitude in
the digestion chemistry requires that the analyst be knowledgeable of the
fundamental chemistry of the digestion acids and matrix. Comprehensive
reviews of acid digestion chemistry have been prepared to support this need
(5-7) and will be maintained and expanded on the world wide web
"SamplePrep Web" site at Duquesne University.
URL: http://nexus.chemistry.duq.edu/sampleprep/prepnet.html
A table providing guidance in selecting acid combinations has been prepared
and will be included in the final draft of Method 3052.
The following two sections are brief examples that describe some of the
attributes of hydrochloric and hydrofluoric acids.
Complexation and Stabilization Effects of Hydrochloric acid
Hydrochloric acid is a non-oxidizing acid which exhibits weak reducing
properties during dissolution. Most metals form soluble metal chlorides with
several notable exceptions: AgCl, HgCl, TiCl, and PbCl2- The solubility of
silver can be dramatically increased by the complexation of silver chloride
with excess chloride ions.
Ag+ + Cl <===> AgCl
AgCl + CI- <===> AgCl;1
Some metals are stabilized by the combination of high temperature acid
digestion and complexation with chloride ions. 70% of antimony can be lost
during a typical open beaker digestion with an oxidizing acid such as nitric
acid. However, using a combination of nitric and hydrochloric acids at an
elevated reaction temperature, the antimony can be oxidized to the Sb(V). It
then readily forms a stable complex anion with chloride ions as described
below {7,8).
Sb (V) + Cl- <===> SbCl,'1
The recovery and stabilization of antimony and silver are illustrated in Table
1.
Removal of Hydrofluoric acid
Hydrofluoric acid is most commonly used in inorganic analysis because it is
one of the few acids that can dissolve silicates, as described in the equation
below:
SiO2 + 6 HF > H2SiF6 + 2 H2O
109
-------
Dissolution by hydrofluoric acid produces primarily soluble fluorides, with
the exception of insoluble or sparingly soluble fluorides of the alkaline earth,
lanthanide, and actinide elements.
Table 1: Stabilization and Recovery of Elements from NIST SRM 2710 with
the addition of HC1.
Element
Antimony
Silver
HNO3 & HF
33.1 ± 2.1
10.6 ±4.5
HNOs, HF
&HC1
39.3 ± 0.9
36.6 ±0.5
Certified
38.4 ± 3.0
35.3 ± 1.5
All analysis reported in ug/g
Once dissolved, many analyses require the removal of all hydrofluoric acid to
prevent damage of the instrumental equipment, or to resolubilize insoluble
fluorides. The fluorosilicic acid can be dissociated by heating the solution
down to fumes in a ventilated system with other acids, such as hydrochloric
acid, nitric acid, or perchloric acid. This procedure is frequently repeated a
second time, for the complete removal of fluoride.
SiF
2 HF
H2SiF6
Care should be taken in the design of a fuming procedure to prevent the
possible volatilization of numerous elements including: As, B, Se, Sb, Hg, Ge,
Cr, Re, Os, and Ru.
Data are presented demonstrating the complete removal of fluoride ions by
fuming off hydrofluoric acid using a microwave evaporation system
(Milestone Corporation) that is beneficial in three ways. The fuming process
results in statistically accurate analytical data (i.e. no loss of analytes through
volatilization or precipitation), resolubilization of some metal fluoride
precipitates, and removal of some ICP-MS interferences.
Another approach to the removal of hydrofluoric acid after digestion is to
complex the fluoride ion with boric acid. The reaction of boric acid with
hydrofluoric acid is a two step process with the slowest step being the
formation of fluoroboric acid (9).
H3BO3 + 3 HF < - > HBF3(OH) + 2 H2O
HBF3(OH) + HF < - > HBF4 + H2O
Cooling the solution after the addition of boric acid will increase the reaction
rate (9). Many knowledgeable analysts suggest using 10 to 50 times
stoichiometric excess boric acid.
110
-------
CONCLUSIONS
The flexibility and robustness of EPA Draft Method 3052 has been
demonstrated for many matrices and many analyte elements. The selection
of decomposition reagents and reaction conditions has been shown to have
dramatic effects on the efficiency of digestion and analysis. With the careful
planning of a digestion protocol, based on the fundamental chemical
reactivity of the acids and matrices, a total decomposition with a wide variety
of matrices can be achieved. Method 3052 has broad applicabilities and
promises to become a widely used method.
ACKNOWLEDGMENTS
The authors gratefully acknowledge Milestone Corporation for its support of
the fundamental research behind EPA Method 3052 and to Mr. Oliver
Fordham, EPA RCRA Inorganic Methods Coordinator, for his leadership and
assistance in the development of EPA method 3052.
REFERENCES
(1) Kingston, H. M.; Walter, P. J. "Comparison of Microwave versus Conventional Dissolution
for Environmental Applications", Spectroscopy 1992, 7,20-27.
(2) Kingston, H. M.; Jassie, L. B. "Microwave Energy for Acid Decomposition at Elevated
Temperatures and Pressures Using Biological and Botanical Samples", Anal. Chem. 1986,
58,2534-2541.
(3) Kingston, H. M.; Jassie, L. B.; "Chapter 6: Monitoring and Predicting Parameters in
Microwave Dissolution" In Introduction to Microwave Sample Preparation: Theory and
Practice; Jassie, L. B., Kingston, H. M., Eds.; ACS: Washington, D.C., 1988, pp 93-154.
(4) "SW-846 EPA Method 3052: Microwave assisted acid digestion of siliceous and
organically based matricies" In Test Methods for Evaluating Solid Waste, 3rd edition, 3rd
update; U.S. EPA: Washington, DC, 1995.
(5) Walter, P. J., "The Development and Validation of Advanced Reaction Control Techniques
for Microwave Sample Preparation", Ph.D. Dissertation, Duquesne University, 1996.
(6) Walter, P. J.; Chalk, S. J.; Kingston, H. M.; "Chapter 2: Overview of Microwave Assisted
Sample Preparation" In Microwave Enhanced Chemistry; Kingston, H. M., Haswell, S.,
Eds.; American Chemical Society: Washington, D.C., 1996.
(7) Kingston, H. M.; Walter, P. J.; Chalk, S. J.; Lorentzen, E. M.; Link, D. D.; "Chapter 3:
Environmental Microwave Sample Preparation: Fundamentals, Methods, and
Applications" In Microwave Enhanced Chemistry; Kingston, H. M., Haswell, S., Eds.;
American Chemical Society: Washington, D.C., 1996.
(8) Hewitt, A. D.; Cragin, J. H. "Comment en "Acid Digestion for Sediments, Sludges, Soils,
and Solid Wastes. A Proposed Alternataive to EPA SW 846 Method 3050"", Environ. Sci.
Technol. 1991,25,985-986.
(9) Sulcek, Z.; Povondra, P. Methods of Decomposition in Inorganic Analysis; CRC Press: Boca
Raton, 1989.
111
-------
23
The Accurate Determination of Species by Speciated Isotope
Dilution Mass Spectrometry: Exemplified by the Evaluation of
Chromium(VI) in Soil
H. M. "Skip" Kingston. Dengwei Huo, Stuart Chalk, and Peter J. Walter
Duquesne University, Department of Chemistry and Biochemistry and the
Environmental Science Program, Pittsburgh, PA 15282
ABSTRACT
A new generally applicable method for the accurate determination of
chemical species in the environment has been developed called Speciated
Isotope Dilution Mass Spectrometry (S-IDMS). The method compensates for
specie transformations that can occur in many of the steps of sample
processing due to unwanted side reactions. It has been specifically developed
to address the problems of accurately quantifying different species in
complicated matrices. It is an excellent diagnostic tool for identifying the
most error prone steps in the measurement, storage, sample preparation and
the sampling process. These unique capabilities will eventually make it
possible to establish standard measurement methods and to develop standard
sampling procedures for speciation.
Preliminary speciation study of the Cr(VI) and Cr(IE) has demonstrated the
unique advantages of this method. These examples demonstrate that up to
53% degradation of Cr(VI) to Cr(HI) can be accurately corrected back to the
original concentration in the sample. In theory it will be possible to corrected
for up to 90% interconversion. These results are used as examples to
demonstrate the S-IDMS concept.
The method is applicable to approximately 80% of the species that must be
measured and extends to oxidation states, organometallics, and molecular
forms of species. Examples of mercury, chromium, tin, selenium, and other
species represent only some of the first applications. A US patent has been
issued for the method of Speciated Isotope Dilution Mass Spectrometry (1). It
discusses how the method can be used as a diagnostic tool for collecting data
that must be legally defensible and definitive with regards to species of
varying toxicity.
INTRODUCTION
Speciation measurements are made for a variety of reasons, including
characterization and evaluation of systems in environmental science,
medicine, biological process monitoring, nutrition, and industrial
troubleshooting. Since the chemistry in these processes is specie specific,
there is a great need to answer these questions at the speciated level. Until
112
-------
now, no definitive diagnostic measurement capability has been available.
Speciated Isotope Dilution Mass Spectrometry (S-IDMS) provides a new
technological advance necessary to make these measurements that are
currently beyond the capability of accurate metrology methods.
Almost every facet of scientific research involves measurement of the
composition of matter. Currently, only bulk measurements of total
elemental concentration and composition can be made accurately and
routinely. It is known that these measurements are not adequate to resolve
the complex questions that are raised for many chemical systems that
undergo different reactions depending on the elemental form. Elements and
molecules react differently in their speciated forms. For example, Cr(III) is a
trace element essential for human health, but Cr(VI) is a poison and
carcinogen to humans and most other animals. The difference in this case is
only the oxidation state of the element. Each of these forms of chromium is
called a species of chromium, and each has unique chemical reactions
associated with it.
Two methods currently used for speciated chromium analysis are
electrochemistry and chromatography. Electrochemistry can distinguish
between the two different forms of chromium in a simple mixture, and
chromatography can separate the mixture before presentation to a species
non-specific detector such as ICP-MS. For complex sample matrices,
chromatography can be used to separate Cr(III) in time from Cr(VI). Because
each specie can react with its surroundings, and even with the separating
agents, detection after a chromatographic separation is only a snapshot of the
specie distribution at that time. Each specie may have reacted with other
sample components, reagents, or been transformed during the storage and
analysis steps. With time resolution, there is no way to determine how much
chromium was actually in each speciated form when the sample was
originally taken.
WHAT IS SPECIATED ISOTOPE DILUTION?
The speciated isotope dilution method is a procedure for the measurement of
elemental species that uses isotope dilution mass spectrometry to determine
the quantity of the species in the sample. The method will permit the
measurement of elemental species in environmental, biological,
pharmaceutical, and industrial samples. It uses the concept of spiking the
sample with one or more separated isotopes that have been chemically
converted into a single specie, equilibrated with the naturally occurring
species, and extracted from the sample material with the naturally occurring
species of interest.
Conventional methods of chromatographic separations are used to separate
the species, and isotope ratios are evaluated for each specie, across the
113
-------
chromatographic peak, to determine the extent of crossover between the
various species prior to and during analysis. It is thus possible to correct for
degradation of the species, or interchange from different species, at each step
in the sampling and analysis protocol. This allows for accurate back
calculation of the original sample concentration(s) of the different chemical
species, something that has previously been impossible. Figure 1 shows a
sample separation for this procedure using two spikes of isotopically enriched
chromium species (5(Cr(ffl) and 5:Cr(VI)).
Figure 1 The Chromatogram of the On-line Separation of Cr(III) and Cr(VI)
): lOOppb
100-] f\ Cr(lll) Ci(Vl): lOOppb
Flow rate: I ml/min
Fluent: 0.06M HNO3
Column :CETAC ANX 4605 Cr
Cr(VI)
25.00 50.00 75.00 100.00
r etent ion t ime ( s)
An Example of Speciated Isotope Dilution
Chromium has several minor stable isotopes described in Table 1 that are
suitable to be transformed into the two important oxidative species Cr(VI)
and Cr(III). By spiking the original sample with separated isotopes in a
speciated form as previously described, the stability and convertibility of the
species can be evaluated and corrected. The isotopically enriched specie will
undergo the reactions of the same "natural" specie, and this alteration will
leave shifted isotopically labeled related species. The natural specie
undergoes the same reactions and is chemically indistinguishable from the
spiked specie. This is a very important additional application of this
technology and no other measurement method can provide this information.
The S-IDMS patent (1) describes a study where evaluations have been
performed without using speciated isotope dilution and would provide an
example where this new technology will provide vital definitive
information. The patent also represents the first time a correction for
transformation has been able to be made and accurate analyses made feasible.
PROCEDURE FOR CHROMIUM SPECIATION IN WATER
Step 1. Spike Preparation
Separated (enriched) isotope spikes for each of two species is prepared for
Cr(ni) and Cr(VI). The isotopically enriched specie spike of 5(Cr for Cr(ni), and
the isotopically enriched specie spike of 5tr for Cr(VI) were used:
114
-------
Table 1. Isotopic Abundance of Chromium
Natural Isotopic Abundance
Cr(Ill) and Cr(VI)
50-4.35%
52-83.79%
53 - 9.50%
54 - 2.36%
* Isotec
Enriched Isotope 5aCr*
Cr(III) spike
50-93.1%
52 - 6.8%
53-0.1%
54 - 0%
Inc. Lot #2691 ** Isotec Inc.
Enriched Isotope 53Cr**
Cr(VI) spike
50-0.03%
52-2.19%
53- 97.7%
54-0.08%
Lot #2692
Step 2. Sample Collection and Spiking
Water is collected from a natural water aquifer. Using the single enriched
specie spike of 5(Cr for €1(111), the sample is spiked with a known quantity of
the spiked (5(Cr, Cr(IQ)) species. The optimal ratio would be 5tr(in) in a
concentration that approaches approximately a 1:1 ratio with the natural
5tr(ni). This permits higher accuracy during the measurement. The same
procedure is applied to the 53Cr spike for Cr(VI)
Step 3. Sample Specie and Spike Specie Equilibration
Equilibrate the sample (natural) with the spikes. For this example we mix
both the natural and enriched material in aqueous form. From this point on
the total Cr(VI) has a isotopic ratio of 5tr(VI) to 5^Cr(VI) in the sample of
approximately 1:1, depending on the amount of Cr(VI) in the original sample.
Step 4. Resolve the Species Temporally or Spatially
For this example, the species were separated using chromatography similar to
that described in Figure 1. This is determined by the chemistry of the species.
Separation in time on-line to an inductively coupled plasma mass
spectrometer (ICP-MS) provides physical resolution of the Cr(IH) and Cr(VI)
fractions in different portions of the chromatogram. Cr(EI) and Cr(VI) are
separated from each other with the equilibrated isotopic distribution based on
any conversion of the spike and the sample. (Note: Time resolution
chromatography is available for these species and has been demonstrated in
the literature.)
Step 5. Isotope Ratio of Each Speciated and Resolved Component
Each sample point on the chromatographic peaks contains the isotopic
distribution of that species allowing replicate measurements of the isotope
ratios for each injection. Isotope ratio measurements of each individual
species isotope resolved component is made separately for Cr(in) and for
Cr(VI). The concentration of the species is determined from isotope dilution
115
-------
calculations. A total concentration for both isotopes can also be performed as
a check.
Step 6. Determination of Specie Conversion
Deconvolurion of each specie in the presence of the other using isotopic ratios
is required at this stage. For example if Cr(VI) is converted to Cr(ni), the
isotope ratio ^r/^r in Cr(III) will show a decrease compared to no reduction
of Cr(VI). If Cr(III) is converted to Cr(VI), then the isotopic ratio of Cr(VI) will
have a lower 5tr/5(Cr ratio compared with no conversion. For more
complicated scenarios where there are conversions in both directions, the
previously described equations can be used to definitively calculate the
concentrations of Cr(in) species and Cr(VI) species in the water, as well as the
amount of interconversion between Cr(III) and Cr(VI) after spiking.
Double-spiking the solutions containing Cr(in) and Cr(VI) with different
isotope enriched spikes is very unique in S-IDMS. If the spikes for different
species have the same isotope enrichment, the crossover among them can
not be determined. In this situation, when IDMS is applied to speciation, the
crossover between the different species is not allowed and the complete
separation of the species from each other must be carried out. Using the
double-spiking technique described here, the crossover between different
species after spiking is permitted. The incomplete separation of the species is
also allowed if the initial concentrations of the species at the time of spiking,
but not the amount of the crossover, are desired because the incompleteness
of the separation can be treated as the interconversion between the species.
The previously described process is being applied and evaluated various
samples such as natural waters and soil samples. The process may be altered
depending on the matrix and each step is being evaluated and altered during
the development stage to optimize the performance, precision, and accuracy
of the method for both matrix types.
RESULTS
An example of some preliminary data showing the interconversion of Cr(DT)
and Cr(VI) is shown Table 1. Isotopically enriched speciated spikes of
chromium were added to NIST SRMs 2108 and 2109 and analyzed by speciated
isotope dilution mass spectrometry (S-IDMS) and isotope dilution mass
spectrometry (IDMS). Even though the samples were very clean (in dilute
nitric acid) some interconversion of the species was seen and identified as is
shown in Table 2.
116
-------
Table 2. Comparison of the Results Obtained from IDMS and Speciated-IDMS
Sample Dates . Speciated-IDMS IDMS
Concentration (ppb)* Conversion (%) Concentration (ppb)**
Cr(III) Cr(VI) (III) to (VI) (VI) to (III) Cr(III) Cr(VI)
Sample 1 Jan 27,96 69.76±0.21 68.8510.20 4.87±0.14 3.57±0.02 72.42±0.21 72.7010.20
lOOppb Jan 30,96 69.23+0.35 69.38+0.18 3.47+0.07 11.92+0.31 78.20+0.36 72.52+0.22
KMnO4 Feb8,96 70.50+0.59 68.54+0.24 2.80+0.08 22.35+0.15 86.74+0.65 71.31+0.27
Sample 2 Jan 27,96 69.87+0.13 68.77+0.25 17.65+0.04 2.95+0.01 72.41+0.13 82.68+0.26
200 ppb Jan 30,96 69.30±0.45 69.63+0.40 14.55+0.84 11.40+0.43 78.8710.46 82.0910.93
KMnO4 Feb8,96 70.6710.25 68.7810.30 12.8210.04 22.0810.20 88.6110.24 81.4810.27
Sample3 Jan 27,96 69.7510.38 68.9810.16 23.7910.21 2.7610.05 72.3210.36 87.6610.20
400ppb Jan 30,96 69.0110.50 69.5610.21 21.6110.15 10.2110.06 78.3410.53 87.7710.10
KMnO4 Feb8,96 70.39+0.32 68.9010.48 17.65+0.21 22.0710.08 89.4110.24 86.34+0.43
True Concentration 69.67 68.63 69.67 68.63
* Interconversion between Cr(III) and Cr(VI) was considered in the calculation
** No interconversion between Cr(III) and Cr(VI) was considered in the calculation
Other data will be presented on actual soils and spiked sand using a modified
EPA proposed method 3060A procedure and microwave extraction. These
additional data support this technique as a successful method for very
difficult and practical speciated environmental samples.
In an attempt to show the power of the S-IDMS method, the method is being
applied to the determination of Cr(VI) in extracts of contaminated soils.
Proposed EPA method 3060A uses a sodium hydroxide/sodium carbonate
leach to selectively extract, and stabilize, Cr(VI) from soils sediments and
sludges.
Preliminary data is shown in Table 3. The extract solutions were spiked with
isotopically enriched Cr(III) and Cr(VI) even though only Cr(VI) is being
determined. Recoveries of chromium(VI) using both hotplate and
microwave heating were statistically indistinguishable for both matrices
indicating that the microwave radiation does not have any additional affect
on the stability of the chromium(VI) species in solution. However, in both
cases the recoveries as determined by the UV-Vis method were not 100%
indicating two points. First, the low recovery of the spiked extract solution
using UV-Vis, indicates interconversion of chromium(VI) to chromium(III)
117
-------
by the analysis procedure (colorimetric method with diphenylcarbazide). In
addition some interconversion of the chromium(VI) is induced by the sand
matrix (probably due to low levels of iron). There is also incomplete recovery
of the chromium(VI) by the SIDMS method for the sand extraction indicating
some chromium(VI) has been lost from solution (e.g. by adsorption onto the
sand particles) before the isotopic spike is added.
Table 3. Results of Chromium Analysis from Spiked Samples
Description
Spiked Sand
Spiked Sand
Spiked Extract
Spiked Extract
Heating
method
Microwave
Hot plate
Microwave
Hot plate
Initial [Cr(VI)]
UV/Vis
867±6
869±6
970 ±7
974 ±7
in extract (ng g"1)
SIDMS
981 ±8
970128
999111
101319
Cr(VI) to Cr(III)
percent conversion
36. 18% 10.76%
40.27% 1 0.29%
14.87% 1 0.13%
13.81% ± 0.17%
* Values listed are the averages from the analysis of three replicates of a sample 1 std dev
The samples were extracted, spiked with isotopic standards, and acidified to 0.06M HNO3.
The samples were analyzed nine days after acidification.
APPLICATIONS OF SPECIATED ISOTOPE DILUTION
The effect of species on chemical reactions point to the need to sample,
separate, and measure different species and determine whether any
interchange occurs during each of the various processes. Also, tracking
species in reactive systems and monitoring their distribution in various
forms requires specific measurement methods. With conventional
measurement techniques, the origin of one specie, and its possible
transformation or distribution between several species, is impossible to
determine.
Another use for this technique is in the chemical processing of
environmental, biological and other samples for speciated components.
Species stability and the precise transformations species undergo are currently
indeterminable by every other method of speciated analysis. It is imperative
that a definitive method of analysis be developed for sample processing or
preparation prior to all forms of analysis. Use of this speciated isotope
dilution technique may fulfill this measurement need.
Currently, almost all speciated data measurements are being discredited in the
legal process due to their reliability. There has not previously been a
diagnostic method that was applicable to provide quality assurance for
speciated measurements such as Cr(VI) and other transformable and highly
reactive species. The method of speciated isotope dilution may permit the
correction for species shifts that have occurred during analysis and also
during other portions of sample handling if they are included in the method
protocol. This could provide both a measurement technique and a diagnostic
118
-------
tool to validate or calibrate other speciation measurement methods for a
variety of different species.
CONCLUSIONS
The preliminary data and the theoretical background both support the
potential of the S-IDMS method. A complete evaluation of this new tool is
necessary to determine where it is appropriate and where it must be altered
for specific applications. New and intermediate species may be found to play
important roles that have not thus far been evaluated for their significance in
specie transformations. Multiple forms of speciated spikes may be required
for some species. However, this is the first time a diagnostic tool has been
available to permit validation of other speciated methods as well as provide
legally defensible data for speciated metrology.
REFERENCES
1. U. S. Patent Number 5,414,259, "Method of Speciated Isotope Dilution
Mass Spectrometry," Granted May 9,1995.
119
-------
24
ON-SITE X-RAY FLUORESCENCE SPECTROSCOPY
FOR THE DETERMINATION OF METALS AND PREDICTING ACID ROCK DRAINAGE
Michael Higgins, Carl Einberger, Anthony Burgess, Ph.D
Colder Associates, Inc., 4104 148th Avenue NE, Redmond, Washington 98052;
Joseph Scheuering, Noranda Minerals Inc., 2501 Catlin, Suite 201, Missoula, Montana 59801
ABSTRACT
On-site measurement (OM) techniques have been recognized as providing equal or better
information compared to a fixed laboratory product. Traditionally, issues regarding the nature
and extent of contamination and the effectiveness of waste removal or other remedial actions
have been addressed using fixed-based laboratory analyses. However, OM techniques have a
number of demonstrated advantages and their use has rapidly gained support. OM techniques
provide analytical data using dedicated project personnel who are on-site and able to respond to
changes as data are generated. Additional sample handling effort and expense, especially at a
remote site, and delays in field activities while waiting for laboratory data are avoided.
The OM used at a copper/cobalt mine is an example of providing cost-effective dedicated
analytical equipment and personnel for a remote site. Analytical support included the design
and implementation of a Spectrace 9000 X-Ray Fluorescence (XRF) laboratory to evaluate waste
rock acid generation or acid rock drainage (ARD) potential. XRF measurements of iron and sulfur
(pyrite minerals) were used to predict waste rock ARD potential, to determine the extent of the
waste rock boundaries, and to provide information for removal decisions in near real-time.
Limited use of off-site analysis of acid:base accounting (ABA) using standard laboratory methods
provided confirmation of decisions as construction activities were occurring.
Copper, cobalt, and arsenic residuals were measured simultaneously with the XRF ARD analyses.
Through the development of site-specific reference samples, accurate and precise XRF
measurements in the ppm (mg/kg) range were determined for samples throughout the site. The
cost efficiencies of OM allowed for a greater density of sampling and analysis than would
otherwise be feasible. This provided more data to characterize the spatial heterogeneity of the
site and increased the confidence of decision-making. High-density sampling and analysis also
provided additional data that were useful in the determination of acceptable decision error
OM XRF analyses were also used to evaluate potential control measures. Both field and
laboratory scale extraction procedures were utilized to delineate waste rock units contributing to
the generation of acid. Short-term leach tests were evaluated to provide information on soluble
and non-extractable metal loads of waste rock, colluvium below waste rock, and sediment
samples. In addition, XRF analyses were used to better understand the nature of copper and
cobalt adsorption and release of mine loadings into receiving waters by determining metals
concentration in creek bed sediments.
XRF correlation with ABA provided a useful, cost-effective, and time-saving method of
identifying waste capable of generating ARD. XRF measurements were also used to determine
120
-------
metals concentrations in sediments and colluvium underlying waste rock, enabling the extent of
impacted areas to be determined and providing near real-time direction of construction activities.
INTRODUCTION
Standard fixed laboratory-based metals and inorganic analyses have some drawbacks that
confine them to off-site fixed labs. First, they require sample dissolution before introduction into
laboratory instruments, this requires additional support equipment for sample preparation.
Second, routine environmental laboratories do not usually have the equipment available to
perform total sulfur analyses. And finally, fixed-based laboratory instruments cannot operate
without well conditioned power, stable temperature and humidity required of critical optical
components and on-board computers often lacking at remediation/construction sites.
OM techniques have a number of advantages over fixed-based laboratory techniques. Chief
among these advantages is in providing dedicated project personnel who are on-site and able to
respond directly to the needs of the project as data is generated and evaluated. Once an action
level is established for site activities, OM techniques are matched to meet these objectives and
used to guide removal actions as they are occurring, enabling contractors to operate more
efficiently. In addition, the cost efficiencies of OM allow for a greater density of sampling and
analysis. This provides additional data to more precisely characterize the spatial heterogeneity of
the site and increased the confidence in the data. High-density sampling and analysis also
provided additional data that were useful in the determination of acceptable decision error. Ten
percent of the field samples (approximately 800 samples) were analyzed by ICP on the
hydrofluoric acid digest and regressed against the XRF results to reveal any relative bias in the
data sets. The relative percent difference (RPD) between the two determinations was calculated
to indicate the average disparity between the analyses.
OM using XRF at the Blackbird Mine site is an example of providing cost-effective dedicated
equipment and personnel for a remote site. The site covers approximately 830 acres of private
patented mining claims held by the Blackbird Mining Company, and 10,000 acres of unpatented
claims previously held by Noranda Minerals Corporation within the Salmon National Forest near
the town of Cobalt, Idaho. The site is located within the Northern Rocky Mountain province and
the topography is characterized by deep stream cut canyons with steep and rocky slopes.
Regional elevations range from 3,000 ft to approximately 9,000 ft. Mining activity has resulted in
construction of approximately 14 miles of underground workings and a 12 acre open pit.
Additionally, the mine site includes graded roads, numerous piles of waste rock, a tailing
impoundment, sedimentation ponds, a water treatment plant, a wastewater sludge pond, and
support structures. Historic mining activity has abo occurred in the area since the late 1800's.
The Blackbird site is drained by two streams: Bucktail Creek to the north, and Meadow Creek to
the south. Both discharge to Panther Creek, a major tributary to the Salmon River.
Prediction, control, and monitoring of waste rock acid generation or acid rock drainage (ARD)
potential is of major importance when considering the environmental aspects at mining sites (1).
XRF measurements of iron and sulfur (pyrite minerals) were used to predict waste rock ARD
potential, to determine the extent of the waste rock boundaries, and to provide information for
removal decisions. Copper, cobalt, and arsenic concentrations were measured simultaneously
with the XRF for iron and sulfur analyses. Residual metals were also measured under a variety of
121
-------
circumstances independent of ARD evaluation, in these cases sulfur was not included in the
analysis in order to reduce sample excitation time and increase analytical throughput.
MATERIALS AND METHODS
Analytical support included the use of Spectrace 9000 field portable X-Ray Fluorescence (XRF)
instrument. The Spectrace 9000 uses three isotopes, ^Fe, l09Cd, and ^Am for the production of
primary x-rays for sample excitation and a high resolution mercuric iodide X-ray detector.
Quantitation was based on a fundamental parameters calibration model.
The majority of excitation time (400 seconds) was due to the excitation of sulfur species using the
^e source. The 10*Cd provided excitation (200 seconds) for the analysis of iron, cobalt, copper,
and arsenic. Total processing time, including the 600 second excitation, was slightly more than 10
minutes. Data processing accomplished with the on-board microprocessor, results and spectral
storage to a PC were accomplished while loading the next sample for analysis.
Prior to initiating and during the OM activities, Blackbird mine reference samples (BMRS) were
selected to provide the appropriate range of concentrations in various matrices that were
expected to be encountered during field work carried out at the site. The three major categories
of sample matrices: waste rock, colluvium below waste rock, and debris flow and alluvial
deposition, were evaluated. The references samples were used to validate the calibration model
initially during project planning and on an on-going basis during field activities.
i Element
llron
rCobaTt"
| Copper
1 Arsenic
1 Sulfur
BMRS 0920
50,100±3,478
T04±79 -
857±195
312±108
ND
BMRS 1950
68300±6320
"l96±822
1,120±105
1,530±131
} 4,570±115
BMRS 1960
180,000± 11,700
"l57±5"78""~
4,710±140
2,720±191
32,700±115
BMRS 0920, 1950 and 1960 were selected based on the concentration of target analytes which
bracket expected soil, sediment, debris, and waste material concentrations. BMRS 1950 and 1960
were prepared from bulk samples collected from test pits from the site. BMRS 0920 was prepared
by compositing surface debris flow collected from an area adjacent to the toe of a waste rock
dump. All BMRS's were pulverized and sieved to 75 mM (200 mesh), HF digested, and analyzed
in triplicate by ICP by the off-site laboratory and confirmed using an additional off-site laboratory
independent of the project. All BMRS samples were within ±16% difference from the reference
values for XRF target analytes as measured throughout the duration of the project.
122
-------
SCREENING VS CONFIRMATORY LEVELS OF ANALYSIS.
Two sample preparation methods were used for the XRF analysis and performed in selected
areas of the site: screening-level sampling including ABA screening, and confirmatory composite
sampling.
The screening-level sampling and analysis were intended to provide rapid-turnaround,
qualitative, inorganic analysis. The screening-level protocol included collection of grab samples.
Sample representativeness was estimated by comparing a large number of individual sample
measurements, in contrast to the more rigorous composite sampling, where reduction and
homogenization of a large composite was performed using splitting techniques, followed by sieve
fractional analysis, pulverization, and additional mixing. The confirmatory composite sampling
protocol was developed for specific site areas where action level criteria were established and the
need for defining the precision and accuracy limits were warranted (2).
SAMPLE PREPARATION
The on-site lab received samples collected specifically for XRF screening analysis consisting of
minus 2-inch waste rock or other solid materials/soil stored in 4 oz HOPE sample jars or other
suitable polymeric containers. Separate samples were also collected as field duplicates at a
frequency of 5% of the total number of field samples.
An aliquot (approximately 10*100 grams depending on particle size distribution), typically
consisting of both fine and coarse material, was removed from sample jars with a disposable
plastic spoon and dried to less than 2% moisture. Where sample crushing was necessary (i.e.
>20% of the sample over 10 mesh), samples were manually crushed using a hand sledge, until an
adequate amount of minus 10 mesh sample material was obtained for analysis. All samples
requiring size reduction were crushed prior to sieving. Samples collected from boreholes (drilled
using air rotary) or from sediments generally did not require the crushing step. Pulverization was
not required by the XRF screening-level protocol. However, a number of samples collected for
evaluation sample preparation protocol were pulverized using a Bico-Braun direct motor-driven
pulverizer. The dried samples were sieved to minus 10 mesh and the over-sized fraction
discarded. Differences in metals concentration between the over-size and sieved fractions were
compared on a periodic basis and were found to be insignificant when analysis was performed on
samples collected from waste areas. XRF sample cups were filled approximately 3/4 full, tapped
to distribute fine particles to the window side of the XRF cups and compacted by tapping with
the window side down prior to analysis.
PRE-OPERATIONAL CHECKS
Element X-ray response checks were performed and recorded each day. A sample of pure iron
for XRF-metals and pure titanium for XRF-sulfur was used for these analyses. The 109Cd source
provided excitation for the XRF-metals (iron, arsenic, cobalt, and copper). To verify response of
this source, the 109Cd spectrum was examined after excitation of a pure iron (Fe) sample for 50
123
-------
seconds. A relative response greater than 0.95 for iron was required. Iron spectral interference
with cobalt and manganese values must be less than 0.006 relative to iron. Similar X-ray response
checks were performed for XRF-sulfur which utilizes the ^Fe source for excitation. The relative
response for titanium in a pure titanium (11) sample measured for 50 seconds must be greater
than 0.95. No spectral interference is present in the Ti spectrum and adjacent energies were not
evaluated.
Energy calibration was checked by examining the 109Cd spectrum. The K alpha line associated
with iron was verified and recorded at 6.4 ± 0.02 keV before proceeding with sample analysis.
After verifying the energy calibration, spectrum resolution was confirmed by verifying that the
number of counts at the centroid of the iron K alpha line (6.4 keV) were more than twice the
number of counts at ± 0.3 keV (resolution at half height). These results were recorded each day.
Goals for energy calibration were met for all analyses. Energy calibration during operation was
maintained automatically by examining X-ray back-scatter in between each analysis.
An analytical background check was performed at least once each day to verify both the
spectrometer operation and the condition of the probe window. A Teflon® (blank) sample was
placed within the sample positioning ring and contaminant-free conditions verified before
proceeding with field sample analyses. Goals for contaminant free blanks were met for all
analyses.
The analytical computations of the Specrrace 9000 are based on a fundamental parameter (FP)
calibration algorithm. When using FP, the highest degree of accuracy is obtained when the entire
sample composition is known. Since many of the elements in the site samples could not be
measured with the instrumentation employed, it was necessary to make assumptions about the
elemental composition of the sample. As a result, certain sample types (i.e. waste rock, overbank,
colluvium,) exhibit minor amounts of bias (deviation from a known reference value) due to
variations in the unmeasurable element balance in the sample. The application used at the site
assumed that use of a SiO2 (silica) balance, accounting for the majority of the unmeasurable
balance in field samples, provided sufficiently accurate analyses without a need for further
adjustment. The accuracy of the measurements could be fine-tuned by adjusting the slope and
off-set values for a given target analyte when reference samples have been characterized for the
site. Calibration adjustments were not considered trivial, but were performed based on
knowledge of the site material being analyzed. Adjustments to slope and off-set were performed
only when: 1) appropriate references samples were available and have been fully evaluated, and
2) the samples collected for analysis had characteristics similar to the reference material. At all
times, 10% of all samples analyzed by XRF were confirmed using ICP. This was the basis for
quality control of reported XRF results and provides the quality assurance that the adjustments
have been applied correctly.
The method minimum detection limit was obtained by measuring the appropriate BMRS
representative of the matrix and level or levels at which decision criteria had been established.
Statistical analysis of XRF BMRS data were performed and a method detection limit was
calculated. The method reporting limit was defined as the standard deviation (rounded up to the
nearest whole number) multiplied by three and reported to two significant figures. The
calculated method detection limits were determined to be: 3,000 mg/kg for iron, 1,500 mg/kg for
cobalt, 300 mg/kg for copper, and 150 mg/kg for arsenic. The detection limits were applied to all
124
-------
measurements reported. The method reporting limit for sulfur was evaluated and established in
conjunction with the XRF instrument manufacturer and determined to be less than 2,000 mg/kg.
PRECISION AND ACCURACY
Data precision and accuracy were evaluated to determine the usability of the data. The primary
concern in this evaluation was elimination or minimization of the decision error resulting from
false negative results; that is, the determination that an area was non-impacted when in actuality
it was. Evaluation of all target elements were considered in this evaluation. However, cobalt was
not detected in any of the duplicate measurements, and sulfur was not quantitatively measured
for the majority of field samples due to the extended measurement time required for this analyte.
Therefore, cobalt and sulfur were not included in the determination of analytical precision.
Precision was measured by evaluating field and laboratory duplicates and BMRS data. Overall
field and laboratory analytical precision were evaluated by comparing the relative percent
difference (RPD) between the field or laboratory sample and the respective field sample. Field
duplicate samples were collected for the majority of test pit locations in order to assess sample
homogeneity within the testing location. To evaluate field sample variation at the specific field
sample location, a field duplicate sample was collected from opposing location, either top and
bottom or opposite sides of the test pit. Borehole samples were similarly collected using the top
and bottom portion of a core sample or a first and second sample of air rotary cuttings.
Combined laboratory (measurement) and field (collection) precision were measured. Duplicate
recoveries for these samples demonstrated an overall average RPD <_ 45%. RPDs were generally
less than 35% when analytes were detected near the proposed level of concern. Repetitive
analysis results of the appropriate BMRS's were evaluated for percent relative standard deviation
(RSD %) to assess analytical precision. The RSD should be within ±40% for the data to be
considered adequately precise. All calibration reference samples were within this limit.
Accuracy was evaluated by performing confirmation analysis on the same samples used for field
XRF analysis. Random samples were selected and submitted for HF digestion and ICP analysis
for iron, cobalt, copper, and arsenic at a frequency of at least 10% of the number of field samples.
Sulfur was analyzed off-site using a LECO sulfur analyzer requiring no additional sample
preparation. Analyte concentrations, as determined from field measurements, ranged from non-
detect to one or two orders of magnitude above the MDL for most analytes. Confirmation
samples were submitted throughout the field program as field data were generated and
evaluated. To evaluate accuracy, the coefficient of determination (r2), defined as the amount of
variation in the dependent variable (XRF analysis) that can be attributed to the independent
variable (ICP analysis), was used. Using the mining model developed for the project, the
coefficient of determination was greater than 0.8 (r >^ 0.8) for all sample results and r2 >0.9 was
achieved when concentrations of target metals were near the proposed level of concern. The
initial sulfur application assumed a high lead content because it was developed using the NIST
SRM 2710 before a suitable site reference sample was available. The NIST standard contained the
appropriate concentration of sulfur (2,400 mg/kg), but because of the elevated lead content (5,500
mg/kg), the initial application model generated assigned an X-ray region for the detection of
sulfur which was too narrow. This error affected sulfur accuracy and precision, and was only
useful for determining the presence or absence of sulfur. Subsequent calibration models were
designed to be used in the absence of lead, as is the case at the site. This allowed for a broader
125
-------
sulfur window that vastly unproved XRF measurement statistics. The use of this new calibration
model is currently being evaluated and the associated sulfur accuracy and precision has improved
dramatically compared to the initial results.
ACID-BASE ACCOUNTING (ABA) PREDICTION
Acid-base accounting (ABA) is an accepted criterion for evaluating the potential of mine waste
rock to create acid drainage and associated mobilization of metals (1). However, ABA testing
cannot be efficiently conducted on-site without extensive laboratory facilities. Therefore, XRF
analysis was employed for rapid-turnaround estimation of metals concentrations to provide an
indication of ABA potential and the amount of copper, cobalt, and arsenic available for potential
release. Detection of high levels of iron > 100,000 mg/kg and total sulfur > 2,000 mg/kg appeared
to be indicative of acid-generation waste rock, most likely related to the presence of pyrite (FeSJ.
Location
TP-95-15
TP-95-57
TP-95-7
TP-95-15
TP-95-7
TP-95-25
TP-95-35
TP-95-7
TP-95-43
TP-95-15 FD
TP-95-35
FP-95-13
TP-95-13
FP-95-58
TP-9S43
TP-95-4
TP-95-57
TP-95-43
TP-95-43 FD
TP-95-5
TP-95-50
TP-95-13
TP-95-9
TP-95-6
OMXRF
|ron
(mg/Kg)
165,500
154,400
147,900
141,440
136,880
136^30
127370
124,000
123,220
119^90
117350
112360
106,150
104,720
103,970
77,800
70,980
68,250
67,550
67,050
66350
63,260
61,400
56,760
Arsenic
(mg/Kg)
6,138
6,422
4,200
5,943
4,055
3,015
1,982
1336
783
3,959
1359
603
1,692
1,760
1,762
1318
212
486
430
1,095
170
404
645
92
Cobalt
(mg/Kg)
ND
ND
ND
ND
ND
1920
ND
1560
ND
ND
ND
ND
ND
ND
ND
5740
ND
ND
ND
ND
ND
ND
ND
ND
Copper
(rag/Kg)
1,650
3,600
2,116
958
1,633
855
1,129
1,778
371
832
287
250
400
1,624
565
7,980
863
682
608
513
528
1,537
440
298
Sulfur
(mg/Kg)
8,600
2,400
9300
7,100
8,200
2,200
4300
6,400
7,200
5,200
7,700
2,200
4,500
4300
3,600
6,600
2,040
2,060
1,190
6,200
4,270
1,260
1,660
5,030
OFF-SITE LAB
Pyr Sulfur
(mg/Kg)
9,600
2,600
9,200
15,400
5,000
43,600
10300
7,000
3,800
18,000
7,600
980
3,900
3,500
23,400
100
400
3,900
A/B
Potential
-21.1
3.4
-18.4
-38.4
-42
-127.4
-26.9
-12.4
-02
-48.4
-21.5
-12.9
2.0
4.6
-60.9
15.6
12.8
63
63
-53
14.4
103
15.8
10.8
126
-------
TP-95-60
1P-95-11
TP-95-53
53350
50,530
49,280
137
2,533
104
ND
5770
ND
242
3,116
207
2,000
7,500
3,070
5,800
100
10.4
-14.4
12.8
SUMMARY
OM XRF instrumental and sample preparation techniques have proved to be a very useful tool
especially in a remote site. XRF versus fixed lab techniques were not statistically different,
especially considering the variation due to sample collection. The cost efficiencies of XRF
allowed for implementation of high-density sampling versus low density sampling which overall
achieved a higher degree of sample representativeness providing increased confidence in
decision making. The strong positive correlation between XRF iron and sulfur and acid
generating potential provides a predictive tool which can be used for remedial activity decisions
in near real time.
ACKNOWLEDGMENT
The authors would like to thank Peter Berry and Todd Rhea for their many useful discussions
and efforts helping to develop the calibration model used for this project.
REFERENCES
(1) SEPS, April 1992, "Mine Rock Guidelines Design and Control of Drainage Water
Quality" Prepared by SRK, Vancouver, B.C., for Saskatchewan Environment and Pubic
Safety, Mines Pollution Control Branch, Prince Albert, Saskatchewan.
(2) USEPA, April 1990 "Quality Assurance Quality Control Guidance for Removal Activities"
EPA 540 G-90/004, OSWER Directive 9360.4-01 U.S. Environmental Protection Agency,
Office of Emergency and Remedial Response, Washington DC ,
NOTICE The views expressed are the authors and do not necessarily reflect the views of the
Blackbird Mine Site Group.
127
-------
25
Extraction of Anions from Solid Phase Samples for Capillary Ion Electrophoresis
Roy-Keith Smith1, David Roth2, Jim Krol3, and Joe Romano4
The analysis of inorganic anions in solid phase samples has been addressed in the US
Environmental Protection Agency's National Sewage Sludge Survey5 and in the context of
analysis of agricultural soils6. The EPA testing needs were addressed as one line sample
preparation modifications to already existing traditional wet chemical methods for
determination of the specific anions total phosphate, chloride, sulfate, nitrite and nitrate.
These for the most part these involved mixing a portion of the solid with reagent grade water,
agitating for up to 24 hours, then filtering and treating the filtrate as a normal aqueous sample.
A number of procedures are available in a methods manual used by agricultural soils testing
laboratories. These involve extraction of the sample with either reagent grade water alone or
reagent grade water with various additives such as calcium sulfate, ammonia, and/or
diethylenetriaminepentaacetic acid (DTPA).
As part of an effort to develop methods for the analysis of inorganic anions using
capillary electrophoresis, attention was directed toward solid phase samples. The problem was
formulated as first evaluating known procedures with an eye toward capillary electrophoresis,
examining possible alternative sample preparations, then generating precision and accuracy
data for a suitable procedure.
Reagents, materials and apparatus
Reagent grade water was prepared using a cartridge purification system from
Continental Water consisting of 2 mixed bed ion-exchange resin cartridges, a series of two
activated charcoal cartridges and finally a 0.45 um filter cartridge.
Class A volumetric glassware was used for all liquid manipulations. The analytical
balance calibration was checked daily with a set of Class S calibration standards.
Anion standards were prepared as follows:
1 Analytical Services, Inc. 110 Technology Parkway, Norcross, GA 30092.
- Analytical Services, Inc. 110 Technology Parkway, Norcross, GA 30092.
3 Waters Corporation. 34 Maple Street, Milford, MA 01757.
4 Waters Corporation, 34 Maple Street, Milford, MA 01757.
5 Analytical Methods for the National Sewage Sludge Survey, USEPA WH-585, September 1990; POTW
Sludge Sampling and Analysis Guidance Document, USEPA August, 1989.
6 Handbook on Reference Methods for Soil Analysis, Soil and Plant Analysis Council, Inc., 1992; Wisconsin
Procedures for Soil Testing, Plant Analysis and Feed A Forage Analysis, Department of Soil Science, College of
Agricultural and Life Sciences, University of Wisconsin-Extension-Madison, 1987; Soil Sampling and Methods
of Analysis, M.R. Carter (Ed), Canadian Society of Soil Science, 1993, CRC Press, Boca Raton FL; Methods of
Soil Analysis: Part 2 - Chemical and Microbiological Poperties, 2nd Edition, American Society of A.gronomy
Soil Science Society of America, Madison WI1982
128
-------
Salt
NaBr
NaCl
NaF
NaNQj
NaNOa
Na2HPO4
N32SO4
CAS
7647-15-6
7647-14-5
7681-49-4
7631-99-4
7632-00-0
7558-79^
7757-82-6
Amount g/L
1.2876
1.6485
2.2102
1.3707*
1.4998*
1.50
1.48
Anion stock standard solution, 1000 mg/L: Prepare a series of anion standard solutions
by weighing the indicated amount of salt (obtained from Aldrich Chemical Company), dried to
a constant weight at 105°C, to 1000 mL. Store in plastic bottles (HOPE) in a refrigerator.
Anion
Bromide
Chloride
Fluoride
Nitrate
Nitrite
Ortho-phosphate
Sulfate
# This gives a 1000 mg/L solution as nitrate, 225.8 mg/L as nitrate-N.
* Do not oven dry, but dry to a constant weight in a desiccator over phosphorus pentoxide. This gives a
1000 mg/L solution as nitrite, 304.3 mg/L as nitrite-N.
Mixed stock standard: Add 5.00 mL of the chloride, sulfate, nitrate and phosphate
anion stock standard and 2.SO mL of the bromide, nitrite and fluoride anion stock solutions to
a 100 mL Class A volumetric then dilute to volume to give 50 mg/L of the first 4 anions and
25 mg/L of the bromide, nitrite and fluoride anions.. Store for up to one month in a HDPE
container in the refrigerator.
Mixed working standards: Prepare a series of 5 different concentration standards by
diluting 1.00, 10.00, 20.00 and 50 mL of the mixed stock standard to 100 mL. The mixed
stock standard is the fifth standard. This results in calibration standards with concentrations of
0.50, 5.00, 10.0, 25.0 and 50.0 mg/L for chloride, sulfate, nitrate and phosphate, and 0.25,
2.50,5.00,12.5 and 25.0 mg/L for the bromide, nitrite and fluoride. Prepare these standards
weekly and store in the refrigerator in HDPE bottles. A fresh 5-point calibration was prepared
daily and checked every 10 samples.
The 50 mg/L mixed matrix spiking solution was prepared by adding 5.00 mL of each
stock anion standard to a 100 mL volumetric flask and then diluting to volume. A 1:10
dilution with extraction solution gave a 5.00 mg/L matrix spike.
Diethylenetriaminepentaacetic acid (CAS 67-43-6, Aldrich Chemical Company
28,556-0) was prepared by adding 0.500 g to a 100 mL volumetric flask containing about 50
mL water, adding 3.00 mL of 1.00 M NaOH solution, then diluting to volume. A 1:10 dilution
of mis stock solution gave the 0.050% extraction solution.
Ammonium hydroxide (CAS 1336-21-6, Fisher Scientific, 28-30%) was diluted 1:10
to give a 3% solution, then diluted 1:10 again to give a 0.3% extraction solution.
129
-------
Chromate electrolyte solution7. 4 mM chromate and 3.0 mM TTAOH; dilute 20 mL of
chromate concentrate and 15 mL TTAOH concentrate with 400 mL water. Add 5 mL CHES
buffer which gives a pH of 9.0. Dilute to 500 mL. Filter and vacuum de-gas daily. Prepare
fresh weekly and store capped at room temperature. Chromate concentrate (100 mM):
Dissolve 23.41 g sodium chromate tetrahydrate8 (CAS 10034-82-9) Mwt 234.04 in 500 mL
water, then dilute to 1 L. This concentrate is stable for 1 year when stored in a capped plastic
or glass container at room temperature. Electroosmotic flow modifier concentrate: Dissolve
3.364 g tetradecyltrimethyl ammonium bromide9 (CAS 1119-97-7) Mwt 336.4 (TTAB) in 50
mL reagent water, dilute to 100 mL, giving a 100 mM solution. Immediately prior to use pass
the required volume through an ion-exchange cartridge10 directly into the 500 mL volumetric
flask used to prepare the working electrolyte, to exhange bromide for hydroxide ion and
generate TTAOH. CHES Buffer (100 mM) Dissolve 20.73 g 2-(N-cyclohexylamino)-
ethanesulfonic acid' » (CAS 103-47-9), Mwt 207.3 (CHES) in 500 mL water, then dilute to 1 L.
This concentrate is stable for up to 1 year when stored in a capped glass or plastic container
The following solid samples were dried at 104°C, then ground and sieved through a 1.0
mm double aluminum screen to prepare a homogeneous sample for testing:
Type Source
Sandy silt Douglasville, Georgia, 1.0 meter subsurface ("Georgia Red Clay")
Silty sand Atlanta, Georgia
Sediment Red water pond sediment, Sandusky Ohio
Clay San Joaquin NIST Standard Reference Material 2709
Clay Montana Soil NIST Standard Reference Material 2711
Wastewater sludge samples were obtained from Greensboro, Georgia (1.4% solids) and
Manchester, Georgia (5.6% solids).
Syringe filters of 0.45 urn pore size (Acrodisk Ion Chromatography, Gelman catalog #
P/N 4585) and a 10 mL gas tight syringe with Teflon® tipped plunger (Hamilton catalog #
1010) were used to filter samples prior to analysis. Syringe and filter were rinsed with 5 mL
water and 1 mL sample prior to collecting the solution for testing in a 0.6 mL polypropylene
capped centrifuge tube.
7 Available pre-prepared from Waters, Com
8 Catalog f21,662-3, Aldrich Chemical Company, 1001 West Saint Paul Ave, Milwaukee WI53233,1-800-558-
9160.
9 Catalog #74762, Sigma Chemical Company, P.O. Box 14508, St Loius MO 63178-9916, 1-800-325-3010.
This solution is available pre-prepared from Waters Corporation as a 100 mM solution of TTAOH.
10 Available from Alltech Corporation part number 30254.
1' Available from Sigma Chemical Company, catalog number C2885.
130
-------
Cavitator ™ Ultrasonic Cleaner obtained from Mettler Electronics, Inc., Model ME
4.6,115 v, 2 amp, 85 watt bath.
Capillary ion electrophoresis was performed on a Waters Capillar Ion Analyzer with
Mellinium software. A 60 cm, 75 um id fused silica capillary column was used with 30
second hydrostatic sample introduction and 15 kV potential with UV detection. Electrolyte
was replaced between each sample.
Methods and Results
An Instrument Dection Limit and a Method Detection Limit study following EPA
protocols (40 CFR 136, Appendix B) on reagent water were conducted. The results are
presented in Table 1.
Reagent blanks and matrix spiked reagent blanks were prepared for each extraction
solution and treated as samples to test for process contamination. Other than substantial
amounts of carbonate in the ammonia reagents and some formate, no laboratory contamination
was found. The results for these blanks and blank matrix spikes are presented in Table 2.
Samples were weighed (1.00 ± 0.02 g) into 20 mL polyethylene scintillation vials with
polyethylene cone-shaped cap liner (Kimble catalog number 58515), then 10.00 mL extraction
solution with or without matrix spike was added. The treatments used were reagent water,
reagent water plus matrix spike, 0.3% ammonia, 0.3% ammonia plus matrix spike, 0.050%
DTPA, 0.050% DTPA plus matrix spike and 0.3% ammonia plus 0.050% DTP A. The
samples were capped and irradiated in the ultrasonic bath containing about 1 cm water for 1
hour. After filtration the sample extracts were analyzed by CE.
Table 1. Instrument Detection Limit (IDL) and Method Detection Limit (MDL) for anions.
Anion IDL (mg/L) MDL (mg/L) MDL spike level (mg/L)
Chloride 0.023 0.046 0.100
Bromide 0.039 0.046 0.100
Nitrite 0.025 0.072 0.200
Sulfate 0.036 0.032 0.100
Nitrate 0.084 0.084 0.200
Fluoride 0.011 0.020 0.100
Phosphate 0.042 0.041 0.100
131
-------
Table 2. Anion blank and blank MS recoveries for a 5.0 mg/L spike into empty sample vials
Reagent water
Anion
Chloride
Bromide
Nitrite
Sulfate
Nitrate
Fluoride
Phosphate
Blank
nd
nd
nd
nd
nd
nd
nd
MS
97
102
101
101
99
115
102
NH3
Blank
nd
nd
nd
nd
nd
nd
nd
MS
87
105
97
94
103
109
102
DTPA
Blank
nd
nd
nd
nd
nd
nd
nd
MS
97
103
101
98
102
111
110
NH3-DTPA
Blank
nd
nd
nd
nd
nd
nd
nd
Results for the extraction and analysis of the dried soil samples are presented in Tables
3 through 9 for the calibrated anions. The values are corrected for the amount of sample
weighed into the vial. Matrix spike results are expressed in terms of percent recovery,
corrected for background target analyte in the sample. An 'nd' is used to indicate that the
target analyte was not detected in the sample.
Table 3. Chloride results in mg/kg and percent recoveries of SO mg/kg matrix spike
Reagent water
NH3
Soil
Sandy silt
Silty sand
ClaySJ
Clay Mon
Sediment
Sample
1.7
1.2
72
24
2.9
MS
97
97
94
98
93
Sample
nd
nd
70
20
1.4
MS
74
89
98
109
85
Samp
nd
nd
71
21
23
DTPA
le MS
98
102
116
105
53
NH3-DTPA
Sample
nd
nd
74
17
2.1
132
-------
Table 4. Bromide results in mg/kg and percent recoveries of 50 mg/kg matrix spike.
Soil
Sandy silt
Silty sand
ClaySJ
Clay Mon
Sediment
Table 5.
Soil
Sandy silt
Silty sand
ClaySJ
Clay Mon
Sediment
Table 6.
Reagent
Sample
nd
nd
nd
nd
nd
Nitrite results
Reagent
Sample
nd
nd
55
11
nd
Sulfate results
water
MS
98
102
103
98
88
in mg/kg
water
MS
98
98
120
105
89
in mg/kg
Reagent water
Soil
Sandy silt
Silty sand
ClaySJ
Clay Mon
Sediment
Sample
58.6
62
240
58
809
MS
165
196
42
81
111
NH3
Sample
nd
nd
nd
nd
nd
MS
82
101
101
105
87
and percent recoveries
NH3
Sample
nd
nd
nd
nd
nd
MS
81
95
99
95
92
and percent recoveries
NH3
Sample
481
564
190
56
864
MS
0
62
71
99
10
DTPA
Sample
nd
nd
nd
nd
nd
MS
102
101
104
91
100
NH3-DTPA
Sample
nd
nd
nd
nd
nd
of 50 mg/kg matrix spike.
DTPA
Sample
nd
nd
57
7.7
nd
MS
95
98
93
99
f 97
NH3-DTPA
Sample
nd
nd
nd
nd
nd
of 50 mg/kg matrix spike.
DTPA
Sample
258
366
229
66
775
MS
36
84
88
74
232
NH3-DTPA
Sample
493
548
202
51
825
133
-------
Table 7. Nitrate results in mg/kg and percent recoveries of 50 mg/kg matrix spike
Reagent water NH3 DTPA NH3-DTPA
Soil
Sandy silt
Silty sand
ClaySJ
Clay MOD
Sediment
Sample
nd
nd
212
14
187
MS
98
99
28
62
108
Sample
10.2
nd
288
30
210
MS
50
88
70
94
67
Sample
nd
nd
208
14
202
MS
99
98
121
62
70
Sample
9.3
nd
311
27
208
Table 8. Fluoride results in mg/kg and percent recoveries of 50 mg/kg matrix spike
Reagent water NH3 DTPA NH3-DTPA
Soil
Sandy silt
Silty sand
ClaySJ
Clay MOD
Sediment
Sample
nd
nd
4.4
3.2
nd
MS
20
5
55
67
102
Sample
15.7
30
6.0
6.7
nd
MS
76
83
63
86
101
Sample
OJ
nd
1.1
2.6
nd
MS
40
26
52
58
97
Sample
14.3
26
7.1
6.5
nd
Table 9. Phosphate results in mg/kg and percent recoveries of 50 mg/kg matrix spike
Reagent water NH3 DTPA NH3-DTPA
Soil Sample MS Sample MS Sample MS Sample
Sandy silt
Silty sand
ClaySJ
Clay Mon
Sediment
nd
nd
nd
3.8
nd
1
5
25
67
54
nd
nd
nd
nd
nd
22
43
41
50
143
nd
nd
nd
15.2
nd
13
5
42
49
109
nd
nd
nd
1.4
38
The San Joaquin SRM and a matrix spike were analyzed 7 times with the 0.05% DTPA
extraction solution to determine repeatability of the extraction/analysis procedure. The results
are presented in Table 10.
134
-------
Table 10. Analysis of 7 replicates of the San Joaquin SRM and 50 mg/kg matrix spike.
Anion
Chloride
Bromide
Nitrite
Sulfate
Nitrate
Fluoride
Phosphate
Sample Mean
67
nd
nd
187
282
1.8
nd
%RSD
2.0
-
-
5.7
2.4
18
-
MS Mean %R
93
97
90
94
90
56
37
%RSD
2.5
1.7
3.2
3.7
3.0
6.5
28
Duplicate samples of approximately 5 g of the wastewater sludges were weighed into
the 20 mL scintillation vials. One of the duplicates was mixed with 10.00 mL of 0.050%
DTPA solution, while the other was mixed with 10.00 mL of 0.050% DTPA containing 5.00
mg/L matrix spike. The samples were irradiated in the ultrasonic bath for 60 min, then filtered
and analyzed by CE. High levels of organic acids and carbonate in the sludges interferred
with the fluoride and phosphorus responses. The results are presented in Table 11.
Table 11. Anion results for sludge samples in mg/kg wet weight and percent recoveries of
matrix spiked sludges
1.4 % sludge 5.6% sludge
anion Sample mg/kg MS % Sample mg/kg MS %
Chloride 271 254# 14.4 63
Bromide 069 93 nd 106
Nitrite nd 0 nd 74
Sulfate 7.2 109 33 o
Nitrate 0.94 50 nd 32
Fluoride int int int int
Phosphate int int int bt
i Sample background about 20 times matrix spike amount.
Discussion
A number of recipes which have been used in the past for extraction of anions from
soils have relied upon additives such as calcium sulfate, ammonium fluoride, hydrochloric
acid, silver sulfate and phosphate buffers to assist the solubilization of the analytes. None of
135
-------
these were considered as acceptable procedures for an anion screening method due to addition
of one or more target analytes to the sample.
A preliminary set of 4 samples was extracted with reagent water, 0.050% EDTA and
0.050% DTPA with 15 minute ultrasonic extraction and 90 minute heating in a water bath (60-
80°C). Some nitrite was detected in one sample which was extracted using ultrasound, while
the nitrite was missing from the heated sample. Otherwise no substantial differences in the
heating and ultrasonic treatments were noted. Heating the samples was not considered fiirther.
EDTA was found to migrate close to the carbonate peak and contributed to
electropherogram distortion, while the migration of DTPA was considerably retarded and
analyte anion migration times were more stable. As no differences were obvious in the
extraction efficiencies of the two chelating agents, use of EDTA was abandoned and further
experiments concentrated on DTPA.
In another preliminary evaluation, total phosphorus analysis using the persulfate
digestion procedure was attempted using CE. The amount of sulfate in the digestate
completely masked the electropherogram and further attempts along these directions were
abandoned.
Ammonia was evaluated as an aid in digesting the samples. The ammonia used
contained substantial carbonate which presented interference problems due to migration time
changes in the fluoride and phosphate signals.
The results obtained from comparative analysis of the five soil samples indicated that
the different extraction solutions exhibited different strengths and weaknesses for the
individual anions.
Chloride
The four extraction solutions exhibited similar results for four of the five test soils,
with acceptable recoveries of the matrix spikes (74-116%). The redwater pond sediment was
the lone outlier, as the DTPA extraction pulled about 10 times more chloride out of the sample
than did the water, ammonia or ammonia-DTPA solutions. Because of this single sample,
DTPA would be a preferred extraction solution for general use.
Bromide
None of the test soils exhibited any background bromide, and overall marix spike
recoveries were excellent for all four test solutions.
Nitrite
136
-------
Similar sample results and good to excellent matrix spike recoveries were obtained for
the water and DTPA extractions. The DTPA extractions exhibited a tighter range of
recoveries (93-99%) than did the water (89-120%). Even though good to excellent (81-99%)
matrix spike recoveries were obtained from all samples with the ammonia solution, no
background nitrite was recovered from the two samples which exhibited nitrite by the water or
DTPA extractions.
Sulfate
Ammonia and DTPA pulled substantially more sulfate out of the silt containing
samples than did the water. Ammonia and the ammonia-DTPA combination are noted to
performed as well or better than DTPA by itself for these samples. The high levels of sulfate
in these samples (greater than five times the matrix spike) make the interpretation of the spike
recoveries difficult. However, the greater than 100% spike recovery of the water extracted
samples suggests that even the modest amounts of anions in the spike solution can serve to
displace sulfate from the silt matrix. Water by itself is shown be be a poor choice for sulfate
isolation.
Nitrate
The ammonia and ammonia-DTPA extractions are noted to isolate more nitrate than
the DTPA or water extractions. Comparison of the nitrite results in Table 5 with the nitrate
results in Table 7 suggests that this may be due to a conversion of nitrite in the sample to
nitrate by the ammonia. This is particularly noticeable with the two clay soils where the nitrite
extracted by the water or DTPA is approximately equal to the increased amount of nitrate
found in the ammonia extractions. The sediment sample which exhibited no nitrite, gives
similar results for nitrate in all four extracts.
Fluoride
The ammonia and ammonia-DTPA solutions extracted substantially more fluoride
from the silt and clay containing samples, as compared to the water and DTPA extracts.
Further the low MS recoveries suggest that fluoride is being actively absorbed by the matrix
from the solution, Le.. the matrix is not saturated with respect to fluoride. The improved
performance of the ammonia solution over that of the DTPA suggests that heavy metals are
responsible for the fluoride absorption rather than alkaline earth metals. The sediment sample
exhibited no difference between the different treatments and further the matrix spike
recoveries were excellent.
137
-------
Phosphate
Substantial interference with the phosphate signal in the electropherogram was noted in
most of the samples due to carbonate and organic anion, particularly formate, presence. The
poor to fair matrix spike recoveries obtained from all the four extraction treatments on the silt
and clay samples further indicates the samples were unsaturated with respect to phosphate and
the extraction treatments were unable to overcome the absorption of phosphate by the matrix.
These results suggest that a simple extraction and CE analysis is not suitable for phosphate
determination in solid samples. This is unfortunate, however the traditional acidic oxidative
digestion followed by molybdate colorimetric determination used for total phosphorus is one
of the few good procedures for analysis of anions in solids.
Sludge samples
The two sludge samples were run as solids without drying. The heterogeneous nature
of the samples makes the interpretation of the matrix spike results difficult. Further the
electropherograms exhibited large amounts of interference with the fluoride and phosphate
anions. A good picture of the chloride, bromide, nitrite, sulfate and nitrate contents of the
sludges were obtained in a short period of time.
Conclusion
Overall what is seen is mat the tradition recipes for anion isolation from soil matrices
may work very well for indivdual anions tested one at a time using conventional wet chemical
procedures. However for the multi-analyte techniques which are the technology of the future,
more generally applicable processes need to be developed. This short survey has indicated
that mere are procedures other than the use of plain water which are substantially more
efficient for analyte isolation. Future promising directions of development include looking at
ethanol amines as extraction additives.
138
-------
26
"Choosing Between Closed Vessel and Atmospheric Pressure Microwave
Sample Preparation for Environmental Analysis"
Leo Collins. Phil Shymanski, Kevin Kelly
01 Analytical, PO Box 9010, College Station, Texas 77842-9010
Closed vessel microwave technology has been well documented showing
fast and precise sample preparation for many types of environmental samples.
The microwave transparent, pressurized vessels used in this study quickly
reached the target temperature of 175°C. The digestion period was only 10
minutes, compared to the traditional 90-100 minute hot plate procedure. Control
of the method was possible through remote PC access. Windows-based
software allowed real-time monitoring, data storage, and more. The study will
demonstrate the improvements in time and precision for closed vessel
microwave sample preparation compared to the standard hot plate sample
preparation methods, for animal tissue, plant tissue, and sediment standard
references.
Atmospheric pressure microwave technology is more appropriate for
larger sample types and samples with high organic content. Atmospheric
pressure microwave sample preparation was performed in quartz vessels with
air reflux columns to eliminate the possible loss of volatile analytes. Programmed
methods allow for sequential reagent addition. The focused microwave
instrument can prepare four samples at a time for Kjeldahl nitrogen and
phosphorus analysis. Catalyst-free digestions were obtained in 20-25 minutes,
much less than the traditional 1-2 hour procedures that often require a metal
catalyst. Speed and precision gains when performing atmospheric pressure
microwave sample preparation for Kjeldahl nitrogen and phosphorus analysis
will also be demonstrated.
139
-------
27
THE EXTENSION OF EPA MICROWAVE
LEACH METHODS 3015 AND 3051
Dirk D. Link. Peter J. Walter, H.M. 'Skip' Kingston
Department of Chemistry and Biochemistry, Mellon Hall,
Duquesne University, Pittsburgh, PA 15282-1503
ABSTRACT
Microwave leach methods have been developed as an alternative to
traditional leach methods that use hot-plate technology. These include EPA
microwave acid leach methods 3015 "Microwave assisted acid digestion of
aqueous samples and extracts" and 3051 "Microwave assisted acid digestion of
sediments, sludges, soils, and oils". Method 3051 was designed as an
alternative to hot plate method 3050, but due to differences in the reagents
specified for use in each method, it does not reproduce the chemistry of
method 3050. This results in chemical biases between methods. Using
hydrochloric acid along with nitric acid in method 3051 provides better
comparability of the chemistries, and leads to better recoveries for those
elements which typically are biased low when using method 3051. As
method 3015 currently uses only nitric acid, similar results in recoveries are
expected upon adding hydrochloric acid to the leach.
INTRODUCTION
Microwave leach methods have been developed as an alternative to
traditional leach methods that use hot-plate technology. These include EPA
microwave acid leach methods 3015 "Microwave assisted acid digestion of
aqueous samples and extracts" and 3051 "Microwave assisted acid digestion of
sediments, sludges, soils, and oils". Each of these methods uses nitric acid
and microwave heating to perform leaches of environmental matrices.
These methods have shown low recoveries for several elements because of
the stipulation that only nitric acid be used.
EPA Method 3051 was designed as a microwave alternative to hot-plate EPA
method 3050. Method 3050 allows for the use of nitric acid, hydrogen
peroxide, and hydrochloric acid, while method 3051 specifies that only nitric
acid be used as the leaching acid. The EPA specified that method 3051 use
only nitric acid in an attempt to avoid potential interferences by high chloride
concentrations on several instrumental methods, including electrothermal
140
-------
vaporization atomic absorption spectrometry and inductively coupled plasma
mass spectrometry. If hydrochloric acid is not included in the microwave
leach procedure, chemical biases can be introduced between the methods.
Microwave method 3051 does not reproduce the chemistry that is taking place
in hot-plate method 3050, resulting in differences in leach recoveries for some
elements. For example, it has been shown that recoveries for the elements
iron and aluminum are 12% and 50% lower, respectively, when using
method 3051 than when using method 3050 for the leaching of a standard
sediment sample (NIST SRM 2704) (1). In order to improve the comparability
of the microwave methods, the chemistry of the microwave methods has
been modified to more accurately reproduce the chemistry of the standard
hot-plate methods. Hydrochloric acid is added to enable the complexation
and stabilization of elements such as aluminum, antimony, iron, and silver.
By including hydrochloric acid in the microwave method, the chemical biases
between microwave and hot-plate leach methods are minimized. A recent
study has shown that for the leaching of standard soils and sediments, the
optimum acid ratio for EPA method 3051 is 9 mL nitric acid to 3 mL
hydrochloric acid (2). By using this ratio to leach standard reference materials
using method 3051, recoveries of elements that were traditionally low, like
antimony, iron, and silver, were increased and are comparable to recoveries
using method 3050. Further studies will demonstrate the range of
applicability of method 3051 using the new 9 mL nitric to 3 mL hydrochloric
acid combination. The altered method will be used to leach, a variety of
different standard reference materials of various matrix types for which the
current method 3051 is applicable, including soil, sediment, oil, and oil-
contaminated soil.
EPA method 3015 is an acid leach of aqueous samples, such as wastewater, for
determination of 23 RCRA regulated elements. This method only uses nitric
acid in the leaching of the aqueous matrices. It is expected that recoveries for
elements which were low using method 3051, namely antimony, iron, and
silver, will also be low using 3015. The chemistry of this method does not
take advantage of the complexation power of hydrochloric acid. It will also be
shown that by including hydrochloric acid in method 3015, recoveries for
traditionally less soluble elements, such as iron, antimony, and silver, will be
improved by complexation and stabilization by hydrochloric acid.
SUMMARY
Current EPA methods 3051 and 3015 only use nitric acid to leach RCRA
regulated elements out of environmental matrices. Exclusive use of nitric
acid results in low recoveries of some elements. In current studies,
hydrochloric acid has been added to the nitric acid, and each leach method
141
-------
was performed on a variety of matrices for which they are applicable.
Improvements in recoveries for elements such as antimony, silver, and iron
are demonstrated.
REFERENCES
(1) Kingston, H.M.; Walter, P.J. "Comparison of Microwave versus
Conventional Dissolution for Environmental Applications", Spectroscopy
1992,7,20-27.
(2) Link, D.D.; Walter, P.J.; Kingston, H.M. "The Extension of EPA
Microwave Leach Methods 3015 and 3051", Abstract 978, Pittsburgh
Conference, March 1996, Chicago, IL.
142
-------
28
Application of Automated Microwave Digestion for
Envrionmental Sample Analysis
By: S. Leikin, B. Moshiri & M. Moses
Presented by: Michael Moses
Microwave Digestion has tremendously reduced the time required for preparation
of samples prior to analysis by AA and ICP. However, Microwave digestion still
remains a labor intensive technique. Several manufacturers have attempted to
automate this process with continuous flow systems. A new technology has
been developed by Questron under a grant from the Ontario Ministry of Energy
and the Environment. This technology involves the technique of Discreet Flow
for Microwave Digestion. This technology permits complete isolation of each
sample without dilution and contamination. It also permits the application to
solids and larger particles.
In this paper we will discuss the application of automated discrete flow
automated microwave digestion including: dispensing, reagent mixing,
predigestion, heating cooling and dilution sample. Applications and results will
be presented for soils, vegetation, geologicals and EPA protocols.
143
-------
29
Heterogeneous Reduction of Nitrate Nitrogen in Water and Wastewater for
Compliance Monitoring
Abstract Submitted for 12th Annual Waste Testing & Quality Assurance
Symposium July 23-26,1996 Washington, DC
Bhupendra R. Patel, Mark Harris
BRAN+LUEBBE 1025 Busch Parkway BUFFALO GROVE, IL 60089
The objective of this paper is to present the data and conclusions of the nitrate determination
performed on water samples using heterogeneous and homogenous reduction of nitrate to nitrite
method.
Several methods have been used for nitrate determination in water and wastewaters. Most of these
methods have limited sensitivity or suffer from serious interference by other constituents present in
the water samples. For example, nitrate determination by nitration of 2,6-xylenol (1) is interfered by
large chloride concentrations. A polorographic method of catalytic reduction of nitrate in the
presence of uranyle ions(2) is not suitable in today's strict OSHA and EPA regulations due to the use
of elemental mercury. Ion Selective Electrode(ISE) methods for nitrate determination have been
used for fresh water samples. Application of ISE method in wastewater and seawater sample is
limited due to the interference from most common anions.
One of the most sensitive method for nitrate determination is reduction of nitrate to nitrite followed
by formation of azo dye. Mullin and Riley(3) proposed the homogeneous reduction of nitrate by
hydrazine in the presence of copper ions as catalyst. However, the hydrazine reduction method is
not quantitative, highly dependent on external conditions and lengthy(4). The reduction of nitrate in
a heterogeneous system using metallic cadmium granules have been reported by several
authors(6,7,8). This paper will discuss the method performance, reaction of nitrate reduction and
applicability of method for different types of samples for nitrate analysis.
The information provided in this presentation would be helpful to water treatment professionals,
wastewater analysts, and water treatment plant engineers.
Reference:
1. Hartley A.M., and Asai R. I. (1965) Anal. Chem. 35,1207
2. Barnes H (1959) Apparatus and methods of Oceanography, Part I, Allen and Unwin, London.
3. Mullin J. B. and Riley J. P. (1963) Analyt. Chim. Acta 29,272-279.
4. Dal Pont G. (1962), CSIRO Australia Ocenogr. Cruise Rept. 4, Oceanographic Obs. in the Indian
Ocean, 1960.
5. Greenberg A.E., Clesceri L.S., and Eaton A.D.( 1992) "Standard Methods for the Examination
of Water and Wastewater" Method 4500-NO3'E.,18th Edition, 1992
6. EPA/600/R-93/100, (Aug. 1993)., Methods for the Determination of Inorganic Substances in
Environmental Samples., Method 353.2.
7. American Society for Testing and Materials(ASTM) method (1990)D 3867
8. BRAN+LUEBBE (1995) Method 901-95(currently being reviewed for approval)
144
-------
QUALITY
ASSURANCE
-------
30
THE DEVELOPMENT OF NIST STANDARD REFERENCE MATERIALS
(SRMs) FOR METHfLMERCURY CONTENT IN COMPLEX
MATRIX ENVIRONMENTAL SAMPLES
frlarv Kate Behlke. Michele M. Schantz, and Stephen A. Wise
Analytical Chemistry Division, Bid 222, Room B208, National Institute of Standards and Technology,
Gaithersburg, Maryland 20899
BACKGROUND
Many mercury-containing compounds are currently used or were previously used by industry and agriculture
for applications such as wood preservatives, seed treatments, thermometers, marine paint additives,
batteries, pesticides, fabric softeners, and furniture polish1. The initial forms and amounts of mercury
introduced into the environment are a poor indication of actual pollution levels and dangers to marine
organisms, however. This is caused by the natural interconversion of mercuric compounds and ions in the
marine environment. Therefore, quantification of the most toxic mercury compounds formed in this cycle
is especially importrnt to correctly assess environmental damage. For this reason, scientists have focused
their efforts on the formation and quantification of methylmercury and other alkylmercury compounds in the
marine environment.
The speciation of toxic metals is vital to the accurate assessment of pollution levels due to the variability
in toxicities of different metallic compounds. Few speciated certified reference materials (CRMs) are
currently available for complex matrix materials similar to those routinely monitored for environmental
contamination. CRMs are necessary to ensure method validation and data quality assurance.
In response to the growing trend toward speciation and the need for method and data validation, an analytical
method based on solid-liquid extraction, gel permeation chromatography (GPC) cleanup, and gas
chromalogruphy with microwave-induced plasma atomic emission detection (GC-AED) was developed at the
National Institute of Standards and Technology (NIST) to quantify methylmercury in biological tissues and
sediments. Ten NIST standard reference materials (SRMs)(SRMs are CRMs from NIST) were screened for
methylmercury content using the GC-AED method to determine which materials would be appropriate for
eventual certification2. The two materials with the highest methylmercury content were SRM 1974a
Organics in Mussel Tissue and SRM 2974 Organics in Freeze-dried Mussel Tissue. A third mussel tissue
reference material, RM 8044, also was chosen for further investigation due to the similarity in matrix
types.
In addition to the three mussel tissue materials being analyzed for total mercury and methylmercury content
at NIST, the materials also were analyzed by the International Atomic Energy Agency (IAEA), Marine
Environment Laboratory (Monaco) and the Institute frr Applied Physical Chemistry, Research Centre
Jiilich (Germany) for the same two analytes. The analytical methods used by all three laboratories are
independent from each other, thus providing three sets of independent data for the determination of certified
methylmercury and tola' mercury values in the three mussel tissue materials. The methods used by each
laboratory are described briefly in the experimental section of this manuscript. More detailed descriptions of
these methods can be found in the references indicated.
EXPERIMENTAL
Determination of Methylmercury by NIST using GC-AED: Samples were mixed with 1 mL 1 M copper
sulfate, 8 mL water (only added for standard solutions and RM 8044), 4 mL acidic potassium bromide
solution, and 2 mL toluene in individual 50 mL centrifuge tubes for 1 h. The toluene layers were then
separated, and the extraction was repeated with a second 2 mL aliquot of toluene. The combined toluene
extracts for each sample were concentrated to 0.5 mL, and high molecular weight pigments and lipids were
removed using preparative GPC. The cleaned extracts were analyzed using gas chromatography (OV-1701
column, 0.53 mm x 15 m, 3.0 Jim film thickness, Quadrex, New Haven, CT) with mercury-specific
microwave-induced plasma atomic emission detection3.
145
-------
Determination of Methylmercury by IAEA using cold vapor atomic fluorescence spectrometry (CV-AFS):
Samples were saponified in closed Teflon vial overnight at 70 °C with 10 mL 25% potassium hydroxide in
methanol and diluted to 25.8 mL with metham 1. Methylmercury in the saponified samples was derivatized
with sodium tetraethyl borate at pH 4.9 and collected on a Tenax trap. Separation of the individual
organomercury species was achieved by isothermal gas chromatography (glass capillary SPB-5 column,
0.75 mm x30 m, 1.0 p.m film thickness) followed by pyrolysis and CV-AFS detection. AH analysis were
done during the same day4.
Determination of Methylmercury by Research Centre Jtilich using cold vapor atomic absorption
spectrometry (CV-AAS): Samples were extracted three times with 6 M hydrochloric acid and the extracts
combined. Methylmercury in the acid extract was decomposed to inorganic mercury with nitric acid or
ultraviolet light followed by CV-AAS detection5-6.
Determination of Total Mercury by NIST using flow injection cold vapor atomic absorption spectrometry
(FI-CV-AAS): Samples were mixed with 5 mL concentrated nitric acid and microwave digested for 10-20
min at a maximum temperature of 228 °C and a maximum pressure of 1242 kPa. After digestion was
complete, the contents of the vessels were transferred to 100-mL volumetric flasks containing 0.75 mL 1%
potassium dichromate solution and diluted to volume with 1% (v/v) sulfuric acid. Samples were then
analyzed by cold-vapor generation using the FIAS-200 flow injection system and a Perkin-Elmer 5000
atomic absorption spectrometer (Perkin-Elmer, Norwalk, CT). Peak height measurement was used for
quantification7.
Determination of Total Mercury by NIST using instrumental neutron activation analysis (INAA): Samples
were weighed into individual quartz vials and flash frozen in liquid nitrogen prior to sealing the vials. The
vials were encapsulated in polyethylene film followed by packaging in sets of eight in polyethylene
irradiation vessels. Each vessel was irradiated for 3 h at a reactor power of 20 MW. The samples were
removed from the vessel approximately 1-2 months after irradiation, the polyethylene films removed, and
the outer vial surface cleaned. Each vial then was encapsulated in clean polyethylene films and gamma
radiations were collected for > 8 h using a germanium detector and associated electronics. Quantification of
mercury was based on comparison with standards using the 279 keV line from 203Hg, corrected for the
interference from 75Se.
Determination of Total Mercury by IAEA using cold vapor atomic fluorescence spectrometry (CV-AFS):
Samples were digested in closed Teflon vials with 4 mL nitric acid and 2 mL sulfuric acid for 3 h at 70 °C
and diluted to 25.8 mL with milliQ water and 500 \lL of a preservative. The digested samples were
analyzed by double gold trap amalgamation using CV-AFS detection. Each sample was analyzed twice and
all analyses were done during the same day8.
Determination of Total Mercury by Research Centre Julich using gold amalgamation cold vapor atomic
absorption spectrometry (GA-CV-AAS): Samples were mixed with 10 mL nitric acid and digested at 150
°C for 10 hours. The digested samples were diluted to 20 mL with distilled water prior to analysis by cold-
vapor atomic absorption spectrometry. A portion of each samples solution was mixed with tin chloride to
reduce any Hg(II) to Hg(0), and the resulting mercury vapor was collected by amalgamation on a gold wire.
The collected mercury was thermally desorbed from the wire by heating the wire to 600 °C, and transported
to the glass cuvette for absorption measurement at 253.5 nm (slit width 2.0 nm, cuvette temperature 50
°C)9.
RESULTS AND DISCUSSION
Table I contains the mean methylmercury and total mercury concentrations for each mussel tissue material
as determined by each laboratory. IAEA has not reported a total mercury value for SRM 1974a or an
uncertainty for methylmercury in RM 8044 as indicated in the table. Methylmercury results are presented
as ng methylmercury expressed as mercury per g of dry material, and total mercury results are presented as
ng total mercury per g of dry material. The mean of the method means is reported for each material as an
indication of the certified value, although the exact certified concentrations have yet to be finalized. The
uncertainties tor these overall mean values are expressed as 95% confidence intervals.
146
-------
The total mercury and methylmercury results for SRM 2974 and SRM 1974a agree within the uncertainties
of the values. These two materials were prepared from the same tissue homogenate, so this agreement
indicates freeze-drying has had little or no effect on the materials' mercury contents. The data for these two
materials are currently being evaluated by MIST statisticians for eventual certification.
The methylmercury value determined by NIST for RM 8044 was lower than the results from IAEA and
Research Centre Jiilich. The NIST extraction procedure may not be quantitatively extracting methylmercury
from this material. Research Centre Jiilich observed an unusual sample size dependence to the measured
methylmercury concentrations in this material, but believe this dependence did not affect their final reported
results. IAEA indicated that a problem with their extraction procedure may have initially caused an
elevation in their measured methylmercury concentrations in RM 8044 which they corrected before
reporting results. These problems encountered by IAEA and Research Centre Jiilich during the analyses of
RM 8044 and the disagreement among the three laboratory results indicates that further investigations of
this material are needed. The total mercury results for RM 8044 agree quite well, however, as compared to
the methylmercury values.
ACKNOWLEDGMENTS
The authors wish to thank Milena Horvat, Vesna Mandic, and Sabine Azemard at the Marine Environment
Laboratory /IAEA and Hendrik Emons, Peter Ostapczuk, and Mechtild Burow at the Research Centre Jiilich
for providing total mercury and methylmercury data. We also thank Elizabeth Mackey, Rabia Demiralp and
Raj Saraswati at NIST for providing total mercury data and Barbara Porter at NIST for providing water
content data on the materials.
REFERENCES
1. Krenkel, PA (1973) CRC Crit. Rev. Environ. Control 3, 303-373.
2. Behlke, MK (1996) Doctoral Dissertation, University of Massachusetts.
3. Behlke, MK, Uden, PC, Schantz, MM (1996) Anal. Comm. 33: 91-2.
4. Horvat M, Liang L, Bloom NS (1993) Anal. Chim. Acta 282: 153-168.
5. May K, Stoeppler M, Reisinger K (1987) Toxicol. & Environ. Chem. 13: 133-159.
6. Ahmed R. May K, Stoeppler M (1987) Fresenius Z. Anal. Chem. 326: 510-516.
7. Saraswati R. Vetter TW, Jr. RLW (1995) Mikrochim. Acta 118: 163-175.
8. Liang L, Bloom NS (1993) J. Anal. Atom. Spectrosc. 8: 591-594.
9. Padberg S, Burow M, May K, Stoeppler M , Uberlingen, Germany 1991; Bodenseewerk Perkin-Elmer
GmbH.
Table I Summary of methylmercury and total mercury results for mussel tissue materials
MeHg
(ng/gasHg)
Mean values:
Total Hg
(ng/g)
GC-AED:
CV-AAS:
GC-AFS:
FI-CV-AAS:
INAA:
GA-CV-AAS:
CV-AFS:
Institute
NIST
Jiilich
IAEA
NIST
NIST
Jiilich
IAEA
SRM 1974a
74.3 ± 15
81.1 ±3
78.1 ±5
77.3 ± 12
184 ± 17
203 ± 18
170 ± 10
-
SRM 2974
80.6 ± 10
71.7+ 3
80.6 ± 5
77.8 ± 8.4
186 ±2
155+ 16
182 ±4
17418
RM8044
20 ± 2
28.1 +
28
-28
61 ±2
59 + 8
62 ±2
64 + 6
1.5
Mean values: 186 ±41 174 ±22 62 + 3
147
-------
31
VARIABILITY AMONG TOXAPHENE REFERENCE STANDARDS
Francis J. Carlin, Jr., Senior Research Scientist, and John M.
Hoffman, Chemist, Analytical Science Division, Hercules
Incorporated, Research Center, Wilmington, Delaware 19808-1599
ABSTRACT
Toxaphene is a chlorinated camphene insecticide which yields
a complex, multi-component gas chromatogram. The accurate
identification and measurement of toxaphene in the environment
depend on the availability of a suitable reference standard.
Toxaphene standards were purchased from nine suppliers. The
comparison of these standards by electron capture gas
chromatography resulted in the discovery of a wide range in
the relative proportions of the gas chromatography peaks among
the materials studied. Such variations will have serious
implications on the qualitative and quantitative determination
of toxaphene in environmental samples. Chromatograms to
illustrate the differences will be presented.
Toxaphene has been a widely used pesticide in the United
States. Over the past several years, Hercules has collected
environmental samples as part of monitoring activities and has
submitted those samples to contract laboratories for the
determination of toxaphene. Recommended methods1 specify that
quantitation is done by using peaks in the latter half of the
toxaphene chromatogram. Because Hercules observed differences
among results of analyses between laboratories, a search for
the cause of the variations was undertaken.
When no obvious differences (which could cause the observed
variations in the results of analysis) in the practice of the
methods were discovered, the question of the uniformity of
toxaphene calibration standards was investigated. In the
summer and fall of 1995, a survey revealed that there were
more than twenty providers of toxaphene reference standards,
but the number of primary suppliers was reduced to nine upon
further investigation.
148
-------
PESULTS
Samples from the various suppliers were purchased as neat
standards. The reference sample for this comparison study was
a retained portion of a toxaphene lot designated as X16189-49,
which was produced in the Hercules Brunswick, Georgia, plant
in 1968. Each standard purchased was compared with technical
toxaphene X16189-49 after injection on 30-meter X 0.53-mm J&W
Megabore columns: DB-5 and DB-1701. Hexane solutions of all
standards were injected into a Hewlett-Packard 5890 gas
chromatograph equipped with an electron capture detector, and
the detector response was monitored by HP ChemStation software
(client/server version).
Figures I through 3 illustrate the differences in the
distribution of the component peaks of the various standards
purchased. The chromatograms shown are from the DB-1701
column, and technical toxaphene X16189-49 is the uppermost
sample in each set. It is obvious that the qualitative
differences among the standards vary greatly. Some standards
have larger peaks in the first half of the chromatograms, and
some have larger peaks in the latter half. In addition to the
obvious differences in the distribution of the GC peaks among
the samples, the physical appearance of the toxaphene is
different, particularly in the case of Supplier #9, where the
normally waxy toxaphene is a viscous liquid at refrigeration
temperature.
Two standards (Figure 4; DB-5 GC column) give an excellent
match for technical toxaphene X16189-49 — a neat standard
from Supplier #8 and a solution in hexane purchased from
Supplier #10.
How standards are prepared may result in differences in
analytical results. The chromatograms in Figure 5 show the
difference between two standards purchased from Supplier #8.
The top chromatogram is technical toxaphene X16189-49, the
middle chromatogram is a neat toxaphene standard, and the
bottom chromatogram is a solution of a toxaphene standard
supplied as a solution in methanol. The differences in the GC
profiles are significant and illustrate the problem of
obtaining consistent standards — even from the same supplier.
In addition to the qualitative comparisons reported here, work
is in progress at the Hercules Research Center to demonstrate
the effects of the variations among commercially available
toxaphene reference standards on calibration curves and on the
quantitative determination of levels of toxaphene in
environmental samples.
149
-------
CONCLUSIONS
This comparison of toxaphene reference standards from
commercial suppliers has demonstrated significant variations
among the materials currently available. These variations
have serious implications for the accurate identification and
quantitation of toxaphene in environmental samples.
Action should be taken by suppliers of reference standards to
establish a true toxaphene reference standard for use by all
laboratories involved in the analysis of environmental samples
for toxaphene.
SUMMARY
Hercules personnel have conducted a survey of the suppliers of
toxaphene calibration standards. The results of that survey
demonstrate significant differences among the qualitative GC
profiles of the "standards" purchased from different
suppliers. Also demonstrated is the significant difference
between two toxaphene standards purchased from the same
supplier.
These differences have serious implications for the accurate
qualitative identification and quantitative determination of
toxaphene in environmental samples.
Work is in progress at the Hercules Research Center to
determine the potential analytical problems caused by such
variations in the toxaphene "reference standards" which are
currently available to analytical laboratories.
REFERENCES
1.) U.S. Environmental Protection Agency, "Test Methods for
Evaluation of Solid Waste, Volume IB: Laboratory Manual,
Physical/Chemical Methods, (SW-846)," Method 8080A,
Organochlorine Pesticides and Polychlorinated Biphenyls
by Gas Chromatography, Revision 1, November 1992.
150
-------
I - COMPARISON OF HERCULES TOXAPHENE AND STANDARDS FROM COMMERICAL SUPPLIERS
counts
160000-
140000-
120000-
100000-
80000-
60000-
40000
20000-
Hercules Technical Toxaphene X16189-49 @ 5.0 ug/mL
Supplier # 1 @ 3.8 ug/mL
10
15
20
25
counts
160000-
140000-
120000-
100000-
80000-
60000-
40000-
2000°'_1
o-
0
Supplier #2 @ 3.2 ug/mL
10
15
20
25
counts
160000-
140000-
120000-
100000-
eoooo-
60000-
40000-
20000-
0J
Supplier # 3 @ 4.7 ug/mL
0
10
15
-25.
min
151
-------
FIGURE 2 - COMPARISON OF HERCULES TOXAPHENE AND STANDARDS FROM COMMERCIAL SUPPLIERS,
counts'
160000H
140000-
120000-
100000-
80000-
60000^
40000-
20000-3
0-i
Hercules Technical Toxaphene X16189-49 @ 5.0 ug/mL
10
15
20
25
counts"
160000J
140000-;
120000-
100000-
80000-
60000-
40000-
20000-
D-i
0
Supplier # 4 @ 4.8 ug/mL
10
is
20
counts
160000-
140000-
120000-
100000-
80000-
60000-
40000
20000
1
counts
160000-
140000-
120000-
100000-
80000-
60000-
40000-
20000^
0
Supplier # 5 @ 5.2 ug/mL
10
15
20
25
mm
Supplier # 6 @ 4.9 ug/mL
10
15
20
25
152
-------
FIGURE 3
- COMPARISON OF HERCULES TOXAPHENE AND STANDARDS FROM COMMERCIAL SUPPLIERS.
counts 1
160000-
140000-
120000-
100000-
80000-
60000-
40000-
20000-
0-
0
Hercules Technical Toxaphene X16189-49 @ 5.0 ug/mL
10
IS
20
25
counts "
160000 -
140000-
120000-
100000-
80000 -
60000-
40000r
20000
3-
u
Supplier # 7 @ 5.1 ug/mL
10
Supplier # 8 @ 5.4 ug/mL
15
20
ccunts
160000-
140000-
120000-
100000-
80000-
60000 ~
40000-
20000-
0-
Supplier # 9 @ 5.5 ug/mL
o
10
15
20
25
153
-------
FIGURE 4 - COMMERICAL STANDARDS WHICH MATCH HERCULES TOXAPHENE.
counts
25000CM
200000-
150000^
100000-
50000 -
Hercules Technical Toxaphene X16189-49 @ 5.0 ug/mL
10
15
20
25
counts
250000-
100000-
150000-
100000-
50000-
0-
LA.
Supplier # 8 @ 5.4 ug/mL
10
15
20
25
counts
250000-I
200000-
150000-
100000-
50000-
0-
Supplier # 10 @ 5.0 ug/mL
10
15
20
25
154
-------
FIGURE 5 - DIFFERENCES BETWEEN TOXAPHENE STANDARDS FROM A SINGLE SUPPLIER.
counts j
250000-
200000-
150000-
100000-
50000
0-
Hercules Technical Toxaphene @ 5.0 ug/mL
10
15
20
25
cour.ts
counts
250000-
200000-
150000-
103GOO-
50000-
Supplier # 8, Neat Standard, @ 5.4 ug/'mL
Supplier #8, Solution of Standard, @ 5.0 ug/mL
10
15
20
25
155
-------
32
GC/MS AUTOMATION TECHNIQUES FOR
THE ENVIRONMENTAL CHEMIST
Philip Cocuzza, GC/MS Task Leader, Steven Bucher. GC/MS Scientist, and Richard
Phillips, Analytical Section Supervisor, Lockheed Environmental Systems and
Technologies Company, 2890 Woodbridge Avenue, Edison, New Jersey, 08837
ABSTRACT
Both semivolatile and volatile gas chromatograph/mass spectrometry (GC/MS)
methods are automated with minimal financial and manpower investment through the
use of automated intelligent sequencing procedures and spreadsheets. The amount of
time required for sample analysis and data processing is greatly reduced thereby
increasing operator efficiency and productivity. This increase in productivity
translates into a reduction in per analysis cost.
INTRODUCTION
The environmental laboratory market has evolved from a large number of independent
laboratories relying mostly on the quality of analytical data to attract and retain clients
to a highly competitive arena with key players concerned about continuing pressure on
cost and profitability. The quality of the data and sample turnaround are still of
paramount importance, however, due to the larger proportion of corporate monies
earmarked for environmental services, cost is and remains a deciding factor when
choosing an environmental laboratory. To compete profitably, laboratory managers
need to implement methods to increase productivity and thus reduce per analysis cost.
The gas chromatograph/mass spectrometry (GC/MS) instrumentation is an integral
part of the modern environmental laboratory and the GC/MS automation techniques
discussed in this paper provide a method to improve the cost-effectiveness of the
method and thus enhance the profitability of the laboratory.
PURPOSE
The daily activities of a GC/MS analyst in an environmental laboratory includes
analyzing an endless backlog of samples and reviewing acquired data. In addition,
many laboratories request their GC/MS departments to perform special analyses or to
156
-------
adhere to specific customer demands, i.e., quick sample turnaround, creating an
additional burden on an already overworked laboratory. To respond to these requests,
the analyst must develop a working routine integrating efficient and maximum use of
the GC/MS instrument with quick but thorough data review, qualification, and
reporting. Automation procedures can significantly reduce analysis time, calculation
and transcription errors, and project turnaround time. Toward this end, several
automation procedures were developed using standard analytical laboratory GC/MS
instrumentation and third party spreadsheet software.
Following sample extraction, GC/MS analysis of aqueous environmental samples falls
into three major categories: 1) instrument analysis of sample extract, 2) data review
and 3) data qualification / reporting. Automation of all these tasks is easily
accomplished as described below using a method of semivolatile analysis to
demonstrate and discuss intelligent sequencing and a volatile method to discuss data
review and data qualification / reporting. The flexibility of these automation
techniques is also portrayed.
PROCEDURE
The order of sample injection is automated by using intelligent sequencing, a fully
automated procedure allowing the program to make logical decisions concerning
instrument analysis of extracts. By employing supplied keywords in a sequence log
table coupled with system macros, decisions can be made by the software concerning
various aspects of instrument analysis of extracts, i.e., sample injection. The
following are examples of the unique capabilities of the sequencing procedure.
Entry number one in the sequence log table (Example 1A) is the keyword SeedName,
an automated sequential numbering system used either to prevent overwriting of data
files upon reinjection or to allow a jump in the sequence. A brief format report of the
sequence log is printed before analysis begins and a final sequence sample log is
printed after analysis is completed to verify sample analysis. The second and third
keywords in Example 1A are Reinject and On_Flag. These keywords are set to either
a pass or fail flag by a system macro. The macro first analyzes the sample data and
makes decisions based on sample type, i.e., standard, method blank or sample. The
sample type is set to a numerical value that is recognized by the macro. After
deciding the sample type, the macro checks the sample data based on criteria set by
the user. If for example the macro decides that a sample does not meet the specified
criteria, the macro sends a fail message to the sequence log table which then carries
out the appropriate action defined by the keyword. On_Flag causes a jump in the
sequence to the keyword Label which initiates a wash blank while Reinject directs the
autosampler to reanalyze the sample.
157
-------
This decision making process by the system macros is very useful when determining
whether or not tuning compounds meet the specified criteria, an essential analytical
consideration prior to sample analysis. When a decafluorotriphenylphosphine
(DFTPP) tune sample is analyzed (number 4 on Example 1A), the instrument
determines whether or not the collected data of DFTPP passes or fails the prescribed
tuning criteria. If a pass condition exists, the analytical sequence continues. If a fail
condition exists, the software determines whether or not the DFTPP sample is
reanalyzed or a target tune procedure is performed . The target tune procedure
retunes the GC/MS instrument, reanalyzes the calibration sample DFTPP and checks
the response against the criteria.
After the DFTPP tune sample passes, a daily calibration check standard is injected.
Again the system macros are employed to check if the daily calibration standard
meets acceptance criteria. If a fail flag is passed to the sequence log table, the
computer directs the autosampler to reinject the continuing calibration standard. If the
daily calibration standard meets acceptance criteria, the response factors, compound
retention times, and reference spectra is updated, printed and saved for sample
quantitation and/or qualification. It should be noted that if the DFTPP or continuing
calibration standard fails after reinjection, the sequence may either be halted,
continued for use in sample screening or instructed to analyze an initial calibration
curve.
Before the analysis of the method blank, the keyword DB_Start is used to inform the
computer to start logging sample data to a spreadsheet (the keyword DB_End is self
explanatory). The spreadsheet is used as a database to automatically plot control
charts and is linked to other spreadsheets for data qualification. The analysis of a
method blank is checked using the system macros for contamination of target
compounds above the allowable limits, internal standard response, and surrogate
recovery. If any of the acceptance criteria is not met, the method blank is either
reinjected or a jump in the sequence occurs to the sample labeled wash blank. The
wash blank is injected and analyzed for contamination above the allowable limits and
is used to help clean up the system from a previous sample run and/or determine if
any carryover exists from a previous injection.
Sample analysis is initiated after all the quality control samples have been injected,
analyzed and checked for criteria acceptance. Samples are analyzed and scrutinized
using the same criteria as the method blank discussed previously. If the acceptance
criteria, i.e., surrogate recovery or internal standard response, is not passed, the
sample may be reinjected or a wash blank analyzed before the next sample injection.
LastRun in the sequence log table informs the macro that this is the last sample
analyzed in this sequence and to log the data to the database and to send the control
charts to the printer.
158
-------
The intelligent sequence in this example was instructed to perform decisions
concerning DFTPP, daily calibration, internal standard recovery, surrogate recovery,
and saturation of the linear range of the initial calibration curve. The intelligent
sequence routine is a powerful tool that can be used for more elaborate procedures as
part of the sample analysis protocol of GC/MS laboratory automation process. For
example, the intelligent sequence can be programmed to retune the instrument or
reinject DFTPP every twelve hours for continuous operation.
The second and third part of the analysis of environmental samples deals with data
review and qualification / reporting. The procedures that follow were developed using
spreadsheets in combination with a personal computer based environmental software
package. The procedures are tailored for environmental sample analysis however are
adaptable to any laboratory procedure.
The first procedure performed using the spreadsheet generated from the analysis of
the quality control and environmental samples is the QC Summary Report ( Example
IB). The report is automatically printed at the conclusion of an analysis sequence.
The QC summary report provides a quick review of important quality control criteria.
For this example, surrogate recovery, internal standard area, internal standard
retention time and injection time are checked for acceptable performance. The
spreadsheet may be modified to criteria limits to meet unique laboratory
specifications. Any data that fails to meet the analyst's specified limits is flagged
indicating a problem with that particular run. Printouts of the QC summary report,
tuning report, and continuing check standard report enable the analyst to determine
instantly the status of the analysis. The QC Summary Sheet can also be used as
documented proof of meeting acceptable QA/QC standards. Along with the QC
Summary Report, control charts for surrogate recovery are also created and printed.
Control charts (Example ID) are essential for monitoring trends in instrument
performance. QC summary sheets save the analyst considerable time and effort in
sifting through data searching for retention times/injection times or performing
menial calculations.
After user defined edits, i.e., false positive, the final automation procedure is used for
the qualification and reporting of the sample data. As part of data qualification, the
reporting limits are automatically corrected for sample dilutions. In the example
spreadsheet 1C, the final report is qualified for sample concentrations detected lower
than corrected reporting limits (J, QM) and non-detects (U). Procedures can also be
designed to qualify data for variances in internal standard recovery, surrogate
recovery, initial calibration %RSD, and continuing calibration percent difference
(%D) acceptance criteria. The capability exists in the software for the batching of the
data files for sequential data processing and generation of the final report. The report
format and the qualification parameters may also be tailored to meet client
specifications.
159
-------
SUMMARY
Although the examples presented are simplistic, the limitations to automation are
confined only by the extent of user knowledge of macro programming or the system
commands available. More elaborate procedures are easily developed to perform
complicated sequences with minimal operator interaction thus allowing the analyst to
perform other functions in the laboratory or conduct the analysis unattended during
nonworking hours. In addition, these procedures are adaptable to the determination of
organic compounds in other matrices, e.g., soil/sediment.
The Hewlett-Packard 5972 GC/MS with EnviroQuant software (version C.01.00) and
Microsoft Excel (version 5.0) was used in this study. Trade and company names are
used for identification purposes only and do not imply endorsement by Lockheed
Engineering Systems and Technologies or the U.S. EPA.
DISCLAIMER
This work has been funded in part by the United States Environmental Protection
Agency contract 68D-10158 to Lockheed Environmental Systems and Technologies
Company. The work does not necessarily reflect the views of the agency and no
official endorsement should be inferred.
REFERENCES
Hewlett-Packard MS ChemStation User's Guide, Copyright 1993
Microsoft Excel User's Guide 1, Copyright 1992
160
-------
EXAMPLE SEQUENCE LOG 1A
Sequence Name: C:\CHEM\1\SEQUENCE\SEQ1.S
Comments:
Operator:
Data Path: C:\CHEM\1\DATA
Pre-Seq Cmd:
Post-Seq Cmd:
Method Sections To Run
(X) Full Method
( ) Reprocessing Only
On A Bartcode Mismatch
(X) Inject Anyway
( ) Don't Inject
Line Type
Vial
DataFile
Method
Sample Name
1 SeedName
2 Reinject
3 On_Flag
4 DFTPP
Sample
5 DailvCal
Standard
6 DB_Start
7 Blank
SSamnle
9Samnle
10 Samnle,
HLastRun
12 DB End
13 End
14 Label
15 Wash Blank
16 Return
A00005
1 time
Run_Blank
2
qcsummdb.xls
4
6
7
qcsummdb.xls
Run_Blank
c
625 DFTPP Tune
o/j Metnou oiank
(\~)^ reliant Cnrnnl^
o/j client sample
o/j L.iient sample
625 Client Sample
625 Client Sample
AO^ \X7ocK donl*-
ozj wasn r>ianK
161
-------
EXAMPLE IB
QC SUMMARY SHEET
DATA FILE
C00407.D
C00408.D
C00411.D
C00412.D
C00413.D
C00414.D
C00415.D
SAMPLE NAME
P-BFB
VSTD050
VBLK474
#096777
#096778
#096779
#096780
1st
IS AREA
358302
328282
333139
327207
326421
145556
Q
#
2nd
IS AREA
1901181
1711740
1730779
1689684
1717440
1666246
Q
3rd 1st 2nd 3rd 1st 2nd 3rd 12HR
IS AREA Q ISRT Q IS RT IS RT SURR Q SURR Q SURR Q CLOCK Q
1511658
1354532
1373301
1332769
1350506
1324785
9.17
9.20
9.22
9.21
9.19
9.70
#
10.90
10.93
10.96
10.95
10.93
10.96
16.19
16.21
16.24
16.24
16.23
16.25
98
108
107
107
107
135
#
99
106
106
107
107
106
98
103
104
104
103
105
4/19/969:24
4/19/96 9:44
4/19/9611:18
4/19/9611:49
4/19/96 12:21
4/19/96 12:52
4/19/96 23:24
#
O)
1st IS = BROMOCHLOROMETHANE
2nd IS - 1,4-DICHLOROBENZENE
3rd IS - CHLOROBENZENE-D5
1st SURR = 1.2-DICHLOROETHANE-D4
2nd SURR = TOLUENE-D8
3rd SURR = BROMOFLUOROBENZENE
AREA UPPER LIMIT = +100% OF INTERNAL AREA
AREA LOWER LIMIT - - 50% OF INTERNAL AREA
LIMITS* 76-114
LIMITS" 88-110
LIMITS* 86-115
-------
33
INCORPORATION OF POLLUTION
PREVENTION PRINCIPLES
IN ENVIRONMENTAL METHODS
Mitchell D. Erickson. Jorge S. Alvarado,
Cheng-Shen (Jeffrey) Lu, David P. Peterson, and James Silzer
Environmental Research Division
Argonne National Laboratory, 9700 S. Cass Ave., Argonne, IL 60439-4837
(MDE telephone: 708-252-7772; MDE Fax: 708-252-9594)
ABSTRACT
The principles of pollution prevention (P2) have not been sufficiently incorporated into
analytical methods. In this paper, we focus on the needs for, the potential of, and the
benefits of incorporating the principles of P2 into environmental analytical methods.
Although the amount of reagent used per experiment is often only a few milliliters, these
small amounts can quickly add up, especially when they are aggregated across the entire
research community. Effective P* requires operational or even fundamental changes in the
techniques and methods employed. "End-of-pipe" recycling is not practical with small
streams. The high potential for P2 in the laboratory will be specifically illustrated with
improvements in routine analytical techniques. Routine analytical methods for
environmental and waste samples are quite prescriptive and often do not include the
principles of waste minimization and P2. Many methods require preparation of 100-fold or
more excess sample for an instrumental determination. Many methods also use reagents
that are not now considered "green."
We have adapted P2 principles, along with other modern analytical approaches, to develop
routine analytical methods that significantly reduce waste generation while they maintain
acceptable analytical figures of merit and achieve cost savings through reduced reagent
consumption and reduced labor costs. Results will be reported, and the significance of and
need for incorporating P2 into environmental methods will be discussed.
INTRODUCTION
The chemical industry has embraced and implemented P2 practices over the past decade.
Conversely, adoption by routine environmental laboratories has been slow because
"regulatory standard methods must be followed." Adoption of P2 practices is increasingly
necessary for the environmental analytical community as regulations tighten, waste disposal
costs escalate, and public scrutiny increases. It is incumbent on the analytical chemist to
become less introverted and "to take a lead towards the preserving of our environment
rather than to [merely] measure its deterioration" (de la Guardia and Ruzicka, 1995).
163
-------
Regulatory approved analytical methods are quite prescriptive and often do not include
principles of waste minimization and P2. A classic example is the procedure currently
approved and in use for the Toxicity Characteristic Leaching Procedure published in the
U.S. Code of Federal Regulations (40 CFR Part 261, Appendix II, Method 1311 [SW-
846]). This procedure tests materials for the leachability of toxic components. The
approved procedure requires that a 100-g sample of the unknown be leached with a solvent
20 times its weight. This procedure generates 2 L of liquid waste, which becomes
hazardous waste if any one of 40 inorganic or organic contaminants exceeds the established
regulatory level. Add to this level any secondary waste associated with instrument
calibration and quality control for the analysis, and the waste stream has multiplied over 20-
fold. Only a small percentage of the sample generated in the previous scenario is actually
used for the analysis. This example is by no means unique.
We are developing and demonstrating techniques for routine environmental analysis that
incorporate P2 principles. Here we report on the adaptation of commonly used routine
methods published by the U.S. Environmental Protection Agency (EPA) for
polychlorinated biphenyls (PCBs) in water, solids, and oils and for metals analyzed by
inductively coupled plasma spectroscopy (ICP). Necessary practical techniques and
methods for implementation in the routine environmental laboratory community have been
considered.
RESULTS
Polychlorinated Biphenyls
Solvent substitution for PCB analysis can achieve comparable results and eliminate
environmentally less-desirable solvents, as illustrated in Table 1. A wide variety of
solvents are used as extractants (Erickson, 1986), generally without significant comparative
evaluations among potential solvent systems. Technically acceptable solvents are those that
yield quantitative extraction of the analyte (as measured by a spiked sample); solubility of
the PCBs and wetting of the soil matrix are contributing factors to the efficacy of a solvent.
Our results indicate that many common solvents or solvent mixtures can yield quantitative
extractions.
Microscale extractions can cut the scale of the analysis by at least a factor of 10, as
illustrated in Tables 2 and 3.
Matching sample to determination size requirements. We can better match the amount of
sample used with the amount needed for the gas chromatographic (GC) determination step,
as illustrated in Table 4.
Waste volume reduction. The volume of waste generated can be cut by at least a factor of
10, as shown in Tables 2 and 3. This reduction is increasingly important as we move
toward full cost accounting, including waste disposal costs, in the analytical chemistry
laboratory.
164
-------
Table 1 — Solvent Recoveries Using Soxhlet Extraction in Soil
Extraction Solvent
Hexane
Acetone
1:1 Hexane/Acetone
3:1 Hexane/Acetone
3:1 Acetone/Hexane
Methylene Chloride
1:1 Methylene Chloride/Acetone
9:1 Hexane/Methylene Chloride
10:1 Toluene/Methanol
Aroclor 1254
101
101
105
109
94
89
104
99
101
Cost considerations. Costs are reduced significantly in apparatus, reagent consumption,
and labor. The apparatus and reagent cost reductions are illustrated in Tables 2 and 3. The
labor costs were not quantified but can be inferred directly from several rows of both
tables. A cost impact will occur during the transition: new glassware will be purchased,
method validation will require some overhead time (see next section), training will require
some downtime, and some efficiencies will be seen only after a break-in period.
Laboratories are advised to plan transitions such that they do not occur during a crisis
period and to phase in changes in methods over a period of time.
Quality assurance. Any adaptation of a method requires some sort of internal validation.
The changes discussed here are no different. Any laboratory adapting their routine
methods to green/microscale techniques needs to validate the changes with the appropriate
quality control samples to demonstrate that the laboratory is providing data of known and
consistent quality. In addition, quality control measures need to be modified as necessary
to clearly monitor the performance of the analyses.
Resource Conservation and Recovery Act Metals
The sensitivity and versatility of analytical instrumentation has improved
dramatically over the last two decades. The mismatch between the size of the sample
prepared in the laboratory (as required by regulatory accepted procedures) and that needed
for the instrumental characterization is often different by a factor of 1,000 or greater. With
this fact in mind, we attempted to adapt currently approved methods for analysis of
Resource Conservation and Recovery Act (RCRA) metals in soils to microscale techniques.
We used two different methods of acid digestion, three different methods for sample
introduction, and two different analytical instruments to investigate whether smaller
samples can be used for environmental characterization of soils without a loss of precision
or accuracy.
165
-------
Table 2 — Comparisons of Soxhlet, Micro Soxhlet, and Shakeout Procedures
CO
O)
Parameters
Sample Size (g)
Sodium Sulfate Used (g)
Extraction Solvent Volume (mL)
Extraction Time
Concentration Technique
Concentration Time (min)
Florisil Used for Cleanup (g)
Solvent Used for Cleanup
Final Concentration Volume (mL)
Waste Volume (mL)b
Apparatus Cost ($)c
Reagent Cost ($)d
Soxhlet
10
10
300
16-24hr
Kuderna-Danish
10-20
20
Methylene Chloride
10
610
2.50
12.76
Micro Soxhlet
1
1
15
5hr
Nitrogen Slowdown
10-20
1
Hexane
1
25
1.40
2.76
Shakeout
0.5
0.5
15a
15 min
Nitrogen Slowdown
10-20
1
Hexane
1
24
0.19
2.76
aThe 15 mL consisted of three 5-mL extractions, each lasting 5 min.
bAssumes no recycling at this point; does not include gloves and other ancillary waste.
cBased on manufacturer's catalog prices or actual purchase requisitions. Soxhlet and micro Soxhlet amortized over 100 uses
(i.e., Soxhlet investment = $250).
dBased on manufacturer's catalog prices or actual purchase requisitions; assumes complete consumption of amount purchased
for sodium sulfate, Florisil, and solvent.
-------
Table 3 — Florisil Extraction of PCBs from Motor Oil:
Macroscale and Microscale Solid Phase Extraction (SPE)
Reagent
Florisil (g)
Hexane (mL)
Oil Sample (g)
TOTAL WASTE (mL)a
Macroscale
20
280
1.5
-300
Microscale
1
25
0.2
-26
Time (min)
Dilution/Cleanup
Eluate Concentration13
GC Analysis Time
TOTAL
120
50
45
215
20
50
45
115
Cost ($)
Florisil
Hexane
Apparatus0
TOTALd
2.61
3.28
0.52
6.41
2.37 (SPE syringe)
.29
-
2.66
Yield
% Yield
100
100
aAssumes no recycling; does not include gloves and other ancillary waste.
bNitrogen blowdown technique used for concentration of eluate. Time required is based
on volume of solvent evaporated.
cQlass chromatography column with reservoir amortized over 100 uses (i.e., column
investment = $52) for the macroscale procedure. Microscale requires no comparable
apparatus.
dBased on manufacturer's catalog prices or actual purchase requisitions; assumes
complete consumption of amount purchased.
167
-------
Table 4 — Scaling Chemistries to What Is Needed
for the Determination Step (PCB Example)
Problem
1 nL/injection
Currently 103tol04of
sample injected
Solution
Use of an internal
recovery standard
Final (small) inexact
volume compensated
Result
Smaller initial sample
Fewer reagents
To better match sample size with the instrumental requirements, we evaluated the use of
microscale techniques to adapt currently approved methods for analysis of RCRA metals in
soils. We used two different, analytical instruments that are currently approved for
environmental characterization of the metal content of soils: (1) the inductively coupled
plasma atomic emission spectrometer (ICP/AES) and (2) the inductively coupled plasma
mass spectrometer (ICP/MS). We chose these instruments because they are capable of
multielement analysis on the same sample. We modeled our analysis after regulatory
approved analytical methods that exist for both of these instruments (method 6010A and
method 6020 in Test Methods for Evaluating Solid Waste, EPA document SW-846 [EPA,
1994]). We also used two different acid digestion methods that are approved for use with
these instrumental methods: (1) method 3050A (hot-plate digestion using nitric and
hydrochloric acid and hydrogen peroxide); and (2) method 3051 (microwave-assisted
digestion using nitric acid alone). Finally, we used pneumatic nebulization (PN),
ultrasonic nebulization (USN), and direct-injection nebulization (DIN) as techniques for
sample introduction to investigate efficiency for the combinations of instrument, digestion
method, and technique of sample introduction.
We digested progressively smaller soil aliquots in the range of 2.0 to 0.1 g by using both
acid digestion methods and analyzed these solutions for the RCRA elements that are
approved for each individual method. The precision and accuracy of the analysis, the cost,
the volume of secondary waste produced, a term for instrumental versatility (ratio of
analytes analyzed to analytes needed for analysis), and the initial sample weight required
were used to estimate and qualify the various methods tested.
The elemental recovery obtained from 2.0 g samples by using hot-plate digestion and
ICP/AES were 85% to 99% for elements such as Pb, As, Cd, Cr, and Ag. Samples under
2.0 g were not analyzed by this procedure due to the concentrations of many of the
elements studied. With the exception of Pb and Ba, concentrations for other elements were
closed to instrumental detection limits.
By using ICP/MS, samples of 0.1 g, 0.25 g, and 0.5 g were digested by hot-plate and
microwave digestion procedures. For most elements, the results showed statistically equal
recoveries when both digestion procedures were used. Percent recoveries of mercury were
highly variable. Barium showed low percent recoveries with both digestion techniques,
and the percent recovery of silver was better when the hot-plate digestion procedure was
used. In general, microwave digestion procedures should yield higher percent recoveries
for most elements compared with the hot-plate digestions.
168
-------
In addition to sample digestion procedures, PN, DIN, and USN were studied as sample
introduction techniques. Each sample introduction technique was evaluated in terms of
sample uptake, method detection limits, and time required per analysis. These parameters
were considered as part of the overall qualifications of the method.
The general trend observed was that the effectiveness of the method increases as the sample
size decreases. Smaller samples produce less secondary waste, reduce the quantity of
hazardous chemicals used, and lower the overall cost of the analysis. The conclusions
presented here are based upon analysis by ICP/AES and ICP/MS using PN, DIN, and
USN; however, sample introduction by electrothermal vaporization was not addressed
during this study. The cost savings are directly proportional to the reduction in waste
volume generated.
General Results
Regulatory considerations. Analyses for PCBs are most often conducted to comply with
regulatory guidance under several statutes, notably the Toxic Substances Control Act
(TSCA). Except in specific instances, TSCA rules do not specify the analytical methods to
be used. Because of this fact, dozens of standard methods have been issued (Erickson,
1986). Among the more broadly applied is the EPA's manual, Test Methods for
Evaluating Solid Waste (EPA, 1986; and any newer updates, editions, or revisions), which
has document number SW-846 and is commonly referenced by that alphanumeric
shorthand. This multivolume document contains methods to be applied under the RCRA
but is widely applied beyond the regulatory requirements. For most applications, SW-846
is a guidance document, and analyses using reasonable adaptations of the
published procedures are permitted. Except for four specific federal applications
and any imposed by states or other authorities, the manual is a "guidance document setting
forth acceptable, though not required, methods to be implemented by the user, as
appropriate, in responding to RCRA-related sampling and analysis requirements" (58 FR
46041, August 31, 1993), and "any reliable analytical method may be used to meet other
requirements under Subtitle C of RCRA" (abstract of SW-846, Revision 1, July 1992).
Lesnik (1992) clearly amplifies this by stating that "the manual is intended to be a collection
of flexible methods, suitable for adaptation to cover the wide range of analytical
applications and matrices required by the RCRA regulations. ... Examples of this
flexibility include adjusting sample sizes to fit the optimum analytical range of methods, or
using alternate glassware or equipment provided that method performance is not
compromised." The disclaimer in the manual states that "SW-846 methods are designed to
be used with equipment from any manufacturer that results in suitable method performance
(as assessed by accuracy, precision, detection limits, and matrix compatibility)" (disclaimer
of SW-846, Revision 0, July 1992). Similar text is found in Chapter 2 (Section 2.1.2,
Rev. 2, November 1992): "... glassware and supplies ... specified in these methods may
be replaced by any similar types as long as this substitution does not affect the overall
quality of the analyses" [emphasis added]. Thus, contrary to the beliefs of many of those
requesting or conducting analyses, PCS analysis according to verbatim SW-846
procedures is not required for many applications, and reasonable adaptations are permitted.
169
-------
CONCLUSIONS
We have demonstrated that P2 can pay off in the laboratory through the adaptation of
commonly used routine EPA methods for soils and oils. We have developed miniature
extractions and cleanups feeding into standard GC analysis for PCBs and demonstrated that
smaller sample sizes are preferable for RCRA metals. The "green chemistries" applied
include:
• Smaller sample sizes that match the instrumental consumption needs,
• Substitute solvents,
• Minimization of reagent consumption,
• Reduction of apparatus costs with microscale glassware, and
• Reduction of energy consumption.
The methods developed here have direct applicability to the routine PCB and RCRA metal
analysis laboratory. The applications presented here can be applied now without awaiting
regulatory approval.
The results reported here are but a few steps along the path toward implementing P2 in the
routine analytical chemistry. Much more remains to be done.
ACKNOWLEDGMENTS
Work supported by the U.S. Department of Energy, Assistant Secretary for Environmental
Management, Office of Technology Development, under contract W-31-109-Eng-38. We
wish to thank Ella Mulford (DOE-Chicago) for her support of this project and Jim Thuot
(Argonne National Laboratory, Environmental and Waste Minimization) for encouragement
and helpful suggestions. Pam Postlethwait and Amrit Boparai are thanked for assistance in
the chemical analyses.
REFERENCES
de la Guardia, M., and J. Ruzicka, 1995, "Towards Environmentally Conscientious
Analytical Chemistry through Miniaturization, Containment, and Reagent
Replacement," Analyst 120:17N.
EPA: See U.S. Environmental Protection Agency.
Erickson, M.D., 1986, Analytical Chemistry of PCBs, Butterworths Publishers, Boston,
Mass., 506 pp.; reprinted by Lewis Publishers, Chelsea, Mich. (1992); 2nd edition in
preparation.
Lesnik, B., 1992, "Perspectives on SW-846," American Environmental Laboratory
4(4):26-27.
U.S. Environmental Protection Agency, 1994, Test Methods for Evaluating Solid Waste,
EPA/SW-846, Office of Solid Waste and Management, Washington, D.C., Sept.
170
-------
34
CONTRACT LABORATORY PROGRAM (CLP) 2000 - A MODEL FOR PROVIDING
SUPERFUND ANALYTICAL SERVICES INTO THE NEXT MILLENNIA
Howard Fribush. Ph.D.r U.S. EPA Office of Emergency and Remedial Response.
401 M Street, SE, Washington DC 20460, 703/603-8831
R. Richard Thacker and Sean J. Kolb. DynCorp Inc.,
300 N. Lee Street, Alexandria, Virginia 22314, 703/519-1470
Abstract
With the impending reauthorization of Superfund, two themes have been prevalent throughout
public debate. These are: 1) the acceleration of the site cleanup process and 2) providing a better
link between the site cleanup activities and the risk to the public. By increasing the focus on these
two areas of concern, the EPA not only protects the public health and but can speed up the
economic recovery of the areas surrounding these sites including the "brownfield" sites located in
urban areas. These concepts can also be applied outside the context of Superfund to the
Department of Defense's base realignment and closure (BRAC) program and to initiatives at the
cleanup of the Department of Energy's laboratory complexes. However, in order to accomplish
these goals, access to a reliable source for obtaining the levels and types of contamination at these
sites in a cost effective and timely manner must be first be identified. While many new
improvements to existing field methods are being implemented, these procedures are still not as
reliable as those that can be provided by fixed off-site laboratories.
The Contract Laboratory Program (CLP) has been providing EPA users data of known and
documented quality to support Superfund site inspection, remedial, removal and enforcement
activities since 1980. During this time, approximately two million environmental samples have
been analyzed through this program. The CLP provides a community of commercial environmental
laboratories that perform a variety of chemical analyses to identify the levels and types of
contamination at these Superfund sites. The CLP is also supported by an infrastructure consisting
of technical EPA staff, procurement officials and support contractors for quality assurance and
tracking support.
This paper discusses why the centralized functions provided by the CLP is the most appropriate
mechanism for ensuring that reliable analytical data can continually be obtained to support
Superfund into the next century. It discusses the overall structure of the CLP, the services it
currently provides and the benefits the CLP can provide data users in the areas of cost-effective
analytical services, data management and quality assurance. It discusses several modifications to
the program which can be quickly implemented to enhance the effectiveness of these services in
terms of the delivery of analytical services, the electronic delivery of data, data review, and
communication. This paper also discusses the potential advantages to EPA by providing CLP
services to other government agencies. Data supporting this paper are obtained from the results of
various internal program reports and reports generated by outside organizations (i.e., GAO, OIG,
etc.).
171
-------
35
HANDLING NONLINEAR CALIBRATION PLOTS
Delphia F. Harris, Laboratory Director, Nicole Klueh, Student
Analyst, and Rose Noyes, Student Analyst, UIW Analytical Lab,
University of the Incarnate Word, 4301 Broadway, San Antonio,
Texas 78209.
ABSTRACT
There are well known analyte, matrix and instrumental
deviations from Beer's Law for absorbance vs concentration
calibration plots. If a nonlinear calibration plot is
obtained during method development or method validation, then
the validity of the plot roust be verified. It must be
established that the plot is 1) reproducible, 2) is not an
anomaly of standard preparation, and 3) accurately represents
the behavior of the analyte within the matrix of interest.
Once these criteria have been met, then nonlinear functional
forms can be used to analyze the absorbance vs concentration
relationship. Common functional dependencies include
quadratic and exponential fits. Alternative functional forms
are presented including the tools necessary to obtain an Ra
fit in order to meet Quality Assurance criteria. If there are
more than two adjustable parameters in the functional form of
the calibration plot, then this must be accompanied by an
increase in the number of standards used in the calibration.
INTRODUCTION
It is accepted practice to use a curve and not restrict
analysis to the linear range as long as standards bracket all
sample concentrations and there are enough standards such that
the curve is well characterized. R2 fits are then based on
the fit to the curve rather than a line. Increasingly,
instruments are computerized and have built in regression
capabilities for calibration curves. What happens if a
reproducible relationship between absorbance and concentration
for standards is not consistent with the instrument's curve
fitting software?
First, troubleshooting must be done to explore the possible
sources of the atypical calibration curve. Aspects of
standard preparation which may contribute to the atypical
calibration curve must be investigated. This may result in a
typical calibration curve.
Second, if these investigations do not produce a typical
calibration curve, then other mathematical models may be more
appropriate. Many instruments use a quadratic function to fit
absorbance calibration curves. Alternative equations will be
provided which include an "S-curve" fit.
172
-------
TROUBLESHOOTING NONLINEAR CALIBRATION CURVES
When performing an analysis which is new to the analyst, there
is much to learn about the method of analysis, the analyte,
and the matrix of the analyte. Clearly, analytical methods
are developed which take into account the nature of each
analyte. If adapting a method to an analyte, the method must
be validated even if the analyte is chemically very similar to
species for which the method was developed. An entire method
validation is beyond the scope of this paper. This discussion
will be limited to calibration plots.
Figure 1 provides an example of an atypical calibration curve.
This figure illustrates a flame atomic absorption calibration
plot for cadmium in water. The relationship between
absorbance and concentration is not well represented by either
a linear or a quadratic function. The absorbance readings of
the low concentration standards fall well below those expected
in a typical calibration plot. If it is determined that such
a profile is reproducible, then an investigation of possible
causes of such an atypical calibration curve would be
necessary.
The first possible cause to consider is the volumetric
glassware and dilution scheme used in the preparation of the
low concentration standards. These same concentrations can be
prepared by using an intermediate stock in Border to verify
that the standards are of the reported concentration. If the
results are consistent with the atypical curve then other
considerations must be investigated.
A calibration curve must accurately represent the response of
the instrument to the analyte within the matrix of interest in
order to provide quantitative results. The Standard Addition
Method can be used to test the concentration dependence of the
analyte absorbance within the sample matrix. Figure 2 clearly
demonstrates that the atypical calibration curve in Figure 1
Bust be an anomaly of standard preparation. The absorbance vs
concentration plot for cadmium within the sample matrix is a
straight line.
Figure 3 is also a calibration plot for cadmium in water. The
only difference in the preparation of this set of standards
and those in Figure 1 is that the standards were diluted to
the mark with 3% nitric acid rather than with water. Since
the cadmium was dissolved in acid in order to prepare the
stock solution, the low concentration standards demonstrate
the effect of not only the dilution of the analyte, but also
the dilution of the acid concentration. Diluting with nitric
173
-------
Concentration Least Squares
0.35H
A
B
S
O
R
B
A
N
C
E -|
0.00
R fit: 0.9745
max. error: 0.08701 ug/ml
0.0
1.0
CONCENTRATION OF Cd ug/ml
FIGURE 1: Flame Atomic Absorption calibration curve for
cadmium. The standards were prepared by dilution
with reagent grade water.
0.40H
Standard Additions
A
B
S
O
R
B
A
N
C
E _
0.00
Correlation Coefficient : 0.999
Sample Concentration : 0.01382 ug/ml
0.0
1.0
CONCENTRATION OF Cd ug/ml
FIGURE 2: Flame Atomic Absorption calibration plot for cadmium
using the standard additions method.
174
-------
0.40H
Concentration Least Squares
R2 fit: 0.9999
max. error: 0.00700 ug/ml
1.0
CONCENTRATION OF Cd ug/ml
FIGURE 3: Flame Atomic Absorption calibration plot for
cadmium. Standards were prepared by dilution
with 3% nitric acid.
acid maintains a near constant acid concentration for all of
the standards and also matches the concentration of acid in
the samples resulting from the digestion procedure.
The comparison of Figures 1 and 3 provides a vivid reminder
that standards must be identical in every respect except
concentration of analyte. The effect of acid dilution is not
as pronounced with other metals as it is with cadmium. This
is a reminder of the unique interactions of each analyte.
MATHEMATICAL MODELS FOR NONLINEAR CALIBRATION CURVES
Some calibration plots which can be validated are not linear.
Many computerized instruments now come with data fitting and
regression analysis capability. Typically, a quadratic
function is used for the data fit, in which case, a linear
relationship simply determines a coefficient of zero for x3.
Use of a quadratic model requires a greater number of
standards to adequately characterize the calibration curve.
Standard Methods1 states that for atomic absorption analysis
of metals a calibration curve should be "composed of a blank
and three or more standards (depending on instrumentation)..."
Three standards and a blank, or four points, is the minimum
needed to determine a linear relationship. Since a line can
175
-------
be placed through any two points, four points would provide
only two additional points for the regression analysis.
A quadratic curve can be placed through any three points.
Therefore, a calibration plot with only four points would not
adequately determine the regression curve. Five standards
plus a blank, or six points, are needed to obtain a good
calibration curve using a quadratic model, or any model with
three adjustable parameters.
The most common regression analysis, or determination of an R2
fit, is based on a linear model. The least squares fit to
data is the most familiar approach. A regression analysis of
a quadratic curve can also be done by least squares when the
equation is transformed to a linear form2. The quadratic
equation
Y = ao + bx(X - X) + ba(X - XV Eq. 1
can be transformed with the relationships Xx = X - X, and
Xa = (X - X")a/ into the linear equation
Y = a0 + bxXx + b^ . Eq. 2
Equation 2 is now in the form of a multiple regression
equation, and an Ra fit can be obtained for the transformed
equation .
A variety of other mathematical models can also be treated in
the standard linear form using multiple regression by a
suitable transformation of the variables3. For example, an
exponential model of the form
Y = e^**bx*» Eq. 3
can be transformed to a linear model by taking the natural
logarithms of both sides. This results in
InY = bc + bA + baXa Eq. 4
The residuals which result from a regression analysis such as
this are for the transformed responses, not the original
instrument responses.
As was seen in Figure 1, a curve may not be adequately fitted
by the curve fitting software which comes with the instrument.
Figure 1 represents an anomalous calibration plot which has
distinctive "s-shaped" character.
176
-------
1.0 H
Concentration Least Squares
i
N
T
E
N
S
I
T
Y
0.0
R fit: 0.9941
max.error: 26.54941 ug/ml
500
CONCENTRATION OF K ug/ml
FIGURE 4: Flame Atomic Emission Calibration plot for high
concentrations of potassium.
Figure 4 illustrates an actual calibration plot which has some
s-type shape to the curve. It is not nearly as pronounced as
that in Figure 1. The shape of the K calibration plot is
attributed to effects of ionization and self-absorption*.
This particular calibration plot has the best fit using a
quadratic model. However, s-type curves can also be fit using
a model which can be transformed to a linear equation.
The mathematical model which exhibits an s-type shape is given
by the equation
Y = aX2/(X2 + b)
Eq. 5
This equation involves only two adjustable parameters. A
third parameter can be introduced for the coefficient of X2 in
the denominator, however, when the equation is transformed to
a linear model, only two parameters can be solved for
independently. The transformation requires taking the
reciprocal of both sides of Eq. 5. This transformation
requires a positive value for the blank as it is undefined
when Y=0. The transformation of Eq. 5 to linear form requires
first taking the reciprocal of both sides.
1/Y = (Xa + b)/(aX2) « I/a + b/(aX2) Eq. 6
Setting X1 = X2, the transformed linear equation is
1/Y = I/a + b/(aX,) Eq- 7
177
-------
Since this transformed linear equation has only two adjustable
parameters, a and b, it has the same requirements for the
number of standards as a linear fit.
SUMMARY
In conclusion, nonlinear calibration plots can be justified,
however, it must be validated that they accurately represent
the response of the instrument to the analyte within the
matrix of interest. If a nonlinear calibration is used, then
it may require a greater number of standards to provide a good
calibration. Five standards plus a blank are needed for
functional forms with three adjustable parameters.
Instruments typically use a quadratic fit when the data is not
linear. This is adequate for most applications, however,
additional mathematical models, an exponential and an s-type
curve, have been presented. These mathematical models can be
transformed to a linear equation which fits the form of a
standard multiple regression.
ACKNOWLEDGEMENTS
This project received partial funding from a University of the
Incarnate Word Program Development Grant.
REFERENCES
Standard Methods for the Examination of Water and Wastewater.
Edited by Andrew D. Eaton, Lenore S. Clesceri, and Arnold E.
Greenberg, 19th edition, Prepared and published jointly by
American Public Health Association, American Water Works
Association and Water Environment Federation, (Washington,
D.C., 1995).
'Applied Statistics: Analysis of Variance and Regression, by
Olive Jean Dunn and Virginia A. Clark, John Wiley & Sons,
Wiley Series in Probability and Mathematical Statistics (New
York, 1974).
aApplied Regression Analysis. Second Edition, by N. R. Draper
and H. Smith, John Wiley & Sons, Wiley Series in Probability
and Mathematical Statistics (New York, 1981).
*Principles ^f Instrumental Analysis. Fourth Edition, by
Douglas A. Skoog and James J. Leary, Saunders College
Publishing, (San Antonio, 1992).
178
-------
36
Comparability of Commercial Environmental Reference Standards
Principal Authors: Dave Henderson, Kenneth J. Herwehe, Supelco Inc., Supelco Park, Bellefonte, PA 16823
phone 814-359-5446 fax 814-359-5750
Jack Criscio, Absolute Standards, PO Box 5585, Hamden, CT 06518
phone 800-368-1131 fax 800-410-2577
Abstract:
It is widely recognized that commercial environmental reference standards vary considerably from manufacturer to
manufacturer. Because of this, laboratories are forced to purchase multiple versions of a similar products from different
vendors and prove that they are equivalent, yet independent, standards. This added laboratory cost is perceived as a
necessary burden practiced to meet the quality assurance requirements of SW-846 and other regulatory methods. While
costs differ considerably from laboratory to laboratory, eliminating the costs of useless standards inventory and non-
revenue generating testing is beneficial to any laboratory operating in today's soft economic climate.
One approach which will minimize the need to repeatedly purchase similar products from multiple vendors, thus reducing
the level of non-revenue generating testing, involves intervendor comparison of standards by commercial producers before
sale. However, it has historically proved difficult to bring competing manufacturers together to accomplish this goal.
Subsequently, existing sources of commercial reference standards are different enough to drive the need for additional
testing on the part of the laboratory.
This paper reviews comparability between two commercial environmental standards vendors. Presented are the results of
a several month project designed to establish and perpetuate comparability between two commercial environmental
standards sources. Comparability of two commercial sources of standards by compound class and the variability across
product lines will be discussed. Data are to be presented illustrating the degree to which environmental standards from
independent commercial different vendors match and do not match. The presentation will present a recommendation on
how commercial vendors may cooperate to minimize the cost to commercial laboratories consuming environmental
standards through incorporation of Reciprocal Data Review in to product quality assurance programs.
179
-------
37
IMPROPER USE OF SIGNIFICANT FIGURES AND NUMBER ROUNDING BY
LIMS JEOPARDIZES CHEMICAL DATA USABILITY
Garabet H. Kassakhian. Ph.D.. Quality Assurance Director, Tetra Tech, Inc., 670 N. Rosemead
Boulevard, Pasadena, California 91107-2190
ABSTRACT
Chemical data usability is inherently based on the proper usage of significant figures and the
application of the principles of number rounding. When a Laboratory Information Management
System (LIMS) is commissioned or modified, particular attention must be paid that all calculations
that generate data take into account the cardinal rules of significant number usage in mathematical
operations. Figures reported in the Quality Control Data section of a Sample Delivery Group
(SDG) must allow for the verification of the calculated percent recovery and relative percent
difference (RPD) values, without triggering an on-site raw data audit. The paper compares the
way several different analytical laboratories report identical data for chemical analyses; it
discusses recent occurrences of jeopardized data during the calculation of spike and surrogate
percent recoveries. Since modification of the LIMS system is onerous once the system is in place
and processing data, it is suggested that laboratory management can avoid the expense and
embarrassment of hand entering the correct data and issuing replacement SDG pages by exercising
strict up-front quality control on the first sample SDG generated by the new or modified LIMS
system. The laboratory's credibility is jeopardized when no action is taken even after the client
alerts the proper quality assurance personnel. It is proposed that continuous trend analysis of
SDGs being produced for various clients and programs be used as a means of ensuring that the
LIMS system and its programmers do not deviate in their proper usage of significant figures.
INTRODUCTION
All laboratories use a form of Laboratory Information Management System (LIMS) to collect and
manage data. There is a proliferation of operating system and application softwares, database
management systems, automated data collection and processing instrumentation. A LIMS system
makes use of all of these in a plethora of available configurations based on the specific needs of
the laboratory. The U.S. Environmental Protection Agency (EPA) defines the LIMS as an
automated laboratory system where the user has "the ability to effect changes to original
observations or measurements."1 Historically LIMS systems were created for the manufacturing
industry, not to collect and process data from environmental laboratories. Most laboratories
purchasing a LIMS, had to custom program it to handle a number of very important processes
such as calculations, number rounding, proper use of significant figures, etc. These functions do
not come in prepackaged versions. For a laboratory using an older system, the programming
becomes a major undertaking since the system may have "maxed out" and may not be able to
accept so many new codes. Besides managing data the LIMS is also used to generate data reports,
a.k.a. Sample Delivery Groups (SDG). Koller and Liesegang recognized as early as 1993 that the
LIMS was limited in its ability to handle significant figures; the data management systems did not
have the calculation capabilities that laboratory instrumentation had.2 Industry sources claim that
as recently as two years ago LIMS systems addressed decimal figures only and were not
programmed to tackle significant figures3
180
-------
SIGNIFICANT FIGURES AND ROUNDING OFF NUMBERS
Since every measurement involves an estimate, numbers resulting from measurement are qualified
as not exact, while numbers that can be counted are considered to be exact numbers, e.g.. the
number of pages in a book, the number of people in a room.
When a competent person makes a measurement, the last digit of the measured number is
considered to be a best estimate. If further experiments or measurements were conducted this digit
is die one that might change. All the other digits of the measured number will not change. Such a
number contains only significant figures, including the last estimated digit. An exact number is
one with unlimited significant figures.4"5>6'7
The number of significant figures that one uses for a chemical measurement is detemined by the
uncertainty inherent in the equipment and instrumentation used to obtain the measurement.
Although modem electronic instrumentation and computers may use a bewildering array of digits
in their automated calculations, it is an axiom of analytical chemistry that analytical results can be
only as accurate as the information that leads to them. The use of digits beyond the acceptable
number of significant figures is unnecessary. Hence the need for the proper rounding of significant
figures.
Accuracy of a measurement indicates how close the latter is to the correct value, while precision
indicates how close are the measurements to each other. Highly accurate measurements can not be
imprecise, although highly precise measurements can be inaccurate. In both instances the role of
proper rounding of significant figures can not be underemphasized. Accuracy and precision are
numerically dependent on the number of significant figures used in the calculations and the manner
in which they are rounded off.
RULES OF USAGE
The following are the simple and well-known rules of usage of significant figures and for rounding
offofmeasurements:4'6'7
1. Zeroes used to position the decimal point are not significant figures, e.g., 0.000123 contains
only three significant figures; it could also be written as 1.23x 1O"4.
2. Zeroes to the right of digits are considered significant figures when a decimal point is placed
after the number, e.g., 120. contains three significant figures, while 120.00 contains five
significant figures.
3. The answers of multiplication and division operations can contain no more significant figures
than the least number of significant figures used in the operation. For example, a 20 ug/L
spike contains one significant figure, its recovery of 24 ug/L contains two significant figures.
24
The percent recovery can not be expressed as —xlOO= 120%, since 120 contains two
significant figures. Unless 20 is expressed as 20. or 20.0 the answer should be rounded off to
100% with a significant loss of information.
181
-------
4. In addition and subtraction the number of significant figures in the answer is determined by the
least number of figures to either the right or left of the decimal point.6
5. During complex calculations involving multiplication or division and addition or subtraction,
the operations are performed serially, and the final result is rounded off to the least number of
significant figures involved.6
6. During rounding off when 6, 7, 8, or 9 is dropped, the preceding digit is increased by one unit;
if the digit 0, 1,2, 3, or 4 is dropped the preceding digit is not altered. If the digit 5 is dropped,
the preceding digit is rounded off to the nearest even number. While 4.45 becomes 4.4, 4.55
is rounded to 4.6.7
7. When a group of two or more figures is to the right of the last significant figure, the number is
considered as a group, such as in 3.4(503) the 503 is considered >5, while in 3.4(498) the 498
is considered to be <5. This is a very important consideration to remember while programming
your new LIMS.
DISCUSSION
The EPA clearly mandates that "computer programs used for data reduction should be validated
before use and verified on a regular basis", and that "analytical results (be) reported with an
appropriate number of significant figures"* Acceptance criteria for the upper control limits for
surrogate and spike recoveries in Laboratory Control Samples (LCS) and Matrix Spike and Matrix
Spike Duplicates (MS/MSD) are expressed in three significant figures; therefore all calculations
that result in numbers that are compared to these limits should always have at least three
significant figures.
The EPA started the trend of disregarding the usage rules of significant figures by lackadaisically
suggesting "spike with 20 ug/L," or proposing control limits with a mixed number of significant
figures, e.g. 80-120%, or ±30% (1). Unfortunately this approach still continues to plague the
analytical community, the most recent example being the brand new February 1996 version of the
Air Force Center for Environmental Excellence (AFCEE) Quality Assurance Project Plan.9
An SDG should contain enough information to allow the user to calculate the reported values, such
as percent recovery and relative percent differences (RPD), from the data provided. This includes
reporting the actual spiking concentrations and the recovered concentrations. Anything less
invalidates the data, is not legally defensible, and, of course, triggers a raw data audit. Even
greater confusion and consternation will result in a few years' time when the archived raw data are
not available anymore.
Most laboratories conscientiously use an adequate number of significant figures in their
calculations of surrogate and spike recoveries. However a nationally known midwestem
laboratory(KP) submitted voluminous data similar to the ones reported in Tables I and II, as
recently as December 1995. The SDGs carried the following misleading disclaimer: "The Quality
Control Sample Final Results listed above have been rounded to reflect an appropriate number
of significant figures. Consistent with EPA guidelines unrounded concentrations have been used
to calculate % Rec and RPD values". Upon closer scrutiny it was discovered that similar
inconsistencies in the reported results affected the LCS and MS/MSD results of almost all the
analytical methods under consideration, i.e., total petroleum hydrocarbons (diesel/gasoline),
182
-------
pesticides and polychlorinated biphenyls (8080), volatile organics (601 and 8260-modified list),
semivolatile organics (8270), metals, etc. The laboratory's LIMS used up to six digits (not all of
them significant figures) to calculate the percent spike and surrogate recoveries, which were then
consistent with the raw data. However the same LIMS reported the spike and surrogate values and
their recovered values after rounding them off to two significant figures, resulting in the
discrepancies noted, in Tables I and II. For example, the diesel MS/MSD spike recoveries in Table
II are both reported as 12 mg/kg but the RPD is reported as 6%, because the LIMS calculates the
RPD based on the percent recoveries - otherwise the RPD will be 0%. When the laboratory
Table I
Quality Control Data for Laboratory Control Sample (Method 8260/Water)
KP Laboratory (Midwest US)
Parameter
Chloromethane
Bromomethane
Vinyl Chloride
Chloroethane
Methylene Chloride
Acetone
1,1-Dichloroelhene
1,1-Dichloroethane
trans- 1 ,2-Dichloroethene
Chloroform
1,2-Dichloroe thane
2-Butanone
1 , 1 , 1 -Trichloroethane
Carbon Tetrachloride
Brornodichloromethane
1,2-Dichloropropane
trans-1 ,3-Dichloropropene
Trichloroethene
Dibromochloromethane
1, 1,2-Trichloroethane
Benzene
cis-l,3-Dichloropropene
Bromoform
4-Methyl-2-Pentanone
Tetrachloroethene
1. 1,2,2-Tetrachloroethane
Toluene
Chlorobenzene
Ethyl Benzene
Styrene
Xylenes (total)
l,2-Dichlorethane-d4 (S)
Toluene-d8 (S)
4-Bromofluorobenzene (S)
Spike Cone.
ug/L
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
60
No datum
No datum
No datum
LCS Result Lab Reported Spike
ug/L % Recovery
19
17
17
17
19
17
17
18
17
17
15
24
17
16
18
19
17
18
18
19
20
17
18
23
17
21
20
18
17
18
52
No datum
No datum
No datum
96
84
87
86
97
87
86
88
84
84
77
119
87
82
92
94
85
92
88
93
100
87
89
117
87
103
98
88
86
88
87
87
106
102
Client Calculated
Spike % Recovery
95
85
85
85
95
85
85
90
85
85
75
120
85
80
90
95
90
90
95
85
90
115
85
105
100
90
85
90
183
-------
programmed its LIMS to take into account three significant figures, the MS and MSD percent
recoveries became consistent with the reported values. It also became obvious that the RPD must
be calculated from the actual recovery values rather than using already rounded percent
recoveries.
Table II
Matrix Spike/Matrix Spike Duplicate Recoveries and RPD for Diesel Fuel (Method OA2/soil)
Using Two and Three Significant Figures
Spike
Cone.
mg/kg
17
16.5
MS result
X, mg/kg
12
11.6
MS Spike
% Rec.
70
70
MSD result
X2 mg/kg
12
12.2
MSD Spike
% Recovery
74
74
Lab RPD
from%
Rec.
6
6
Calculated
RPD from
X, and X2
0
5
The laboratory had failed to exercise adequate quality control on its products because it considered
this practice acceptable. This was compared to SDGs from seven nationally known laboratories as
shown in Table III. The majority used an adequate number of significant figures in their reports
and calculations. The DC laboratory even reported all acceptance limits with six significant
figures. CT1 on the Pacific Coast used only two significant figures and yet the reported percent
recovery and RPD were consistently justified.
Table III
Number of Significant Figures Used in LCS and MS/MSD Percent Recovery Reports
by Various US Laboratories
Laboratory
QP (US East Coast)
DC (US West)
PHB (Pacific Coast)
CTI (Pacific Coast)
CTB (Pacific Coast)
LA (US West)
SL (Midwest US)
KP (Midwest US)
No.of Significant
Figures Used
4
4-5
1-4
2
4
4
4
2
Actual Spike and
Recovery Values
Reported
Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
Reported Data and %
Recovery Are Consistent
Yes
Yes
No
Yes
Yes
Yes
Yes
No
The KP laboratory acknowledged that its ChemWare Horizons™ LIMS had been modified from a
commercial application using an Oracle™ database. It did not address QC adequately. The LIMS
had been programmed to round off to two significant figures, which meant that, in a worst case
scenario, 95 could be rounded to 100. If three significant figures had been used then 99.5 will be
rounded to 100 - a major reduction in uncertainty. Although the laboratory was equipped with
184
-------
Hewlett Packard™ instrumentation capable of reporting up to four decimal places and transmitting
the raw data measurements directly into the LIMS, a special code had to be added to report
surrogate recoveries with three significant figures, hi fact, each analytical method had to be coded
separately into a QC reference file.. A new code was not used in the production database until it
had been successfully tested on the laboratory's development data base. The laboratory was asked
to replace the pages with the correct information; this request was only partially fulfilled by hand
entering the most glaringly erroneous calculations, leaving a pile of invalid data pages still to be
accounted for.
The PHB laboratory reported recoveries using from one to four significant figures, depending on
the method. It did not report the actual recovered concentrations. The percent recoveries contained
minor discrepancies of up to 1%, that only an on-site raw data audit could clarify'.
In another instance the LIMS of the CTB laboratory had been programmed to round off
intermediate results for percent moisture calculations. This resulted in up to 5% discrepancies
from the results that were obtained when rounding took place only at the conclusion of the
mathematical operations.
SUMMARY
It is obvious that all automated calculations should be carried through with the maximum
permissible number of significant figures, without intermediate rounding off of the results of
mathematical operations. The principles of usage for significant figures and rounding off numbers
must be scrupulously adhered to. When results are close to the outer limits of acceptance criteria
even a few percent deviation may affect the acceptance or rejection of data.
All LIMS users should implement the principles and guidance enunciated in EPA's 1995 Edition of
Good Automated Laboratory Practices (GALP).'
When a new LIMS is installed, or an old one upgraded or modified the laboratory quality
assurance unit (QAU) should conduct an in-depth audit of the LIMS for each analytical method
and the SDGs that are produced using the LIMS data. The LIMS programmer must address the
fact that uncertainties inherent in instrumentation and methodology, including preparation and
extraction, are accurately reflected in the proper use of significant figures and the rounding off of
fa final data.
To minimize erroneous RPDs the latter must be calculated from actual recovery values, i.e.,
concentrations, not from already rounded off percent recoveries.
Laboratories must use trend analysis of their final product - the SDGs, to quality control the LIMS
performance for accuracy and consistency. Reports from clients regarding inconsistent SDG data ,
should be immediately investigated, as they may be indicative of LIMS problems that may have
escaped the QAU audits.
185
-------
ENDNOTES
1. In 1980-1981 the EPA was involved in a historic imbroglio involving significant figures. A
highly placed EPA official(RL) responded favorably and in writing to the request from two
petroleum refining corporations to increase the concentration of tetraethyl lead (TEL) in
gasoline to the maximum concentration which (for monthly reporting purposes) could still be
rounded down to the EPA mandated two significant figure limit. The actual amount of TEL
emissions and the accompanying profits increased significantly while the refiners stayed within
the letter of the law. The ensuing congressional hearing ended up with the dismissal of the
official from her position and the imposition of a jail sentence.
REFERENCES
/. U.S. Environmental Protection Agency, "2185 - Good Automated Laboratory Practices -
Principles and Guidance to Regulations For Ensuring Data Integrity In Automated
Laboratory Operations with Implementation Guidance", 1995 Edition, p. 1-2, 10 August,
1995, USEPA, Research Triangle Park, North Carolina 27771
2. Roller, A.J., and G.W. Liesegang, " Managing Data in the Lab ", Environmental Testing and
Analysis, 2(4), pp.32-38. 1993.
3. Richardson, L., Telecation, Inc., Lakewood, Colorado 80235, personal communication, May
1996.
4 Whitten, K.W., and K.D. Gailey, R.E. Davis, "General Chemistry", Ch. 1, pp.22-25, Fourth
Edition, 1992, Saunders College Publishing, Harcourt, Brace Jovanovich Publishers, Orlando,
Florida 32887.
5_. Petrucci, R. H., and W.S. Harwood, "General Chemistry, Principles and Modern
Applications", Ch. 1, p.23, Sixth Edition, 1993, Prentice Hall, Englewood Cliffs, N.J.07632
-------
38
How to Use U.S. Environmental Protection Agency World Wide Web
and Listserv Internet Resources to Increase Environmental
Compliance and Reduce Environmental Compliance Costs
Leon Lazarus, U.S. Environmental Protection Agency, Region 2,
Edison, New Jersey 08837; Peter Savoia, U.S. Environmental
Protection Agency, Region 2, Edison, New Jersey 08837; Phil
Flax, U.S. Environmental Protection Agency, Region 2, New York,
New York 10278.
The U.S. Environmental Protection Agency (EPA) World Wide Web
(WWW) site has been recognized as one of the best sites on the
WWW. For example, McKinley's Magellan Internet Directory gives
the EPA WWW site a four star rating, which is its highest rating.
The EPA WWW site allows individuals to search EPA data bases,
review regulations, and focus explorations of EPA on: Key Word
Searches, Offices, Programs, Regions, Initiatives, News, Grants,
Contracts, Publications, etc. The EPA WWW site is remarkable
because it allows individuals myriad search preferences.
In addition to the WWW, EPA operates approximately 80 Internet
Listservs. These Listservs are not widely used because most
environmental professionals are not aware they exist. Listervs
are sometimes called Listervers, Internet Mailing Lists, and
Electronic Mailing Lists, but this paper will only use the term
"Listserv". Individuals who subscribes to an EPA Listserv
receive e-mail from EPA on one topic, such as federal register
notices of new and proposed hazardous waste regulations.
Although non-government Listservs allow subscribers to post notes
to all other subscribers, EPA Listervs do not allow subscribers
to transmit comments to other subscribers. EPA Listervs only
distribute e-mail generated by EPA.
Why are Listservs so helpful? Listservs are narrowly focused
because people only receive e-mail on topics they select. In
addition, listservs do not require anyone to spend time "surfing"
the net. All Listserv information is delivered via e-mail. The
following are examples of EPA Listervs, which anyone with outside
e-mail service can subscribe to. Subscribing to EPA Listservs is
extremely easy, and it is free.
mercury@unixmail.rtpnc.epa.gov (U.S. EPA MERCURY mailing list)
epa-press@unixmail.rtpnc.epa.gov (U.S. EPA Press Releases)
epa-meetings@unixmail.rtpnc.epa.gov (U.S. EPA FEDERAL REGISTER
documents for MEETINGS)
epafr-contents@unixmail.rtpnc.epa.gov
(U.S. EPA FEDERAL REGISTER CONTENTS documents)
epa-sab@unixmail.rtpnc.epa.gov
(U.S. EPA FEDERAL REGISTER SCIENTIFIC ADVISORY BOARD)
187
-------
39
Comparison of the Region 2 QA Program to the Air Force Center for
Environmental Excellence: CERCLA Outreach in Region 2
Amelia Jackson, U.S. Environmental Protection Agency, Region 2,
Edison, New Jersey 08837
Increased flexibility in designing and implementing a practical
environmental program has become an absolute necessity in this
era of shifting and tightening resources. At the same time, the
OIG has identified the need for improved QA oversight by EPA of
Superfund Federal Facilities, that is, smarter application of QA
principles, in their September 1995 audit report. Region 2 has
consistently maintained an open relationship with the QA
personnel from the Army Corps of Engineers as well as with the
Army Environmental Center, both of whom are responsible for many
federal facilities in this region. This has aided Region 2's
ability to determine the reliability of each entity's Quality
Assurance Programs in addition to ensuring the quality of the
environmental data produced.
An effort will now be undertaken to enhance this Region's
oversight responsibility to include the Air Force Center for
Environmental Excellence, the Quality Assurance "nerve center"
for AF projects. This initiative is intended to facilitate
communication with the federal facility, their consultants, and
the analytical lab, thus solidifying the QA partnership. In
addition, relevant QA/QC issues and concerns will be addressed
thus achieving appropriate and cost effective environmental
monitoring.
Region 2 QA staff have also developed a Home Page on the
information superhighway of the World Wide Web. This valuable
on-line resource will provide the ability to contact the
appropriate technical QA experts in Region 2 directly on-line.
Thus, the cross training of its nationwide audience to the EPA
Region 2 QA and the various Federal Facility QA Programs will be
easily accomplished.
188
-------
40
«QA in Cyberspace": The Region 2 Quality Assurance Outreach
Program For The '90s
Patricia Sheridan, U.S. Environmental Protection Agency, Region
2, Edison, New Jersey 08837; Peter Savoia, U.S. Environmental
Protection Agency, Region 2, Edison, New Jersey 08837
In an effort to keep Regional personnel informed of current U.S.
EPA Region 2 quality assurance policies, the quality assurance
(QA) office of the Environmental Services Division has developed
an outreach program suitable for the '90s, "QA in Cyberspace".
To promote the CERCLA and RCRA outreach efforts, the QA office
has uploaded numerous guidance documents electronically through
CLU-IN (formerly the OSWER Bulletin Board System (BBS)) and the
information superhighway of the World Wide Web (WWW). A recent
initiative of the Region 2 QA Office is the quarterly QA
newsletter via the Region 2 Home Page, which may be of particular
interest to WWW site users. The QA newsletter is intended to:
(a) facilitate communication with the regulated community,
environmental consultants, testing laboratories, and state and
federal regulators, (b) forum relevant Quality Assurance/Quality
Control (QA/QC) issues and concerns, and (c) promote appropriate
and cost effective environmental monitoring.
The USEPA Region 2 QA newsletter contains: (a) one or more
feature articles, (b) three ongoing task specific columns (Field
QA/QC Support, Laboratory QA/QC Support, & RCRA QA/QC support
notes) , (c) a QA fact or definition, (d) a summary of our
outreach efforts and upcoming training events, and (e) a listing
of active Contract Laboratory Program (CLP) Routine Analytical
Services (RAS) contracts. In addition, a brief biography of each
author is provided to supplement each article and establish
points of contact for the user.
The USEPA Region 2 QA newsletter will provide first hand QA/QC
information to anyone undertaking an environmental monitoring
project. This will enable one to access QA/QC support from
USEPA1 s many technical experts directly on-line. Working in
partnership with the regulated community, testing laboratories,
and consultants will ensure environmental compliance is
undertaken in a correct and cost effective manner. Increased
flexibility in designing and implementing a practical
environmental program is a current goal of the Agency. The
USEPA Region 2 QA newsletter is one of the Agency's many
innovative on-line information resources which can positively
impact the way we do business.
189
-------
41
NON-TRADITIONAL USE OF ENVIRONMENTAL SITE ASSESSMENTS:
A TOOL FOR DETERMINING POTENTIAL ENVIRONMENTAL
IMPAIRMENT/COMPLIANCE
Authors: Scott McCone. CHMM, Senior Environmental Scientist
Robert C. Najjar, Ph.D., Senior Environmental Scientist
URS Consultants, Inc., 282 Delaware Ave., Buffalo, NY 14202
ABSTRACT
Phase I environmental site assessments (ESAs) traditionally have been the tool used by buyers and
sellers to identify potential environmental liabilities associated with property transfers. However, ESAs
also can be used by property owners to assess their present and past site activities that may have resulted
in environmental impairments. The U.S. Environmental Protection Agency (USEPA) recently has
promulgated policy that encourages the voluntary discovery, disclosure, correction, and prevention of
regulatory deficiencies (60 FR 66706, December 22, 1995). The use of an ESA to compliment an
environmental compliance audit comprehensively addresses past, as well as present use in evaluating a
facility's regulatory compliance.
As part of a voluntary compliance audit program for the Owner, we performed ESAs at twelve large
Connecticut facilities involved in industrial activities. Current and past practices were identified and
evaluated for their possible impact on the environmental quality of the sites. The ESA process was
modified to meet the criteria of both American Society of Testing and Materials (ASTM) Standard E
1527-94 and Connecticut Transfer Act Site Assessment (TASA) guidance.
This case study presents the modified ESA process, as well as findings which included: asbestos pipe
insulation, lead paint, abandoned leach fields that received chemical wastes, leaking underground storage
tanks, and potential contamination of soil and groundwater. It also presents a discussion of
recommended corrective actions and management practices to prevent future deficiencies in the areas
identified. The Owner is committed to correct the problems that were identified.
INTRODUCTION
Environmental compliance is a key part of any industry's goals. Obeying the law is understood, but
managing industrial environmental issues in a proactive mode moves those enlightened firms into the
"Good Corporate Citizen" category. To that end, the Owner of twelve facilities which perform industrial
activities in Connecticut is in the process of having URS Consultants, Inc. (URS) render comprehensive
environmental compliance audits. In addition, the Owner wanted a review of potential environmental
impacts from past property uses and practices so that these potential impacts could be addressed
proactivery, using a program like USEPA's Voluntary Site Cleanup Program.
Most environmental compliance audits (EGAs) are a snapshot of compliance at a facility when the audit
is performed and historic uses/practices usually are not addressed. URS' approach to meet the Owner's
expectations was to perform Phase I ESAs to evaluate the Owner's more immediate need for identifying
potential environmental impairments for the sites, followed by the comprehensive ECAs.
190
-------
The ESAs were performed to ASTM Standard E 1527-94, which establishes minimum criteria for parties
seeking a definition of environmental due-diligence for Phase I ESAs. In addition, the Owner wanted
to comply with Connecticut Department of Environmental Protection (CTDEP) TASA guidance to
follow that agency's recommendations. The ASTM E 1527-94 standard allows for flexibility and
professional judgement in performing Phase I ESAs. Also, the ASTM standard is consistent with the
TASA guidance, provides more detail and guidance in performing proper Phase 1 ESAs, and has been
nationally adopted by the financial, commercial, and industrial sectors. The major departure of TASA
from typical ASTM Phase I ESAs was to require an in-depth review the federal, state and local
enforcement history of each site.
Newer facilities ranged in age from 2 to 5 years old; the older facilities dated back to use during the Civil
War. Sites varied in size from a large city block to several hundred acres. Typical current industrial
operations at these site included machine shops, metal fabrication, motor pool, farming, and
woodworking. The potential environmental impacts also varied with historic use; that is, older sites
used different fuel sources over the years for heating such as wood, coal, oil, and natural gas.
THE ENVIRONMENTAL SITE ASSESSMENT PROCESS
Prior to conducting the ESAs at the twelve facilities, URS personnel met with the Owner and outlined
die extent (and limitations) of the ESAs. URS prepared due-diligence site walkover forms that addressed
the requirements of both the TASA guidance and the ASTM standard, and submitted the walkover forms
to the client for review and comment prior to finalizing the forms.
Approximately two weeks prior to the scheduled site walkover inspections, the Owner provided URS
with their comments on the site walkover forms along with accurate site location maps and a contact
person at each facility. Since access to the facilities was restricted, URS personnel involved with the site
walkover inspections were required to obtain photo identification security passes.
Upon receipt of the site location maps, URS contacted the ERIIS database service1 to perform a federal
and state database search to identify and locate reported known environmental problems within the radii
recommended in the ASTM standard. ERIIS provided URS with information on sites listed on: the
federal National Priorities List (NPL) Site List; the federal Comprehensive Environmental Risk, the
Compensation and Liabilities Information System (CERCLIS) List; the federal Resource Conservation
and Recovery Act (RCRA) Treatment, Storage and Disposal (TSD) Facilities List; the federal RCRA
Generator List; the federal Emergency Reporting and Notification System (ERNS) List; the Connecticut
Hazardous Waste Site List; the Connecticut Landfills List; the Connecticut Leaking Underground
Storage Tank (UST) List; and the Connecticut Registered UST List.
URS verified the ERIIS database information by submitting Freedom of Information Law (FOIL)
inquiries to the appropriate federal agencies and conducting a manual file search of the CTDEP files.
Additional information was obtained by reviewing State of Connecticut Department of Public Works
(CTDPW) records for asbestos abatement information and local government files. The information
compiled helped identify areas of potential environmental concern, disclosed historic environmental
compliance enforcement actions, and provided information on investigations involving spills of
191
-------
hazardous materials and hazardous wastes. Copies of appropriate documents were obtained for inclusion
in the final ESA reports.
In addition to environmental information, URS also developed historic information and property
ownership information. Historic information was developed by reviewing aerial photographs (when
available), Sanborn fire insurance maps, and cross-reference telephone directories. Property ownership
information, in general, was established to the early 1900s. However in one instance the ownership of
one of the properties was tracked to the early 1800s. Interviews also were conducted to help verify and
provide additional detail to the significant/relevant historic and property information. URS used U.S.
Geological Survey (USGS) 7.5 minute topographic maps to describe the physical setting of each
property. Soil survey reports, geological maps, and water resource maps also were utilized to assess the
environmental setting at each property.
URS personnel experienced in conducting ESAs completed the site walkover inspections. Prior to
conducting the site walkover inspection, facility personnel knowledgeable in the operation of the facility
were identified and due-diligence interviews were conducted. Information obtained during the records
search was provided to facility personnel who then addressed questions concerning historic hazardous
materials and hazardous waste handling and disposal practices. Site maps indicating the locations of
USTs, aboveground storage tanks (ASTs), historic waste disposal areas, and active hazardous material
storage areas were prepared prior to conducting the site walkover inspection. URS also reviewed
available engineering drawings to note the possible presence of historic subsurface structures, such as
past leach fields or landfills, that may be of concern.
The site walkover inspections included an inspection of both the interior of any onsite structures and the
surrounding exterior of each property. The URS team made note of adjoining property uses and general
land use of the surrounding area, hi addition, any land uses that had the potential for impact on/near the
subject properties were identified.
Following the completion of each site walkover inspection, copies of the due-diligence assessment form
were made for facility personnel. Upon the return of the URS team to their office, individual ESA
reports were prepared that were consistent with the TASA guidance document and ASTM standard.
Since the ESA was conducted to identify potential historic problems at the facilities, any areas where the
potential for historic environmental impairment or current environmental compliance were identified and
an aggressive Phase n ESA sampling program was recommended to determine the presence, nature and
extent, and impact of a potential environmental hazard.
FINDINGS AND DISCUSSION
The findings and deficiencies identified at each of the facilities, as well as recommendations to correct
any deficiencies, were presented in individual ESA reports prepared for each facility. Following a review
of the individual ESAs by the Owner, a formal presentation was made to the Owner that summarized the
ESAs findings and recommendations.
Hazardous Material Issues: The types and quantities of hazardous materials varied by facility. In
general, limited quantities of hazardous materials for building maintenance (such as paint and cleaning
192
-------
compounds) were present in each building. Buildings housing industrial activities contained the greatest
variety of hazardous materials including oil-based paints, degreasmg solvents, inks, chlorinated solvents,
flammable liquids, corrosive liquids, and oils. Heat at the majority of the facilities was provided by
oil-fired boilers.
Areas where hazardous materials were stored and used were inspected to determine if spills would be
contained or controlled. The majority of the non-industrial storage areas inside the buildings did not
have floor drains and there were no indications of spills. Also, floor drains were not present or had been
sealed in the majority of the storage buildings associated with industrial activities. Buildings that had
floor drains discharged into the sanitary sewer line. One of the storage buildings was found not to be
properly ventilated which resulted in a buildup of flammable vapors; the building also was found not
to meet the building code for electrical wiring.
Except for three facilities, USTs were used for the storage of fuel oil for the boilers and emergency
electrical generators. Records disclosed that a few of the USTs had been taken out of service after leaks
were discovered. While monitoring wells were installed near a few of the USTs, the facilities generally
depended on inventory controls to determine if a UST has a possible leak. URS recommended that the
soils adjacent to the leaking USTs be tested to determine if residual petroleum products were present.
URS also recommended that a facility-wide UST leak detection program be developed, or the USTs be
removed and replaced by ASTs. Since a significant number of the USTs had not been registered with
flic CTDEP, URS recommended that all USTs be registered. At facilities with ASTs, no significant oil
spills were noted inside the secondary containment structures. Since it was discovered that spill
prevention control and countermeasure (SPCC) plans were not prepared for these facilities, URS
recommended that SPCC plans be prepared in accordance with 40 CFR 112.
Wastes- Industrial wastes generated at the majority of the facilities consisted of waste oil,
paint wastes, and air compressor blowdown fluids. These industrial wastes, in most instances, were
shipped oflsite for recycling. Designated hazardous waste storage areas met the RCRA criteria for small
quantity generators as outlined in 40 CFR 262. Even though recycled industrial wastes are not
considered hazardous wastes, copies of non-hazardous waste manifests were kept on file at most of the
facilities.
One of the older facilities recently replaced ten old electrical transformers which had been placed
temporarily outside the designated RCRA storage areas. While dielectric fluid samples were collected
from each transformer and tested for polychlorinated biphenyls (PCBs), none of the transformers had
been marked as required in 40 CFR 761. URS recommended that the transformers be marked as
"NON-PCB" and disposed of as non-hazardous waste. Other hazardous wastes stored at this facility
included 5-gallon drums of a chlorinated solvent (perchloroethylene) and lead-acid batteries. URS
recommended that arrangements be made to dispose of the perchloroethylene at a RCRA TSD facility
as soon as possible. URS also recommended that procedures be developed to ensure that all hazardous
wastes brought to the RCRA storage area be characterized, stored, marked, and labeled in accordance
with 40 CFR 26 1-262, and that wastes be removed from the facility within 90 days.
Discharges to the Environment: Spills of gasoline and fuel oil from USTs represented the greatest
percentage of discharges to the environment. However, historic waste disposal practices represent the
193
-------
most significant potential environmental impacts. Sanitary sewer lines at two of the facilities discharged
into leach fields before being connected to regional sewer systems. Historic disposal of hazardous
materials and wastes into the sanitary sewer system may have resulted in hazardous materials percolating
into the groundwater. Results of historic groundwater sampling down-gradient of the former leach fields
contained chlorinated solvents and other organic and inorganic materials. URS recommended that soil
and groundwater samples be collected in the vicinity of the abandoned leach fields to determine the types
and concentrations of organic and inorganic materials. The analytical results will be compared against
the CTDEP and federal water quality standards.
Historic solid waste disposal areas, some of which covered 3 acres, also represent a potential source of
groundwater and surface water contamination. Surface water, groundwater, and sediment samples
collected in and around the former solid waste disposal areas were found to contain high levels of heavy
metals. URS recommended that additional soil and groundwater samples be collected from the former
solid waste disposal areas and that hydrogeologic information be obtained to better define the potential
for oflsite migration of contaminants. The analytical results will be compared against the CTDEP and
federal water quality standards.
At one of the facilities, boiler cooling water that had been treated with biocides was being discharged to
the storm sewer which emptied into a nearby pond. URS recommended that the boiler cooling water be
tested and that arrangements be made to discharge the boiler water into the sanitary sewer lines. Other
uncontrolled or unpermitted discharges included a paint spray booth discharge and an air compressor
blowdown onto the ground.
Health Issues: The ESAs disclosed that some of the facilities still had asbestos-containing materials
(ACM) as insulation on hot water and steam pipes; possible lead-based paint on walls, ceilings, and
fixtures; and possible PCB-containing electrical ballasts. URS recommended that the insulation on the
pipes be tested to determine if it posed an airborne hazard and that any ACM be removed. URS further
recommended that paint samples be collected and tested for lead and that information on the electrical
ballasts be obtained to determine if the ballasts contained PCBs.
Permits and Registrations: The types of permits required varied greatly for each facility. The records
of facilities with oil-fired boilers were checked to determine if air discharge permits had been obtained
and were current Other point sources of air discharges in manufacturing areas also were identified and
the records checked to determine the monitoring requirements for these discharges. At one of the older
facilities, it was discovered that a paint booth was not permitted; it was recommended that a permit be
obtained prior to using the paint booth.
A review of the facility records disclosed that a number of the older USTs had been taken out of service
and either replaced with new USTs or ASTs. The majority of the facilities had registered their USTs
when Connecticut initially required registration. However, a significant number of the facilities had not
informed the state when USTs had been taken out of service. URS recommended that a
organization-wide UST management program be instituted to correct the discrepancies, maintain current
registrations, and verify UST compliance.
194
-------
CONCLUSIONS AND SUMMARY
As seen in the process and findings sections, ESAs compliment EGAs with some overlap. ESAs and
EGAs both deal with the site's current use, storage, handling and disposal of hazardous materials.
However, ESAs also focus much more on historic uses and practices and less on recordkeeping than
EGAs.
The Owner's need for an assessment of potential environmental impacts from past and present uses and
activities was met efficiently and effectively by performing Phase I ESAs on the twelve industrial activity
sites. This proactive approach by the Owner will result in several Phase II ESA studies to determine the
presence, nature, extent, and impact of potential contamination. The Owner is committed to voluntarily
cleaning up areas that exceed regulatory thresholds under programs like the USEPA's Voluntary Site
Cleanup Program
FOOTNOTE
1. ERIIS stands for Environmental Risk Information & Imaging Services, Hemdon, VA 22070.
Phone: (800)834-0600.
REFERENCES
1. "Phase I Assessments," ASTM Standard E 1527-94, American Society of Testing and
Materials, 1994.
2. "Transfer Act Site Assessment Guidance Document," Connecticut Department of
Environmental Protection, 1991 revision.
3. Code of Federal Regulations (CFR) Title 40, USEPA, 1996.
195
-------
42
A Report on Sample Precisions in Hanford Tank Waste Samples
H.K. Meznarich, Ph.D. and D.A. Dodd. Analytical Services,
Westinghouse Hanford Company, P. 0. Box 1970, T6-16, Richland, WA. 99352.
High-level radioactive mixed wastes generated at the Hanford site during
nuclear defense production were stored in underground tanks. Tank wastes are
characterized for safety, health, and tank stability concerns. Data quality
requirements for the tank characterization program are established by the Data
Quality Objective process for confirmation of historical information, waste
tank safety, expected future processing for immobilization, and regulatory
requirements. Data quality control parameters and acceptance criteria are
specified in Tank Sampling and Analysis Plans for each tank. However, there
is no appropriate basis to establish data quality requirements on sample
precision for the tank waste samples. A reasonable amount of precision
information from different tank samples have been accumulated in the 222-S
Laboratory data management system, but have not been evaluated. The
objectives of the present report are to compare sample precision between
liquid and solid matrix samples obtained from various tank wastes, to provide
performance based precision information, and to compare against the current
regulatory basis.
Core tank wastes were sampled by the Tank Waste Characterization System
(TWRS) and submitted to the 222-S Laboratory for analyses. Tank samples were
extruded from the core sampling apparatus in the Hot Cell and were
subsequently subsampled. Subsamples were either subjected to water leaching
(for anion analysis), fusion digestion (for total alpha and total beta
analysis, plutonium 239/240), or subject to direct analyses e.g., cyanide
and/or total organic carbon analyses. The data analyzed for the present study
were generated by the analysts at the 222-S Laboratory in 1995. These data
were entered into the data management system (MULTI-LIMS) and were verified by
the responsible chemists. Precision in the tank waste samples is calculated
as the relative percent difference (RPD) between laboratory sample duplicates.
Laboratory sample duplicates were generated for each sample after sample
extrusion in the Hot Cell, but prior to sample digestion. Sample precision
data were retrieved from the MULTI-LIMS system and subjected to Histogram
analysis. Sample precisions were evaluated for chloride, fluoride, nitrite,
nitrate, phosphate, sulfate, bromide, total organic carbon, total alpha, and
plutonium 239/240.
For all the analytes that were evaluated in this study, the following
trends have been observed. Majority of RPD for the liquid samples falls into
the category of less than 20% RPD, except for plutonium. Approximately 50% to
75% of the RPD of solid samples falls into the category of less than 20%.
Approximately 70 to 90% of the RPD of solid samples falls into the category of
less than 35%. The larger RPD in the solid samples than in the liquid samples
indicates that issues e.g., inhomogeneity, subsampling, dissolution are
present in the solid samples. For Hanford tank waste samples, quality control
acceptance criteria for evaluating data quality should be established based on
performance in the samples. A larger precision acceptance criteria is
recommended for solid tank waste samples as long as the analytical system
performance on standards is maintained at the tighter limits. (Sponsored by
the Department of Energy/Richland operations office, Analytical Services
Program, WPD99L10).
196
-------
43
Jerry L. Parr. Quanterra Environmental Services; John J. Austin Jr., U.S. Environmental
Protection Agency's Office of Solid Waste; Charles W. Carter, Lockheed Analytical Services;
Roger Claff, American Petroleum Institute
COMPARISON OF EPA AND API LABORATORY RESULTS
AS PART OF THE 1992-1996 PETROLEUM REFINERY LISTING STUDY
In 1992, as part of a consent agreement, EPA's Office of Solid Waste (OSW) initiated a program
to characterize 29 residuals from petroleum refineries. As part of this effort, EPA performed
sampling and analyses of a variety of refining residuals at 25 petroleum refineries across the
United States. Because EPA's activities could result in a regulation that could significantly
impact the petroleum industry, the American Petroleum Institute (API) decided to provide
independent analyses of split samples.
EPA had prepared a draft Quality Assurance Project Plan (QAPP) and selected a contractor for
the laboratory services (Lockheed Analytical Services). API provided constructive input into
EPA's QAPP, especially relative to the sampling and analysis protocols, and selected a
laboratory to provide the split sample analyses (Quanterra Environmental Services).
With the EPA QAPP as the framework, each laboratory independently performed analyses on a
variety of complex petroleum refining residuals, including tank bottoms, spent catalysts, off spec
solutions, and process sludges. These samples are probably some of the more challenging and
complex samples analyzed in environmental laboratories.
The EPA and API laboratories were both expected to utilize considerable judgment in how to
optimize the laboratory procedures to achieve the best possible data suitable for regulatory
decisions.
This presentation will illustrate the differences in analytical procedures used by the two
laboratories to address these complex matrices. The presentation will also provide a comparison
of the EPA and API data sets, focusing on differences in qualitative and quantitative accuracy
between the two data sets.
197
-------
44
COMPARABILITY OF MEASUREMENTS OF SELECTED PCB CONGENERS,
PAHS, AND CHLORINATED PESTICIDES IN THE MARINE ENVIRONMENT —
RESULTS OF A PERFORMANCE-BASED QUALITY ASSURANCE PROGRAM
Reenie M. Parris. Michele M. Schantz, and Stephen A. Wise
Analytical Chemistry Division, National Institute of Standards and Technology, B208 Chemistry
Bldg., Gaithersburg, MD 20899 USA
ABSTRACT; NIST efforts in providing mechanisms for assessing data quality and comparability
for laboratories monitoring selected organic contaminants in the marine environment include
method development, production of certified reference and other control materials, conduct of
intercomparison exercises, and the coordination of quality assurance workshops for laboratories.
Results from the intercomparison exercises can be used to assess current levels of PAH, PCB
congener, and chlorinated pesticide measurement comparability among participating laboratories.
INTRODUCTION
Tools and mechanisms for the assessment of data produced by laboratories providing
environmental analyses are critical because decision-making based on inaccurate results or data of
unknown quality can have significant economic and health consequences. NIST provides a variety
of activities in support of environmental monitoring programs for organic contaminants. The
largest of these programs has been in cooperation for the past ten years with the U. S. National
Oceanic and Atmospheric Administration (NOAA) National Status and Trends (NS&T) Marine
Monitoring Program and the U. S. Environmental Protection Agency (EPA) Environmental
Monitoring and Assessment Program1"1. For this program, NIST efforts focus on providing
mechanisms for assessing the interlaboratory and temporal comparability of data, and on
improving measurements for the monitoring of over 60 organic contaminants such as polycyclic
aromatic hydrocarbons (PAHs), polychlorinated biphenyl congeners (PCBs), and chlorinated
pesticides in marine bivalve, sediment and fish samples. Table 1 lists the analytes targeted in these
exercises. This program includes the development of improved analytical methods, production of
needed NIST Standard Reference Materials (SRMs) and other control materials, conduct of yearly
interlaboratory comparison exercises, and the coordination of workshops to discuss the results of
these exercises. Since 1993, private sector and other laboratories that are not associated with these
NOAA or EPA programs and that reimburse NIST for the participation costs of these exercises
and workshops have subscribed to the "NIST Intercomparison Exercise Program for Organic
Contaminants in the Marine Environment." Current participants represent a number of multi-
laboratory monitoring programs and include over forty federal, state/municipal, university/college,
private sector and international laboratories. In this performance-based program, participants use
methods currently in use in their laboratories for these analyses.
Natural matrix SRMs, NIST SRM 1941 and the renewal SRM 194la, Organics in Marine
Sediment, and SRM 1974 and the renewal SRM 1974a, Organics in Mussel Tissue; PAH, PCB,
and chlorinated pesticide calibration solution SRMs at two concentration levels; and sediment and
tissue control materials for routine use were developed in response to the needs of this program.
198
-------
jfTTERCOMPARISON EXERCISES; MATERIALS AND CONDIJgT
To facilitate the isolation of sources of analytical variability and bias, the materials chosen for the
intercomparison exercises have progressed from simple hexane solutions of target analytes to
enriched extracts of natural materials to unenriched extracts of natural materials to natural-matrix-
based materials. Since 1991, samples of two natural-matrix-based homogeneous materials
derived from the marine environment that have not been fortified with any of the target analytes
have been used for the annual intercomparison exercises. Exercise materials are chosen and
prepared to be as similar as possible to the samples. Table 2 shows typical concentration ranges in
exercise materials. The homogeneity of each material is assessed by NIST. One of the most
difficult tasks of the exercise coordinators is the choice of the "assigned value," the concentration
of each analyte in a particular material to which comparisons will be made. Candidate SRMs, true
unknowns during the time frame of the exercise, are used as exercise materials whenever possible.
This enables eventual comparisons of the exercise assigned values and the results of the individual
laboratories with certified concentrations for these reference materials.
Z-scores and p-scores4 are used to assess and track performance (for accuracy and precision,
respectively) of federal, state/municipal, university/college, private sector and international
laboratories participating in these exercises.
RESULTS AND DISCUSSION
Results from 1994 Mussel Tissue VI, 1993 Mussel Tissue V, and 1992 Marine Sediment III
intercomparison exercises in which candidate reference materials were analyzed can be compared
with concentrations of target analytes certified in these materials. These results and those from the
1994 and 1995 exercises can be used to illustrate current levels of measurement comparability
among the participating laboratories for these selected analytes present at trace level concentrations
in natural-matrix materials.
As discussed previously, when candidate SRMs are used as exercise materials, the assigned values
for the exercise can be compared with the certified values when issued. Figure 1 shows this
comparison for PAHs in the Mussel Tissue V exercise material which were relabeled bottles of
SRM 1974a. In general, there is good agreement between the exercise assigned values and the
certified concentrations.
Typically, the performance of laboratories that have participated in this program for a number of
years was better than that of the laboratories in their first or second year of participation. As
expected, laboratories reporting concurrent reference material analyses results (perhaps an
indicator of regular reference material use) typically showed better performance than those who
did not analyze the reference materials. The percentage of PAH results from laboratories
participating more than two years in the program that were within 50% of the accepted
concentrations (i.e., z-scores (25%) in the -2 to +2 range) was 90% for the Sediment IV and 75%
for the Mussel Tissue VI. The percentage of acceptable results for those laboratories participating
in the intercomparison exercise program for one to two years was 58% and 53% for the Sediment
IV and Mussel Tissue VI, respectively.
The use of the z-scores for assessing performance in a number of exercise materials is illustrated in
Figure 2 which shows phenanthrene z-scores (25%) for four laboratories for the NOAA NS&T
Program on two sediment and three mussel exercise materials analyzed over a three-year period.
One can see by the results of Lab 4 for the Mussel VI in comparison with results for the previous
four materials which were well within the Program's acceptability criteria that the demonstration
199
-------
of acceptable performance should be an ongoing process of which participation in intercomparison
exercises is but one part. Charts of z-scores as illustrated in Figure 3 are useful for assessing
intra- and interlaboratory bias trends and relative performance.
For the 1994 Marine Sediment IV exercise, 97.4% of the p-scores (precision indicators) of
reported PAH results were acceptable. In these exercises, interlaboratory variability is a greater
contributor to PAH measurement incompatibility than is the intralaboratory variability.
CONCLUSIONS
Intercomparison exercises provide an important mechanism for assessing the comparability of data
being produced by the participating laboratories. Exercise materials similar in matrix, form and
analyte concentration to typical samples being analyzed by the laboratories are most useful for
demonstrating the level of comparability and for revealing potential problem areas. The use of
candidate CRMs as exercise materials is very useful in evaluating the procedures used to
determine the assigned values for an exercise material.
For the determination of the target analytes in these complex marine matrices with relatively low
levels of these analytes, the levels of bias and reproducibility of many of the participating
laboratories meet their current acceptability requirements; however, there is certainly room for
improvement. Minimizing the between-laboratory bias such that the analytical variability is
significantly less than the sampling variability should be an achievable goal.
At the beginning of this activity, the NOAA NS&T and EPA EMAP program managers had no
tools to assess the comparability and quality of the measurements made by the various laboratories
providing data for their monitoring programs nor could they effectively document changes in the
quality of the data produced. With continuous participation in these annual intercomparison
exercises, laboratories can now demonstrate the quality and comparability of their results. Two
natural matrix SRMs (i.e., sediment and mussel tissue) and six calibration solution SRMs have
been developed at NIST in response to the needs of this program. The availability of these SRMs
provides the opportunity for laboratories to evaluate their performance on a continuous basis.
REFERENCES
1. A. Y. Cantillo and R. M. Parris, "Evaluation of Trace Organic NOAA Status and Trends
Quality Assurance Project Performance," in Quality Assurance for Analytical Laboratories,
M. Parkany (ed.). Royal Society of Chemistry, Spec. Publ. No. 130 (1993).
2. A. Y. Cantillo and R. M. Parris, National Status and Trends Program Quality Assurance
Project: Trace Organic Intercomparison Exercise Results 1986-1990, NOAA Tech. Memo.
NOS/ORCA69(1994).
3. A. Y. Cantilio, NS&T Quality Assurance Project Intercomparison Exercise Results 1991-
1993, NOAA Tech. Memo. 79 NOS/ORCA (1995).
4. IUPAC "The International Harmonized Protocol for the Proficiency Testing of (Chemical)
Analytical Laboratories," Pure&Appl. Chem., 65, 2123-2144 (1993).
200
-------
Table 1. Analytes" of Interest in NIST Intel-comparison Exercise Program for
Organic Contaminants in the Marine Environment
Chlorinated Pesticides
hexachlorobenzene
alpha-HCH (alpha-BHC)
gamma-HCH (gamma-BHC, Lindane)
heptachlor
heptachlor epoxide
cw-chlordane (alpha-chlordane)
fra/«-chlordane (gamma-chlordane)
oxychlordane
cis-nonachlor
/rans-nonachlor
mirex
Polvchlorinated Biohenvl Congeners
2,4'-DDE
4,4'-DDE
2,4'-DDD
4,4'-DDD
2,4'-DDT
4,4-DDT
aldrin
dieldrin
endrin
endosulfan I
endosulfan n
PCB No. Compound Name
8 2,4'-dichlorobipheny!
18 2,2',5-trichlorobiphenyI
28 2,4,4-trichlorobiphenyl
44 2,2',3,5'-tetrachlorobiphenyl
52 2,2',5,5'-tetrachlorobiphenyl
66 2,3',4,4'-tetrachlorobiphenyl
101 2,2',4,5,5'-pentachlorobiphenyl
105 2,3,3',4,4'-pentachlorobiphenyl
118 2,3',4,4',5-pentachlorobiphenyl
Polycyclic aromatic hydrocarbons (PAH)
naphthalene
2-methyInaphthalene
1 -methylnaphthalene
biphenyl
2,6-dimethylnaphthalene
acenaphthylene
acenaphthene
1,6,7-trimethylnaphthalene
fluorene
phenanthrene
anthracene
1 -methylphenanthrene
153
170
180
187
195
206
209
PCB No. Compound Name
128 2,2',3,3',4,41-hexachlorobiphenyl
138 2,2',3,4,4',5'-hexachlorobiphenyl
2,2',4,4t,5,5'-hexachlorobiphenyl
2,2',3,3',4,4',5-heptachlorobiphenyl
2,2',3,4,4',5,5'-heptachlorobiphenyl
2,2',3,4',5,5',6-heptachlorobiphenyl
2,21,3,3t,4,41,5,6-octachlorobiphenyl
2,2',3t3',4,4',5,5',6-nonachlorobtphenyl
decachlorobiphenyl
fluoranthene
pyrene
benz[a] anthracene
chrysene
benzo[fluoranthenes [b+j+k]
benzo[^]pyrene
benzo[a]pyrene
perylene
mdeno[ 1,2,3-o/lpyrene
dibenz[a,/z]anthracene
benzo[g/?/]perylene
1 The following are typically reported by participants and evaluated as sums of the indicated
components:
PAH
PCB congeners
chrysene + triphenylene
benzo[b]- + benzojj]- + benzo[k]fluoranthene
dibenz[a,h]anthracene + dibenz[a,c]anthracene
PCB 66 + PCB 95
PCB 101 + PCB 90
PCB 138 + PCB 163 + PCB 164
PCB 187 +PCB 182 +PCB 159
PCB 170 +PCB 190
201
-------
200
150
100 -
50
0
fluoranthene
pyrene
chrysene/triphcnylenc
bcnzo[b+j+k jfluoranthene
benzo[e Jpyrenc
o
M
TD
tn
•H)
30
20
10
0
20
15
10
5 H
naphthalene
phenanthrcnc
I-methyl-
phenanthrene
benz[a]-
anthracene
indeno(l,2,3-c
-------
Table 2. Typical Concentration Ranges of Determined Compounds
Great Lakes Fish Homogenate Marine Sediment
ng/g wet wt. ng/g drv wt.
PAHs
PCB Congeners
Chlorinated Pesticides
not determined
1 to 150
1 to 150
40 to 10000
1 to 90
1 to 150
Mussels
ng/g drv wt.
1 to 300
1 to 150
1 to 100
Lab
Lab 2
5 -
4 -
3 -
^ 2
(N
^•s
2
O
« 0 -i
N
-1 -
' *
*
^
+
'
*
:«
.
. . .
*
4
A
;*
Lau t
+ •
• »
. *
*
*
S3 M4 M5 S4 M6 S3 M4 M5 S4 M6 S3 M4 M5 S4 M6 S3 M4 M5 S4 M6
FIGURE 2 Phenanthrene z-scores of four NS&T laboratories for three mussel tissue (M4, M5,
M6) and two sediment (S3, S4) intercomparison exercises conducted during a three year period.
7 -
6 -
4 -
•^^*> O
^ J "
W)
»N T -
V_> Z
§ 1-
u
"^ n -j
N U
-1 -
X
• A
X
- i 'ft
i11 f ' 'Z'i'fi1
• Ml
• A B A +
A si
i • i
•
9
t
A
"7
: o
' A '
A A
a: _
f'Ali'
A
!* v i
1 V ' *P
•I 1
A
•
1 1 A 1
1 i 1 O 1
ii
«•
•
A
la 2 3 4 5 6 7 8 9 10 11 12 13 14 15 17 18 19 20 21 22
Laboratory Number
FIGURE 3 Z-scores by laboratory for 23 PAHs in 1994 Sediment IV intercomparison
203
-------
45
ABSTRACT
The Effectiveness of the Procedure for Estimating Instrument Detection Limits for
Inorganic Analyses in the Contract Laboratory Program
G. A. Laing and F. C. Gamer
Lockheed Environmental Systems and Technologies Company
and
G. L. Robertson
U.S. Environmental Protection Agency
National Exposure Research Laboratory
Characterization Research Division
Las Vegas, NV 89193
The U.S. EPA, National Exposure Research Laboratory, Characterization Research Division (Las
Vegas, NV) conducts studies of the quality assurance and quality control (QA/QC) measures
required in EPA methods. These studies determine whether specific QA/QC measures add value
to the data and whether these QA/QC measures are the most effective way of controlling a given
parameter. The QA/QC of the Contract Laboratory Program (CLP) inorganic analyses are
currently being tested. This presentation covers the evaluation of the requirement for
determining the instrument detection limit (IDL). The CLP inorganic contracts require that
laboratories determine and report IDLs quarterly. These determinations are made by analyzing a
standard at three to five times the estimated IDL seven times on each of three consecutive days
and averaging the standard deviation of the results for each day. This investigation evaluated
whether the procedure measures a realistic IDL, whether the frequency of measuring the IDL is
appropriate, and whether alternative procedures could provide the same or better information at
lower cost Data from the CLP analytical results database (CARD) were used to perform the
evaluation. Study results showed that the current procedure provides an optimistic IDL, that the
IDL varies significantly from day to day, and that similar IDL information can be obtained from
QA/QC analyses that are already performed on a daily basis. Alternatives to the current IDL
procedure are suggested.
204
-------
46
AUTOMATED DATA VALIDATION:
WHAT ARE THE LIMITATIONS?
Karen A. Storne. Project Scientist
O'Brien & Gere Engineers
5000 Brittonfield Parkway
Syracuse, New York
13221
INTRODUCTION
In the past, the validation of environmental samples has been performed manually. The
reports may have been generated using a computer, but the data evaluations were done
using a trusty pen and a hand-held calculator. However, now that computers are so much
apart of analytical processes, why not computerize data validation? Since most of the
validation tasks involve comparison of numbers, it makes sense to automate these
sometimes mundane reviews. But what about the tasks that involve more than simply
comparing control limit windows to sample results? Those validation processes that may
involve making judgment calls are beyond the capabilities of current computer programs.
The USEPA has successfully used one computer program to automate certain parts of
validation but other aspects of data validation may not be easily performed by a
computer.
VALIDATION PROCESS
Data validation has typically been performed by an experienced chemist who diligently
inspects page after page of a data package which is generated by a laboratory during the
analysis of environmental samples. The process of validating laboratory data involves
evaluating the sample data by recreating the analysis process. Each chain of a custody
form, sample preparation log, instrument injection log, sample quantitation form,
chromatogram, mass spectrum, quality control form, and sample result form is examined
to determine whether the requirements of the analysis have been met. The validator then
determines whether the accuracy, precision, comparison, reproducibility, and
completeness of the analysis meets requirements set by the project Sample and Analysis
Plan (SAP), or the Quality Assurance Project Plan (QAPP), and finally a report listing the
excursions and conclusions is generated.
205
-------
COMPLICATIONS
The validation process can be straightforward if the sample matrix has little affect on the
instrument's ability to generate data, and if analyses follow the required methods without
deviations which affect the sample results. However, with complicated sample matrices,
and deviations from the methods, data validation can become more complicated. For
example, if a sludge sample has been analyzed for polychlorinated biphenols (PCBs), it
may contain any of several mixtures of Aroclors, one of the most common of which were
trade named Aroclors; it may be difficult to identify the Aroclor peak pattern in the
cnromatograms produced by the instrument because the Aroclor may be masked by
interfering peaks. Many Aroclor patterns that samples exhibit will not match the standard
pattern exactly since the Aroclors in the samples have been weathered. If a weathered
Aroclor pattern is present, the identification of the Aroclor will require experience and
judgment since the Aroclor pattern will appear altered.
If that same sample is analyzed for semivolatile organics by gas chromatograph/mass
spectroscopy (GC/MS), the mass spectra may contain fragments from other interfering
compounds that disguise the spectra of the target compound. Even after using the best
techniques to clean the samples up before running them on the instruments, matrix
interferences may be present in the samples and can cause interpretation to be difficult.
In the case of samples that are analyzed for volatile organics, if a sample is purged in the
same instrument vessel that was previously used to analyze a sample that contained a
high concentration of volatile organic, such as trichloroethene, the trichloroethene result
for that second sample may be high because carryover has occurred. A review of the
injection log (which is typically hand written) can determine whether the trichloroethene
concentration is erroneously high; if applicable, a qualifier can be applied to warn of this
analysis error.
Due to the pressure of schedules associated with getting sample results to the data user
on time, analysts occasionally make decisions during the analysis process that affect the
final sample results. Samples are analyzed outside of holding times (since the client
requested the analysis after the holding times have been exceeded), samples are analyzed
on instruments with poor calibration response factors (because the method listed in the
QAPP is not current), and sample results associated with low laboratory control sample
recoveries are reported (because the recovery criteria listed in the QAPP and the
laboratory control limits differ). While these decisions may seem reasonable at the time,
the overall affect of these decisions are evaluated during validation.
206
-------
Though laboratories continue to improve data processing and reporting processes, many
still use manual entry of data results into computer systems to generate final data
packages. While quality control measures are in place to check for data entry errors in
the laboratory, occasionally data which contain transcription errors are reported. These
errors can involve using the wrong initial calibration factor when several initial
calibrations are involved, adjusting the compound areas if manual integrations are used,
forgetting to include dilutions used, and the lack of adjustment to account for percent
solids in reporting detection limits. These mistakes are easily corrected by the laboratory,
but the data user may not realize that a mistake has been made.
COMPUTER VALIDATION
The ability of computer systems to validate environmental sample data depends on the
extent of the tasks that the computer would be expected to perform. If number
comparisons are involved such as evaluation of quality control sample recoveries, blank
contamination, calibration relative standard deviations, response factors, and percent
differences, holding times, and retention time windows, the computer program can be
used to perform the validation. As long as data is available for the computer to review,
and hand written logs are provided in electronic form, the program can evaluate the
analysis process.
However, if complications are detected in the sample analysis process, it would be
difficult for a computer system to thoroughly and properly evaluate the resultant sample
data. While the criteria that the computer are capable of reviewing are important to the
evaluation process, other factors are as important to reporting accurate data to the data
user. These may include providing a second opinion of identified Aroclors with patterns
that appear to be weathered or mass spectra with erroneous fragment patterns, judging
whether carryover affects a particular compound result, or determining whether detection
limits seem reasonable considering the type of sample matrix involved in the analysis. It
would be difficult for current computer programs to make these important judgment calls.
THE USEPA CADRE SYSTEM
The USEPA Computer Aided Data Review and Evaluation (CADRE) system is an
automated system that compares analytical results to quality control criteria established
for the Quick Turnaround Methods (QTMs) [1]. The QTMs are methods that involve
single column gas chromatography (GC) analysis, therefore situations that involve
evaluating second column confirmations or the interpretation of GC/MS spectra are not
involved in the review of this sample data. After the CADRE program evaluates the data,
it applies data qualifiers to indicate usefulness of the data.
207
-------
Table 4 - Qualifiers added to QTMdata by the CADRE program
Qualifier Quality control problem detected by the program
U Sample is affected by method blank contamination
B Sample is affected by instrument blank contamination
L Compound is outside LCS recovery criterion
P Compound is outside PVS recovery criterion but inside expanded criterion
T Absolute retention time is used to the identify compound
S SMC in the sample is outside recovery criterion
E Compound concentration is above initial calibration range
M Retention time shift is detected during sample analysis
The CADRE performs the evaluations of quality control sample results, calibrations,
laboratory control samples, performance verification standards, method and instrument
blanks, and system monitoring compounds. This system compares holding times to
analysis and extraction dates, determines whether retention time window criteria were
met for compound identification, and evaluates peak resolution. These evaluations all
involve comparing numbers and performing calculations, which a computer is fully
capable of doing.
SUMMARY
While computers can automate the process of determining whether sample results are
within control limits and meet method criteria, situations may arise during the validation
process that would prove to be challenging to current computer programs. In many cases
judgments must be made during the validation process that involve much more than
number comparisons. These judgments often involve investigations that utilize a
validator's knowledge of anlaytical chemistry, past experience and the application of
laboratory common sense. Whether the judgment involves deciding not to reject sample
results because the data user acknowledges that the concentration reported may be
inaccurate due to exceeded holding times, or determining if a weathered Aroclor pattern
can be reported as Aroclor 1248 or Aroclor 1254, limitations in the validation process
exist. While data users may trust the computers to perform part of the validation, as long
as these limitations exist, data validators will need to manually review at least some
analytical data.
REFERENCES
[1] USEPA Quick Turnaround Methods, Statement of Work, Exhibit D, Washington,
D.C.
208
-------
47
Automation of ICP Data Validation for Environmental Analysis.
MJatro
209
-------
48
12th Waste Testing and Quality Assurance Symposium
New Quality Systems at the U.S. Environmental Protection Agency
Nancy W. Wentworth, P.E.
Director
Quality Assurance Division (8724)
National Center for Environmental Research and Quality Assurance
U.S. Environmental Protection Agency
Washington, DC 20460
Abstract
The U.S. Environmental Protection Agency (USEPA) established
the first mandatory quality assurance (QA) program in 1979.
Guidance on the preparation of QA program plans and QA project
plans was issued in 1980. While these documents laid the
foundation for today's quality systems, they focused on
analytical quality control with only limited recognition of the
contribution of network design, field operations, and data
assessment to the quality of environmental data used in decision
making.
In the late 1980s, there was wide recognition that the 1980
QA program description was not serving the environmental
community well. Environmental restoration programs were
expanding in both the public and private sectors. More holistic
approaches to environmental testing and monitoring programs were
needed to meet the competing interests of limited resources and
growing expectations for performance of monitoring systems.
The USEPA Quality Assurance Division (QAD) was instrumental
in the development of the American National Standard
Specifications and Guidelines for Quality Systems for
Environmental Data Collection and Environmental Technology
Programs (ANSI/ASQC E4-1994). This standard provides the basis
for revisions to the USEPA quality system.
The USEPA quality system has a number of components that
will be outlined in this presentation. Included will be
discussion on the content and status of proposed revisions to the
EPA Order that provides the basis for the internal USEPA quality
program, the Quality Manual that contains the requirements for
the internal Agency programs, and the regulatory changes that
will affect contractor/cooperator/grantee operations. Training
in the details of these documents will also be discussed.
210
-------
49
12th Waste Testing and Quality Assurance Symposium
Achieving Flexibility Through Effective QA Practices
John Warren, Ph.D.
Senior Statistician
Quality Assurance Division (8724)
National Center for Environmental Research and Quality Assurance
U.S. Environmental Protection Agency
Washington, DC 20460
It is an axiom of good science that effective alternatives,
or flexibility, can only be achieved if it is anticipated or
planned for at the very early stages of development; the same
axiom applies equally well to environmental monitoring
activities. The U.S. Environmental Protection Agency (USEPA)
recognizes this need for flexibility in its data collection
activities by insisting on a planned approach to data
acquisition, for example, use of the data quality objectives
(DQO) process.
This paper outlines the DQO process as it applies to the
overall goal of flexibility and integrates this planning aspect
with the latest quality assurance tool from the Agency, "Data
Quality Assessment" (DQA) . Attention will be given to the
analytical features of DQA, and also to the way in which DQA can
be used to efficiently reformulate DQOs to achieve the desired
flexibility. Copies of Guidance for the Data Quality Objectives
Process (EPA QA/G-4) and Guidance for Data Quality Assessment
(EPA QA/G-9) will be available for participants.
211
-------
50
DATA QUALITY MANAGEMENT
FOR EMERGENCY RESPONSE CLEANUP SERVICES
Julia A. Fields. Corporate Quality Assurance Director, Environmental Quality
Management, Inc., 1310 Kemper Meadow Drive, Cincinnati, OH 45240
George C. Schupp. Regional Quality Assurance Manager, U.S. EPA Region V,
111 W. Jackson Blvd., Chicago, IL 60604
ABSTRACT
Data Quality Management (DQM) is a particular challenge for U.S. EPA
Superfund Emergency Response Cleanup Services (ERCS). The ERCS
mechanism of the Office of Superfund provides cleanup services for active and
abandoned industrial sites classified as an imminent threat to human health and
the environment. Upon initial notification, rapid onsite mobilization (typical
response within 3 to 6 hours) can be required at a site over 500 miles away. The
emergency nature of these projects preclude opportunities for traditional data
quality management such as preparation, review, and implementation of a site-
specific sampling and analysis plan. Analytical laboratory support for these
efforts must typically be chosen based on the location for sample delivery and
availability to meet rush turnaround times. The laboratories providing analytical
support for this service are often required to provide appropriate quality,
defensible data within the minimum possible time. All analytical results must be
generated and reviewed and even validated without delay to the project, so that
work at the site may continue without interruption.
In the face of these challenging circumstances, typical laboratory performance
and usability of data have yielded marginal results. As a corrective action, DQM
program improvements were implemented for the U.S. EPA Region V ERCS
Contract. These program improvements help ensure that data collected in
support of cleanup efforts are defensible, of appropriate quality, and meet the
rapid turnaround time requirements.
The elements of this improved emergency response DQM program include:
Enhanced Comprehensive Program-wide Quality Assurance Plan (QAPP)
Emergency Response Laboratory Approval Program (ERLAP)
Site Control Samples
Program-Specific Analytical Method Standard Operating Procedures (SOPs)
Quality Assurance oversight performed by the Analytical Services Coordinator
Standard Reporting Procedures
212
-------
During the development and implementation of the DQM program improvements
over a 6-quarter period, performance was measured in terms of the percentage of
samples rejected or qualified as estimated due to quality control criteria not being
met. Sample data were simplified to conduct a trend analysis. One sample may
have been analyzed for several different parameters, however, if any one of the
parameters was rejected then the entire sample was counted as rejected.
Additionally, data qualified as estimated due to sample matrix effects were not
included in this measurement. The percentage of samples with rejected data
decreased from 25% to 0% over the period, and percentage of samples with data
qualified as estimated decreased from 75% to 2%.
INTRODUCTION
The ERCS Contract mechanism provides for rapid response service for removal
and remediation of oil, petroleum, and hazardous substances releases that pose
an imminent threat to human health or the environment. Projects are conducted
under Section 311 of the Clean Water Act and Section 104 of the Comprehensive
Environmental Response, Compensation and Liability Act of 1980 as amended by
the Superfund Amendments and Reauthorization Act of 1986 (SARA), and
Subtitle I of the Resource Conservation and Recovery Act as amended by SARA
of 1986.
The ERCS mechanism requires rapid onsite mobilization, typically within 3 to 6
hours upon initial notification, at a site that can be over 500 miles away. The
emergency nature of the project precludes opportunities for traditional data quality
assurance planning such as preparation, review, and implementation of a site-
specific sampling plan. Analytical laboratory support is chosen based on
laboratory location and the ability to meet rush turnaround times. The
laboratories providing analytical support are often required to provide appropriate
quality, defensible data within the minimum possible time. All analytical data must
be generated, reviewed, and even validated without delay to the project, so that
work at the site can proceed without interruption.
DQM Program improvements were implemented for the U.S. EPA Region V
ERCS. These improvements ensure that data collected in support of cleanup
efforts are defensible, of appropriate quality, and meet the rapid turnaround time
requirements.
The DQM program is driven by litigation for cleanup cost recovery and is specified
by the Office of Solid Waste, Emergency Response, Quality Assurance/Quality
Control Guidance for Removal Activities, Sampling Plan and Data Validation
Procedures, EPA/540/G-90/004 (OSWER Directive).
213
-------
IMPLEMENTATION OF DQM IMPROVEMENTS
Working with a prepared, standard QAPP in support of the ERGS program
resulted in marginal laboratory performance and data usability concerns. As a
result, the ERGS QA staff determined that a more proactive approach was needed
in order to bridge all site-related activities and achieve QA objectives. To that
end, program elements were added to improve quality performance. Through
continuous improvement, the program quality performance measurement
improved dramatically over a 6-quarter period. Prior to implementation of
improved program elements, virtually all samples had at least one analytical result
that was qualified as estimated or rejected because quality control criteria
established by the OSWER Directive were not being met.
Quality Assurance Performance
re
a
1
1995
2
1995
3
1995
1995
5
1996
6
1996
257
Samples
352
Samples
615
Samples
400
Samples
139
Samples
399
Samples
1^
0 10 20 30 40 50 60 70 80 90 100
- % Rejected
- % Estimated
• % Accepted
Percent of Samples
214
-------
The implementation of the Emergency Response Laboratory Approval Program
(ERLAP) provided a dramatic improvement in data quality for results provided to
ERGS. Over a 2-quarter period, rejected data decreased 15% and data qualified
as estimated dropped by 70%. With the implementation of an upgraded
program-wide QAPP, standard reporting procedures, and increased QA
responsibility by the analytical coordinator, QA performance again improved as
rejected data decreased an additional 5% and data qualified as estimated
decreased by an additional 10% over the next 2-quarter period. With the
implementation of program-specific analytical method SOPs, QA performance
again improved as rejected data decreased an additional 10% and data qualified
as estimated also decreased by an additional 10%.
ELEMENTS OF DQM IMPROVEMENTS
COMPREHENSIVE PROGRAM-WIDE QUALITY ASSURANCE PLAN (QAPP)
A generic program-wide QAPP is applied to all sites where environmental
measurements are conducted. This QAPP must consider all potential program
activities including site characterization and extent of contamination assessment,
waste compatibility testing, waste characterization, efficiency of applicable
treatment procedures and systems, and air monitoring for health and safety.
This program-wide QAPP is a tool for ensuring that the data quality objectives
(DQO) of all data collected under the ERCS program are known and documented.
To that end, the QAPP presents the organization, objectives, functional activities,
and specific QA/QC procedures associated with the program. The QAPP also
describes specific protocols to be followed for sampling, sample handling and
storage, chain of custody, and laboratory and field analysis. Additional
information in the QAPP includes laboratory evaluation and selection criteria and
QA program performance measurements. All sampling and analytical method
SOPs and data review and validation checklists are provided in the appendices of
the QAPP.
A copy of the program-wide QAPP is provided to each ERCS program site,
approved laboratories, and analytical coordinators. The QAPP is updated
annually to include new requirements and upgraded procedures developed
through continuous improvement.
EMERGENCY RESPONSE LABORATORY APPROVAL PROGRAM (ERLAP)
ERLAP was developed and implemented as a corrective action for addressing
marginal laboratory performance. The laboratory approval program consists of a
preliminary capability review to determine qualifications and a formal technical
215
-------
systems audit to review procedures and documentation. Those laboratories
successfully completing the audit, which includes resolution of all audit exceptions
and re-audit, as appropriate, are approved in ERLAP for the ERGS program.
Prior to the selection of a laboratory for providing service to ERGS, a preliminary
review is conducted that consists of an evaluation of the following: state and
federal certifications including audit sample results; run logs or bench sheets for
analytes of interest; a laboratory prepared analyst experience table; and the
laboratory's Quality Assurance Management Plan (LQMP).
The LQMP can provide evidence of the existence of an acceptable quality
management program. The LQMP should, at minimum, contain descriptions of
the following: organization and policy, facilities and equipment, document control,
analytical methodology and acceptance criteria, data and report generation, QA
and QC procedures. The following factors are evaluated in the LQMP: visibility
of the QA Organization, QA policy, facilities and equipment capabilities,
documentation control, definition of QA objectives for every analyte and method,
data and report preparation procedures, QG sample procedures, internal systems
audits, and performance evaluation samples.
A formal systems audit is performed of those laboratories successfully meeting
the requirements of the preliminary review. This audit includes both form and
facility evaluations, review of laboratory procedures and documentation against
the requirements of SW-846, and the ability and willingness to meet OSWER
Directive requirements, regional requirements, the program-wide QAPP and the
laboratory's own internal SOPs. An audit checklist, standardized to address all
specific requirements, is utilized to maintain consistency and objectivity across all
laboratories that are evaluated.
Laboratories are provided a 1-week advance notification of the audit, when
possible. An inbriefing meeting is held to introduce audit participants, discuss the
goal of the audit, identify specific areas for review, and to introduce laboratory
personnel to the ERGS program. Facility evaluations and form auditing are
conducted, along with laboratory staff interviews. At the conclusion of the audit,
an outbriefing is held to discuss findings and observations.
In general, the areas for review include sample receipt and storage, chain of
custody, sample container and glassware cleaning requirements, sample
preparation and analysis, preparation and analysis documentation, detection limit
studies, instrument calibration, instrument maintenance, data management,
document control, Laboratory QA and QC procedures, organizational structure,
and personnel experience.
216
-------
Following the audit visit, a written report is provided to the laboratory detailing
audit exceptions. A forum is established with the laboratory and the ERCS
program QA staff to address and resolve all exceptions. Upon successful
resolution of audit exceptions, the laboratory is revisited to review all corrective
action that was taken and to ensure that all compliance issues have been met.
The laboratory is classified as ERLAP approved for the ERCS program only after
successful resolution of all audit exceptions.
SITE CONTROL SAMPLES
Site control samples were implemented to address QA for data validation
purposes on smaller sites when an approved laboratory is not available and
auditing is not cost-effective and for other laboratory performance evaluation
requirements. Three different types of control samples have been designed to
meet ERCS program requirements.
The site media control sample, which is least expensive, is prepared by the ERCS
QA staff by spiking a QC standard into sample media from the site in a controlled
environment. This control is then shipped to the laboratory with other site
samples for analysis. Although statistical data is not available to evaluate
absolute laboratory performance, certain qualitative information can be derived.
In the event of poor recovery or other QC issues concerning this control sample, a
reference laboratory may also be asked to evaluate the sample. This approach is
utilized for a site where samples submitted to the laboratory are comprised of an
extremely complicated matrix. The site media control sample used in conjunction
with matrix spike and duplicate sample data can be used to extrapolate the
laboratory's qualitative performance in recovering the analytes of interest.
Another type of site control sample is the standardized performance evaluation
(PE) sample. This sample is provided by a supplier and is prepared with standard
media and established acceptance ranges. The supplier ships the samples
directly to the laboratory. Purchased PE samples generated by an external
source provide statistical data for comparability studies with multiple laboratories.
This approach is more expensive, however, and the laboratory will know the true
nature of the sample and may treat it differently than the site samples. This
approach is appropriate for a relatively easy sample matrix, such as drinking
water, for performing a preliminary evaluation of a laboratory, or when the
samples are used in an on-going study with multiple participants.
A third option is the "double blind" PE sample in which the site sends a supplier a
bottle identical to those used in the field. The supplier prepares a PE sample and
sends it to the site to be labeled and delivered to the laboratory with other
samples for analysis. The control is designed to appear similar to other site
217
-------
samples and permits a relatively absolute evaluation of laboratory performance.
Although this option is the most expensive, it is appropriate for certain
confirmational analyses.
The decision to use a specific option is site specific based on many factors,
including the data quality objectives for the site. Ultimately, the choice of a
particular type of control sample may result in obtaining qualitative information
concerning data usability, rather than a strict determination of laboratory
performance. The needs of the site and cost-effectiveness may warrant no
additional requirements for laboratory performance data.
PROGRAM-SPECIFIC ANALYTICAL METHOD SOPs
A program interpretation for ERCS sample testing, based on requirements
recommended by the OSWER Directive, regional requirements,.and the guidance
of the SW-846 protocols, is detailed in parameter specific preparation and
analysis method SOPs. These SOPs specify requirements for sample
preservation, preparation and cleanup, initial and continuing calibration, method
performance (check and/or laboratory control samples), precision and accuracy,
method blanks, and quantitation. The SOPs define acceptance criteria for each
parameter as well as required laboratory corrective action when the criteria are
not met.
The SOPs were generated by using the SW-846 protocol as the basis for sample
preparation and analysis and modifying where appropriate with OSWER Directive
specifications, regional requirements, and program-wide standard method
interpretation of the protocol. Additionally, the SOPs specify a notification
requirement for the laboratory to report any modifications prior to performing the
sample analysis.
The SOPs are provided to potential laboratories during the bidding process, along
with the request for pricing and procurement terms and conditions documentation.
This provides each potential laboratory with the knowledge of the exact
preparation, analytical, deliverable, and reporting requirements so that the service
may be appropriately priced. The SOPs are also included in the program-wide
QAPP available to all site personnel, ERLAP Laboratories, and analytical
coordinators.
ANALYTICAL SERVICES AND QA/QC COORDINATION
The analytical coordinator plays a critical role in DQM for ERCS. The coordinator
is the point of contact for both the site and the laboratory, and as such provides
technical support for sampling and analytical activities. The coordinator
218
-------
continuously monitors the laboratory to ensure that all data meets the data quality
objectives. The coordinator also works with site personnel to interpret data
quality objectives and provide guidance during sampling activities to ensure that
sample handling will not compromise data quality. The coordinator also performs
the function of data review and validation, and determines corrective action as
necessary.
STANDARD REPORTING PROCEDURES
Documentation is an essential part of the QMP. As such, standard reporting
forms have been created with a convenient, user-friendly checklist format. These
checklists are used for the initial notification QA/QC requirements, and the daily
laboratory coordination and QA/QC status form, analytical testing requirements,
and data review and data validation. These checklists are a tool by which all
information can be easily recorded, reviewed, and audited as necessary. All
checklists are included in the program-wide QAPP provided to each site, ERLAP
approved laboratory, and the analytical coordinator.
SUMMARY
The implementation of the DQM program corrective action elements for marginal
laboratory performance and usability of data concerns has resulted in significant
performance improvement. The development of ERLAP and improved
laboratory auditing procedures, as well as implementation of analytical method
SOPs are elements that are particularly significant in the program's success.
The modified procedures with the additional form and checklist requirements not
only provide an appropriate level of documentation but facilitate effective
communication between the site, program management, and the laboratory. The
program requirements necessitate a team approach to working the project, and
when accomplished properly, the DQM program performs at an optimum level.
REFERENCES
U.S. EPA Office of Solid Waste and Emergency Response, Quality Assurance/
Quality Control Guidance for Removal Activities; Sampling QA/QC Plan and Data
Validation Procedures, EPA/540/G-90/004, April 1990.
U.S. EPA Office of Solid Waste and Emergency Response, Test Methods for
Evaluating Solid Waste, Physical/Chemical Methods (SW-846), November 1986.
219
-------
51
DOCUMENTATION AND RECORD KEEPING GUIDELINES
Ann Rosecrance. Corporate Quality Assurance Director
LaDonna Kibler, Corporate Quality Assurance Administrator
Core Laboratories, 5295 Hollister Road, Houston, Texas 77040
ABSTRACT
In order for analytical data to be considered legally defensible, the general rule is that the
sampling and analysis activities must be performed according to published analytical
methods and that adequate documentation is available to support the sample results. A
previous publication addressed guidelines for generating defensible data.1 Although
documentation and record keeping are major components in the generation of analytical
data used for environmental assessment and monitoring, the emphasis has historically been
placed on the adherence to technical methods for performing the field sampling and
analysis activities. While technical requirements are usually documented in published
analytical methods and procedures, the criteria for proper documentation and record
keeping is typically not clearly defined or readily available. Analytical data may be
technically acceptable but not defensible due to poor documentation and record keeping
practices. Without adequate and accurate records, data that is subpoenaed for litigation
purposes may be disqualified and any subsequent decisions that were to be made based on
the data may be dismissed.
This paper provides guidance on the proper documentation procedures and record keeping
practices that should be used for environmental field sampling and analysis activities.
General do's and don'ts for documentation as well as ways to streamline the process using
electronic notebooks, bar coded sample labels, and electronic chain of custody records are
also addressed. Specific information that should be recorded are provided for the
following: document control logs, signature lists, field logs, sample bottle labels, chain
of custody records, sample receipt logs, sample receipt checklists, sample preparation logs,
sample analysis logs, instrument run logs, instrument maintenance logs, chemical/standard
receipt logs, standards/reagent preparation logs and standard/reagent labels.
INTRODUCTION
This paper provides general guidelines for documenting information and a list of minimum
information that should be included when recording data for the following categories: (1)
Document Control, (2) Sampling and Sample Handling, (3) Sample Preparation and
Analysis, and (4) Chemical Handling.
GENERAL GUIDELINES
All information related to environmental sampling and analysis activities should be
accurately and adequately documented in order to allow reconstruction and verification by
220
-------
a third party at a future time. General guidelines on how to document information are
listed below.
o Record all data in permanently bound logbooks that have consecutively
numbered pages. Computerized log sheets may be used in place of a bound
logbook if the sheets are numbered and kept in a bound format.
o Record all entries promptly and legibly using indelible ink. Pencil or non-
permanent ink pens should not be used.
o Record information accurately and completely, identifying the who, what,
where, when and how of each activity.
o Do not erase, obliterate or overwrite information. Correction fluid should
not be used.
o Make corrections to errors by drawing a single line through the error and
dating and initialing the correction.
o Cross out blank pages or sections of pages in logbooks with an "X" or "Z"
to prevent the addition of data at a later date.
o Review logs on a regular basis by supervisory personnel to verify
adherence to required analytical and documentation procedures.
DOCUMENT CONTROL
All records and documentation related to environmental sampling and analysis should be
included in a document control system.
Document Control Logs
Document control logs are used to maintain a record of all documents (i.e., logs, forms,
QA Manuals, SOPs, etc.) that should be controlled. Record the following minimum
information in a document control log:
o Title of document
o Assigned document control number
o Location or distribution of document
o Date issued
o Inclusive dates of use
o Date archived
221
-------
Signature Lists
Signature lists are used to keep a record of the signatures and initials of all personnel that
record data. Record the following minimum information in a signature list:
o Employee's typed or printed name and initials
o Employee's handwritten signature and initials
o Employment start date
o Employment termination date
SAMPLING AND SAMPLE HANDLING
Information on the collection and handling of samples should be completely documented
to allow the details of sample collection and handling to be recreated and used with the
sample results.
Field Logs
Field logs are used to document events that occurred during field sampling and to identify
information on individual field samples. For projects involving drilling operations, boring
logs should be used. Record the following minimum information in a field log:
o Project name/ID and location
o Sampling Personnel
o Geological observations
o Atmospheric conditions
o Field measurements
o Sample dates, times, and locations
o Sample identifications
o Sample matrix
o Sample descriptions (e.g., odors and colors)
o Number of samples taken per location
o Description of any QC samples
o Any deviations from the sampling plan
o Any difficulties in obtaining sample or any unusual circumstances
Sample Bottle Labels
Sample bottle labels are used to identify sample bottles and to link each sample with the
information documented in the field log and Chain of Custody Record. The sample bottle
label should contain enough information to avoid misidentification. The use of a bar code
system is recommended for electronically creating the sample bottle labels. Record the
following minimum information on a sample bottle label:
222
-------
o Sample ID (or bar code with ID number)
o Date and time of collection
o Sampler's signature
o Parameter(s)
o Preservative (if applicable)
Chain Of Custody Records
Chain of Custody Records are used to document the custody and exchange of sample(s)
as they are transported from the point of collection to receipt by the laboratory. In some
situations, chain of custody may begin when clean containers are provided by the
laboratory. The use of a bar code system is recommended for completing the Chain of
Custody Record. Record the following minimum information on a Chain of Custody
Record:
o Project name/ID and location
o Sample specific information: sample ID, date/time collected, sample matrix,
container type, preservatives, parameter and method(s) of analysis, etc.
o Sampler's signature
o Signatures of all personnel who had custody of the samples
o Date and time of each transfer
o Air bill or earner ID number
Sample Receipt Logs
Sample receipt logs are used to document the receipt of the samples by the laboratory.
The documentation must be chronological and permanent. Record the following minimum
information in a sample receipt log:
o Date and time of receipt
o Sample collection date
o Client sample ID
o Laboratory sample ID
o Number of samples
o Sample matrices
o Requested analysis, including method number(s)
o Signature or initials of the sample custodian or designee
o Sampling kit code (if applicable)
Sample Receipt Checklists
Sample receipt checklists are used to document the condition of the samples as received
by the laboratory. Any unacceptable conditions or nonconformances with the samples
should be documented and the client notified immediately. Record the following
223
-------
minimum information on a sample receipt checklist:
o Client sample ID
o Laboratory sample ID
o Presence or absence of Chain of Custody Record with samples
o Presence or absence of custody seals
o Condition of custody seals (if applicable)
o Temperature of cooler
o Sample condition at time of receipt
o Acceptability of sample container, volume, and preservation
o Presence of headspace in samples for volatile organics analysis
o Agreement between Chain of Custody Record and sample labels
o Radioactivity screen (if applicable)
SAMPLE PREPARATION AND ANALYSIS
Information on the preparation and analysis of samples should be completely documented
to allow the details of sample preparation and analysis to be recreated and reviewed. Data
generated on laboratory instruments should correspond with the sample information
entered in the preparation and analysis logs. Electronic logs generated through the
laboratory information management system are recommended.
Sample Preparation Logs
Sample preparation logs are used to document the preparation of samples by a specific
method or procedure (e.g., metals digestions or organic extractions). Record the
following minimum information in a sample preparation log:
o Parameter/analyte
o Method number
o Date and time of preparation
o Analyst's initials or signature
o Laboratory sample ID
o Type/matrix of samples
o Initial sample volume or weight
o Final sample volume
o Concentration and amount of spiking solutions used
o Quality control samples included with the sample batch
o ID for reagents, standards and spiking solutions used
o Comments
Sample Analysis Logs
Sample analysis logs are used for documenting information related to the analysis and
224
-------
calculation of the final analytical results. The sample analysis log may either be
parameter specific or may contain several related parameters. Data should be recorded
in the order in which it was generated. Record the following minimum information in a
sample analysis log:
o Parameter/analyte
o Method number
o Date and time of analysis
o Analyst's initials or signature
o Laboratory sample ID
o Sample aliquot
o Dilution factors and final sample volumes (if applicable)
o Absorbance values, peak heights, or initial concentration reading
o Final analyte concentration
o Calibration data (if applicable)
o Correlation coefficient, slope, and y-intercept (if linear regression is used)
o Calculations
o Comments on interferences or unusual observations noted during the
analysis
o Quality control information, including percent recovery for laboratory
control standards and matrix spikes and RPD for sample duplicates.
Instrument Run Logs
Instrument run logs are used to record the analysis of all calibration standards, field
samples and quality control samples processed during an analytical run. An instrument
run log is generally used for recording the analyses performed on all major
instrumentation, such as an ICP, GFAA, FLAA, GC and GC/MS. The data should be
recorded on the instrument run log in chronological order based on the actual analytical
run sequence. Record the following minimum information in an instrument run log:
o Instrument ID
o Parameter/analyte
o Method number
o Date/time analyzed
o Analyst's initials or signature
o Laboratory IDs for standards, field samples, and quality control samples
o Comments
Instrument Maintenance Logs
Instrument maintenance logs are used to document the maintenance and repair activities
performed on major instrumentation. The instrument maintenance log should be kept near
the instrument to facilitate the documentation of all repairs and maintenance activities.
225
-------
Record the following minimum information in an instrument maintenance log:
o Name/type of instrument
o Instrument manufacturer and model number
o Serial number
o Date received and date placed in service
o Instrument ID assigned by the laboratory (if used)
o Service contract information, including the phone number and name
of the service representative
o Description of each maintenance or repair activity performed
o Date and time when each maintenance or repair activity was
performed
o Initials of person who performed the maintenance or repair activity
If an instrument must be removed from service due to a non-routine maintenance
procedure or repair, then the date and reason for removing the instrument from service
should be recorded. Document in the log the repair or non-routine maintenance which
was performed, including the date, service representative's name and company, and the
date the instrument was returned back to service.
CHEMICAL HANDLING
Information on chemical handling should be completely documented to allow the details
of standards preparation to be recreated and traceable from the analysis back to the source
materials. The accuracy and validity of chemicals used for preparing reagents and
standards used for calculating analytical results must be verifiable.
Chemical/Standard Receipt Logs
Chemical/standard receipt logs document the receipt of analytical standards and chemicals
used for the preparation of standards and reagents used in the laboratory. Record the
following minimum information in a chemical/standard receipt log:
o Laboratory control number
o Date of receipt
o Initials or signature of person receiving chemical
o Chemical name and catalog number
o Vendor name and lot number
o Concentration or purity of standard
o Expiration date
All standards and chemicals received by the laboratory should have a chemical receipt
label affixed to the bottle or container. Record the laboratory control number, date of
receipt, and expiration date on the chemical receipt label.
226
-------
Standards/Reagent Preparation Logs
Standards/reagent preparation logs are used for documenting the preparation of standards,
reagents, spiking solutions, surrogate solutions, and reference materials. Record the
following minimum information in a standard/reagent preparation log:
o Date of preparation
o Initials of the analyst preparing the standard solution or reagent
o Concentration and identification of stock solution or neat materials
o Volume or weight of the stock solution or neat materials
o Final volume of the solution being prepared
o Final concentration of the solution being prepared
o Laboratory ID/control number assigned to the new solution
o Standardization of reagents, titrants, etc. (if applicable)
o Expiration date
Standard/Reagent Labels
Standard/reagent labels are used to label standards and reagents. Labels should be affixed
to the container for each standard or reagent prepared by the laboratory. Record the
following minimum information on standard/reagent labels:
o Laboratory ID/control number
o Name of standard or reagent
o Concentration or purity of standard or reagent
o Preparation date
o Initials of preparer
o Expiration date
SUMMARY
As environmental enforcement actions increase, the importance of documentation of
sampling and analysis activities has escalated. By ensuring that technically sound
practices are employed and that information is adequately documented, environmental data
can be verifiable and withstand the intense scrutiny during legal review or litigation.
Good documentation and record keeping procedures not only ensure that environmental
measurement results are accurate and traceable and but also allow appropriate decisions
to be made based on the correct sample results.
REFERENCES
1. Rosecrance, A. and Kibler, L., "Generating Defensible Data", Environmental
Testing & Analysis, May/June 1994.
227
-------
52
NIST SRMs FOR ORGANIC ENVIRONMENTAL ANALYSES -
CURRENT AVAILABILITY AND APPROACH TO CERTIFICATION
Reenie M. Parris. Dianne L. Poster, Lane C. Sander, Michele M. Schantz, and Stephen A. Wise
Analytical Chemistry Division, National Institute of Standards and Technology, B208 Chemistry
Bldg., Gaithersburg, Maryland 20899 USA
ABSTRACT: NIST has many Standard Reference Materials (SRMs) for use by the organic
environmental community ranging from simple solutions containing a number of analytes to
natural matrix materials. SRMs are certified reference materials (CRMs) issued by NIST. NIST
SRMs for use in this field are listed. Certified values for NIST SRMs are generally based on
agreement of results obtained from two or more independent and reliable analytical procedures.
For NIST certification measurements of organic components in complex natural matrix materials,
the extraction and cleanup/isolation steps as well as the final analytical measurements are generally
based on different separation characteristics to minimize the possibility of similar biases in both
measurements. The use of this approach is shown in the certification design of a number of
recently issued NIST SRMs.
Currently Available NIST SRMs
In the past 15 years, the National Institute of Standards and Technology (NIST) has developed a
number of SRMs to assist the environmental community in validating measurements of organic
environmental contaminants such as aliphatic hydrocarbons, polycyclic aromatic hydrocarbons
(PAHs), polychlorinated biphenyl congeners (PCBs), and chlorinated pesticides.Ii2) Table 1 lists
NIST SRMs for this field that are simple solutions, each containing a number of these analytes of
interest for uses such as calibration of chromatographic instrumentation for retention times and
detector response factors or in analyte recovery studies. Sixteen natural matrix NIST SRMs that
are representative of different environmental samples are described in Table 2. These natural
matrix SRMs can be used in the development and validation of analytical methods and can also be
analyzed on a regular basis in conjunction with unknown samples with similar matrices to
continually monitor the accuracy and precision of the analytical measurements.
Certification Measurements of Recently Issued SRMs for Organic Contaminants
Introduction: Certified values for NIST SRMs are generally based on agreement of results
obtained from two or more independent and reliable analytical procedures. For the determination
of organic components in complex natural matrix materials, the extraction and cleanup/isolation
steps as well as the final analytical measurements are based on different separation characteristics
to minimize the possibility of similar biases in both measurements2'. The use of this approach can
be shown in the certification design of a number of recently issued NIST SRMs. For example, the
certification of polychlorinated biphenyl (PCB) congeners and chlorinated pesticides in the
recently issued SRM 1945, Organics in Whale Blubber, included using different
extraction/sample preparation procedures and analytical techniques based on two gas
chromatographic (GC) capillary columns with different selectivity and two different detectors,
electron-capture (ECD) and mass spectrometric (MSD).
228
-------
Determination ofPCBs and Pesticides in SRM 1945: An example of the use of this approach is
shown in the certification of the recently issued NIST SRM 1945, Organics in Whale Blubber",
for PCB congener and chlorinated pesticide concentrations that is described in detail in Reference
4. Two groups of this frozen blubber tissue homogenate samples, were selected from the SRM
sample pool according to a stratified random scheme. Analytical techniques used for these two
groups of samples utilized two different detectors, ECD and MSD. Schantz el al.5\ have reported
and compared the results of the validation of these analytical procedures for the analysis of five
existing natural-matrix CRMs (a marine sediment, a marine tissue, and three fish [cod and
mackerel] oils). The ECD measurements included using two GC capillary columns with different
selectivity (a 5% phenyl-substituted methylpolysiloxane stationary phase and a
dimethylpolysiloxane phase containing 50% methyl C-l 8). For the ECD analyses, samples were
Soxhlet-extracted with dichloromethane, the majority of the lipid and biogenic material was
removed using size exclusion chromatography, and normal-phase liquid chromatography was used
to isolate two fractions containing (1) the PCB congeners and lower polarity chlorinated pesticides
and (2) the more polar chlorinated pesticides. For the GC/MS analyses, the second group of
samples were Soxhlet-extracted with 1:1 hexane/acetone (v/v), sulfuric acid was used to remove
lipid interferences, and a silica solid-phase extraction column was then used to remove the polar
interferences in the extracts. Aliquots of a solution of five internal standards were added to the
blubber prior to extraction for quantification purposes. Calibration response factors for the
analytes relative to the internal standards were determined by analyzing aliquots of the following:
SRM 2261 (Concentrated Chlorinated Pesticides in Hexane), SRM 2262 (Concentrated PCB
Congeners in Iso-octane), gravimetrically prepared solutions of additional analytes not contained
in SRMs 2261 and 2262, and the internal standards. (The purities of the components in the
solution of "additional analytes" were assessed and the concentrations verified
chromatographically.) Samples of SRM 1588, Organics in Cod Liver Oil, were analyzed similarly
and concurrently with the whale blubber samples. Experiments for each technique were designed
so that sources of possible random error in the measurements would be replicated to reduce the
bias they could cause. The results from these three analytical procedures were in good agreement
and were combined to provide certified concentrations for 27 PCB congeners and 15 chlorinated
pesticides4*. A summary of the analytical results by method and the resulting certified
concentrations and uncertainties for 36 of the 42 certified constituents is shown in Table 2.
Determination ofPAHs in SRM 1974a: Similarly, the certification of PAHs in SRM 1974a,
Organics in Mussel Tissue6', is based on results obtained using different extraction/sample
preparation procedures with four different analytical techniques: (1) reversed-phase liquid
chromatography with fluorescence detection (LC-FL) of the total PAH fraction; (2) reversed-phase
LC-FL of isomeric PAH fractions isolated by normal-phase LC (i.e., multidimensional LC); (3)
GC/MS of the PAH fraction on a 5% phenyl-substituted methylpolysiloxane stationary phase; and
(4) GC/MS of the PAH fraction on a smectic liquid crystalline stationary phase that provides
excellent shape selectivity for the separation of PAH isomers. Details of the use of these
techniques have been reported previously for the certification of PAHs in SRM 194la, Organics in
Marine Sediment7-8'.
Certification in progress - SRM 1944: Independent and validated procedures are being used for
the determination of selected PAHs, PCB congeners, and chlorinated pesticides in a new NIST
reference material, SRM 1944, a highly contaminated sediment. In addition, the candidate SRM
229
-------
1944 sediment is being analyzed by a large number of national and international laboratories in
intercomparison exercises for both selected organic and elemental analytes. These data, from
laboratories using a number of different extraction, cleanup, and analytical techniques, will be
evaluated for use in the certification.
CONCLUSIONS
In the past 15 years, NIST has produced many SRMs for trace organic contaminants, especially in
support of measurement activities for the marine environment. Through the use of validated
procedures designed to differentiate among potential sources of measurement bias by exploiting
differences in separation and selectivity characteristics in the extraction, isolation, and
measurement steps, the possibility of having similar biases in these certification measurements is
minimized. Even though these analytical procedures have some similarities (e.g., nonpolar solvent
extraction and GC separations for PCB congeners and chlorinated pesticides) and are therefore not
totally independent, this approach is appropriate for NIST SRM certification given the current
state of the art for these measurements.
REFERENCES
I) NIST Standard Reference Materials Catalog 1995-1996, NIST Spec. Publ. 260, 1995.
2) Wise S.A., M.M. Schantz, R.M. Parris, R.E. Rebbert, B.A. Benner Jr. T.E. Gills (1992): Standard
Reference Materials for Trace Organic Contaminants in the Marine Environment. Analusis, 20, M57-
61.
3) Certificate of Analysis for Standard Reference Material 1945, Organics in Whale Blubber, National
Institute of Standards and Technology, Gaithersburg, MD, 1994.
4) Schantz M.M., B.J. Koster, L.M. Oakley, S.B. Schiller, and S.A. Wise (1995): Certification of
Polychlorinatcd Biphenyl Congeners and Chlorinated Pesticides in a Whale Blubber Standard
Reference Material. Anal. Chem., 67,901-910.
5) Schantz M.M., R.M. Parris, J. Kurz, K. Ballschmiter, S.A. Wise (1993): Comparison of Methods for
the Gas-Chromatographic Determination of PCB Congeners and Chlorinated Pesticides in Marine
Reference Materials. FreseniusJ. Anal. Chem., 346, 766-778.
6) Certificate of Analysis for Standard Reference Material 1974a, Organics in Marine Sediment, National
Institute of Standards and Technology, Gaithersburg, MD, 1995
7) Wise S.A., M.M. Schantz, B.A. Benner, Jr., M.J. Hays, and S.B. Schiller (1995): Certification of
Polycyclic Aromatic Hydrocarbons in a Marine Sediment Standard Reference Material. Anal. Chem.
67, 1171-1178.
8) Schantz M.M., B.A. Benner, Jr., M.J. Hays, W.R. Kelly, R.D. Vocke, R. Demiralp, R.R. Greenberg,
S.B. Schiller, G.G. Lauenstein, and S.A. Wise (1995): Certification of Standard Reference Material
(SRM) 1941a, Organics in Marine Sediment. FreseniusZ. Anal. Chem., 352, 166-173.
9) Ballschmiter K., M. Zell (1980): Analysis of Polychlorinated Biphenyls (PCB) by Glass Capillary Gas
Chromatography, Fresenius Z Anal. Chem., 302, 20-31.
230
-------
Table 1. NIST Calibration Solution SRMs for the Determination of Organic Environmental Contaminants
no
w
Date
SRM No. Title Issued
1491 Aromatic Hydrocarbons in Hexane/Toluene 1989
1492 Chlorinated Pesticides in Hexane 1992
1493 Chlorinated Biphenyl Congeners in 2,2,4-Trimethylpentane 1995
1494 Aliphatic Hydrocarbons in 2,2,4-Trimethylpentane "
1583 Chlorinated Pesticides in 2,2,4-Trimethylpentane 1985
1584 Priority Pollutant Phenols in Methanol 1984
1585 Chlorinated Biphenyls in 2,2,4-Trimethylpentane 1986
1586 Isotopically Labeled and Unlabeled Priority Pollutants 1984
in Methanol
1587 Nitrated PAHs in Methanol 1985
1596 Dinitropyrene Isomers and 1 -Nitropyrene in Methylene Chloride 1987
1614 Dioxin (2,3,7,8-TCDD) in Iso-octane 1985
1639 Halocarbons (in Methanol) for Water Analysis 1983
1647c Priority Pollutant PAHs (in Acetonitri le) 1992"
2260 Aromatic Hydrocarbons in Toluene 1991
(Nominal Concentration 60 (Jg/mL)
2261 Chlorinated Pesticides in Hexane 1992
(Nominal Concentration 2 ug/mL)
2262 Chlorinated Biphenyl Congeners in 2,2,4-Trimethylpentane 1995
(Nominal Concentration 2 ug/mL)
2269 Perdeuterated PAH-I
2270 Perdeuterated PAH-II
2273 DDTs and Metabolites in Solution
2274 PCB Congener Solution-II
2275 Chlorinated Pesticide Solution-II
2276 Three Planar PCBs in Solution
Certified
Constituents
PAHs (23)
Pesticides (15)
PCBs (18)
Hydrocarbons (20)
Pesticides (5)
Phenols (10)
PCBs (8)
Priority pollutants (10)
Nitro-PAHs (6)
Nitro-PAHs (4)
Dioxins (2)
Halocarbons (7)
PAHs (16)
PAHs (23)
Pesticides (15)
PCBs (25)
Perdeuterated PAHs (5)
Perdeuterated PAHs (6)
DDTs and Metabolites (7)
PCBs (11)
Pesticides (9)
PCBs (3)
Noncertificd
Constituents
PAHs(l)
PCBs (2)
Pesticides (1)
Phenols (1)
Nitro-PAHs (1)
Dioxins (2)
PAHs(l)
PCBs (4)
Certification in progress.
SRM 1647 was first issued in 1981.
-------
Table 2. NIST Natural Matrix SRMs for the Determination of Organic Contaminants in Environmental Samples
SRMNo. Title Date Issued Certified Constituents Noncertified Constituents
ro
o
ro
1580 Organics in Shale Oil
1581 Polychlorinated Biphenyls in Oils
1582 Petroleum Crude Oil
1588 Organics in Cod Liver Oil
1589 PCBs (as Aroclor 1260) in Human Serum
1597 Complex Mixture of PAHs from Coal Tar
1648 Urban Paniculate Matter
1649 Urban Dust/Organics
1650 Diesel Paniculate Material
1939 PCBs (Congeners) in River Sediment
194la Organics in Marine Sediment
1944 Highly Contaminated Marine Sediment
1980
1982
1984
1989*
1985
1987
1978
1982*
1985*
1990
1994
**
1945 Organics in Whale B lubber (Frozen) I993
1974a Organics in Mussel Tissue (Mytilus edulis) (Frozen)
1975 Diesel Paniculate Extract
2974 Organics in Mussel Tissue (Mytilus edulis)
(Freeze Dried)
PANH = Polycyclic aromatic nitrogen heterocycles
PAQ = Polycyclic aromatic quinones
PASH = Polycyclic aromatic sulfur heterocycles
**
PAHs (5); Phenols (3)
PANH(l)
Aroclor 1242/1260
PAHs (5); PASH (1)
PCBs (5); Pesticides (10)
Aroclor 1260
PAHs (12)
Trace elements (9)
PAHs (5)
PAHs (5)
Nitro-PAHs(l)
PCBs (3)
PAHs (23); PCBs (21)
Pesticides (6)
PAHs; PCBs; Pesticides
Trace Elements
PCBs (27); Pesticides (15)
1995
Trace Elements (28)
Pesticides (7)
PAHs; Nitro-PAHs
PAHs; PCBs; Pesticides
Methylmercury
Phenols (6); PANH (1)
PAHs (5); Phenols (2); PANH (1)
PCDDs/PCDFs (7)
Dioxins (2)
PAHs/PASH/PANH(18)
Trace elements (25); PAH (13)
PAHs (9)
PAHs (6);
Nitro-PAHs (3); PAQ (1)
PCBs (14); Pesticides (5); PAHs (5)
PAHs (14); PCBs (7); Pesticides (3)
Aliphatics (17); Trace Elements (27)
Aliphatics, PCDDs/PCDFs
PCBs (2); Pesticides (2)
PAHs (15); PCBs (20)Aliphatics (16);
PAHs (18); PCBs (4); Pesticides (4)
Methylmercury
Aliphatics, Trace Elements
PCDD = Polychlorinated dibenzodioxins
PCDF — Polychlorinated dibenrofurans
= Certification of additional analytes in progress.
* = Certification in progress
-------
Table 3. Certified Concentrations and Summary of Analytical Results for 36 of the 27 PCB
Congeners" and 15 Chlorinated Pesticides Certified in SRM 1945, Organics in Whale
Blubber31
Concentration (ug/kg wet basis)
Certified"
GC-ECDC
Constituent Concentration (CP Sil 5 CIS CB)d
PCB 18
PCB 44
PCB 49
PCB 52
PCB 66
PCB 95
PCB 66/95
PCB 87
PCB 99
PCB 101/90
PCB 105
PCB 118
PCB 128
PCB 138/163/164
PCB 149
PCB 156
PCB 170/190
PCB 180
PCB 183
PCB 187
PCB 194
PCB 206
PCB 209
2,4'-DDE
4,4'-DDE
2.4'-DDD
4,4-DDD
2,4'-DDT
4,4'-DDT
HCB
Y-HCH
ct-HCH
heptachlor epoxide
oxychlordane
cis-chlordane
mwii-nonachlor
Mirex
4.48
12.2
20.8
43.6
23.6
33.8
16.7
45.4
[65.2
30.1
74.6
23.7
131.5
106.6
10.3
[40.6]
106.7
36.6
105.1
39.6
31.1
10.6
12.28
445
18.1
133
106
245
32.9
3.30
16.2
10.8
19.8
46.9
231
28.9
±
+
±
±
+
+
±
+
+
+
±
±
±
±
±
±
+
±
±
±
±
±
±
±
+
+
+
±
±
±
±
±
±
±
±
±
0.88
1.4
2.8
2.5
1.6
1.7
1.4
5.4
5.6]
2.3
5.1
1.7
7.4
8.4
1.1
2.6
5.3
4.1
9.1
2.5
2.7
1.1
0.87
37
2.8
10
14
15
1.7
0.81
3.4
1.3
1.9
2.8
11
2.8
4.91
12.25
18.54
45.1
23.2
34.1
16.25
41.5
61.6
31.6
74.9
23.2
131.7
101.6
9.76
39.3
105.5
39.5
111.7
39.2
32.9
10.75
12.46
421
20.0
128
113.0
238
32.2
3.90
18.6
10.74
20.6
47.2
229
30.1
(0.32)
(0.64)
(0.84)
(2.5)
(1.8)
(1.7)
(0.68)
(2.9)
(3.3)
(2.4)
(4.7)
(2-2)
(8.0)
(6.6)
(0.69)
(2.6)"
(6.2)
(2.9)
(7.0)
(2.4)
(1.9)
(0.87)
(0.57)
(31)
(1.3)
(10)
(7.0)
(46)
(2.1)
(0.37)
(1.2)
(0.87)
(1.5)
(3-2)
(16)
(2.3)
GC-ECDC
(DB-5)d
3.78
13.17
22.64
42.8
[59.7]'
17.71
47.3
[66.3]
30.0
75.6
24.6
127.7
106.6
10.93
[40.2]
108.8,
36.0
103.1
41.3
30.7
11.16
11.69
453
16.2
132.4
98
33.1
2.63
14.4
10.1
20.7
46.0
233
29.6
(0.30)
(0.74)
(0.74)
(2.9)
(2-8)
(0.91)
(2.8)
(4.3)
(2.4)
(4.8)
(2-1)
(9-0)
(7-5)
(0.66)
(2.6)
(6.7)
(2-8)
(6-7)
(2.6)
(1.9)
(0.94)
(0.70)
(25)
(1.4)
(9.1)
(10)
(2.2)
(0.37)
(1.3)
(1.2)
(1.0)
(3.4)
(16)
(2.3)
GC-MSD0
(DB-5)"
4.76
11.25
21.28
42.4
23.7
33.6
16.66
47.3
[68.3]
29.8
71.7
22.9
134.2
110.6
10.07
[42.4]
109.5
34.4
100.9
39.4
29.7
9.95
12.27
440
17.96
138.3
101.6
246
33.8
3.37
15.3
11.2
18.61
47.2
232
26.8
(0.33)
(0.94)
(0.95)
(2-1)
(1.9)
(1.6)
(0.47)
(2.3)
(2.3)
(1.3)
(3.7)
(1.5)
(5.7)
(4-7)
(0.72)
(1.9)
(4.9)
(1.8)
(5.1)
(1.8)
(1.6)
(0.85)
(0.54)
(34)
(0.93)
(6-2)
(6.3)
(12)
(1.8)
(0.24)
(1.0)
(1.0)
(0.83)
(2-5)
(10)
(2.2)
233
-------
PCB congeners are numbered according to the scheme proposed by Ballschmiter and Zell91.
When two or more congeners are known to coelute, the PCB congener listed first is the major
component and the additional congeners may be present as minor components. Quantitative
results are based on the response of the congener listed first.
The certified values are weighted means of results from three analytical techniques. The
uncertainty is based on a 95% confidence interval for the true concentration and includes an
allowance for differences between the analytical methods used.
Numbers in parentheses are one standard deviation of a single measurement.
Capillary column stationary phase
Numbers in [ ] indicate known coelution of two or more congeners. PCB 66 and 95 coelute
under the GC-ECD (DB-5 column) conditions used; PCB 164 is separated from PCB 138 and
PCB 163 when using the CP Sil 5 CIS column; PCB 190 is separated from PCB 170 when
using the CPSil 5 CIS column.
234
-------
53
ABSTRACT
DATA USABILITY: THE NEXT STEP TO DATA VALIDATION
FOR SITE REMEDIATION
URS CONSULTANTS, INC.
282 DELAWARE AVENUE
BUFFALO, NEW YORK 14202-1805
(716) 856-5636
(716) 856-2545 (fax)
AUTHOR: PETER R. FAIRBANKS
In order to properly evaluate analytical data and determine its usability for site remediation,
the data reviewer must not only understand the analytical methods and how to apply USEPA
validation qualifiers to the data, but must be familiar with the project-specific documents
(e.g., investigative reports and sampling and analysis plans) and fully understand the
objectives of the project. The objectives to consider include: the type of investigation taking
place (e.g., site characterization, remedial investigation/feasibility study, human/ecological
risk assessments, engineering design, etc.); the principal contaminants of concern; the
regulatory agencies reviewing the data; and the comparison of reported values/detection
limits to regulated cleanup limits and/or health-based criteria. These items must be
considered before making a complete assessment of the analytical results. Based on our
experience, we will present how these factors affect the validation process and the ultimate
usability of the data.
Oftentimes, data usability is determined solely on the basis of the data validation process.
Published data validation guidelines, such as USEPA CLP National Functional Guidelines
for Data Review, depict methods to alert the data user of potential problems by applying
qualifiers to the data. An overview of the input of flags on data and their interpretation and
impact on usability will be discussed.
Rejection of data would be the worst case scenario involving data usability. This could lead
to resampling/reanalysis, thus driving up the cost of investigations. However, data which
are rejected based upon USEPA data validation criteria may be deemed conditionally usable
when applying professional judgement in conjunction with project-specific information.
Rejecting data solely on data validation guidelines may be premature. The data reviewer
must understand the project data quality objectives and the intended use of the data before
deciding to qualify the data. Properly evaluated data potentially reduces the need for
resampling/reanalysis, thereby minimizing unnecessary project costs. We will describe the
validation process with a focus on specific case studies from a user's perspective.
235
-------
54
NELAC Update.
/?. Trovato
236
-------
55
THE SHELL REQUIREMENTS FOR CHEMISTRY DATA GENERATED FOR USAGE
PROJECTS
Joseph F.Solskv. Chemist, U.S. Army Corps of Engineers, CEMRO-HX-C, 12565 West Center
Road, Omaha, Nebraska, 68144
ABSTRACT
The purpose of the 'Shell' is to establish the basic approach to be used when performance-based
methods, especially the SW-846 methods, are employed by the U.S. Army Corps of Engineers
(USACE) for the analytical testing of environmental samples. These methods are flexible and can be
readily adapted to individual project-specific requirements. The chemistry data generated for
USAGE projects must be produced by a process or system of known quality to withstand scientific
and legal challenge relative to the use for which the data are obtained. The 'Shell' outlines such a
process. A related document, the 'Method's Compendium,' is available that applies the concepts as
presented in the 'Shell' to specific SW-846 methods for relatively critical data uses. The USACE
has also developed laboratory specific standard operating procedures (SOPs) consistent with the
'Shell' for many SW-846 methods and general laboratory operations.
Project-specific data quality objectives (DQOs) must be established for both the field and laboratory
operations. Each laboratory would normally define their own set of laboratory-specific objectives for
general day-to-day use for their implementation of any given method. Any differences between
project DQOs and lab operational criteria must be reconciled before project execution. For each
project, data quality must be demonstrated for the analytes of concern at the levels of concern.
laboratories are required to maintain written, approved SOPs for all methods and operations. The
demonstration of method proficiency begins with establishing the basic sensitivity of the method by
determining the method detection limit (MDL). The relationship between the MDL, the method
quantitation limit, the initial multi-point calibration curve, and the laboratory's method reporting
limit is established. A laboratory cannot claim to reliably quantitate values below the low standard
or above the high standard. A given method is suitable when the laboratory's low standard is below
the site action level for each analyte of concern.
The 'Shell' describes the requirements for instrument calibration. Initial calibration curves are
verified with standards from an independent source. Continuing calibration verification is performed
for all method target analytes at the beginning and end of each analytical sequence. Each preparation
batch is to contain a method blank and a laboratory control sample containing all of the project-
specific analytes of concern spiked at the levels of concern to monitor laboratory performance. Each
preparation batch would typically contain additional QC samples to monitor the effect of the matrix
on the method. Corrective actions are carefully detailed and involve interaction with project
managers to avoid the generation of a significant amount of flagged data.
The intent of the 'Shell' is to ensure the generation of chemistry data of known quality. Laboratories
should not be treated as black boxes where samples go in and only numbers come out. Laboratories
employ chemists and others who are experts in the interpretation of analytical data. Much is to be
gained by enhancing the interaction between the laboratory and project personnel. The 'Shell'
encourages this.
237
-------
INTRODUCTION
The U.S. Army Corps of Engineers (USAGE) currently executes remedial and compliance activities
under several environmental regulatory programs. The analytical testing of various environmental
samples is often a significant part of these activities. The data must be produced by a process or
system of known quality to withstand scientific and legal challenge relative to the use for which the
data are obtained. All USACE projects require a comprehensive and multifaceted approach to
quality assurance and quality control (QA/QC) to achieve and document attainment of appropriate
quality for the intended data usage. The project technical team must establish project-specific quality
standards for chemical data early in the planning process. These standards are documented as
project-specific data quality objectives (DQOs).
The purpose of this document (the 'Shell') is to establish the basic approach to be used when
performance-based methods, especially the SW-846 methods, are employed by the USACE during
the execution of its projects. The strategies presented in the 'Shell' shall be used by laboratories in
conjunction with the project-specific DQOs when providing chemistry analytical services to the
USACE. A related document, the 'Method's Compendium1, is available that applies the concepts as
presented in the 'Shell' to specific implementations of the SW-846 methods. Default target analyte
lists are presented along with default acceptance criteria for relatively critical data uses. The USACE
also has available laboratory specific standard operating procedures (SOPs) for many SW-846
methods and general laboratory operations. These SOPs implement the concepts as presented in the
'Shell' and are now being carried out by the USACE Division Laboratories.
As stated in the Final Rule that incorporated the Third Edition of SW-846 (and its updates) into the
RCRA regulations, this document is required to be used for certain activities in the RCRA program.
hi other situations, this EPA publication functions as a guidance document setting forth acceptable,
although not required, methods to be implemented by the user, as appropriate, in satisfying RCRA-
related sampling and analysis requirements.
SW-846 covers many separate test methods addressing hundreds of analytes. For any given analyte,
multiple methods, with varying detection limits, are potentially available. SW-846 is a dynamic
document that is constantly changing as new information and data are developed. Any of these
promulgated or draft SW-846 methods or any other methods may be used by the USACE to support
the project-specific requirements.
The 'Shell' is not a stand alone document. The 'Shell' does not reference the use of specific
analytical methods, does not cite specific target analyte lists for individual methods, and does not
provide default QA/QC criteria for individual methods. The 'Shell' does outline the strategy for
implementation for any given method and must be used in conjunction with the project-specific
DQOs. The use of the 'Shell' provides for a uniform implementation of the various analytical testing
methods by the laboratory community and clarifies the ambiguities that are present when laboratories
implement the SW-846 and other performance-based methods. The 'Shell' and 'Method's
Compendium' shall be incorporated into EM200-1-3 (Requirements for the Preparation of Sampling
and Analysis Plans dated 1 September 1994) referenced by ER1110-1-263 (Chemical Data Quality
Management for Hazardous, Toxic, Radioactive Waste Remedial Activities dated 1 April 1996).
ER1110-1-263 is the umbrella regulation that defines the chemical data quality management and
integrates all of the other USACE guidance on the topic.
238
-------
QUALITY ASSURANCE OBJECTIVES FOR DATA MEASUREMENT
Project-specific DQOs must be established for both the field and laboratory operations. The DQO
process will assist in determining the appropriate reporting limits, extraction/digestion methods,
clean-up methods, analytical methods, target analytes, method quality control samples, quality
control acceptance ranges, and corrective actions. Chemical data is typically procured by separate
contracts between the architect/engineering (A/E) firm and the laboratory. The contract laboratories
are bound by the DQOs as approved in the final documents for each project. Project-specific DQOs
should not be confused with laboratory-specific objectives. Each laboratory would define their own
set of laboratory-specific objectives for general day-to-day use for its implementation of any given
performance-based method, including the SW-846 methods. Any differences shall be reconciled
prior to project execution. If differences are noted, then either a different method must be chosen that
does meet the project-specific DQOs or the current laboratory method must be modified.
The determination of data quality includes the evaluation of the PARCC parameters (i.e., precision,
accuracy, representativeness, comparability, and completeness) and sensitivity. The PARCC
parameters, including sensitivity, are quantitative or qualitative limits that, when exceeded, generate
data that is questionable for the intended use. It is important that data quality be demonstrated for
the analytes of concern at the levels of concern. Of these parameters, sensitivity is the one that has
caused the greatest confusion. The 'Shell' interrelates the concepts of the method detection limit
(MDL), method quantitative limit (MQL), and method reporting limit (MRL). Under the 'Shell'
strategy, the laboratory shall perform MDL studies annually, as a minimum, and whenever the basic
chemistry of the procedures are changed using the procedures presented in 40 CFR, Part 136,
Appendix B. The MDLs shall be extraction/digestion method specific and shall include all method
target analytes. The MDLs shall be demonstrated on individual instruments when multiple
instruments are used for any given method. To ensure that reasonable MDL values are determined,
the laboratory should analyze an MDL check sample quarterly by spiking an interference free matrix
with all method target analytes at about two times the determined MDL. If any of the method target
analytes are not recovered, then the MDL study shall be repeated. The determination of method
detection limits in site-specific matrices may be required for certain projects. The method
quantitation limit (MQL) is defined as the lowest calibration standard used during the initial
calibration. The lowest calibration standard would normally be set at between three and ten times the
MDL for each method target analyte. The method reporting limit (MRL) can be set no lower than
the value of the low standard used during initial calibration. The low standard would typically be
below the project-specific action levels or regulatory levels. This item is critical since it will directly
affect the use and interpretation of laboratory-specific quality control samples. A laboratory shall
not claim to reliably quantitate values below the low standard. Analyte values can be reported below
the low standard but shall be reported as estimated (flagged) values.
A laboratory shall prepare a written Quality Assurance Plan (QAP) (or Laboratory Quality
Management Manual (LQMM)) which describes the general and specific procedures used within the
laboratory to achieve scientifically sound and legally defensible data.
ANALYSIS REQUIREMENTS
The laboratory shall have, in-house, the appropriate standards for all method target analytes.
239
-------
Laboratories shall be required to maintain written, approved laboratory-specific standard operating
procedures (SOPs) for all methods and general operations. Laboratory-specific SOPs are mandatory
that fully detail the actual procedures and documentation used to implement performance-based
methods that are often presented as guidance. Simply referencing a given method or method number
is not sufficient.
For each method performed, the laboratory shall maintain documentation that demonstrates the
laboratory's ability to perform the method within the sensitivity and precision/bias limits as stated in
the method, if applicable. For each method performed, the laboratory shall maintain documentation
that demonstrates each employee's ability to perform the method within the precision/bias limits as
determined by the laboratory or as stated in the method, if applicable.
Under certain conditions, due to the introduction of new technology or when an unusually difficult
matrix is encountered, modifications to the methods may be necessary. The MDL and the
precision/bias noted for a modified method shall be no worse than the performance criteria as given
in the original method unless the project-specific DQOs can still be achieved. The demonstration of
equivalency shall be required whenever the basic chemistry of the original method has been altered
and may need to be repeated when samples from different matrices are received. This decision will
be based on project-specific DQOs and will be made on a case by case basis.
The data produced by a laboratory typically provide the primary basis for environmental cleanup
decisions and enforcement actions. The data may also end up in court. To be legally defensible, the
data must be produced according to the project-specific requirements as specified in the final
approved project documents. The laboratory must be aware of these requirements and be able to
show that these requirements were followed. Documentation that would clearly show how all
analytical values were obtained must be maintained. There is a significant difference between
scientifically valid and legally defensible data. Intentional falsification of process results or quality
control parameters, or failure to document actual conditions for the purpose of misrepresentation,
constitutes fraud. Inappropriate use of manual integrations, for example, to meet calibration or
method QC criteria would be considered fraud. Mistakes happen, but mistakes must be documented
and corrected.
CALIBRATION PROCEDURES AND FREQUENCIES
Calibration of instruments is required to ensure that the analytical system is operating correctly and
functioning at the proper sensitivity. All reported method target analytes shall be within the high and
low initial calibration standards. All samples analyzed shall be sequentially bracketed by calibration
verification standards that typically would include all method target analytes.
Linearity is best determined using linear regression analysis for each method target analyte by
calculating the 'Coefficient of Determination' (r2, the squared correlation coefficient). The use of the
'Coefficient of Determination' should be used in preference to the use of average calibration or
response factors. A visual inspection of the calibration curve should be performed. If the visual
inspection indicated problems or if the acceptance criterion was not met, then the laboratory shall
evaluate each of the following items in the listed sequence: (1) The first option would be to check the
instrument operating conditions or the standards used and make adjustments to achieve a linear
calibration curve; (2) The second option would be to narrow the calibration range using the same
240
-------
number of standards as required by the individual method; (3) The third option would be to use a
nonlinear calibration curve. The statistical considerations in developing a nonlinear calibration
model require more data than the more traditional linear approaches.
The initial calibration curve shall be checked with a standard purchased or prepared from an
independent source. This initial calibration verification (1C V) involves the analysis of a standard
containing all of the method target analytes each time the initial calibration is performed.
Continuing calibration verification (CCV) involves the analysis of a single standard at the beginning
and end of each analytical sequence. For methods that contain multi-component target analytes,
typically only a subset of these analytes would be used in the CCV. Calibration Verification1 differs
in concept and practice from 'continuing calibration1. Calibration verification shall be used for all
analytical methods.
As an alternative for ICP initial multipoint calibration, a single high standard and a calibration blank
may be used for initial calibration if calibration is verified using both mid and low level standards.
QUALITY ASSURANCE AND QUALITY CONTROL PROCEDURES
Laboratory internal quality control checks shall be designed to determine if laboratory operations are
in control and to determine the effect of the sample matrix on the data being generated. Laboratory
performance QC shall be based on the analysis of a laboratory control sample to generate precision
and accuracy data that are compared to control limits and on the analysis of a method blank. Matrix
specific QC shall be based on the use of an actual environmental sample for precision and accuracy
determinations and commonly relies on the analysis of matrix spikes, matrix spike duplicates, matrix
duplicates, and surrogate spikes, if appropriate.
The basic unit for laboratory quality control is the batch. Samples shall be prepared and analyzed in
batches and be traceable to these individual batches. Batch sizes are normally limited to twenty field
samples of a similar matrix. Samples prepared together would normally be analyzed together on a
single instrument. Samples taken from the same site would normally be grouped together. If
samples from multiple clients are grouped into a single batch, additional batch QC samples may be
needed that evaluate the effect of the matrix from each site on method performance.
The preparation batch shall be defined as samples that are prepared together by the same person, or
group of people, using the same equipment/glassware with the same method sequence and the same
lots of reagents and with the manipulations common to each sample within the same time period or in
limited continuous sequential time periods, usually not to exceed one analytical shift. The analytical,
or instrumental, batch shall be defined as samples that are analyzed together within the same time
period or in sequential continuous time periods. Within this period of time, each batch may be
subdivided into individual run sequences. Run sequences would be bracketed by the appropriate
continuing calibration verification standards and other QC samples (i.e., tunes, instrument blanks,
etc.) as defined by each method.
A summary of the minimum required QC samples for each preparation batch follows. (1) Method
blanks (MB) are analyzed to assess background interference or contamination that exists in the
analytical system. The method blank is defined as an interference-free blank matrix similar to the
241
-------
field sample matrix to which all reagents are added in the same volumes or proportions as used in
sample preparation and carried through the complete sample preparation, cleanup, and analytical
procedures. At least one method blank shall be analyzed with each preparation batch of samples. (2)
The laboratory control sample (LCS) is analyzed to assess the ability of the laboratory's method to
successfully recovery the method target analytes from a control sample of known composition. The
LCS is similar in composition to the method blank and is spiked with all of the single component
method target analytes, and a subset of the multi component method target analytes, before it is
carried through the complete sample preparation, cleanup, and analytical procedures. The spiking
levels would normally be set at the project specific action limits. At least one LCS shall be analyzed
with each preparation batch of samples. Control charts, or tables, shall be maintained for these
samples to establish the accuracy and precision of the method. (3) Matrix spike samples are
analyzed to assess the ability of the method to successfully recover the method target analytes from
an actual matrix. A MS is an environmental sample to which known concentrations of the method
target analytes, or a subset of representative method target analytes, have been added before it is
carried through the complete sample preparation, cleanup, and analytical procedures. The levels
would normally be set at the project specific action limits. If method target analytes were known to
be present in samples from a given site, then the spiking level should be adjusted to a concentration
that would approximately double the concentrations of the original method target analytes. The
sample to be used for the MS would normally be specified in the field. The requirements for the
frequency and quantity of MSs would be specified in the project specific DQOs. For batches
consisting of samples from the same site for the same matrix, the frequency of analysis of the MS
would normally be one per preparation batch. A MS from one site could not be used to evaluate the
matrix effects on samples from other sites. (4) The matrix duplicate (MD) and/or matrix spike
duplicate (MSD) is analyzed to assess the precision of the method in an actual matrix. The
requirements for the frequency of MDs or MSDs would normally be specified in the project-specific
DQOs. A MD would normally be included with each preparation batch of samples processed where
method target analytes were expected to be present. An MSD would normally be included with each
preparation batch of samples processed where method target analytes were not expected to be
present. (5) Certain methods may require additional QC procedures or samples besides those already
identified.
The laboratory is encouraged to analyze additional standard reference materials (SRMs) and
participate in external performance evaluation (PE) programs.
CORRECTIVE ACTIONS
When errors, deficiencies, or out-of-control situations exist, the laboratory's QA program shall
provide systematic procedures, called corrective actions, to resolve problems and restore proper
functioning to the analytical system(s). Problems noted during sample receipt shall be documented.
The project manager, or their designee, shall be contacted immediately for problem resolution. If
samples cannot be prepared and/or analyzed within the method required holding times, the project
manager, or their designee, shall be immediately notified, such that an appropriate corrective action
plan can be generated. Sample analysis shall not be allowed until all initial calibrations and initial
calibration verifications meet the appropriate requirements. All continuing calibration verifications
that do not meet method requirements shall result in a review of the calibration. Continued failure of
the CCV shall result in the construction of a new initial calibration curve followed by the reanalysis
of all samples affected. All method QC shall meet the appropriate project-specific requirements and
242
-------
associated corrective actions. Failure of method QC shall result in the review of all affected data. If
no errors can be noted, the affected sample(s) may need to be reanalyzed or reprepped then
reanalyzed within method holding times, if possible. The project manager, or their designee, shall be
notified as soon as possible to discuss possible corrective actions should unusually difficult sample
matrices be encountered.
For method blanks, the concentration of all method target analytes shall be below the MDL for each
method target analyte, or less than 5 percent of the regulatory limit associated with that analyte, or
less than 5 percent of the sample result for the same analyte, whichever is greater. The first step of
corrective action is to assess the effect on the samples Typically, the contaminated samples of the
batch would be reprepped and reanalyzed with a new method blank and batch specific QC samples.
The LCS is evaluated by comparing the recovery for all of the method target analytes to the recovery
windows as determined by the laboratory or as specified in the project-specific DQOs. A batch of
samples shall be considered acceptable only for those analytes that had acceptable recoveries in the
LCS. The first step of corrective action is to assess the effect on the samples. Typically, the LCS,
method blank, and all associated samples of the batch would be reprepped and reanalyzed for the
failed analytes.
The MS is evaluated by comparing the recovery for all method target analytes to the recovery
windows as determined by the laboratory for the LCS or as specified in the project-specific DQOs.
MS data evaluation is more complex than method blank or LCS data evaluation. In general,
laboratories will not base batch control on the results of the MS unless a general method failure is
indicated. MSs that fail to meet the appropriate acceptance criteria would indicate that a potential
matrix effect is present. The associated sample(s) would typically be reprepped then reanalyzed to
verify the effect. Corrective action shall include the immediate notification of the project manager, or
their designee, to discuss the implementation of additional cleanup procedures, method
modifications, etc. to compensate for the matrix effects noted.
The MSD is evaluated using the same accuracy criteria as described for the MS. The MD or MSD is
evaluated by comparing the precision for all method target analytes to the windows as determined by
the laboratory for the LCS or as specified in the project-specific DQOs. Corrective actions shall be
performed as described for the MS.
DATA REDUCTION. VALIDATION. AND REPORTING
All analytical data generated by the laboratory shall be extensively reviewed prior to report release to
assure the validity of the reported data. This internal data evaluation process shall include a
minimum three levels of documented review.
Level 1 review shall be performed by the analyst based on an established set of guidelines. Level 1
analyst review shall be documented by using a check list and by the signature and date of the
reviewer.
Level 2 review shall be performed by a supervisor or data review specialist whose function is to
provide an independent, peer review of the data package. Level 2 review shall be structured so that
all calibration data and QC sample results are reviewed and all of the analytical results from at least
243
-------
25% of the samples are checked back to the raw data and/or bench sheets. If no problems are found
with the data package, the review is complete. If any problems are found with the data package, then
all sample results shall be checked. Level 2 peer review shall also be documented on a check list
with the signature and date of the reviewer.
Level 3 review shall be performed by the program administrator at the laboratory. This review shall
provide a total overview of the data package to ensure its consistency and compliance with project-
specific requirements. Level 3 administrative review shall also be documented on a check list with
the signature and date of the reviewer.
QA review shall be performed by the QA Officer or QA Branch. This review is not part of the
normal production data review process. The QA Officer would typically review at least 10 percent
of the data produced by the laboratory using the procedures as outlined in the Level 3 data review.
Additional technical details could be reviewed based upon the results of this QA review.
Data qualifiers shall be added by the laboratory during the data generation/review process. These
qualifiers would be applied when acceptance criteria were not met and corrective action was not
successful or when corrective action was not performed. All flags used by the laboratory shall be
defined. These flags should also identify any suspected bias in the data. The project manager, or
their designee, shall be notified as soon as possible to discuss possible corrective actions should data
be qualified. Additional data flagging may be performed by the USAGE based upon overall project-
specific requirements, by using external data review or validation.
The USAGE does require a minimum level of data reporting that includes the reporting of all sample
data along with the supporting QC information. This format allows for the review and validation of
the data by an independent organization. This format does not allow for the complete independent
reconstruction of the analytical data. An electronic deliverable would also be highly recommended.
The format of this electronic deliverable shall be specified in the project-specific documents.
SUMMARY
The intent of the 'Shell' is to ensure the generation of chemistry data of known quality. The strategy'
as presented in the 'Shell' is applicable to the SW-846 methods and to any performance-based
method. This strategy should reduce the ambiguities that presently exist between the implementation
of methods between different laboratories. The interaction of project and laboratory staff is critical
at the beginning and during project execution to ensure that project-specific DQOs can be and are
being achieved. This interaction should reduce the amount of flagged data being generated for any
given project
244
-------
56
INTEGRATION OF THE DQO PLANNING PROCESS IN THE CLOSURE OF MIXED WASTE
STORAGE UNITS
Terrv Y. Hosaka Project Manager, Michael H. Schlender Program Manager.
Brian J. Day Environmental Compliance Representative, Del ores K. Lutter
Senior Environmental Engineer. Battelle, Pacific Northwest National
Laboratory. Battelle Boulevard P.O. Box 999. Richland. Washington, 99352;
Mitzi S. Miller President, Environmental Quality Management, 10801 Fox
Park. Knoxville, Tennessee. 37931
ABSTRACT
The Data Quality Objectives (DQO) Planning Process was used to facilitate
the completion of a Resource Conservation Recovery Act (RCRA) Closure Plan
for the 324 Radiochemical Engineering Complex (REC) and High Level Vault
(HLV) tanks.
INTRODUCTION
The Data Quality Objectives (DQO) planning process was used to define
requirements in the development of a RCRA Closure Plan for unpermitted
mixed-waste storage units at the Hanford Site. The DQO process is a
facilitated decision-making method; it develops in logical step-wise
fashion a solution to which all key decision makers and involved parties
agree. Developing a Closure Plan for an unpermitted mixed waste storage
unit carries with it many issues that do not lend themselves to resolution
by industry convention or standards. In the development of the Closure
Plan, the DQO process proved to be a valuable tool in resolving all the
key issues in a timely and mutually acceptable manner.
Background
The Pacific Northwest National Laboratory (PNNL) in support of the U.S.
Department of Energy - Richland Operations (DOE-RL) is preparing a RCRA
Closure Plan for unpermitted waste storage units in the 324 Building. 300
Area. Hanford Site. The waste is stored in the Radiochemical Engineering
Complex (REC) in the 324 Building. The REC consists of four hot-cells
with a common airlock, two underground vaults that contain stainless-steel
liquid-storage tanks . and ancillary support components. The wastes are
highly radioactive and are stored in the REC because no facility now in
245
-------
operation can accept high activity mixed waste for Interim storage or
final disposition. The facility Is being closed under a consent order
manifested as a Tri-Party Agreement (TPA) Milestone. The TPA is a Hanford
Facility Clean-up agreement between the U.S. Department of Energy (DOE).
U.S. Environmental Protection Agency (EPA), and Washington State
Department of Ecology (Ecology).
In the REC the mixed wastes are stored in B Cell, tanks in the High Level
Vault (HLV). and tanks in the Low Level Vault (LLV) (See Figure 1.0). B
Cell is 7.6 m long x 6.7 m wide x 9.3 m high and is the largest of the
four hot cells in the complex. The floor and walls, up to 8.2 m high, are
lined with stainless steel plate that is fully welded at the seams. The
HLV is located under the floor of the cask handling area and contains four
stainless steel tanks, three of which contain stored mixed waste
solutions. The LLV is also a sub-basement vault housing four stainless
steel tanks, one of which contains a residue of a mixed waste solution.
The B-cell waste is consists primarily of dispersible debris that
accumulated in the cell during many years of operation and during clean-up
activity initiated in 1988. The debris is made up of dust that has
settled from intake air. process fines, decomposition materials, small
tools, and pieces of equipment. The debris was designated as hazardous
from heavy metals through process knowledge. Additional waste materials
Operating Gattery
Basement Level
Second Floor
Figure 1 - Plan View of the 324 Building REC
246
-------
being stored in B-cell include dried melter feed material, elemental lead.
and liquid metal seal.
The HLV tanks contain approximately 4.000 liters of highly radioactive
solutions. Two tanks house 3.400 liters of a solution containing 137CsN03
and 90Sr(N03)2. The remaining 600 liters is a process feed-stock solution
that contains higher levels of 137Cs and90 Sr. These solutions are
considered hazardous due to both a corrosive characteristic and the
presence of heavy metals.
One of the LLV tanks at one time contained a radioactive-only waste
solution. That solution has evaporated, and the concentrations of
hazardous constituents rose above designation levels. At this time the
solution has evaporated completely, and a mixed waste residue remains.
Closure Plan Development Strategy
The primary objective of the Closure Plan is to describe the closure of
the unpermitted storage units in the 324 Building. A secondary objective
is to meet the milestone agreements in the TPA. which act as consent
orders. To meet the objectives, we used a simple strategy that integrated
the schedule required by the TPA and the regulatory requirements that
drive the development of the closure plan.
Our first Closure Plan milestone was to deliver a draft to Ecology by
December 29. 1995. To meet this objective a subcontractor was hired to
assemble the information and develop a draft closure plan. A draft was
delivered on December 22. 1995.
The draft Closure Plan acted as a basis for negotiation and discussion
with decision makers. In cases where no documentation has been gathered,
it is essential to compile and summarize data, facility information.
process history and any other information that may be pertinent to
decision making and sampling design. Participants received an information
package that included the draft Closure Plan, regulatory documents
applicable to the situation, and other documents pertinent to various
clean-up work underway in the facility. To facilitate, document, and
streamline these discussions and negotiations, we initiated a DQO Planning
Process, which will help guide the revision of the draft closure plan. On
completion of the DQO process, the Closure Plan will be revised, and the
247
-------
first revision will be finalized through workshop reviews with Ecology.
Central to our strategy is the DQO Planning Process. By facilitating and
documenting discussion among decision makers, we can obtain closure of the
more difficult issues related to the Closure Plan before open
forum/workshop discussions are initiated. A timely, mutually agreeable
framework for closure will save precious time and money in the completion
of the 324 REC Closure Plan.
THE DQO PLANNING PROCESS
Our DQO process started with the hiring of an experienced facilitator, Ms.
Mitzi Miller, President of EQM. The facilitator was a critical factor in
bringing credibility and independence to the process. An experienced
facilitator guided each meeting and prevented distraction by agendas or
tangents from participants. The facilitator also helped to focus
discussion on issues and persistently induced feedback from the group
until issues were resolved or a clear path for resolution was defined. In
the hands of an able facilitator our DQO process proceeded as planned with
few surprises.
Ground rules
Our DQO Process followed the following Ground rules as established by our
facilitator:
1. All decision makers must be at the scheduled meeting, or the meeting
will not proceed.
2. All attendees will have a specific purpose at the meeting or will be
asked to leave.
3. All scheduled speakers and presenters must be prepared for each
meeting.
4. Decisions must be based on logic and regulatory requirements.
5. Attendees may make multiple agreements regarding listening behavior
and discussion to avoid conflict.
248
-------
Roles and Responsibilities
Every attendees at our DQO meeting had one of three primary roles. The
most critical participants are the decision makers. Usually they
represent the regulatory agency and the facility/unit/site being
discussed. For our DQO, representatives from Ecology and the DOE acted
as the decision makers.
The second category of attendees are support staff. They support the
decision makers, and for the most part gather and present the information
at the DQO meetings.
The third category of attendees for our DQO were observers. They are
interested in learning about the DQO process and are invited to the
meetings to gain first hand DQO experience.
In addition the these three primary categories, subject matter experts.
were asked to attend specific meetings to provide detailed information
related to a complex issue or to provide information that which could not
be supplied by other participants. A court reporter attended to document
the discussion at each meeting.
Meeting Format
Meetings are the heart of the DQO process. Through presentation and
dialog, critical issues can be addressed and resolved. The DQO process
helps to streamline discussions and addresses the issues in a logical
format. The meetings are not open discussion forums. Instead the DQO
:meeting is a series of presentations and focused discussions used to
isolate issues and resolve them before moving forward. Preparation is
therefore the key to any DQO meeting. Presenters must understand their
subject and objectives and should format their information to be readily
assimilated; data should be tabulated, key points highlighted, and lengthy
text summarized.
To support several of our DQO meetings, dry runs were held, and
presentations critiqued. These dry runs are important if clear
communications is to occur at the DQO meeting.
In any decision-making meeting, issues may be skirted and decision
249
-------
postponed. The facilitator controls this problem, forcing the decision
makers to engage each issue and either resolve it or clear a path to
resolution. This policing by the facilitator prevents unnecessary
posturing by the participants so that the issues are addressed and
resolved.
Documentation of Decision
Two forms of documentation were used to support our DQO process. The
primary form is meeting minutes. Documentation provided by the court
reporter is the basis for a meeting summary that identifies each major
decision made and any assigned action items. The minutes are accepted by
the decision makers at the following meeting. Some decision makers prefer
to sign the agreements from each meeting: others prefer to sign the final
DQO Summary Document, which includes a list of agreements. In this case.
no formal signatures are applied to the minutes.
The second form of documentation is the DQO Document. This document,
which is prepared by the facilitator with assistance from subject matter
experts, outlines each decision made as a result of the DQO Process. This
document is reviewed and signed by each decision maker and is the official
documentation of a formal agreement by each party.
Results of the DQO Process
The DQO process helped to resolve many of the big issues related to the
closure of the 324 REC. The standardized DQO format provided ready
resolution of such issues as the boundary of an unpermitted unit, closure
standards for highly radioactive storage units, and applicable sampling
and analysis to determine if closure standards were met. Most
importantly, with each of these discussions thoroughly documented, and the
Closure Plan poised for revision, any moves to revisit or reopen these
closed issues can be met with a signed DQO agreement.
SUMMARY
To achieve those remarkable results from the DQO process required a
remarkable amount of work from all participants. Preparation is the key
for each DQO meeting, information must be formatted appropriately for easy
250
-------
assimilation, and all participants must review material assigned to them
(including previous meeting minutes). In addition to a highly
participative group, the presence of a skilled facilitator is critical to
completing the process successfully. A facilitator will make sure that
the logical format of the DQO is followed, will detect hidden agendas, and
will abort low-value discussions. With proper preparation, participants
who understand their roles and responsibilities, and a capable
facilitator, the DQO process can add significant value in addressing even
the most intractable environmental problem.
REFERENCES
B Becker-Khaleel and K Schlick. 1995. "324 Building REC and HLV Tanks
Closure Plan".Scientific Ecology Group. Richland. Washington.
US EPA. Guidance for the Data Quality Objectives Process. EPA QA/G-4.
September 1994.
251
-------
57
PROPER SELECTION OF PERFORMANCE EVALUATION SAMPLES
FOR CHEMICAL DATA ACQUISITION
BASED ON VENDOR ACCEPTANCE LIMIT CRITERIA
Garabet H. Kassakhian. Ph.D.. Quality Assurance Director, Tetra Tech, Inc., 670 N. Rosemead
Boulevard, Pasadena, California 91107-2190
ABSTRACT
Performance Evaluation (PE) samples are widely used by laboratories and their clients to verify the
accuracy and precision of the analytical methods employed by the laboratory. Although the U.S.
Environmental Protection Agency (EPA) supplies some PE samples for its programs, the bulk of
PE samples is usually purchased from commercial vendors as either off-the-shelf or customized
samples. Interpretation of the results of the PE sample analyses is based on the acceptance limits
provided by the vendor. Different vendors offer significantly varying acceptance limits for the
same analytes to be analyzed by the same analytical methods. When, for a specific analyte, the
vendor acceptance limits exceed the suggested method recovery limits the usability of the analytical
result, i.e., the PE sample, is usually jeopardized. The paper discusses the advantages and
disadvantages of using single-blind versus double-blind samples, and the pitfalls in receiving,
repackaging and shipping of PE samples to the laboratory. It also discusses and compares the
acceptance limits of various vendors to analytical method limits. A method is proposed for the
proper selection of PE samples to satisfy the requirements of the data quality objectives of a
program and maximize the analytical information that can be derived from the interpretation of the
PE analytical result.
INTRODUCTION
The environmental laboratories are going through a period of economic and professional turmoil.
The ubiquitous requirement in government contracts for the mandatory use of small and
disadvantaged business spawned the proliferation of new laboratories that operated with poorly
trained personnel using antiquated or inadequately maintained equipment. As the focus of the
Federal and State environmental efforts changed from Remedial Investigations/Feasibility Studies
(RI/FS) to actual remediation, the number of environmental field samples drastically diminished.
This resulted in a price cutting scramble for the dwindling supply of analytical dollars. Mergers
and acquisitions removed respected and renowned laboratory names from the analytical stage,
replacing them with regional or international players, some lacking experience in US laboratory
management.
The surviving environmental laboratories are undergoing wrenching reductions in personnel, up to
and including total closure and liquidation of assets. The seasoned analytical technicians and
chemists who, due to their intimate knowledge of the quirks of their instruments, were able to coax
excellent results from them, are increasingly being replaced by temporary professionals. The latter
come into a project unfamiliar with the specific analytical instrument at hand, with arguably
ambiguous corporate loyalty and commitment to quality control (QC). Their lack of knowledge of
company memory makes their use in the middle of large and long term projects a most risky
proposition.
252
-------
Meanwhile the rostrum of debarred laboratories or ones currently under Federal or State
investigation keeps mounting. Data integrity has become an area of major QA concern, so much
so that the EPA recently issued guidelines for Good Automated Laboratory Practices (GALP).1
Unrealistic regulatory or client expectations of laboratory conformance and method compliance are
increasing the potential for fraudulent manipulation of data. The U.S. Air Force Center for
Environmental Excellence (AFCEE) recently issued its generic Quality Assurance Project Plan
(QAPP) for use in its upcoming programs.2 The QAPP requires the laboratory to perform data
validation on its deliverable, with the implicit understanding that rejected data will not be paid for,
and the laboratory may be required to pay for resampling and reanalysis. It also requires the
spiking of laboratory control samples (LCS), matrix spike/matrix spike duplicates (MS/MSD) with
all the analytes of the method. When the recoveries of the LCS spiked analytes are outside the
AFCEE specified acceptance limits, the QAPP requires the repreparation and reanalysis of all the
samples in the corresponding analytical batch. In the case of a method, such as SW 8270B
(Semivolatile organics) with 237 analytes (of which only 64 have method specified limits) the
chances of non-compliance are a statistical certainty. Such radical increases in the analytical effort
may lead to sharply higher prices or lower quality, or both. Laboratories are already switching to
cheaper "non-traceable" reference materials and standards rather than using fully documented
certified and "traceable" ones.
Thus, the evaluation and selection of analytical laboratories have become a hybrid of quality
assurance(QA) and risk management. Not only are we concerned whether the selected laboratory
can analyze the samples correctly, providing legally defensible data, but we also have to investigate
and evaluate its financial stability, as well as its viability to stay in business for the mandatory or
contractual five year period for archiving records.
EVALUATING AND MONITORING LABORATORY PERFORMANCE
Traditionally laboratory performance is evaluated and monitored through:
1. On-site evaluations.
2. Raw data and magnetic tape audits.
3. Proficiency in analyzing PE samples.
4. The potential client's past experience with the laboratory.
All four provide a past status report of the laboratory's performance. All have to be extrapolated
into predicting future performance.
As contracting officers of Federal and State programs keep insisting that expenses for laboratory
selection can not be charged to the respective program/project costs, more and more environmental
engineering/remediation firms are eliminating or greatly abridging the in-depth on-site laboratory
evaluation. It is being replaced by preaward visits and reviews of greatly expanded statements of
qualifications. The latter now include, but are not limited to, the following:
1. The PE results of the past two years' EPA administered Water Pollution (WP) and Water
Supply (WS) studies.
2. On-site evaluation reports by third parties and the laboratory's responses to them.
253
-------
3. Method Detection Limit (MDL) studies and their backup documentation.
4. Interlaboratory PE study results, such as those administered by Analytical Standards, Inc.
(ASI), Environmental Resource Associates (ERA) and others.
A cost effective way of evaluating and monitoring laboratory performance is through the use of:
two types of PE samples, namely,
1. Preaward samples* usually off-the-shelf single-blind commercial samples.
2. Post-award monitoring samples, preferably double-blind commercial samples, or program
specific ones, supplied and evaluated by the EPA, US Army Corps of Engineers (USAGE),
etc.
The pros and cons of using PE samples for laboratory monitoring have been extensively discussed
in recent literature.3'4> 5> 6- 7
SINGLE-BLIND versus DOUBLE-BLIND PE SAMPLES
Guidelines for administering and scoring PE samples are, as a rule, not found in project specific
QAPPs. The recent U.S. Navy "...Laboratory Quality Assurance Guide" discusses the
administration of single-blind samples only, while the AFCEE QAPP completely bypasses the
issue.2*8 Single-blind samples are those recognized by the laboratory as PE samples with unknown
content. Usually they are delivered in ampules to be diluted to a volume of one liter. Double-blind
PE samples are delivered already diluted and in identical sample containers as those from the field.
They are shipped in field coolers together with routine samples. Customized double-blind PE
samples contain only those analytes that have been identified at the site and their concentrations
have been adjusted to mimic field hits.
Each signed contract with an environmental laboratory should contain stipulation on the use and
scoring of PE samples, both for single and double-blind samples. Depending on the data quality
objectives (DQO), the contract should define what constitutes "non-acceptance7' and the corrective
actions that both parties will undertake in such a case. For example, when an off-the-shelf single
blind sample is analyzed, it is assumed that the laboratory knows it is being tested and is on its
"best behavior".7 Therefore misidentification or misquantification of analytes of concern should
be considered more detrimental than similar inaccuracies relating to other analytes which are
routinely spiked into the "off-the-shelf commercial samples.
David Kennedy, a 25 year veteran of the environmental laboratory industry, asserts that
"laboratories have been known to "cheat" on single blind PE samples by analyzing the contents
of an ampule directly rather than preparing a dilute aqueous sample and going through the
entire analytical process".1
Use of customized double-blind PE samples resolves the above issues, since they are
indistinguishable from routine field samples, and are already diluted to 1 liter volume, or whatever
is the appropriate volume. The containers should be identical to those of the field samples.
254
-------
When the laboratory misidentifies or misquantifies the contents of the PE samples, the supplier and
the field technicians will be the ones to answer for the integrity of the sample preparation and
shipping process. The following precautions should be taken in the preparation of the double-blind
PE samples:
1. The labeling and shipping of the empty sample bottles to the PE supplier should be witnessed
by at least one other person, preferably from the QA/QC organization of the shipper.
2. Photograph or videotape the process of opening the cooler (shipped by the PE sample supplier)
and relabeling of the PE sample containers (if necessary). Frozen samples, broken containers,
and mislabeled samples are infrequent but real-life occurrences.
3. Photograph or videotape the process of placing the PE samples in the cooler that contains the
regular field samples.
4. All documentation relating to the PE samples, including the "key" to the sample concentrations
should be kept by the project QA/QC officer under strict security until the results are reported
by the laboratory.
5. Customized double-blind PE samples should contain only those analytes of concern that the
laboratory may have already encountered in previous samples from the same site. The
concentrations should also be in the range of interest, but should never be below the practical
quantitation limit of the method.
SCHEDULING OF DOUBLE-BLIND PE SAMPLES
It is important to schedule the double-blind PE samples correctly. In cases when the laboratory has
been unable to demonstrate proficiency in the requested analytical method, one PE sample per field
batch is the norm. This is an expensive effort used only as a last resort.
It is more common to schedule an off-the-shelf commercial PE sample at the beginning of the field
effort. The drawback is that off-the-shelf samples contain more analvtes than would be expected
from a field sample. When a "picket fence" chromatogram starts plotting, it is immediately
obvious to the laboratory that a PE sample has arrived. In the case of metals, it will be the one
with major hits across the whole ICP list. It is still not possible to manufacture homogenous PE
samples using site soils. The commercially available solid matrix samples use Ottawa-sand type
free flowing media, which are immediately recognizable by the laboratory. Shipping one aqueous
PE sample in a batch of soil samples also tips off the laboratory.
For short term projects scheduling customized double-blind PE samples at the beginning and at
75% completion of the field effort will be adequate. Since the results will not be known for 10 to
21 business days, this schedule will provide only ex-post facto information - confirmation that the
"bracketed" field samples were analyzed properly. On longer term projects it would be advisable
to use a quarterly PE sample, or one at the beginning , 30%, 60% and 90% completion of the
project.
When the laboratory fails to analyze accurately a customized or an off-the-shelf PE sample, this
should be followed by a customized PE verification sample. Shipping a single-blind verification
sample for purposes of claiming adequate laboratory performance and "saving the data" verges on
255
-------
the manipulative and even the unethical. A raw data audit will yield far more legally defensible
information than the laboratory's "acceptable" analysis of the single-blind sample.
VENDOR ACCEPTANCE CRITERIA
When selecting a PE sample vendor, the vendor's acceptance criteria must be compared to the
method limits. ERA, AS I and Analytical Products Group Inc. (APG) supply PE samples with
varied acceptance limits. Since some vendors determine their acceptance and warning limits based
on the performance of the laboratories that have participated in the analysis, some criteria may be
much wider than recommended by the method. While this practice of using wider limits may be
statistically justified, and definitely laboratory friendly, it may not satisfy the DQO of the project.
Hence, staying within the method limits is a more credible alternative, and legally easier to defend.
Table I presents acceptance criteria for selected semivolatile organic parameters for off-the-shelf
PE samples from two commercial vendors, as compared to method QC acceptance criteria.9
Table I
Vendor Acceptance Criteria for Selected Semivolatile Organic Analytes
in Percent Recoveries Compared to Method SW 8270
Parameter
Acenaphthene
Butylbenzyl phthalate
2-Chlorophenol
Di-n-butyl phthalate
1 ,4-Dichlorobenzene
Diethyl phthalate
Hexachlorobenzene
Pentachlorophenol
Phenol
Acenaphthylene
Pyrene
Fluorene
Vendor A
60.1-139
12.2-188
43.9-156
31.2-169
58.0-142
29.7-170
27.7-172
11.2-189
7.73-192
Vendor B
38.1-117
43.1-112
31.1-126
10.2-121
46.3-114
44-126.6
49.2-122.9
Method
47-145
D-139.9
23-134
1-118
20-124
D-114
D-152
14-176
5-112
33-145
52-1 15
59-121
It is obvious that Vendor B has acceptance criteria closer to the method ones.
If a PE sample is used with vendor specified acceptance criteria significantly wider than the
method QC acceptance criteria then it is strongly recommended that two identical PE samples be
purchased. The second one should be analyzed by a referee laboratory.
SUMMARY
Customized double-blind PE samples are a cost-effective alternative for evaluating and monitoring
the performance of laboratories engaged in long-term projects. When selecting PE sample vendors,
the vendor acceptance criteria should be compared to the method recommended QC acceptance
criteria. The vendor's acceptance criteria should be very close or tighter than the method ones. In
256
-------
cases where PE samples with wider limits have to be administered, it is recommended that an
identical PE sample be analyzed by a referee laboratory.
Labeling, packaging and shipping of PE samples require strict QC documentation, since any "non-
acceptance" may lead to costly resampling and reanalysis.
REFERENCES
1. U.S. Environmental Protection Agency, "2185 - Good Automated Laboratory Practices -
Principles and Guidance to Regulations For Ensuring Data Integrity In Automated
Laboratory Operations with Implementation Guidance", 1995 Edition, p. 1-2, 10 August,
1995, USEPA, Research Triangle Park, North Carolina 27771
1. HQ Air Force Center for Environmental Excellence, "Quality Assurance Project Plan",
Section 7, Version 1.1, February 1996, AFCEE, Brooks Air Force Base, San Antonio, Texas.
j. Robertson, G.L., and K.S. Kumar, J.R.Donnelly, F.C. Garner, "Causes of Problems in
Analyzing PE Samples", Environmental Testing and Analysis, 4(1), pp. 44-49, 1995.
4. Coyner, T.V., "Performance Evaluation Review", Environmental Testing and Analysis, 3(5),
pp. 31-33, 1994.
j. Morton, S., and G.M. Marlette, M.C. Verwolf, R.R. Newberry, "Mixed-Analyte PE
Program", Environmental Testing and Analysis, 4(3), pp. 34-37 and 66, 1995.
6. Moore, J.M., and J.G. Pearson, "Quality Assurance Support for the Superfiind Contract
Laboratory Program" in "Quality Control in Remedial Site Investigation: Hazardous and
Industrial Solid Waste Testing", pp. 85-102, Fifth Volume, STP-925, 1986, American Society
for Testing and Materials, Philadelphia, Pennsylvania 19103
7. Kennedy, D.C., "A Look at Commercial Providers of Performance Evaluations",
Environmental Laboratory, 7(6), pp. 3-5,1996.
8. Naval Facilities Engineering Service Center, "Naval Installation Restoration Laboratory
Quality Assurance Guide", Section 2, pp. 1,4-6; Section 3, pp. 10-11, Interim Guidance
Document, February 1996, Naval Facilities Engineering Service Center, Port Hueneme,
California 93043-4328
9. U.S.EPA, "Test Methods for Evaluating Solid Waste, Physical/Chemical Methods, SW-846",
3rd Edition, Final Update II, September 1994, U.S. EPA, Washington,DC 20460
257
-------
58
PERFORMANCE-BASED METHODOLOGY FOR THE
TRANSURANIC WASTE CHARACTERIZATION PROGRAM
Michael P. Maskarinec, Senior Research Staff Member
Wayne H. Griest. Group Leader
John M. Keller, Group Leader
Chemical and Analytical Sciences Division
Oak Ridge National Laboratory
Oak Ridge, Tennessee. 37831-6120
ABSTRACT
The Transuranic Waste Characterization Program (TWCP) is charged with the disposal of
mixed waste at the Waste Isolation Pilot Plant (WIPP) in New Mexico. Requirements for
characterization of this waste have been developed based on regulatory requirements,
knowledge of the processes used to generate the waste, and performance criteria of the
disposal site. Based on these considerations, a sampling and analysis program has been
evolving. A Quality Assurance Program Plan has been developed which sets forth the
data quality objectives of the program, and the criteria to be used to assure that analytical
data generated by the program is useful. Performance criteria exist for all analytes in these
matrices, and guidance is taken from SW-846 in choosing methodology. This paper will
describe the development of analytical methodology which meets the QAPP objectives and
assures occupational safety as well.
Transuranic wastes contain primarily alpha emitters, and the work on these samples is
performed mainly in glove boxes. Analytes to be determined include purgeable organics, a
limited number of semi volatile organics, non-purgeable volatiles, and metals. Using
Method 8260 as guidance, an automated purge and trap device has been installed in a
glove box and used to perform the analysis of purgeables. A modification of Method
3550 using a closed system with ultrasonic and thermal energy input has been used for the
extraction of semivolatiles, with the determination by Method 8270. The non-purgeable
organics are extracted with water and analyzed using a gas chromatograph located in a
radiochemical hood. Metals are digested using a microwave system in a glove box, and
can be analyzed using AA or ICP. The development of these methods was constrained
primarily by the performance-based criteria found in the QAPP, as well as by the need to
limit exposure to radioactivity.
Because these samples provide unique analytical challenges, they represent a strong
example of the advantages of performance-based analysis. Modifications to prescriptive
methods are possible as long as the QAPP objectives are satisfied. These objectives
258
-------
include for example: control sample recovery criteria, blank criteria, calibration criteria,
surrogate and matrix spike recovery criteria, holding time criteria, and others. In fact, all
aspects of the analytical methods are controlled. Still, there is no proscription for the use
of methods. As a result, high quality analytical data are generated using relatively flexible
methodology. While the advantages of this performance-based approach are clear, the
downside is in the degree to which documentation must be established. The actual data
package takes more time to produce than do the results.
INTRODUCTION
Beginning in the early 1970's, the promulgation of environmental protection legislation led
to a proliferation of new analytical methods. Because of the fact that these legislative acts
tended to be directed at specific environmental matrices (e.g. Clean Air Act, Clean Water
Act), the methods which gradually became associated with each piece of legislation were
developed with that particular matrix in mind. While methods for the Clean Water Act,
the Safe Drinking Water Act, and the Resource Conservation and Recovery Act
(groundwater) differed little in terms of analytical technology, they did differ in the various
quality assurance measures required. With the advent of the Superfund program, the
analytical methods became imbedded in contracts (the Contract Laboratory Program) and
became, at least for this program, completely proscriptive. As more and more states took
control of their own environmental programs, the methods suggested for the other
legislative acts became equally proscriptive, because the states lacked the expertise to
evaluate analytical data generated using non-standard methods. This placed burdens on
the analytical laboratory community, which was forced to comply with a variety of slightly
different requirements depending on the legislation under which samples were collected.
These burdens became oppressive in a business which is highly competitive, and has led to
the reevaluation of the methods policies.
Recently, the realization that analytical costs were becoming a major issue in
environmental compliance has led to an active effort to reduce the proscriptive nature of
: the various methods and to substitute instead performance-based criteria which are
developed in advance and driven by the needs of the program. Data quality objectives are
elucidated early in the program development phase, and these objectives are supplemented
with criteria for the analyses which will allow the data generated to support the objectives.
Methods which were formerly proscriptive are used more as guidance, and the usefulness
of the resulting data is judged based on its conformance with the performance criteria
rather than on to-the-letter compliance with the written methods. The net result is
expected to be the generation of data of known quality at reduced cost and increased
efficiency.
The Transuranic Waste Characterization Program (TWCP) is charged with the disposal of
259
-------
transuranic mixed waste at the Waste Isolation Pilot Plant (WIPP) in New Mexico. Large
volumes of waste will be stored at WIPP, and there are, of course, waste acceptance
criteria. In order to assure that waste destined for WIPP meets these criteria, the program
has chosen a performance based system for the generation of the associated analytical
data. This sytem has been implemented in the form of a Quality Assurance Program Plan
(QAPP)1, which was developed using prior knowledge of the wastes, performance criteria
at the disposal site, and with the active participation of the relevant regulatory authorities.
Analytical methods, while tied loosely to SW-8462, are expected to meet the Data Quality
Objectives (DQOs) of the QAPP rather than the precise language of their promulgated
counterparts. DQOs include analyte lists, method detection limits, program required
quantitation limits, precision, accuracy, and holding times. The central issue in this
program is that these wastes contain radioactivity in the form of alpha-emitters, which
requires that analytical operations be performed under containment (glove boxes or
radiochemical hoods). Thus, a prescriptive approach to the methods would have a high
probability of failure. This paper reports on the implementation of the QAPP
requirements, from method development to validation, to ensure that the data generated
support the waste acceptance criteria.
METHOD DEVELOPMENT
An examination of the analyte lists (see Tables 1-4) indicates that three organic methods
are required. Many of the analytes are the traditional volatile organic class, which suggest
the use of methods 5030 (purge and trap) and 8260 (capillary GC/MS). A second group
includes water-miscible solvents, suggesting the adaptation of method 8015. Finally, a
limited number of semi volatile organics are listed. The analysis of these compounds
requires a sample preparation step, followed by method 8270 (capillary GC/MS).
Metals Analysis
The primary methods used for the TWCP include 6010A (Inductively Coupled Plasma-
Atomic Emission Spectrometry, or ICP-AES), 6020 (ICP-Mass Spectrometry,or ICP-
MS), 7471A (Hg-Cold Vapor Atomic Absorption, or CVAA), and the 7000 series for
Graphite Furnace Atomic Absorption (GFAA). All of the analytical techniques required
the containment of instruments in radiochemical hoods to ensure contamination control
and worker safety. Many of the analytical problems observed with the TWCP work are
typical of samples with high dissolved solids which can result in poor precision for both
ICP-AES and ICP-MS. Frequently the TWCP sludges contain high levels of uranium
which causes severe spectral interference with other metals by ICP-AES. At higher levels
the uranium must be separated from the matrix prior to analysis to ensure satisfactory
data. High levels of nitrate salts can quickly cause deteroriation of graphite furnaces used
for GFAA measurements which limits the number of samples that can be analyzed in a
260
-------
batch. All of these "real world" problems tend to broaden the QC acceptance criteria and
degrade overall performance. Most of these matrix effects can be characterized to
determine the attainable data quality and performance checks can be developed to ensure
consistent data. If acceptance criteria can not be designed to handle complex matrix
problems, the analytical costs can get out-of-control trying to meet unreasonable QC
limits.
Volatile (purgeable) Organics
The main issue in the application of methods 5030 and 8260 for these samples was the
containment of the radioactivity. With the assistance of the manufacturer, a purge and
trap autosampler was constructed which allowed the physical separation of the carousel
and purge head from the trap and desorption system. The carousel was installed in a glove
box, with the transfer of the purge gas occurring via a heated line through the wall of the
box. Samples collected in VGA vials can be placed in the autosampler without sample
manipulation, and can be analyzed without ever being opened. The performance of this
methodology was confirmed, although it was clear that potential existed for contamination
of the purging head. Such contamination would make routine maintenance difficult.
However, an examination of the performance requirements indicated that the method
detection limits could be met using a methanol extraction, followed by purge and trap
GC/MS analysis of the methanol. This approach reduces the total radioactivity in the
glove box and in addition allows for reanalysis of the sample as needed. Thus, the method
chosen involves an extraction of a sample, preweighed (ca. 2 g) and shipped in a VGA
vial, with 2 mL of methanol. The methanol is removed from the sample in a glove box,
and a 50uL aliquot is added to water in the purge and trap device. Using this method, the
performance data in Table 2 were generated. Recoveries did not always meet the
established criteria, though the data could then be used to establish acceptable limits.
Performance of the method with respect to the established criteria was acceptable with the
exception of a consistently low response factor for bromoform. However, because of the
fact that all other performance criteria were met, this became insignificant. The QAPP can
be revised to reflect the lack of significance of this parameter. Furthermore, in selecting a
method for the dichlorobenzenes, it was shown that this method was capable of meeting
all performance criteria, and was superior to the semivolatile analysis. The analysis of
these compounds as purgeables rather than semivolatiles simplified the sample preparation
steps for the semivolatiles. These are examples of the advantage of performance based
criteria for data generation.
Non-pugeable Volatile Organics
As noted earlier, the method developed for these compounds was based on method 8015.
Again, preweighed samples shipped in VGA vials were extracted, in this case with 10 mL
261
-------
of water, the water was separated from the waste in a glove box, and placed in a GC
autosampler vial. The method of direct injection was used. A gas chromatograph
equipped with dual injectors, columns and dual flame ionization detectors was installed in
a radiochemical hood. Samples were then analyzed in a conventional manner. Table 3
shows the results of the performance data generated using this method. As this method
was developed and sample analysis began, it became apparent that some of the
performance criteria could not be met. One of the analytes which was originally targeted
was ethyl acetate. However, the high pH of the samples caused hydrolysis of the ethyl
acetate spikes, resulting in the formation of ethanol (not a target compound) and acetic
acid, which interfered with the chromatography. The ethyl acetate was subsequently
deleted from the target analyte fist. Ethyl ether, also a target analyte, showed consistently
poor performance. It was determined that the poor performance was due in part to
instability of the aqueous standards and in part to the low water solubility (low extraction
efficiency) of the compound. As a result, ethyl ether was deleted from this analysis and
added to the purgeable analyte list. Finally, pyridine, listed as a semivolatile analyte, was
found to meet all of the performance criteria when analyzed using this method. Again,
these three examples illustrate the improvement in data quality when using performance
based criteria rather than proscriptive methods.
Semivolatile Organics
The advantages offered by the use of performance based criteria for data generation are
most striking with respect to the semivolatile organics. First, by limiting the list of target
compounds to those with a probability of occurrence, the associated QC becomes
manageable. Second, by establishing method detection limits in advance, the required
sample size can be kept to a minimum - a necessity when dealing with radioactive samples
in containment. Finally, by establishing precision and accuracy requirements the sample
preparation techniques can be evaluated on performance rather than proscription. This
last issue becomes more important when considering the standard preparation techniques
for wastes. In this case the use of a Soxhlet extractor (method 3540 would incur
operational difficulties, including the generation of large amounts of waste from apparatus
cleaning, the generation of heat in the contained area, and the general diffculty in
assembling and disassembling the apparatus in a glove box. The use of a sonicator
(method 3550), while minimizing operational difficulties, would result in the spread of
contamination throughout the glove box, and would likely result in cross contamination of
the samples. Even the use of an automated soxhlet extractor (method 3541) had
operational problems. The availability of the performance criteria established in the QAPP
led to the development of the following preparation technique. Using the program
required detection limits, it was determined that 10 g of sample was required. The sample
was weighed into a standard soxhlet thimble, and mixed with sodium sulfate. The thimble
was placed in a wide mouth bottle of slightly larger dimension and 100 mL of methylene
262
-------
chloride was added. The bottle was tightly capped, and placed in an ultrasonic bath for 60
min at 23° C. The thimble was removed from the solvent and the solvent concentrated
using evaporative concentration. This technique minimizes sample handling, waste
generation, and cross-contamination. More importantly, all of the performance criteria
can be met. The resulting extract concentrate is then analyzed by method 8270.
Performance data are given in Table 4.
SUMMARY
Using method performance criteria based on established data quality objectives, methods
have been developed for the characterization of transuranic wastes. There has been an
evolving relationship between the performance based requirements of the TWCP Quality
Assurance Program Plan and the capabilities of the analytical technology. Where
unexpected analytes have been detected, new DQOs have been established. Where the
limitations of the analytical techniques have precluded meeting the DQOs, adjustments
have been made to the performance criteria. This type of evolution allows for the
incorporation of technological improvements in near real time, rather than waiting for new
methods to be approved. Extremely positive impacts on data quality are observed in such
a system, as compared to the use of prescriptive methods. Continuous review of the
DQOs in conjunction with improvements in analytical methods offers the best hope of
creating high quality data in an efficient manner.
One difficulty noted here is in the degree to which data need to be documented. Because
of the nonproscriptive methods, extra documentation may be required. However, as the
use of performance based criteria increases, it is likely that simplified methods will evolve
for demonstrating compliance with the DQOs.
REFERENCES
I. Quality Assurance Program Plan for the Transuranic Waste Characterization Program.
Revision B, CAO-94-1010, Carlsbad Area Office, U.S. Department of Energy, 1994.
2. Test Methods for Evaluating Solid Waste. Physical/Chemical Methods. EPA 600-
SW/846. Third Edition.
263
-------
Table I.
Performance data for metals analysis
Analyte
Antimony
Arsenic
Barium
Beryllium
Cadmium
Chromium
Lead
Mercury
Nickel
Selenium
Silver
Thallium
Vanadium
Zinc
PRQL* (QAPP)
(mg/Kg)
100
100
2000
100
20
100
100
4
100
20
100
100
100
100
IDL (Exp)
(mg/Kg)
n=7
5.0
5.0
2.4
2.6
20
5.5
5.0
1.0
38
5.0
22
5.0
6.9
49
Ace. (QAPP)
(%)
80-120
80-120
80-120
80-120
80-120
80-120
80-120
80-120
80-120
80-120
80-120
80-120
80-120
80-120
Ace. (Exp)
(%)
n=7
82-87
93-108
86-107
82 -102
78-104
75-102
90-116
100-111
81-106
94-103
84-101
95-108
84-106
81-104
PRQL assumes a 100X dilution of the sludge sample.
264
-------
Table 2
Performance data for purgeable organics analysis
Compound
Vinyl Chloride
Trichlorofluoro-nwlhane
Elhyl Ether
112-Trichloro-122-
trifluomelh.ine
Melhylene Chloride
Carbon Disulfide
Chloroform
1.2-[>ithlor»elhane
1.1.1 -Trichloroet hane
Carbon Telrachloride
Benzene
Trichloroelhene
1 . 1 .2-Trichloroetliane
Broinolbnn
Toluene
Telrachloroethene
Clilorobeiizene
Ellis 1 Henzene
imp-xylene
o-xylenc
1 . 1 .2.2-Tetracitloroclhane
1 .4-Dich!oruhcnzcne
1 .2-DicliU>n>lieiizene
MDL(QAPP)uK/g
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
2.0
1.0
1.0
1.0
1.0
MDL(EXP)'ug/g
0.5
0.80
0.3
0.2
0.3
0.3
0.3
0.3
0.3
0.1
0.3
0.2
0.5
0.7
0.2
0.3
0.2
0.1
0.5
0.2
0.3
0.3
0.1
Ace. (QAPP) %
D-251
17-181
80-120
60-150
D-221
60-150
51-138
49-155
52-162
70-140
37-151
71-157
80-120
45-169
47-150
64-148
37-160
37-162
60-150
60-150
46-157
18-190
18-190
Ace. (Exp)2 %
34-79
47-103
54-92
43-95
67-108
36-85
72-111
76-112
71-110
54-115
70-109
80-120
78-110
61-115
78-109
79-109
82-109
79-110
82-108
80-108
67-117
81-114
77-106
1 Calculated according to (1), n=4
1 Calculated according to (1), n=5
265
-------
Table 3
Performance data for non-purgeable volatiles analysis
Compound
Methanol
Acetone
Methyl Ethyl Ketone
i-Butanol
n-Butanol
Pyridine
MDL(QAPP)ug/g
10.0
10.0
10.0
10.0
10.0
50.0
MDL(EXP)'ug/g
6.5
4.4
2.4
3.5
3.8
6.6
Ace. (QAPP) %
50-150
50-150
50-150
50-150
50-150
50-150
Ace. (Exp)2 %
49-145
61-136
62-134
52-126
50-110
64-122
1 Calculated according lo (1), n=4
2 Calculated according to (1), n=5
Table 4.
Performance data for semivolatile organics analysis
Compound
2-Melhyl Phenol
F lexachloi oethane
4-Methyl Phenol
Nitrobenzene
2,4-Dinitrotoluene
MDL(QAPP)ug/g
5.0
5.0
5.0
5.0
0.3
MDL(EXP)'ug/g
0.26
0.23
.023
0.37
0.23
Ace. (QAPP) %
60-150
40-113
60-150
35-180
39-139
Ace. (Exp)2 %
46-104
38-93
46-1 14
49-98
54-146
1 Calculated according to (1), n=4
2 Calculated according to (1), n=5
266
-------
59
Streamlining the Data Quality Process Without Compromising Defensibility
Al Verstuyft, Consulting Scientist, Analytical Sciences Unit, Chevron Research and
Technology Company, P.O. Box 1627, Richmond, CA 94802-0627
Patricia V. Cline, Risk Management Technology Leader, CH2M HILL, P.O. Box 147009,
Gainesville, FL 32614-7009
ABSTRACT
The purpose of this presentation is to challenge the prescriptive approach to selection of
quality control samples and the level of data quality evaluation. The methods program
with performance based methods recognizes the need for streamlining the data quality
process. More cost effective sampling and analysis programs can be developed without
compromising legal defensibility through a critical review of each component of the data
quality process. Knowledge of site history, chemicals of potential interest, potential matrix
interferences, and other information and data obtained during initial stages of
investigations at large facilities can be used to focus the data collection efforts while
continuing to provide usable data of known quality. For, example, the number and
frequency of use of trip and field blanks should be adjusted in large projects. Similarly, the
level of QC samples may be adjusted to correlate with the levels of uncertainty that may be
acceptable in the decisions to be made. This results in greater documentation in sensitive
areas/samples (e.g. perimeter wells as compared to interior wells). The EPA's Contract
Laboratory Program criteria for selection of data quality evaluators at facilities that are not
under Superfund and have a more narrow range of issues results in unnecessary data
collection that to not improve our ability to make defensible risk management decisions.
Similarly, the requirement for 100 percent data validation for data to be used in risk
assessments also needs serious review. These procedures focus on analytical uncertainties,
which are frequently much lower than other factors that are impacting the decisions. A
tiered data evaluation/validation process that focuses on the decision process may include
review of selected subsets of the data. Improved computer based evaluation tools
(COMPARE by A. Sauter) are available to address the mechanics of process. Proposed
streamlined tiered evaluation process is more cost effective than complete validation and
does not compromise def ensibility.
267
-------
ORGANIC
•$•,*"/-
> «c **, > ,
t"f.. \
'"*{&*
-------
60
COMPARISON OF INERTNESS PROPERTIES OF TUBING IN GAS CHROMATOGRAPHY
Scott E. Adams. GC Product Development Chemist, Joseph W. Walsh, GC Product Development
Manager, William S. Cooke, GC Product Development Chemist, Alltech Associates Inc., 2701
Carolean Industrial Drive, State College, PA 16801.
ABSTRACT
Advancing technologies are lowering detection limits for various methods. Even with the
introduction of new detectors that are sensitive down to parts per trillion, sample integrity can be
compromised before it reaches the detector. Sample introduction and transportation can have
adverse effects on detection of low level analytes. This paper investigates the interaction of parts
per billion (by volume) levels of volatile organic gas standards with the inner pathways of various
tubing used in gas chromatography. AT-Steel™, stainless steel tubing with a fused silicalike
lining, demonstrated the least interaction with the gas-phase molecules studied.
INTRODUCTION
A trend in environmental analysis is to achieve lower detection limits. Advances in detector
design make it possible to achieve these lower limits. However, sample integrity can be
compromised before reaching the detector, therefore sample introduction and transportation
pathways benefit from treatment to minimize analyte losses.
This paper investigates the interaction of low level volatile organic gas standards with tubing
commonly used in gas chromatography. A switching valve and a flame ionization detector (FID)
were connected by the tubing being analyzed. The tubing was purged with nitrogen, and
switched in line with a low level volatile organic gas standard for a set period of time. The
response recorded from the FID for various low level volatile organic gas standards indicated the
amount of interaction between the tubing and the gas-phase molecules. Volatile organic gas
standards, in the parts-per-billion (ppb) concentration range, were used in this investigation.
EXPERIMENTAL
The volatile organic gas standards were made using a static dilution bulb. A volume was
transferred from the static dilution bulb to a SUMMA® canister. The SUMMA® canister was
pressurized with nitrogen to 29.4 psig to give a final concentration in the parts per billion (by
volume) range. The analytes chosen for the volatile organic gas standards were benzene,
hexachlorobutadiene, methyl tert-butyl ether, and 1,2-dichloropropane.
The types of tubing tested were AT-Steel, nickel, stainless steel, aluminum, and copper. All of
the tubing was 1/8" OD by 0.085" ID and had a length of 12'. The tubing was initially rinsed with
pentane, methylene chloride, and methanol. They were purged with dry nitrogen overnight.
The apparatus used to test the tubing was constructed from a switching valve and a flame
ionization detector (FID). A variation of this apparatus has been described elsewhere (1). A piece
of tubing was connected between the valve and the FID. The valve was plumbed with a supply of
pure nitrogen and the SUMMA® canister containing the tow level volatile organic gas standard.
The run was initiated with pure nitrogen and after one minute the valve was switched to the
volatile organic gas standard. The valve was left in this position for one minute allowing the gas
269
-------
standard to flow through the tubing. Two minutes after beginning the run, the valve was switched
back to the original position. The responses from the FID were recorded and plotted.
RESULTS AND DISCUSSION
The first analyte studied was benzene. A standard of 125ppb of benzene was made. Figure 1
shows the plot of the FID response for this analyte on each types of tubing (i.e. AT-Steel, nickel,
stainless steel, aluminum, and copper). By comparing peak shape, you can clearly see that the
stainless steel, nickel, and aluminum tubing each exhibit interaction with benzene. Surprisingly,
copper tubing exhibited very little interaction. Upon closer scrutiny, it was evident that AT-Steel
did perform better than copper. The plot for the copper tubing shows a small amount of rounding
on the top front of the peak and more tailing than AT-Steel.
The next analyte studied was methyl /erf-butyl ether (MTBE). A gas standard of 200ppb was
made for this analyte. Figure 2 shows the plot of the FID response for MTBE using AT-Steel,
nickel, stainless steel, aluminum, and copper tubing. Comparison of these peak shapes show
the adsorbance of MTBE on nickel, stainless steel, and aluminum tubing. Once again, copper
and AT-Steel show inertness to this analyte. Comparing peak shapes between AT-Steel and
copper is further evidence that AT-Steel tubing is more inert.
Another analyte chosen to investigate the activity of the inner surfaces of various tubing was
200ppb 1,2-dichloropropane. Rgure 3 shows the response from the FID for AT-Steel, nickel,
stainless steel, and copper tubing. Again adsorbance of this analyte occurred on the nickel and
stainless steel tubing. AT-Steel showed the least interaction with 1,2-dichloropropane, with
copper the next least active.
Finally, Figure 4 shows the FID response for 200ppb hexachlorobutadiene on AT-Steel, nickel,
and copper. As expected, AT-Steel showed the most inertness of the tubing tested.
Various levels of benzene standards (125,50,15, and 5ppb) were subjected to the same
analysis. Figure 5 shows the FID responses for these analytes. AT-Steel tubing provides the
same general peak shape for all analyte concentrations. Since, for the other volatile organic gas
standards studied, copper showed similar inertness, a 5ppb standard was ran on the copper
tubing. The peak shape for this low level analyte shows the superiority of AT-Steel over copper.
SUMMARY
By comparing peak shapes for various analytes using different types of tubing, we conclude that
AT-Steel provides the most inertness of the tubing studied (nickel, stainless steel, aluminum, and
copper). For a completely inert pathway, the ideal peak shape would be rectangular. It would
have sides orthogonal to the baseline and a top parallel to it. The only peaks that resemble this
shape were the peaks generated using the AT-Steel.
Surprisingly, copper exhibited remarkable inertness. Copper is not usually considered for
transfer lines, but may be acceptable as a low cost alternative when working with high levels of
analytes.
As seen by the results for nickel, it does not make the best tubing when it comes to inertness.
Analysts considering nickel as a transfer line should perhaps consider AT-Steel in its place.
270
-------
REFERENCES
[1] McAndrew, J.J.F.; Znamensky, D.; DeBord, D.; Inman, R. Anal. Chem. 1995, 67, 3075-
3078.
AT-Steel™
24 I • 10 12 14 11Mb.
Figure 1
I25ppb Benzene
Interact/on With Various
Tubing
Nickel
Stainless Steel
I 2 4
I 19 12 14 1IMIO.
24 6 I 10 12 14 16*11.
Aluminum
52168
XI
Copper
24 f I 10 12 14 HMIn.
0 24 < • 10 12 14 18MIO.
271
-------
AT-Steel™
Jl
0 2 4 I I 10 12 14 liMll.
Figure 2
lOOppb MTBE
Interaction With Various
Tubing
Nickel
Stainless Steel
0 24 I I 10 12 14 ICMta.
0 24 0 8 10 12 14 16Mta.
Aluminum
0 24 t I 10 12 14 16Mb!.
Copper
0 24 S 8 10 12 14 16 Mil.
272
-------
At-Steel™
2 4 8 I 10 12 . 14 1SMIB.
Figure 3
200ppb 1,2-
Dichloropropane
Interaction With Various
Tubing
Nickel
-_ -^
Stainless Steel
024 6 8 10 12 14 16Mb.
0 2 4 6 8 10 12 14 16Mln.
I2I8I
Copper
0 2 4 6 8 10 12 14 16 Mm.
273
-------
52182!
AT-Steel™
0 2 4 I • 18 12 14 11Mb.
Figure 4
200ppb
Hexachlorobutadiene
Interaction With Various
Tubing
Nickel
Copper
• 2 4 6 I IB 12 14 16 Mil.
I I I
0 24 6 8 10 12 14 16 Mil.
274
-------
125ppb Benzene
AT-Steel™
0 2 4 6 S 10 12 14 1SMII.
Figure 5
Various Concentrations
of Benzene
521691
SOppb Benzene
AT-Steel
I Sppb Benzene
AT-Steel
UJL
24 « 8 10 12 14 16Mln.
0 24 6 S 10 12 14 16 Mia.
Sppb Benzene
AT-Steel
(y-ixis expanded)
5ppb Benzene
Copper
(y-axis expanded)
u
0 2 4 6 I 10 « 14 ISMifl.
fl 2 4 S I 10 12 14 18 Mia.
275
-------
61
Title: EASIER AND FASTER GC/ECD ANALYSES OF PESTICIDES AND PCB'S
Authors: Susan M. Brillante, Phil Wiley and Tom Stark
Affiliation Hewlett-Packard Co.
2850 Centerville Road
Wilmington, DE 19808
Abstract: The generation of environmental data for pesticides and PCB's in various matrices can
be very a time-consuming and expensive process for laboratories and engineering firms.
In order to keep a GC/ECD system operating within control limits, analytical time will most likely
be spent on tasks, such as: recalibrations, reinjection of samples, cleaning detectors, reintegration
of chromatography peaks, as well as analyzing billable samples. In addition, the reporting
software may not be sufficiently optimized to generate reports in a timely manner.
We have been using an improved ECD on the Hewlett-Packard 6890 GC. This detector has been
modified to provide several new features: the linear dynamic range covers several orders of
magnitude, it stays cleaner longer, and it is more robust. All of this results in calibrations lasting
for quite a long period of time. Also, fewer samples need to be diluted and reinjected because they
were originally out of the linear range. The use of new Chemstation software results in better data
reduction and easier report generation. Improved injection techniques provide greater method
sensitivity and lower MDL's. Finally, the use of "fast GC" techniques results in a shorter analytical
run-time. Data will be presented to demonstrate these features
276
-------
62
COST AND QUALITY IMPROVEMENTS USING A MASS SPECTROMETER
AS THE DETECTOR FOR METHODS 8010, 8020, AND 8021
Richard Burrows. Principal Scientist, and Allen N. Quick, Jr., Scientist, Quanterra
Environmental Services, 4955 Yarrow Street, Arvada, Colorado 80002; Fred Feyerherm,
GC/MS Specialist, Analytical Instrumentation, Hewlett-Packard Company, 2000 West
Loop South, Houston, Texas 77072.
ABSTRACT
The analytes listed in SW-846 methods 8010, 8020, and 8021 can also be determined by
GC/MS, method 8260. GC/MS offers much better confidence in correct identification of
analytes than the GC methods and reduces the possibility of false positives. Customers of
environmental laboratories often prefer and specify the GC methods because of perceived
benefits in price, sensitivity, and quantitation accuracy. Problems with the GC methods
include: false positives due to the limited selectivity of the PID, false positives due to the
crowded chromatogram that results when extensive lists of target analytes are of concern,
poor stability of the ELCD, and quantitation difficulty due to co-elution.
We show that the quantitation limits and calibration criteria for methods 8010, 8020, and
8021 can be met using MS detection. General performance characteristics such as
accuracy and precision, linearity, analyte specificity, and sensitivity to water are
improved. The impact of improved chromatographic performance on productivity
measures such as tum-around-time, overall cost, and capacity is demonstrated.
INTRODUCTION
Volatile aromatic and halogenated compounds are of considerable environmental
significance, and several methods for their determination appear in SW-846. Two
approaches have been used for measurement of volatile organics in SW-846: the
GC/selective detector approach (methods 8010, 8020, and 8021) and the GC/MS
approach (methods 8240 and 8260). The GC/MS methods provide better confidence in
identification due to the combination of retention time and mass spectral data. The GC
methods rely on retention time alone and confirmation, normally on a second column, is
necessary for confidence in identification. Even with the second column, complex
matrices may lead to false positives. This is particularly true for the PID, which has
limited selectivity.
The number of analytes listed as within the scope of methods 8010, 8020, and 8021 has
typically increased with each revision of SW-846. The currently proposed method 802IB
lists 57 RCRA and 13 additional non-RCRA analytes (Table 1). Separation of this
number of analytes (plus the required surrogate and internal standards) presents severe
277
-------
difficulties. Even when resolution is obtained for a standard, matrix effects may lead to
retention time shifts in the sample, possibly leading to false positives or false negatives.
Data users have often preferred the GC methods over the GC/MS methods because of a
belief that lower detection limits are possible and because of more stringent calibration
criteria (Tables 2 and 3).
TABLE 1
Number of Analytes Listed in SW-846
Third Edition
Update I (July 1992)
Update II (September 1994)
Update III (Proposed)
8010
39
34
49
-
8020
7
7
8
-
8021
-
60
60
70
TABLE 2
Calibration Criteria
Initial calibration
Calibration verification
Bracketing standard
8021
< 20% RSD all analytes
+/- 15%, all analytes
Generally required
8260
< 30% RSD, subset of 6 analytes
+/- 20%, subset of 6 analytes
Generally not required
TABLE 3
SW-846 MDLs and PQLs
8010
8020
8021
8260
MDL (jig/L)
0.001 -0.34(l)
0.2-0.4
0.01 -3
0.02 - 0.5
PQLOig/L)
0.01-3.4(l)
2-4
0.1-30
0.2-5
(I)
It is generally recognized that these levels cannot be met in routine operation.
278
-------
EXPERIMENTAL
All work was performed on a Hewlett-Packard 6890 gas chromatograph with a single
split/splitless electronic pneumatic controlled (EPC) injector coupled to an HP 5972 mass
spectrometer. A Tekmar 3000 purge and trap and Tekmar 2016 autosampler were
interfaced to the GC by cutting the carrier gas line to the split/splitless injector
approximately 1-2 inches from the injector. The carrier supply line was then routed to the
desorb flow input on the back of the purge and trap. The heated transfer line from the
purge and trap was then connected to the previously cut line on the injector using a 1/16"
Swagelock union. This configuration allowed complete electronic control of the desorb
flows and pressures.
Needle spargers were used on the Tekmar 2016 autosampler with a purge flow of 40
mL/min. The purge time was set to 11 min. The water management system on the purge
and trap was bypassed. Desorb time was set to 2 min. and the desorb flow was controlled
by setting the split flow on the GC.
The split/splitless injector was operated in the split mode with a split ration of 10:1. A
capillary column was installed the normal way in the injector and coupled directly to the
mass spectrometer with no jet separator or splitter. The column used was a newly
available 60m X 0.32mm X 1.8 urn film HP-VOC-MS (HP part #19091R-316). Column
flow was set at a constant 1.5 mL/min using EPC. No cryogens were used for this work.
The mass spectrometer was tuned using the maximum sensitivity autotune and operated
in the full scan mode with a scan range of 35-300 amu. The complete purge and trap
GC/MS system was controlled by a Hewlett-Packard Chemstation computer networked to
an HP Chemserver. Teklink software was used to communicate with the purge and trap.
RESULTS AND DISCUSSION
Method Detection Limits and Calibration Stability
A method detection limit study was performed for the mass spectrometer at a spiking
level of 0.1 ug/L. These results are presented in Table 4. Reasonable spectra were
obtained at this level.
279
-------
TABLE 4
Compound
Dichlorodifluoromethane
Chloromethane
Vinyl chloride
Bromomethane
Chloroethane
Trichlorofluoromethane
Freon 1 13
1,1-Dichloroethene
Methylene chloride
trans- 1 ,2-Dichloroethene
1,1-Dichloroethane
c/.y- 1 ,2-Dichloroethene
Chloroform
1,1,1 -Trichloroethane
1 ,2-Dichloroethane
Carbon tetrachloride
Trichloroethene
1 ,2-Dichloropropane
Bromodichloromethane
2-Chloroethyl vinyl ether
cis-l ,3-Dichloropropene
trans-] ,3-Dichloropropene
1 , 1 ,2-Trichloroethane
Tetrachoroethene
Dibromochloromethane
1 ,2-Dibromoethane
Chlorobenzene
Bromoform
1 . 1 ,2,2-Tetrachloroethane
1 ,3-Dichlorobenzene
1 ,4-Dichlorobenzene
1 .2-Dichlorobenzene
Spiking Level
Hg/L
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
Average of 7
Determinations
Hg/L
0.09
0.12
0.08
0.13
0.08
0.10
0.10
0.10
0.36
0.10
0.11
0.10
0.11
0.10
0.13
0.09
0.09
0.10
0.10
0.28
0.09
0.08
0.09
0.08
0.07
0.09
0.09
0.05
0.09
0.09
0.09
0.08
MDL
Hg/L
0.02
0.04
0.03
0.03
0.02
0.03
0.02
0.02
0.06
0.03
0.02
0.03
0.02
0.04
0.12
0.02
0.02
0.01
0.03
0.55
0.01
0.02
0.04
0.02
0.02
0.02
0.01
0.04
0.03
0.02
0.02
0.02
280
-------
Calibration Stability
Calibration stability on the GC/MS system was excellent. In all cases, the method 8000
criteria for a maximum of 15% drift from the initial calibration for all components was
met. In addition, all samples were bracketed by a closing calibration that also met this
criteria.
Field Samples
Thirty-six field samples were analyzed by both GC/ELCD and GC/MS. Twenty-six
analytes were included on the target list, and the reporting limit was 0.5 ppb. Dual
column confirmation was used in the case of electrolytic conductivity detection in order
to improve the reliability of compound identification. Thirty-one analytes had positive
results, ranging from 0.41 ppb to 62,000 ppb. These detected analytes are listed in Table
5. Generally, results are quite comparable. In two cases, low levels of a target analyte
were identified by MS and not by ELCD. This is because the MS provided reasonable
spectra at a level that appeared to be baseline noise on the ELCD.
TABLE 5
47894-01
47894-02
47894-06
47894-07
47894-08
47894-09
47904-01
47904-02
47904-05
47904-06
47904-07
47904-08
48032-01
48032-02
48032-04
48032-05
48032-17
U-
Dichloroethene
ELCD
26
0.4
98
MS
28
0.2
113
Chloroform
ELCD
1.4
ND
630
360
1200
530
MS
2.5
14
750
420
1370
620
Trichloroethene
ELCD
37
140
ND
MS
56
192
33
1,1,2-
Trichloroethane
ELCD
1100
1.4
14
3600
18
750
1300
MS
1300
1.5
21
4100
22
690
1800
1,2-
Dichloroethane
ELCD
110
600
54000
34000
62000
13000
27000
18000
25000
17000
22000
MS
119
520
62000
39000
71000
15000
31000
20000
25000
17000
36000
281
-------
CONCLUSIONS
Modern benchtop GC/MS systems can provide the sensitivity of the ELCD in the
detection of halogenated volatiles. Use of the MS instead of the ELCD carries a number
of benefits for the laboratory and the data user, including:
• More confidence in compound identification and lower risk of false positives and
false negatives.
• Less difficulty in meeting stringent calibration criteria.
• Easier data review and validation.
• Faster analysis time.
• No co-elution problems.
It could be claimed that disadvantages of MS include greater initial cost of the
instrumentation and higher skill level needed by the operator. However, the greater initial
cost is compensated by the much greater potential sample throughput of the MS system;
and it is the authors' belief that production of high quality data with selective detectors
requires a higher skill level than for GC/MS due to the more difficult data interpretation.
282
-------
63
COMPARISON OF EXTRACTION METHODS FOR HEAVY PETROLEUM
HYDROCARBON MEASUREMENT IN SOILS
P. Calcavecchio. B.A. Kelley, A. Felix, G.C. Van Gaalen and E.N. Drake
Exxon Research and Engineering Company, Rt 22 East, Annandale, NJ 08801
ABSTRACT
This poster provides a summary of method development to optimize recovery from soxhlet,
supercritical fluid (SFE), accelerated solvent (ASE) and microwave extraction systems for
petroleum hydrocarbon measurements in refinery soil. A comparison of the three methods is
included along with extract molecular analyses by high pressure liquid chromatography
(HPLC) and gas chromatography/mass spectrometry (GC/MS).
The results show that under optimum conditions, both the SFE and ASE have equivalent
Total Petroleum Hydrocarbon (TPH) recoveries and reproducibility of results relative to the
soxhlet after a silica gel clean-up step. The SFE and ASE were shown to have a number of
advantages over the soxhlet including lower solvent requirements, easy set-up, automated
operations and compact designs. The ASE was found to be mechanically more reliable than
the SFE in the TPH application as well as better suited to accommodating larger sample
sizes.
The conclusions of the above study indicate that the ASE is a practical alternative to the
soxhlet for research applications requiring the monitoring of TPH. The ASE was also
shown to be flexible in its capability to use various solvents (including mixtures) and has the
potential for extraction applications in a wide range of environmental analyses.
INTRODUCTION
The use of soxhlet extraction methods for the quantitative removal of organic contaminants
from soils are effective and reproducible when matrix interferences from moisture content
and particle size distribution are controlled. There are incentives to finding alternative
methods due to inefficiencies associated with soxhlet use which include the manipulation of
glassware, the monitoring of solvent cycle rates and long extraction times (4-24 hours).
Other draw backs of the soxhlet include the requirements of large areas of vented hood
space, handling large volumes of solvent and solvent boiling point/azeotrope limitations.
There are a number of compact commercial alternatives designed to shorten the extraction
time and automate portions of the procedure based on SFE, ASE and microwave
technologies. These technologies were evaluated by extracting refinery soils containing
heavy aged petroleum contaminants. Hydrocarbon recoveries were compared to modified
soxhlet methods (based on EPA 3540 and 418.1) designed to minimize matrix interferences.
283
-------
SUMMARY
The evaluation of SFE and ASE showed nearly equivalent TPH recoveries relative to the
soxhlet (see Table 1) when extracting conditions were optimized. The SFE recovery was
optimized using a high density (0.99g/cc @ 25 C) initial extraction step followed by a
higher temperature (0.80g/cc @ 83 C) second step. The critical limiting variable found with
the SFE was the capacity of the solid analyte trap which could be exceeded when the
sample size was increased above 1 gram. The ASE provided the most flexibility in this
application allowing a larger sample size (10 grams soil) with TPH recovery equivalent to
the soxhlet over a range of 1.0 - 4.5 wt % TPH on dry weight of soil (see Figure 1).
The ASE also provided equivalent gravimetric recoveries to the soxhlet in a comparison of
a methylene chloride extraction based on EPA 3540 (see Table 2). This solvent provides a
higher recovery of 4+ ring aromatics and is used for quantitative GC/MS analyses for
polynuclear aromatic hydrocarbons (PAH's) and semi-quantitative characterization of
molecular classes by HPLC. The evaluation of a microwave system for this application
showed somewhat higher variability and required more solvent handling steps.
The methylene chloride extracts from soxhlet and ASE extractions were compared using
HPLC and the results showed comparable compositions of saturates and 1-4 ring aromatics
with a somewhat higher polars content in the ASE extract (see Figure 2). These extracts
were also analyzed by GC/MS after EPA 3611 alumina clean-up with initial results
indicating comparable recoveries of semivolatile PAH's (see Table 3).
CONCLUSIONS
The ASE provides equivalent recoveries relative to the soxhlet in a TPH application and
provides the following advantages: short extraction times, low solvent use, compact design,
large sample size and automated operations.
The ASE permits the substitution of Freon 113 with the less volatile tretrachloroethylene
(PCE) in the TPH application. The incentives for this change in solvent include regulatory
phase out of Freon and the much lower cost of PCE.
Evaluation of the ASE in a methylene chloride extraction for quantitative recovery of PAH's
shows high recoveries and low variability. These results were achieved using a 20 minute
ASE extraction compared to a standard 16 hour soxhlet extraction.
The SFE provides equivalent TPH recovery relative to the soxhlet under optimum
conditions but required more method development than the ASE. The SFE was also limited
by its smaller sample size capacity and mechanical servicing requirements.
The microwave results had higher variability than the soxhlet, SFE and ASE.
284
-------
Soxhlet and ASE Sample Preparation and Extraction for TPH
• Thoroughly mix the soil sample and sieve through a Tyler 4 mesh sieve.
• Weigh the soil into a solvent rinsed beaker and mix with a 3/1 ratio of
sodium sulfate to soil. Mix well and allow to dry for 1 hour.
• Break up agglomerates in sample with a pestle and quantitatively transfer
into the extraction thimble. Clean beaker and pestle with glass wool
and place on top of sample.
• Soxhlet extract with Freon 113 for 4 hours at a minimum 20 cycles/hour.
ASE extract with PCE for two 10 minute cycles at 150 C and 2000 psi.
• Filter the extract through solvent rinsed sodium sulfate into a volumetric
flask and mix well.
• Prepare standards from a reference oil containing 15 ml n-hexadecane,
15 ml isooctane and 10 ml chlorobenzene per EPA method 418.1 diluted in
the extracting solvent (soxhlet Freon 113, ASE tetrachloroethylene).
• Calibrate an infrared spectrometer with the standards using the absorbance
at 2930 wave numbers after subtracting the solvent background. Use a
minimum 5 point calibration to plot absorbance vs. mg TPH per 100 ml.
• Remove 10 ml of extract and treat with 0.5 g of 60-200 mesh silica gel.
Read the IR absorbance after 30 minutes and determine TPH.
References Washington State WTPH-418.1, EPA Methods 418.1, 9073, and 3540
-------
ro
00
O)
Alternative Extraction Methods for TPH
Method->
Solvent- >
Replicate Extractions
1
2
3
4
Soxhlet (4 hr.)
Freon 113
mg/kg
9800
10800
11000
9900
SFE (45 min.)
Carbon Dioxide
mg/kg
10300
11100
10400
10600
ASE (20 min.)
Tetrachloroethylene
mg/kg
10000
10200
10200
10500
Mean
RSD*
* relative standard deviation
10400
5.91 %
10600
3.36 %
10200
2.02 %
Table 1
-------
Alternative Extraction Methods for Molecular Characterization
N>
00
Method- >
Sol vent- >
Replicates
1
2
3
4
Soxhlet (24 hr.)
Methylene Chloride
mg/kg
17500
18000
16500
17500
ASE (20 min.)
Methylene Chloride
mg/kg
18000
18500
17900
18500
Microwave (15 min.)
Hexane/Acetone
mg/kg
15600
18700
18400
18500
Mean
RSD*
17400
4.85 %
18200
1.76 %
17800
8.27 %
Table 2
-------
GC/MS Quantitation of PAH's from Soxhlet and ASE Extracts
00
Soxhlet Replicates
1
2
3
Mean
RSD
ASE Replicates
1
2
3
Mean
RSD
Benzo(a)pyrene
(PPb)
8064
6635
7312
7337
9.74 %
8797
8232
8458
8496
3.35 %
Dibenz(a,h)anthracene
(PPb)
1917
1353
1466
1579
18.90 %
1692
1579
1466
1579
7.16%
Table 3
-------
TPH Comparison of Soxhlet and ASE Extractions
Solvents - Freon 113 and Tetrachloroethylene (PCE)
ASE TPH (wt%)
00
CO
(Freon)
(PCE)
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
Soxhlet Freon TPH (wt%)
Figure 1
-------
ro
CD
o
HPLC Molecular Classes from Soxhlet and ASE Extracts
wt % Detected
50
40
30
20
10
Saturates 1 Ring 2 Ring 3 Ring 4+Ring Polars
Molecular Class
Soxhlet
ASE
Figure 2
-------
64
OPEN-VESSEL MICROWAVE EXTRACTION
OF SOIL AND SEDIMENT SAMPLES
Kevin P. Kelly. Ph.D.; David L. Stalling, Ph.D., Leo W. Collins, Ph.D., OI Analytical,
Sample Preparation Products Group, 555 Vandiver Drive, Columbia, Missouri 65202
ABSTRACT
The application of microwave technology can make organic trace-level extractions and
digestions much more rapid and reproducible. Heating of and chemical reactions within
mixtures can proceed more rapidly because energy is input directly to the sample and
reagents using radiation rather than transferred through vessel walls and into the reagents
by convection heating. Additionally, rapid localized heating can cause temperatures to
be reached that are above the normal boiling points of reagents or solvents.
Higher temperatures are also reached when the microwave operation is conducted in a
closed vessel, due to increases above atmospheric pressure in the system. This also
contributes to faster kinetics for extraction and digestion. Higher pressures produced
during closed-vessel microwave operations can introduce procedural complications.
Open-vessel microwave assisted extraction (MAE) is explored here as an alternative to
closed-vessel procedures. In this work extraction vessels at atmospheric pressure were
positioned in a microwave system and samples were impinged by a focused microwave
beam conducted through a wave guide. Recovery and precision data are presented for
soil samples containing various target analytes to demonstrate applicability of the open-
vessel extraction process to trace environmental analyses.
INTRODUCTION
Removing target analytes by extraction from solid matrices is generally a matter of kinetic
rather than themwdynamic limitations. In nearly all extractions there is more than enough
solvent present to dissolve all trace level contaminants or even percent level impurities
(e.g. fats). The speed of reaction is limited, however, by how quickly target compounds
can be dislodged from active sites on or sequestered areas within the matrix material.
Microwave processing inputs energy directly to the target analytes (if they are polar) and
the matrix, as well as the solvent. Therefore it greatly accelerates kinetics of extraction.
Microwave heating has been applied to organic extractions primarily through the use of
closed-vessel systems in cavity ovens with radiation reflected about the compartment.
Pressures above atmospheric are achieved in closed vessels, so that the digestion or
extraction is very rapid, but a drawback is that sample size must be very limited or a very
expensive high pressure vessel is required.
291
-------
An alternative is microwave processing in vessels that are open to atmospheric pressure.
In this type of system solvent does not reach the temperatures produced in a closed-vessel
operation, yet extraction is still very rapid and requires only enough solvent to adequately
contact the sample. For example, crushed rock core from a petroleum drilling operation
was processed with automated Soxhlet extraction using 120 mL of methylene chloride for
a period of six hours (the same sample processed by traditional Soxhlet extraction
requires 24 hours or longer for extraction). Evaporating and weighing the resulting
residue produced an extractables content measurement of 0.273%. Another aliquot of the
same sample was processed with open-vessel MAE for 45 minutes (at a 30% power
setting in a 300 Watt system) using only 60 mL of methylene chloride. Measurement of
extractables content produced a result of 0.287%, demonstrating that the open-vessel
extraction yields results equivalent to exhaustive Soxhlet extraction.
DISCUSSION
Some forms of extraction can produce conditions which breakdown targets analytes during
the extraction process. Matrix materials may contribute to such side effects, by acting as
acidic or basic catalysts, for example. Higher temperatures accelerate extraction, but also
can accelerate such side effects. This may be the reason compounds such as benzidine,
a,a-dimethylphenethylamine, 2-picoline, dibenzo[a,j]acridine, and the organophosphorus
pesticides exhibited some low or poor recoveries during closed-vessel MAE in 1:1
acetone/hexane mixture1. Also, MAE of moistened soils showed decreasing recoveries
of some compounds when the soil samples were aged (seven or fourteen days).
Another example of differences between techniques for recoveries of certain compounds
is generally lower recoveries observed for endrin aldehyde and chlorothalonil during
sonication extraction (SW-846 Method 3550) when compared with results from automated
Soxhlet extraction (Method 3541) or closed-vessel MAE during acetone/hexane extraction
of 10 gram samples (sand, clay, or sediment)2. MAE (101%) equaled automated Soxhlet
(98%) for chlorothalonil at a high spiking level (100 ug = lOug/g) and both were superior
to sonication (69%). At a lower spiking level (500 ng = 50 ng/g) sonication recovered
no chlorothalonil from sand or clay, whereas automated Soxhlet and MAE produced equal
recoveries (69% each). Endrin aldehyde recoveries were similar for all techniques at both
high (17% MAE, 25% SOX, 22% SON) and low (26%, 32%, 29%) levels. This work
studies effectiveness of open-vessel MAE for recovery of analytes known to be easily lost
at low trace levels. Results were not available at the time the manuscript was prepared.
111 Beckert, W. F. et. al, "Microwave Assisted Extraction from Soil of Compounds Listed in SW-846 Methods
8250, 8081, and 8141 A", 1995, Eleventh Annual Waste Testing & Quality Assurance Symposium, Paper 34,
pp. 228-229. Washington, D.C.
|21 R. McMillin, et. a!., "Abbreviated Microwave Assisted Extraction of Pesticides and PCBs in Soil", 1996,
Pittsburgh Conference, Paper 477. Chicago, Illinois.
292
-------
65
BIAS FACTORS IN RECOVERY OF SPIKED
ORGANIC COMPOUNDS FROM AQUEOUS SAMPLES
Kevin P. Kelly, Ph.D.. David L. Stalling, Ph.D., OI Analytical,
Sample Preparation Products Group, 555 Vandiver Drive, Columbia, Missouri 65202; Rick
McMillin, Michael Daggett, US EPA Region 6 Laboratory,
10625 Fallstone Road, Houston, Texas 77099;
Laura Palmer, Rocky Mountain Arsenal, AM CPM-RML-A,
Building 130, Commerce City, CO 80022
ABSTRACT
In organic trace analysis, spiking experiments are system checks (i.e. can you recover what you
put in) but don't necessarily reveal whether a technique recovers incurred residue from a matrix.
Analysis of standard reference materials (known matrix that is analyzed multiple times using a
reference method) provides more accurate assessments of whether bound contaminants are
recovered. Nevertheless, judgement based on spike recoveries is widely used when comparing
techniques for analytical equivalency. This is particularly so for aqueous samples, since reference
materials are not available, have inadequate shelf life, or cannot be homogeneously subsampled.
This work describes occurrence and prevention of bias during separatory funnel (SF), continuous
liquid/liquid (CL), and electrically assisted extraction (EA). The degree of bias observed can be
affected by: 1) choice of extraction technique, 2) choice of solvents used for preparation of spiking
solutions, 3) method of introducing spiking solution to the sample, and 4) the atmospheric pressure
of the laboratory. For example1, 56 BNA analytes spiked into seven replicates of reagent water
at 10 ug/L and processed with EA extraction produced average recovery of 77% when the spiking
solutions were composed of methanol, versus an average of only 61% when they were composed
of methylene chloride (DCM). The corresponding figures for the same experiment but with CL
extraction were 72% for both cases. These results were assigned to a wall effect during EA
extraction, wherein spiked compounds were carried onto an upper surface of the glass sample
bottle in the presence of DCM only, rather than dispersed into the water sample in the presence
of methanol. In another bias example, a surface layer volatility effect was observed when two
laboratories performing EA extraction obtained similar results, but only after the one at higher
altitude (ca. 5000 ft. elevation) began injecting spiking solution under the surface of the water
instead of on top of the sample. Average recoveries for seven chlorinated compounds at the
higher altitude laboratory were 26%, versus 78% recovered at the lower altitude laboratory under
similar extraction conditions. Average recoveries for those analytes at the higher laboratory rose
to 73% after the change in spiking procedure.
INTRODUCTION
All liquid/liquid extraction techniques are based on some equilibrium process which partitions
extractable materials between aqueous phase (sample) and organic phase (extraction solvent). This
is especially true of batch-sequential extractions such as separatory funnel (SF) (e.g. SW-846
Method 3510). Continuous liquid/liquid extraction (CL) also depends on such equilibria, but for
293
-------
many small droplets of solvent introduced over a long time. This is still batch-sequential in a
sense, but what is non-equilibrium about CL is that recovered droplets are recycled by distilling
solvent and thus reintroducing fresh organic phase throughout the extraction.
For spiking recoveries to be an accurate measurement of extraction efficiency, it is necessary that
spiked compounds participate in whatever liquid/liquid equilibria are essential to that technique.
Procedural artifacts which prevent this from happening can invalidate a spiking experiment. At
levels typically used to evaluate extraction techniques most analytes are above their solubility in
water and prone to biases resulting from failure to participate fully in establishment of phase
equilibria. Some possibilities include (Type 1): wall effects - hydrophobic analytes adsorb to
portions of glassware and become more difficult to recover; (Type 2): surface layer volatility
effects - analytes having substantial vapor pressure at room temperature, when trapped in a layer
on a liquid surface, can evaporate more rapidly from mixtures of water and organic solvent than
they do from solutions in solvent alone; and (Type 3): non-equilibrium bias effects - which can
occur when water and solvent are both present at the beginning of an extraction, analytes are
offered an opportunity to pass directly into a solvent aliquot, and that aliquot is removed from
contact with the aqueous solution relatively quickly thus preventing the analyte from ever
establishing equilibrium concentrations in the two phases. The purpose of this study is to report
occurrences of bias, comment on possible mechanisms (and thus reveal ways to control it), and
demonstrate the potential for bias present in three techniques (SF, CL, and EA extractions).
DISCUSSION
US EPA methods generally call for spiking solutions made up in a water-miscible solvent such
as acetone or methanol2. These materials help relatively insoluble compounds to disperse into
aqueous sample. Theoretically, liquid/liquid equilibria are path independent. In other words,
whether compounds are dispersed into the aqueous phase when extraction begins is immaterial,
provided all compounds somehow reach equilibrium between the two phases. In SF extraction
it is easy to guarantee that condition, since all materials are present and exposed when the system
is thoroughly homogenized. For the other two techniques, this is not a sure bet.
Type 1 effect: Immediately after spiking solution is introduced each compound may be present
in more than one phase. Some may be dissolved or dispersed within the sample. There may also
be globules of spiking solution solvent plus analytes, and if a bulk layer of extraction solvent is
present some of a compound may migrate and dissolve in that as well. If there isn't opportunity
to disperse or dissolve in aqueous or solvent layers, spiked compounds may conglomerate and
form a separate layer, which may find a solid surface (e.g. glassware) upon which to deposit.
Consider how three extraction techniques accomplish mixing of aqueous and solvent phases. SF
extraction takes place in a closed vessel and all surfaces are wetted during shaking, therefore the
type 1 effects should not occur and equilibria should be path independent. This is not so for CL,
since spiked material not dispersed in the aqueous phase or fallen into the solvent pool below the
sample (or to the membrane, in the case of accelerated CL) may "climb walls" of the extractor
barrel at the top of the sample. Even so, a CL system is nearly closed, with a condenser at the
top, and through the long extraction time (at least 6 and possibly as long as 24 hours) and reflux
action analytes that were not dispersed in the beginning may still be recovered . In contrast.
294
-------
during EA extraction, the spiking compounds are introduced to sample residing in a glass bottle
(see diagram). Then the water sample (plus solvent aliquot) is conducted through a PTFE tube
(1/4 inch diameter) into a PTFE funnel where phase mixing takes place. Although the (now
solvent saturated) water sample returns to the sample bottle along with a fresh aliquot of
extraction solvent, extreme type I effects occurring in this system might not be completely
corrected during subsequent extractions.
Type 2 effect: In addition to adsorption in inaccessible spots, another mechanism for loss of
analytes is evaporation. During these investigations extraordinary losses were noted for a few of
the more volatile compounds from the semivolatiles target list. The losses were more severe than
and out of proportion to losses of other analytes close in order of GC elution (and therefore
presumed to be similar in volatility). Typical compounds exhibiting such losses include di-, tri-,
andhexachlorobenzenes, hexachloroethane, hexachlorocyclopentadiene, hexachlorobutadiene. and
naphthalene. Some are solids with enough vapor pressure at room temperature to sublime.
The unexplained losses seemed only to affect samples processed using EA technique; however,
it eventually was shown that such losses could also occur in SF extraction and could be avoided
during EA extraction. The problem was traced to the way spiking solution was introduced to the
sample. When it was dropped on top of a sample ("top spiking") a film layer was observed atop
the water, and fragrant naphthalene was detected by smell. However, when the same spiking
solution was injected directly into the sample ("direct spiking") no film layer was observed and
no odor was detected. Any extraction method can be affected by this problem, but SF extraction
is usually "forgiving", since materials are contained in a vessel with a narrow neck that is often
sealed shortly after spiking. An experiment involving top spiking and leaving water samples in
a jar in the hood for an extended time followed by SF extraction produced the same pattern of
losses that had been seen during EA extraction (see Table 3). Such losses can also occur during
CL extraction. Examination of the CL data in a published study3 reveals a pattern of recoveries
that is consistent with the type of volatility losses reported here.
Type 3 effect: So far we mentioned only negative factors; however, it's possible to have positive
bias during extraction. This occurs if a portion of spiked compound, instead of dispersing within
the sample, enters an extraction solvent aliquot that has no oppoitunity tc participate in phase
equilibria. In that case the percentage of the target compound contained in the recovered extract
is not representative of the ability of the technique to extract compounds from water samples.
This situation is most likely to occur during CL extraction, where there is generally a layer of
solvent at the bottom of the extractor to ensure proper operation without introducing aqueous fluid
into the boiling flask connected to the extractor.
EXPERIMENTAL
Facilities where these experiments were performed are identified as Lab 1 (US EPA Region 6
laboratory in Houston, Texas), Lab 2 (OI Analytical^ laboratory in Columbia. Missouri), or Lab
3 (Rocky Mountain Arsenal laboratory near Denver, Colorado). SF and CL extractions were
performed according to US EPA SW-846 methods 3510 or 3520, except some spiking solutions
were dissolved in DCM instead of the methanol specified in EPA methods. All samples were
dosed with 1 mL of each spiking solution used. For CL extraction the accelerated version was
295
-------
SOLVENT A
RINSE INPUT
EXTRACTION
CELL
BELLOWS
PUMP
HIGH
VOLTAGE
ELECTRODE
SAMPLE
INTERFACE
DETECTOR
VALVE
TO WASTE
CONTAINERS
VALVE
A
EXTRACT
RECEIVING
6OTTLE
AQUEOUS
SAMPLE
SOTTLE
Figure 1.
Schematic Diagram of ExCell Extraction Module
296
-------
employed, using the One-Step system (Corning), wherein a polymer membrane retains water
sample while allowing passage of solvent droplets. This modification eliminates the need for a
solvent "pad" below the sample and allows extraction to be completed in about 6 hours. EA
extractions were performed using the ExCell system (OI Analytical), which completes automated
extractions in about 2 hours.
Extracts from SF or EA were evaporated using the RapidVap N, (Labconco). Extracts from CL
were evaporated using the integrated Kudema-Danish glassware of the One-Step system. All
results reported from Laboratory 2 were obtained using GC/FID quantitation, while those from
Laboratories 1 and 3 were obtained using GC/MS. Results from three groups of experiments (A,
B, and C) are presented to demonstrate bias effects. Conditions for the sets of experiments are
summarized in Table 1.
Observance of type 1 effect: Group A experiments were conducted in Lab 1 using either CL or
EA extraction with a spiking solution containing fifty-six semivolatile analytes in either DCM or
methanol. Matrix was 1-liter samples of reagent water, wastewater, or groundwater. Sequential
extraction was performed at two pH values as per Method 3520. These results (Table 2) show
a possible wall effect which had no significant impact on CL extraction, but which seemed to
introduce negative bias during EA extraction. The results for EA using spiking solution in DCM
at the 10 ug spiking level averaged 58% for all 56 analytes over the three matrices tested, versus
69% for CL. When the spiking solution was changed to methanol the CL results changed hardly
at all, whereas the EA results jumped to an average recovery of 73%. At the higher spiking level
(300 ug) there was no significant difference in EA recoveries from reagent water upon changing
spiking solution solvent; however, the corresponding CL results dropped from 97% to 85%,
suggesting that when DCM was used as spiking solvent at the 300 ug spiking level the CL method
may have been subject to a positive bias from a type 3 effect.
Observance of type 2 effect: Group B experiments were performed in Laboratory 2 on acidified
reagent water samples spiked with 25 ug each of eleven semivolatile analytes in DCM. This
group shows how volatility losses were either produced or avoided using either EA or SF
extraction depending upon conditions. Results are presented in Table 3. Note that recoveries of
phenol and 2-fluorophenol, not susceptible to volatility loss, are determined by their water
solubility and change very little between the three experiments. This serves as a control measure
for validity of the extraction procedure. In contrast, recoveries of the compounds subject to
volatility losses vary drastically depending on the spiking procedure. Top spiking produced poor
results for both EA and SF (poor SF results were "forced" by top spiking one liter of water in a
1.25 liter jar and leaving the jar open in the hood for 80 minutes before the SF extraction). When
EA was conducted with direct spiking, recoveries for susceptible analytes (marked with a "*" in
the Table) rose from an average of 50% to 78% while recoveries of those not susceptible changed
hardly at all.
Group C experiments were conducted in Laboratory 3 using EA extraction only for reagent water
samples that were either top spiked or direct spiked with a methanol solution containing 50 ug/mL
of numerous semivolatile compounds. Results are presented in Table 4 for compounds prone to
type 2 effect and they show a dramatic improvement in recoveries for those analytes when top
spiking was replaced by direct spiking. It is believed that altitude of the laboratory (above 5000
feet) played a role in the severity of the losses associated with top spiking.
297
-------
SUMMARY & CONCLUSIONS
Three types of bias resulting from failure of the spiked analytes to completely participate in
liquid/liquid phase equilibria were discussed. Examples of two types of negative bias, wall effects
and volatility losses, were shown to occur by comparing: I) average recoveries when using spiking
solutions made from either methanol or DCM (Group A experiments); or 2) the average recoveries
from experiments (Groups B and Q where spiking solution was either placed on top of the sample
(top spiking), or injected directly into the sample (direct spiking). A third (positive) type of bias
results from shunting off solvent into which a portion of the spiked analytes has dissolved so that
they have no opportunity to reach equilibrium. This type of bias may occur during CL extraction,
as is suggested by experiments where at a level of 300 ng of each compound CL produced
significantly higher recoveries with spiking solution made in DCM than it did with a solution
made of methanol.
Overall, the results indicate that all three extraction types, SF, CL, and EA, can be subject to bias
(negative or positive) which derives from spiking technique or the type of solvent used to prepare
spiking solutions. Logically, bias effects are more severe at higher spiking levels because spiked
compounds become more difficult to disperse. SF, and less so CL, appear to be fairly "forgiving"
of variations in technique. Valid EA spiking results seem to be more strongly dependent on use
of correct technique.
Those laboratories which deviate from published methods by using spiking solutions prepared in
a non-miscible solvent may wish to reexamine their procedures. Direct spiking is recommended
over top spiking for superior recovery of spiked compounds which are subject to volatility losses.
Although direct spiking is not aesthetically pleasing, since the instrument for delivering spiking
solutions must be dipped into the sample, the use of a syringe with needle for adding the spiked
compounds makes it a fairly simple procedure to wipe the tip clean with a tissue and remove
matrix material before the next aliquot of spiking solution is drawn up.
REFERENCES:
[1| R. McMillin et.al; "Performance Evaluation and Method Equivalency Study for Automated Electrically
Assisted Liquid-Liquid Extraction"; PitCon '96. Paper 1269, Chicago, March 1996.
12] 'Test Methods for Evaluating Solid Waste". SW-846 3rd Ed., Update 2; 1995, US EPA, Washington. D.C.
[3] C. A. Valkenburg et.al; "Evaluation of Modifications to the Extraction Procedure Used in the Analysis of
Supcrfund Sites": J. AOAC1989. 72(4), pp. 602-608.
298
-------
Table 1. Conditions Used for the Cited Extraction Experiments
Experimental group
Laboratory
Extraction Techniques
Matrix
Spiking Solution
Spiking Technique
Spiking level
PH
number of replicates
A
1
EA or CL
RW, GW. WW
methanol or DCM
direct
lOpg, 300 Mg
base then acid
7, 3
B
2
EA or SF
RW
DCM
top or direct
25 Mg
acid
5 or 2 (SF)
C
3
EA
RW
methanol
top or direct
50 Mg
neutral
2
Table 2. Wall Effect (Group A): Comparison of Methanol and DCM Spiking at Lab 1
Average Recovery for 56 Semivolatile Analytes
Spiking Solution in ==>
Spike Level
300ug/L
lOug/L
Matrix
reagent water
groundwater
waste water
DCM
EA
16%
61%
55%
59%
CL
97%
72%
61%
75%
Methanol
EA
78%
77%
75%
67%
CL
85%
72%
63%
65%
299
-------
Table 3. Group B: Evaporative Losses in Lab 2 During EA or SF Extraction
Analyte
2-fluorophenol
phenol
1 ,4-dichlorobenzene(*)
1 ,2-dichIorobenzene(*)
hexachloroethane(*)
nitrobenzene
frw-(2-chloroethoxy)methane
naphlhalene(* )
hexachJorobenzene(*)
2-fluorobiphenyl(*)
acenaphthene
SFtop
60
46
59
67
45
90
94
75
49
71
79
EA top
58
43
44
51
27
94
93
71
29
75
86
EA direct
52
41
77
78
75
81
82
79
79
81
83
Table 4. Group C: Comparison of Top and Direct Spiking at Lab 3 Using EA Extraction
Analyte
1 ,3-Dichlorobenzcne
1 ,4-Dichlorobenzene
1 ,2-Dichlorobenzene
fci'j-(2-Chloroisopropyl) ether
Hexachloroe thane
2.3-Dibromochloropropane
1 ,2,4-Trichlorobenzene
Naphthalene
Hexachlorobutadiene
2-Methylnaphthalene
Hexachlorocyclopentadiene
2-Chloronaphthalene
Acenaphthylene
Acenaphthene
2-Fluorobiphenyl
Top Spiking
16
17
19
55
12
45
19
36
10
45
27
55
76
68
52
Direct Spiking
74
75
75
77
72
77
76
81
76
83
47
86
87
88
86
300
-------
66
PAH EXTRACTION FROM RIVER WATER USING NEW NOVO-CLEAN CIS
EXTRACTION MEMBRANES
Theodore R. West, Solid-Phase Extraction Chemist, Gary L. Nixon. Solid-Phase
Extraction Specialist, Alltech/Applied Science Labs, 2701 Carolean Industrial Drive, State
College, Pennsylvania 16801
ABSTRACT
Novo-Clean C18 PTFE extraction membranes were evaluated, and a procedure based on
EPA method 550.1 was developed for quantitating polycyclic aromatic hydrocarbons in
river water.
INTRODUCTION
Novo-Clean C18 PTFE (Teflon) extraction membranes were evaluated, and a procedure
based on EPA method 550.1 was developed for quantitating polycyclic aromatic
hydrocarbons (PAHs) in river water. These 47 mm membranes are twice as thick (54 mil,
about 1 mm) as the only other commercially available PTFE membrane and are 80%
bonded silica by mass. Because of their greater thickness and high CIS content, they have
more capacity than the other PTFE membrane. This was verified by measuring the
volume of a PAH solution that could be passed through the membranes before analyte
breakthrough.
The procedure for extracting PAHs from river water using these membranes was tested by
extracting spiked Susquehanna river water (20 ng/L) and spiked reagent water (2 and 20
Hg/L). The procedure has the advantage of not employing chlorinated solvents; only
acetonitrile, fcrf-butylmethylether (MTBE), methanol, and water are required. Analysis is
done with a gradient controlled HPLC pump, a Shandon hypersil PAH column, and a UV
detector.
EXTRACTION PROCEDURE
Equipment, Reagents, and Consumables
• Novo-Clean CIS extraction membrane, 47 mm
• Kontes 47 mm membrane manifold
• HPLC grade acetonitrile, te/7-butylmethylether (MTBE), methanol, and deionized water
• Alltech model 525 gradient HPLC pump with a UV detector and a Shandon hypersil
green PAH column
• PAH standard solution (200 u,g of 16 PAHs in 1 mL of methylene chloride diluted to
10 mL with acetonitrile)
301
-------
Sample Preparation: For laboratory spiked samples, the appropriate amount of PAH
standard solution was added to a 1 liter flask. To this, 5 mL of methanol were added and
the solution was swirled to ensure homogeneity. Then 1 liter of river or reagent water
was added to the flask. It was not necessary to filter sediment from river water samples.
Membrane Preparation: The membrane was placed in the filtration apparatus and 10 mL
of methanol were added to the solvent reservoir. After the methanol had soaked into the
membrane for 3 minutes, vacuum was applied to draw most of it through, leaving 5 mm of
solvent above the membrane. Then 10 mL of MTBE were added to the reservoir, allowed
to soak into the membrane for 3 minutes, and then drawn through by vacuum. The
membrane was dried thoroughly by applying vacuum.
Extraction: The membrane was conditioned by adding 15 mL of methanol to the solvent
reservoir and letting it soak into the membrane. Most of the solvent was drawn through
and 15 mL of deionized water were added to the reservoir. After the water soaked into
the membrane, most of it was drawn through. The water sample was applied and drawn
through the membrane by vacuum in 10 to 20 minutes. The membrane was dried by
vacuum for 3 minutes. At no time during the extraction procedure was the membrane
allowed to dry until all of the sample had been drawn through.
Elation: Ten mL of methanol were placed in the solvent reservoir and allowed to soak
into the membrane for 3 minutes. The solvent was drawn through the membrane at a rate
of <5 mL/min and the eluent was collected. Then 10 mL of MTBE were added to the
solvent reservoir and allowed soak into the membrane for 3 minutes. The solvent was
drawn through the membrane and the eluent was collected. The extracts were combined
and concentrated under an Nz flush to a total volume of 1 mL. The solution was
reconstituted to 10 mL using a 50:50 mixture of acetonitrile and deionized water.
Analysis: HPLC conditions for analysis of extract (chromatogram presented in Figure 1):
Column: Shandon Hypersil Green PAH 5u, 100 x 4.6 mm
Mobile phase gradient: A = CH3CN:H2O (1.99), B = CH3CN:H2O (99:1)
Time (min)
%B
0
50
5
50
25
100
27
50
30
50
Flow rate: 2.0 mL/min
Detector: UV254nm
Sample size: 100 ^L or less
DISCUSSION
Recoveriest for extractions from river water and reagent water are given in Tables I and
II. The recoveries of five sequential extractions were used to calculate the mean
recoveries from river water. The recoveries are within the limits specified as acceptable by
EPA method 550.1.
302
-------
Breakthrough of a solution of 0.5 ng/mL PAHs in 80:10:10 water:acetonitrile:methanol
was determined for a Novo-Clean CIS membrane and the only other commercially
available PTFE membrane. After 50 mL of solution was eluted through the other
manufacturer's conditioned membrane, the eluent contained naphthalene, phenanthrene,
and anthracene whereas the Novo-Clean membrane under identical conditions had no
PAHs in the eluent. This greater capacity may be attributed to the increased thickness of
the Novo-Clean membrane. HPLC chromatograms of each eluent are given in Figures 2
and 3.
SUMMARY
A procedure, based on EPA method 550.1, for extracting PAHs from river water using
Novo-Clean CIS extraction membranes has been developed. Recoveries of each of the 16
analytes are within the limits specified as acceptable by the EPA method. The Novo-Clean
membranes were shown to have a higher sample capacity over the only other commercially
available C18 PTFE membrane by measuring the volume of PAH solution required to
breakthrough the membranes.
Table I: Average Recoveries* of PAHs from Laboratory Spiked Susquehanna River Water
Peak Number PAH MeanRecoygry Standard Deviation
1 Naphthalene 75.7% 39.9%
2 Acenaphthylene 67.0% 10.5%
3 Acenaphthene 68.2% 10.4%
4 Fluorene 75.9% 6.8%
5 Phenanthrene 92.3% 3.3%
6 Anthracene 89.7% 1.3%
7 Fluoranthene 92.5% 4.9%
8 Pyrene 94.9% 5.4%
9 Benzo(a)anthracene 95.5% 4.1%
10 Chrysene 95.4% 3.8%
11 Benzo(b)fluoranthene 97.5% 6.3%
12 Benzo(k)fluoranthene 94.7% 3.8%
13 Benzo(a)pyrene 92.0% 8.4%
14 Dibenzo(ah)anthracene 93.6% 7.6%
15 Benzo(ghi)perylene 96.8% 3.9%
16 Indeno(123cd)pyrene 91.7% 2.8%
303
-------
Table II: Recoveries of PAHs from Spiked Reagent Water
Peak Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
PAH
Naphthalene
Acenaphthylene
Acenaphthene
Fluorene
Phenanthrene
Anthracene
Fluoranthene
Pyrene
Benzo(a)anthracene
Chrysene
Benzo(b)fluoranthene
Benzo(k)fluoranthene
Benzo(a)pyrene
Dibenzo(ah)anthracene
Benzo(ghi)peryiene
Indeno( 123cd)pyrene
20ue/L
92.2%
82.9%
84.6%
85.3%
83.6%
79.4%
83.4%
74.2%
82.3%
80.0%
75.1%
77.7%
60.8%
83.7%
82.2%
83.1%
2 ng/L
81.2%
62.4%
55.6%
79.4%
83.5%
85.1%
78.6%
69.9%
77.8%
79.1%
72.4%
78.5%
74.1%
64.9%
66.9%
69.0%
NOTES
* Average of five extractions, calculated using an external standard.
f (20 ug PAH/L « 100 ppb) River water was sampled from the West branch of the
Susquehanna River at Lock Haven, Pennsylvania, on 27 February 1996; acidified to
pH=4.5 with 1.0 M HCl<»q) on the following day; and stored at 4°C until used. Samples
were spiked immediately prior to extraction.
REFERENCE
J. W. Hodgeson, W. J. Bashe, T. V. Baker, "Determination of Polycyclic Aromatic
Hydrocarbons in Drinking Water by Liquid-Solid Extraction and HPLC with Coupled
Ultraviolet and Fluorescence Detection, Method 550.1." Environmental Monitoring
Systems Laboratory, U. S. Environmental Protection Agency, Cincinnati, Ohio, July 1990.
304
-------
Figure 1: HPLC Analysis of PAHs
Extracted from River Water
2 3
10
16
11
12
13
15
14
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Time (min)
305
-------
0
Figure 2: Breakthrough with Novo-Clean
mL
6 8 10 12 14 16
Time (min)
0
Figure 3: Other Membrane
46 8 10 12 14 16
Time (min)
306
-------
67
SCREENING FOR SILVEX BY A MAGNETIC PARTICLE ENZYME IMMUNO-
ASSAY IN TCLP EXTRACTS FROM SOIL AND WASTEWATER.
P.M. Rubio. T.S. Lawruk, A.M. Gueco and D.P. Herzog, Ohmicron Environmental
Diagnostics, Inc., 375 Pheasant Run, Newtown, Pennsylvania 18940
ABSTRACT
Use of immunoassays as field-screening methods to detect environmental contaminants
has increased dramatically in recent years. Acceptance of immunoassay methods by the
EPA has accelerated, enabling their use in applications where SW-846 methods are
required or desirable. Immunochemical assays are sensitive, reliable, cost-effective and
can be used for lab or field analysis. A commercially available magnetic particle-based
enzyme immunoassay (ELISA) kit has been adapted to screen the extracts of
environmental samples prepared according to the Toxicity Characteristics Leaching
Procedure (TCLP) at the 1.0 ppm regulatory level for 2-(2,4,5-Trichlorophenoxy)
propionic acid (also known as Silvex or 2,4,5-TP). The enzyme immunoassay was
originally designed to quantitate Silvex in water samples at concentrations around the
MCL (50 ppb) established by the the Safe Drinking Water Act (SWDA). The kit includes
calibrators at 5, 25 and 250 ppb. A simple one-step 1:40 dilution of a TCLP extract with
diluent is performed to bring the analytical range of the immunoassay within the regulatory
level of 1 ppm established by the EPA for Silvex TCLP leachate. Subsequent analysis by
the magnetic particle-based ELISA indicates an absence of matrix effects on various
environmental waste matrices. The method was found to be accurate, achieving 100%
agreement between positive and negative samples around the regulatory level of 1.0 ppm
on the samples tested. Comparison of 30 water samples using the Silvex ELISA versus an
USEPA GC method gave a correlation (r) of 0.995. The application of this ELISA
method in combination with the standard TCLP extraction permits cost-effective
evaluation of samples without the solvent disposal required by traditional methods.
Details of the studies to evaluate the results of combining the immunoassay with the TCLP
extraction method are presented.
INTRODUCTION
Silvex was used as a plant growth regulator for the control of woody plants on uncropped
lands; as a selective pre- and post-emergence herbicide used on crops, right-of-ways and
golf courses; and as an aquatic herbicide in lakes, streams, and irrigation canals. Silvex
has been banned in the U.S. since 1983 because of adverse effects to the liver and kidneys
of laboratory animals (USEPA, 1991).
Silvex is persistent in soil at levels which can contaminate drinking water by runoff into
surface water or by leaching into groundwater. It is regulated under the Safe Drinking
Water Act (SDWA) which set a Maximum Contaminant Level (MCL) of 50 ppb in
307
-------
drinking water and under the Resource Conservation and Recovery Act (RCRA) which
set an MCL of 1.0 ppm for Toxicity Characteristic Leaching Procedure (TCLP) (USEPA,
1992).
The Methods Section of the EPA Office of Solid Waste (OSW) has stated a need for more
rapid, less expensive, field screening procedures that do not compromise the accuracy of
the analysis. The availability of commercial immunoassays, which are reliable, sensitive,
and which can be utilize for inexpensively screening large sample loads of potentially
contaminated samples, coupled with the subsequent confirmation of positive samples by
established instrumental methods is an approach that addresses this need. The OSW has
accepted a number of immunoassays (Series 4000) for inclusion in its SW-846
compendium of analytical and test methods.
The principles of enzyme immunoassays have been described (Hammock and Mumma,
1980). Magnetic particle-based ELISAs have previously been applied to the detection of
environmental contaminants (Lawruk et al., 1996). The performance characteristics of a
2,4-D magnetic particle-based enzyme immunoassay on TCLP extracts have previously
been demonstrated (Hayes et al., 1993); subsequently, the OSW validated and accepted
the method as Method 4015 for use in hazardous waste analysis of water and soil matrices
as required by federal, state and local governments.
The TCLP was designed to determine the mobility of both organic and inorganic analytes
present in liquid, solid, and multiphasic wastes. If the analysis of any one of the liquid
fractions of the TCLP extract indicates that a regulated compound is present at a
concentration equal to or above the regulatory level for that compound, then the waste is
considered hazardous and should be disposed accordingly.
All wastewater and soils used to demonstrate the capability of the immunoassay for TCLP
analysis were subjected to the Method 1311 extraction procedure prior to analysis. The
Silvex RaPID Assay® kit, a magnetic particle-based immunoassay developed and
manufactured by Ohmicron Environmental Diagnostics, Inc., for the determination of part
per billion (ppb) Silvex in water and in soil, was used during this evaluation. This study
shows that the immunoassay can be readily adapted for use as a screening method for the
analysis of TCLP extracts for the presence of Silvex at the regulatory level of 1.0 ppm.
MATERIALS AND METHODS
Equipment and reagents: Silvex RaPID Assay kits, Silvex Sample Diluent, precision
pipettes, magnetic separator, RPA-I photometric analyzer and vortex mixer are available
from Ohmicron Environmental Diagnostics, Inc., Newtown, PA. Silvex for the
gravimetric preparation of spiking stock solutions was obtained from Chem Service
(Westchester, PA). All stock solutions were prepared in methanol
308
-------
Environmental Samples: Several water samples were studied for matrix effect. They
included municipal water from Philadelphia, PA and surface water ("runoff") collected
from Newtown, PA. Wastewater ("effluent") was collected from a wood treatment
facility in North Dakota. Two soil types were evaluated: a sandy soil and an organic soil
from New Jersey.
Procedures: TCLP extractions were performed on soils and wastewaters as described in
Method 1311. Silvex spikes were made volumetrically into the final extract after filtration.
The detailed TCLP extract screening procedure is published as an Application Procedure
(T00077) and available from Ohmicron Environmental Diagnostics, Inc. All immunoassay
results are converted to ppm Silvex in the waste extract for evaluation vs. the regulatory
level. Detailed directions on use of the immunoassay kit are given in the kit package
insert.
RESULTS AND DISCUSSION
Establishing a cutoff concentration for the screening procedure. To use the quantitative
result obtained from the immunoassay in a qualitative way, a cutoff that could reliably
discriminate the regulatory action level (1.0 ppm) from one half the concentration level
(0.5 ppm) was established. The TCLP extraction buffer (sodium acetate, pH 4.9) and
TCLP extracts of the sandy soil were diluted (1:40) with Silvex sample diluent and spiked
with Silvex at 0.5, 1.0 and 2.0 ppm. Ten replicates of each solution were then tested in
one immunoassay run. Figure 1 shows that the use of a cutoff concentration of 0.75 ppm
provides excellent discrimination between 1.0 and 0.5 ppm. This cutoff concentration falls
near the mid-point on the calibration curve where precision is optimal.
In a similar manner, the TCLP extract of a Silvex-free organic soil was spiked at 0.5, 1.0
and 2.0 ppm and tested at each level of interest in an immunoassay run. Figure 2
demonstrates a similar pattern of result distribution between the various spiked matrices:
buffer, TCLP extracts of sandy and organic soils; no overlap of the regulatory level could
be seen with half its concentration.
To test the utility of the method as a screening procedure for organic type of soils, ten
replicates each of a TCLP organic soil extract and buffer spiked at 0.50, 0.75, 1.0 and 2.0
ppb Silvex were run in an immunoassay run, when the definition of "positive" as greater
than 0.75 ppm and "negative" as less than 0.75 ppm are applied to this data (Figure 3), the
utility of the method as a screening procedure can be demonstrated.
Matrix specific performance of the screening method. A variety of water and soil matrix
types were studied to determine their interference, if any, with the immunoassay screening
method. All matrices were first treated as solid or liquid waste according to the
procedures given in Method 1311 (TCLP Procedure). Prior to spiking, each extract was
diluted, tested in the immunoassay and determined to contain a non-detectable
309
-------
concentration of Silvex. Aliquots of the extract were then spiked volumetrically with 0.5,
1.0 and 2.0 ppm Silvex, prior to the 1:40 dilution in Silvex Sample Diluent and subsequent
analysis by immunoassay.
Table 1 shows that the screening method accurately discriminates negative and positive
regulatory levels in all matrices tested. All results on matrices spiked at one-half the
regulatory level (0.50 ppm) were less than 0.54 ppm Silvex, showing no false positives
(results > 0.75 ppm). Results on matrices spiked at the 1.0 ppm regulatory level ranged
from a low of 0.76 to a high of 1.75 ppm with no false negatives (results < 0.75 ppm).
Samples spiked above the regulatory level at 2.0 ppm gave quantitative results from 1.96
to 2.72 ppm Silvex (no false negatives).
Correlation with a Reference Method. Results from a method comparison performed in
30 samples (7 groundwater, 9 municipal, 13 surface, TCLP extract) using GC by EPA
Method 515.1 (x) and the Silvex RaPID Assay (y) is illustrated in Figure 4. The methods
agreed well; the regression analysis yielded a correlation coefficient (r) of 0.995 and a
slope of 1.17.
SUMMARY
A magnetic particle based enzyme immunoassay for Silvex has been evaluated for use as a
screening method with TCLP extracts of some commonly encountered forms of liquid and
solid waste. Using a 0.75 ppm cutoff concentration, the method was found to be accurate,
achieving 100% agreement between positive and negative samples around the regulatory
level of 1.0 ppm. No interferences were exhibited by the method when TCLP extracts
from an organic soil, effluents and runoff water samples were tested. The immunoassay
compares favorably to GC Method 515.1. The Silvex RaPID Assay provides real-time
data for on-site decisions and permits timely initiation of corrective process control
measures in compliance situations. Fifty results can be available in less than one hour
compared to conventional methodology which requires several days to weeks to provide
results. The test is easy to use and can be performed in the laboratory or in the field by
non-laboratory personnel and technicians. Extraction, derivitization and cleanup are not
required, making the immunoassay safe to handle and easy to dispose of.
REFERENCES
Hammock, B.D.; Mumma, R.O. Potential of Immunochemical Technology for Pesticide
Analysis. Pesticide Identification at the Residue Level: Gould, R.F., Ed.; ACS Symposium
Series 136; American Chemical Society: Washington DC, 1980; pp 321-352.
Hayes, M.C.; Jourdan, S.W.; Lawruk, T.S.; Herzog, D.P. Screening of TCLP Extracts of
Soils and Wastewater for 2,4-D by immunoassay. Proceedings of the 9th Annual Waste
Testing and Quality Assurance, July 12-16, 1993, Arlington, VA.
310
-------
Lawruk, T.S.; Lachman, C.E.; Jourdan, S.W.; Flecker, J.R.; Hayes, M.C.; Herzog, D.P.,
Rubio, P.M. Quantitative Determination of PCBs in Soil and Water by a Magnetic
Particle-Based Immunoassay Environ. Sci Technol 1996, 30, 695-700.
USEPA, National Primary Drinking Water Regulations; Final Rule, Federal Register 40
CFR Parts 141, 142 and 143, Washington DC, January 30, 1991
USEPA, Method 1311: Toxicity Characteristic Leaching Procedure, 40 CFR Ch. 1, July
1, 1992.
311
-------
0)
D)
C
ro
tr
to
CD
cr
5 -
4 -
3 -
2 -
1 -
0
Negative
Positive
0.5 ppm spike
1.0 ppm spike
2.0 ppm spike
1
2
Silvex Concentration by Immunoassay (ppm)
Figure 1. Demonstration of the sensitivity of the selected cutoff for Silvex in TCLP
Buffer Matrix and Sandy Soil TCLP Extract. Silvex was spiked into TCLP extraction
buffer and sandy soil TCLP extract at concentrations of 0.5, 1.0 and 20 ppm. Each
spiked solution was then diluted 1:40 in Silvex sample diluent. Ten immunoassay
determinations were then made on each diluted solution. The frequency of occurence of
the results is plotted against the range of Silvex concentration measured.
312
-------
0)
O)
ro
a:
0)
a:
5 -
4 -
3 -
1 -
Negative Positive
0.5 ppm spike
1.0 ppm spike
2.0 ppm spike
0123
Silvex Concentration by Immunoassay (ppm)
Figure 2. Demonstration of the sensitivity of the selected cutoff for Silvex in TCLP
extract of an organic soil. Silvex was spiked into TCLP extract of an organic soil TCLP
extract at concentrations of 0.5, 1.0 and 2.0 ppm. Each spiked solution was then diluted
1:40 in Silvex sample diluent. Ten immunoassay determinations were then made on each
diluted solution. The frequency of occurence of the results is plotted against the range of
Silvex concentration measured.
313
-------
f
0.
Q.
in
r^
c>
2
•o
£
CO
Q.
-------
300
50
100 150 200 250
USEPA GC Method 515.1 (ppb)
300
Figure 4. Correlation of the Silvex RaPID Assay with method 515.1. Regression
analysis yields the following equation: y = 1.17x - 1.45 ppb, correlation coefficient (r) =
0.995, n = 30.
315
-------
Table 1. Silvex spiking results on TCLP extracts of environmental waste matrices.
Silvex was spiked into buffer, water or a TCLP extract of the matrices shown after the final filtration step. Each matrix was diluted 1:40
and tested by the immunoassay five times. Each result was then compared with the 0.75 ppm cutoff to determine if its status was positive
or negative.
O)
Ig Matrix/Spike
1 TCLP buffer
2 TCLP buffer+ 2.0 ppm
3 TCLP buffer + 1.0 ppm
4 TCLP buffer + 0.5 ppm
5 Sandy soil extract
6 Sandy extract+ 2.0 ppm
7 Sandy extract + 1.0 ppm
8 Sandy extract + 0.5 ppm
9 Organic soil extract
10 Organic extract + 2.0 ppm
11 Organic extract + t .0 ppm
12 Organic extract + 0.5 ppm
13 Effluent #1
14 Effluent #1+2.0 ppm
15 Effluent #1 + 1.0 ppm
16 Effluent #1+0.5 ppm
17 Effluent #2
18 Effluent #2 + 2.0 ppm
19 Effluent #2+1.0 ppm
20 Effluent #2+0.5 ppm
21 Runoff
22 Runoff+2.0 ppm
23 Runoff + 1.0 ppm
24 Runoff+0.5 ppm
25 Municipal water*
26 Municipal + 2.0 ppm
27 Municipal + 1.0 ppm
28 Municipal + 0.5 ppm
Silvex concentration by immunoassay (ppm)
Determination #
i
nd
2.21
1.16
0.51
nd
2.30
1.11
0.35
nd
2.26
1.07
0.34
nd
1.60
0.81
0.43
nd
2.11
1.10
0.53
nd
1.82
0.75
0.44
nd
2.17
0.85
0.50
2
nd
2.12
1.01
0.54
nd
1.96
0.98
0.40
nd
2.22
1.06
0.32
nd
1.43
0.89
0.47
nd
2.13
0.95
0.48
nd
1.76
0.81
0.41
nd
1.99
0.81
0.39
3
nd
2.23
1.05
0.47
nd
2.20
1.18
0.29
nd
2.58
1.16
0.38
nd
1.54
0.79
0.40
nd
2.08
0.86
0.48
nd
1.66
0.81
0.37
nd
1.83
0.79
0.47
4
nd
2.66
1.08
0.48
nd
2.72
1.49
0.34
nd
2.33
1.50
0.33
nd
1.55
0.78
0.46
nd
2.10
1.05
0.51
nd
1.60
0.86
0.49
nd
1.85
0.76
0.45
5
nd
2.37
1.07
0.52
nd
2.30
1.75
0.35
nd
2.48
1.23
0.27
nd
1.57
0.76
0.46
nd
2.24
1.06
0.54
nd
1.68
0.88
0.49
nd
1.85
0.90
0.44
Mean
.
2.32
1.07
0.50
.
2.30
1.30
0.35
m
2.37
1.20
0.33
.
1.54
0.81
0.44
.
2.13
LOO
0.51
.
1.70
0.82
0.44
.
1.94
0.82
0.45
%POS
0
100
100
0
0
100
100
0
0
100
100
0
0
100
100
0
0
100
100
0
0
100
100
0
0
100
100
0
%NEG
0
0
0
100
100
0
0
100
100
0
0
100
100
0
0
100
100
0
0
100
100
0
0
100
100
0
0
100
Not applicable to wastewater regulatory limit
-------
68
EVALUATION OF THE ACCELERATED SOLVENT EXTRACTION SYSTEM™ FOR
THE EXTRACTION OF ENVIRONMENTAL MATRIX REFERENCE MATERIALS
Michele M. Schantz and Stephen A. Wise, Analytical Chemistry Division,
National Institute of Standards and Technology, Gaithersburg, Maryland 20899
ABSTRACT
The Accelerated Solvent Extraction (ASE) System™ was evaluated for extraction of selected
polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyl (PCB) congeners, and
chlorinated pesticides from the following certified reference materials: Standard Reference
Material (SRM) 1649a (Urban Dust/Organics), SRM 1941a (Organics in Marine Sediment),
SRM 1944 (Highly Contaminated Marine Sediment), SRM 2974 (Organics in Freeze-dried
Mussel Tissue), and CARP-2 (Ground Whole Carp). The materials were extracted using
methylene chloride, acetonitrile, and a mixture of hexane/acetone (1:1, v:v). For the three
solvent systems evaluated, the ASE™ system showed comparable extraction efficiency to
Soxhlet extraction for selected PAHs, PCB congeners, and chlorinated pesticides in a variety
of matrices.
INTRODUCTION
Accelerated solvent extraction (ASE)™ is a new extraction procedure that uses elevated
temperature (50-200 °C) and pressure (500-3000 psi) for solvent extraction of solid matrices
[1], It has appeared as proposed EPA Method 3545 in Update III of EPA's SW-846 manual
[2,3]. Two of the method's advantages are: (1) it uses less solvent than traditional Soxhlet
and sonication extractions and (2) it is automated [4]. Richter and co-workers [1] investigated
the effect of experimental parameters on recovery of total petroleum hydrocarbons, PAHs, and
PCBs from a variety of samples.
This paper details an evaluation of ASE™ for the extraction of PAHs, PCBs, and chlorinated
pesticides from several certified reference materials: SRM 1649a (Urban Dust/Organics),
SRM 1941a (Organics in Marine Sediment), SRM 1944 (Highly Contaminated Marine
Sediment), SRM 2974 (Organics in Freeze-dried Mussel Tissue) and a Carp reference material
(Carp-2). The solvents used for extraction were methylene chloride, hexane/acetone (1:1,v:v),
and acetonitrile.
EXPERIMENTAL
Materials. SRM 2260 (Concentrated PAH in Toluene), SRM 2261 (Concentrated Chlorinated
Pesticides in Hexane), SRM 2262 (Concentrated PCB Congeners in Iso-octane), SRM 1941a
(Organics in Marine Sediment), SRM 1649a (Urban Dust/Organics), and SRM 2974 (Organics
in Freeze-dried Mussel Tissue) were obtained from the Standard Reference Materials
Program, NIST. Candidate SRMs 2269 (Perdeuterated PAH-I), 2270 (Perdeuterated PAH-II),
2275 (Supplemental Pesticides), and 1944 (Highly Contaminated Marine Sediment) were also
used. Perdeutered 4,4'-DDT was obtained from Cambridge Isotope Laboratories (Andover,
MA) while PCB 103 (2,2',4,5',6-pentachlorobiphenyl) and PCB 198 (2,2',3,3I,4,5,5',6-
317
-------
octachlorobiphenyl) were obtained from Ultra Scientific (New Kingston, RI). The carp
sample was obtained from the National Research Council of Canada. All solvents were
HPLC-grade.
Sample extraction. The conditions used for ASE™ are similar to those given in proposed
EPA Method 3545 [2] and used by Richter and coworkers [1]: 100 °C, 5 min equilibration, 5
min static time, 2000 psi. Since dead volume of the cells can be a problem, precleaned
sodium sulfate was used to fill the cell volume and to dry the wet samples. Two samples each
of SRM 1944 (2-3 g dry) and SRM 1941a (3-5 g dry) were extracted with methylene chloride
using the above conditions. Also extracted using methylene chloride were one sample of SRM
1649 (1.78 g dry), two samples of SRM 2974 (1-1.5 g dry), and two samples of the Carp-2
material (5-7 g wet). One sample of each of the above mentioned materials was also extracted
using acetonitrile and one sample of each using hexane/acetone (1:1, v:v). At least one of the
samples for each of the materials was extracted a second time using the same conditions.
None of the second extracts contained measurable quantities of the analytes of interest. The
internal standard solution (containing PCB 103, PCB 198, and perdeuterated 4,4'-DDT,
naphthalene, acenaphthene, biphenyl, phenanthrene, fluoranthene, pyrene, benz[a]anthracene,
benzo[0]pyrene, perylene, dibenz[a,/j]anthracene, and benzo[g/»]perylene) was added directly
to the cells prior to extraction. The extracts were concentrated to approximately 0.5 mL using
an automated evaporation system.
In the case of the sediment samples, copper powder that had been activated using hydrochloric
acid and then washed with methanol and methylene chloride was added to the extracts. The
sediment extracts and the urban dust extracts were then eluted through a silica solid phase
extraction cartridge that had been precleaned with the eluant, 15 mL of 10% methylene
chloride in hexane. This fraction was then concentrated using an automated evaporation
system to approximately 0.8 mL. This 0.8 mL was divided into two 0.4 mL portions, one
portion for the determination of the PCB congeners and pesticides and the other portion for
the determination of the PAHs.
In the case of the mussel tissue and carp extracts, size exclusion chromatography (SEC) on a
preparative-scale divinylbenzene-polystyrene column (10 fim particle size, 100 A pore size,
2.5 cm i.d. x 60 cm, PL-Gel, Polymer Labs, Inc., Amherst, MA) was used to remove the
majority of the lipid and biogenic materials. Using a mobile phase of 100% methylene
chloride at 9.9 mL/min for the SEC, the majority of the lipid and biogenic material elutes
immediately after the void volume of the column while the compounds of interest are retained
longer. The eluant (approximately 70 mL) was concentrated using an automated evaporation
system to approximately 0.8 mL for the mussel tissue samples, one 0.4 mL portion for the
determination of the PCB congeners and pesticides and the other 0.4 mL portion for the
determination of the PAHs, and to approximately 0.4 mL for the carp samples for the
determination of the PCB congeners and pesticides.
Determination of the chlorinated pesticides and PCB congeners. One 0.4 mL portion was
subjected to normal-phase liquid chromatography (LC) on a semi-preparative-scale
aminopropylsilane column to isolate two fractions containing (1) the PCB congeners and the
lower polarity chlorinated pesticides and (2) the more polar chlorinated pesticides. For the
normal-phase LC fractionation, n-hexane was used as the mobile phase for the isolation of the
318
-------
PCB and lower polarity pesticide fraction, and 5% methylene chloride in /i-hexane was used
for the isolation of the second fraction. The two fractions were analyzed using gas
chromatography (GC) with electron capture detection (ECD). GC-ECD was performed using
a 0.25 mm x 60 m fused silica capillary column containing a 5% phenyl-substituted
methylpolysiloxane phase (DB-5), 0.25 /*m film thickness.
Determination of the PAHs. The other 0.4 mL portion was concentrated and further
fractionated using LC on a semi-preparative-scale aminopropylsilane column for GC/MS
analysis. The mobile phase used for the normal-phase LC fractionation was 2% methylene
chloride in hexane. A similar GC column (DB-5 MS) as above was used for the GC-MS
analysis. The major ions monitored during different time periods were 128, 136, 142, 152,
154, 156, 164, 166, 170, 176, 178, 184, 188, 190, 192, 202, 212, 228, 240, 252, 264, 276,
278, 290, and 292.
RESULTS AND DISCUSSION
The concentrations for selected PAHs, PCB congeners, and chlorinated pesticides determined
using ASE™ with methylene chloride as the solvent are summarized in Table 1 for a variety
of reference materials and are compared to the certified concentrations where available or to
concentrations determined using Soxhlet extraction. The concentrations determined using the
ASE™ are comparable to those determined using Soxhlet extraction for the analytes of
interest. The sediment samples (SRM 1944 and SRM 1941a), the urban dust sample (SRM
1649), and the freeze-dried mussel tissue (SRM 2974) contain very little water (< 10% by
mass). The Carp-2 sample is a whole carp tissue sample and therefore contains approximately
85% (by mass) water [5]. The sodium sulfate that was added to the sample was sufficient to
handle this amount of water. Losses were not observed for the most volatile compounds of
interest, naphthalene and hexachlorobenzene. Similar extraction efficiencies were obtained
using hexane:acetone (1:1, v/v) and acetonitrile.
SUMMARY
ASE™ showed comparable extraction efficiency to Soxhlet extraction for the PAHs, PCB
congeners, and chlorinated pesticides in a variety of environmental matrices. The advantages
of ASE™ over Soxhlet extraction include speed of extraction and concentration of the extracts,
automation, and reduced solvent usage.
ACKNOWLEDGEMENT AND DISCLAIMER
We wish to thank Dionex for the loan of an Accelerated Solvent Extractor System™ for a
period of one month.
Certain commercial equipment, instruments, or materials are identified to specify adequately
the experimental procedure. Such identification does not imply recommendation or
endorsement by the National Institute of Standards and Technology, nor does it imply that the
materials or equipment identified are the best available for the purpose.
319
-------
REFERENCES
1. Richter, B.E.; Jones, B.A.; Ezzell, J.L.; Porter, N.L.; Avdalovic, N.; Pohl, C. Anal.
Chem. 1996, 68, 1033-1039.
2. Test Methods for Evaluating Solid Waste, Method 3545, USEPA SW-846, 3rd ed.,
Update III; U.S. GPO: Washington, DC, July 1995.
3. Lesnik, B.; Fordham, 0. Environ. Lab 1995, December/January, 25-33.
4. Majors, R. LC/GC 1996, 14, 88-96.
5. Certificate of Analysis for Certified Reference Material Carp-1. National Research
Council Canada, Ottawa, Ontario, Canada, 1995.
6. Certificate of Analysis for Standard Reference Material 1941a, Organics in Marine
Sediment. National Institute of Standards and Technology, Gaithersburg, MD, 1994.
320
-------
Table 1. Concentrations (pg/kg) of Selected PAH«, PCB Congeners, and Chlorinated Pesticides in Various Reference Materials extracted using ASE™ with Methylene Chloride
00
ro
Compound
PAHs
Naphthalene
Phenanthrene
Anthracene
Fluuranthene
Pyrene
Benzo[a]pyrene
Perylene
Benzolgto'lperylene
PCB Congeners
PCB 28
PCB 52
PCB 101/90
PCB 118
PCB 138/163/164
PCB 153
PCB 170/190
PCB 180
SRM 1944
Soxhlef ASE*
2152
4896
2570
8954
9721
4082
953
2749
75.8
78.9
73.3
57.6
59.7
73.5
23.2
41.7
(34)
(77)
(56)
(233)
(301)
(85)
(42)
(91)
(2.2)
(1.7)
(1.5)
(1.3)
(1.5)
(1.5)
(0.8)
(1.0)
2224
4906
2257
8953
978 1
4281
925
2818
77.8
80.8
72.5
62.1
62.9
76.9
24.1
44.5
SRM 1941a
Certificate' ASEb
1010 ±
489 ±
184 ±
981 ±
811 ±
628 ±
452 ±
525 ±
140
23
14
78
24
52
58
67
ND«
6.89 ±
11.0 ±
10.0 ±
13.38 ±
17.6 ±
3.00 ±
5.83 ±
0.56
1.6
1.1
0.97
1.9
0.46
0.58
1114
460
180
1005
849
639
393
528
ND
6.90
10.9
10.5
12.4
16.9
3.16
5.67
SRM
Certificate/Soxhlet*
4140
432
6450
5290
2509
646
3990
18.0
24.4
51.7
26.0
70.8
86.8
30.0
84.0
ND
± 370
± 82
± 180
± 250
± 87
± 75
± 890
(0.4)
(0.5)
(0.6)
(0.4)
(1.2)
(1.0)
(1.3)
(1.7)
1649a
ASE'
ND
4494
405
6674
5223
2375
642
3795
19.2
24.7
52.4
25.0
72.7
84.5
32.5
85.0
SRM 2974
Soxhlct'
9.7 ± 1.5
20.8 ± 1.6
6.49 ± 0.45
159 ± 10
146 + 9
16.5 ± 1.6
7.60 ± 0.59
20.6 ± 0.9
74.8 ± 5.2
118 ± 5.8
120 ± 3.3
115 ± 6.5
111 ±8.6
145 ± 9.8
6.38 ± 0.37
16.4 ± 0.82
ASE*
9.25
25.5
6.25
164
156
15.7
8.09
22.8
75.3
107
119
114
99.7
138
5.49
17.6
Carp-2
Soxhlet" ASE'
ND
ND
ND
ND
ND
ND
ND
ND
31.6 (1.4)
139 (6)
142 (8)
148 (7)
124 (3)
104 (5)
20.6 (0.6)
51.3 (0.9)
ND
ND
ND
ND
ND
ND
ND
ND
30.4
133
143
155
121
101
21.2
53.1
Chlorinated Pesticides
4,4'-DDE
4,4'-DDD
4,4'-DDT
cu-chlordane
ironj-nonachlor
hexaehlorobenzene
95.6
120
125
15.7
7.36
5.54
(1.9)
(8)
(6)
(0.5)
(0.25)
(0.45)
95.3
It8
121
15.3
7.41
5.68
6.59 ±
5.06 ±
ND
2.33 ±
1.26 ±
70 ±
0.56
0.58
0.56
0.13
25
6.26
4.65
ND
2.09
1.35
67.5
40.3
72.1
1120
10.1
8.88
ND
(0.7)
(1.3)
(54)
(0.8)
(0.7)
42.8
73.0
1187
9.89
8.56
ND
40.2 ± 1.8
44.9 ± 3.3
4.00 ± 0.31
15.5 ± 0.59
14.5 ± 1.1
ND
44.6
38.0
3.93
17.0
16.4
ND
147 (4)
97.1 (1.8)
4.31 (0.15)
7.56(0.21)
10.8 (0.7)
4.20 (0.08)
148
99.2
4.32
7.78
11.3
4.14
Three samples were extracted using melhylene chloride and analyzed in duplicate. Concentrations are the means, and the numbers in parentheses are one standard deviation of a
single measurement.
Two samples were extracted using melhylene chloride and analyzed in duplicate. Concentrations are the means.
The values are as reported in the Certificate of Analysis for SRM 1941a [6]. The uncertainty is based on a 95% confidence interval of the true concentration.
The PAH values are as reported in the Certificate of Analysis for SRM I649a with the uncertainty based on a 95% confidence interval. The chlorinated values are the means from
six samples Soxhlet extracted using methylene chloride and analyzed in duplicate. The numbers in parentheses are one standard deviation of a single measurement.
One sample extracted using melhylene chloride.
Eight samples were extracted using methylene chloride and analyzed in duplicate. Concentrations are the means, and the numbers in parentheses are one standard deviation of a
single measurement.
ND = not determined
-------
69
METHODS OF PREPARING SOIL SAMPLES FOR
HEADSPACE ANALYSIS OF VOLATILE ORGANIC COMPOUNDS:
EMPHASIS ON SALTING OUT
Alan D. Hewitt, Research Physical Scientist
U.S. Army Cold Regions Research and Engineering Laboratory
72 Lyme Road, Hanover, New Hampshire
ABSTRACT
Three equilibrium and two solvent extraction methods of preparing and analyzing volatile
organic compounds (VOCs) in soil by headspace gas chromatography (HS/GC) were com-
pared. The samples studied were triplicates of four different soil types spiked with an aque-
ous solution containing BTEX and four chlorinated compounds. Solvent extraction was
found to be superior for recovering spiked VOCs, followed by: direct heating; an aqueous
solution preserved with NaHSO4; and lastly, an aqueous solution saturated with NaCl and
acidified with phosphoric acid. The findings indicated that correction factors may be neces-
sary for equilibrium HS/GC determinations of VOCs in soils.
INTRODUCTION
This study evaluated several different methods by which discrete soil ("grab") samples can
be prepared for headspace gas chromatography (HS/GC) analysis of volatile organic com-
pounds (VOCs). In all, two solvent extraction methods and three equilibrium HS proce-
dures were assessed. Of particular interest among the HS equilibrium methods is one that
calls for soil samples be placed into an aqueous solution saturated with sodium chloride
(NaCl) and acidified phosphoric acid (H3PO4). This salting-out method of HS/GC analysis
was recommended in the initial draft of Method 5021. This method, Volatile Organic Com-
pounds in Soil and other Solid Matrices using Equilibrium Headspace Analysis, is cur-
rently scheduled for the third update of the Test Methods for Evaluating Solid Waste (1).
Indeed, salting out has long been recognized as a means of increasing the sensitivity of HS/
GC analysis of VOCs in aqueous samples (2), and acidification is an effective means of
limiting biodegradation of labile compounds in water (3), soil and soil-water slurries (4-5).
However, none of the experiments addressing the matrix effects of soils and soil-water
slurries on HS analysis (6-7), studied or recommended salting out.
To assess soil matrix effects and potential confounding effects due to salting out during
equilibrium HS/GC analysis, a comparative sample preparation study was performed. Re-
coveries of VOCs from four different soil types, as determined by HS/GC analysis, were
compared after the samples were heated, dispersed in water saturated with NaCl and acidi-
fied with H3PO4, dispersed in water acidified with sodium bisulfate (NaHSO4), extracted
322
-------
with methanol (MeOH), or extracted with tetraethylene glycol dimethyl ether (tetraglyme).
Merits and potential problems when using these various sample preparation methods, and
the trends established in physical characteristics of the soil matrices and organic compounds
tested, are discussed.
EXPERIMENTAL
a. Soil Subsample Preparation
A spiking solution was prepared by adding microliter volumes (3.1-5.8 p.L) of neat analytes
benzene (Ben), toluene (Tol), ethylbenzene (E-Ben), p-xylene (p-Xyl), o-xylene (o-Xyl),
trans-1,2-dichloroethylene (TDCE), cis-l,2-dichloroethylene (CDCE), trichloroethylene
(TCE) and tetrachloroethylene (PCE) to a 100-mL volumetric flask containing about 103
mLof groundwater. Once all the analytes were transferred, this aqueous solution was mixed
for 48 hours using a stirring bar. The target concentration for each analyte was 50 mg/L;
however, some volatilization losses occurred.
Table 1. Soil characteristics.
Ottawa sand > 99% sand, 0.035% organic carbon
Fort Edwards clay > 90% clay, 0.5% organic carbon
CRREL loam silly/sand, 1.5% organic carbon
Point Barrow. Alaska peat silty/clay, 7.1 % organic carbon
Laboratory samples were prepared from four soils that varied in texture (clay, silt, and sand)
and organic carbon content (Table 1). The method of laboratory subsample spiking and
handling has been presented elsewhere (5). Briefly, 15 air-dried 2-g subsamples of each soil
type were first transferred to 1-mL glass ampoules. Then, placing one ampoule at a time in
a metal tension clamp, they were spiked with the aqueous solution. After transferring a
0.200-mL aqueous spike using a 0.500-mL glass syringe (Hamilton), each ampoule was
quickly heat-sealed with a propane torch. All spike aliquots were taken from well below the
water-air interface, and the stainless steel needle was wiped prior to insertion into the
ampoule's neck. In addition to preparing the soil subsamples, 0.200-mL aliquots of the
spiking solution were placed into three separate auto sampler headspace vials (22 mL,
Tekmar) containing 10 mL of Type 1 water, and immediately capped with crimp-top caps
and Teflon-faced butyl rubber septa (Wheaton). These autosampler vials with 10 mL of
water were spiked at the beginning, middle and end of the soil subsample spiking process,
and served as a means of establishing the spike concentration and homogeneity. These three
autosampler vials containing spikes of the aqueous solution were analyzed within 24 hours
of preparation.
All of the 60 sealed ampoules containing treated soil were placed in a refrigerator for two
days to allow the analytes to sorb to the matrix. Then all of the ampoules were removed
from the refrigerator and triplicates of each soil type (3 X 4) were randomly taken for each
of the five sample preparation and analysis protocols. All of the soil subsamples were dis-
persed into the appropriate solutions at this time, then returned to the refrigerator for stor-
age until analysis. By using this procedure, VOC losses due to volatilization and/or biodeg-
radation were minimized.
323
-------
b. Preparation for Subsample Analysis
Heated equilibrium HS/GC analysis: Ampoules containing the spiked soil subsamples were
transferred to empty autosampler vials and capped. Once the vials were hermetically sealed,
the ampoules were broken and the soil completely dispersed by carefully hand-shaking the
autosampler vials. Sample analysis was performed the same day the ampoules were broken,
after heating the sample to 60°C for 50 minutes and removing a HS vapor sample. This
procedure is consistent with that described in the Draft Statement of Work for Quick Turn-
around Analysis (8).
Aqueous dispersion in a NaCl-saturated solution acidified with H3PO4, equilibrium HS
analysis: The aqueous dispersion solution was prepared by acidifying 500 mL of water with
H3?O4 to pH 2, then adding 180 g of NaCl. Ten mL of this aqueous solution was transferred
to 12 autosampler vials and an ampoule of spiked soil was added to each. Once the vials
were sealed, the ampoules were broken and their contents completely dispersed. These
samples were analyzed one day after the ampoules were broken, after heating the sample to
85°C for 60 minutes and removing an HS vapor sample. This analysis procedure is consis-
tent with that currently described in draft Method 5021.
Aqueous dispersion in a solution acidified with NaHSOj*. equilibrium HS analysis: Ten mL
of water and 0.25 g of NaHSO4 were placed into twelve autosampler vials, then an ampoule
was placed in each vial. Once the vials were sealed, the ampoules were broken and then-
contents completely dispersed. Samples were analyzed two days after the ampoules were
broken. HS vapors were removed after samples had been held for at least 20 minutes at
25°C.
MeOH extraction: Five mL of HPLC grade MeOH was transferred to twelve autosampler
vials and an ampoule was placed in each vial. After capping the vials, the ampoules were
broken, and the soil was completely dispersed by hand shaking. Five days after initiating
the extraction process the autosampler vials were opened and a 100-mL aliquot was re-
moved and transferred to another autosampler vial containing 10 mL of Type 1 water, which
was then capped. HS vapors were removed after the aqueous MeOH solutions had been
held for at least 20 minutes at 25°C.
Tetraglyme extraction: Samples were extracted with tetraglyme using the same procedure
as described for MeOH (5 mL/ampoule). Six days after initiating the extraction process the
auto sampler vials were opened and a 0.500-mL aliquot was transferred to another
autosampler vial containing 9.5 mL of Type 1 water saturated with NaCl. HS samples were
removed from the aqueous tetraglyme solution saturated with NaCl after they had been held
for at least 20 minutes at 25°C.
HS ANALYSIS
All samples were analyzed with an HS autosampler (Tekmar 7000) coupled to a GC (SRI
model 8610-0050) equipped with a 15-m DB1 0.53-mm capillary column and sequential
PID/FID detectors. Just prior to placing auto sampler vials in this system, each was shaken
for approximately 2 minutes. Appropriate settings for the autosampler were used for the
expected in-vial pressures obtained during the equilibration period. For each sample prepa-
324
-------
ration procedure, analyte amounts were established relative to HS standards prepared by
adding small (< 2 u.L) quantities of an MeOH stock solution to autosampler vials containing
the same solution composition and volume as the samples. However, since the 2 g of soil
and broken glass ampoules were present for the three equilibrium methods, these samples
contain three phases (air, water, and soil), as opposed to only one or two for the standards
(air, and air and water). In addition, the samples for the three equilibrium HS methods had
less air space (i.e., the glass ampoule and soil occupied approximately 2 cm3) than the
standards. Although not presented here, this decrease in vapor phase volume causes a sig-
nificantly (5-10%) different analyte response for samples in the heated HS and salting-out
method of HS analysis.
RESULTS AND DISCUSSION
Spike amounts and recovery from the different soil matrices, as achieved by the five differ-
ent subsample preparation and analysis protocols, appear in Table 2. The means and stan-
dard deviations of the analyte determinations (ug VOC) demonstrate that each sample prepa-
ration and analysis procedure was precise. A one-way analysis of variance test (ANOVA)
was performed at the 95% confidence level for each sample preparation and analysis method,
Table 2. Octanol-water partition coefficients of analytes, and means and standard de-
viations of triplicate determinations. For each analyte with the same method of sample
preparation and analysis, values with the same letter (or no letter at all) are not significantly
different from each other.
Compound (Abbreviation)
trans-1,2-dichloroethyIene (TDCE)
cis-l,2-dichloroethylene (CDCE)
benzene (Ben)
trichloroethylene (TCE)
tetrachloroethylene (PCE)
toluene (Tol)
o-xylene {o-Xyl)
ethylbenzene (E-Ben)
p-xylene (p-Xyl)
Octanol-water partition coefficient
2.09
2.13
2.53
2.60
2.65
2.95
3.13
3.18
Analyte Concentration (\ig)
Analyte
Spike
Ottawa
Ft. Edwards
CRREL Pt. Barrow, Alaska
L MeOH extraction
TDCE
CDCE
Ben
TCE
PCE
Tol
o-Xyl
E-Ben
P-Xyl
8.26 ±0.25
8.45 + 0.24
5.83 ±0.1 8
9.98 ±0.29
9. 10 ±0.28
6.54 + 0.20
6.71 ±0.20
6.23 + 0.26
6.22 + 0.14
8.23 + 0.21
8.15 + 0.41
5.60 ±0.21
9.58 + 0.46
8.66 ±0.35
6.58 + 0.28
7.03 ±0.57
6.08 + 0.27
6.27 + 0.20
8.57 ±0.10
8.70 ±0.1 6
5.93 ±0.11
9.75 ±0.14
9.24 ±0.1 6
6.69 ±0.1 8
7.20 ±0.08
6.50 ±0.16
6.50 + 0.16
8.23 ±0.46
8.42 ±0.44
5.75 ±0.30
9.64 ±0.40
8.95 ±0.45
6.64 ±0.43
7. 10 ±0.29
6.38 ±0.48
6.55 ± 0.65
8.33 ±0.1 7
8.53 ±0.28
5.86 ±0.20
9.82 ±0.33
8.94 ±0.46
6.56 ±0.1 8
6.92 ±0.1 9
6.29 ±0.40
6.34 ±0.31
325
-------
Analyte Concentration (]ig)
Analyte
Spike
Ottawa
Ft. Edwards
CRREL Pt. Barrow, Alaska
II. Tetraglyme extraction
TDCE
CDCE
Ben
TCE
PCE
Tol
o-Xyl
E-Ben
p-Xyl
8.26 ± 0.25a
8.45 ± 0.24a
5.83 + 0.18
9.98 ± 0.29a
9.10±0.28a
6.54 ± 0.20a
6.71 ± 0.20a
6.23 ± 0.26a
6.22±0.14a
8.30±0.12a
8.28 ± O.lOa.b
5.82 ±0.1 2
10.0 ± 0.26a
8.87±0.16a,b
6.42±0.14a,b
6.58±0.18a
5.87±0.14a
6.02±0.13a,b
7.76 ± 0.03b
7.93±0.16b
5.571006
9.42 ±0.1 Ob
8.42±0.36b
6.13±0.15b,c
6.32±0.42a
5.48±0,20b
5.67±0.25b,c
7.68 ± 0.32b
7.87±0.40b
5.55 ±0.17
9.56±0.40a,b
8.39 ±0.1 Ob
6.14±0.15b,c
6.32±0.04a
5.47 ± O.OSb
5.53±0.12c
8.02 + 0.28a,b
8.17±0.18a,b
5.62 + 0.19
9.38 + 0.35
7.64 ± 0.5 Ic
5.87±0.27c
5.03 + 0.35b
4.56 + 0,25c
4.64±0.28c
ID. Heated-headspace analysis
TDCE
CDCE
Ben
TCE
PCE
Tol
o-Xyl
E-Ben
p-Xyl
8.26 ± 0.25a,b
8.45 ± 0.24b
5.83±0.18a,b
9.98 ± 0.29a,b
9.10±0.28a,b
6.54 ± 0.20b
6.71±0.20b
6.23 ± 0.26b
6.22±0.14b
8.68 ± 0.22a
9.24±0.16a
6.10±0.14a
10.5±0.15a
9.74 ± 0.32a
7.00 ± 0.23a
7.38 ± 0.16a
6.93 ± 0.20a
6.80±0.16a
8.12±0.41b
8.56±0.38b
5.75±0.26b
9.72 ± 0.44b,c
8.86±0.46b,c
6.18±0.39b
5.98±0.28c
5.68±0.40c
5.70±0.16c
8.22±0.20a,b
8.30±0.27b
5.73±0.09b
9.28±0.42c
8.45 + 0.39c
6.15±0.05b
5.48 + 0.04d
5.58±0.18c
5.40 ±0.1 3d
8.09 + 0.20b
6.94 + 0.08C
4.47±0.06c
7.17±0.07d
5.38±0.29d
3.84±0.10c
2.24 + 0.13e
2.72±0.06d
2.57±0.14e
IV. Headspace analysis of soil-water slurry preserved with NaHSO4
TDCE 8.26±0.25a 8.35±0.09a
8.37±0.06a
5.83±0.02a
10.2±0.21a
9.10±0.07a
6.51±0.05a
6.62±0.16a
6.37±0.08a
6.00 + 0.16a
8.45±0.24a
5.83±0.!8a
9.98 + 0.29a
9.10 + 0.28a
6.54±0.20a
6.71±0.20a
6.23±0.26a
6.22±0.14a
5.56±0.37d
5.58 ±0.4 Id
3.95±0.25c
6.60±0.74c
5.71±0.34b
4.34 + 0.28c
4.37±0.20b
4.09±0.23b
3.87±0.26b
7.81±0.21b
7.83±0.17b
5.41±0.12b
8.71±0.21b
5.85±0.13b
5.46±0.17b
3.91±0.15c
3.99±0.18b
3.51+0.14c
6.54 ± 0.2 Ic
6.40±0.24c
4.03 ± 0.1 3c
5.21+0.10d
2.32 + 0.06c
2.95±0.08d
1.46±0.08d
1.53 + 0.05c
1.31±0.07d
CDCE
Ben
TCE
PCE
Tol
o-Xyl
E-Ben
P-Xyl
V. Headspace analysis of soil-water slurry saturated with NaCl and acidified with H3PO4
TDCE 8.26 + 0.25a 8.85 + 0.46a 6.74±0.69b 6.67 + 0.10b 4.58 + 0.09c
CDCE 8.45±0.24a 8.66±0.29a 6.45±0.75b 6.10 +0.2 Ib 3.42±0.04c
Ben 5.83±0.18a 6.33 + 0.31a 4.71+0.55b 4.24 + 0.17b 2.09±0.20c
TCE 9.98±0.29a 10.8±0.61a 7.37±0.95b 5.91±0.39c 2.51±O.I5d
PCE 9.10±0.28b 9.99 + 0.51a 5.39±0.34c 3.45 + 0.37d 1.24±0.13e
Tol 6.54±0.20b 7.16±0.28a 4.69±0.32c 3.20±0.22d 1.18±0.08e
o-Xyl 6.71±0.20a 6.93±0.32a 3.33±0.27b 1.60±0.10c 0.53±0.02d
E-Ben 6.23±0.26b 6.94±0.30a 3.40±0.18c 1.84±0.16d 0.62±0.02e
p-Xyl 6.22±0.14a 6.54±0.37a 3.07±0.29b 1.61+0.14c O.S4±0.03d
and for each analyte, to determine if there were any significant differences between the
theoretical (spiked) and measured concentrations for the various soil matrices tested. In
addition, a Fisher's Protected LSD was performed to determine which values were signifi-
cantly different from each other. Table 2 is arranged with analytes of increasing octanol/
water partition coefficient going down a column, and soil matrix with increasing organic
carbon content going across a row.
326
-------
Independent of sample preparation and analysis method, there was good recovery of all the
analytes from the Ottawa sand. However, MeOH extraction was the only sample prepara-
tion method that was able to achieve quantitative recoveries of all of the spiked analytes
from the other three soils. Tetraglyme showed good recoveries of TDCE, CDCE, and Ben;
however, recovery of the other analytes from the three other soils was significantly lower
than expected. This discrepancy between the expected and measured amounts often in-
creased with increasing percent organic carbon in the test soil, and with increasing octanol/
water partition coefficients (Table 2). To compensate for the extraction deficiencies of
tetraglyme, correction factors up to 1.34 would be necessary. The greater extraction effi-
ciency of MeOH in comparison to tetraglyme for VOCs in soils is consistent with a previ-
ous study (9).
In general, the pattern established for tetraglyme was followed by each of the equilibrium
HS sample preparation and analysis methods, except that the number and magnitude of
correction factors increased. For the heated HS method, correction factors as great as 3.00
would be necessary. Likewise, correction factors as great as 4.75 and 12.7 would be neces-
sary for the aqueous dispersion acidified with NaHSO4, and the aqueous dispersion satu-
rated with NaCl and acidified with H3PO4, respectively.
During equilibrium HS/GC analysis, the partitioning of VOCs into the vapor phase from
solution is more strongly increased by salt addition than by increases in temperature (2).
Usually, in the analysis of aqueous solutions by equilibrium HS/GC, both parameters are
used in concert for optimal conditions. Accordingly, the standards prepared for equilibrium
HS/GC analysis that used a saturated NaCl solution acidified with H3?O4, and were ana-
lyzed after heating to 85°C, had analyte responses some 2 to 5 times greater than the others
(all three procedures used the same detector and signal attenuation). However, this salting-
out approach to preparing soil samples for equilibrium HS analysis failed to achieve a simi-
lar enhanced analyte response. Instead, the matrix-analyte interactions cited previously in-
creased. A possible explanation for this phenomenon is that organic carbon, which serves as
a separate phase into which VOCs can partition, is more favorable than the vapor state
under salting-out conditions.
Method 5021 and Method 5035 (also under consideration for the third update of the SW-
846) both recommend in-vial methods to solve the volatilization and preservation issues
that plague VOC determinations in solid waste matrices. These two loss mechanisms have
been shown to cause up to 99.9% reduction in VOCs between collection and analysis (10-
11). Clearly, in-vial methods are necessary for obtaining site-representative VOC concen-
trations for vadose-zone samples. These two methods, along with the currently used Method
5030, recommend an aqueous dispersion/extraction method for low-level (< 1 ng/g) and
MeOH extraction for high-level (> 1 ug/g) VOC determinations in soils. The water-based
sample analysis procedure for Methods 5030 and 5035 is performed by dynamic purging,
while static headspace is used for Method 5021. Studies have shown that MeOH is a supe-
rior solvent, in comparison to water, for extracting VOCs from soils (6, 12), and that dy-
namic purging by mass transfer may be more efficient than the static equilibrium methods
described here (6).
327
-------
The findings of this study should make users of equilibrium HS/GC methods aware of
potential matrix-analyte interactions. On a site-per-site basis, these interactions can easily
be identified and corrected for by surrogate analyte and/or matrix spike recoveries. Often,
these interactions will be small, perhaps even insignificant, because vadose-zone soils rarely
have high (> 1%) organic carbon contents.
SUMMARY
Except in the case where there are no matrix effects (i.e., little or no organic carbon or clay
content), such as was shown for Ottawa sand in this study, correction factors are likely to be
necessary either when purging a soil-water slurry or performing an equilibrium HS analysis
of a soil or soil-water slurry for determining VOC concentrations in soils. Appropriate cor-
rection factors for soil matrices can be established by either surrogate analytes or matrix
spike recoveries. In general, soil matrix effects will increase with the analyte octanol/water
partition coefficient and the organic carbon content of the soil. Furthermore, because analyte-
organic carbon matrix effects increase with increasing solution electrolyte concentrations,
using salting out for equilibrium HS is not recommended.
ACKNOWLEDGMENTS
Funding for this work was provided by the U.S. Army Environmental Center, Martin H.
Stutz, Project Monitor. The author thanks Thomas Ranney and Jane Mason for critical re-
view of the text.
This publication reflects the view of the author and does not suggest or reflect policy, prac-
tices, programs, or doctrine of the U.S. Army or of the Government of the United States.
REFERENCES
1. U.S. Environmental Protection Agency (1986) Test Methods for Evaluating Solid Waste
Vol. IB. SW-846.
2. Friant L.F. and I.H. Suffet (1979) Interactive effects of temperature, salt concentration,
and pH on headspace analysis for isolating volatile organics in aqueous environmental
samples. Anal. Chem., 51: 2167-2172.
3. Maskarinec M.P., L.H. Johnson, S.K. Holladay, R.L. Moody, R.A. Jenkins (1990) Sta-
bility of volatile organic compounds in environmental water samples during transport
and storage. Environ. Sci. TechnoL, 24: 1665-1670. •
4. Hewitt A.D. (1995) Preservation of soil subsamples for the analysis of volatile organic
compounds. U.S. Army Cold Regions Research and Engineering Laboratory, Hanover,
New Hampshire, Special Report 95-5.
5. Hewitt A.D. (1995) Enhanced preservation of volatile organic compounds in soil with
sodium bisulfate. U.S. Army Cold Regions Research and Engineering Laboratory,
Hanover, New Hampshire, Special Report 95-26.
6. Hewitt A.D., P.H. Miyares, D.C. Leggett, T.F. Jenkins (1992) Comparison of analyti-
328
-------
cal methods for determination of volatile organic compounds. Envir. Sci. Technol.,26:
1932-1938.
7. Kolb B., C. Bichler, M. Auer and T.C. Voice (1994) Simultaneous determination of
volatile aromatic and halogenated hydrocarbons in water and soil by dual-channel
ECD/PID equilibrium headspace analysis. J. High Resol. Chrom., 17: 299-302.
8. U.S. Environmental Protection Agency (1993) Draft statement of work for quick turn-
around analysis, Analytical Operations Branch: Washington, DC, EPA/540/R/94/086.
9. Jenkins T.F. and P.W. Schumacher (1987) Comparison of methanol and tetraglyme as
extraction solvents for determination of volatile organics in soil. U.S. Army Cold Re-
gions Research and Engineering Laboratory, Hanover, New Hampshire, Special Re-
port 87-22.
10. Hewitt A.D., T.F. Jenkins, C.L. Grant (1995) Collection, handling, and storage: Keys
to improved data quality for volatile organic compounds in soil. Am. Environ. Lab.
Jan-Feb.
11. Hewitt A.D. and N. J.E. Lukash (1996) Obtaining and transferring soils for in-vial analy-
sis of volatile organic compounds. U.S. Army Cold Regions Research and Engineer-
ing Laboratory, Hanover, New Hampshire, Special Report 96-5.
12. Minnich M.M. and J.H. Zimmerman (Accepted) Extraction methods for recovery of
volatile organic compounds from fortified, dry soils. J. Assoc. Off. Anal. Chem.
329
-------
70
Detection of Cyclodiene Insecticides in Water and Soil Using a Magnetic
Particle-Based Immunoassay
Michele Y. Selisker, David P. Herzog, William A. Day and Jeanne A. Itak
Ohmicron Environmental Diagnostics, 375 Pheasant Run, Newtown, Pennsylvania 18940.
Abstract
Although banned in most countries, the cyclodiene insecticides still cause concern because
of their high toxicity and environmental persistence. Half-lives ranging from two months
to ten years have been reported for various cyclodienes in soil and water. A method for
the quick and simple determination of cyclodiene insecticides in water using a magnetic
particle based enzyme linked immunosorbent assay (ELISA) is described. Results are
reported in parts per billion dieldrin. Recovery from dieldrin spiked waters averaged 97%.
A comparative study of dieldrin recovery from water using immunoassay versus EPA
Method 505 showed good correlation yielding an r2 = 0.958. An extraction method for
use on soil is also described. Soils fortified with the various cyclodienes and extracted
using a methanolic solution containing stabilizers and surfactants averaged 97% recovery.
Introduction
The cyclodiene insecticides comprise a large group of polychlorinated, cyclic
hydrocarbons including two pairs of stereoisomers: aldrin and isodrin, dieldrin and endrin.
Until the discontinuance of production and limitations on domestic usage, the cyclodiene
insecticides most commonly used in the United States were heptachlor, chlordane, aldrin,
dieldrin, endrin, toxaphene and benzene hexachloride (BHC) (Stanker, 1994; Smith,
1991). Though usually considered to be nonsystemic insecticides, they have been shown
to be absorbed by root tissue like carrots and potatoes (Felsot and Pederson, 1991).
The previous widespread use of these compounds has led to concern about the persistence
of these pesticide residues in foodstuffs and the environment. These highly lipophilic
compounds have soil half-lives ranging from 2-10 years (Kidd and James, 1991; Stanker,
1994) which has resulted in severe environmental contamination. Although banned from
use, bioaccumulation has resulted in their prevalent distribution in milk and human fat.
Fish also have accumulations of these compounds in their bodies as a result of
contaminations in the water (Stanker et al., 1994).
Traditional methods for the determination of cyclodiene insecticides in water and soil
matrices require lengthy extraction procedures. Magnetic particle based ELISAs for the
determination of pesticides in water and soil have been previously reported (Itak et al.,
1993 and Lawruk et al., 1993). Described within is a quick and easy-to-use method for
the determination of cyclodiene insecticides in water based on the use of a magnetic
330
-------
particle ELISA. Also described is a short method for the extraction of the cyclodiene
insecticides from soil.
Materials and Methods
Reagents
a. Cyclodienes RaPID Assay ®. RaPID Prep® Soil Collection Kit, RaPID Prep
Cyclodienes Sample Extraction Kit, Cyclodiene Water Stabilizer and Sample
Diluent, (Ohmicron Corp, Newtown, PA).
b. Dieldrin was purchased from Chem Service (West Chester, PA).
c. Compounds Tested for Cross Reactivity: Heptachlor, toxaphene, lindane,
perthane, DDT, DDE and ODD were also purchased from ChemService.
Chlordane, a-endosulfan, aldrin, endrin, isobenzan and heptachlor-endo-
epoxide were purchased from Cresent Chemical (Hauppauge, NY). Isodrin
was purchased from Aldrich (Milwaukee, WI).
Apparatus
a. RaPID Magnetic Separation Unit (Ohmicron Corp, Newtown, PA).
b. RPA-I Analyzer™ (Ohmicron Corp, Newtown, PA).
c. Vortex Mixer (VWR Scientific, South Plainfield, NJ).
d. Adjustable pipettes, Gilson P-1000 and P-200 (Rainin, Woburn, MA).
e. Eppendorf Repeater pipette 4780 (Hamburg, Germany).
Water Preparation
Water samples were collected in glass vessels with teflon cap liners. Immediately upon
collection, samples were diluted with Cyclodienes Water Stabilizer (1 part Water
Stabilizer: 3 parts water sample) to prevent adsorptive losses. After samples dilution,
those samples containing gross paniculate matter were centrifuged in order to remove
particles.
Soil Extraction
Ten grams of soil were weighed or measured by packed volume into a Soil Collection
tube. The contents of one vial of Cyclodienes Extraction Solution was poured into a
collector. The sample was capped and shaken for one minute. The collection tube was
positioned upright in the rack and the contents were allowed to settle for 5 minutes. The
cap was removed from each tube and a filter cap was secured onto each collector. The
soil collector was inverted over a collection vial; the plunger was attached, and pressure
was applied to the handle. The collection vial was filled with a minimum of 30 drops (1.5
mL) of extractant and capped. Using a pipet, 250uL of the soil extract was transferred
directly into a vial of Cyclodienes Extract Diluent (12.25mL). The samples were then
assayed according to Cyclodienes RaPID Assay protocol (see below).
331
-------
The actual Cyclodienes concentration expressed as ng/g dieldrin (ppb) was calculated by
using the following equation.
(1) (assay result) x (dilution) x volume of extract (mL) =
weight of sample (g)
(2) (ug/mL) x (50) x (20mL/10g) = assay result x 100 =
(3) cyclodienes concentration (ng/g or ppb dieldrin)
RaPID Assay Protocol
250uL of the standard, control, or sample to be tested was added to a disposable test tube
along with 250uL of enzyme labeled cyclodiene analog and SOOuL of magnetic particles
with cyclodiene specific antibodies attached. After mixing, the samples were incubated for
30 minutes at room temperature. The reaction mixture was separated using the magnetic
separation rack and washed twice with washing solution. SOOuL of color solution was
then added and allowed to develop for twenty minutes at room temperature. The color
reaction was stopped with SOOuL of stopping solution. Photometric analysis of the final
colored product was made using the RPA-I Analyzer set at 450nm to determine ppb levels
of cyclodienes (as dieldrin) in the samples.
Results and Discussion
The Cyclodienes RaPID Assay is calibrated to dieldrin but detects ten of the main
cyclodiene insecticides. Table 1 summarizes the reactivity of the Cyclodienes RaPID
Assay to dieldrin and other cyclodienes in buffer. The percent cross-reactivity was
determined as the amount of dieldrin analog required to achieve 50% B/Bo divided by the
amount of cross reactant analog required to achieve 50% B/Bo.
Recovery studies were performed using four water samples collected from creeks and
wells from the east coast of the United States. Each water was treated with water
stabilizer and spiked with four levels of dieldrin (0, 3, 6, 10 and 15 ppb). The mean
recovery from water was 97% (n=16). The recovery data is summarized in Table 2.
Inter and intra assay precision studies were performed with four surface and ground water
samples diluted 1:3 with water stabilizer and fortified with one level of dieldrin. Dieldrin
samples were analyzed over 5 days, 5 assays each per day, in singlicate (n=25). The
within- and between-day and total variations were determined according to the method of
Bookbinder and Panosian (1986). Results are shown in Table 3. Coefficients of variation
(%CV) for within day and between day were less than 13%
Twenty dieldrin fortified water samples were analyzed by immunoassay and EPA Method
505. The coefficient of correlation was r^ = 0.958, with a best fit line equation of y =
332
-------
0.778x + 1.15. The correlation graph is shown in Figure 1. The immunoassay recoveries
ranged from 80 - 99% while the GC reported recoveries ranging from 66 - 170%.
The reactivities of the cyclodiene insecticides in soil have been characterized for the
described soil method. The least detectable dose (LDD) for the extraction method was
estimated at the first standard which is equivalent to 100 ng per gram dieldrin in soil.
Table 4 shows the analyte specific soil equivalents for each of the 3 assay standard levels.
While the assay reports results as ppb dieldrin, the soil equivalency table can be used to
semi-quantitate or rank the other cyclodienes. For example, if a heptachlor soil sample
gives an assay response equivalent to 750 ppb dieldrin, it can be estimated to contain
approximately 1200 ppb heptachlor.
In a soil precision study, ten samples of two soils fortified with dieldrin were weighed on a
balance or measured by packed volume in the soil collector prior to analyzing. Results are
shown in Table 5. The overall coefficient of variation for cyclodienes measurement using
the Soil Collection and Sample Extraction Kits for both cases was less than 17%.
Fourteen neat and spiked soil samples collected from around the world were extracted
with the RaPID Prep Cyclodienes Soil Extraction Kit and run in the Cyclodienes RaPID
Assay. Out of the fourteen samples, four were found to be positive (greater than 100 ng/g
dieldrin). Dilutions of these positive readings showed them to be linear. One positive
sample was also confirmed by GC for 700 ppb chlordane.
Conclusions
This study demonstrates the feasibility of applying magnetic particle-based immunoassays
to the detection of pesticide residues in water and soil. The speed of the described
protocol makes it suitable as a screening method for large numbers of samples in short
periods of time. This assay also compares favorably with the EPA recommended method
for the determination of cyclodienes in water.
333
-------
Table 1. Cross reactivity in the Cyciodienes Immunoassay
COMPOUND
Dieidrin
Aldrin
Isodrin
Heptachlor-endo-epoxide
Heptachlor
Endrin
Chlordane
oEndosulfan
Isobenzan
Toxaphene
Lindane
p.p'-DDD
p.p'-DDT
p.p'-DDE
Perthane
90% B/Bo
LDD (ppb)
0.45
0.22
0.32
0.47
0.50
0.51
0.81
0.87
1.04
1.95
15.2
2040
6562
NR
NR
IC50
(ppb)
9.91
7.88
9.54
17.8
15.9
12.3
26.9
44.6
45.8
116.0
846.0
NR
NR
NR
NR
% CROSS
REACTIVITY
100.0
125.4
103.9
55.7
62.3
80.6
36.8
22.2
21.6
8.5
1.2
<0.001
<0.001
<0.001
<0.001
Table 2. Dieidrin recovery from water.
Dieidrin Added
(ppb)
3.0
6.0
10.0
15.0
Mean
(ppb)
3.20
5.72
9.57
13.68
S.D.
(ppb)
0.72
0.78
0.76
1.08
% Recovery
107
95
96
91
Average
97
334
-------
Table 3. Immunoassay precision for the recovery of dieldrin from water.
Sample +3.0ppb +6.0ppb +10.0ppb
Replicates 5
Days 5
N 25
Mean ppb 2.90
% CV Intra 12.8
% CV Inter 10.9
% CV Total 16.2
5
5
25
6.03
11.8
<0.1
11.6
5
5
25
10.00
8.4
<0.1
8.3
+ 15.0 ppb
5
5
25
14.49
9.8
3.8
10.4
Table 4. Cyclodienes in Soil Equivalents.
Analyte
Dieldrin
Aldrin
Isodrin
Endrin
Heptachlor
Heptachlor-endo-epoxide
Chlordane
Isobenzan
ocEndosulfan
Toxaphene
STANDARD 1
( ppb soil)
100
100
100
125
150
150
200
300
300
830
STANDARD 2
(ppb soil)
750
600
700
900
1200
1300
1550
3200
3100
8700
STANDARD 3
(ppb soil)
2000
1700
2000
2500
3400
4000
5000
10,600
10,700
28,000
335
-------
Table 5. Cyclodienes in Soil Precision by Weight and by Volume.
Collection by Weight
Collection by Volume
Soil
replicates
assayed soil concentration
(ppb)
C.V. (%)
Holland Mars Bluff
10 10
971 761
Holland Mars Bluff
10 10
748 626
13.8
11.6
16.0
11.5
Figure 1.: Cyclodienes Correlation: Method 505 vs. RaPID Assay
e
XI
Q.
25
20
15
< 10
y-'KJ^Sx'+l.lS
r2=0.958
10
15
20
25
GC Method 505 (ppb dieldrin)
336
-------
References
The Agrochemicals Handbook (1991) 3rd Ed., H. Kidd and D.R. James, (Eds), Royal
Society of Chemistry, Cambridge, England.
Bookbinder and Panosian, 1986). Clin. Chem., 32/9, pages 1734-37 (1986)
Felsot, AS. and Pedersen, W.L. (1991) Pesticidal Activity of Degradation Products, In:
Pesticide Transformation Products, Fate and Significance in the Environment, L.
Somasundaram and J.R. Coates, Jr., (Eds), American Chemical Society, Washington,
DC.
Itak, J. A.; Selisker, M.Y.; Jourdan, S.W.; Flecker, J.R.; Herzog, D.P. Determination of
Benomyl (as Carbendazim) and Carbendazim in Water, Soil, and Fruit Juice by a
Magnetic Particle-Based Immunoassay. Journal of Agricultural and Food
Chemistry, 1993, 41, 2329-2332.
Lawruk, T., Lachman, C, Jourdan, S., Flecker, J., Herzog, D. and Rubio, F.
Quantification of Cyanazine in Water and Soil by a Magnetic Particle -Based ELISA.
Journal of Agricultural and Food Chemistry, 1993, 41, 747-752.
Rubio, P.M.; Itak, J.A.; Scutellaro, A.M.; Selisker, M.Y.; Herzog, D.P. Performance
Characteristics of a Novel Magnetic Particle Based ELISA for the Quantitative
Analysis of Atrazine and Related Triazines in Water Samples. FoodAgric. Immunol.
1991, (3), 113-125.
Smith, A.G. (1991) Section 15.5: Cyclodiene and Related Compounds, In: Handbook of
Pesticide Toxicology, W.J. Hayes and E.R. Laws, (Eds), Academic Press, Inc.
Stanker, L.H., Vanderlanan, M. and B.D. Watkins. (Aug. 2, 1994) Monoclonal
Antibodies to Cyclodiene Insecticides and Method for Detecting the Same. United
Stales Patent # 5,334,528.
337
-------
71
The Determination of Specific Polychlorinated Biphenyl Congeners in Fish Samples
by Gas Chromatography and Electron Capture Detection
John L. Snyder, Lancaster Laboratories, 2425 New Holland Pike, Lancaster, PA
17601-5994
Because of the persistence of Polychlorinated Biphenyls {PCBs) in the environment,
the detection and accurate quantitation of PCBs remain an integral part of many
environmental studies and remedial actions. In classical USEPA methodology,
PCBs are reported as technical Aroclor mixtures which are recognized by their
distinctive peak pattern on the gas chromatogram. PCB congener specific analysis,
which has been made possible by advances in high resolution gas Chromatography
and the availability of all possible 209 PCB congeners as pure standards, allows
for the individual detection of the specific more toxic planar congeners, has been
shown to be less biased in quantifying PCBs than the classical Aroclor methods,
and permits better tracking of PCBs in the environment because of weathering
which occurs to Aroclor patterns.
In this study a gas chromatographic method was developed to separate and
quantify all PCB congeners. Two dual capillary columns (a non polar Supelco SPB-
Octyl 60 m x 0.25 mm. df 0.25 /ym and a more polar J&W DB17 60 m x 0.25 mm,
df 0.25//m) connected to a short length of guard column were used to separate the
congeners; dual electron capture detectors (ECDs) were used for detecting and
quantifying the congeners. Tetrachlorometaxylene (TCMX) and Decachlorobipheny
(DCB) were added to all standards and samples as retention time markers for
calculating relative retention times and as surrogate standards. Quadruplicate
spiking studies (56 congeners) using Soxhlet extraction showed overall recoveries
of 114% with an overall average RSD of 5.4%. Fish samples (approximately 30)
collected from the Trenton Channel between Canada and the United States by the
US Army Corps of Engineers were analyzed for these 56 PCB congeners. After the
fish samples were Soxhlet extracted, the extracts were cleaned up using gel
permeation Chromatography, sulfuric acid, and florisil solid-phase cartridges. PCB
congeners were found in all of the fish samples and ranged from the detection limit
(less than 1 ppb for most congeners) to a high of 210 ppb for congener 153 in a
Carp sample.
Composites of the fish sample were made to compare the efficiencies of the
Soxhlet extraction, the sonication extraction. Accelerated Solvent extraction (ASE),
and Supercritical fluid extraction. The recoveries and precisions for the PCB
congeners in the fish tissue using these extraction techniques (n = 3 for each
method) were compared.
338
-------
72
NEW AND SPECIALLY DEVELOPED BONDED SILICONE PHASES FOR PESTICIDES
TRACE ANALYSIS WITH CAPILLARY GC.
J. de Zaeuw, P. Heynsdijk and C. Duvekot
Chrompack Int., P.O.Box 8033, 4330 EA Middeifaurg, the Netherlands
Pesticides can be quantified at concentrations down to ppt levels using special
sensitive detectors such as an Electron Capture Detectors. As the absolute amount of
component that is injected on the column is in the order of picograms, the inertness
of the chromatographic system is extremely important.
It is already known that certain pesticides will decompose in a hot injection port. p,p'-
DDT and Endrin are such components.
Also the column must have a high degree of inertness to be able to elute picograms
of pesticides as symmetrical peaks. A very popular column for pesticides analysis is
the CP-Sa 8 CB for Pesticides, which separates the most important pesticides with
excellent resolution, polysiloxane.
In many environmental applications a second, more polar column is often used to
confirm the separation and identification of the apolar column.The commercially
available OV-1701 phases and columns turned out to be quite irreproducible.
especially for the recovery of pesticides like the p.p'-DDT.
A special high purity polymer was synthesized and the polarity was adjusted to be
close to the polarity of OV-1701. This new polymer turned out to be very inert and it
makes the quantitative elution possible of p,p'-DDT at sub-picogram levels. This new
phase is called CP-Sil 19 CB for Pesticides, as it shows unique applications for these
types of compounds.
In this poster the chromatographic behavior of this new phase will be discussed.
339
-------
73
A NEW HPLC STATIONARY PHASE FOR THE ANALYSIS OF POLYCYCLrC AROMATIC
HYDROCARBONS (PAh'S).
J. de Zeeuw, N. Lammers, W. Verstraeten, J.W. Harinissen
Chrompack International B.V., P.O. BOX 8033, 4330 EA Middelburg,
The Netherlands.
A new stationary phase has been developed to meet specific
requirements towards PAH-analysis according to EPA- and ASTM methods:
CP-EcoSpher 4 PAH.
on this phase the EPA-PAfl's are well spread over the ehronatogram due
to it's specific selectivity. Using optimum conditions the
'^peakcapacity11 is over 40, which means that there is enough space
left for other PAH's to elute between the EPA-PAH's. For instance,
this enables better quantification of for instance the components
perylene and benzo(e)pyrene.
Besides that, because of the higher bonding percentage and efficiency
of the material, it is possible to use methanol/water mixtures
instead of the more "popular" acetonitrile/water mixture without a
drastic decrease in separation power. As methanol is less expensive
and less toxic than acetonitrile, the use of methanol offers some
nice advantages.
The day to day deviation of the retention factor k is less than 1%
and the day to day deviation of the selectivity factor a is less than
0.2%
The optimum temperature range for the new phase is lower and wider
(15-35 *C) compared with most other phases as will be shown by a
selectivity test.
340
-------
74
Overview of the Current Status of the RCRA Organic Methods
Development Program.
B. Lesnik
341
-------
75
FREON ALTERNATIVES FOR THE ENVIRONMENTAL LAB
Mark L. Bruce. Quanterra, 4101 Shuffel Dr. NW, North Canton, Ohio 44720.
Jack R. Hall,Quanteifa,9000ExeeutiveParkDrive, Suite A110, Knoxville, TN 37923.
ABSTRACT
The Accelerated Solvent Extractor (ASE) has been used to extract total petroleum
hydrocarbons (TPH) from solid matrices with tetrachloroethene. The extraction
conditions were 200°C and 1500 psi for 10 minutes. Analysis was performed with an
infrared filter photometer following the procedure described in EPA SW-846 Method
8440. ASE extraction typically recovered 50% more TPH than Soxhlet extraction with
Frcon-113®. The exceptions were a group of wet clay samples. For many of these
samples, the result from the initial ASE extract was significantly lower than the Soxhlet
extract A second ASE extraction of the same sample aliquot yielded high levels of
hydrocarbons. Some secondary extracts actually had higher TPH results than the initial
extracts. This illustrates the extraction problem caused by water when only a non-polar
extraction solvent is used. Using hexane in the ASE provides an alternative for
gravimetric analysis of TPH in solid matrices.
INTRODUCTION
As Freon-113® usage is being phased out many people within the environmental testing
community have been searching for alternative solvents and test procedures.
Unfortunately no single alternative solvent has been able to match all the key
characteristics that have made Freon-113® so popular. It has been the primary solvent for
the extraction and analysis of "Oil & Grease" and 'Total Petroleum Hydrocarbons" from
both aqueous and solid matrices. The extraction is usually performed in a separatory
funnel (waters) or Soxhlet (solids). The most common analysis methods are gravimetric
or infrared measurement around 2950cm-1. Selecting the best alternative solvent and
extraction procedure depends on the matrix and type of analysis.
Infrared analysis can be carried out in high purity tetrachloroethene (PCE). This nonpolar
solvent is similar to Freon-113® in polarity and IR background around 2950cm-1. It's
boiling point (121°C) is much higher than the boiling point of Freon-113® (48°C) which
makes it unsuitable for use in most Soxhlet systems. The higher boiling point is not a
problem when using Method 3545, Pressurized Fluid Extraction.
Many gravimetric analyses are switching to hexane. This nonpolar solvent can also be
used in the Accelerated Solvent Extractor system. After removing water from the extract.
silica gel may be added to remove polar extracted materials. The hexane can then
evaporated for gravimetric analysis similar that described in Method 1664.
EXPERIMENTAL
Pressurized fluid extraction (PFE) will be included in SW-846 Update HI as Method
3545. The extraction time is 10 minutes, with sample-to-sample cycle time of about 13
minutes. The required solvent volume ranges from 15 to 50mL depending on the amount
of sample extracted. Sample amounts up to 20g can be routinely extracted. If the sample
is dry enough that very little drying agent is required then 30g quantities can be extracted.
The Accelerated Solvent Extractor system (ASE) has been described previously at this
symposium (1,2). In this application the ASE has been used to extract total petroleum
342
-------
hydrocarbons (TPH) from solid matrices with tetrachloroethene or hexane. The
extraction conditions were 200°C and 1500 psi for 10 minutes unless stated otherwise.
Analysis was performed with an infrared filter photometer following the procedure
described in EPA SW-846 Method 8440. Hexane was investigated as an alternative for
gravimetric analysis of TPH in solid matrices. A few matrix spiked clay samples were
studied. Also, 72 real TPH samples were extracted and analyzed. Most samples were
only processed once. The exceptions being the wet clay samples discussed below.
RESULTS and DISCUSSION
Figure 1 shows the results from several matrix spiked petroleum hydrocarbon samples.
Hydrocarbon recovery from wet samples is more difficult when only non-polar extraction
solvents (such as Freon-113® or tetrachloroethene) are used. It is necessary to dry the
sample either before or during the extraction to achieve good analyte recovery. Both wet
and dry sample matrix spikes of diesel and motor oil were sequentially extracted until
quantitative recovery was achieved. The two dry samples were quantitatively extracted
with a single 10 minute extraction with tetrachloroethene at 200°C.
The wet samples were more difficult to extract. The fresh diesel spiked sample required a
second 10 minute, 200°C extraction to reach quantitative recovery. The aged motor oil
spike although easy to extract when dry became very difficult to extract when the
moisture content of this clay sample was brought to 50%. Three 10 minute extractions
were required to achieve quantitative recovery. The first extraction had very little
hydrocarbons recovered, but the extract contained significant amounts of water.
Subsequent extractions of the same clay sample aliquot recovered more hydrocarbons and
less water. These wet clay results illustrate the problem encountered when extracting wet
samples with nonpolar solvents. Fortunately most real samples were not this difficult, as
described below.
120 -,
100 -
8
G>
cc
dry (aged oil)
dry (diesel)
wet (fresh diesel)
wet (aged oil)
extraction number
Figure 1 Recovery of TPH from spiked clay
343
-------
A wide range of real samples were extracted with both Soxhlet and ASE. The IR analysis
results are shown in Figure 2. ASE extraction typically recovered 50% more TPH than
Soxhlet extraction with Freon-113®. The diagonal line in the center indicates the region
where the ASE concentration data corresponds exactly to the Soxhlet data. The shaded
region covers the area from ASE results being 50% higher to 50% lower than the Soxhlet
results. The X and Y axes are shown in log-log space since the concentration data cover
many orders of magnitude.
TPH-IR
I
UJ
u
a.
ui
CO
10000 T
1000 .-
100 ••
10 ••
1
1
10 100 1000 10000
Soxhlet: Freon-113 (mg/kg)
Figure 2 Comparison of TPH Results for ASE and Soxhlet Extractions
The exceptions to the general trend of higher recovery with the ASE were a group of wet
clay samples. For many of these samples, the result from the initial ASE extract was
significantly lower than from the Soxhlet extract. A second ASE extraction of the same
sample aliquot yielded high levels of hydrocarbons. Some secondary extracts actually
had higher TPH results than the initial extracts. The average Soxhlet data and the sum of
the two ASE results are shown in Figure 3. With one exception, the combined ASE
results were equivalent or higher than the Soxhlet results. Usually the first ASE
extraction yielded l-2mL of water and 30-70% of the TPH. The second extraction had no
water and the remainder of the hydrocarbons. This illustrates the extraction problem
caused by water when only a non-polar extraction solvent is used. The water reduces
analyte recovery until it is removed from the surface of the sample. It is possible the
water forms an insoluble barrier between the analytes on the particle surface and the non-
polar extraction solvent.
344
-------
ASE 2nd pass
ASE 1st pass
ox h let Avg
#1 #2 #3 #4 #5 #6 #7 #8 #9 #10 #11 #12
Sample
Figure 3 Comparison of Wet Clay TPH Results for ASE and Soxhlet Extractions
Unfortunately the 200°C ASE extractions wear out the polymeric seals of the extraction
vessel more quickly than the usual 100°C extractions (1,2). A few of these wet clay
samples were repetitively extracted at 100°C for 10 minutes. Figure 4 illustrates that
TPH recovery was significantly lower at 100°C than it was at 200°C. The amount of
water in the initial passes of the 100°C extractions was much lower than in the first pass
of the 200°C extraction. It appears that water was not being efficiently removed from the
surface of the sample and thus inhibited recovery of the hydrocarbons. The single 200°C
extraction after 5 extraction passes at 100°C demonstrates that significant levels of
hydrocarbons were still available for recovery. Presumably water elimination (and thus
TPH recovery) was enhanced by the increased solubility of water in tetrachloroethene at
200°C relative to 100°C. This water interference effect was very pronounced even
though all samples had been mixed with at least an equal amount of sodium sulfate
drying agent.
345
-------
200°C 6th pass
iiiniiiiiiii 100°C 5th pass
100°C 4th pass
100°C 3rd pass
100°C 2nd pass
100°C 1st pass
200°C double pass
#1 #2 #3 #4 #5
#6 #7 #8
Sample
#10 #11 #12
Figure 4 Comparison of Wet Clay ASE TPH Results for 100°C and 200°C Extractions
Eight of the wet clay samples were extracted with hexane at 200°C for 10 minute passes.
Analysis was performed gravimetrically. Figure 5 shows the results were similar to the
tetrachloroethene extraction results. However, the hexane extraction with gravimetric
analysis was not equivalent to the tetrachloroethene extraction with 1R analysis. This was
expected since the solvents and analysis methods were quite different.
12000T
3rd pass hexane
2nd pass hexane
1 st pass hexane
double pass PCE
#1 #2 #3 #4 #5 #6 #7 #8 #9 #10 #11 #12
Sample
Figure 5 Comparison of Wet Clay ASE TPH Results for Gravimetric and IR Analysis
Figure 6 summarizes the results from Soxhlet extraction with Freon-113® and IR
analysis, ASE extraction with PCE and IR analysis and ASE extraction with hexane and
gravimetric analysis. The trends in the analytical data are similar but are sufficiently
different to demonstrate that the three combinations of extraction and analysis methods
are not likely to produce equivalent results for all samples.
346
-------
12,000
10,000
8,000 •
X
Q.
6,000 •
4,000
2,000
#2
ASE (PCE)2x10min
ASE (Hexane) 2x1 Omin
Soxhlet (Freon) 18hr
#11
#5 ^
Sample
Figure 6 Summary of Wet Clay TPH Results for ASE IR & Grav and Soxhlet IR
#12
CONCLUSION
ASE extraction efficiency is generally equivalent or better than the traditional Soxhlet
techniques. A single 10 minute. 200°C extraction is sufficient for most soil samples. The
difficult to extract wet clay samples require two extraction passes at 200°C to complete
the extraction. The ASE PCE-IR results are similar but not equivalent to the ASE
hexane-gravimetric results. There do not appear to be any replacement solvents or
methods which will produce equivalent results to the Soxhlet extraction with Freon-113®
followed by IR analysis for all sample types. Since Freon-113® is being phased out
changes to alternative solvents/methods will necessitate changes in the way the data are
interpreted and used.
ACKNOWLEDGMENTS
Many people have contributed to the success of this study. In particular the authors
would like to thank Bruce Richter. John Ezzell, Dale Felix and Brent Middleton from the
Dionex Corporation. Essential support was provided by Sarah Braxter, Tami Stephens,
Dave Cox and Bruce Hart from Quanterra Environmental Services.
REFERENCES
(1) Richter, Felix, Ezzell, llth Waste Testing & Quality Assurance Symposium 1995,
p!21-127.
(2) Bruce, Hall, llth Waste Testing & Quality Assurance Symposium 1995, pi 14-120.
347
-------
76
EVALUATION OF SOLID PHASE EXTRACTION FOLLOWED BY
SUPERCRITICAL FLUID ELUTION OF SEMI-VOLATILE ORGANIC
COMPOUNDS IN TOXICITY CHARACTERISTIC LEACHATES.
William E. Corl III. PWC Environmental Laboratory, Norfolk Naval Base, Norfolk Virginia 23511-3095;
and Department of Chemistry and Biochemistry, Old Dominion University, Norfolk, Virginia 23529-4628.
ABSTRACT
The Environmental Protection Agency (EPA) developed the toxicity characteristic leaching
procedure (TCLP) to determine the potential for unknown wastes to leach hazardous constituents
into the environment Current methods being used for the extraction of semi-volatile organic
compounds from aqueous TCLP leachates involve acid/base partitioning into methylene chloride.
The concentrated extract is then analyzed by gas chromatography/mass spectrometry (GC/MS).
Continuous liquid/liquid extractions require up to 48 hours for completion and consume up to 500
mis of methylene chloride. Separatory funnel extractions are faster and require less solvent but
are labor intensive and prone to forming emulsions, resulting in poor recoveries. Ironically, the
wastes generated from these extractions are themselves a listed hazardous waste. Methylene
chloride has also been targeted for production elimination due to its ozone depleting properties.(l)
Considering the hazards associated with exposure to chlorinated solvents and the disposal costs
of methylene chloride wastes, great interest has been elicited by the EPA in other extraction
techniques which would reduce the volume of solvent wastes being generated by environmental
laboratories. A hyphenated technique utilizing solid phase extraction (SPE) followed by elution
with supercritical fluid carbon dioxide (SFE) has been developed. Method development included
the optimization of both the SPE and SF elution variables. The modified method has been found
to show comparable detection limits, accuracy, and precision for most of the TCLP compounds
while drastically decreasing extraction time and virtually eliminating methylene chloride.
Comparison of SPE/SFE extractions of fortified TCLP leachate solutions with currently used
extraction procedures have been compared.
INTRODUCTION
Two extraction methods which offer the benefit of reduced solvent consumption are liquid-solid
extraction or solid-phase extraction (SPE) and supercritical fluid extraction (SFE). In addition to
reducing the amount of methylene chloride being used, these two methods may also offer
drastically reduced extraction times, and reduction in the manual labor involved with liquid solvent
extraction procedures.
Solid phase extraction has been widely used in environmental applications where a
preconcentration and/or an isolation step is required prior to analysis. Reversed phase sorbents
are the most applicable for extracting organic compounds from aqueous samples, with a variety of
reversed phase sorbents available. Alkyl bonded silica's such as C18 and C8 have been shown to
exhibit quantitative extraction capabilities for large non-polar compounds such as polycyclic
aromatic hydrocarbons (PAH) (2) and polychlorinated biphenyls (PCB) (3). However, polymeric
sorbents such as styrene-divinyl benzene (SDVB) have been shown to exhibit retention values
20-40 times higher than C18 for moderately polar compounds in aqueous samples (4,5). Recent
advances in solid phase technology has produced the solid phase extraction disks (6,7). The solid
phase Empore™ disks were first introduced by 3M (St. Paul, MN) in 1990. The characteristics of
a smaller particle size with a high surface area, combined with a uniform particle distribution within
348
-------
the disk, results in extremely high flow capabilities while maintaining extraction efficiencies
comparable to or exceeding conventional solid phase extraction cartridges. Solid phase extraction
disks have already been examined for a variety of environmental applications and will also be
utilized in proposed hazardous waste method 3535 in the up-coming third update of SW-846 (8).
Other methods for wastewater and hazardous waste using solid phase disk extraction are
currently being reviewed for validation by the EPA.
Supercritical fluid extraction (SFE) is an additional analytical technique which is being heavily
evaluated by the EPA as a replacement for classical methods using liquid solvents. Supercritical
fluids have intermediate properties of liquids and gases, providing perhaps the most influential
characteristic of SFE; very high mass transfer capabilities. With viscosity's similar to that of gases
and solute diffusivities higher than that of liquid solvents, supercritical fluids show much faster
extraction kinetics than conventional liquid solvent extraction techniques. While CO2 is the most
commonly used fluid due to its mild critical constants (73 atm @ 31° C). the addition of polar
modifiers is usually used to aid in increasing the extraction efficiencies of more polar compounds.
SFE has been limited to mostly solid and semi-solid matrices. SFE applications for the direct
extraction of organic compounds in aqueous matrices have been slow to gain acceptance.
Although Hedrick and Taylor (9,10) have demonstrated the extraction of organic compounds
directly from aqueous samples, only small sample volumes could be extracted. The mobility of the
water through the extraction cell to the restictor outlet can result in restrictor plugging. A more
widely accepted approach for extracting organic compounds from aqueous samples using SFE is
performed by first immobilizing the analyte(s) on a stationary support and then extracting the
support using SFE. This type of technique using solid phase extraction coupled to supercritical
fluid elution (SPE/SFE) has also been used for the extraction and measurement of organics in
aqueous matrices.(11) Recently, Messer and Taylor (12) reported the use of 47-mm C18 disks for
the recovery of trace semivolatile analytes from water using a three step, acetone-modified SFE-
CO2 elution technique. Quantitative recovery of over 90% of the EPA Method 525.1 analytes,
including PNA's, PCB's, phthalates, and orgonochlorine pesticides (OOP's) were determined in
fortified reagent water with the majority of the RSD's found to be below 15%.
In the following study, a method has been developed for the extraction of the TCLP-regulated
semivolatile compounds in TCLP leachates by isolating the compounds onto a solid phase
extraction disk followed by elution with modified supercritical fluid CO2. The procedure required
that the optimization of both the solid phase extraction and the supercritical fluid elution be
examined separately. The two phases of solid disk extraction followed by supercritical fluid elution
were sequentially evaluated, optimized, and then combined. This strategy simplified the approach
by reducing the number of co-variate parameters involved when integrating the two methods. The
main goal of the extraction procedure was to quantitatively recover the TCLP semi-volatile
compounds from TCLP leachates while minimizing the amount of hazardous waste produced.
Further requirements of the method that were deemed essential were that the extraction
procedure must meet the detection limits, recoveries, and additional QC requirements mandated
using current protocols. The extraction procedure was found to be completely compatible with the
current method of quantitation (M8270) using gas chromatography/mass spectrometry (GO/MS).
Optimization of Solid Phase Extraction. Initial experimentation was performed to maximize the
retention of the TCLP analytes on the solid phase extraction disk. The first portion of the SPE
optimization was to establish the type of solid phase disk which had the highest retention for the
TCLP analytes. Afterwards, optimum conditions of pH and ionic strength were evaluated as well.
Determination of the capacity factor of the solid phase disk was also performed in order to prevent
analyte loss when filtering larger volumes of the fortified leachate. A mass balance determination
of eluent and filtrate extracts was also performed using all of the optimized conditions as a
confirmation of the solid phase extraction efficiency.
349
-------
Optimization of Supercritical Fluid CO2 as an Elution Solvent. A Suprex Prepmaster
supercritical fluid extractor with Accutrap module for sub-ambient trapping of the analytes was
used for all SF-elutions. Evaluation of the trap desorption (efficiency of solvent elution of the
compounds from the trap) was the initial task at hand. This was performed by spiking the top of
the solid trapping material with the TCLP targets and surrogates prior to sealing the trap. After
inserting the trap into the Accutrap module, solvent was pumped through the trap at 1 ml
increments and collected into separate 2-ml autosampler vials until a total volume of 10 mis had
been eluted through the trap. This initial procedure established the optimum type and amount
elution solvent required for quantitative removal of the compounds from the trap. The next step in
optimizing the trapping procedure was to evaluate the isolation of the compounds from the
decompressing fluid onto the solid trapping media. A 1 gram portion of hydromatrix was fortified
with the TCLP target and surrogate compounds prior to dynamic supercritical fluid extraction. The
trap was rinsed with solvent after 10 volumes (10mls as fluid) of CO2 had been used. This
procedure was repeated using several trapping temperatures and flow rates. After the
optimization of trapping and collection had been completed, the optimum conditions for
supercritical fluid efution of the TCLP compounds and the related surrogates from a solid phase
extraction disk were investigated. For this portion of the study, 47-mm disks using the previously
determined, optimized-SPE conditions were used. The TCLP and associated surrogates were
spiked at SOug each into 50 ml aliquots of TCLP extraction fluid prior to filtering through the disk.
The disk was then dried prior to SF elution. Disks were either dried by vacuum directly on the
filtration assembly or by pressure drying using nitrogen gas. Pressure drying was performed by
placing the disk between two stainless steel support screens which were then placed into a
stainless steel zero headspace extraction assembly or ZHE (Associated Design and
manufacturing, Alexandria, VA). Optimization of SF elution required evaluation of pressure,
temperature, modifier addition, and volume of fluid required. The CO2 flow rate was adjusted
manually by manipulation of the expansion fitting on the Duraflow restrictor while modifier was
injected directly into the extraction vessel prior to extraction. All other extraction variables were
manipulated via direct programming of the extractor unit.
RESULTS AND DISCUSSION
Solid Phase Extraction. The initial evaluation of which phase of these two would be more
efficient at retaining the TCLP analytes is summarized in figure 1. The absolute peak areas were
used as indicators of disk performance. 47-mm disks were used for extraction of TCLP leachates
fortified with SOug of the TCLP semi-volatile compounds prior to solvent elution with 1:4 ethyl
acetate/methylene chloride. Quantitative recoveries were not of concern at this point in that other
optimization variables had not yet been investigated. Overall extraction efficiencies were higher
for the SDVB disk for most compounds. The SDVB disks showed surprising increases in
extraction efficiences for the nitre-containing compounds relative to the C18 disks. A 24% and
28% increase in recovery for nitrobenzene and 2,4-dinrtrotoluene respectively, were found using
the SDVB disks. The strong electron withdrawing nitro-groups favor delocalization of the aromatic
pi-electrons through resonance and inductive effects. This results in an increased pi-electron
cloud area which may increase the availability of pi-pi interactions with the aromatic moiety of the
SDVB ligand. In contrast, an almost quantitative loss of pyridine was observed using both disks.
Effects of pH and ionic strength on extraction efficiencies were also evaluated. The salting-out
caused by the addition of 10% NaCI resulted in an increase in the recoveries of all of the
compounds. A significant increase in recoveries for the phenolic compounds was observed when
350
-------
the pH of the extraction fluid was decreased. The lowered pH ensures that the acidic compounds
exist mainly in their unionized forms, with a subsequent reduction in water solubility. The decrease
in pH is especially important for pentachlorophenol; the most acidic of the phenolic compounds
(pKa 4.9).
The data in table 1 shows the recoveries of the TCLP targets and base/neutral surrogates spiked
at 100ug and acid surrogates spiked at 200 ug, in 50, 100, 250, and 500 mis of extraction fluid
filtered through 90 mm SDVB disks. The non-polar, neutral organic compounds were found to
have little break through. Conversely, some of the more polar compounds showed rapid break
through between 250 and 500 mis of sample filtered. Establishment of the breakthrough volume is
particularly critical here considering that a final extract volume of 1 ml results in a considerable
concentration factor. This concentration factor will play an important role in maintaining method
detection limits below the regulatory limits of the TCLP compounds. A mass balance evaluation
was performed as a final evaluation of the effectiveness of the 90 mm SDVB disk using all of the
optimized conditions and summarized in table 2. As was expected, a quantitative recovery of
pyridine was recovered from the filtrate indicating that the sorbent failed to retain this compound.
Poor retention was also found to be responsible for the discrepancies for the two acidic
surrogates; phenol-d6 and 2-fluorophenol. A mass balance for each of the other compounds
showed discrepancies to be within experimental error except for 1,4-dichlorobenzene and
hexachlorobutadiene.
Supercritical Fluid Elution. Evaluation of trapping efficiencies for the TCLP analytes from a
fortified inert solid (hydromatrix) are summarized in figures 2 and 3. Desorption from the trap was
performed by gentle heating of the trap followed by rinsing with an appropriate solvent. A
minimum trapping temperature of -5° C was needed in order to obtain quantitative recoveries for
1,4-dichlorobenzene and hexachloroethane while a trapping temperature of-20° C was required
to achieve a semi-quantitative recovery of 86% for pyridine. The extreme trapping conditions
required for pyridine recovery is believed to be due to the combined effects of its high vapor
pressure and poor adsorption on the trapping media. The other compounds were quantitatively
recovered at slightly higher temperatures although precision decreased with increasing trap
temperatures for most compounds. Initial poor recoveries for some of the phenolic compounds
were recovered upon an addition of a methanol modifier. Subsequently, a methanol modifier
concentration of 3% was added directly to the extraction vessel prior to extraction. SFE extraction
conditions of pressure and temperature were selected at 450atm and 50°C to maintain high fluid
densities at lower temperatures. Temperature limitations were mandated due to disk degradation
and water plugging considerations. The conditions of lower temperatures and higher pressures
equated to higher solvent densities which increased the solvating power of the fluid, decreased
restrictor plugging, and decreased the potential of thermal degradation of analytes. The length of
extraction time or total amount of fluid used was found to be vital to quantitative recoveries of
analytes, with 3 to 4 extraction vessel volumes usually found to be sufficient to quantitatively
sweep solubilized species from the vessel. A summary of the final, optimized conditions for the
solid phase extraction and supercritical fluid elution procedures are given in tables 3a and 3b
respectively.
CONCLUSIONS
Final evaluation of the optimized procedure is summarized in tables 4,5, and 6. A recovery
comparison using typical liquid solvent extraction procedures on fortified TCLP leachates to the
solid phase extraction procedure followed by solvent elution and SF-elution is found in table 4.
The solid phase techniques were found to be comparable to the liquid solvent procedures for all of
the TCLP semi-volatiles and surrogates except for pyridine. The optimized SPE/SFE procedure
showed the greatest reduction in the extraction time and amount of solvent required. The
351
-------
procedure was also found to show acceptable surrogate recoveries relative to current method QC
requirements. Finally, a series of fortified samples were extracted at the 5ug level to determine the
method detection limit The calculated values are given in table 6 which show that the modified
method not only met the quantification levels relative to the regulatory limits, but showed
comparable detection limits relative to the liquid techniques as well.
The failure of the solid phase extraction to isolate pyridine realizes the importance of further
evaluation of additional solid phase sorbents. As a matter of logistics, one disk with the capability
of simultaneously extracting all of the semi-volatile compounds is preferable. To augment this
study, current research is being performed on actual leachates using the modified method in order
to evaluate if the method is robust enough for leachates with high organic loading. Finally, direct
coupling of the supercritical fluid eluent to the GC inlet of the chromatographic system is the
desired end-product The future on-line method will allow for an extremely fast extraction,
separation, identification, and quantitation procedure with a greater sensitivity and will literally
eliminate the need for solvent
A CKNOWLEDGEMENTS
The author gratefully acknowledges the assistance of Craig Marked of 3M Corporation, Lori
Dolatta and the other staff of Suprex Corporation, and Nick Ringo of NAStech Systems.
REFERENCES
(1) "Looming Ban on Production of CFC'S, Halons Spurs Switch to Substitutes," Chemical &
Engineering News, 1993, Vol. 71 (48), pp 12-18.
(2) Loconto, P.R., I.C.-GC. 1991, Vol. 9, pp 752.
(3) Ozretich, R.J. and Schroeder W.P., Analytical Chemistry 1986, Vol.58, pp2041.
(4) Hennion, M-C; Scribe, P. Environmental Analysis: Techniques, Applications and Quality
Assurance; Barcelo, D., Ed.; Elsevien Amsterdam, 1993; Vol. 13, pp. 23-77.
(5) Hennion, M-C.; Coquart, V. Journal ofChromatography 1993, Vol. 642, pp.211-214.
(6) Hagen, D. R., Markell, C., Schmitt, G., and Blevins, D. B., Analytica Chimicta Acta 1990, Vol.
236, pp. 157-164.
(7) Markell, C., Hagen, D. F., Bunnelle, V. A., LC.-GC. 1991, Vol. 9, pp.332-337.
(8) Marsden, P. M., Environmental Testing and Analysis, 1995, Vol. 4, pp. 30-34.
(9) Hedrick, J. L., and Taylor, L. T., HRC & CC, 1990, Vol. 13, pp. 312-316.
(10) Hedrick, J. L., and Taylor, L. T., HRC& CC, 1992, Vol. 15, pp. 151-154.
(11) HO, J. S., Tang, P. H., Eichelberger, J. W., and Budde, W. L., Journal of Chromatographic
Science, 1995, Vol. 33, pp. 1-13.
(12) Messer, D. C., and Taylor, L. T., LC-GC, 1996, Vol. 14 (2), pp. 134-142.
352
-------
Comparison of Active Phase Sorbents
DC18 AREA
DSDVB AREA;
5.00E+04
4.50E+04
4.00E+04
3.50E+04
£ 3.00E+04
Q.
Figure 1. Peak areas of TCLP compounds recovered from a fortified leachate.
Extractions performed on 47-mm disks with 1:4 ethyl acetate/methylene chloride used
as the elution solvent.
353
-------
TABLE 1 EXTRACTION EFFICIENCES OF 90MM SDVB DISKS
% RECOVERY AS VOLUME FILTERED (MLS)
BASE/NEUTRALS 5Q IflQ 25fl 5QQ
Nitrobenzene-dS (Surrogate) 89.2 84.6 82.6 78
Nitrobenzene 92.4 88.2 78.2 76
2-Fluorobiphenyl (Surrogate) 85.7 85.4 76.7 81
Terprenyl-d14 (Surrogate) 113.8 104.9 108.7 101
1 ,4-Dichlorobenzene 87.4 83.1 77 65
Hexachloroethane 80.8 76.5 75 65
Hexachlorobutadiene 81.7 78.8 76 77
2,4-Dinitrotoluene 98.7 95.5 96.6 76
Hexachlorobenzene 97.3 87.5 85 65
Pyridine 5.4 3 3.1
ACIDS
2-Fluorophenol (Surrogate) 79.5 76.7 66.7 37
PhenoW6 (Surrogate) 66.1 55.1 45.1 21
O-Cresol 86 80 76.8 37
M/P-Cresol 82.3 75.5 72.2 33
2.4,6-Trochlorophenol 99.8 75.8 68.5 53
2,4,5-Trichlorophenol 88.7 77.6 72.3 47
2,4,6-Tribromophenol (Surrogate) 99.3 93.8 83.5 74
Pentachlorophenol 99 94.8 89 75
Acid surrogates were spiked at 200ug, Base/neutral surrogates and TCLP compounds
spiked at 100ug. Solvent elution with 1:4 ethyl acetate/methylene chloride.
354
-------
Table 2
MASS BALANCE OF SPE
Nitrobenzene-d5 (Surrogate)
Nitrobenzene
2-Fluorobiphenyl (Surrogate)
Terprenyl-d14 (Surrogate)
1,4-Dichlorobenzene
Hexachloroethane
Hexachlorobutadiene
2,4-Dinitrotoluene
Hexachlorobenzene
Pyridine
2-Fluorophenol (Surrogate)
Phenol-dS (Surrogate)
0-Cresol
M/P-Cresol
2,4,6-Trochlorophenoi
2,4,5-Trichlorophenol
2,4,6-Tribromophenol (Surrogate)
Pentachlorophenol
Three replicates of 250 ml aliquots of TCLP extraction fluid fortified with 200 ug of
acid surrogates, 100 ug of base/neutral surrogates and TCLP target compounds.
pH adjusted to 2, 10% w/v NaCI added.
% recovered
90mm SDVB
solvent elution
(RSm
82.6(9.1)
78.2(7.7)
85.4(5.7)
108.7(11)
77(11)
75(10.4)
76(5)
96.6(7.6)
84(13.2)
3.1(67)
66.7(12.7)
45.1(23)
76.8(9.9)
72.2(12.7)
68.5(15.6)
72.3(8.4)
83.5(9.6)
89(15)
% unretained
(recovered
from filtrate)
fRSD)
12.7(3.7)
15(5.5)
3.8(7.8)
-
3.6(2)
20.9(5.5)
<1(6)
4(3.8)
4.1(2)
95(13)
30(9)
44(10.1)
18.2(6.6)
24(4.7)
21.7(13)
24.8(8.8)
11(4.4)
-
%
balance
4.7
6.8
10.8
-
I 194 |
4.1
I 24 |
11.9
1.9
1.4
10.9
5
3.8
9.8
2.9
5.5
11
355
-------
ACID TRAPPING EFFICIENCIES
01
o
UJ
105
95
85
75
65
55
45
-O-Cresol
-M/P-Cresol
- 2,4,6-Trochlorophenol
- 2,4,5-Trichlorophenol
- Pentactilorophenol
40 20 10 5 0 -5
TRAP TEMPERATURE (C)
-10
-20
Figure 2. Trapping efficiencies of TCLP acid compounds on a solid trap eluted from 1 gram
hydromatrix using 10 mis SFE-CO2 at 80°C and 300 atm (d=.7495)
\u
§
u
Hi
te
BASBNEUTRAL TRAPPING EFFICIENCIES
20 10 5 0 -5
TRAP TEMPERATURE (C)
Nitrobenzene
1,4-Dichlorobenzene
Hexachtoroethane
Hexachlorobutadtene
2.4-Dinitrotoluene
Hexacti lorobenzene
Pyridine
Figure 3. Trapping efficiencies of TCLP base/neutral compounds on a solid trap eluted
from 1 gram hydromatrix using 10 mis SFE-COZ at 80°C and 300 atm (d=.7495).
356
-------
Parameter (Values
Sorbent
Size
Sample Volume
Sample Conditions
Drying
SDVB Disk
90-mm
200 mis
pH=2, 10% NaCI
N2@ 10 psi, 5 min.
Table 3.a Optimum Conditions for
Solid Phase Extraction
Parameter (Values
Extraction Conditions
Vessel Size 10ml
Temperature 50 °C
Pressure 450 atm (psi)
Density 0.987
Extraction Fluid 3% methanol-CO,
Flow Rate 2 mls/min
Static Time 20 min
Dynamic Time 30 min
Trapping Conditions
Solid Trap 1:1 Glass Beads/
ENVI-Chrom P
Trap Temperature -5 C
Restrictor Temp. 50 C
Desorption Conditions
Rinse Solvent 1.2 mis of 1:4 Ethyl
Acetate/CH2Cl2
Rinse Flow Rate 1 ml/min
Trap Temperature 40 C
Table 3.b Optimum Conditions for SFE
Elution of SDVB Disk
357
-------
Comparison of Extraction Procedures
90-mm SOVB
solvent elution
77(8.8)
82(12.1)
89(11.3)
75(8.2)
93(15.4)
98(7)
79(10.6)
85(7.8)
83(5.3)
97(7.1)
84(12.7)
3(67)
1.3 Hours
50 mis
90-mm SDVB
SF elution
72(2.2)
78(11.1)
88(2.3)
83(5)
91(13.8)
108(12.8)
79(5.1)
88(9.3)
89(6.4)
90(6.1)
82(7)
n/a
50 Minutes
1.2 mis
% Recovery (RSD)
Sep. funnel
EPA 3510
79(4.2)
88(9.8)
91(4.6)
90(5.8)
88(10)
93(8)
83(15.5)
88(5.5)
85(14.7)
100(2.9)
88(8.3)
64(14.3)
5 Hours
350 mis
Cont. Liq.;Liq.
EPA 352Q
88(3.9)
92(4.8)
88(1.5)
98(7)
92(4.9)
102(4.2)
90(3)
97(4.9)
89.2(5.8)
98(5.2)
94(3.9)
78(9.5)
36 Hours
500 mis
O-Cresol
M/P-Cresol
2,4,6-Trochlorophenol
2,4,5-Trichlorophenol
Pentachlorophenol
Nitrobenzene
1,4-Dichlorobenzene
Hexachloroethane
Hexachlorobutadiene
2,4-Dinitrotoluene
Hexachlorobenzene
Pyridine
Extraction Time
CH2CI2 Required
Table 4. Extractions of 200 ml, TCLP leachates fortified with 100 ug of base/neutrals,
and 200 ug of acids as EPA Certified Reference Standards. (n=7)
Surrogate Recoveries
Surrogate
90-mm SOVB
solvent elution
90-mm SDVB
SF elution
Sep. funnel
EPA 3510
Cont. LiqJLiq.
EPA 3520
Recovery Criteria
EPA 8270/Water
Nitrobenzene-d5 95.9
2-Fluorobiphenyl 86.4
Terphenyl-d14 108.7
PhenokJ6 61.3
2-Fluorophenol 75.6
2.4,6-Tribromophenol 88.3
91.7
88.4
77.5
69.2
78.8
98.6
95
89.2
105
73.7
76.1
100
95.1
98.2
101.4
83.2
89
95.4
23-120
30-115
18-137
24-113
25-121
19-122
Table 5. Mean surrogate recoveries for extraction procedures. (n=7)
Acid surrogates fortified at a concentration of 200 ug/L
Base/Neutral surrogates fortified at 100 ug/L
358
-------
PYRIDINE
1,4-DICHLOROBENZENE
0-CRESOL
HEXACHLOROETHANE
NITROBENZENE
M/P-CRESOL
HEXACHLOROBUTADIENE
2,4,6-TRICHLOROPHENOL
2,4,5-TRICHLOROPHENOL
2,4-DINITROTOLUENE
HEXACHLOROBENZENE
PENTACHLOROPHENOL
CALCULATED METHOD DETECTION LIMITS*
CONCENTRATIONS IN ug/L
METHOD 3510B METHOD 3520B SPEWITH REGULATORY
SEP. FUNNEL CONT. LIQ./LIQ. MECL2 SPE/SFE LIMIT
71
21
21
25
15
51
18
30
24
48
32
50
29
26
27
38
16
42
8
17
26
27
16
70
23
35
32
21
19
27
18
20
23
20
45
-
29
35
24
59
35
46
36
37
45
43
42
5000
7500
200,000
3000
2000
200,000
500
2000
400,000
130
130
100,000
Table 6.
•METHOD DETECTION LIMIT WAS BASED UPON SEVEN, 200-ML REPLICATE ANALYSES
OF TCLP EXTRACTION FLUIDS SPIKED WITH 20ug EACH OF BASE/NEUTRAL TARGET
COMPOUNDS AND 40ug OF THE ACID TARGET COMPOUNDS.
VALUES WERE CALCULATED BY MULTIPLYING THE STANDARD DEVIATION BY ONE-SIDED
STUDENTS t-VALUES DETERMINED BY N-1 DEGREES OF FREEDOM AT THE 99% CONFIDENCE
INTERVAL.
359
-------
77
EVALUATION OF THE MICRODISTILLATION METHOD (METHOD 5031) FOR MEASURING
VOLATILE WATER-SOLUBLE COMPOUNDS IN PULP MILL TREATMENT INFLUENTS AND
EFFLUENTS
Alex Gholson. Diana Cook, and Dean Hoy, West Coast Regional Center, NCASI, PO BOX 458,
Corvallis, Oregon 97339
ABSTRACT
A variety of low boiling point, water-soluble compounds are produced as byproducts of the chemical
pulping process. Analysis of these compounds has been complicated because they are too soluble to either
extract or purge from water. Method 5031 - Microdistillation, which provides a simple method for
concentrating these volatile water-soluble compounds from a water matrix, is proposed in the third update
of SW-846. This report presents an evaluation of the microdistillation method using GC-FID analysis for
analysis of pulp mill wastewater. General method performance was evaluated by determining recovery,
precision, sensitivity and the operating range of the method. Specific studies examined the effect of
sample pH, sample preservation, storage stability, choice of internal standard, optimization of GC column
and injector conditions, and the effect of sample and distillate volume. Of the nineteen compounds
studied, seventeen were found to be recovered and quantified with good precision at levels ranging from 5
ug/L to 20,000 ug/L. Several compounds in effluent samples were found to be unstable unless preserved
at low pH. Distillation of samples at high pH was found to result in the loss of four compounds. When
samples were neutralized before analysis, no loss was observed. The volume of the sample distilled (10 to
40 mL) had an effect on the results, while the volume of distillate collected (0.1 to 0.4 mL) had no
observable effect A summary of QA results and suggestions on how to analyze samples by direct aqueous
injection is also presented.
INTRODUCTION
A variety of low boiling point, water-soluble compounds are known to be produced in the kraft pulping
process (1). Among those identified, methanol, acetone, and 2-butanone are common and are often
present at parts per million concentrations. Many of these compounds have been identified in regulatory
statutes such as the Clean Air Act and RCRA. Components of these statutes may require the pulp and
paper industry to perform analyses for these compounds in a variety of matrices and at levels in the parts
per billion range. For low-level work, purge and trap methods have typically been employed. However,
even with the benefit of isotopic dilution (EPA Method 1624), purge and trap methods perform poorly for
water-soluble compounds, and for methanol, purge and trap will not work at all. Purge efficiencies are
low and can be quite variable depending on sample matrix and purge conditions (2). Solvent extraction
methods are not effective for the water-soluble volatile compounds because of the poor partitioning from
water into immiscible solvents and volatile loss during solvent concentration steps. Direct aqueous
injection is suggested in EPA Methods 8240 and 8260 and has been characterized by several researchers
(3,4,5,6), but it has limited sensitivity and is prone to artifacts due to nonvolatile compounds dissolved in
water. Therefore, a brief literature survey was conducted to search for alternative analytical approaches
capable of yielding reliable low-level data for water-soluble volatiles.
One alternative that shows promise is microdistillation (7), which has been proposed as Method 5031 in
the third update of SW-846. Microdistillation removes compounds from the sample which either distill
zeotropically at lower boiling points than water (methanol), or that form azeotropes with water that distill
below the boiling point of pure water. In the case of compounds present in low concentrations (<1%), it
has been found that most of the water-soluble volatile organic compounds are present in the initial
milliliter of distillate, resulting in concentration factors of between 70 and 230 for a 40 mL sample (7).
As the distillate condenses, most water-soluble organics remain in the condensate, while the hydrophobic
volatile organics escape into the atmosphere. This eliminates or reduces interferences from
methylsulfides, monoterpenes and chloroform, which are hydrophobic volatile compounds present in some
mill wastewaters. The concentrated condensate can be analyzed by direct aqueous injection GC-FID.
360
-------
NCASI initiated an evaluation of the method as it pertains specifically to influents and effluents of pulp
mill treatment systems. The goal of the project was to determine the accuracy, precision, and sensitivity
for the target analytes. The presence of artifacts, interferents, matrix effects and sample instabilities was
also investigated.
ANALYTICAL OPTIMIZATION
During method evaluation, the chromatographic procedures were optimized. The parameters that were
investigated included the columns (megabore DB-Wax and DB-624), and injection parameters
(temperature, purge time, and rinse solvent). The DB-Wax column was chosen because of key separations
which could not be resolved using the DB-624 column. Target compounds which coeluted using the DB-
624 column were: ethyl ether, ethanol, and acrolein; acetonitrile, 2-propanol, and methyl acetate; and 2-
butanone and propionitrile. Since some of these analytes are known to be present in pulp mill
wastewaters and others were of interest due to lack of data, the DB-624 column was not selected for this
study. The DB-Wax column was not very stable and retained the water until midway through the elution
of target analytes, but it separated a larger number of the target analytes than the DB-624 column. Only
acetonitrile and MIBK were found to coelute using the DB-Wax. Based on the data available from purge
and trap analysis, it seemed unlikely that acetonitrile would be detected in the matrices of interest. The
DB-Wax column was chosen for this study because it was best suited for the target analytes. Other
applications may be better suited by a different column.
The injection of water on the DB-Wax column was found to be critical. Water elutes during the analysis
of the target analytes, rather than before as is common with most solvents. The water acts as a moving
phase on the column, and selectively affects both the retention time and peak width of the target analytes.
The amount of water introduced onto the column was controlled by the injection port temperature and the
purge vent time. While conditions were selected to minimize the amount of water on column, some water
was necessary to resolve some target analytes. Injecting water with the automatic injector periodically
resulted in freezing of the syringe plunger. An organic solvent (dimethyl formamide and, later, ethyl
ether) was added as a second rinse to help lubricate the syringe plunger between analyses. At the end of
the sequence, a methylene chloride rinse was used to keep the syringe from freezing. Table 1 lists the
optimized GC conditions used for the study.
The choice of internal standard for the microdistillation method is critical, because the internal standard is
spiked into the sample before distillation. This resultes in correction for standard recovery. Therefore, it
is important that recovery of the internal standard correlates well with the recovery of target analytes.
During this study 2,2,2-trifluoroethanol, 2,2,2-trichloroethanol, 1-propanol, 2-chloroacetonitrile, methyl
heptafluoropropyl ketone, and hexafluoroethanol were tested as internal standards. 1-Propanol was found
to work well with effluents, but is a native component in some influents. Methyl heptafluoropropyl ketone
and hexafluoroisopropanol were not acceptable because of coelution with methanol and low water
solubility, respectively. 2-Chloroacetonitrile elutes well after the target analytes, and an interferent was
found in one of the three mill influents tested. 2,2,2-Trifluoroethanol eluted in the middle of the
chromatographic run and did not interfere with any target analytes. However, an unidentified interferent
was found to be native in five of the eighteen mill influents tested. 2,2,2-Trichloroethanol eluted after all
the target analytes and no interferent was found for the three mills checked using this standard. As the
method states, each new matrix must be checked for interferences before an internal standard can be used,
and it is wise to calibrate with several so a standby is available when interferences are encountered. 2,2,2-
Trifluoroethanol was used as an internal standard for most of this study, with 2-chloroacetonitrile used as
a backup and surrogate for later studies. 2,2,2-Trichloroethanol is added after distillation and is used as a
recovery standard for the internal standard.
METHOD EVALUATION
Microdistillation was performed as described in Method 5031, with the exceptions described in each
section below. A collection volume of 300 mL was used because of difficulties encountered with 100 mL
361
-------
volume collection. The changes were made either to study the effect of the method parameter (e.g.
condensate volume) or to optimize the procedure (e.g. preservation techniques).
Distillation Recovery - For the microdistillation method to be effective for a particular compound, a
significant percentage of the compound should be distilled and collected in the initial drops of the
condensate. The percent of each target compound distilled was calculated by comparing the distilled
response to standards prepared without distillation and correcting for the internal standard's absolute
recovery. Table2 lists the recoveries found for each analyte compared with those reported by Bruce (7).
The recoveries were similar to those found previously, and the recoveries found for compounds not
previously reported were above 20 percent, except for ethyl ether. Ethyl ether was found to give highly
variable recoveries, and was eliminated from the target analyte list.
Effects of Sample pH on Microdistillation - Untreated pulp mill wastewaters can be quite variable in pH.
Standard preservative techniques for biologically active samples such as pulp mill effluents include pH
adjustments. Some of the target analytes are known to be susceptible to hydrolysis when heated at
extreme pH. Therefore, the effects of pH on microdistillation were investigated. The native
concentrations and matrix spike recoveries were compared for a mill influent analyzed at its native pH of
10.6 and after pH adjustment to 7. The same comparison was made for an influent that was preserved at a
pH of 2. The results of this comparison are shown in Table 3. Methyl acetate and ethyl acetate showed a
large decrease in matrix spike recovery at pH 10.6 with a smaller decrease found at pH 2. Percent
recovery for acrylonitrile and acrolein were found to be lower in the pH 10.6 analysis compared to pH 7,
but were not affected by the lowest pH distillation. Acrolein was not recovered well in the pH 7 analysis of
the pH 10.6 influent. Native concentrations did not show a significant effect with analysis pH, possibly
because compounds that are not stable at extreme pHs are not likely to be found in these influents.
Similar experiments were performed for calibration standards using DI water (pH 5), DI water pH
adjusted to 7 with strong base, and DI water buffered at pHs 4, 7, and 9.2. The acetate and acrolein
recoveries decreased as the pH increased. The levels of methanol and ethanol increased when the acetates
decreased.
To prevent both the loss of target analytes due to hydrolysis and the interference of other target analytes
such as methanol and ethanol, a neutralization step was added to the procedure to bring the pH to between
6 and 7 using diluted NaOH or H2SO4. No neutralization is recommended for preparing standards in DI
water. If alkaline water is used for calibration, a pH adjustment should be made when preparing
standards
Storage and Preservation of Samples • Storage and preservation of effluent and influent samples for
microdistillation are critical because some of the target analytes are known to be chemically and
biologically unstable. Standard volatile collection and storage techniques require storage at 4°C and a
sample free of headspace. For biologically active samples, adjustment to pH 2 is recommended.
Headspace-free sampling can be difficult to accomplish when samples foam and when adjustment to a
specific pH is required. The need for headspace-free samples was investigated, as well as preservation of
samples by several methods.
The effect of headspace was investigated by spiking a pH 2 preserved effluent and storing it at 4°C with
no headspace, with a 10 percent container volume of headspace, and with a 20 percent container volume
of headspace. Each sample was analyzed in duplicate at day one and at day eight. The difference in the
recovery at day eight and day one was found to be smaller than the range in the duplicate determinations,
which indicated no measurable volatile loss during storage. These results indicate that headspace in the
sample for these compounds does not have a significant effect on the analysis.
The need for preservation of samples was demonstrated with a spiked mill effluent stored for 23 days, both
preserved at pH 2 and unpreserved . After 23 days of storage, six analytes could not be detected in the
unpreserved effluent, and six more analytes' recoveries were less than 40 percent Acetone recovery
increased to 159 percent, while only acrylonitrile and propionitrile had recoveries between 70 and 130
362
-------
percent. All but three compounds had recoveries between 70 and 130 percent in effluent preserved at pH
2 for 23 days. Acrolein and ethyl acetate had recoveries of 12 and 69 percent, respectively. Methanol
recovery was 135 percent, but the methanol spike level was 60 percent of the native level, which would
result in a greater variability in the recovery values.
In unpreserved influent after 24 days, three compounds (methyl acetate, acrolein and ethyl acetate) could
not be detected, and acrylonitrile recovery was less than 25 percent. All other analytes' recoveries were
between 70 and 120 percent. In the pH 2 preserved influent stored for 24 days, all analytes' recoveries
were between 70 and 120 percent except acrolein, which was not detected.
Most of the target analytes were not stable in unpreserved effluent, while four of the analytes were not
stable in unpreserved influent. Acrolein was not stable under any storage condition in either matrix. It
was observed that acetone recovery increased with storage in the unpreserved effluent. Both influents and
effluents could be preserved at pH 2 for up to three weeks for all target analytes except acrolein.
Sample Size and Dilution - The concentrations of components found in pulp mill influents and effluents
ranged from nondetect to over 100,000 ug/L for methanol. The distillation and GC-FID method was
found to be linear over the range of 10 to 200,000 ug/L. However, at the highest concentration, column
overloading caused an unacceptable shift in retention times and a loss of compound resolution. A more
practical upper response range would be 20,000 ug/L. Because a typical influent has methanol values
above this range, a practical method for dilution was needed. The approaches investigated were varying
the sample volume, diluting the sample before distilling, and varying the distillate volume. An influent
was spiked with the target analytes and three volumes were analyzed (10, 20, and 40 mL), the sample was
diluted twice (1:1 and 1:3), and 100,200, 300, and 400 uL of distillate were collected.
The volume of sample distilled correlated (r2 >0.95) with the measured concentration for ten of the
compounds. Table 4 shows the results of this study. Of the ten analytes that showed a strong correlation,
five were found to increase as volume increased, and five decreased. Methanol showed the largest change,
with a decrease of 73 percent going from 10 to 40 mL sample size. When samples were diluted and
analyzed at 40 mL each, no change in concentration was found except for methanol, which was above the
practical range of the method (200,000 ug/L).
Distillate volume is difficult to control because drop size is relatively large compared to volume collected.
If distillate volume is a critical factor in the analysis, control of the volume collected must be improved.
During this study a volume of 300 uL was routinely collected, rather than the 100 uL specified in the
method, to minimize fluctuations in the volume collected. To study the effect of distillate volume, spiked
influent was distilled and 100, 200, 300, and 400 uL were collected. A recovery standard was spiked into
the distillate to determine the recovery of internal standard at each volume. No concentration change was
noticed in the results, but the recovery of the internal standard was 30,58, 87, and 96 percent for the 100,
200,300, and 400 uL distillate volumes, respectively. Because internal standard recovery is proportional
to the volume collected, sensitivity of the method does not decrease as distillate volume increases up to a
volume of 300 uL. Above 300 uL, internal standard recovery is no longer proportional with volume
collected, so a dilution of the collected analytes occurs, decreasing the sensitivity of the method. This
study shows that the volume collected is not critical between volumes of 100 and 300 uL, and 300 uL was
chosen for this study because it was easier to collect.
METHOD PERFORMANCE
The optimized method was used to collect data for 18 mills representing a cross-section of the kraft and
sulfite pulp and paper industry. The QA/QC data were compiled for this study to get an indication of the
method performance. Sensitivity was determined for a single effluent using two procedures, the method
detection limit (MDL) as defined in chapter one of SW-846 and a regression-based detection limit
(RBDL) procedure (8). The precision was estimated by pooling duplicate sample analyses for each
matrix. Accuracy was determined from percent recovery of matrix spike analyses.
363
-------
Table 5 shows the detection limits for pulp mill effluent using microdistillation. The MDLs were
determined twice, once during the initial method evaluation and again during an evaluation of the RBDL
procedure. Effluent from the same mill was used, but the actual samples were collected at different times.
Most of the detection limits range from 5 to 50 ng/L, with the higher levels found for the compounds
elating before water. The difference between MDL and RBDL were, in most cases, no greater than a
factor of two.
Precision for the effluent and influent was estimated using the standard deviation from duplicate results.
Some of the values were based on native levels, while others were based on duplicate spikes. Table 6 lists
the relative standard deviations found and the number of duplicate sets used. The relationship between
standard deviation and concentration was determined for one effluent sample during the RBDL study.
Using this relationship, the relative standard deviation (RSD) at any concentration (C) can be calculated
by solving for the equation RSD = (Ae^VC, where A and B are the exponential regression constants
given in Table 6. Precision was estimated to be less than 25 percent for all compounds except acrolein.
Most compounds had RSDs of less than 10 percent, except compounds that eluted before water, whose
RSDs were generally >10 percent
Method accuracy was evaluated by performing matrix spike experiments. Table 7 shows the mean
recovery and standard deviations for effluents and influents. Mean recoveries were found to be within 75
and 125 percent for all target anarytes except acrolein. The standard deviations of the recoveries were
higher for the influents than for the effluents. The majority of the standard deviations in the effluent were
10 percent or less, while most influent standard deviations were above 20 percent. To investigate the
effect of spike concentration on recovery in influent, the results of the RBDL study were used. Multiple
spikes of the same effluent were made at levels covering the linear range of the method. Standard
additions were used to determine the native level and estimate a true concentration. Most of the target
anarytes' recoveries fell between 70 and 130 percent recovery, with no observable trend with
concentration. Five compounds had values that fell outside the 70 to 130 percent recovery range, and they
are plotted in Figure 1. Acrolein shows a trend to lower recoveries at higher concentrations, while the
other four compounds' recoveries are scattered. All five of these compounds elute before the water peak
in the region where chromatographic performance is less than ideal.
SUMMARY
The microdistillation method provides a sensitive, nigged method for measuring water-soluble volatile
organic compounds in pulp mill effluents and influents. Detection limits were found to be in the range of
5 to 50 ug/L, with RSDs of 30 percent or less, and mean recoveries of 75 to 120 percent for the target
anah/tes. Method performance for acrolein and ethyl ether was found to be unacceptable.
Several changes or additions to the method should be made to improve its performance. Samples outside
the pH range of 6 to 7 should be adjusted into this range before distillation to prevent the hydrolysis of
labile compounds, which results in both poor recovery and addition of the hydrolysis products methanol
and ethanol. Samples should be preserved by adjusting the pH to 2.0 and storing at 4°C to prevent both
biological activity and chemical reactivity. Distillation volume has been found to be an important
variable, and the volume of sample distilled should be the same as the volume of calibration standard
distilled. When necessary, samples should be diluted before distillation so that the diluted sample volume
distilled is the same as the calibration standards. Distillate volume collected is not a critical parameter,
and collection volumes between 100 and 300 uL would result in the same sensitivity, while larger volumes
would result in a loss of sensitivity. 2,2,2-Trifluoroethanol was shown to be a good internal standard
when no native interferences were present. All new matrices should be screened before selecting an
internal standard. Alternative internal standards 2-chloroacetonitrile and 2,2,2-trichloroethanol were
shown to be recovered by the distillation process and did not coelute with any target analytes. These
standards have been used as backup internal standards/surrogates and as recovery standards for the
internal standard for a limited number of samples.
364
-------
The major source of variability found with the method was due to chromatographic problems encountered
with compounds which eluted before the solvent, water. Developing a more rugged chromatographic
determination step for the method would provide a more consistent level of data quality for all the
anarytes.
REFERENCES
1. Wilson, D.F. and Hrutfiord, B.F., "SEKOR IV. Formation of Volatile Organic Compounds in
the Kraft Pulping Process," JAPPI, 54, 1095-1098, (1971).
2. Lin, D.P., Falkenberg, C., Payne, D.A., Thakkar, J., and Elly, C., "Kinetics of Purging for the
Priority Volatile Organic Compounds in Water," Anal. Chem. 65,999, (1993).
3. Fox, M.E., "Rapid Gas Chromatographic Method for Determination of Residual Methanol in
Sewage," Environ. Sci. Technol.. 7, 838, (1973)
4. Harris, L.E., Budde, W.L., and Eichelberger, J.W., "Direct Analysis of Water Samples for
Organic Pollutants with Gas Chromatography-Mass Spectrometry," Anal. Chem.. 46, 1912,
(1974).
5. Pfaender, F.K., Jonas, R.B., Stevens, A.A., Moore, L., and Hass, J.R., "Evaluation of Direct
Aqueous Injection Method for Analysis of Chloroform in Drinking Water," Environ. Sci.
Technol.. 12,438, (1978).
6. Gurka, D.F., Pyle, S.M., and Titus, R., "Environmental Analysis by Direct Aqueous Injection,"
Anal. Chem.. 64, 1749, (1992).
7. Bruce, M.L., Lee, R.P., and Stephens, M.W. "Concentration of Water-Soluble Volatile Organic
Compounds from Aqueous Samples by Azeotropic Microdistillation," Environ. Sci. Technol.. 26,
160-163, (1992).
8. Gholson, A., Cook, D., and LaFleur, L., "Application of a Regression Based Detection Limit
Determination for Volatile Water Soluble Compounds in Pulp Mill Effluent using
Microdistillation (SW-846 Method 5031)," presented at EPA's 19th Annual Conference on
Analysis of Pollutants in the Environment, Norfolk, VA, May, (19%).
365
-------
TABLE 1 METHOD CONDITIONS FOR MICROD1ST1LLATTON ANALYSIS
Pinmetcr Setting
Sample Conditions'
Sampktforage pH2*t4°C
Sample vohunc 40 mL
Distillation pH between 6 and 7
Distillate collection volume 300 uL
GC Condition!
Injection On* splhless, at 110°C with 0.35 purge vent
Cotunn DB-Wax,30mxO.S3mmi.d with 1.0 |im film
Carrier He at 4 mL/min
^taen program
Initial 10°C for 1 min
Ratcl 200C/ironto20°C
Rate 2 6°C/minto 120°C
Rate 3 12°C/minto200°C
Detector FID with Na make-up at 250°C
' Unless otherwise specified as part of the experimental plan.
TABLE 2 DISTILLATION RECOVERIES FOR TARGET ANALYTES
Recovery (%)
Compound Measured Literature
Ethyl ether 16.5 na
Aeetaldehyde 21.1 na
Acetone 41.0 na
Methyl acetate 23.7 na
Acrolein 28.7 17
Ethyl acetate 22.6 20
2-Butanone 35.8 na
Methanoi 39.9 35
2-Propanol 46.0 na
Ethanol 45.3 na
AcryJonitrile 27.1 28
MfflK 24.0 na
Propmutrile 37.0 48
1,4-Dknaoe 33.0 38
bobutanol 46.1 58
l-Butand 51.5 58
Cydonexanone 39.1 na
Aoetophcnone 38.5 na
2^2'TrifluorDdhanoIk 48.5 na
' Recoveries reported in reference 7.
' Internal standard
366
-------
TABLE 3 EFFECT OF MICRODISTILLATION PH ON TARGET AN ALYTE IN INFLUENTS
Concentration (mg/L)'
Mill A
Compound
AceUldehyde
Acetone
Methyl acetate
Acrolein
Ethyl acetate
2-Butanone
Methanol
2-Propanol
Ethinol
Aciylonitrile
MIBK
Propioiiitnle
1,4-Dioxane
Isobutanol
1-BuUnol
Cyclohexanone
Acetophenooe
.PH 10.6
2730 (115)
1280 (103)
nd(14)
nd(0)
nd(38)
498 (105)
59200 (89)
254 (84)
2890 (89)
nd(56)
18.2 (122)
nd (Ml)
nd (104)
16.6 (104)
47.3 (1 12)
24.9 (102)
21.4 (90)
_PH7
2430 (67)
1160 (85)
nd(64)
nd(32)
nd(75)
469 (87)
68000 (109)
176 (106)
3020 (103)
nd (80)
17.6 (88)
nd (96)
nd (101)
17.5 (100)
49.4 (112)
27.4 (111)
21.5 (115)
MilIB
PH2
185 (72)
690 (98)
nd(73)
nd(79)
nd (80)
273 (55)
89700 (68)
nd(91)
1710 (73)
nd (108)
nd(93)
46.4 (78)
388 (67)
nd(73)
10.3 (87)
23.9 (62)
IU
PH7
258 (97)
861 (99)
nd (102)
nd(86)
nd (106)
323 (56)
71400 (67)
83.8 (86)
1380 (81)
nd (US)
nd (108)
42.9 (100)
286 (64)
nd(85)
8.7 (84)
19.3 (76)
na
' Matrix spike recoveries as percent are in parenthesis.
nt Compound not analyzed.
nd Compound not detected
TABLE 4 RESULTS OF DISTILLATION VOLUME EXPERIMENT STUDY FOR SPIKED INFLUENT
Concentration (rngfL)
Compound
AceUldehyde
Acetone
Methyl acetate
Acrolein
Ethyl acetate
2-Butanone
Methanol
Isopropanol
Ethanol
Aciylonitrile
MIBK
Propioahnle
1,4-Dioxane
Isobutanol
l-Butanol
Cyclohexanone
Acctophenone
40 mL Sample
2090
3140
2310
1920
2400
2790
134000
2770
4840
2210
2460
2320
2890
2550
2690
2550
2520
20 mL Sample'
1840 (7.7)
2840 (3.8)
1960 (3.2)
1700 (6.2)
1980 (0.1)
2350 (5.3)
195000 (14)
2810 (1.5)
5950 (16)
1870 (3.0)
1990 (5.4)
2170 (3.6)
4050 (3.0)
2470 (5.4)
3800 (6.1)
2970 (7.1)
2920 (6.6)
10 mL Sample
1940
2840
1940
1690
1850
2280
232000
3210
6120
1810
1800
2140
4420
2620
2920
3100
3010
' Mean value of duplicate with the relative percent difference in parenthesis.
367
-------
Compound
AceUkkhyde
Acetone
Methyl acetate
Acroletn
Ethyl acetate
2-Butanonc
Mcthanol
bopropAnol
Ethanol
AcryknitrUe
MIBK
Propionilrile
1-ProfNmol
1,4-Diosne
Isobutanol
1-Butanol
Cydohemone
Acetophenone
MDL1
(mg/L)
20.3
11.4
17.7
14.5
14.2
7.0
26.6
9.4
16.0
20.0
15.2
4.6
na
14.9
3.9'
8.5
8.9
na
MDLk
91.0
34.1
17.2
28.3
21.1
6.9
na
8.0
8.0
3.7
9.3
5.9
4.1
9.7
5.4
4.6
3.4
4.8
RBDL'
45.4
26.6
35.7
na
24.8
6.9
na
5.9
7.8
7.6
7.8
6.1
5.1
12.8
5.8
6.2
6.2
8.6
' Method detection limit as defined in SW-846 determined during method evaluation study.
' Method detection limit as defined in SW-846 determined during the regression based
detection timil study.
' Regression based detection limit as described in reference 8.
' Mean concentration was peater than five times the MDL.
na n not available.
TABLE 6 METHOD PRECISION ESTIMATES FROM DUPLICATE RESULTS
Effluent precision Influent precision
Compound
Acetaldehyde
Acetone
Methyl acetate
Acrolem
Ethyl acetate
2*Butanone
Metfaanol
Ixopropanol
Ethanol
Acryknitnle
MIBK
1% " "i *!_•
1,4-Dioxac*
Isobutanol
I-Butanol
PveMiennm
Acetophenone
USD1
21.5
11.4
13.0
na
11.6
5.3
21.7
2.3 '
4.4
5.7
9.1
3«
.j
10.0
2.4
3.9
6.5
8.0
n'
7
8
4
na
4
4
9
4
4
4
4
4
4
4
4
5
Ac
19.3
8.79
5.62
6.50
6.40
1.85
na
1.52
2.64
1.94
2.32
i
-------
3LE7 MATRIX SPIKE RECOVERY DATA
Effluent Recovery
Compound
Acetaldehyde
Acelone
Methyl acetate
Acrolein
Ethyl acetate
2-But»none
Methanol
Isopropanol
F.thanol
Acrylonitrile
MIBK
Propionitrile
1,4-Dioxane
hobutanol
1 -Butanol
Cyclohexanone
Acetophenone
Mean
86.9
93.4
75.8
65.7
83.1
89.2
105
105
112
87.8
101
91.5
99.6
98.4
102
102
102
SI)1
5.8
9.3
29.J
36.0
23.3
7.2
19.4
11.5
8.7
9.0
10.0
7.7
SO
8.2
10.1
11.0
2.7
Nk
7
7
7
7
7
7
6
7
6
6
7
7
7
7
7
7
2
Mean
99.0
89.4
77.6
32.5
998
85.6
100
83.0
88.4
77.9
124
98.4
85.1
94.9
928
96 1
105
Influent Recovery
SD1
21.1
22.8
26.1
31.4
21.7
24.2
32.7
36.3
19.1
28.1
35.8
9.0
27.3
15.1
26.1
19.2
81
N"
12
11
12
12
11
12
11
11
11
10
12
12
12
12
12
12
(,
Standard deviation expressed H a percent.
Number of recovery determinations.
130
Elhanol
2-Butanone
Acetaldchyde
Methyl acetate
10
100
Concentration (pg/L)
1000
Figure 1 Percent recovery versus concentration for microdislillation for pulp mill effluent
369
-------
78
Extraction of Poly chlorinated Dibenzo-p-Dioxins and
Polychlorinated Dibenzo-Furans from Environmental
Samples Using Accelerated Solvent Extraction (ASE)
B. E. Richter. J. L. Ezzell, D. E. Knowles, and F. Hoefler
Dionex Corporation, Salt Lake City Technical Center, 1515 W. 2200 S.,
Suite A, Salt Lake City, Utah 84119
A. K. R. Mattulat and M. Scheutwinkel
Dr. Scheutwinkel GmbH, Berlin, Germany
D. S. Waddell, T. Gobran and V. Khurana
Ontario Ministry of Environment and Energy, 125 Resources Road,
Etobicoke, Ontario, Canada, M9P 3V6
Method 3545 is a new extraction method that utilizes accelerated solvent
extraction (ASE) to perform rapid and efficient extractions of analytes from
solid and semi-solid matrices. The method applies to the following
analytes: organochlorine pesticides (OCPs), polychlorinated biphenyls
(PCBs), semivolatiles (BNAs), organophosphorus pesticides (OPPs),
polycyclic aromatic hydrocarbons (PAHs), and chlorinated herbicides. The
only extraction techniques that have applicability and acceptance as
broad as ASE are Soxhlet, automated Soxhlet and sonication.
ASE applies temperature and pressure to accelerate extraction processes
and improve the efficiency of solvent extraction. Sample extractions can
be done faster and with less solvent than by current methods. Since
organic solvents are used in ASE, method development can be very rapid,
and no significant matrix effect is seen with ASE. This paper reports on a
study to compare the results obtained with Soxhlet to those obtained by
ASE for various samples containing polychlorinated dibenzo-p-dioxins
(PCDDs) and polychlorinated dibenzo-furans (PCDFs). Contaminated
samples (soil, sediments, chimney brick, urban dust and fly ash) were
extracted by ASE and Soxhlet. A review of the data indicates that ASE
gives essentially equivalent data to Soxhlet extraction. However, ASE
extractions are performed in less time and with less solvent than by the
classical extraction techniques (about 15 mL and less than 25 min for 10-
g samples by ASE as compared to 250 mL and 18 hours for Soxhlet).
370
-------
79
A CONFIRMATORY HOLDING TIME STUDY FOR PURGEABLE VOCS EV
WATER SAMPLES
O. R. West, C. K. Bayne, R. L. Siegrist, W. H. Holden, Oak Ridge National Laboratory,
Oak Ridge, TN, 37831; D. W. Bortrell Department of Energy, Washington, DC, 20874
Abstract
Analyte stability during pre-analytical storage is essential to the accurate quantification of
contaminants in environmental samples. This is particularly true for volatile organic
compounds (VOCs) which can easily volatilize and/or degrade during sample storage.
Recognizing this, regulatory agencies require water samples be collected in vials without
headspace and stored at 4°C, and that analyses be conducted within 14 days, even if
samples are acid-preserved. Since the selection of a 14-day holding time was largely
arbitrary, the appropriateness of this requirement must be re-evaluated. The goal of the
study described here was to provide regulatory agencies with the necessary data to extend
the maximum holding time for properly preserved VOC water samples to 28 days. An
extensive stability experiment was performed on freshly-collected surface water spiked
with a suite of 44 purgeable VOCs. The samples were contained in 40-mL glass vials
with no headspace, preserved with 250 mg of NaHSO4 and stored at 4°C. For a majority
of the 44 VOCs included in this study, concentration changes were <10% of the initial
values after 28 days of storage. Maximum holding times calculated from the stability data
using the "practical reporting time" approach [Bayne et al., 1994] were predominantly
greater than 28 days. This study showed that a 28-day holding time for properly
preserved VOC water samples would not jeopardize the measurement of target VOCs.
This holding time extension would benefit the regulated community, particularly
government agencies with large-scale compliance sampling programs such as the
Department of Defense and Department of Energy. The suggested modification of
holding times can also improve the efficiency of commercial laboratories through
simplified sample management. Application of this study's results to data review would
also improve the analytical data validation process by providing an alternative to the
currently "one-size-fits-all" accept or reject approach that is very costly but not
technically defensible.
Introduction
Analyte stability during pre-analytical storage is essential to the accurate quantification of
contaminant levels in environmental samples. This is particularly true for volatile organic
compound (VOC) analysis, since some of these target analytes can volatilize and/or
degrade during sample storage. To reduce the impact of these transformation mechanisms
on VOC analyses, regulatory agencies require that water samples be collected without
headspace in 40-mL vials with Teflon-lined septum caps, acidified to pH~2, and stored at
4°C. Furthermore, analytical data are considered valid only if the analyses are conducted
within 14 days of sample collection (7 days if samples are not acidified). This maximum
371
-------
holding time was arbitrarily set and specified in 40CFR Part 136 [1979], and has since
been adopted by other regulatory programs and for application to other environmental
media [40CFR Part 136, 1984; EPA, 1986]. The appropriateness of this requirement
must be re-evaluated since compliance with this 14-day holding time can and has been
difficult and costly for sample collectors, data users and analytical laboratories. Recent
Superfund Guidance [EPA, 1994] attempts to address the problem by relying on data
validators' judgment to assess the impact of missed holding times on analytical
measurements. However, this has still led to unequivocal rejection of data when
prescribed holding times are missed, and more specific guidance backed by scientific data
is needed from regulatory agencies [Bottrell, 1995].
Previous stability studies [Maskarinec et al., 1989; Bottrell et al., 1989] have
demonstrated that a majority of purgeable volatile organic compounds in properly
preserved VOC water samples (acidified, no headspace, 4°C storage) are stable for time
periods well over 14 days. The goal of this study is to confirm these previous studies, as
well as to provide regulatory agencies with the necessary data to extend the maximum
holding time for properly preserved VOC water samples to 28 days
Experimental Methods
The stability study was performed on surface water collected from a tributary of the
Clinch River in Oak Ridge, TN. Water samples were prepared following the procedure
described by Maskarinec et al. [1989]. A clean 3-L Tedlar bag was filled with two liters
of surface water. Measured aliquots of VOC standard solutions were injected into the
water-filled Tedlar bag through the bag's septum port. The water-filled Tedlar bag was
shaken for 1-min, and allowed to equilibrate at room temperature for 20 min. After
equilibration, the spiked water was distributed into an appropriate number of pre-cleaned
40-mL VOA vials with Teflon-lined (0.010-in thick) silicone septum caps. Two hundred-
fifty milligrams of NaHSO4 were placed in each vial prior to filling. Each vial was
completely filled (i.e., with no headspace) and stored at 4°C prior to analysis. Two sets
of samples were prepared, one set was spiked to 20-ppb (Wl) while a second set was
spiked to 200-ppb (W2). At 1, 8, 15, 22, 29, 35, and 71 days after sample preparation,
four samples from each set were'analyzed for VOCs. Analyses followed the purge-and-
trap (PT) method in SW846-8260A [EPA, 1986], except that all calibrations were
performed in reagent water that had been acidified to pH~2 with reagent-grade NaHSO4
(resulting pH was 2.4-2.6). Further experimental details are given in a forthcoming report
and publications [West et al., 1996].
Results
Regression lines were fitted to the data for concentration vs analysis day (see Table 1 for
a select number of compounds). Measurement variability [i.e., relative measurement error
(%RME)] for each compound within each sample set was estimated as follows:
372
-------
%MfE-5-xl00%. (1)
c
o
where S0 is the square root of the mean-square error for the linear regression residuals, and
C0 is the extrapolated concentration on Day 0. Calculated values for %RME were
predominantly less than 15% in both sets Wl and W2, with values being lower in set Wl
(20 ppb spike). These low values indicate that scatter in the data was generally minimal,
and that concentration trends with time were less likely masked by measurement
variability. Such masking of concentration trends may have occurred with vinyl chloride,
which exhibited the highest %RME in both sample sets (43% and 33%). Statistically
insignificant changes in vinyl chloride concentration with time may have been due to large
measurement variability. However, a statistically significant negative slope was observed
in vinyl acetate in both sets Wl and W2, despite a relatively large %RME (21% and
29%). For this compound, the concentration change with time was large enough to offset
the masking effects of data scatter.
Changes in concentration after 28 days of storage (last column in Table 1) were
predominantly low relative to the initial concentrations. Out of 44 analytes,
concentration changes exceeded 10% for only three compounds in set Wl [vinyl acetate
(42%), cis-l,3-dichloropropene (11%), and trans-l,4-dichloro-2-butene (20%)].
Concentration changes exceeded 10% for a larger number of compounds in set W2,
including trichlorofluoromethane (15%), acrolein (25%), carbon disulfide (17%), vinyl
acetate (42%), cis-J,3-dichloropropene (14%), trans-I,3-dichloropropene (12%),
tetrachloroethene (16%), and trans-l,2-dichloro-2-butene (26%). For compounds
exhibiting non-significant slopes (Table 2), maximum holding times (MHTs) were set to
71 days, i.e., the duration of the stability experiment. For compounds with significantly
negative slopes, MHTs were calculated from the stability data using "practical report
time" analysis [Bayne et al. 1994]. Results of the analysis are presented in Table 2;
details of the analysis are given in [West et al. 1996]. The compounds were subdivided
into 3 groups (see Table 2): (1) VOCs which have MHTs greater than 28 days (Group 1),
(2) VOCs which have MHTs less than 28 days but for which the relative change in
concentration on the 28th day was <10% of the initial value, and (3) VOCs which have
MHTs less than 28 days and the relative change in concentration on the 28th day was
373
-------
Table 1,
Summary of linear regression on stability data for select compounds.
No. of
data
points
SAMPLE SET Wl
Acetone
Benzene
Caibon disulfide
Carton tetrachloride
Chlorobenzene
Chloroform
DichJoroethene, trans-1,2-
Methylcne chloride
Pentanone,4-melhyl-2-
Styrcne
Tetrachloroethene
Toluene
Trichloroelhane, 1,1,1-
Trichloroethene
Vinyl acetate
Vinyl chloride
Xylene, m.p-
Xytene, o-
SAMPLE SET W2
Acetone
Benzene
Carbon disulfide
Carbon tetrachloride
Chlorobenzene
Chloroform
DichloFoethene, trans- 1,2-
Methylene chloride
Penlanone,4-methyl-2-
Styrene
Tetrachloroethene
Toluene
Trichloroethane, 1,1,1-
Trichloroethene
Vinyl acetate
Vinyl chloride
Xylene, m,p-
Xylene. o-
16
28
28
28
28
28
28
28
24
28
28
28
28
28
20
24
28
28
16
27
27
27
27
27
27
27
23
27
27
27
27
27
19
23
27
27
Regression parameters
Intercept Slope
(ppb) (ppb/dav)
21.2
16.1
15.2
16.6
16.8
18.0
15.6
17.5
18.3
17.7
18.1
15.2
15.6
16.9
23.5
10.9
37.4
18.7
194.5
177.7
200.8
199.0
182.0
194.0
176.5
179.3
210.6
174.7
166.5
178.6
182.3
193.7
235.4
144.0
343.6
172.2
0.0124
0.0033
-0.0501
-0.0168
-0.0134
0.0056
-0.0211
-0.0007
0.0082
0.0099
-0.0654
-0.0005
0.0034
-0.0103
-0.3486
0.0915
-0.0752
-0.0268
-0.2472
-0.2523
-1.2258
-0.6786
-0.3897
-0.3069
-0.5698
-0.1591
-0.0675
-0.6391
-0.9265
-0.1996
-0.5152
-0.5018
-3.5267
-0.0040
-1.1289
-0.4257
Significant
Negative
Slope
(1 -sided 5%
significance)
NO
NO
YES
YES
YES
NO
YES
NO
NO
NO
YES
NO
NO
NO
YES
NO
YES
NO
NO
NO
YES
YES
YES
YES
YES
NO
NO
YES
YES
NO
YES
YES
YES
NO
YES
YES
Relative
Meas. Error
10%
5%
6%
6%
5%
6%
5%
5%
7%
12%
9%
6%
8%
5%
21%
43%
10%
11%
9%
10%
8%
13%
9%
10%
9%
9%
5%
5%
8%
16%
12%
11%
19%
33%
7%
6%
Change in
cone, at 28
days
(ppb)
—
—
-1.4
-0.5
-0.4
—
-0.6
-
-
-
-1.8
—
-
—
-9.8
-
-2.1
-
_
—
-34.3
-19
-10.9
-8.6
-16
-
-
-17.9
-25.9
-
-14.4
-14.1
-98.7
-
-31.6
-11.9
%Change in
cone, at 28
days
—
—
-9%
-3%
-2%
—
•4%
-
-
-
-10%
—
-
—
-42%
-
•6%
-
.
-
-17%
-10%
-6%
•4%
-9%
-
-
-10%
-16%
-
-8%
-7%
-12%
-
-9%
-7%
374
-------
Table 2. Maximum holding times for compounds in sample sets Wl and W2. Also
includes relative change in concentration after 28 days of storage.
Sample Set Wl f20ppb Sample Set W2 (200 ppb~
spike) spike>
Compound Max. %Change in Max. holding %Change in
holding time cone, at 28 time (days) cone, at 28
(days) days days
Group 1: Max, holding time greater than or equal to 28 days
Acetone 71 .71
Acrylonitrile 71 - 71 -
Benzene 71 - 71
Bromomethane 71 - 71 -
2-Butanone 71 - 71
Dichloroethane, 1,1- 71 71
Dichloroethane,l,2- 71 71
Dichloroethene, 1,1- 71 71
Dichloropropane, 1,2- 71 - 71 -
Dibtomomethane 71 - 71 -
Pentanone, 4-methyl-2- 71 - 71
Tetrachloroethane, 1,1,2,2- 71 - 71
Toluene 71 - 71 -
Trichloroethane, 1,1,2- 71 71
Vinyl chloride 71 - 71
Bromodichloromethane 71 - 45 -
Carbon tetrachloride 43 - 28
Chlorobenzene 40 - 28
Chloroethane 71 - 34
Chloroform 71 - 44
Trichloroethane, 1,1,1- 71 - 30
Trichloroethene 71 - 31
Methyl iodide 71 -37
Group 2: A .far, holding time less 28 davs and %change at 28 davs < 10%
Chloromethane 25 -5% 22 -10%
Bromomethane 25 -4% 20 -9%
1,1-Dichloroethene 30 - 23 -10%
trans-U-Dichloroethene 24 -4% 20 -9%
Ethylbenzene 40 - 25 -8%
Dibromochloromethane 71 - 14 -6%
m,p-Xylene 36 - 14 -9%
o-Xylene 71 - 17 -7%
Styrcne 71 - 9 -10%
Bromoform 71 - 15 -6%
2-Hexanone 71 - 7 -9%
1.2,3-Tricnloropropane 71 - 12 -6%
Cirnun 3- \far holdine time less than 28 davs and %chanee at 28 davs >
Trichlorofluoromethane
Acrolein
Carbon disulfide
Vinyl acetate
cis-1 ,3 -Dichloropropene
trans- 1 ,3-Dichloropropene
Tctrachloroethene
trans- 1.4-Dichloro-2-butene
16
—
13
10
12
17
17
16
-8%
—
-9%
-42%
-11%
-9%
-10%
-20%
14
4
9
9
13
16
9
3
-15%
-25%
-17%
-12%
-14%
-12%
-16%
-26%
375
-------
Calculated MHTs were very short for some analytes with very low measurement
variability, even though concentration changes on the 28th day were relatively small (e.g.,
styrene in set W2: %RME = 5.0%, MHT= 9.4 days, %change on the 28th day relative to
initial concentration = -10%). In such cases, factors other than calculated MHTs should
be considered when assessing the effects of holding times on measurement validity.
Statistical definitions of significant concentration change, such as the practical report time
approach [Bayne et al., 1994], must be complemented with "practical" definitions of
"acceptable" concentration change. Ideally, specifications for "acceptable" concentration
changes should be tied into the eventual use of the analytical data. For example,
"acceptable" concentration changes for analytical data used in quantitative risk
assessments can be determined by the sensitivity of the risk assessment results to
variations in the input analytical data. Since, the selection of a generic "acceptable"
concentration change was beyond the scope of this study, 10% was chosen as a
reasonable value to assess the holding time effects on analyses.
Based on calculated MHTs and an "acceptable" concentration change of 10% for low-
variability analytes, the stability study showed that the measurement of 36 out of 44
purgeable VOCs in properly preserved water samples will not be affected by sample
storage for 28 days. Larger changes in concentration (>10%) and low MHTs were
observed for a few analytes (see Group 3 in Table 2). However, additional analytical
problems for some of the latter compounds exist (e.g., inconsistent purging) which can
confound the analytical process and which can not be addressed by restricting maximum
holding times alone.
Summary
This study demonstrates that a 28-day holding time for properly preserved water
samples would not jeopardize the measurement of VOCs. This holding time extension
would benefit the regulated community, particularly government agencies with large-scale
compliance sampling programs such as the Department of Defense and Department of
Energy. Stringent holding times result in logistical difficulties further complicated by
additional requirements for sample screening (e.g., for radioactivity). The suggested
modification of holding times can also improve the sample through-put of commercial
laboratories through simplified sample management. Application to data review of the
database generated by this study would also improve the analytical data validation
process by providing an alternative to the currently "one-size-fits-aH" accept or reject
approach that is very costly but not technically defensible.
This study also demonstrated a methodology for conducting a stability study and
practical reporting time analysis of the stability data. The latter approach would be
useful for establishing site-specific maximum holding times which, depending on the
compounds of interest, can be longer than 28 days.
376
-------
Acknowledgments
The authors would like to acknowledge the following who contributed towards the
successful completion of this work: Barry Lesnik, Ray Bath, and Alan Hewitt for
reviewing our experimental plan; EPA Region VI for conducting the data validation; EPA
Region IV for participating in the holding time study; and Analytical Resources Inc. for
providing high-quality and timely VOC analyses. This project was funded by the
Department of Energy's Analytical Programs Office.
References
[1] Bayne, C. K.; Schmoyer, D. D.; Jenkins, R. A. 1994. Practical Reporting Times for
Environmental Samples. Environ. Sci. Technol. 28, 1430-1436.
[2] Bottrell, D. W.;Fisk, J. F.; Dempsey, C. 1989. Pre-analytical Holding Time Study -
Volatiles in Water. Proceedings Fifth Annual Symposium on Solid Waste Testing and
Quality Assurance. U.S. Environmental Protection Agency. 11:24.
[3] Bottrell, D. W. Suggested Modification of Pre-analytical Holding Times - Volatile
Organics in Water Samples. Proceedings Eleventh Annual Waste Testing and Quality
Assurance Symposium. American Chemical Society and U.S. Environmental Protection
Agency. 507-516.
[4] Federal Register. 1979. 40CFR Part 136. Proposed Rules, Vol. 44, No. 233: 69534.
Dec. 3.
[5] Federal Register. 1984. 40CFR Part 136. Rules and Regulations. Vol. 49, No. 209:
145. Oct. 26.
[6] Maskarinec, M. P.; Bayne, C. K.; Johnson, L. H.; Holladay, S. K.; Jenkins, R. A.
1989. Stability of Volatile Organics in Environmental Water Samples. Storage and
Preservation. ORNL/TM-11300. Oak Ridge National Laboratory, Oak Ridge, TN.
[7] U.S. Environmental Protection Agency. Test Methods for Evaluating Solid Waste,
3rd ed.; SW-846; U.S. EPA: Washington, DC, 1986. (Method 8260-A, Revision 1,
1994).
[8] U. S. Environmental Protection Agency. 1994. Contract Laboratory Program
Functional Guidelines for Organic Data Review, EPA-540/R-94/012. U.S. Environmental
Protection Agency, Office of Solid Waste and Emergency Response. Washington, D.C.
[9] West, O R.; Bayne, C. K.; Holden, W. L.; Siegrist, R. L.; Scarborough, S. S.; Bottrell,
D. W. 1996. A Confirmatory Holding Time Study for Purgeable VOCs in Water
Samples. ORNL/TM-13240. Oak Ridge National Laboratory, Oak Ridge, TN.
[10] West, O. R.; Bayne, C. K.; Holden, W. L.; Siegrist, R. L.; Bottrell, D. W. 1996.
The Stability of Volatile Organic Compounds in Water Samples, (manuscript in
preparation).
377
-------
80
How to Modify Existing Methods for New Applications.
B. Lesnik
378
-------
81
What To Expect From Immunoassay Methods
And How To Obtain Their Benefits
Stephen B. Friedman, Ph.D.
President
Sylvanus Environmental, Inc.
P.O. Box 897
Pittsboro, NC 27312
Abstract
Immunoassay testing can significantly reduce the cost and improve the overall
quality of environmental projects when they are integrated into a
comprehensive program that includes analytical testing. Unlike traditional
methods they are able to generate reliable data within hours of sample
collection.
Claims regarding immunoassay simplicity, appropriate applications and
general performance vary considerably. Those interested in using these tests
must be willing to address the practical issues of labor, net cost calculations,
training and product selection. A specialized immunoassay testing center has
recently been established (Sylvanus Environmental, Pittsboro, NC) that
provides the public a convenient way to access their benefits. Whether through
self testing, or the services of a specialized laboratory, immunoassay methods
can become the cornerstone of an efficient, state of the art, testing program.
Introduction
More than 50 compound-selective immunoassay methods have been developed
by a number of manufacturer's. These tests are used to detect chemicals (e.g.
PCB's, PCP, PAH's, TPH, pesticides, herbicides) that may be present in a variety
of matrixes (e.g. soil, water, oil, sludge, surfaces). They provide data within
hours of sample collection, and because they are portable, can be performed
anywhere. They become the method of choice once the identity of a
compound(s) has been established.
[mmunoassay testing is used to improve the efficiency of site characterization
(i.e. mapping), remediation and monitoring activities. They provide the data
needed to produce a comprehensive map of a contaminated area rapidly. They
can be used to help track a plume of contamination, direct excavation activities,
assist in the design of an effective treatment process and monitor the treated
waste. They provide the real time data needed to guide drilling activities and
are used for the influent/effluent testing associated with drinking water and
379
-------
industrial wastewater regulations. Once a compound has been identified
testing using immunoassay methods if often beneficial. They promote overall
project efficiency by providing the rapid data needed to keep field crews and
equipment continuously, and productively, occupied. Their lower cost
provides more data and improved quality.
Immunoassay vs. Analytical
cost vs. response time
$/Test
d Analytical
Q Immunoassay
2 24 240
Response Time (hours)
Obtaining Immunoassav Benefits
The acceptance of immunoassay methods by the EPA began five year ago when
OSW accepted an immunoassay for the detection of pentachlorophenol in water
(Method 4010). A comprehensive program to review subsequent method
submissions considered beneficial to RCRA activities has also been
implemented. Developers submit documentation relating to the regulatory
application of a method and justify performance claims with supporting
internal validation and external field trial information. The time and resources
required to complete the process are considerable, usually taking several years.
The quality and reliability of the methods accepted has been well established.
The performance of an EPA-accepted immunoassay method should be
consistent with the claims made by the manufacturer in the package insert.
EPA acceptance, however, does not suggest that competing products perform
equivalently. Different "EPA-accepted " immunoassay methods for the same
compound(s) exist (e.g. EnSys RISc PCB, Ohmicron RaPID PCB, EM Science
DTech PCB, Milhpore Envirogard PCB). They each contain different, and
usually proprietary, reagents (i.e. antibody, conjugate, buffers, substrates, etc.)
that influence the tests overall sensitivity, specificity, interference's and general
performance. An evaluation of the claims made for a method will reveal
performance variations that can determine the one that will work best for any
given project.
380
-------
Quantitation and Specificity
Immunoassay methods are by nature quantitative procedures that can
accurately and reproducibly quantify substances in a sample. Most of the
available tests produce a signal that is inversely, and logarithmically, related to
the concentration of the target compound. For example, the higher the PCB
concentration in a sample the lower will be the intensity of color produced by
the PCB test. By testing a series of reference solutions (i.e. standards), a dose-
response curve of optical density (plotted on the ordinate) vs. concentration
(plotted on the abscissa) can be developed and used to quantify the
concentration of a sample. As an alternative, a single standard containing the
compound at, or below, the action level can be tested and used as a comparative
reference for the samples. The latter version does not provide the quantitative
data of the former, but simply a "contaminated/not contaminated"
interpretation at the action level of interest.
The sensitivity of an immunoassay is a function of the binding constant of the
antibody population used. The maximum binding affinity of an antibody is
approximately 10nL/M. The limit of sensitivity for a competitive
immunoassay method (i.e. without ligand concentration) is a function of the
mass action equation and limited to approximately 10 M M. Environmental
immunoassay methods, because of the relatively small size of the molecules
being detected, rarely attain a sensitivity greater than 10'8 - 1O9 M without prior
concentration.
The compound detection profile of an immunoassay is also a function of the
antibody used. Each manufacturer uses a proprietary reagent that provides
performance unique to their product. The following chart illustrates the
comparative detection characteristics of EPA method 4035 for PAH. A
knowledge of the target compound and coexisting compound(s) at the site,
action level(s) and comparative product performance will help ensure the
selection of an appropriate method.
381
-------
Comparative PAH Immunoassay Recognition
Method 4035
350.0
300.0
100.0
o
Comparstlvite PAH Immunoassay Recognition
Method 4035
- 2500.0
2000.0
i 1 500.0
u
f
I
JL
«
I
£
o
I
382
-------
Immunoassay methods detect substances by means of an antibody binding
reaction. Antibody binding can be either highly specific or selective to varying
degrees. For example, an immunoassay that uses an antibody that specifically
binds xylene(s) will reliably detect xylene to the exclusion of other compounds.
An immunoassay using an antibody that detects both xylene(s) and toluene will
detect both to differing degrees. If the intent is to specifically quantify xylene,
the presence of toluene in the sample will cause a false positive response. The
greater the recognition of toluene by the antibody, and the higher the
concentration of toluene in the sample, the greater will be this incidence. On
the other hand, if toluene is not present at the site the issue of cross reactivity
becomes inconsequential and the immunoassay will provide the intended
benefit.
Immunoassay methods are specific detection tools. They are used to detect
either a single, or structurally-related group, of compounds. They differ from
chromatographic methods that can fractionate, identify and quantify a variety
of chemicals in a single sample simultaneously. Special care must be taken
when using an immunoassay method to detect a family of molecules (e.g.
PAH's, PCB's). Immunoassay's for PCB detection do not target a single
congenor, but instead recognize the mixture of congeners found in the more
highly chlorinated aroclor preparations. The sensitivity of a PCB immunoassay
for different aroclors may vary significantly. If a specific aroclor is identified at
a site, and other cross-reacting aroclors are absent, the assay can be calibrated
appropriately. Similarly weathered aroclor samples can be analyzed if field
references are incorporated.
The recognition of different polyaromatic compounds by a PAH immunoassay
will also vary. These tests predominantly target the compounds having 3, 4
and 5 rings. They use this collective group of molecules as an indicator of
total PAH contamination at a site. If the intent is to use a test to quantify a
single PAH compound, reliable data can only be produced if the target
compound is the only recognizable specie present in the sample, the assay can
be effectively blanked for other detectable species and the test is calibrated
appropriately.
Each of the group-specific immunoassay methods (i.e. PCB, PAH, TPH) have a
unique detection profile. They need to be calibrated appropriately for the
intended target. Positive field samples and field blanks, should be integrated
into the testing routine whenever possible. Immunoassay methods from
different manufacturer's will differ in their specificity, sensitivity and general
performance characteristics.
383
-------
Obtaining Immunoassay Benefits
Unlike the medical industry, where hundreds of different immunoassay tests
are used to detect disease, substance abuse and pregnancy within a $20B/year
industry, the amount of environmental immunoassay testing being currently
performed is relatively low (i.e. appx. 500,000 tests in 1995). In part this reflects
the different ages of the two industries. The medical immunoassay industry
has matured over the better part of the twentieth century to a point where
immunoassay products are routinely used. Environmental immunoassay
technology, for the most part, has emerged this decade. A look at the
evolution of the medical diagnostics industry, however, may provide insight
into our future with this technology and guidance on the best way to obtain
their benefits.
Immunoassay testing for medical applications is performed by trained
laboratory professionals and the general public. Hundreds of millions of
immunoassay tests are performed each year, but the Food and Drug
Administration (FDA) allows the public to perform less than 5% of the methods
available (i.e. pregnancy, streptococcus). The FDA mandates that the vast
majority of immunoassay testing be performed by experienced professionals.
Commercial laboratories, hospitals and doctors office laboratories provide the
public rapid, reliable, cost-effective immunoassay data.
While environmental immunoassay methods are less complicated and more
portable than analytical testing, claims that infer simplicity comparable to a
"home pregnancy test" are exaggerated. Sample matrix heterogeneity dictate
that environmental immunoassay tests contain protocols to process aqueous,
solid, oily, and mixed wastes. Differing state and federal action levels require
that dilution steps be added to address variations in regional regulations.
Consider the additional complexity of a pregnancy test if soil had to be
analyzed, or if each state had a different definition for pregnancy.
Immunoassay tests contain many components and use protocols that contain
numerous steps. Some are sensitive to time, technique and temperature. The
user of an immunoassay method must have a basic understanding of the
elements that can influence performance. Training is necessary and proficiency
should be demonstrated. For those considering the use of a quantitative
immunoassay method the statistical conventions applicable to any analytical
procedure, and those unique to immunoassay methods, should be understood
and incorporated.
The use of a specialized immunoassay testing center (e.g. Sylvanus
Environmental, Pittsboro, NC) can simplify the process of obtaining rapid,
reliable, cost-effective data. Specialist select and perform the test most
appropriate for a project and integrate their benefits into a program that
includes the appropriate use of analytical methods. The result of an integrated
384
-------
program of innovative and traditional testing is the significant reduction in the
time and cost required to complete a project. Trained personnel can provide
data on-site within hours of sample collection or produce data from a
centralized laboratory within a day. The emergence of an immunoassay
laboratory removes many of the obstacles that have complicated their use and
provides a convenient, practical, way for individuals and corporations to obtain
their benefits.
Summary
Immunoassay testing can benefit most programs. On-site testing can benefit
projects in remote locations or those that need results within hours.
Individuals who perform these methods should become familiar with the
technology, proficient with the protocols and review the product claims in
order to select the best testing method. The use of a specialized immunoassay
laboratory should be considered as a convenient, cost-effective way to gain the
benefits of this valuable new technology.
385
-------
82
Method 4670: A Quantitative Immunoassay Method for the Determination
of Triazine Herbicides as Atrazine
Harry B. McCartv. Ph. P.. Senior Scientist, Environmental and Health Sciences Group, Science Applications
International Corporation, 11251 Roger Bacon Drive, Reston, VA, 22190; Baldev Bathija, Ph. D., Environmental
Scientist, Drinking Water Standards Division, Office of Water; and Barry Lesnik, Chemist, Methods Section,
Office of Solid Waste, USEPA, 401 M Street, SW, Washington, DC 22046
While immunochemical reactions have been employed in the medical testing community for more than 20
years, recent efforts to expand the utility of immunoassay techniques into environmental testing have met with mixed
success. Problems included lack of testing products for environmental contaminants of interest, lack of compound
specificity for some testing products, and product formats that yielded semi-quantitative, or screening, results.
Manufacturers of immunochemical testing products have worked on resolving these issues, resulting in the first set
of EPA immunoassay methods proposed in the SW-846 methods manual from the Office of Solid Waste. These
methods, the "4000 Series* methods, are specifically designed as semi-quantitative screening procedures.
The EPA Office of Ground Water and Drinking Water (OGWDW), in conjunction with the Office of Solid
Waste (OSW), are about to propose an immunoassay method for Atrazine for use in monitoring compliance with
the Safe Drinking Water Act. This method, Method 4670, represents several "firsts' for both OGWDW and OSW.
It is UK first truly cooperative methods development effort between these two EPA Offices and it represents the first
truly quantitative immunoassay method to be proposed by EPA.
The adaptation of a screening immunoassay for quantitative determinations involves a variety of efforts.
The testing product developers have to develop a product capable of die accurate and precise determination of the
analyte of interest in a given matrix. They must also develop appropriate calibration standards and procedures that
are consistent with the fundamentals of me immunochemical techniques. EPA's role has been to not only provide
guidance on the matrices of interest, the regulatory levels of concern, and the validation data required, but also to
develop written method documentation that incorporates the traditional quality assurance and quality control aspects
of EPA methods and provides sufficient information to allow the data user to develop confidence in the results.
Given the goals of allowing the use of as many testing products as possible without rewriting the method and
minimizing the need for product-specific descriptions in the procedure, the development of the written method is
not a trivial task.
In preparing Method 4670, EPA sought to provide the necessary level of quality assurance and written
detail in several ways, including:
• Providing general background and guidance on the mechanics of immunoassay testing in the
written method.
• Describing the potential limitations of the testing products relative to the intended use.
• Directing the user to product inserts and instructions provided by the manufacturer and reviewed
by EPA for the details of the testing procedures.
• Incorporating modifications to die traditional EPA approach to quality control so that the analyst
and die user can determine that the results are valid and usable.
Written in EPA's EMMC format and numbered for inclusion in SW-846, both Offices are planning to propose
Mediod 4670 in late 19%. It's primary use will be in SDWA compliance monitoring, but it may find application
is ground water monitoring associated with RCRA requirements. EPA hopes that the approach employed in Method
4670 will become the model for the incorporation of additional quantitative immunoassay methods.
386
-------
83
METHOD 8085: PESTICIDES BY GC-AED
Speaker/Presenter: Rov Y. Araki. chemist, EPA Region 10 Manchester Laboratory, Port
Orchard, Washington
Authors: Randy K. Cummings and Robert Rieck, chemists, EPA Region 10 Manchester
Laboratory, Port Orchard, Washington; Norman L. Olson and Robert Carrell, chemists,
Washington State Department of Ecology Manchester Laboratory, Port Orchard, Washington
ABSTRACT
Current EPA pesticide methodology focuses on a few target analytes using rigorous
calibration techniques. Often a great deal of effort is used to demonstrate acceptable multi-
point calibrations for compounds not present in the sample. The rigorous calibration
techniques may discourage efforts to identify and quantitate non-target detections.
Furthermore, the various clean-up procedures used in the methods limit the analysis to a few
target compounds. Proposed RCRA's SW-846 Method 8085 specifically uses gas
chromatography with atomic emission detection (GC-AED) for the general screening for
hetero-atom containing pesticides in environmental samples. The screen monitors for
nitrogen, sulfur, iodine, bromine, chlorine, phosphorous and carbon contained in the
compounds. A single point calibration of expected pesticides at the limit of quantitation
(LOQ) and a special AED compound independent calibration (CIC) mixture are utilized to
assure that quantitation levels are met, identify and quantitate compounds detected, and
provide dilution factors for further analysis if required. Quantitation of pesticides at levels
above the LOQ that do not meet specific CIC criteria may require the use of traditional EPA
methodology. Application of this technique is presented.
INTRODUCTION
Currently the EPA uses a myriad of analytical methods to cover the analysis of synthetic
pesticides. These methods use an assortment of instruments and detectors. The primary
instruments utilized are high performance liquid chromatography (HPLC) and gas
chromatography (GC) of which GC is the most commonly used.
Of the available detectors for GC analysis the electron capture detector (ECD), the flame
photometric detector (FPD), the nitrogen/phosphorus thermionic detector (NPD), Hall
Electrolytic Conductivity Detector (ELCD) and mass selective detectors (MSD) are widely
used in the various EPA methods. With the exception of the MSD, these detectors use their
selectivity to analyze particular classes of pesticides. For this reason their target compound
lists are limited to pesticides containing specific functional groups. The MSD is a universal
detector. Its' non-specificity makes investigating every peak in the resulting chromatogram
very time consuming and it is usually pursued in a cursory fashion. Thus the MSD is most
often limited functionally to a targeted list of pesticides.
387
-------
In addition to the limited scope of target compounds within each EPA method, these methods
have a set of quality assurance/quality control (QA/QC) criteria that include multi point
calibrations and rigorous calibration checks. The results obtained using these methods are
usually accurate and precise. However, so much effort is put into quality control of a target
list that investigation of unknown peaks is often neglected. Therefore the best use of these
methods is when the target list of compounds is thoroughly known, i.e. all sources of
pesticide contamination have been determined.
Currently a complete analysis for pesticide contamination of an environmental sample may
require several EPA methods that cover an assortment of often overlapping pesticide target
lists. In a restrictive senerio an investigator or laboratory client will request a "tried and true"
method that may not apply to the different pesticide contamination possibilities in his/her
samples. Method 8085 uses the selectivity of the atomic emission detector (AED) to screen
for and quantitate an extended list of pesticides in an effort to address the previous situations
using a single methodology. Where applicable the AED can also be used as the detector for
any of the other EPA methods following an initial demonstration of capability.
The AED uniquely offers the combination of the following capabilities not offered by any
other detector:
1) Multi-element response,
2) High selectivity for most elements common to pesticides,
3) and, compound independent calibration capability.
METHOD OVERVIEW
Method 8085 is proposed under the EPA's SW 846 RCRA methodologies. It is designed to
provide a flexible approach to general pesticide screening in a manner similar to that used
by the FDA. A minimum of QA/QC is used for non-detected target compounds whereas
detected target and non-target compounds follow the conventional QA/QC requirements of
the other EPA methods.
One function of the target list of compounds described in Method 8085 is to furnish
information about a wide range of pesticides comprised of differing functional groups.
Group characteristics can be used as a way to generalize method applicability. That is,
compounds with similar functional groups should generally behave the same within the
confines of the method. So even if a compound is not on the method list, method
applicability can be assumed — compound analysis can go beyond a routine list.
The method allows for the most efficient extraction procedures but priority is given to SW-
388
-------
846 methods. Even with this flexibility, not all compounds that can chromatograph via GC
can be extracted from a sample or remain stable during extract cleanup. Therefore, cleanup
of the resulting extracts is kept to a minimum until the sample has been screened.
The sample is screened for compounds containing phosphorus, nitrogen, sulfur, iodine,
bromine and chlorine. In addition, carbon is monitored for background assessment. These
elements are common hi most synthetic pesticides. Fluorine, oxygen and other elements can
be monitored, if desired, but they require additional analysis steps. Since the AED is very
selective to elements even in a difficult background matrix, the analyst has high confidence
in the presence of a hetero-atom containing compound. Therefore, more time and effort can
be dedicated to investigating or confirming the compound's identification and specific
cleanup steps, focusing on isolating the target compound can be implemented. Monitoring
the carbon channel aids in peak identification when mass spectral confirmation is desired.
Even with the selectivity the AED provides, its use in combination with mass spectrometry
optimizes this screening technique.
COMPOUND INDEPENDENT CALIBRATION (CIO
It has been demonstrated that GC-AED provides a linear response to elements that is
independent of molecular structure.(l,2) This phenomenon allows detector calibration based
upon elemental response. Thus a compound independent calibration (CIC) mixture is used
to calibrate the AED for all the elements of interest.
The suggested CIC mixture described in the method contains 15 compounds, but different
CIC mixtures may be used. The compounds in the mixture were chosen for their elemental
composition, solvent and GC stability, and retention times. Concentrations of the analytes
were adjusted to define the calibration range of the elements determined.
To demonstrate an application of the CIC mix, the calibration curve for element chlorine in
the following table shows that the AED response is linear and compound independent. Even
though the six compounds describing the curve elute in random concentration order a high
degree of linearity (RSD=2.5%) is observed. Area, not peak heights, must be used.
Compound
1,2,3-Trichlorobenzene
Dichlobenil
Silvex, methyl ester
PCNB
Chlorpyrifos
Decachlorobiphenyl
RT
6.157
7.903
15.067
15.350
19.520
ng/uL
681.0
614.0
40.0
169.0
568.0
33.18749.3
area(3_uL)
49558
30648
1921
12512
21605
4202
StdDv
ngCl/uL
4.00
2.54
0.15
1.01
1.72
0.351
310
AERF
RSD
ERF
12390
12066
12807
12388
12561
11971
12364
2.5%
389
-------
The CIC is key to the implementation of Method 8085. Therefore the quality of this standard
must be very high. After the CIC standard has been prepared, and during each analysis
thereafter, the linearity for each element must be tested. First, the average elemental
response factors (AERFs) are calculated for each element by averaging their elemental
response factors (ERFs). Second, the standard deviation of the AERF is determined. Last,
the relative standard deviation (RSD) is calculated by dividing the standard deviation by the
AERF. The linearity, for each element other than phosphorus is considered acceptable if the
RSDs are less than 10%. Phosphorus will be discussed in more detail later, but RSDs should
be less than or equal to 20%.
The CIC must also be verified against a standard of high confidence. Certified standards
from private companies make good reference standards. Most of the elements in the CIC are
interrelated through common compounds, thus a standard containing only one element may
be sufficient to provide quality assurance for the entire CIC. Once the CIC is verified, it can
be used to measure the reliability of most other analytical standards, as well as quantitate
analytes. In fact comparison of the single point calibration standard to the CIC is an ongoing
part of analysis using Method 8085.
Even though the CIC is a versatile calibration standard, to obtain full quality assurance for
an analysis, the compound specific standard (the single point calibration standard) must also
be of high quality. Multi-component mixes can be checked internally against the assay of
each compound. This provides information about the relative composition of the standard.
If all of the components of the mix provide relatively equivalent proportional elemental
responses, the only other error likely is the dilution of the entire standard. A dilution error
can be checked using the CIC.
THE SINGLE POINT CALIBRATION
Method 8085 employs a single point calibration using several multi-component standard
mixes. This single point calibration serves two purposes- The primary aim is to establish
a calibration point at the practical quantitation limit (PQL) for a defined target list. The
secondary goal is to provide a validation for CIC calculations.
The single point calibration is performed before and after sample analyses. When there is
a non-detect of targets, the only purpose of the single point calibrations is to insure that the
targets can still be detected. In this case no response criteria is required between the first and
last calibrations. By performing the single point calibration, this method provides as much
reliability as any EPA method of a compound's absence above a PQL.
A verified CIC mixture is used to calibrate the AED for carbon, phosphorus, sulfur, nitrogen,
chlorine, bromine and iodine. Fluorine and oxygen can be calibrated as required. The
calibration is used to determine dilution estimates, elemental ratios (empirical formula), or
390
-------
quantitation. Elemental ratios may be used for compound identification confirmations (along
with GC retention time). Quantitation can be made when:
1) the RSDs of the AERFs are less than or equal to 10%
(except for phosphorous at 20%),
2) a minimum of five points is used to calculate the AERF,
3) a single point analyte calibration produces an elemental
response factor within 20% of the AERF, before and after
sample injection and,
4) the elemental concentration falls in the range defined by the CIC.
The GC system should be in a condition to meet all of the above criteria for most compounds
on the target list before analysis begins. As a result of chromatographic difficulties some
analytes do not always meet the third condition. This is not a problem when the
compound(s) is not present in the sample and the minimum quantitation level can be
achieved. If a compound is identified in the sample but does not meet the single point
calibration criteria for response, then analyte calibration may be required if unqualified data
is desired. System performance and degradation can be assessed from CIC comparisons to
the single point calibrations. It should be emphasized that no other EPA method need be
employed unless a positive detection is made above the PQL and calculations using the CIC
cannot be made.
Information about retention time, matrix background, and hetero-atoms present are
invaluable for mass spectral investigations of unknowns. If an unknown is identified with
relative certainty it can be quantitated using the CIC AERFs. If extraction and GC
performance for the compound is unknown, the measurement is only an estimate. However,
this method provides the best estimate available without a reference standard.
METHOD DRAWBACKS
Quantitation limit requirements are occasionally too low for the AED to achieve. This is
especially true for some of the SW 846-8080 chlorinated pesticides, particularly for
Toxaphene and PCBs. In general multi-component analytes are not amenable to CIC
calibration techniques. However, that does not preclude analyzing for them. Patterns
derived from these analytes are often quite different from those found from ECD analysis,
thus providing additional information for their identification.
Sample preparation/analysis methods may need modification if lower detection limits are
required, i.e. use larger injection volumes, larger sample sizes, and/or smaller extract
391
-------
volumes. Such measures can also increase matrix interferences. For instance, tissue samples
may not be amenable to the method due to a combination of the lipids present and the low
quantitation limits commonly required. Modified sample preparation/analysis techniques
tend to concentrate the sample lipids that interfere with the gas chromatography portion of
the analysis.
Very often it is the GC portion of the instrument that limits the scope of the method, not the
detector.
Multiple injections are required to screen for all desired elements (see table). Each of these
injections requires time. If the sample has already been characterized, more definitive
methods may be cheaper and easier for a few target compounds.
Injection Sequence (element emission wavelength)
1st injection 2nd injection 3rd injection
193 carbon 478 bromine 186 phosphorous
181 sulfur 479 chlorine 178 phosphorous
174 nitrogen
206 iodine
Compounds containing phosphorus usually require extra effort when quantitating using
phosphorus emission lines. This is due to bias resulting from interaction of phosphorus with
the discharge tube.(3,4) If measures are not taken to reduce the interactions, RSDs for
phosphorus may fall outside the CIC calibration criteria. This is because the interactions
with the tube yield a disproportionate affect on the lower concentrate standards. Suppression
of the tube interactions is required to obtain phosphorus quantitation with minimum bias.
For screening purposes suppression of this effect is generally not necessary. However, if
required, full suppression can be accomplished with addition of 0.1% gas oil to the standard
and sample extracts. Partial suppression is obtained by introducing 2-methyl-naphthalene
diffused into helium through the AUX gas line of the AED. But, the use of 2-methyl-
naphthalene is not presented in Method 8085.
REFERENCES
1. N. L. Olson, R. L. Carrell, R. K. Cummings, and R. H. Rieck, "Gas Chromatography
with Atomic Emission Detection For Pesticide Screening and Confirmation," LC-
G£,vol. 12, p. 142(1994).
2. N. L. Olson, R. L. Carrell, R. K. Cummings, R. H. Rieck, and S. Reimer, "Atomic
Emission Detection for Gas Chromatographic Analysis of Nitrogen-Containing
Herbicides in Water," J. Assoc. Off. Anal. Chem.. vol. 78, No. 6 (1995).
392
-------
3. B. D. Quimby and J. J. Sullivan, Anal. Chem. vol. 62, p. 1027 (1990).
4. J. P. J. van Dalen, P. A. de Lezenne Coulander, and L. de Galan, Anal. Chim. Acta.
vol. 94, p 19(1977).
393
-------
84
VOLATILES PRODUCTIVITY ENHANCEMENT
WITH AN XSD® DETECTOR
Mark Bruce. Tim Lavey, Quanteira, 4101 Shuffel Dr. NW, North Canton, Ohio 44720
Richard Burrows,Quanterra,4955 Yarrow St., Arvada, Colorado, 80002
ABSTRACT
The XSD is a new halogen specific GC detector based on thermionic emission from an
alkali activated platinum electrode when interacting with chlorine, bromine and other
halogens. This paper will present comparative results from ELCD and XSD® detectors
for spike and real world volatiles samples. GC peak shape will be examined in particular
as well as general performance characteristics such as accuracy and precision, linearity,
analyte specificity and sensitivity to water. The impact of improved chromatographic
performance on productivity measures such as turn-around-time and capacity will also be
examined.
INTRODUCTION
Electrolytic conductivity detectors (ELCD) have been used for many years in the GC
analysis of volatile organic analytes. Methods 8010 and 8021 list the ELCD as the
detector of choice. While the ELCD is more specific than the electron capture detector
(ECD), chromatographic performance is degraded by the liquid transfers involved. Thus,
short term reproducibility may suffer and raise detection limits. The limited linear range
of the detector can require frequent dilutions for common samples. Long term stability is
limited and leads to more frequent recalibration than other common GC detectors.
Transfer of water from the purge-and-trap system to the GC is a common problem and is
particularly troublesome for the ELCD. General sensitivity declines and responses for
early eluting compounds are especially impacted. A relatively new detector technology
based on thermionic emission (Halogen Specific Detector, XSD®) promises improved
performance for analysis of volatile organohalogen compounds based on methods 8010
and 8021.
An OI Analytical application note describes the XSD (1). 'The ... XSD is a thermionic
device based upon the work of Rice (2) and Roberts (3). The XSD requires a negatively
biased platinum electrode (cathode) to be activated by an alkali, which lowers the work
function of the platinum metal. Prior combustion is required to oxidize the sample
[analytes] to carbon dioxide, water, and halogen atoms. Since halogen atoms are in the
gas phase, they exist in an equilibrium towards the dissociated halogen atoms." The
signal current is probably generated when halogen species adsorb onto the alkali activated
cathode surface, emission of a thermal electron, followed by desorption of a neutral
halogen atom or halogen ion or alkali-halide molecule.
Figure 1 shows the principle components of the detector. The GC column effluent is
introduced through the jet tube into the reactor core. The analytes are oxidized at high
temperature (1000°C) to produce the halogen atoms which interact with the platinum
probe. Several types of interactions produce the detector signal which produces the
chromatographic peak. The detector is reported to be linear from lOpg to lOng. From
lOng to lOOOng the calibration response is quadratic (!)• The detector is most sensitive to
chlorine. Bromine response is about 9X less sensitive than chlorine. Since the XSD
detection limits are much lower than required for normal environmental work this is not
expected to be a problem.
394
-------
Probe Assembly
Reactor Assembly
Jet Tube
Figure 1 XSD Diagram.
EXPERIMENTAL
A Hewlett Packard 5890 Series II was equipped with both an XSD Model 5360 and an
ELCD Model 5220 detectors. The injection port fed into a Valco Y splitter which was
connected to two Restek RTX-502.2 105m, 0.53 mm ID, 3 (im film columns. One
column was plumbed to the XSD and the other to the ELCD. An OI Analytical Sample
Concentrator Model 4560 was used. An Arcon Model 5100 autosampler from Dynatech
was attached to the sample concentrator.
RESULTS and DISCUSSION
This paper will present comparative results from ELCD and XSD® detectors for spike
and real world samples. GC peak shape will be examined in particular as well as general
performance characteristics such as accuracy and precision, linearity, analyte specificity
and sensitivity to water. The impact of improved chromatographic performance on
productivity measures such as turn-around-time and capacity will also be examined.
REFERENCES
(1) OI'Analytical. Application Note 07671095.
(2) Rice. Chester W. U.S. Patent 2,550,498. June 24, 1951.
(3) Roberts, John A. U.S. Patent 2,795,716. June 11, 1957.
395
-------
85
Fast GC Capillary Columns applied to Environmental Separations and
Analysis - Screening Methods
Dennis Gere. W. Dale Snyder, Vince Giarroco and Richard Kolloff, Hewlett-Packard,
Little Analytical Division, 2850 Centerville Rd. Wilmington, DE 19808,
phone 302-633-8162 fax 302-633-8908
Bernhard Rottweiler, Hewlett-Packard, Walbronn Analytical Division, Hewlett-
Packard Strasse 8, 76337 Waldbronn, Germany
There are some environmental analytical problems which would appear to benefit
from additional speed. In some cases, this is a direct enhancement of sample
throughput. In others, rapid screening of a potential hazardous sample area can aid
in the decisions about the ultimate analysis plan.
Although narrow-bore capillary columns ( 100-200 ) have been available for some
time, the narrow peak width was almost wasted because the time constant of the
GC and GC detector could not keep up. Recently, faster GC equipment has
become widely available providing the entire system desired for fast
chromatography.
There are several commercial GC systems approaches for fast gas chromatography.
They include ; cryofocus with fast desorption, cryogenic cooling and
multidimensional separations with rapid column heating, miniaturized
(micromachined) isothermal systems and small internal diameter columns with fast
temperature programming. In this talk, we will discuss the last approach listed, smalt
ID GC columns with fast temperature programming.
There are specific requirements in the hardware in order to take advantage of the
speed and resolution of small ID columns. These requirements are listed in table 1.
Table 1. Requirements for Fast GC columns Screening Methods
Requirement
High speed injection
High inlet pressures
Fast oven temperature programming
High speed detectors
High speed data acquisition
Inlet flow conditions
Parameter values
100 millisecond
up to 150 psi
35-120 degrees per minute
200 Hertz
1-200 Hertz
Electronic Pressure Control
396
-------
The gas chromatography applications presented here include data for the GC of
volatile organic compounds (VOC), total petroleum hydrocarbons (TrPH), and
organochlorine pesticides. The intent is to present very fast separations in each case
to be used for screening methods.
The first example is a separation of 60 VOC's at the 40 ppb level in 10 minutes. This
work was carried out with a 10 m X 100 urn X 0.34 mm capillary column using the HP
6890 Series GC. The speed is accomplished using electronic pneumatic control
(EPC). Both the FID and MSD detectors were utilized. This is seen in Figure 1.
The 100 um column provides a very adequate screening procedure that is perhaps
2-4 times faster than the conventional 250 um ID column. This will be extended to
compare with a 25 m X 200 mm X LI um column, with separation of most of the 60
VOCs at the 10 ppb level in 17 minutes. This larger bore, thicker film column provides
a sensitivity optimized method especially for the early eluting most volatile
compounds particularly for the mass spectral detector.
Figure 2 shows another example of the fast screening separation of the VOCs. In this
case, 84 VOCs are separated in 14 minutes. In each of the two examples, the
sample was introduced by purge and trap apparatus and the detection device was
a GC/MSD. In the more conventional separation on a 250 um ID column where full
quantitation would be accomplished, the separation time would be in the order of
35 to 40 minutes under analogous flow and temperature programs with the same
phase ratio.
Figure 3 depicts the separation of 44 semi-volatile compounds with a 100 um ID
column HP-1 column in 12 minutes. The HP 6890 GC with fast temperature
programming and splitless injection utilized the electronic pressure control with
constant flow starting at 48 psi inlet pressure.
In Figure 4, the fast screening separation of organochlorine pesticides was carried
out in just 6 minutes. A small volume electron capture detector was used in this
application. This detector has a wider than usual dynamic range, so quantitation on
a 250 um ID column was carried out in the same linear dynamic region.
Finally, in the last chromatogram. Figure 5, we see the group separation of a diesel
range of organic compounds resulting from the SFE of a soil sample. The entire
group separation was carried out in seven minutes.
397
-------
We will discuss the advantages of using these small diameter fast columns with the
comparative larger ID columns used for the determinative methods. In summary, it
will be shown that a current method used for determinations of environmental
samples can be used as the templates for scaling to small ID fast GC columns for
convenient screening of large numbers of samples in a short period of time.
Illustration will be made of a method translation computer program which allows the
conversion back and forth between screening methods and determinative methods
while maintaining equivalent separation and identical elution order.
398
-------
Figure 1. 60 VOCs separated in 9.5 minutes
1.00
100
3.00
4.00
S.OO
COO
7.00
1.00
1.00
399
-------
Figure 2. 84 VOCs separated in 14 minutes
Instrument: HP P&T,5890GC&5972MSD
Sample Name: 84 VOC
30 cm/sec He Carrier
M<*si.v* TIC:99581D.D
1400000
1200000
-
.
i
1000000
800000
600000
.
400000"
200000
' '
;TC 1.00
Hill
I
!
2.00 3.00
\
r-r-q
!
J!
1 '
4.00
L
-" 1
1
L
L
1
• • 1 1 1 1 1 1 1 1 • [ • 1 1
t
1
1
1
UJJ
.1
1
L
j,
"T j T T-T i i • i r i | 1 i i | ' i ' .
|
L
1
. JO
i 1 " ~? I ~~ ' " ' T '
5.00 6.00 7.00 8.00 9.00 10.00 11.00 12.00 13.00 14.00
400
-------
Figure 3.
40 Semi-Volatiles separated in
12 minutes
10m x 0.10mm, 0.4jim HP-1
JUh
40°C (0.5min)
120°C/min to 90°C (0 min)
5°C/min to 100 (0 min)
30°C/min to 320°C (hold)
0.5{il splitless
48.6 psi helium (0.6 ml/min)
0
8
10
12min
401
-------
Figure 4.
11 Organo Chlorine Pesticides
separated in 6 minutes
ECD1 B. (uECDS_08\SIG10Oae 0)
5H2
8000
700O-
0000-
2000-
1000-
S
402
-------
Figure 5.
Diesel Range Organics TrPH group
separation in 7 minutes
IOmeter x 0.10mm x 0.17Mm HP-5
pECD
PA
1200
1000
800
600
400
200
0
methylene
chloride
40°C(1.5min)
40°C/min to 325°C
helium 0.6 ml/min
0.1 \i\ splitless
-•—r
6
nun
403
-------
*> f I
GENERAL/AIR
&
GROUNDWATER
-------
86
THE APPLICATION OF HIGH SPEED GAS CHROMATOGRAPHY
TO AIR ANALYSIS
Norman Kirshen. Senior Chemist, David Coe, Senior Chemist,
Craig Hodges, Senior Chemist and Yihan Bao, Senior Chemist
Varian Chromatography Systems, 2700 Mitchell Drive, Walnut Creek, California, 94598
ABSTRACT
The demand for frequent sampling (2 to 5 minutes) approaching a continuous mode has generated interest
in high speed GC analysis. This paper describes an apparatus which produces a minimum injection
bandwidth (10 to 50 msec) which provides the primary means of attaining good column efficiency while
achieving High Speed GC Analysis. This FastGC inlet which is mounted on a Varian 3400 Gas
Chromatograph is described and its operation is detailed from the cryofocussing of the sample onto a short
Nickel tube to the 100,000°C/sec desorption to the 8 Meter x 0.32 mm capillary column. The applications
described in the paper include the analysis of chlorinated hydrocarbons, polar organics, an ozone
precursor standard (C2 to C13), and BTEX's. Data regarding the recovery, response linearity, and
retention time and area count precision are also reported
INTRODUCTION
The demand for faster analyses and increased sample throughput when using High Resolution Capillary
Gas Chromatography has traditionally required developing application specific chromatographic methods
which sacrifice column efficiency in order to use shorter columns, higher oven temperatures, higher
carrier gas flows, or less retentive coatings. The tremendous effort required to develop and modify
numerous application specific methods has pointed up the need for more general methodology yielding
fast GC analyses. This has lead to a re-examination of the means to preserve overall column efficiency
while reducing analyses cycle times. This paper describes the general applicability of minimizing injection
bandwidth as the primary means of retaining good column efficiency and simultaneously achieving High
Speed GC analyses. The technology is applied to the chromatographic analysis of several classes of air
pollutants.
THEORETICAL CONSIDERATIONS
Elution times can easily be reduced by shortening the length of the analytical column. However, the total
resolving power of that column is also reduced. Approaches to High Speed GC have attempted to improve
column efficiency in order to get useful performance from shorter columns. This can be seen
mathematically in the Giddings-Golay Equations [1-3] , used to evaluate column efficiency in terms of
minimizing plate height (H). Capillary column efficiency is related to longitudinal diffusion through the
column (B term ) and to resistance to mass transfer of the gas and liquid phases (C terms, Figure 1). For
low volume columns, such as short columns, performance is better described by modifying the Golay
equation by adding a D term [4] which accounts for band broadening by the system's dead volume.
One approach to obtain lower H values needed for short columns (<10M) has been to decrease column
diameter (dc) in the Cg term. If a particular analysis uses 0.25 mm ID columns, typical of today's
capillary' work, then 0.05 mm or 0.1 mm ID columns (microbore) would be required to gam any
significant speed reduction while maintaining resolving power. An example is shown in Figure 2. The
405
-------
higher pressures needed for these columns increases fl and decreases f2 (the pressure correction factors)
working against the gains made by decreasing dc. These columns are also somewhat limited, having a
decreased sample capacity, a smaller selection of available phases, and several additional instrument
hardware requirements.
FastGC inlet technology uses sample introduction as a means to reduce the requirements of column
efficiency to effect a usable separation. This technology improves column efficiency by significantly
reducing t (unrelained peak time) in the D term. The FastGC inlet system [5] cryofocusses the sample
onto a short nickel trap and then desorbs the sample using heating rates up to 100,000°C/sec. This results
in samples being introduced to the column in 2-10 msec bandwidths compared to the 50 to 500 msec
bandwidths of conventional GC. The injection dead volume is thereby reduced by a factor often to several
hundred. Theoretically, the D term is reduced, decreasing plate height, and the resolving power of the
desired short column is now sufficient to effect the same separations obtained using longer columns with
broader injections.
The reduction of injection bandwidth also permits minimizing analysis times by increasing carrier gas
flow rates. With a conventional split injector, even a very fast 50 ms injection requires reduced column
gas flow to obtain a low H value and good column efficiency (Figure 3).The 2 ms to 10 ms injection
bandwidths obtained by the FastGC inlet provides more tolerance to column flow increases, almost as
favorable as the theoretical minimum (0 ms). With these injections, column flow rates can be increased 2
to 4 times the standard rates.
APPARATUS AND ANALYTICAL CONDITIONS
All chromatograms were acquired using the following instrumentation and conditions:
Varian 3400CX Gas Chromatograph equipped with FastGC Accessory
Trap temp: -70 to -90°C (LN2 cryogenics)
Column: 8M x 0.25 mm ID x 0.25 ^M film thickness DB-1 or specified
Oven: 50°C Isothermal or temperature programmed as specified
Carrier Gas: H2,150 cm/second
Inlet: Direct Air Sample Inlet
Detectors: FID at 300°C (50 ms Time Constant)
P1D at 250°C (50 ms Time Constant)
s
Data acquired using a Varian Star Workstation ADC Board running at 40 Hz.
FASTGC INLET OPERATION
The Varian 3400CX GC was configured with the low volume FastGC inlet system. The basic inlet system
has been described previously [5]. As shown in Figure 4, samples may be introduced via a split injection
or the special air inlet that was used in this study. The microsolenoid valves, VI and V2, control the
carrier gas flow pathway. The deactivated, fused silica tubing restrictors (Rl, R2, R3) balance gas flows.
In the Standby mode, VI and V2 arc closed allowing carrier to flush the trap, restrictors and inlet line.
During sampling (Figure 4), VI is opened and carrier gas flows from the GC injector with the injected
sample to the trap, or in the case of air sampling, air is drawn through the trap to the vacuum source for a
user-specified time interval. Sample vapor is condensed in the trap, which may be held at temperature to -
99°C. Next, carrier purges any remaining sample from the restrictors and purges any air from the trap.
406
-------
For the sample desorption onto the column, VI is closed (Figure 5) reversing the gas flow direction
through the trap which is then heated rapidly (at approximately 100,000°C / second) by applying a current
pulse from a capacitive discharge circuit. The desorbed sample flows with the carrier gas onto the head of
the analytical column.
RESULTS AND DISCUSSION
Several standards and one sample were analyzed using this new approach to High Speed Gas
Chromatography. Most of these separations would require at least 20 to 40 minutes using standard GC
columns and injectors. But with the FastGC inlet, analyses times are approximately one minute.
In Figures 6 and 7 are shown the chromatographic separations of an ozone precursor standard by
traditional capillary Chromatography and FastGC. In both cases, recoveries of 100% were obtained
through CIO and fell off to approximately 60% for C12. Retention time precisions were similar at
approximately 0.05% . Peak area count precisions were approximately 1% and 3% for the traditional and
FastGC separations.
Figure 8 is a 0.65 mL sample of 10-20 ppm each paint booth standard in air. This mixture of polar
components was cryofocussed at -70°C on the nickel trap for approximately 30 seconds, desorbed to the 8
M x 0.25 mm DB-1 column, and chromatographed in less than one minute. Peak tailing normally
associated with polar components was not observed.
Figure 9 exhibits a mixture of ten halocarbons focussed at -70°C and subsequently chromatographed with
an initial column temperature of 35°C, programming immediately to 80°C. An air sample was drawn into
the FastGC module during carpet installation with the resulting chromatogram shown in Figure 10. It can
be seen that toluene was an obvious solvent in the carpet glue.
In the last figure, Figure 11, 0.16 mL of 50 PPBV BTEX's are detected on a PID. The sample was
concentrated over a 30 second time interval. Since samples of up to 2 mLs at 100% humidity are possible
this chromatogram shows that FastGC has application to industrial hygiene as well as ambient air analysis
where high ppb sensitivity is required.
SUMMARY
Dramatic reductions of elution times while maintaining good resolution and peak shape were observed
using shorter columns, faster carrier gas flows, and the reduced injection dead volume provided by the
FastGC inlet system. This was accomplished with standard 0.25 mm ID columns and standard GC
hardware.
Several gaseous standards including air toxics and a wide range of hydrocarbons from ppb to ppm levels
were preconcentrated on the FastGC cryotrap over fixed timed intervals and resolved in very short
analysis times.
407
-------
= 2Dgflf2
_(11k24
Cn =
96(k+1)2 Dgf2
2 k df2
3 (k+1)2 15T
D =
L(k+1)2
Figure 1. Modified Golay Equations for High Speed GC
75
25
[
D
1
1
0
lviMe*>o*Mtt^«7M1 >un
•Mi
7S
50
25
H,l
l
li
jL
ll I
i
• • «d
I
1 to 4 mm
4 5 to 8 min
Standard Column:
70
DB-1 BOM X 0.25mm ID
X0.25umfilm He flow: 1.2 ml/min @
24 psi Program: 31 °C
hold 15 min.
ramp2°C to 15O°C Sample: Gasoline 1 ul.
split 1 :830
1 078 inj:
Microbore Column
250°C; FID
DB-1 20MX01mmlD
X 0.4 un film He flow. 0.7 ml/min @ 95 psi
Pgm:35°C,
hold 10 min, ramp 4°C to 150°
— 0 = — ""T^ Gasoline 1ul, split 1:
1237 lnj:250°C
Figure 2. Standard Column (0.25 mm ID) vs. Microbore (0.1 mm ID) Analysis of Gasoline
408
-------
50ms
0
0 50 100 150 200 250 300 350 400 450
Average Linear Velocity (cm/s)
Figure 3. Dead Time Effect on Column Efficiency
Carrier-
Carrier
Purge Out
*- Split Vent
^ Pressure
Regulator
R2 <-\
Air
Sample
Inlet
V2
R3
\\WW)
Column
Figure 4. Sample mode for FastGC
409
-------
Carrier-
Carrier
Purge Out
Split Vent
Pressure
^Regulator
R2
; Air
Sample
Inlet
v:
VI
Column
Figure 5. Analyze Mode on FastGC
89
Conditions 7
Sample: 100 mL @100 ppb 6
Trap:TenaxTA/30«C
Column: 60M x 0.53 mm, 1.5(1 Film, ,
DB-1
FID: 12x10-"
2
j\JL
1
I
V
4
,
10
11
jk
•> /•*
1
UJ
3 3. Benzene
4. C7
I
5. i-C,
6. Toluene
7. C8
8. p-Xylene
1
„
9. o-Xylene
4 1"- C9
11. TMB
12. C10
13. Cn
14. C,2
15 „ r12
IS. C13
/• -i -^ ~*"
10
Time, min
20
Figure 6. Ozone Precursor Standard by Standard GC
410
-------
1 2 3
Conditions
Sample: 0.60 mL @ 5 ppm ea
Trap: -90°C
Column: 15M x 0.
FID:20x10-12
10
1.
2.
3.
3pm ea 4
5.
m, 0.25fiFilm, DB-1 g
7.
8.
9.
12J 10'
11
11.
12.
13 13.
14.
14 115.
L
C5
C6
Benzene
C7
l~vxg
Toluene
c«
p-Xylene
o-Xylene
c,
TMB
C,o
c,,
C12
C13
I
Time, min
Figure 7. Ozone Precursor Standard by FastGC
o 1-
C 0
Sample: 0.65 mL @ 10-20 PPM
Trap: -70°C
12 1 Methanol
Column: 35°C/9 sec. 5°C/min to 80°C
FID
3
1
4
5
9
a
_ 7
V .
6
!
I
I
-li
w
4 4
10 1,1
1
1 ti
\
.11 R . «» _
2. Ethanol
3. Acetone
4. 2-Propanol
5. 2-Butanone
6. Ethyl Acetate
.f 7. 2-Methyl-1-Propanol
•3
I
I
U
8 1-Butanol
9. 2-Propanone
10. n-Heptane
11. Toluene
12.p-Xylene
13.2-Heptanone
15 14.o-Xylene
L,A IS.ButvlCellusolve
01 0.3 05 07 09
Time, minutes
Figure 8. Polar Organics by FastGC and Flame lonization Detector
0.2 -
0.1
0.0-
6
3
1 _
I 2.
JLl
5
4
7
Sample: 0.66 mL @ 20-50 PPM
Trap: -70-C
Column: 35°C/0 min,
50°C/min to 80°C
FID
in 1 . Methytene Chloride
2. Chloroform
3. 1 ,2-Dichloroethane
9l 4. Carbon Tetrachloride
j| 5. Trichloroethylene
6. 1 -Chloropentane
7. Tetracriloroethylene
J|[ ^ 8. Chlorobenzene
10. 1,2.3-Trichloropropane
01
0.3
0.5
Time, minutes
0.7
09
Figure 9. Chlorinated Hydrocarbons by FastGC and FID
411
-------
ra 1.5
g>
W
0.5
Column: 50°C
10 15
Time (s)
20
25
30
Figure 10. Direct Lab Air Sample During Carpet Installation
Conditions
Sample: 0.16 mL
Trap: -90°C
PID:1x10-11
(g 50 PPBV
1. Benzene
2. Toluene
3 Ethyl Benzene
4. m,p-Xylene
5. o-Xylene
0 Time, min
Figure 11. BTEX's by FastGC and Photoionization Detection
REFERENCES
1 Giddings, J. C., Anal. Chem., 1964 36, 741
2. Giddings, J. C.. Seager, S. L.. Stucki, L. R.. Stewart, G. H., Anal. Chem., 1960 32, 867 (1960)
3. Giddings, C . Anal. Chem., 1962 34, 314
4. Caspar, G.. Arpino, P., Guicochon, G.J., J. Chromatogr. Sci., 1977 15, 256
5. Klcnip, M. A., Akard, M. L., Sacks, R. D., Anal. Chem., 1993 65, 2516
412
-------
87
A RAPID, COST-EFFECTIVE METHOD FOR DETECTION OF
TRfflALOMETHANES IN DRINKING WATER
Karen McKenzie - Manager of Technical Services, William Studabaker - Senior Research
Scientist, and Kevin Carter - Vice President of Sales, Marketing and Technical Services
EnSys Environmental Products, Inc.
P. O. Box 14063
Research Triangle Park, North Carolina 27709
ABSTRACT
EnSys Environmental Products, Inc. has developed the first commercially available, simple,
colorimetric technology that can supplement trihalomethane (THM) testing using EPA
methodology. This rapid, low-cost method enables water treatment facilities to respond
quickly to changes in water quality. The test method and required resources will be
described demonstrating its usefulness as a rapid, inexpensive and easy-to-use method for
in-plant total trihalomethane (TTHM) determinations. Conventional analysis of TTHM
require the use of gas chromatography methods, which require expensive equipment,
extensive user training, and significant run time (30-45 minutes). The data presented will
demonstrate the comparability of sensitivity, precision and performance to GC methods.
INTRODUCTION
Trihalomethanes (THMs) have been found to be the most widespread organic contaminants
in drinking water, and occur at higher concentrations than other disinfection by-products.
The four THMs (chloroform, bromodichloromethane, dibromochloromethane and
bromoform) are formed when chlorine-based disinfectants are added to source water with
fairly high organic content, such as surface water. THMs are included among the 25 volatile
organic compounds regulated under the Safe Drinking Water Act (SDWA) of 1987. These
compounds are persistent and mobile, and pose a cancer risk to humans. '
Regulations for the control of THMs in drinking water were promulgated by the EPA in
1979, setting the maximum contaminant level (MCL) of 100 jag/L (ppb) for systems serving
populations of greater than 10,000 people. Since then, the increasing awareness of
microbial risks in drinking water have caused disinfection and disinfection by-products to
become more of an issue. Stage 1 of the proposed Disinfectants/Disinfection By-Products
(D/DBP) Rule represents a lowered TTHM regulatory limit of 80 ng/L. Recently the
AWWA Water Industry Data Base reported that a safety margin of 15% below the
regulatory limit for total trihalomethanes (TTHMs) should be targeted; therefore,
compliance with an 80 ng/L TTHM standard would require that a treatment process achieve
< 64 ug/L to reliably stay below the regulatory limit. The stage 2 proposal, which will be
413
-------
reevaluated in a second negotiated rale-making process, currently includes lowered MCLs
of 40 ng/L for TTHMs.3
Because TTHMs are formed in proportion to the amount of organic matter in the source
water, changes in TTHM concentrations may indicate changes in source quality. Timely
measurement of TTHM concentrations can be a valuable tool in monitoring source water
quality and making relevant adjustments to treatment processes.
CURRENT ANALYTICAL METHODS
TTHMs are analyzed in the laboratory using gas chromatography methods. The most
widely used is EPA Method 502, volatile organohalogens by purge and trap/electrolytic
conductivity detection. Two other methods, EPA Method 501 and the new Information
Collection Rule Method 551 for determination of disinfection by-products, employ solvent
extraction/electron capture detection. Use of GC methods requires expensive equipment,
extensive user training, and significant ran time (30-45 minutes). Furthermore, these EPA
methods are relatively expensive, typically $40-100 per sample, and may involve laboratory
turnaround times (including data analysis and recordkeeping) of 2-4 weeks.
TTHM WATER TEST SYSTEM
A number of investigators have studied simple spectrophotometric methods for determining
THMs or other chlorinated organic compounds. "* These colorimetric methods all involve
in the color-forming step the Fujiwara reaction, which is a reaction between organic halides
and pyridine (or a pyridine derivative) in an alkaline medium. The analytical challenges in
developing a viable method using Fujiwara chemistry included extracting THMs from the
water sample, concentrating them in an appropriate medium, and developing reaction
conditions specific for the detection and quantitation of THMs in the presence of other
chlorinated organic compounds. EnSys Environmental Products, Inc. has developed the
first commercially available, simple, colorimetric technology, which can supplement THM
testing using EPA methodology. The test is simple to use and does not require sophisticated
equipment or chemistry training. Twelve samples can be tested in one batch generating
results within 15 minutes of sample collection.
In the EnSys TTHMs Test System, trihalomethanes are extracted from a water sample using
an Empore™ active carbon filter developed by 3M Corp., (St. Paul, Minn). Active carbon
is a ubiquitous sorbent for the removal of organic contaminants from water. There are also a
number of analytical devices, referred to as solid-phase extraction devices, filters, or
membranes, that consist of a non-polar (reverse) organic phase covalently bound to an
immobile surface, and that are effective in the removal of certain organic compounds from
water. However, such reverse-phase sorbents are not as efficient as active carbon, and the
use of active carbon in solid-phase extraction devices of the type used in the TTHMs Test
System has not been explored extensively. Other extraction techniques reported in
conjunction with Fujiwara chemistry deliver poorer sensitivity.7'8
414
-------
The sample is collected either by attaching the Empore filter directly to a tap and using line
pressure to filter the sample, or by collecting the sample in a suitable container (bottle or
vial) and pressure-filtering using a syringe and either manual pressure or a peristaltic pump.
After filtration, excess water entrained in the filter is removed by aspiration. The extraction
method provides a separation of analyte from some potential interferents and cross-
reactants, including inorganics and halogenated carboxylic acids, which are typically poorly
adsorbed to the filter. THMs are men eluted from the filter with pyridine, using a manual
dispenser. A developer reagent (containing hydroxide ion) is then added to the eluent, the
mixture is incubated in boiling water for two minutes, then the solution is cooled to room
temperature. A pink color indicates the presence of THMs; its intensity is proportional to
the concentration of THMs.
The Fujiwara reaction and its modifications involve the reaction of pyridine, hydroxide, and
chlorinated hydrocarbons to give products that absorb strongly in the visible or long
ultraviolet regions. The ultimate product in all reactions is (colorless) glutaconaldehyde,
obtained by N-alkylation of pyridine followed by complete ring hydrolysis.9 However,
depending on the analyte and on the specific reaction conditions, a number of Schiff base
derivatives of glutaconaldehyde may form as reaction intermediates, and more than one
molecule of glutaconaldehye may be obtained per molecule of analyte.
In the case of trihalomethanes a pink solution may be formed under certain conditions; the
structure of the compound responsible for the color has been determined by Uno et al.
Although the compound is a reaction intermediate, its stability can be controlled.
Conditions in the EnSys method have been optimized to yield the pink solution so that an
approximately equivalent, linear response on a weight basis is obtained for each of the four
THMs. Under the conditions of the present method, other chlorinated compounds give less
response; most give little or none.
Sample concentrations are determined by reference to a single kit calibrator or a standard
curve. A standard curve is obtained by analyzing a freshly prepared aqueous THM standard
in the method. Absorbance is measured in a spectrophotometer at a wavelength of 53Iran.
415
-------
TEST SYSTEM PERFORMANCE
SENSITIVITY
The minimum sensitivity is expressed in terms of the method detection limit, which is the
concentration at which 95% of the samples are correctly reported at that concentration.
This concentration was determined as three standard deviations above a mean blank sample.
The method detection level is usually regarded as the lowest concentration that could be
measured under ideal circumstances and for the TTHM Water Test System is 5 ng/L.
SELECTIVITY
The complete set of disinfection by-products (DBFs) and selected Method 502.1 volatile
organohalogens were tested for chemical cross reactivity with the EnSys method. The
cross-reactivity was determined relative to the chloroform kit calibrator. In addition to the
four individual THMs, the method detects (>10% response relative to an identical
concentration of chloroform) the following organic compounds: trichloroacetonitrile,
trichloroacetone, chloral hydrate, trichloroethylene and trichloroacetic acid. These
compounds are typically not present or found in very low concentrations in treated water.
As shown in Table 1 the test exhibits little cross-reactivity (<5%) with other organic
compounds.
TABLE 1
Organic Compound
Chloroform
Chlorodibromomethane
Bromoform
Bromodichloromethane
Trichloroacetonitrile
Trichloroacetone
Chloral hydrate
Trichloroethylene
Trichloroacetic acid
Other Chlorinated Compounds
Response %
100
92
79
74
70
61
43
28
15
<5
416
-------
88
Overview of Air Sampling and Analysis Methods
Gale G. Sutton, CIH - Galson Corporation
The benefits of understanding the different sampling and analyses methods available for air
contaminants should not be underestimated. Methods were developed for specific air matrices;
although some methods can be modified for alternative applications, others have limited use.
This paper will discuss the different validated methods presently published, their original purpose,
and some alternative applications.
The NIOSH and OSHA methods of air sampling and analysis were developed to evaluate
personal exposure to workplace chemicals and nuisance particles. The sample trains are designed
to be worn by an employee and collect air within his/her general breathing zone. The sampling
volume and collection media are designed to simulate the breathing capacity of an individual
during a normal working day.
Air sampling methods in SW846 were designed for stack or emission sampling. As an example,
method 0050 was written for the heated and/or humid matrix characteristic of stack emissions.
But SW846 is fairly limited in the types of compounds it includes. The Federal Register, 40-
CFR Part 60 and 61, contains not only on-site methods of contaminant determination, but also
contains additional laboratory methods that cover a broader variety of analytes.
Still other publications, such as the EPA Methods for Determination of Toxic Organic
Compounds in Air (TO Methods) and those listed in 40-CFR-Parts 50 and 51 were designed to
evaluate ambient air; generally designed to target very low detection limits.
The California Air Resource Board (CARB), EPA Methods Manual for compliance with the BIF
Regulations also contain air sampling and analyses methods. There are several commercially
available data basis (as well as some produced by EPA) which can be utilized to select the proper
method.
Each type of method has distinct advantages and disadvantages. Understanding the effects of
humidity, air sampling volume and co-contaminants can help you choose the correct method
modification when a specific procedure is not available.
In conclusion, the choice of method can affect usability of the data. Selecting the best method
for a project is essential in obtaining usable and meaningful data from the laboratory.
417
-------
PRECISION
The TTHM Water Test System was validated to ensure precision of results over the range of
regulatory interest. The precision of the test was measured by spiking twenty replicate
blank water samples at 5ug/L and 80ug/L and analyzed as two batches of ten each.
TABLE 2
Individual THM
Chloroform
Bromodichloromethane
Chlorodibromomethane
Bromoform
Relative Standard Deviation (%)
5ufi/L
3.8
10.2
5.7
9.1
80ng/L
1.6
3.2
4.3
3.9
As shown here die method reproducibly (RSDs _<_ 10%) detected concentrations between 5
and 80 ng/L.
RECOVERY
Blank water was spiked with each of the four THMs at various standard concentrations. The
graph below represents the concentrations measured after conducting the TTHM Water Test
method. The concentrations were calculated relative to the chloroform calibrator provided
hi the EnSys procedure.
FIGURE 1
Spike Recovery of Trihalomethanes
400
* Chloroform
Q Bromodichloromethane
O Chlorodibromoin ethane
x BI ouioxui HI
100 200 300
[THM], tpiked (ppb)
400
Chloroform: y = 1.0824* + 2.6157
RJ = 0.9993
Bromodidilonmietiujie y = 0.8S43x + 0.1902
R2 = 0.9993
CUorodibromometlunc: y = 1.1234x - 2.6397
R2 = 0.9981
Bromoform: y = 0.8077x + 1.3839
R1 = 0.997$
418
-------
The EnSys TTHMs test has a somewhat different response factor for each of the THMs,
however the response factors of individual THM spikes are within 15% of the mean
response.
Chloroform was selected as the basis for calibrating the test because most treatment plants
disinfect with chlorine, and as a result chloroform is the predominant analyte in the finished
water. Analyses of such samples using the EnSys procedure will correlate well with results
using gas chromatography (GC) methods.
Certain source water and/or water that has been treated with oxidants other than chlorine
will tend to contain more of the brominated THMs. As shown in the figure above, a
bromoform sample must contain greater concentrations of THMs to give the same amount
of color (absorbance) as the calibration sample. The method can be calibrated using a
standard curve representative of the relative concentration of each of the THMs in a
particular distribution system to account for this response factor difference. Alternatively,
the chloroform calibrator concentration can be multiplied by a correction factor to reflect
the different THM composition of the samples. For example, if water from a treatment
system tends to have a mixture of similar amounts of each analyte, then a correction factor
can be used that reflects an average response of the EnSys method to the four THMs.
CORRELATION WITH STANDARD ANALYTICAL METHODS
The TTHM Water Test can provide a high degree of accuracy when used to analyze water
samples contaminated with TTHMs. To evaluate the relative accuracy samples were
analyzed by the EnSys method; analysis by EPA GC Methods 501.1 and 502.2 was
performed on concurrently collected same samples utilizing two different certified labs. The
data presented below represent samples collected from 10 different water systems (71
observations) with variable treatment processes and a range of concentrations from not
detected to 116 ppb. Results for the same sample show average percent differences of about
20% between each method. The difference between the EnSys method and each GC
method is approximately the same as the difference between the two GC methods. On
average the EnSys method has not given higher or lower concentration values than either
GC method.
The absolute percent difference between methods for all observations in the four trials are
displayed in Table 3 below. For each pair of observations the magnitude of the percent
difference between concentration values was calculated; the mean of these values is the
absolute percent difference. The GC method is used as the reference method to calculate
the percent difference between the TTHM method and each GC method. For the
comparison between GC methods, the percent difference was calculated relative to the mean
of the two values. The results indicate that the percent difference between the EnSys method
and each GC method is approximately the same as the difference between the two GC
methods.
419
-------
The absolute difference between each method was examined for all observations in the four
trials. Percent differences for each pair of observations was again calculated, but this time
the actual difference (positive or negative) was used in calculating the mean; this is the
relative percent difference. The following results indicate that none of the three THM
methods is biased - high or low - relative to another.
TABLE 3
Absolute Percent Difference
Relative Percent Difference
502.2-501.1
(n=23)
21.5%
- 2.8%
501.1-TTHMs
(n=30)
19.0%
0.0%
502.2-TTHMs
(n=50)
22.4%
+1.4%
Field data show excellent correlation with laboratory methods on real samples. As
demonstrated by the slope of the regression line in Figure 2, there is a 1:1 correspondence
between the EnSys method and Method 502.2 with a correlation coefficient of 0.85.
FIGURE 2
•o
o
5.
CO
UJ
Correlation of EnSys TTHMs
Method with GC Method 502.2
120
100
80
60
40
20
0
y=1.1062x-3.5588
R2 = 0.8475
20 40 60 80
Method 502.2
100 120
420
-------
SUMMARY
EnSys Environmental Products, Inc. has developed an accurate, rapid, low-cost method for
measuring total trihalomethane concentrations in drinking water. The results demonstrate
good comparability of sensitivity, precision and accuracy to GC methods. The data further
show that the TTHMs Water Test can be used as an in-plant alternative to the conventional
laboratory-based methods. The method enables water treatment facilities to respond quickly
to changes in water quality and serves as an effective monitoring tool on a routine basis.
NOTES
^ereira, M.A. 1983. Carcinogenicity of chlorination by-products: trihalomethanes.
Water Chlorination: Environmental Impact and Health Effects, 4(Book 2): 1165-76.
2Munro, N.B. and Travis, C.C. 1986. Drinking-water standards. Risks for chemicals and
radionuclides. Environ. Sci. Technol., 20:768-769
3Roberson, J. A. et al. The D/DBP Rule: where did the numbers come from? Jour.
AWWA, 87:10:57 (Oct. 1995).
4Leibman, K.C. and Hindman, J.D. 1964. Modification of the Fujiwara Reaction for
Determination of Polyhalogenated Organic Compounds. Anafyt. Chem. 36:348-351.
5Reith, J.F.; van Ditmarsch, W.C.; de Ruiter, T. 1974. An Improved Procedure for
Application of the Fujiwara Reaction in die Determination of Organic Halides. Analyst,
99:652-656.
6Okumara, K.; Kawada, K.; Uno, T. 1982. Fluorimetric Determination of Chloroform in
Drinking Water. Analyst, 107:1498-1502.
7Reckhow, D. and P. Pierce. 1992. A Simple Spectrophotometric Method for the
Determination of THMs in Drinking Water. AWWA Research Foundation and American
Water Works Association.
"Miller, H.; Milanovich, P.P.; Hirschfeld, T.B.; Miller, F.S. 1987. Optrode For Sensing
Hydrocarbons. U.S. Patent 4,666,672.
9(a) Uno, T., Okumura, K.; Kuroda, Y. 1981. Mechanism of the Fujiwara Reaction. Structural
Investigation of the Reaction Products from Benzotrichloride. J. Org. Chem. 46:3175-3178.
(b) Okumura, K.; Wada, T.; Yamaoka, K.; Uno, T. 1984. Determination of Rate Constants
for formation and Decomposition of Color Products in the Fujiwara Reaction Using
Benzotrichloride. Chem. Pharm. Bull. 32:174-178.
10Uno, T., K. Okumura, K.; Kuroda, Y. 1982. The Fujiwara Reaction: Isolation and Structural
Investigation of the Reaction Product from Chloroform. Chem. Pharm. Bull., 30(5): 1876-
1879.
421
-------
89
DETERMINATIONS OF N-NITROSODIMETHYLAMINE (NDMA) AT PART-PER-
TRILLION (ng/L) CONCENTRATIONS IN CONTAMINATED GROUNDWATERS
AND DRINKING WATERS FEATURING CARBON-BASED MEMBRANE
EXTRACTION DISKS
Bmce A. Tomkins and Wayne H. Griest, Organic Chemistry Section, Chemical and Analytical
Sciences Division, Oak Ridge National Laboratory', P. O. Box 2008, Oak Ridge, Tennessee
37831-6120; Gaydie Connolly and Heidi C. Hayes, Laboratory Support Division, Building
130, Rocky Mountain Arsenal, Commerce City, Colorado 80022
"Tb« mbntiOtti manuscript KM box
•uttmd by • contactor al «w U.S.
Gewnmwnt undw eomnct No. 06-
ACaS4eOR22484. Accon*nglr.«MU.S.
Oe»«ninnnt nb)in« i HUHUI tatt*.
np*r*M «e»M to puMMi at raemduc*
•w puUWwd terni al Mi anMbutton, or
•raw oMn to do to. to U.& Go»«R»wnl
This research was sponsored by the U. S. Army, Program Manager Rocky
Mountain Arsenal, under U. S. Department of Energy Interagency Agreement
No. 1989-H077-A1. Oak Ridge National Laboratory is managed by
Lockheed Martin Energy Research Corp. for the U. S. Department of Energy
under contract DE-AC05-960R22464.
422
-------
DETERMINATIONS OF N-NITROSODIMETHYLAMINE (NDMA) AT PART-PER-
TRILLION (ng/L) CONCENTRATIONS IN CONTAMINATED GROUNDWATERS
AND DRINKING WATERS FEATURING CARBON-BASED MEMBRANE
EXTRACTION DISKS
Bruce A. Tonkins and Wayne H. Griest, Organic Chemistry Section, Chemical and Analytical
Sciences Division, Oak Ridge National Laboratory", P. O. Box 2008, Oak Ridge, Tennessee
37831-6120; Gaydie Connolly and Heidi C. Hayes, Laboratory Support Division, Building
130, Rocky Mountain Arsenal, Commerce City, Colorado 80022
ABSTRACT
A new solid phase extraction procedure extracts N-Nitrosodimethylamine (NDMA) at part-
per-trillion (ng/L) concentrations from aqueous samples using a C]g (reversed-phase)
membrane extraction disk layered over a carbon-based extraction disk. The reversed-phase
disk removes nonpolar water-insoluble neutrals and is set aside; the carbon-based disk is
extracted with a small volume of dichloromethane. NDMA is quantitated in the organic
extract using a gas chromatograph equipped with both a short-path thermal desorber and a
chemiluminescent nitrogen detector (CLND). The Method Detection Limit for the procedure
is 2 ng of NDMA/L; the analyte recovery is approximately 57%. A related procedure
substitutes a standard automatic sampler for the short-path thermal desorber, and is suitable
for determining NDMA in heavily-contaminated aqueous samples. The Method Detection
Limit for the "high-level" procedure, which employs an automatic sampler, is 180 ng of
NDMA/L, with an analyte recovery of approximately 64%.
The detection limits and measured recovery values are comparable to those observed in earlier
work in which a conventional continuous overnight extraction with dichloromethane was used
to remove NDMA from the aqueous samples. The newer procedures described herein offer
a fiftyfold savings in extraction time and a one-hundredfold reduction in dichloromethane
consumed per sample, while maintaining the wide range (three to four orders of magnitude
concentrations of NDMA) observed for the original procedures used in a tandem.
The proposed procedure was challenged during a sampling campaign in which 100-200
groundwater samples were analyzed. In most cases, concentrations ranging between <5 to
200 ng NDMA/L were measured without difficulty, and were confirmed by GC/MS operated
in the single-ion monitoring mode. Some samples required an extra C18 membrane extraction
disk to reduce the concentrations of neutral contaminants. High concentrations of sulfur-
bearing species present in groundwater samples may suppress NDMA recovery.
This research was sponsored by the U. S. Army, Program Manager Rocky
Mountain Arsenal, under U. S. Department of Energy Interagency Agreement
No. 1989-H077-A1. Oak Ridge National Laboratory is managed by
Lockheed Martin Energy Research Corp. for the U. S. Department of Energy
under contract DE-AC05-96OR22464.
423
-------
INTRODUCTION
In previous work1, we demonstrated that Af-nitrosodimethylamine (MDMA, CAS Registry
No. 62-75-9) can be quantitated readily at the part-per-trillion (ng/L) level in groundwaters
after they are extracted continuously overnight with dichloromethane2. Briefly, a substantial
portion of the resulting dichloromethane concentrate is then spiked onto a dual-sorbent
carbon trap and dried before the residues (including MDMA) are desorbed onto a fused-silica
gas chromatographic column. These residues are then separated in the usual manner; NDMA
is detected selectively using a chemiluminescent nitrogen detector (CLND) optimized in its
nitrosamine-selective mode. This procedure, while effective, exhibits a lengthy sample
turnaround time (approximately three days per batch of eight samples, including a blank and
appropriate QA/QC sample) and generates a substantial volume (approximately 300 mL per
1-L aqueous sample) of dichloromethane waste. For these reasons, it was most desirable to
develop alternate procedures which would be effective, rapid, and generate smaller volumes
of dichloromethane waste3.
Both Taguchi et al ° and Jenkins et al6 have described extraction procedures for NDMA
in which a small mass (200 mg) of Ambersorb 572, a carbonaceous sorbent, was used to
extract NDMA effectively from 500 mL drinking water. We have extended this procedure
by employing a carbon-based membrane extraction disk for removing NDMA from aqueous
samples7. The procedure described herein employs a preconditioned CI8 membrane
extraction disk, laid over the carbon-based disk, to remove the unwanted neutral water-
insoluble species. The former disk is set aside; the latter is dried and extracted with a small
volume of dichloromethane. NDMA may then be quantitated using the various procedures
described in Reference 1. Both the extraction time for a 1-L aqueous sample and the volume
of waste dichloromethane are reduced approximately one hundredfold when the membrane
disks are used instead of the continuous overnight extraction procedure. The calculated
Method Detection Limit (MDL)* concentrations are 2 and 180 ng NDMA/L for the "low-
level" (manual injection) and "high-level" (automatic injection) procedures, respectively. The
use of the two procedures in a tandem permits the quantitation of NDMA concentrations
ranging over approximately three orders of magnitude, from ng/L to the ng/L levels.
EXPERIMENTAL
Sample Collection and Storage, Groundwater samples should be collected in commercially
available predeaned 1-L narrow-neck amber glass bottles, stored at 4°C, and extracted within
7 days of sampling.
NDMA Standard Solutions. Stock solutions with certified concentrations of NDMA in
methanol may be purchased from several vendors. These include NSI Environmental
Solutions, Inc. (Research Triangle Park, NC; 5000 ng NDMA/mL); AccuStandard (New
Haven, CT; 1000 ng NDMA/mL); Polysciences (Niles, IL; 1000 ug NDMA/mL); and Ultra
Scientific (North Kingstown, RI; 100 ug NDMA/mL). Aliquots are diluted sequentially to
final concentrations ranging between 2 and 240 ng of NDMA/mL in highest purity
dichloromethane (GC2 grade, Burdick & Jackson, Muskegon, MI, or equivalent) to form
424
-------
calibration standards for the "low-level" method, which employs the short-path thermal
desorber, or between 50-4000 ng of NDMAAnL dichloromethane, prepared similarly, for the
"high-level" method, which employs the automatic sampler. Fresh standards should be
prepared every 60 days.
Preparation of Blanks and Spiked Samples for Method Certification and Quality Assurance.
Aliquots (1-L) of synthetic groundwater are prepared from HPLC-grade water (Burdick &
Jackson, J. T. Baker (Phillipsburg, NJ), or equivalent) which is fortified to 100 mg/L each in
sulfate and chloride, as described elsewhere9. These are spiked to concentrations ranging
between 2 and 40 ng of NDMA/L using a master spiking solution containing 0.1 ug of
NDMA/mL of water for the "low-level" procedure. Spiked samples for the "high-level"
procedure are prepared from a master spiking solution containing 10 ug of NDMA/mL water;
the final concentrations range between 100-4000 ng of NDMA/L. Water produced from
typical "on-demand" systems may not be used, as described in Reference (1).
Sample Preparation and Extraction. Aqueous samples are filtered through Empore C,g filter
disks (47 mm diameter, J. T. Baker or Varian Sample Products Division, Harbor City, CA)
layered over a carbon-based Empore disk10'11. An all-glass funnel/support assembly
(DELTAWARE, with a 1-L filtering flask and a 300-mL reservoir, handles 47 mm diameter
disks, VWR, part no. KT 93825-47 or equivalent) is used to support the disk and hold the
filtrate. The two disks are preconditioned simultaneously with two 10-mL portions each of
methanol and water. Each aliquot remains on the disk for 1 min prior to removal. Do not
permit the disk to become dry during the conditioning sequence. A separate all-glass
funnel/support assembly should be arranged for drying the carbon-based disks. Here, the 1-L
flask is replaced with a custom-prepared "cap" fashioned from an "inner" 1 (standard taper)
40/35 glass joint; dimensions are 37 mm o.d., 60 mm high. These "caps" are large enough
to accommodate pre-cleaned tared 20-mL screwcap vials.
Desorption Equipment. The sorbent trap contains 125 mg each of Carbotrap and Carbotrap
C (Supelco, Inc., Bellefonte, PA), which are packed and conditioned as described in
Reference 1. The short-path thermal desorber (SPTD), Model TD-1 (Scientific Instrument
Services, Ringoes, NJ), is installed and arranged as described in Reference 1 Sample
residues are first dried at room temperature using ultrapure (99.9999%) helium for 2 min at
25-30 mL/min, then desorbed for 5 min at 150°C using ultrapure helium (3 mL/min).
Gas Chromatograph, Automatic Sampler, Chemiluminescent Nitrogen Detector (CLND),
and Data System. An Antek Model 705D pyro-chemiluminescent nitrogen detector was
retrofitted to a gas chromatograph (described below), with details given in Reference 1.
A Hewlett-Packard Model 5890 Series II gas chromatograph equipped with a split/splitless
injector both supported and was physically interfaced to the SPTD ("low-level" procedure),
CLND, data system, and automatic sampler ("high-level" procedure). One of the existing
data interface boards was replaced with a Hewlett-Packard analog input board so that the
output from the CLND could become part of the overall system network. The gas
chromatographic parameters and conditions were optimized and are reported in Reference
425
-------
(1). All gas chromatography was performed with a fused-silica capillary column, Rtx®-200
(Crossbond trifluoropropylmethyl), 3.00 um film thickness, 0.53 mm i.d. x 30 m (Restek
Corp., Bellefonte, PA) connected to a deactivated and uncoated fused silica guard column,
0.53 mm i.d. x 5 m using a Universal Press-Tight connector (Restek).
When the "high-level" procedure was used, a Hewlett-Packard Model 7673 Automatic
Sampler replaced the short-path thermal desorber and injected 2 uL sample on column. The
gas chromatographic operating parameters are the same as those described above with the
exception of the purge valve settings, which are "OFF" at 0.00 min and "ON" at 0.25 min.
The presence of NDMA in contaminated groundwaters was verified using a Hewlett-Packard
Model 5972 GC/MS equipped with a DB-5 analytical column (30 mm x 0.25 mm i.d., 1 urn
film thickness). The column oven temperature was programmed from 40 °C (hold for 4 min)
to 310°C (hold for 14 min) at 10°C/min. Dwell time was maintained at 1000 msec. The
presence of NDMA was established by its retention time (8.3 min) and two major m/e ratios
(74 and 42). The presence of the surrogate d4-dichloromethane was verified using its
retention time (14.7 min) and major m/e ratios (150 and 152).
Procedure: Aqueous samples are filtered simultaneously through C,g and carbon-based
Empore disks under vacuum at approximately 40-50 mL/min (a 1-L sample such as drinking
water or filtered groundwater, which contains little or no paniculate matter, is typically
finished in approximately 20 min). The extraction apparatus is then dismantled; the C,g disk
is set aside or may be used for related independent analyses. The carbon-based Empore disk
is transferred to the second all-glass filter support where the "cap" has replaced the normal
1-L filtration flask, then dried for 10 min under vacuum. When the disk is thoroughly dry,
a precleaned tared 20-mL screwcap vial is placed in the "cap," which is then reattached to the
all-glass filter support. A 3-mL aliquot of highest-purity dichloromethane is added to the
reservoir. After the solvent remains over the carbon-based disk for 1 min, it is drawn
through the disk under vacuum (slowly, if at all possible); approximately 2 mL is collected in
the screwcap vial. The vial is then reweighed to determine the mass (and later, the accurate
volume) of the dichloromethane extract.
The determination of NDMA at ultratrace concentrations (between 2 and 40 ng of NDMA/L
water) proceeds in the manner described in Reference 1. A 100 uL aliquot of the
dichloromethane concentrate'is added slowly to the Carbotrap C end of the dual sorbent trap,
allowing sufficient time for the liquid to soak into the trap bed. The trap is then mounted into
the short-path thermal desorber (Carbotrap C end is connected to the "top tube" without the
needle attached) and dried with ultrahigh purity helium for 2 min at 25-30 mL/min. This
procedure is repeated, leaving the residues from 200 uL of concentrate on the trap. The trap
is then disconnected from the "top tube" and inverted. The needle is connected to the
Carbotrap C end of the trap, and the remaining Carbotrap end is screwed into the "top tube."
The residues present on the trap are then desorbed directly from the sorbent trap onto the
guard and analytical gas chromatographic columns, both maintained at 35°C, using a 150°C
desorption temperature maintained for 5 min and a helium desorption flow rate of 3 mL/min.
The column oven temperature program commences immediately after completion of
426
-------
desorption.
The extraction procedure for "high" levels of NDMA, typically exceeding 300 ng of
NDMA/L water, is the same as that described above. Upon completion of the carbon disk
extraction, the dichloromethane extract is transferred to a 2-mL amber autosampler screwcap
vial equipped with a Teflon®/Silicone/Teflon® septum (National Scientific Co.,
Lawrenceville, GA). A 2 uL aliquot is injected onto the column using the automatic sampler^
and analyzed using the gas chromatographic conditions described above.
RESULTS AND DISCI ISSTON
The MDL values for both the "low-" and "high-" level procedures were calculated as
described in Reference (8). In the "low-level" procedure, eight (minimum seven required) 1-L
synthetic groundwater samples were fortified to 10 ng NDMA/L and analyzed as described
above. The standard deviation, which was calculated for both seven and eight replicates, was
then multiplied by the appropriate one-tailed Student's-t statistic for 99% confidence. The
resulting value is the MDL, approximately 2 ng of NDMA/L when all eight experimental data
points. When the "best seven" values were employed (that is, those with the smallest spread),
were employed, the MDL was reduced to 1 ng of NDMA/L groundwater. A similar approach
was taken for the "high-level" procedure, where the eight groundwater samples were fortified
to 1000 ng NDMA/L. The MDL calculated using all eight data was 180 ng of NDMA/L; that
using the "best seven" data (that is, those with the smallest spread) was 100 ng of NDMA/L.
All raw and calculated data are presented in Table 1.
The performance of the disk-based procedure was evaluated initially by quantitating NDMA
in two small batches of drinking and contaminated groundwater samples. In both cases,
additional QA/QC samples were analyzed to ensure the quality of the results. The latter
included dichloromethane solvent blanks (no detectable NDMA), synthetic groundwater
blanks (NDMA detected at levels at or below the MDL), and synthetic groundwater samples
which had been fortified to 10 ng of NDMA/L (NDMA detected at a minimum recovery of
55%). These results were deemed sufficient to consider the resulting NDMA concentration
data valid.
This procedure was challenged rigorously during a sampling campaign at a military defense
site during the fall of 1995. Approximately 100-200 individual groundwater samples, with
accompanying blanks and matrix spikes, were withdrawn from monitoring wells and analyzed.
While most of these samples were processed without difficulty, a modest number required
additional modifications to the procedure described above. Heavily-contaminated samples
frequently required an additional C|g Empore disk to reduce the number of extraneous peaks.
Furthermore, the presence of mustard gas by-products and degradation products, such as
dithiane and thioxane (oxathiane), in substantial concentrations in the groundwater samples
suppressed the chemiluminescent signal for NDMA. The reasons for this effect are not well
understood at this time.
The concentration of NDMA in most of the samples examined was below the MDL calculated
427
-------
above. A number of the monitoring wells were resampled in February 1996. The resulting
groundwaters were not only re-analyzed, but the presence or absence of NDMA was
confirmed by GC/MS in the single-ion monitoring (SIM) mode. The resulting data are
presented in Table 2. In general, there is good agreement between the chemiluminescence
and GC/MS-SIM data, particularly for the resampled data set. Furthermore, the agreement
between all data for a given site (Fall 1995 and February 1996) is generally acceptable if
NDMA was detected above approximately 20 ng/L.
SUMMARY
The above discussion demonstrates that analytical procedures based on carbon-based
extraction membrane disks are capable of quantitating NDMA in aqueous samples with
recoveries equal to or greater than existing procedures which employ conventional overnight
dichloromethane extractions. The detection limits using the disks were calculated using a
procedure given in the Code of Federal Regulations; the values were virtually
interchangeable with those obtained previously1. When the "low-level" (2-40 ng of
NDMA/L) and the "high-level" (100-4000 ng of NDMA/L) procedures are used in a tandem,
it is possible to quantitate NDMA over approximately three to four orders of magnitude. In
cases where the aqueous NDMA sample concentration falls between the upper bound of the
"low-level" procedure and the lower bound of the "high-level" procedure, an accurate
quantitation may be performed by simply spiking dichloromethane concentrate volumes
smaller than the usual 200 uL on the dual-sorbent trap. It has been demonstrated1 that the
trap performance is independent of the volume of dichloromethane applied to it, up to at least
200 uL. Procedures which employ the membrane extraction offer considerable savings in
both labor and the volume of hazardous waste organic solvent produced compared to
conventional methodology. In this case, the volume of dichloromethane employed was
reduced one-hundredfold compared to that required by conventional continuous extractions.
Furthermore, the overall turnaround time for a lot of eight samples was improved from three
days to a single day. It was also feasible to process a single "emergency" sample within three
hours.
ACKNOWLEDGMENT
The authors gratefully acknowledge the assistance of Craig G. Markell (3M Industrial and
Consumer Sector, St. Paul, MN) for productive discussions and an initial gift of the carbon-
based Empore disks. This work was sponsored by the U. S. Army, Program Manager for
Rocky Mountain Arsenal, under U. S. Department of Energy Interagency Agreement 1989-
H077-A. Oak Ridge National Laboratory is managed by Lockheed Martin Energy Research
Corp. for the U. S. Department of Energy under contract DE-AC05-96OR22464.
428
-------
Table 1 . Calculation of MDL Values for the "Low-Level" and "High-Level"
Procedures for MDMA in Groundwater
Replicate No.
1
2
3
4
5
6
7
8
Standard deviation, all
samples
MDL, all samples'1, ng/L
Standard deviation, "best
seven" samples
MDL, "best seven"
samples6, ng/L
"Low-Level" Procedure'
Calculated NDMA (ng/L)b
7.4
7.0
6.8
6.6
8.6
7.0
7.0
7.6
0.6
2
0.3
1
"High-Level" Procedure'
Calculated NDMA (ng/L)b
710
680
660
710
680
740
560
750
60
180
30
100
'Spiked water concentration is 10 ng NDMA/L.
bValues not corrected for recovery
'Spiked water concentration is 1000 ng NDMA/L
•"Student's t-value (one-tailed) at 99% confidence, eight samples = 2.988.
'Student's t-value (one-tailed) at 99% confidence, seven samples = 3.143.
429
-------
Table 2. Summary of MDMA Results on U. S. Geological Survey Resampling by
Chemiluminescence (CLND) and by GC/MS in the SIM Mode
Sample ID
Method blank
50 pptr spike/H2O
A
B
C
D
E
F
G
H
I
J
Matnx spike
K
Method Blank
50 pptr spike/H2O
L
M
N
O
Matrix spike
40 ng/mL standard
Original Sampling
CLND
ngNDMA/L'
<5
40
20
20
<5
<5
<5
<5
29
34
26
<5
42
21
<5
39
21
43
<5
<5
'36
Not performed
CLND
ngNDMA/L
1
29,32
2
2
4
5
1
1
24
27
19
2
26
11
1
35
12
68
2
1
26
Not performed
Resampling
GC/MS SIM,
ng NDMA/L, by
area
<5b, ]'
52,74
<5,2
<5,3
5,9
5,8
<5,2
<5,1
33,54
27, 39
24, 4 1
<5,2
41,61
\2,18
<5, 7
50,68
14,17
104,775
<5, <5
<5, /
31,40
53,56
GC/MS SIM,
ng NDMA/L, by
height
<5, 1
81, 720
<5,3
<5,3
5,10
5,10
<5,2
<5,1
38,85
33, 65
25,54
<5,2
49, 100
11,77
<5,7
50,67
14, 19
134,225
<5,2
<5, 7
30,34
47,75
Chemiluminescence method employs method of external standards exclusively for calculation
of MDMA concentrations
First number given uses the method of external standards to calculate NDMA concentration by
GC/MS SIM
Italicized number uses the method of internal standards to calculate NDMA concentration by
GC/MS SIM.
430
-------
REFERENCES
1. Tomkins, B. A.; Griest, W. H.; Higgins, C. E. Anal. Chem., 1995, 67, 4387-4395.
2. Method 3520A, Continuous Liquid-Liquid Extraction. Test Methods for Evaluating
Solid Waste, Physical/Chemical Methods, SW-846, 3rded,, Final Update 1, U. S.
Environmental Protection Agency: Washington DC, Jul 1992; Revision 1.
3. Ho, J. S.; Tang, P. S.; Eichelberger, J. W.; Budde, W. L. J. Chromatogr. Set. 1995,
33, 1-8.
4. Taguchi, V. Y.; Jenkins, S. W. D.; Wang, D. T.; Patmentier, J.-P. F. P.; Reiner, E. J.
Can. J. Appl. Spectrosc. 1994, 39, 87-93.
5. Taguchi, V. The Determination ofN-Nitrosodimethylamine (NDMA) in Water by Gas
Chromatography-High Resolution Mass Spectrometry (GC-HRMS), Environment
Ontario (Canada), Laboratory Services Branch, Quality Management Office: Ontario,
1994; Method catalogue code NDMA-E3291A, originally approved Feb 19, 1993;
revised and approved Apr 7, 1994.
6. Jenkins, S. W. D.; Koester, C. J.; Taguchi, V. Y.; Wang, D. T.; Palmentier, J.-P. F.
P.; Hong, K. P. Environmental Science and Pollution Research, 1995, in press.
7. Craig G. Markell, 3M Industrial and Consumer Sector, St. Paul, MN, private
communication, 1995.
8. Appendix B to Part 136—Definition and Procedure for the Determination of the
Method Detection Limit—Revision 1.11. Code of Federal Regulations: Protection
of the Environment, Parts 100-149; Title 40; U. S. GPO: Washington, DC, Revised
Jul 1, 1990.
9. Program Manager for Rocky Mountain Arsenal: Chemical Quality Assurance Plan.
Version 1, Sep 1993.
10. The carbon-based Empore disks are currently not available through normal vendors.
They should be ordered directly through: Mr. Chuck Haskins, National Sales
Manager; 3M; Building 220-9E-10; St. Paul, MN 55144-1000.
11. Physical parameters of the disks: Carbon type, acid-washed coconut charcoal.
Surface area > 1000 m2/g. Nominal particle diameter, 15-20 urn. Weight of carbon
in the disk, ca. 90% w/w of the disk (about 400-450 mg). Pore size distribution is
proprietary.
431
-------
90
Limited Life Cycle Analysis
Tom Barber. Senior Staff Chemist
Corporate Environmental Technology
Ciba
410 Swing Road
Greensboro, North Carolina 27409
ABSTRACT
Limited Life Cycle Analysis (LLCA) is a technique for developing a quantitative estimate of the
environmental impact of any action which affects the environment in a positive or negative
manner. It is generally used to compare the environmental impact of various possible actions
such as remediation alternatives for contaminated soil or water, for control options for
emissions sources in various media, or even manufacturing processes. LLCA is based on a
European model developed by Schaltegger and Sturm (S&S) and has been successfully used
in several countries. This paper gives a brief overview of the LLCA process, the mechanism
of the S&S model, and summarizes the results of two case studies where this technique has
been implemented.
INTRODUCTION
The technique considers both direct and indirect emissions created by a given action in a
comprehensive way and examines the environmental impact of each step of the process to
be evaluated from "cradle to grave". LLCA incorporates the elements of full life cycle
analysis (raw materials and energy balances), except it is limited in scope and can be applied
to a set of potential alternative solutions to an environmental problem. The S&S model used
is a quantitative assessment, based upon regulatory limits.
LLCA estimates the impact of each pollutant released to the environment by multiplying the
mass of each pollutant released by a Pollution Factor (PF) for that pollutant. The product is
the impact of that release in dimensionless Environmental Impact Units (ElUs). For a given
action, the ElUs for all pollutants are summed to find the overall environmental impact of the
action. ElUs are directly additive and comparable across pollutants and across environmental
media. The PFs are specific for each pollutant and fpr each environmental medium (air,
water, or soil) and are calculated from the legally or socially accepted limit of concentration
of the pollutant in that medium.
This section provides an overview of how the LLCA process is applied. The most important
step in performing this analysis is to define the boundaries of the process to be tested.
Listed below are the types of information necessary to complete a successful LLCA for a soil
or water remediation project.
• the pollutants of interest
• the total mass of pollutants to be remediated
• the total mass of pollutants produced during construction, operation, and
decommission (the amount of pollutants produced from vehicular emissions, materials
manufacture, electric power generation, etc., are published in various national and
international reference documents)
• the required pollutant cleanup levels in soil and/or water
• regulatory limits for emissions in all media
432
-------
• a flow diagram with equipment specifications
• all process and construction materials
• process energy flows
For example, emissions generated by the following are considered in analyzing an excavate
and thermal treatment soil remediation option:
• transportation of materials and equipment to and from the site
• installation of equipment
• soil excavation (emissions from equipment and contaminants from the soil to air)
• production of fuels consumed in the treatment process
• generation of electrical power used in the treatment process
• production of materials consumed by the remediation
• release of pollutants to groundwater during and after the remediation
SUMMARY
Mechanism
The Schaltegger & Sturm model, which is used as the basis for all LLCA efforts described in
this paper, compares each step of the process for environmental impact using regulatory-
derived discharge, ambient, or emission limits. This section provides a brief overview of the
process with example calculations.
• Pollution factor values are calculated for each compound of interest for all appropriate
media. Compounds of interest include the site contaminants as well as emissions produced
as a result of transportation, construction, and operation of the process.
• PFs are derived from regulatory limits in all appropriate media, and are converted to
mass/mole of substrate. This conversion is quite simple for air and water media; calculation
of mass/mole in soil is particularly difficult, since the molecular weight of soil is not well
defined. However, if the soil in the test area is well characterized, an empirical molecular
weight of the soil can be estimated. Once this has been done, the PF (a dimensionless
number) is calculated based upon the normalization to a reference compound. The
compound used as a reference in the Schaltegger & Sturm Model is carbon dioxide;
however, any reference will produce PFs which are inversely proportional to their regulatory
limits, i.e., the lower the allowable concentration, the higher the PF. The table below
illustrates this concept:
Media & Compound
AIR
Carbon dioxide
so,
WATER
Mercury
Limit Value
579 mg/M3"
0.08 mg/M3
0.002 mg/L
Standardization (mg/mole)
13.7
0.00192
0.000036
PF = ElU/Kg
1
7,140
381,000
The carbon dioxide limit value is the mid 1960's normal sea level concentration in air.
433
-------
Conversion from the limit value to mg/mole for each compound is calculated as follows:
For air: Mg/mole = limit value in mg/M3/42.3 (number of moles in one cubic meter of
air at 15° C & 1 atmosphere)
For water: Mg/mole = limit value in mg/L/55.5 (number of moles in 1L water @ 1 5° C)
Standardization of the limit value for each compound is calculated as follows:
PF = 13.7/limitvalue in mg/mole = PF units = EIU/Kg
• In those cases where federal, state, or local emission or discharge limits have not been
established, alternate sources may be utilized. For example, for a compound for which
there are no ambient air limits, US industrial hygiene threshold limit values (TLVs) may be
utilized with modification. Although TLVs have been established for industrial workers
exposed over a nominal eight-hour time weighted average, some state regulatory
agencies have adopted these modified TLV values (divided by 42 to "normalize" them for
a 7-day, 24 hr/day week with a 10-fold safety factor) as an acceptable emission
alternative in the absence of federal ambient standards. As another viable alternative,
site-specific cleanup limitations derived as part of ongoing investigations may be used in
the same manner.
Once pollution factors are determined for each compound, environmental impact units can be
determined for each discharge or emission source based upon the total mass of component.
The following worked hypothetical example of pollution present in groundwater illustrates this
concept:
Compound
Carbon
Dioxide
Benzene
Toluene
Total Impact
US (NJ) H2O
Cleanup Level
579 mg/MJ
(Air)
0.2 ug/L
1 ,000 ug/L
US Cleanup
Level
(mg/mol)
13.7
3.61X10"
0.018
PF ,or
EIU/Kg
1
3.80X10°
761
Quantity in
Ground (Kg)
8.2
2860
Impact
(EIU)
3.12X10'
2.18X10°
3.34X10'
Note that the PFs derived for each compound are based upon the molar concentration in the
medium of interest, e.g., the ambient limit for CO2 in air is converted to mg/mole as are limits
for compounds in aqueous media. This ensures that equal comparison of the PFs can be
made for air, water, and soil (when appropriate) for each compound of interest. Unless the
soil in the cleanup area is well characterized, molecular weights are not usually assigned.
Rather, it is assumed that the mass of contaminants present in the soil will be completely
transferred to groundwater or air, based upon the remediation mechanism under
investigation. In this manner molecular weight assignments to these media are known.
Case Studies - Site A
Site A is a manufacturing facility in which soil contamination was discovered beneath a
former building. Due to the possibility of groundwater contamination by organic materials,
remediation of the area was required.
434
-------
Four remediation technologies were considered: 1) cap & containment, 2) in-situ aerobic
bioremediation, 3) in-situ anaerobic bioremediation, and 4) excavation and thermal treatment.
LLCA was performed on each remediation alternative and compared to the do nothing or no
action as a baseline.
Major assumptions made during the LLCA included the following:
• PF soil = PF groundwater because a) soil cleanup levels were not established, b)
it was assumed that all pollutants would be transferred to the groundwater, and
c) the molecular weight of the soil in the area was not well established
• the remediation area consisted of an area approximately 200' X 150' X 30'
• Total mass of the pollutants in the area = 3,660 Kg
• remediation times vary with technology
The final results of the study are summarized in the table below; total and local
environmental impacts were also determined. (Local impacts occur at the site itself).
Site A Results
[Million EIU]
Component
Total Impact
Release to GW
Electric consumption
Fuel consumption
Process materials
Construction materials
Local Impact
Release to GW
Electric consumption
Fuel consumption
Process materials
Construction materials
No Action
167
167
0
0
0
0
167
167
0
0
0
0
Cap & Contain
25
17
0
5
0
3
21
17
0
4
0
0
Aerobic
In-situ Bio
61
17
42
<1
<1
2
17
17
0
<1
0
0
Anaerobic
In-situ Bio
75
17
56
<1
<1
2
17
17
0
<1
0
0
Excavate
& Thermal
182
0
49
132
0
1
132
0
0
131
0
1
Based upon the results of the LLCA, cap & containment provides the lowest overall impact,
but not the lowest local effect. However, since this option will not remove the
contaminants, the preferred option is in-situ aerobic bioremediation.
Data was also examined in two ways: 1) a sensitivity analysis was performed on the three
most prevalent EIU contributors and varied by _+_ 10%. The slope of the resulting lines
indicate which components are the most sensitive to change (see example on page 7). 2) a
breakeven analysis for operating times was performed on each option. That is, the impact
for each option was determined on an annual basis and plotted against time. The breakeven
operating time is defined as the point where the total (or local) environmental impact
intersects the EIU for the no action option, i.e., operating times shorter in duration than the
breakeven time less negatively impact the environment than the no action alternative (see
example on page 8).
435
-------
Examination of data from the sensitivity analysis shows electricity consumption to be the
most sensitive to change for the in-situ bioremediation options; fuel consumption was the
most critical to excavation followed by thermal treatment.
For the breakeven analysis, the bioremediation options could operate for > 15 years before
more negative environmental impact is realized. This time, however, is well beyond the
expected cleanup times of 5 and 7 years for aerobic and anaerobic systems respectively.
The thermal treatment option exhibits an operational breakeven time of approximately 90
days, which is less than the expected remediation time of 100 days.
Case Studies - Site B
Site B is an abandoned hazardous waste landfill in which organic compounds were leaching
into a public drinking water supply. A series of purge wells contain the contaminated plume
for onsite remediation.
Two remediation technologies are feasible: 1) vacuum steam stripping and 2) granulated
activated carbon (GAC) adsorption. LLCA was performed on each option and compared to
the do nothing or no action as a baseline.
Major assumptions made during the LLCA included the following:
• operations annualized since the total mass of pollutants in the area is unknown
(annual mass = -12,500 Kg)
• no air emissions observed for either treatment option
• construction was not considered (already built)
• well pumping power was not considered (common to all options)
The final results of the study are summarized in the table below; total and local
environmental impacts were also determined.
Site B Results
[Million EIU]
LLCA Component
Total impact [million EIU]
from groundwater
from process materials
from operating electricity (incl C regen)
from organics destruction
from carbon regeneration
from operating fuel consumption
Local impact [million EIU]
from groundwater
from process materials
from operating electricity (incl C regen)
from organics destruction
from carbon regeneration
from operating fuel consumption
Baseline
(Annual, Reg
Agency limits)
17
17
0
0
0
0
0
17
0
0
0
0
0
0
Vacuum Steam
Strip (Annual)
214
0
2
198
0.5
0.8
12
12
0
0
0
0
0
12
GAC (Annual)
46
0
5
35
0
7
0
0
0
0
0
0
0
0
436
-------
Although the baseline (no action) scenario exhibits the lowest EIU value, this option is
unacceptable. Therefore, based upon the LLCA results of the remaining alternatives, GAC
provides the lowest overall impact with essentially no local impact, and was chosen as the
preferred option.
Examination of data from the sensitivity analysis shows electricity consumption to be the
most sensitive to change for both alternatives.
Since the total mass of contaminants at Site B is not reliably known, a true breakeven
analysis in the strictest sense could not be performed. However, when annualized, it can be
demonstrated that the vacuum steam stripper shows breakeven occurring at approximately 4
weeks operation; GAC at approximately 19 weeks operation.
Reference:
Okologieorientierte Entscheidungen in Unternehmen, Stefan Schaltegger & Andreas Sturm,
2nd Ed. 1994
437
-------
Site A -Aerobic In-situ Bioremediation Sensitivity Analysis
66 T.
CO
CO
-10
•Release to GW
Eleclric Consumption
-Construction Materials
-------
Site A Bioremediation Cleanup Level (Time) Sensitivity Analysis
200 r-~
W
CO
7 10
Years Operating
12
15
-*—Aerobic total impact
•^••••Aerobic local impact
-*—Anaerobic total impact
-X—Anaerobic local impact
•-#?-••- No action
-•— Cap 8 contain
-------
91
REDUCING WASTE GENERATION AND RADIATION EXPOSURE BY
ANALYTICAL METHOD MODIFICATION
Amy A. Ekcchukwu. Senior Research Scientist - A
Savannah River Technology Center
Weslinghouse Savannah River Co.
Aiken, SC 29808
SUMMARY
The primary goal of an analytical support laboratory has traditionally been to provide
accurate data in a timely and cost effective manner. Added to this goal is now the need to
provide the same high quality data while generating as little waste as possible. At the
Savannah River Technology Center (SRTC), we have modified and reengineered several
methods to decrease generated waste and hence reduce radiation exposure. These method
changes involved improving detection limits (which decreased the amount of sample
required for analysis), decreasing reaction and analysis time, decreasing the size of
experimental set-ups, recycling spent solvent and reagents, and replacing some methods.
These changes had the additional benefits of reducing employee radiation exposure and
exposure to hazardous chemicals. In all cases, the precision, accuracy, and detection limits
were equal to or better than the replaced method. Most of the changes required little or no
expenditure of funds. This paper describes these changes and discusses some of their
applications.
INTRODUCTION
Our laboratory provides analytical support for many different types of research programs
within SRTC and throughout the Savannah River Site. A wide variety of sample types
including ground water, organics, laboratory waste, process control, sludge, soils, and
others are received for many different analyses. These samples are both radioactive and
non-radioactive and may contain hazardous materials such as RCRA metals, organics, and
flammable solvents. The sample size is often limited by these chemical hazards or by the
level of radioactivity. Hazardous substituenls present in the sample may prohibit its
disposal to the laboratory drain system and therefore may limit the size of the sample or, in
some cases, may prevent its analysis entirely.
The nature of an R&D support organization is to perform accurate analyses on unknown
and uncharacterized samples and to be able to adapt analysis methods to compensate for
difficult or hazardous sample matrixes. To enable analysis of all samples received while
minimizing hazardous waste generation and radiation exposure, we performed a systematic
evaluation of all methods performed in our Wet Chemistry and Ion Chromatography
Laboratories. Evaluation of the more than thirty methods performed in these labs took
approximately two years and is still continuing. Each method was evaluated for reagent
content, detection limits, sample size, and its necessity in being offered. Improvements
were tested and evaluated on both standards and actual samples. Method changes were
implemented once the validity of the method was proven and documented. The changes
that were made fell into three categories: method modification, reagent recycle or
replacement, and method replacement.
440
-------
DISCUSSION
Modification of Existing Methods
• Ion Chromatography
Ion chromatographic analysis is one of the most frequently requested analyses in our
laboratory, with over 7000 samples being submitted in 1995. This method is used to
determine common anions and cations in aqueous sample matrixes. In this analysis, a
sample is introduced into a flowing stream (mobile phase) which carries the sample
through an ion exchange resin. Ions in the sample are separated by differential
interaction with the resin and quantified through conductivity detection. Originally, the
analysis took approximately 20 minutes per sample run and generated approximately
100 mL of aqueous waste per run. On average, complete determination of the
commonly requested anions (fluoride, formate, chloride, nitrite, nitrate, phosphate,
sulfate, and oxalate) required three different dilutions (three runs) so a typical sample
analysis generated 300 mL of aqueous waste. Cation analysis followed a similar
pattern. Several method changes were tested and implemented which lowered the
waste generated to 20 mL per sample.
The existing suppression system, which was used to decrease or "suppress" the
conductivity of the eluent mobile phase, was a counter-current ion exchange system.
Once the sample had passed through the anion exchange resin (for anion analysis) and
the anions in solution were separate, the eluent stream was passed through a cation
exchange resin. The separated anions passed through unaffected, but the cations in the
eluent (predominantly sodium) were exchanged for hydrogen ions contained in a
counter-current (regenerant) stream. This converted the eluent to a weak acid with very
low conductivity. The regenerant stream flowed at 4 to 5 mL per minute and
represented 80% of the liquid waste generated.
This suppression system was replaced with an electrochemical suppression system
which produces a regenerant stream by electrochemically reducing the waste from the
eluent stream. The self regenerating suppressor (SRS) employs a self-contained
electrochemical membrane and thus requires no preparation of reagents. It is therefore
less time consuming and less costly that the previous system. The cost of the SRS
membranes is the same as that of the counter current suppressor membranes. Minimal
instrument and procedural modification was required. The suppressor system is more
effective at decreasing the background conductivity and so enhances detection limits.
The lower background conductivity enabled modification of the detector to increase the
sensitivity of the analysis by an order of magnitude. This lower detection limit
decreased the required sample volume by 90% with a corresponding decrease in
radiation exposure.
The separation of ions is effected by passage through a low capacity ion exchange
column. The existing columns required 20 minutes for complete separation of anions
and 15 minutes for complete separation of cations. These columns were relplaced by
lower capacity ion exchange columns (called "fast-sep" columns) which effect the same
separation in 5 minutes. Since the instrument generates waste based on the length of an
analysis run, this change decreased the waste generated per sample by 75%. The
precision of the data, +/- 3 to 5% depending on the sample matrix, is equal to that of the
original columns. Since the column is of lower capacity that the previous column,
column damage was of concern when running concentrated samples. However, the
column has proven to be fairly resilient and can be easily cleaned and regenerated after
poisoning by concentrated samples. The column lifetime, 12 to 18 months on average,
is comparable to the previous column. Currently, our lab is testing a smaller diameter
441
-------
column which effects the same separation using one fourth the eluent flow rate and one
fourth the sample volume.
The combined effect of these changes was to decrease the liquid waste generated by
95%, decrease the required sample volume by 90% with a corresponding decrease in
radiation exposure, lower the detection limits by an order of magnitude, and cut the
analysis time by 75%. The precision and accuracy were virtually unchanged.
• Titrations • Microelectrodes and Buffer Volume Reduction
Many of the analyses performed in the wet chemistry laboratory are titrations which use
either potentiometric, pH, or ion selective electrodes to determine the titration endpoinL
Waste reduction can be accomplished by decreasing the size of the analysis system and
by using smaller, more sensitive electrodes (microelectrodes).
The most common titrations performed in the lab are pH titrations to determine acid and
base concentrations and halide determinations using ion selective electrodes. The
standard sample cup used required a minimum sample volume of 25 mL to adequately
cover the surface of the electrode. This cup was replaced with a narrower cup which
required only 10 mL to cover the electrode surface. Data validation tests on standards
and samples showed the precision and detection limits to be comparable to that of the
25 mL volume method. This change, which decreased the liquid waste generated by
60%, was simple to make and required minimal procedure modification and virtually no
analyst training.
Microelectrodes are small working surface electrodes, anywhere from 5 mm to less
than 1 mm in diameter. A standard size electrode, which is usually 12 mm in diameter,
typically requires clearance in solution of twice the electrode diameter in order to obtain
accurate data. If a smaller solution volume is used, the electrode experiences
electrochemical feedback from the sides of the reaction vessel. This feedback distorts
the electrochemical response and corrupts the aquired data. Smaller electrodes require
proportionally less clearance in solution and thus smaller volumes of solution. In
addition, reducing the size of the electrode surface decreases the electrochemical noise
caused by the unequal distance of all points on the surface of the electrode to the
reference or counter electrode. For these reasons, smaller electrodes offer lower
detection limits and require correspondingly lower sample volumes. Micro pH
electrodes have been implemented in pH measurements, as described in the next
section. They are still being evaluated for routine use in the titrations performed in the
laboratory.
• pH Determinations
The measurement of pH, hydrogen ion concentration, is one of the simplest laboratory
measurements to make. It is also one of the easiest to do incorrectly, particularly with
unbuffered samples. The most frequently made error is not allowing the electrode to
equilibrate sufficiently in the sample. Therefore it is important to measure an adequate
sample volume (20 to 25 mL with a standard size 12 mm electrode) and to repeat the
measurement with fresh volumes of sample until successive measurements agree within
0.1 pH unit. In general, 50 to 100 mL of sample arc required to accurately determine
pH.
We replaced the standard size pH electrode which had been in use in the lab with a
smaller 5 mm diameter electrode. This electrode requires only 5 mm clearance to obtain
accurate data. Using a 10 mL vial, accurate pH measurements can be made using 1.5
mL of sample. The waste generated by this analysis was reduced by 95%. A wide
variety of samples between pH 2 and pH 14 was measured using both the standard size
442
-------
electrode and the microelectrode. The pH results agreed within 0.1 pH unit and so
were essentially equivalent.
Reagent Substitution and Recycle
• Karl Fisher Titrant
Determination of volume percent water in liquid samples is accomplished by a
potentiomelric titration commonly known as the Karl Fisher method. An aliquot of
sample is introduced into anhydrous methanol and titrated with a pyridine containing
titrant. The titrant complexes with water, thus changing the solvent potential which is
measured electrochemically. Pyridine is a RCRA listed waste and cannot be introduced
in to the 773-A high or low activity drain systems. To reduce the hazard of the waste
generated, we replaced the pyridine containing Karl Fisher Titrant with a pyridine-free
litrant, hydranal. Hydranal contains a non-hazardous amine which complexes with
water in the same manner as pyridine does in the Karl Fisher titrant. No method
modification or instrument reprogramming was required and so the transition to this
new solvent was essentially effortless. The precision and accuracy of the titration using
hydranal is the same as that obtained using the pyridinc-containing solvent. The
relative standard deviation of the method is 3% in both cases.
The solvent used in the Karl Fisher Method, methanol, is also hazardous from a
flammability standpoint. However, less than 20% of the waste from the method is
methanol, so the total waste stream from the method contains less than 10% total
organic carbon. The total waste stream therefore is non-hazardous. Development work
in progress on this method involves reducing the volume of methanol required by
employing a microvoltammetric system.
• Strontium Chloride
Determination of carbonate and free-hydroxide is accomplished by first precipitating the
carbonate out of solution and then determining the carbonate and free hydroxide
concentrations separately with pH titrations. The standard method in use involved
precipitation of the carbonate with excess barium chloride. Barium is a RCRA listed
waste due to its toxicity and can no longer be introduced into the laboratory's high or
low activity drain system. We replaced the barium chloride with strontium chloride
which is non-hazardous. Method validation experiments showed the strontium to be
equivalent in a variety of sample matrices. Samples commonly run for carbonate and
free-hydroxide were analyzed in duplicate using both methods and the results were
found to be equivalent in all cases. As with the Karl Fisher titrant replacement, no
method changes or instrument modifications were required.
• Freon Recycle
The standard analysis method for determining total petroleum hydrocarbons (commonly
known as oil and grease determination) involves solvent extraction of the hydrocarbons
using Freon followed by quantitation using infrared detection. This had been the
method of choice because it was simple, rugged, inexpensive, and applicable to solid
and liquid samples and radioactive samples. Due to its deleterious effect on the ozone
layer, the use of Freon and other chloro-fluorocarbons (CFCs) has been greatly
restricted. Freon has become very expensive ($800/liter) and will soon be unavailable
entirely. Proposed replacement methods include solid-phase extraction, solvent
extraction, and supercritical fluid extraction all of which use gravimetric determination
or infrared analysis of the extracted hydrocarbons. These methods are not as precise or
as sensitive as the Freon extraction method, they require a larger amount of sample due
to the decreased sensitivity, and the solid phase extraction method cannot be used with
solid samples. Supercritical fluid extraction requires expensive instrumentation
443
-------
($ 100,000), facility services (high pressure ultrapure carbon dioxide), and a lot of
bench top and hood space. This method is therefore not feasible for radioactive sample
analysis. All replacement methods would require procurement of new equipment and
analyst training.
We opted to keep the existing Freon method and recycle the solvent. An inexpensive
solvent reclamation system was procured to reclaim the spent Freon. This reclaimer
removes hydrocarbons from the Freon solvent by passage through an activated carbon
bed. The operation is simple: spent Freon is poured into the top of the instrument and
clean Freon is obtained from a tap at the bottom. The hydrocarbon content of the
reclaimed Freon is measured to ensure its purity. The capacity of the carbon bed is 5 g
of hydrocarbon. This bed is changed out routinely every 3 months or if the
hydrocarbon content of the reclaimed Freon is above 0.2 ppm (twice the minimum
detection limit of the analyzer).
Using the carbon bed unit, 95-97% of Freon can be reclaimed and reused. The
reclaimer has been used successfully for non-radioactive sample analyses for over two
years. A contained unit is currently being tested for radioactive samples.
Instrument Replacements
• High Temperature TOC
The existing instrument for determining total organic carbon (TOC) and total inorganic
carbon (TIC) used a persulfate oxidation method. A sample was oxidized with sodium
persulfate and the carbon in the sample was evolved as carbon dioxide gas. The
evolved gas was quantified using infrared detection. This method had several
undesirable characteristics. First, it used a flow cell reaction chamber which required a
large sample volume (50 to 100 mL) per analysis. The instrument could only
accommodate liquid samples, so the frequently requested analyses on sludges and soil
samples could only be accommodated by making a slurry of the sample and then
aspirating this slurry, while stirring, into the instrument. Sampling variability and
absorption of carbon dioxide while the sample was stirring had a deleterious effect on
both the precision and the accuracy of the data. In addition, the aspirated particulates
frequently clogged the flow cell which contributed to a high percentage of down time
while the instrument was being cleaned and repaired. The analysis method required
hazardous chemicals (sodium persulfate) and generated a high volume of hazardous and
radioactive waste. The instrument was also ten years old and was at the end of its
useful life.
We replaced this system with a Rosemount DC-190 High Temperature Total Organic
Carbon Analyzer. This instrument oxidizes samples by combustion in a 900*C
furnace, so there are no reagents to prepare or dispose. The analyzer is automated and
requires only 50 microliters per analysis. In addition, ADS reconfigured the instrument
by placing a small sampling system and analysis chamber in a radiologically contained
hood. This enabled the analysis of radioactive samples, using as little as 25 microliters
of sample. The contained module can accommodate solid samples so the need to dilute
the samples into a slurry, and the accompanying detection limit and instrument damage
problems, is eliminated. The detection limits of this instrument are 0.1 parts-per-
million, which represents an order of magnitude improvement over the previous
system. The instrument is also capable of determining purgeable organic carbon and
non-purgeable organic carbon, methods which were unavailable with the previous
instrument
444
-------
The overall benefit from this instrument change was a 98% reduction in the total liquid
waste generated, a thousand fold decrease in the required sample size, and an order of
magnitude decrease in the detection limits. The precision was slightly improved, from
+/- 5% to +/- 3% RSD. The contained module is approximately one third the size of
the previous system's contained module. This compactness is a huge benefit
considering the scarceness of radioactive hood space. Unlike the previous instrument,
in which 100 mL of radioactive sample was placed in an open flask while being
aspirated into the instrument, the new contained analysis requires that 25 microliters of
sample be introduced into a sealed furnace. This smaller sample volume combined with
the shielding provided by the furnace significantly decreases radiation exposure to the
analyst.
Flashpoint Tester
Flashpoint is a frequently requested waste characterization analysis. Liquids which
have a flashpoint below 140*F are considered hazardous due to flammability and must
be treated as hazardous waste. We had a Pensky-Martens closed cup flashpoint tester
which is the standard method for flashpoint determination. The temperature of the
sample was raised at a specified rate and a flame was periodically introduced into the
space just above the sample. The flash was detected by a heat sensor and recorded.
This open cup system required 60 mL of sample. Since many of the samples received
for flashpoint are organic, such as diesel fuel and solvents, this method presented a
particular waste disposal problem, especially with radioactive samples. As with the
TOC analyzer, the flashpoint tester had been put into service in 1980 and was at the end
of its useful life. When the unit was taken out of service in 1993, it had failed and
could not be repaired.
This instrument was replaced by a Pelrolab multi-position mini-flashpoint tester. This
unit is an open cup flashpoint tester requiring only 1 mL of sample per analysis. The
flash is sensed by a pressure transducer which monitors the pressure change when the
sample flashes. The system is much smaller (approximately one fourth the bench top
footprint of the previous instrument) and requires no facility services other than 110 V
electrical power. The unit has an autosampler which can accommodate eight different
samples and can be programmed to perform different analysis protocols on each
sample. The single sample unit is even smaller than the multiposition unit and one of
these smaller units has recently been procured to be placed in a radiologically contained
hood.
Open cup flashpoint values are slightly higher than closed cup values. For example,
the closed cup flashpoint of dodecanc, which is the check standard used with the
analysis method to ensure that the instrument is functioning properly, flashes at 165*F
in a closed cup tester and 174"F in an open cup tester. The standard laboratory method,
upon which all hazard data is based, requires a closed cup value. The programming in
the open-cup tester was therefore modified to give closed cup values. The accuracy of
this modified temperature program is verified before each set of samples is analyzed.
The precision of the method is +/- 1 to 3% when analyzing standards and samples.
Reducing the sample size had no effect on the precision or accuracy of the data as long
as the sample cups in the tester are well cleaned between samples.
The overall benefit was a 98% reduction in the amount of waste generated. The
automated system greatly reduced the hands-on analysis time which decreased the
analyst's exposure to chemicals and increased productivity.
445
-------
CONCLUSION AND PATH FORWARD
We have successfully decreased the waste generated by these methods and decreased the
total waste generated by the Ion Chromatography and Wet Chemistry laboratories by 60%.
In all cases, the analytical precision and accuracy remained the same or improved and in
most cases the detection limits were improved. The cost per sample decreased not only by
streamlining methods but also by decreasing the costs associated with chemical
procurement and waste disposal. The benefits of waste reduction and minimizing radiation
exposure were achieved. Current efforts to further decrease waste generated include
additional implementation of microelectrodes and buffer volume reductions in titrations and
solvent recycling.
446
-------
INDEX
Jpe|t$£> <'-
Authors
Page
Numbers
Authors
Page
Numbers
Authors
Page
Numbers
Adams, S. (6O) 269
Alvarado, J. (33) 163
Araki, R. (83) 387
Atkinson, D. (7) 46
Austin, Jr., J. (43) 197
Bao, Y. (86) 405
Baratta, E. (14) 67
Barber, T. (9O) 432
Bathija, B. (82) 386
Bayne, C. (79) 371
Behlke, M. (3O) 145
Bender, S. (10) 49
Billets, S. (9) 48
Bottrell, D. (79) 371
Brar, G. (4) 30
Brillante, S. (61) 276
Bruce, M. (75,84) 342,394
Bucher, S. (32) 156
Burgess, A. (24) 120
Burnett, W. (is) 77
Burrows, R. (62,84) 277,394
Cahill, M. m 15
Calcavecchio, P. (63) 283
Carlin, Jr., F. (31) 148
Carrell, R. (83) 387
Carter, C. (43) 197
Carter, K. (87) 413
Chalk, S. (23) 112
Claff, R. (43) 197
Cline, P. (59) 267
Cocuzza, P. (32) 156
Coe, D. (86) 405
Collins, L (26,64) 139,291
Connolly, G. (89) 422
Cook, D. (70,77) 330,360
Cooke, W. (60) 269
Corbett, D. (is) 77
Cori, III, W. (76) 348
Craig, T. (S) 37
Criscio, J. (36) 179
Crockett, A. (3&7) 22,37,46
Cummings, R. (83) 387
Daggett, M.(6S) 293
Day, W. (70) 330
de Zeeuw, J. (72,73) 339,340
Dodd, D. (42| 196
Drake, E. (63) 283
Duvekot, C. (72) 339
Einberger, C. (24) 120
Ekechukwu, A. (91) 440
Erickson, M. (33) 163
Ezzell, J. (78) 370
Fairbanks, P. (53) 235
Felix, A. (63) 283
Fern, M. (15) 77
Feyerherm, F. (62) 277
Fields, J. (so) 212
Filliben, J. (19) 90
Rax, P. (38) 187
Foust, D. (21) 96
Fribush, H. (34) 171
Friedman, S. (ai) 379
Gagner, S. (3) 22
Gamer, F. (45) '. 204
Gere, D. (85) 396
Gholson, A. (77) 360
Giarroco, V.fssj 396
Gobran, T. (78) 370
Grant, C. (4) 30
Greaves E. (18) 89
Griest, W. (58,89) 258,422
Griggs, J. (i 7) 88
Grosser, Z. (20) 91
Gueco, A. (67) 307
Gui, J. (21) 96
Hall, J. (75) 342
Hanby, J. (8) 47
Harris, D. (35) 172
Harris, M. (29) 144
Hayes, H. (89) 422
Henderson, D. (36) 179
Herwehe, K. (36) 179
Herzog, D. (67,70) 307,330
Hewitt, A. (69) 322
Heynsdijk, P. (72) 339
Higgins, M. (24) 120
Hiller, J. (8) 47
Hodges, C. (86) 405
Hoefler, F. (78) 370
Hoffman, J. (31) 148
Holden, W. (79) 371
Holloway, R. (17) 88
Hosaka, T. (56) 245
Hoy, D. (77) 360
Huo, D. (23) 112
Inn, K. (19) 90
Itak, J. (70) 330
Jackson, A. (39) 188
Jacobs, L. (6) 45
Jenkins, T. (4,5,7) 30,37,46
Kassakhian, G. (37,57) 180,252
Keller, J. (58) 258
Kelley, B. (63) 283
Kelly, K. (26,64,65) 139,291,293
Khurana, V. (78) 370
Kingston, H. (22,23,27) 104,112,140
Kibler, L. (si) 220
Kirshen, N. (86) 405
Klueh, N. (35) 172
Knowles, D. (78) 370
Koglin, E. (9) 48
Kolb, S. (34) 171
Kolloff, R. (85) 396
Krol, J. (25) 128
Laing, G. (45) 204
Lammers, N. (73) 340
Lavey, T. (84) 394
Lawruk, T. (67) 307
Lazarus, L. (38) 187
Leikin, S. (28) 143
Lesnik, B. (74,eo,82) 341,378,386
Link, D. (22,27) 104,140
Litman, R. (16) 87
Liu, C. (17) 88
Lu, C. (33) 163
Lutter, D. (56; 245
Maney, J. (i) 1
Marinissen, J. (73) 340
Maskarinec, M. (58) 258
Mattulat, A. (78) 370
McCarty, H. (82) 386
McCone, S. (4i) 190
McCurdy, D. (17) 88
McKenzie, K. (87) 413
McMillin, R. (65) 293
Melberg, N. (6) 45
Meznarich, H. (42) 196
Miller, M. (2,56) 15,245
Minnich, M. (ii) 50
Moses, M. (28) 143
Moshiri, B. (28) 143
Paper Numbers (OO)
-------
. continued
Page
Authors
Najjar, R. 141) 190
Nixon, G. 166) 301
Noyes, R. (35) 172
Olson, N. (83) 387
Palmer, L. (65) 293
Parr, J. (43j 197
Parris, R. (44^2) 198,228
Patel, B. (29) 144
Peterson. D. (33). 163
Phillips, R. (32) 156
Poster, D. (52) 228
Quick, Jr., A. (82) 277
Ranney, T. m 30
Reitmeyer, C. (6). 45
Richter, B. (78) 370
Rieck, R. (83) 387
Robertson, G. (9,10*5) 48,49,204
Romano, J. f2S) 128
Rosecrance, A. (17,51) 88,220
Roth, D. (25). 128
Rothweiler, B. (85). 396
Rubio, F. (67) 307
Rucker, T. (17) 88
Sajc-Bohus. L. (18) 89
Saiaymeh, S. (IT) 88
Page
Numbers Authors
Page
Numbers
Sander, L (52) 228
Savoia, P. (sa^o) 187,189
Schabron, J.(12) 51
Schantz, M. (30^4.52,68)
145,198,228,317
Scheuering, J. (24) 120
Scheutwinkel, M. (78) 370
Schlender, M. (56)~. 245
Schulz, M. (is) 77
Schumacher, B. (ii) 50
Schumacher, P. (4) 30
Schupp, G. (50) 212
Selisker, M. (70) 330
Sheridan, P. (4O) 189
Shymanski, P. (26) 139
Siegrist, R. (79) 371
Sisk, W. (5) 37
Silzer, J. (33) 163
Smith, R. (25) 128
Snyder, J. (7i) 338
Snyder, W. (85) 396
Solsky, J. (55) 237
Sorini, S. (12) 51
Stalling, D. (64,65) 291,293
Stark, T. (61) 276
Stome, K. (46) 205
Studabaker, W. (87) 413
Sutton, G. (88) 417
Tatro, M. (47) 209
Thacker, R. (34) 171
Thome, P. (4) 30
Tomkins, B. (89) 422
Trovato, R. (54) 236
Turriff, D. (6) 45
Vance, D. (17) 88
Van Gaalen, G. (63) 283
Verstraeten, W. (73) 340
Verstuyft, A. (59) 267
Waddell, D. (78) 370
Walsh, J. (60) 269
Walter, P. (22,23,27) 104,112,140
Warren, J. (49) 211
Wells, R.fiT) 88
Wentworth, N. (48) 210
West, O. (79) 371
West, T. (66) 301
Whetzel, Jr. J. (13) 59
Wiley, P. 161) 276
Wise, S. (30^4,52,68) 145,198,228,317
Wolf, R. (20) 91
Zhichao, L. (19) 90
Paper Numbers (OO)
------- |