Mortal Symposium on
Measuring and Interpreting
VOCs in Soils: State of the Art
and Research Needs
January 12-14,1993, Las Vegas, Nevada
f- WWM I I
Sponsored by
U.S. Environmental Protection Agency
Organized by
Oak Ridge National Laboratory
University of Wisconsin—Madison/Extension
U.S. Environmental Protection Agency
U.S. Department of Energy
U.S. Army Toxics and Hazardous Materials Agency
American Petroleum Institute
-------
National Symposium on
Measuring and Interpreting
VOCs in Soils: State of the Art
and Research Needs
January 12-14,1993, Las Vegas, Nevada
Sponsored by
U.S. Environmental Protection Agency
Organized by
Oak Ridge National Laboratory
University of Wisconsin—Madison/Extension
U.S. Environmental Protection Agency
U.S. Department of Energy
U.S. Army Toxics and Hazardous Materials Agency
American Petroleum Institute
In Cooperation with
Geological Society of America
-------
Section I. Symposium Purpose and Program
-------
CONTENTS
Section I. Symposium Purpose and Program
Symposium Purpose
Symposium Planning Committee
Final Symposium Program
Discussion Questions
Section II. Presenter Information
Where provided by each presenter, enclosed are an extended abstract,
supporting publications, and a presenters biography (in approximate order of
presentation)
Platform
Keynote Remarks
VOC measurements in soils: the nature and validity of the process
Effect of VOC measurement uncertainty on the risk assessment process
Panel Discussion
Measurement needs and uncertainty in the risk assessment process
Processes controlling the transport and fate of VOCs in soils
Soil sampling strategies and the decision making process
Data quality objectives and statistical treatment of soil VOC data
Sampling and analyses for soil VOCs
Field screening and soil gas measurement techniques for VOCs
Standard model for volatilization of chemicals from soil at Superfund sites
VOC contamination in ground water: sources of variability and comparison
soil, well and hydropunch results
Statistical simulation and 3-dimensional visualization for analysis and
interpretation of soil VOC datasets
Comparison of collection and handling practices for the analysis for the
analysis of volatile organic compounds in soils
Experimental determination of maximum pre-analytical holding times for
volatile organics in selected soils
- Siegrist
- Wong
- Lincoln
- Hutzler
- England
- Haeberer
- Barcelona
- Spittler
- Dinan
- Barcelona ...
- Mitchell...
- Hewitt
- Jenkins...
National Symposium on Measuring and Interpreting VOCs in Soil
January 12-14,1993 Las Vegas, Nevada
-------
Section II. Presenter Information (continued)
Evaluation of sample holding times and preservation methods for gasoline
in fine-grained sand
Development of an ASTM standard for sampling soils for VOCs
Soils, synthetics, and screening: may the odds be with you
Geoprobe soil sampling and field VOC analyses by gas chromatography
King
Triegle
Rajagopal
Chapman.
An evaluation of four field screening techniques for measurement of BETX - Amick ...
Application of field VOC data in quantitative risk assessment at CERCLA sites - Kramel ...
Site investigations: the role of field screening & analytical tools - Cornell
Advances in on sit and in situ VOC measurement techniques - Koglin
Laboratory analyses and quality assurance for soil VOCs - Bentley
Interlaboratory study of analytical methods for petroleum hydrocarbons - Parr ...
Modifications to EPA procedures for soil VOC analyses - Lesnik
Posters
Purge-and-trap GC/MS method modifications - Ward
Slow desorption dynamics for volatile organic compounds from five
ion-exchanged smectites - Fairly...
Review of VOC sorption behavior in soils - Minnich
Experimental determination of non steady-state diffusion of o-xylene
from a sandy soil - Lindhardt..
Performance of a new soil sampling tool for use with methanol preservation
of samples containing volatile organic compounds - Turriff
Pre-analytical holding times: advanced data treatment - Bayne...
Active soil gas sampling - collection by air withdrawal - Johnson...
Field identification and quantitation of volatile organics in soils
utilizing fourier transform infrared (FTIR) spectroscopy - Demirgian .
The inadequacy of commonly used risk assessment guidance for determing
whether solvent-contaminated soils can affect ground water at arid sites - Korte ...
Referee analyses - a better approach than data validation - Korte ...
Vapor fortification: a method to prepare quality assurance soils
for the analysis of volatile organic compounds - Hewitt
Field observations of variability of soil gas measurements - Fancher
Estimation of potential VOC emissions during trial excavation
activities via flux chamber and fourier transform infrared open path - Simon
National Symposium on Measuring and Interpreting VOCs in Soil
January 12-14,1993 Las Vegas, Nevada
-------
Section II. Presenter Information (continued)
The effect of barometric pumping on the migration of volatile organic
compounds from the vadose zone into the atmosphere
Use of risk assessment ground water model in installation restoration
program site decisions
A fiber optic chemical sensor for the measurement of TCE
Biodegradation of high concentrations of TCE and effects on
aquifer permeability
Field analysis of VOC's by photoacoustic detection
Evaluation of headspace method for volatile constituents in soils
and sediments
On-site analysis of VOCs in soils by transportable GC/MS
Measuring the flux of chlorinated, volatile organic compounds from
the soil surface
- Pirkle ...
- Goldblum...
- Klainer...
- Rowland...
- McClelland..
- Looney...
- Christenson.
- Daley...
National Symposium on Measuring and Interpreting VOCs in Soil
January 12-14,1993 Las Vegas, Nevada
-------
National Symposium on Measuring and Interpreting VOCs in Soil
January 12-14,1993 Las Vegas, Nevada
-------
SYMPOSIUM PURPOSE
Volatile organic compounds (VOCs) such as trichloroethylene, methylene
chloride, and benzene are among the most prevalent contaminants at hazardous
waste and underground storage tank sites across the United States. Substantial
funds are being expended on VOC sampling and analysis and significant site
assessment decisions and remedial actions are occurring. The conventional
measurement and interpretation process, however, may not adequately address
sources of error that can severely hamper the overall effectiveness of site
assessment and remedial action at VOC contaminated sites.
This symposium was designed to bring together VOC data users and the
generators of that data. Risk assessors and other decision-makers are often
unaware of the inaccuracies and variability in VOC data while the field sampling
and laboratory personnel are often unfamiliar with the basic questions risk
assessors and others must address using the data. This symposium will focus on
the need for VOC data that are appropriate and meaningful for risk assessment
and decision making. The symposium consists of a mixture of platform and
poster presentations, panel discussions, open microphone periods, and special
interest group discussions. During three days, symposium participants will
o Explore the foundation of the conventional VOC measurement and
interpretation process,
o Examine results from research and practice that have advanced the
understanding of this process, including sample collection,
preservation and pre-analytical handling, onsite and in situ
measurement techniques, and data analysis and interpretation, and
o Attempt to develop consensus on current practices, recommendations
for alternative procedures, and critical research needs.
This notebook constitutes the working proceedings for the symposium (Figure 1).
Contained herein are the final program, a series of discussion questions, and
materials provided by each presenter. The information exchanged at the
symposium will be incorporated into a white paper which will be made available
to all symposium attendees later in 1993. In addition, contributions may be
solicited and a peer-reviewed monograph may be prepared to further document
the processes and principals of measuring and interpreting VOCs in soils.
National Symposium on Measuring and Interpreting VOCs in Soil
January 12-14,1993 Las Vegas, Nevada
-------
National Symposium on Soil VOCs
Las Vegas, NV
January 1993
I
Platform and Poster Presentations
Panel Discussions
Open Microphone
Special Interest Group Discussions
Consensus Presentations
and Discussion
White Paper Preparation
March 1993
Proceedings Monograph Preparation
w/ Peer-Reviewed Contributions
• • • • •
Figure 1. Conceptual plan for information exchange, documentation, and use.
National Symposium on Measuring and Interpreting VOCs in Soil
January 12-14,1993 Las Vegas, Nevada
-------
SYMPOSIUM PLANNING COMMITTEE
Robert Siegrist (Symposium Chair)
Oak Ridge National Laboratory,
Environmental Sciences Division
P.O. Box 2008, Oak Ridge, TN 37830-6038.
Telephone = 615-574-7286
Telefax = 615-576-8646
Jeff van Ee
U.S. Environmental Protection Agency
Environmental Monitoring Systems Lab
P.O. Box 93478, Las Vegas, NV 89193
Telephone = 702-798-2367
Telefax = 702-798-2454
Pat Eagan
University of Wisconsin
Dept. of Engineering Professional
Development
432 North Lake Street
Madison, WI 53706
Telephone = 608-263-7429
Telefax = 608-263-3160
Ruth Bleyler
U.S. Environmental Protection Agency
Toxics Integration Branch, Regional
Support Section Mail code 5204G
Washington, DC
Telephone = 703-603-8816
Telefax = 703-603-9104.
David Bottrell
U.S. Department of Energy
Office of Technology Development,
Laboratory Management Branch (EM-532)
Washington, DC 20585-0002
Telephone = 301-903-7251
Telefax = 301-903-7613
Bruce Bauman
American Petroleum Institute
Health & Environmental Sciences
Department
1220 L Street, NW, Washington, DC 20005
Telephone = 202-682-8345
Telefax = 202-682-8270
Martin Stutz
U.S. Army Toxic and Hazardous Materials
Agency, CETHA-TS-C, Bldg.E4435
Aberdeen Proving Ground, 21010-5401
Telephone = 410-671-1568
Telefax = 410-671-1680
Duane Gueder
U.S. Environmental Protection Agency
Superfund QA Office (OS-240)
Washington, DC.
Telephone = 202-260-4027
Telefax = 202-260-3265
Joan Fisk
Los Alamos National Laboratory
Los Alamos, NM
Telephone = 505-667-3269
Telefax = 505-665-5982
Roger Jenkins
Oak Ridge National Laboratory
Analytical Chemistry Division
P.O. Box 2008
Oak Ridge, TN 37831-6120
Telephone = 615-576-8594
Telefax = 615-576-7956
Eric Koglin
U.S. Environmental Protection Agency
Environmental Monitoring Systems Lab
P.O. Box 93748, Las Vegas, NV 89193-3478
Telephone = 702-798-2432
Telefax = 702-798-2637
Mike Maskarinec
Oak Ridge National Laboratory
Analytical Chemistry Division
P.O. Box 2008, Oak Ridge, TN 37831-6120
Telephone = 615-576-8594
Telefax = 615-576-7956
National Symposium on Measuring and Interpreting VOCs in Soil
January 12-14,1993 Las Vegas, Nevada
-------
National Symposium on Measuring and Interpreting VOCs in Soil
January 12-14,1993 Las Vegas, Nevada
-------
FINAL PROGRAM
National Symposium on
Measuring and Interpreting VOCs in Soils:
State of the Art and Research Needs
January 12 - 14,1993, Las Vegas, Nevada
Day 1 - Tuesday, January 12,1993
7:30 am Onsite Registration
Session 1: Opening Session
Session Chairs: Bob Siegrist, Oak Ridge National Laboratory
Jeff van Ee, U.S. Environmental Protection Agency
8:30 am Welcome and Introduction
Wayne Marchant, Director, U.S. EPA Environmental Monitoring
Systems Laboratory, Las Vegas, NV
8:40 am Symposium Organization and Purpose
Bob Siegrist, Oak Ridge National Laboratory
Jeff van Ee, U.S. EPA Environmental Monitoring Systems Lab
9:00 am Keynote Remarks
Dave Bennett, Chief, Toxics Integration Branch, Hazardous Sites
Evaluation Branch, U.S. EPA Office of Emergency and Remedial
Response
Joan Fisk, Chairperson, Interagency Steering Committee for
Quality Assurance for Environmental Measurements, U.S. EPA
Office of Solid Waste and Emergency Response
Session 2: VOC Measurement Needs, Issues, and Concerns
Session Chairs: Jeff van Ee, U.S. Environmental Protection Agency
Pat Eagan, University of Wisconsin-Madison
9:30 am VOC measurement in soils: the nature and validity of the process
Bob Siegrist, Environmental Sciences Division, Oak Ridge
National Laboratory, Oak Ridge, TN
National Symposium on Measuring and Interpreting VOCs in Soil
January 12-14,1993 Las Vegas, Nevada
-------
10:00 am Effect of VOC measurement uncertainty on the risk
assessment process
Jeff Wong, Office of the Science Advisor, Toxics Substances
Control, California EPA, Sacramento, CA
10:30 am Break - refreshments provided
10:50 am Panel Discussion and Open Microphone
Diane Easley, Environmental Scientist, U.S. EPA Region 7, Kansas City, KS
Dan Stralka, Toxics Integration Coordinator, Region 9, U.S. EPA
Barry Lesnik, U.S. EPA Office of Solid Waste, Washington, DC
Charles Van Sciver, Chief, Environmental Measurements Section,
Department of Environmental Protection, Trenton, NJ
Allen W. Verstuyft, Chevron Research and Technology Company,
Richmond, CA
David Lincoln, Director of Risk Assessment, CH2M-HU1, Bellevue, WA
Elly Triegel, President, Triegel & Associates, Inc., Pittsburgh, PA
James Bentley, Vice President, Enseco Laboratories, Sacramento, CA
A1 Tardiff, Program Manager, DOE Office of Technology Development,
Washington, DC
12:00 pm Lunch - meal provided at the hotel
Session 3: Soil VOC Measurements and Decision Making
Session Chairs: Ruth Bleyler, U.S. Environmental Protection Agency
Duane Gueder, U.S. Environmental Protection Agency
1:00 pm Measurement needs and uncertainty in the risk assessment process
David Lincoln, CH2M-Hill, Bellevue, WA
1:30 pm Processes controlling the transport and fate of VOCs in soils
Neil Hutzler, Michigan Technological University, Houghton, MI
2:00 pm Soil sampling strategies and the decision making process
Evan England, U.S. EPA Environmental Monitoring Systems
Laboratory, Las Vegas, NV
2:30 pm Data quality objectives and statistical treatment of soil VOC data
Alfred Haeberer, Quality Assurance Management Staff, U.S. EPA
Office of Research and Development, Washington, DC
3:00 pm Break - refreshments provided
3:30 pm Sampling and analyses for soil VOCs
Michael Barcelona, Western Michigan University, Kalamazoo, MI
National Symposium on Measuring and Interpreting VOCs in Soil
January 12-14,1993 Las Vegas, Nevada
-------
4:00 pm Field screening and soil gas measurement techniques for VOCs
Thomas Spittler, Region I Laboratory, U.S. Environmental
Protection Agency, Lexington, MA
4:30 pm Special Interest Group Sessions (Concurrent)
1. Sampling and Analysis Planning
IA. Facilitator = Ruth Bleyler
IB. Facilitator = David Lincoln
IC. Facilitator = Dean Neptune
5:30 pm Poster Session
Social hour and informal discussion (cash bar)
7:00 pm Adjourn for the day
Day 2 - Wednesday, January 13,1993
Session 4: Soil VOC Behavior and Measurement Implications
Session Chairs: Bruce Bauman, American Petroleum Institute
Martin Stutz, U.S. Army Toxics & Hazardous Materials
Agency
8:00 am Review of VOC sorption behavior in soils
Marti Minnich, Lockheed Environmental Systems and
Technology, Las Vegas, Nevada
8:20 am The persistence of several volatile organic compounds in a low organic
carbon calcareous soil from southern Nevada
Spencer Steinberg* and David Kreamer, ^Department of
Chemistry, University of Nevada, Las Vegas, NV
8:40 am Standard model for volatilization of chemicals from soil at Superfund sites.
Janine Dinan, U.S. EPA Office of Emergency and Remedial
Response, Washington, DC
9:00 am VOC contamination in ground water: sources of variability and
comparison of soil, well and hydropunch results
Michael Barcelona*, Allan Wehrmann, Jane Denne, and Dannette
Shaw, ^Western Michigan University, Kalamazoo, MI
9:20 am Statistical simulation and 3-dimensional visualization for analysis and
interpretation of soil VOC datasets
Toby Mitchell*', Olivia West, R.L. Siegrist, *Eng. Physics & Math
Division, Oak Ridge National Laboratory, Oak Ridge, TN
National Symposium on Measuring and Interpreting VOCs in Soil
January 12-14,1993 Las Vegas, Nevada
-------
9:40 am Open microphone
10:00 am Break - refreshments provided
Session 5: Sample Collection and Handling for Soil VOCs
Session Chair: Dave Bottrell, U.S. Department of Energy
10:20 am Comparison of collection and handling practices for the analysis of volatile
organic compounds in soils
Alan Hewitt, U.S. Army Cold Regions Research Laboratory,
Hanover, NH
10:40 am Experimental determination of maximum pre-analytical holding times for
volatile organics in selected soils
Roger Jenkins, Chuck Bayne, Mike Maskarinec, L.H. Johnson, S.K.
Holladay> and B.A. Tomkins, Oak Ridge National Laboratory,
Oak Ridge, TN
11:00 am Evaluation of sample holding times and preservation methods for gasoline
in fine-grained sand
Paul King, P&D Environmental, Oakland, CA
11:20 am Development of an ASTM standard for sampling soils for VOCs
Elly Triegle, Triegle & Associates, Inc., Pittsburgh, PA
11:40 am Open Microphone
12:00 pm Lunch - meal provided at the hotel
Session 6: Measurements of Soil VOCs by In Situ & Onsite Techniques
Session Chair: Roger Jenkins, Oak Ridge National Laboratory
1:00 pm Soils, synthetics, and screening: may the odds be with you
R. Rajagopal, University of Iowa, Iowa City, IA
1:30 pm Geoprobe soil sampling and field VOC analyses by gas chromatography
Hunt Chapman and Jeff Tuttle, Envirosurv, Inc., Arlington, VA
1:50 pm An evaluation of four field screening techniques for measurement
ofBETX
E.N. Amick and J.E. Pollard, Lockheed Engineering & Sciences
Co., Las Vegas, NV
National Symposium on Measuring and Interpreting VOCs in Soil
January 12-14,1993 Las Vegas, Nevada
-------
2:10 pm Application of field VOC data in quantitative risk assessment at CERCLA
sites
Ruth Kramel and Anthony Armstrong, Health and Safety
Research Division, Oak Ridge National Laboratory, Oak Ridge,
TN
2:30 pm Site investigations: the role of field screening & analytical tools
Fred Cornell, Environmental Management, Inc., Princeton, NJ
2:50 pm Advances in on site and in situ VOC measurement techniques
Eric Koglin, U.S. EPA Environmental Monitoring Systems
Laboratory, Las Vegas, NV
3:10 pm Break - refreshments provided
Session 7: Laboratory Sample Analyses for Soil VOCs
Session Chairs: Barry Lesnik, U.S. Environmental Protection Agency
Angelo Caraseo, U.S. Environmental Protection Agency
3:30 pm Laboratory analyses and quality assurance for soil VOCs
Jim Bentley, Enseco Labs, Sacramento, CA
3:50 pm Interlaboratory study of analytical methods for petroleum hydrocarbons
Roger Claff*, Dianna Kocurek, Jeff Lowry and Jerry Parr,
* American Petroleum Institute, Washington, DC
4:10 pm Modifications to EPA procedures for soil VOC analyses
Barry Lesnik, U.S. EPA Office of Solid Waste, Washington, DC
4:30 pm Open microphone
5:00 pm Special Interest Group Sessions (Concurrent)
2. Sample Collection and Handling
2A. Facilitator = Dave Bottrell
2B. Facilitator = Duane Gueder
3. Data Assessment
3A. Facilitator = Joan Fisk
3B. Facilitator = Martin Stutz
5:30 pm Poster session
Social hour and informal discussion (cash bar)
7:00 pm Adjourn for the day
National Symposium on Measuring and Interpreting VOCs in Soil
January 12-14,1993 Las Vegas, Nevada
-------
Day 3 - Thursday January 14,1993
Session 8: State of the Art and Research Needs
Session Chairs: Bob Siegrist, Oak Ridge National Laboratory
Jeff van Ee, U.S. Environmental Protection Agency
8:00 am Special Interest Group Session Presentations
1. Sampling and Analysis Planning
Ruth Bleyler
David Lincoln
Dean Neptune
2. Sample Collection and Handling
Dave Bottrell
Duane Gueder
3. Data Assessment
Joan Fisk
Martin Stutz
9:30 am Break - Refreshments provided
10:00 am Panel Commentary
Diane Easley, Environmental Scientist, U.S. EPA Region 7, Kansas
City, KS
Dan Stralka, Toxics Integration Coordinator, Region 9, U.S. EPA
Barry Lesnik, U.S. EPA Office of Solid Waste, Cincinnati, OH
Charles Van Stiver, Chief, Environmental Measurements Section,
New Jersey Department of Environmental Protection,
Trenton, NJ
Allen W. Verstuyft, Chevron Research and Technology Company,
Richmond, CA
David Lincoln, Director of Risk Assessment, CH2M-Hill,
Bellevue, WA
Ely Triegel, President, Triegel & Associates, Inc., Pittsburgh, PA
James Bentley, Vice President, Enseco Laboratories, Sacramento,
CA
A1 Tardiff, Program Manager, DOE Office of Technology
Development, Washington, DC
11:30 am Open Microphone
12:00 pm Closing Remarks and Symposium Adjournment
National Symposium on Measuring and Interpreting VOCs in Soil
January 12-14,1993 Las Vegas, Nevada
-------
POSTER SESSION
The poster session will run from 1:30 pm on Tuesday through 7:00 pm on
Wednesday. The formal poster session hours are from 5:30 to 7:00 pm on
Tuesday and Wednesday.
Purge-and-trap GC/MS method modifications
Steve Ward, Harry Reid Center for Environmental Studies, University of
Nevada, Las Vegas, NV
Slow desorption dynamics for volatile organic compounds from five ion-exchanged
smectites
Jerry Fairly* amd Spencer Steinberg, "Department of Geoscience and the
Water Resources Management Program, University of Nevada, Las Vegas,
NV
Experimental determination ofnon steady-state diffusion of o-xylene from a sandy soil
B. Lindhardt and T.H. Christiansen, Technical University of Denmark,
Department of Environmental Engineering, Lyngby, Denmark
Performance of a new soil sampling tool for use with methanol preservation of samples
containing volatile organic compounds
David E. Turriff, En Chem, Inc., 1795 Industrial Drive, Green Bay, WI
Preanalytical holding times: advanced data treatment
Chuck Bayne*, Denise Schmoyer, Roger Jenkins, "Computing and
Telecommunications Division, Oak Ridge National Laboratory, Oak Ridge,
TN
Active soil gas sampling - collection by air withdrawal
Samuel Johnson and T.V. Prasael, The Advent Group, Inc., Brentwood, TN
Field identification and quantitation of volatile organics in soils utilizing fourier
transform infrared (FTIR) spectroscopy
J. Demirgian*, M. Clapper-Gowdy, G. Robitaille, ""Analytical Chemistry
Division, Argonne National Laboratory, Argonne, IL
The inadequacy of commonly used risk assessment guidance for determining whether
solvent-contaminated soils can affect groundwater at arid sites
Nic Korte", Pete Kearl, T.A. Gleason, and J.S. Beale, "Environmental Sciences
Division, Oak Ridge National Laboratory, Grand Junction, CO
Referee analyses - a better approach than data validation
Nic Korte* and David Brown, Environmental Sciences Division, Oak Ridge
National Laboratory, Grand Junction, CO
National Symposium on Measuring and Interpreting VOCs in Soil
January 12-14,1993 Las Vegas, Nevada
-------
Vapor fortification: a method to prepare quality assurance soils for the analysis of volatile
organic compounds
Allan Hewitt, U.S. Army Cold Regions Research and Engineering Lab,
Hannover, NH
Field observations of variability of soil gas measurements
Jon Fancher, Westinghouse Hanford Co., Richland, WA
Estimation of potential VOC emissions during trial excavation activities via flux
chamber and fourier transform infrared open path transform
Michelle Simon, U.S. EPA Risk Reduction Eng. Lab, Cincinnati, OH
The effect of barometric pumping on the migration of volatile organic compounds from
the vadose zone into the atmosphere
Robert Pirkle*, Douglas Wyatt, Van Price, and Brian Looney, "Microseeps,
Pittsburgh, PA
Use of risk assessment groundwater model in installation restoration program site
decisions
David Goldblum*, John Clegg, John D. Erving, Sverdrup Environmental, San
Antonio, TX
A fiber optic chemical sensor for the measurement of TCE
Marcus Butler, Stanley Klainer*, Kisholov Gosvami and Jonahtan Tussey,
TiberChem, Inc., Las Vegas, NV
Biodegradation of high concentrations of TCE and effects on aquifer permeability
M.artin A. Rowland, New Orleans, LA
Field analysis of VOC's by photoacoustic detection
John McClelland, R.W. Jones,, and S. Ochiai, Ames Laboratory, Iowa State
University, Ames, IA
Evaluation ofheadspace method for volatile constituents in soils and sediments
Brian Looney, C.A. Eddy, W.R. Sims, Westinghouse Savannah River
Company, Aiken, SC
On-site analysis of VOCs in soils by transportable GCfMS
Jeff Christensen and Dave Quinn, Viking Instruments, Reston, VA
Measuring the flux of chlorinated, volatile organic compounds from the soil surface
Paul Daley* and Stan Martins, "Environmental Restoration Division,
Lawrence Livermore National Laboratory, Liver more, CA
National Symposium on Measuring and Interpreting VOCs in Soil
January 12-14,1993 Las Vegas, Nevada
-------
Section II. Presenter Information
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 1
Symposium Organization and Purpose
Bob Siegrist, Environmental Sciences Division, Oak Ridge National
Laboratory, Oak Ridge, Tennessee; and Jeff van Ee, USEPA,
Environmental Monitoring Systems Laboratory, Las Vegas, Nevada
January 12-14, 1993
Las Vegas, Nevada
-------
VOC MEASUREMENT IN SOILS: THE NATURE AND
VALIDITY OF THE PROCESS
Robert L. Siegrist
Oak Ridge National Laboratory1
Environmental Sciences Division
EXTENDED ABSTRACT
A wide variety of volatile organic compounds (VOCs) such as toluene and
trichloroethylene routinely appear as principal pollutants in contaminated sites across the
USA and abroad (Table 1). As part of the site assessment and environmental restoration
process, determination of soil VOCs is required. While accurate and precise
characterization of a region of interest is often desired, this is difficult due to inherent
uncertainty in measuring and interpreting VOCs in soils. Soil VOC concentrations are
subject to large spatial and temporal variations and significant measurement errors (Figure
1). For example, short-range spatial variations in soil properties important to VOC
transport and fate of an order of magnitude or more can occur within a contaminated region
(Table 2). In addition, measurement of some soil VOCs at a discrete point and time have
been shown to be subject to -100% to +25% bias (Figures 2 and 3). Despite the
uncertainty and error potential of the current process, the assumption is often implicitly
made that soil VOC data are sufficient for the intended purpose.
The current VOC measurement process is illustrated in Figure 4. Underlying this
illustration is the current soil VOC paradigm that emphasizes the character and quality of
discrete data points by employing bulk soil sampling for laboratory analyses and data
validation (Figure 5). This paradigm is flawed and in need of change. Is it reasonable to
collect a 1 to 5 g soil subsample from a discrete location, analyze it for a suite of organic
compounds of widely different properties, and then scrutinize the data point to determine its
"quality"? The costs associated with soil VOC measurements can be great (e.g., $150 to
$500 or more per sample) and seemingly not commensurate with the overall quality of the
data generated (e.g., -100 to +25% measurement error). Due to the time and cost
associated with conventional methods, limited observations (of overall questionable
accuracy and precision) are often made, and inferences about large regions remain
uncertain. Nevertheless, far-reaching decisions are being made based on sparse VOC
datasets, comprised of data of questionable quality.
Changes in the VOC paradigm are necessary and appropriate. While there is not
complete understanding, there are insights which should be incorporated now. These
insights affect all aspects of the measurement and decision-making process. A few
thoughts follow.
1 Oak Ridge National Laboratory is managed by Martin Marietta Energy Systems, Inc.
under contract DE-AC05-840R21400 with the U.S. Department of Energy. This abstract
was for presentation at the National Symposium on Measuring and Interpreting VOCs in
Soils: State of the Art and Research Needs, Las Vegas, NV, January 12-14, 1993.
1
-------
First, it seems appropriate to reconsider and refine the definition and categorization of
VOCs. The current class of VOCs was developed based on analytical considerations, yet
sampling considerations would suggest that multiple classes may be appropriate. This
categorization should be based on the environmental behavior of each soil VOC as well as
its environmental and public health significance. For example, trichloroethylene would
likely be grouped in a separate category from ethylbenzene. Such a re-categorization would
support field screening techniques as outlined below.
We should recognize that we do not have the ability or resources to characterize VOCs
in soil accurately and precisely. Moreover, we probably don't need to in order to support
most decision making. Soil VOC data vary widely in space and time; an accurate measure
at one location and time provides only limited insight into concentrations at an adjacent
location or time. Soil VOC data are log-normally distributed and it seems reasonable to
consider that we should temper our view of discrete data points to account for this fact.
Perhaps VOCs should be viewed much the same as soil bacteria or soil pH.
Changes in measurement planning need to include far greater emphasis on collection of
spatially disperse data of "acceptable quality" to enhance the overall understanding of the
question(s) to be answered. This can only be accomplished using streamlined sampling
and analysis strategies, such as probe techniques, in situ detectors, and field analytical
methods. Older screening procedures (e.g., simple headspace screening with hand-held
photoionization detectors) which have been mostly qualitative, should not be used to
characterize the nature and quality of field generated data. Advancements in sample
acquisition equipment and field analytics need to be incorporated in quantitative decision
making; we must recognize that meaningful data can be achieved with improved field
instruments and procedures (e.g., field-portable gas chromatographs). Continuing
developments in chemical and immunobioassay test kits will likely provide attractive
advantages for field analysis of VOCs, including some of the most problematic. Finally,
samples for VOC analysis should also be analyzed for other characteristics which are
important to understanding the occurrence, transport, and fate (e.g., water content, organic
carbon content).
Improvements in sample collection and handling for soil VOC quantitation need to be
incorporated into the measurement process. Recent research and practice have revealed
alternative methods which can improve VOC measurement accuracy by an order of
magnitude or more for the more volatile/low solubility compounds (e.g.,
trichloroethylene). One method involves collection of undisturbed soil cores in sleeve-
lined, split-barrel samplers. The relatively undisturbed soil cores are sealed within the
sleeves onsite and then transported to a laboratory for controlled subsampling and transfer
to an analysis vessel. This method eliminates field subsampling and containerization and
maintains the 1 to 5 g subsample used for analysis in continuity with a bulk soil volume
until analysis is imminent. However, this approach requires shipment of larger quantities
of material, subsampling by someone unfamiliar with the site and a small subsample (i.e.,
1 to 5 g) is still analyzed. Another method involves onsite subsampling with a micro-
coring device to minimize soil disturbance. The soil from the micro-cores (e.g., 5 to 10
mL) can be extruded into a 40-mL glass vial with an o-ring sealed cap that is designed to
attach directly to a purge and trap instrument. This approach eliminates laboratory
subsampling, maintains low detection limits and does not requiring field handling of
chemicals. However, the sample volume analyzed is quite small (1 to 5 g) and compositing
of soil samples is precluded. Yet another method involves immediate onsite immersion of a
soil sample in an organic solvent (e.g. methanol) contained in a Teflon-sealed glass vial or
jar. The methanol approach has the advantage of increasing the sample size analyzed
(thereby attenuating short-range spatial variability) and also enables sample compositing.
However, the methanol addition can increase the detection limits by a factor of 10 to 100
2
-------
and requires field handling and transportation of hazardous chemicals. An alternative
solvent (e.g., water) could mitigate this latter problem.
Minimizing pre-analytical holding time and variability of conditions is critical to help
reduce measurement error. Soil samples must either be analyzed upon collection (e.g.,
field laboratory) or more rigorously preserved than that provided by simple 4°C
refrigeration (e.g., infield solvent immersion)Improved and expanded use of onsite
analytical instruments and techniques provide great benefits in this regard.
Sample analysis should focus on robust methodologies that provide reasonable
accuracy and precision for problem situations (e.g., soils with non-aqueous phase liquids,
tightly sorbed VOCs, etc.) rather than the ability to reach ever lower detection limits.
Simplified, but effective, field instruments and procedures should be employed as far as
possible.
Data assessment and interpretation must be done carefully and include review of not
only sample collection, handling and analysis procedures, but also all information about the
site physical condition, contamination release, and exposure scenario. VOC measurements
made years ago are probably more suspect than recent data due to recent improvements in
practices. The emphasis on formal data validation on a discrete sample/analyte basis is ill-
founded and does little to enhance overall decision making. Alternatives to formal data
validation of laboratory analyses need to be conceived and evaluated (e.g., performance
evaluation materials, laboratory splits). Finally, measurement interpretation must integrate
and reconcile the various elements of a soil VOC dataset (e.g., field screening, on-site lab
analysis, and off-site lab analysis). After all, the interest is for high quality decisions, not
just isolated high quality data.
Knowledge of VOC behavior in soils, the measurement process, and its relationship to
decision making has continued to expand and advances have been made in equipment and
instrumentation. Yet, the soil VOC measurement paradigm remains largely unchanged.
The stakeholders (e.g., researchers, practicing professionals, regulators, private industry)
need to collectively change the understanding of and expectations from die measurement
process. This presentation will provide an overview of VOC contaminated soils and the
current measurement process and discuss its validity and need for improvement.
3
-------
ORNl OWG 91M-152K
spatial
100 m2
Figure 1. Example spatial and temporal variability and measurement error potential within
contaminated regions of interest
Figure 2. VOC concentrations in soil samples as a function of sampling method as
determined in a laboratory experiment with a sandy soil (Sampling method
attributes: D = disturbed soil, U = undisturbed soil, PB = plastic bag,
TG=Teflon-sealed glass, LHS=low container headspace, HHS=high container
headspace, 4C=4°C holding, and 4C-M=methanol immersion plus 4°C)
[Siegrist and Jenssen, 1990].
4
-------
Soil VOCs (ug/kg)
10000
1000
100 --
10
Soil Headspace GC (ppm TCE)
100
10
Sampling Method
H Conventional RFI
EH Modified Purge and Trap
^3 Infield Immersion in MeOH
"O" Headspace GC
0.1
A[07] B[07] C[07] A[17] B[17] C[17]
Boring Location [Depth (ft)]
Figure 3. VOC concentrations in soil samples from a silt and clay deposit as a function of
sample collection and handling methods [Unpublished data of Siegrist et al,
1992].
Figure 4. Conventional measurement process for VOCs.
5
-------
ORNL DWG91M-12513
Figure 5. Representation of the current paradigm for soil VOC measurement and
interpretation.
6
-------
Table 1.
Some volatile organic compounds included within the Hazardous Substance
List (HSL) in the USA and their occurrence at Superfund Sites.
Prrvrv»rtif»cl
Hazardous
Compound
Molecular
weight
Boiling
point
Vapor
pressure
Aqueous
solubility
waste site
rank &
occurrence^
g/mol
°c
mm
mg/L
Vinyl Chloride
62
-13.9
2660 (25°C)
1100 (25°C)
23 [8%]
Methylene Chloride
85
40
349
20000
18 [10%]
Acetone
58
56.2
270 (30°C)
Miscible
1,1 -Dichloroethene
97.0
31.9
500
20 [9%]
1,1 -Dichloroethane
99.0
57.3
180
5500
19 [10%]
trans-1 ^-Dichloroethene
97.0
48
200(14°C)
600
17 [12%]
Chloroform
119.4
62
160
8000
6 [20%]
1,2-Dichloroethane
99.0
83.5
61
8690
25 [7%]
2-Butanone
72.1
79.6
78
353000 (10°C)
1,1,1 -Trichloroethane
133.4
71/81
100
4400
8 [17%]
Carbon Tetrachloride
153.8
76.7
90
800
27 [7%]
Trichloroethene
131.5
86.7
60
1100 (25°C)
1 [35%]
Benzene
78.1
80.1
76
1780
5 [23%]
2-Hexanone
100.2
128
2
35000
Tetrachloroethene
165.8
121.4
14
150 (25°C)
9 [17%]
1,1,2,2-Tetrachloroethane
167.9
146.4
5
2900
Toluene
92.1
110.8
22
515
3 [27%]
Chlorobenzene
112.6
132
8.8
500
26 [7%]
Ethylbenzene
106.2
136.2
7
152
15 [12%]
Styrene
104.1
145.2
5
300
m-Xylene
106.2
139
6
14 [13%]
o-/p-Xylene
106.2
144.4
5
175
ti
2 Rank (highest = 1) and prevalence (% of sites) based
substances found at the 888 Superfund sites (as of October
in ( ).
on a total of 466 different
1986).
Table 2. Typical distributions of soil properties important to VOC behavior [Smith and
Charbeneau, 1990].
Soil property
Distribution
Mean *
Std. Dev. 1
Total volumetric porosity Normal 0.373 0.09
Volumetric water content Log-normal 0.048 2.
Fractional organic carbon Log-normal 0.028 3.17
1 Geometric mean and standard deviation are given for log-normal distributions.
7
-------
REFERENCES
1. Barcelona, M.J. 1989. In: Principles of environmental sampling. Keith, L.H.,
Ed., American Chemical Society, Washington, D.C. pp 3-23.
2. Barth, D.S., Mason, B.J., Starks, T.H. and Brown, K.W. 1989. Soil sampling
quality assurance user's guide. EPA 600/8-89/046. U.S. EPA Environmental
Monitoring Systems Lab, Las Vegas, NV.
3. Black, C.A. et al. 1965. Methods of Soil Analysis. American Society of
Agronomy, Inc., Madison, Wisconsin, pp 1-72.
4. Chiou, C.T. 1989. Theoretical considerations of the partition uptake of nonionic
organic compounds by soil organic matter. In: Reactions and movement of organic
chemicals in soils, Sawhney, B.L. and Brown, K., Eds. Soil Science Society of
American; Madison, WI, 1989; pp 1-29.
5. "Field Screening Methods Catalog, User's Guide", EPA/540/2-88/005.
Environmental Protection Agency, September 1988. (FM-5) Volatile Organic
Compound Analysis Using GC with Automated Headspace Sampler.
6. Flotard, R.D., Homsher, M.T., Wolff, J.S., and Moore, J.M. 1986. Volatile
organic analytical methods - performance and quality control considerations. In:
Quality Control In Remedial Site Investigations: Hazardous and Industrial Solid
Waste Testing, Fifth Volume, ASTM STP 925, C.L. Perket, ed., ASTM,
Philadelphia, pp. 185-197.
7. Hern, S.C. and Melancon (ed.). 1986. Vadose zone modeling of organic pollutants.
Lewis Publishers, Inc., Chelsea, MI. 295 p.
8. Kjeldsen, P. and Larsen, T. 1988. Sorption af organiske stoffer i jord og
grundvand. Laboratoriet for Teknisk Hygiejne, Danmarks Tekniske Hojskole,
Lynby. 85 p.
9. Lewis, T.E., Crockett, A.B., Siegrist, R.L., and Zarrabi, K. 1991. Soil Sampling
and Analysis for Volatile Organic Compounds. EPA/540-4-91/001. Superfund
Ground-Water Issue. U.S. Environmental Protection Agency, Environmental
Monitoring Systems Laboratory, Las Vegas, NV.
10. Maskarinec, M.P., Baiyne, C.K., Jenkins, R.A., Johnson, L.H., and Holladay,
S.K. 1992. Stability of volatile organics in environmental soil samples. ORNL/TM-
12128, Oak Ridge National Laboratory, Oak Ridge, TN 37831. 81 p.
11. Mason, B.J. 1983. Preparation of soil sampling protocol: Techniques and
strategies. EPA-600/4-83-020. U.S. Environmental Protection Agency,
Environmental Monitoring Systems Laboratory, Las Vegas, NV.
12. Nyquist, J.E., Wilson, T.H., Norman, L.A. and Gammage, R.B. 1990. Decreased
sensitivity of photoionization detector total organic vapor detectors in the presence of
methane. Am. Ind. Hyg. Assoc. J. 51(6): 326-330.
13. Robbins, G.A., Bristol, R.D. and Roe, V.D. 1989. A field screening method for
gasoline contamination using a polyethylene bag sampling system. Ground water
monitoring review, National Water Well Association, Dublin, OH. Fall, 1989, pp
87-97.
14. Roy, K.A. 1990. Analytic technique measures aromatics in soil and water. Hazmat
World. December, 1990, pp 52-54.
15. Sawhney, B.L. and Brown, K. (eds). 1989. Reactions and movement of organic
chemicals in soils, Sawhney, B.L. and Brown, K., Eds. Soil Science Society of
American; Madison, WI, pp 271-304.
16. Siegrist, R.L. and Jenssen, P.D. 1990. Evaluation of sampling method effects on
volatile organic compound measurements in contaminated soils. Environmental
Science & Technology. 24(9): 1387-1392.
17. Siegrist, R.L. 1991. Measurement error potential and control when quantifying
volatile hydrocarbon concentrations in soils. In: Hydrocarbon Contaminated Soils,
Lewis Publishers, Inc., Chelsea, MI. pp 205-215.
8
-------
18. Siegrist, R.L. 1992. Volatile organic compounds in contaminated soils: The nature
and validity of the measurement process. J. Hazardous Materials. Vol. 29, p. 3-15.
September 1991.
19. Smith, V.J. and Charbeneau, R.J. 1990. Probabalistic soil contamination exposure
assessment procedures. Journal of Environmental Engineering. Vol. 116, No. 6,
1143-1163.
20. Smith, P.G. and Jensen, S.L. 1987. Assessing the validity of field screening of soil
samples for preliminary determination of hydrocarbon contamination. Proc.
Superfund '87, Washington D.C. Hazardous Materials Control Institute, Silver
Spring, MD, 20910. pp. 101-103.
21. Urban, M.J. et. al. 1989. In Fifth Annual Waste Testing and Quality Assurance
Symposium, U.S. Environmental Protection Agency: Washington, D.C, 1989; pp
11-87 to n-ioi.
22. U.S. EPA. 1986. Test Methods For Evaluating Solid Waste. SW-846. 3rd. ed.
Volume IB: Laboratory manual, Physical/Chemical Methods, Chapter Four -
Organic Analytes. Office of Solid Waste and Emergency Response, Washington,
D.C. 20460.
23. Van Ee, J.J., Blume, L.J. and Starks, T.H. 1990. A rationale for the assessment of
errors in the sampling of soils. EPA/600/4-90/013, U.S. Environmental Protection
Agency, EMSL, Las Vegas, NV.
-------
BIOGRAPHICAL SYNOPSIS
Robert L. Siegrist, Ph.D., P.E.
Oak Ridge National Laboratory, P.O. Box 2008, Oak Ridge, TN 37831-6038
Telephone = 615-574-7286. Telefax = 615-576-8646. E-mail = BS7@ORNLSTC
Robert Siegrist received his Ph.D. in Environmental Engineering from the University of
Wisconsin in 1986 after completing a M.Sc. and B.Sc. in 1975 and 1972, respectively.
He has held positions with the University of Wisconsin, Ayres Associates, Inc., the
Norwegian Centre for Soil and Environmental Research, The University of Tennessee, and
Oak Ridge National Laboratory. Since joining the Environmental Sciences Division of
ORNL in 1990, he has been responsible for direction and conduct of environmental
engineering research with sponsorship by the U.S. Department of Energy, Department of
Defense, Environmental Protection Agency, and others. Current research efforts are
focused on contaminant behavior and measurement in environmental systems, and in situ
treatment techniques including vapor stripping, chemical oxidation, and bioremediation.
Dr. Siegrist has published over 60 works in symposia proceedings and professional
journals and has participated as conference chair and planning committee member for
several national symposia. He has served as an advisor and technical expert for state and
federal agencies in the USA, Canada and abroad and is providing broad technical support
to the DOE Integrated Program for In Situ Remediation. A registered Professional
Engineer, Dr. Siegrist is a member of several professional societies and organizations.
10
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 2
Keynote Remarks
Dave Bennett, USE PA Office of Emergency Response and Remedial
Response, Washington, DC; and Joan Fisk, Los Alamos National
Laboratory, Los Alamos, New Mexico
January 12-14, 1993
Las Vegas, Nevada
-------
c) '("3c?
National Symposium on Measuring and
interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 3
VOC Measurement, Interpretation, and Decison-Making: The Nature
and Validity of the Process
Bob Siegrist
Environmental Sciences Division, Oak Ridge National Laboratory,
Oak Ridge, Tennessee
January 12-14, 1993
Las Vegas, Nevada
-------
fh&^do Xl. A^ytOcJ^
&-?A JU^uJ^sl^ttLzrcPj Jjr<3~^0
^T_
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 4
Effect of VOC Measurement Uncertainty on the Risk Assessment
Process
Jeff Wong
Department of Toxics Substances Control, California EPA,
Sacramento
/^U
%
r /
^ 7 lj iU] /^U~ '
tfh
January 12-14, 1993
Las Vegas, Nevada
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 5
Panel Discussion and Open Microphone
January 12-14, 1993
Las Vegas, Nevada
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 6
Measurement Needs and Uncertainty In the Risk Assessment
Process
David Lincoln
CH2M HILL
Bellevue, Washington
January 12-14, 1993
Las Vegas, Nevada
-------
)\v°
0
MEASUREMENT NEEDS AND UNCERTAINTY
IN THE RISK ASSESSMENT PROCESS
David Lincoln
CH2M HILL
P.O. Box 91500
BeUevue, WA 98009
INTRODUCTION
Uncertainty is a major technical and social challenge for decisionmaking at hazardous
waste sites. It leads to many rounds of sampling when the objectives of data gathering
are unclear. It leads, to indecision and lack of progress when multiple remedial
alternatives appear to meet the objectives of a cost-effective remedy. This paper
addresses critical risk assessment questions that must be addressed during site
remediation, sources of significant uncertainty in addressing those questions, and risk
management strategies containing different definitions of data sufficiency and limitations,
which have been developed to respond to uncertainty. It concludes by describing the
observational method, a risk management strategy that has been applied for decades by
geotechnical engineers for construction projects in the subsurface soil, an environment
with substantial uncertainty.
PROBLEM STATEMENT
There are three principal questions that must be answered during the remedial action
process:
• Does the site warrant remedial action? This is typically addressed by the
baseline risk assessment, where the "no-action" scenario is considered
during the remedial investigation.
What is the degree of remediation required? This is typically addressed
at several points of the process through the establishment of remediation
action goals. EPA in the National Contingency Plan set an excess lifetime
cancer risk range of 10* to 10"1 as a goal. In April, 1991, the Office of
Solid Waste and Emergency Response changed the recommended decision
criterion for taking action to .a cancer risk level of about 104 or more
(EPA, 1991a).
-------
• What treatment/containment schemes should be put into place to achieve
the required level of remediation? To answer this question, four critical
issues must be identified: critical parameters that control the selected
technologies; the range of critical parameter values (i.e., those that will
significantly affect technology performance) at the site; the response of the
technologies to the site's range of critical parameter values; the standards
of performance, reliability, efficiency, and effectiveness; and methods for
confirming the performance during remediation. These risk management
factors should be addressed during the feasibility study. They may also
be addressed during the remedial action if new conditions are discovered.
Each of these questions requires a form of a risk assessment, which is a systematic
means of addressing the potential human health and environmental consequences of an
action. The assessment methods applicable for Superfund have been described by EPA
(1989b). The method requires the following elements, each of which has substantial
uncertainties (Water Science and Technology Board, 1990):
• Locate and quantify all significant contamination. Subsurface drilling and
direct sampling are still the primary methods for locating and identifying
quantities of wastes, and the physical characteristics of the subsurface
environment. These methods provide a general stratigraphic site
characterization, but even extensive sampling covers only a small fraction
of the total site volume, making it impossible to say with confidence that
all significant contamination has been identified. Improved laboratory
analytical methods are reducing the detection limits, but the application of
the results is still strongly dependent on the representativeness of the
samples collected.
• Estimate contaminant fate and transport. Current fate and transport
models are subject to substantial uncertainties, leading to differences
between predictions and observed results. Processes are often poorly
understood, site conditions are often inadequately specified, and parameter
values for site conditions are likely largely unknown. Degradation
products may be more hazardous than the chemicals originally released.
Problems in locating and characterizing nonaqueous phase liquids are a
recent example of our state of knowledge.
-------
• Estimate exposure. The dose depends on the receptor activities (e.g,
residential, commercial industrial), conditions of the exposure media (soil,
air, water, biota), and the exposure route (ingestion, inhalation, dermal
contact). Uncertainties cover inadequate knowledge of receptor activities,
lack of models for some exposure routes, and unknown chemical-specific
parameter values.
• Estimate hazardous/lexicological properties. The science of toxicology is
still in its infancy. There are a variety of potential effects, but most
studies to date have focused on cancer. Substantive toxicological
information is currently available for only a small fraction of the total
number of chemicals in commercial use (National Research Council,
1984). Synergistic and antagonistic effects among the multiple chemicals
found at a hazardous waste site can be expected, but are largely unknown.
• Estimate risks to humans and the environment. This element integrates
the information from the above elements, and thus" will be subject to all
their uncertainties.
• Predict the performance of the remedial.technologies. Waste site remedial
technologies, in general, have,not been proven to be effective, and remain
in their infancy. Performance with waste streams of known composition,
or laboratory tests does not guarantee field performance.
These substantial uncertainties have allowed opposing parties to develop different, but
equally valid remedial actions because current technologies cannot provide defensible
answers to the three questions above. Because "high stakes" (tens of millions of dollars
in remedial costs, human health and environmental effects) are involved, the remedial
decisions are highly contentious and prolonged.
RISK MANAGEMENT
There are three common strategies to risk management for decisions under uncertainty;
• Retreat, or return to a "risk-free" life by designing on the basis of average
conditions, for example. Sufficient data, for example, have been collected
when the estimate of the coefficient of variation for the average meets the
design objective. This strategy can be dangerous, however, if
heterogeneity is a major influence in system response.
-------
• Control, or overwhelm Ihe uncertainties by assuming "worst-case"
conditions, for example. Data sufficiency, for example, is obtained when
the likelihood of detecting an extreme event meets the design objective.
In hazardous waste site remediation this strategy can be very expensive for
data collection and remediation.
• Comprehend, or study the issues and their uncertainties. This strategy is
discussed below in some more detail.
"More research" is a common means of implementing the "comprehend" strategy. In
the case of hazardous.waste sites, however, delay in remediation will likely make the
problem worse.
Methods have been developed, to aid In decisionmaking, under uncertainty in support of
the "comprehend" strategy. For project scoping, the EPA has developed the data quality
objectives (DQO) process (EPA, 1991b) to enhance the specification of survey design
requirements. It provides an orderly process to address a thorough and efficient
consideration of current information, specific objectives to be addressed, constraints, and
decision objectives in developing a sampling design. Data sufficiency is obtained in
planning when the lowest cost design is obtained to meet acceptable decision error rates.
Decision analysis (Raiffa, 1970; Freeze, et ai., 1990) also focuses on the. specific
decision to be addressed while providing a probabilistic framework for considering
existing information and the responses of nature. Data sufficiency is obtained when the
incremental cost of the next data collection exceeds the value obtained in distinguishing
between decision options.
These methods are useful tools to organize information and address uncertainty explicitly,
but by themselves they contain several limitations: fundamental data are inadequate,
causal processes are poorly understood, future conditions are difficult to address, and the
inclusion of social values is difficult. Another risk management strategy, the
observational method, has been developed since the early 1900's by geotechnical
engineers to respond to these limitations (Peck, 1969). It has been used successfully for
decades on geotechnical projects (e.g., tunnels, dams), and has been adapted to the
hazardous waste sipe remediation (Brown, et aL, 1990).
OBSERVATIONAL METHOD
The key elements of the observational method have been described (Peck, 1969):
a. Exploration sufficient to establish at least the general nature, pattern and
properties of the deposits, but not necessarily in detail.
-------
b. Assessment of the most probable conditions and the most unfavorable
conceivable deviations from those conditions.
c. Establishment of the design based on a working hypothesis of behavior
anticipated under the most probable conditions.
d. Selection of quantities to be observed as construction proceeds and
calculation of their anticipated values on the basis of the working
hypothesis.
e. Calculation of values of the same quantities under the most unfavorable
conditions compatible with the available data concerning the subsurface
conditions.
f. Selection in advance of a course of action or modification of design for
every foreseeable significant deviation of observational findings from those
predicted on the basis of the working hypothesis.
g. Measurement of quantities to be observed and evaluation of actual
conditions.
h. Modification of design to suit actual conditions.
The observational method recognizes uncertainty explicitly and therefore does not use the
"retreat" strategy. It avoids the limitations of the "control" strategy by designing on the
basis of probable conditions and including a careful consideration of alternative
conditions (deviations), continued investigation through design implementation, and the
provision of pre-determined contingency plans. It avoids the limitations of the
"comprehend" strategy by its emphasis on continued data collection during design
implementation. "Full site characterization" is unnecessary for data- collection.
Limitations of its own, however, have been described (Peck, 1969). The observational
method has also been recently expanded to include the operations phase following
construction (D'Appolonia, 1990).
Under the observational method, data sufficiency for proceeding with a remedial action
is obtained when, the critical uncertainties can be addressed by deviations that can be
detected in the field, and a contingency plan for each deviation. EPA has endorsed the
observational method (EPA, 1989a), and the U.S. Department of Energy has recently
integrated the observational method and the DQO process under the term SAFER
(Dailev, et al, 1992).
-------
CONCLUSIONS
Risk assessments contain substantial uncertainty, but are a systematic means of
summarizing the available information. A risk management approach that explicitly
recognizes these uncertainties shouSd be chosen as toe rr.ast cost-effective means of
prcvidisg, health and eiwirotiraentai protecTion. Befmtkjm of data needs will be
determined by thechosen risk rnaragenent Mrategy. Tbe observational method has been
described for Hazardous waste site rernediations based on its iuccesiful impl&r similar.
?ot itcaoej \n geotethcACsl eaguoeericg, projects
-------
BIBLIOGRAPHY
Brown, S.M., D.R. Lincoln, and W.A. Wallace. 1990. Application of the
Observational Method to Hazardous Waste Engineering. 3. Management in Engineering
6:479-500.
Dailey, R., D. Lillian, and D. Smith. 1992. Streamlining Approach for Environmental
Restoration (SAFER): An Overview. Proceedings of the 1992 Waste Management and
Environmental Science Conference, San Juan, Puerto Rico. April 9-11.
D'Appolonia, E. 1990. Monitored Decisions. J. Geoiechnicai Engineering 116:4-34.
Freeze, R.A. J. Massman, L. Smith, T. Spelling, and B. James. 1990.
Hydrogeological Decision Analysis. 1. A Framework. Ground Water. 28:738-766.
National Research Council. 1984. Toxicity Testing: Strategies to Determine Needs and
Priorities. National Academy Press, Washington, D.C.
Peck, R. B. 1969. Advantages and Limitations of the Observational Method.
Geotechnique. 19:171-187.
Raiffa, H. 1970. Decision Analysis: Introductory Lectures on Choices Under
Uncertainty. Addison-Wesley, Menlo Park, California.
U.S. Environmental Protection Agency, Office of Solid Waste and Emergency Response.
1989a. Rl/FS Streamlining. OSWER Directive 9355.3-06. February 14.
U.S. Environmental Protection Agency, Office of Solid Waste and Emergency Response.
1989b. Risk Assessment Guidance for Superfund. Human Health Evaluation Manual,
Part A. OSWER Directive 9285.7-0la.
U.S. Environmental Protection Agency, Office of Solid Waste and Emergency Response.
1991a. Role of the Baseline Risk Assessment in Superfund Remedy Selection Decisions.
OSWER Directive 9355.0-30. April 22.
U.S. Environmental Protection Agency, Quality Assurance Management Office. 1991b.
Planning for Data Collection: The Quality Objectives Process for Environmental
Decisions. Draft Guidance. October.
Water Science and Technology Board, National Research Council. 1990. Ground Water
and Soil Contamination Remediation: Toward Compatible Science, Policy, and Public
Perception. National Academy Press, Washington, D.C.
-------
APPLICATION OF OBSERVATIONAL METHOD
to Hazardous Waste Engineering8
By Stuart M. Brown,1 Member, ASCE, David R. Lincoln,'
and William A. Wallace3
Abstract: Uncertainly is a key technical factor in hazardous waste site re me-
diations. It can lead to unreasonable data gathering exercises if the point of di-
minishing information returns is not recognized. Engineering under uncertainty,
however, is not unique to hazardous waste site remediation. Approaches have been
used elsewhere to recognize and respond to substantial uncertainty. The observa-
tional method, traditionally applied in geotechnical engineering, has a number of
key elements applicable to hazardous waste site remediation. The key contributions
of the observational method are: (I) Remedial design based on most probable site
conditions; (2) identification of reasonable deviations from those conditions; (3)
identification of parameters to observe so as to detect deviations during remedia-
tion; and (4) preparation of contingency plans for each potential deviation. This
paper describes an approach for incorporating the observational method into the
current USEPA Superfund process and provides a detailed discussion of that pro-
cess in the context of ground-water remediation. Explicitly recognizing uncertainty
in a proper application of the observational method offers the opportunity to reduce
project time and costs as well as risks.
Introduction
The objective of this paper is to describe the application of the observa-
tional method to hazardous waste site remediation. Although this paper fo-
cuses on the Superfund process developed by the U.S. Environmental Pro-
tection Agency (USEPA), the observational method is more broadly applicable
to other site remediations. It is important to note, however, that both reg-
ulatory requirements and our ability to implement the observational method
will change over time.
The principal feature of the observational method is its explicit recognition
of uncertainty. We believe that this contributes positively to the remediation
process by recognizing a "fact of life." It is not possible to characterize a
site "fully" before remediation begins. It is more realistic to perform a rea-
sonable investigation and prepare a contingency plan for detecting and re-
sponding to the new information that will invariably change the site under-
standing during remediation.
'Presented at the November 28-30, 1988, Superfund '88 Conference, held at
Washington, D.C.
'Div. Mgr., Industrial and Envir. Services, CH2M HILL, Inc., Portland Office,
2020 S.W. Fourth Ave., Portland, OR 97201.
3Dir. of Risk Assessment, CH2M HILL, Inc., Seattle Office, 777 108th Avenue
NE, Bellevue, WA 98004.
3Vice Pres. and Dir. of Hazardous Waste Mgmt., CH2M HILL, Inc., Seattle Of-
fice, 777 108th Avenue NE, Bellevue, WA.
Note. Discussion open until March I, 1991. To extend the closing date one month,
a written request must be filed with the ASCE Manager of Journals. The manuscript
for this paper was submitted for review and possible publication on December 27,
1989. This paper is part of the Journal of Management in Engineering, Vol. 6,
No. 4, October. 1990. ©ASCE, ISSN 0742-597X/90/0004-0479/S1.00 + $.15 per
page. Paper No. 25153.
479
-------
In the following sections we:
1. Describe the current Superfund paradigm of study-design-build and the'
technical uncertainties in the remediation process.
2. Describe the observational method, which we believe holds particular promise
for site remediation under Superfund and other hazardous waste site remedia-
tions.
3. Compare the current Superfund process with the value-added features of
the observational method for ground water remediation.
Study, Design, Build
The current management process For Superfund site remediation developed
by the USEPA follows a traditional engineering paradigm of study, design,
and build. Following a discussion of project scope, the client's expressed
objectives, budget, operating assumptions, and initial data, the client ini-
tiates a remedial investigation (Rl) of the situation and prepares a feasibility
study (FS) that compares the alternatives for remediation. The client pro-
poses and the USEPA selects an alternative (involving the USEPA record
of decision—the ROD), and a consultant designs it (remedial design—the
RD). Firms offer bids on the facility's construction (remedial action—the
RA), and then construction proceeds according to the remedial design (see
Fig. I). Both RD and RA are likely to involve supplemental site investi-
gations.
The objective of this paradigm is to reduce uncertainties early in the life
of the project. It is reasoned that funds invested at the investigation and study
phases can potentially save large expenses later. A body of experience and
standard practice has developed in the traditional engineering services (e.g.,
designing and building roads, bridges, airports, and sewage treatment plants)
such that uncertainty is reduced to manageable levels at the feasibility study
phase.
At hazardous waste sites, however, the high potential risk to public health
and the environment combined with the high cost of remediation demand
extraordinary accuracy in the work performed. Any errors in the choice or
implementation of the-remedial alternative can lead to substantial. residual
health and environmental risks, or to substantial financial liability for the
client. The high cost of being wrong defines the degree and intensity of
effort one should put into site characterization. The current process is a man-
ifestation of this urgent need for precision and accuracy in site character-
ization.
Uncertainties in Hazardous Waste Remediation
The process becomes lengthy, cumbersome, and expensive as standard
practice comes up against technical uncertainties (hat arc characteristically
480
REMEDIAL INVESTIGATION
Propel Ptormtng
Collect exUlta
dolo.
Identify Inlilal
project/operable
urtf. response
scerxjilos A
remedial octlon
objectives.
initiate ARAR
Identification.
Piepate project
ptons.
Site Characterization
Conduct field
Investigation.
Define natue and
extent of
contamination.
Identify contcrolnant
ft location of specific
Aft AGs.
Conduct baseOne risk
assessment.
Refine remedkJ
ociiongocrfs.
Treatability
Investigations
Perform bench or
pilot treata&roty
tests.
Development of I Screening of
Alternatives | Alternative*
Oetaited Analysis
of Alternatives
Identify potentlcrt i Screen cftematlves.
treatment tecfnotogles, '
contatnment/ctsposat | Preserve ranoeof
requirements. 1 options.
Screen technologies. 1 identify
. 1 octtorvspectnc
Develop alternatives, j ARAR*.
1—
Refine
tflernatlves.
Analyze
alternatives.
FEASIBILITY STUDY
Monitoring
Coflecl data during
Implementation.
Remedld Action
Prepare bid
documents.
Construct action.
Operate action.
Remedki Design
Design remedy.
Perform additional
support activities.
Design tnvesllgcrton
Perform treat aba ty
studies.
Collect oddllonctf
field data.
Perform oddtlond
support services.
Record of Decision
Select remedy.
SoBcllpUbOc
comment.
Adopted from USEPA 1988.
FIG. 1. Current Superfund Process
several orders of magnitude greater at hazardous waste sites than for other
kinds of projects. Uncertainty is a major technical and societal issue for
Superfund, beginning with site characterization. From a technical perspec-
tive, the subsurface environment presents very substantial uncertainty. It is
a heterogeneous, complex environmen* in which small subsurface features
or changes in geologic conditions can have substantial impacts on water and
chemical movement. Major uncertainties also plague source characterization,
assessment of chemical fate and transport in the environment, assessment of
exposure risks and health effects, and remedial action performance. Taken
481
-------
Time -*
FIG. 2. Comparison of Levels of Unoertalnty
together, these factors make uncertainty an inherent feature of hazardous
waste sites.
The consequences of this uncertainty for the traditional engineering par-
adigm should be considered early in Superfund site remediation.
1. It is generally assumed that more study will reduce uncertainty. But to date
it has not been fully recognized that the marginal value of further studies at
Superfund sites declines rapidly. At some point, more study does not lead to
better information. Fig. 2 qualitatively compares hazardous waste engineering to
traditional engineering.
2. The implicit goal has been to design the "ultimate remedy" that can be
"walked away from" following construction. But in most cases, it will not be
possible to walk away from a Superfund site. No matter what the chosen alter-
native, continued monitoring.will be required.
This state of uncertainty should not lead to inaction. Uncertainty is not
unique to hazardous waste problems. Engineering has frequently had to re-
spond to similar situations. The engineering community must bring the cor-
rect paradigm to Superfund site remediation.
Observational Method
Karl Terzaghi, a soil mechanics engineer, first developed systematic pro-
cedures for engineering under conditions of uncertainty. He called these pro-
cedures "the observational," "experimental," or "leam-as-you-go" method.
Geotechnical engineers have used the observational method for many years
to work with the physical uncertainties in soils and foundations problems.
The following lengthy excerpt from Terzaghi's 1945 text on applied soil
4B2
mechanics is repeated here because of its applicability to the Superfund con-
text (Peck 1969):
In the engineering for such works as large foundations, tunnels, cuts,
or earth dams, a vast amount of effort and labor goes into securing
only roughly approximate values for the physical constants that appear
in the equations, Many variables, such as the degree of continuity of .
important strata or the pressure conditions in the water contained in
the soils, remain unknown. Therefore, the results of computations are
not more than working hypotheses, subject to confirmation or modi-
fication during construction.
In the past, only two methods have been used for coping with the
inevitable uncertainties: either to adopt an excessive factor of safety,
or else to make assumptions in accordance with general, average ex-
perience. The designer who has used the latter procedure has usually
not suspected that he was actually taking a chance. Yet, on account
of the widespread use of the method, no year has passed without sev-
eral major accidents. It is more than mere coincidence that most of the
failures have been due to the unanticipated action of water, because
the behavior of water depends, more than on anything else, on minor
geological details that are unknown.
The first method is wasteful; the second is dangerous. Soil mechanics,
as we undertand it today, provides a third method which could be called
the experimental method. The procedure is as follows: Base the design
on whatever information can be secured. Make a detailed inventory of
all the possible differences between reality and the assumptions. Then
compute, on the basis of the original assumptions, various quantities
that can be measured in the field. For instance, if assumptions have
been made regarding pressure in the water beneath a structure, com-
pute the pressure at various easily accessible points, measure it, and
compare the results with the forecast. Or, if assumptions have been
made regarding stress-deformation properties, compute displacements,
measure them, and make a similar comparison. On the basis of the
results of such measurements, gradually close the gaps in knowledge
and, if necessary, modify the design during construction.
Soil mechanics provides us with the knowledge required for practical
application of this "leam-as-you-go" method.
Peck (1969) summarized the key elements in the practice of the obser-
vational method (Peck 1969):
a. Exploration sufficient to establish at least the general nature, pattern
and properties of the deposits, but not necessarily in detail.
b. Assessment of the most probable conditions and the most unfavor-
able conceivable deviations from these conditions.
c. Establishment of the design based on a working hypothesis of be-
havior anticipated under the most probable conditions.
483
-------
d. Selection of quantities to be observed as construction proceeds and
calculation or their anticipated values on the basis of the working
hypothesis.
e. Calculation of values of the same quantities under the most unfa-
vorable conditions compatible with (he available data concerning
the subsurface conditions.
f. Selection in advance of a course of action or modification of design
for every foreseeable significant deviation of the observational find-
ings from those predicted on the basis of the working hypothesis.
g. Measurement of quantities to be observed and evaluation of actual
conditions.
h. Modification of design (o suit actual conditions.
The nature and complexity of ihe work will determine the degree to which
all of these elements are included. Some engineering projects have been
initiated with the observational method, and it has been used on others as
the only way out of a current situation (e.g., construction has started and
some unexpected event has occurred).
The observational method is not applicable if a design cannot be altered
during construction. It also should not be applied if the monitoring and re-
sponse to one of the potential deviations costs more than a more conservative
design.
Failures of the observational method can occur under several conditions:
1. Failure lo anticipate unfavorable conditions. This failure will leave a proj-
ect without a course of action identified in advance, and there may be no avail-
able response to the current situation. A corollary of ihis is that the observational
method should nol be started if a contingency plan cannot be identified for all
potenlial, significant deviations.
2. Failure lo choose and interpret the correct quantities to observe. If the mea-
sured quantity does not address what is of real concern, then it may fail to give
appropriate warnings. The results of the observations must also be reliable. (Peck
explicitly suggests that whoever plans the monitoring program should have sub-
stantial field experience.) The fietcf results must be examined promptly, and the
field team should not feel compelled to wait for a fully documented report to
be prepared. The results must be presented in a thoughtful manner, reflecting
on potentially significant events, not just filling in a table.
3. Failure to consider the influence of progressive failure. Progressive failures
may be relatively small and undetected until something snaps and a massive
failure occurs.
The observational method fundamentally recognizes that uncertainty is
present and uses a structured approach to determine the appropriateness of
the design as it is being implemented. It requires planning for potential un-
favorable conditions and potenlial design modifications.
484
The next section describes how the observational method can be applied
to ground-water remediation within the structure of Ihe USEPA Superfund
program.
Ground-Water Remediation
Here we will compare the observational method to the usual approach to
ground-water remediation under Superfund. We hope to demonstrate that the
observational method offers many advantages.
Current Process
Groundwater remediation currently proceeds by the following steps (see
Fig. t).
Project Planning
This activity typically involves a review of available hydrologic and geo-
logic information. Based on this information, an initial conceptual model of
the site hydrogeology is developed, with a work plan for detailed charac-
terization of site conditions.
Remedial Investigation
The remedial investigation is initiated by installing soil borings and mon-
itoring wells to further characterize site hydrogeology. Multiple rounds of
well installation and analysis of data may be required before site hydro-
geology is "fully characterized."
Further rounds of installing and sampling monitoring wells will be nec-
essary to characterize the nature and extent of ground-water contamination.
Because the interpretation of site hydrogeologic conditions often changes as
more information becomes available, nature and extent characterization is
rarely completed until site characterization is completed. As will be dis-
cussed later, the RI rarely has a clear endpoint because of the inherent dif-
ficulty in characterizing site hydrogeology and nature and extent of contam-
ination at the level of detail required to address questions brought out in
later steps of Ihe process.
The remedial investigation concludes by documenting site characterization
results. One difficulty that often occurs at this point is that the process of
documenting results and analyzing site hydrogeology can lead to Ihe iden-
tification of uncertainties and associated data gaps that must be filled prior
to moving forward in the process.
Feasibility Study
The feasibility study involves screening remedial technologies and assem-
bling feasible technologies into remedial alternatives for detailed analysis.
485
-------
At this point alternative methods for ground-water remediation arc eval-
uated, ranging from no action (i.e., natural attenuation) to ground-water ex-
traction, wiili or without the help of barriers or in situ treatment. The FS
has evolved from an evaluation of conceptual alternatives la, in some cases,
an evaluation of alternatives at the predesign level. Issues that normally need
to be addressed -during the FS are:
1. Number and location of wells and pumping rales required to capture and
remove contamination from ground water.
2. Variation in contaminant concentrations in extraction well water with time
tfot use m treatment plant design).
3. Time required to complete grouitd-water remediation.
Our current ability to- address each issue quantitatively decreases as one
moves from determining the well configuration required to capture and con-
tain a plume to estimating the time required for remediation. Increasingly,
ground-water models are being used to predict contaminant concentrations
and remediation times. The modeling results, however, are often compro-
mised by a lack of data on site bydrcgeology and (he processes controlling
chemical migration and fate.
Record of Decision
When the FS is complete, tJne USEPA prepares a Record of Decision (ROD)
and a public hearing is held.
Remedial Design
The remedial design step includes predesign and final design of the se-
lected remedy.
If ground-water pumping is part of the remedy, one or more test extraction
wells and monitoring wells are often installed to evaluate aquifer response
to long-term pumping.
Aquifer testing is used to refine estimates of well spacing and extraction
rates. Water quality sampling during testing is used to refine treatment plant
sizing and design, and potential discharge requirements. Ground-water mod-
eling will probably be used at this point to readdress the three issues noted
in the feasibility study section, with, it is hoped, a higher level of certainty.
In some cases, modeling. will be used 10 optimise well locations and pump-
ing rates, as well as to define monitoring requirements.
The RD also involves completion of the final design. AH remaining un-
certainties need tq be addressed at this point. For this reason, there is a high
probability that some additional site investigation work may be required.
Remedial Action
When the RD is completed , contract documents and bid specifications can
be prepared. The remaining activities include procuring a contractor, con-
structing the remedy, testing the system, placing the sysiem in operation,
486
and initialing monitoring. This step ends when it is determined that reme-
diation is complete.
Two principal factors have affected the cost and time required to conduct
ground-vjatet remediation under the current process.
1. Uncertainty in subsurface conditions. Hydrogeologic systems are highly
complex and difficult to characterize. Although hydrogeologists can develop rea-
sonable conceptual models of site-specific conditions based on a knowledge of
hydrologic and geologic processes and limited investigation, detailed models are
difficult to develop within reasonable time and budget constraints. The desire to
"Jully characterise" site conditions establishes an ongoing need to investigate
throughout (he entire process because there are always uncertainties.
2. Uncertainties in chemical behavior. Our current understanding of how
chemicals behave in soil and ground water is limited. In most cases, the key
driving force far chemical migration is the movement of water, which is difficult
to characterize for the reasons just mentioned. Small-scale heterogeneities in soil
and aquifer materials can affect the rate and direction of both water and con-
taminant movement. Compounding the problem ere poorly understood chemical
transport and fate mechanisms, such as:
a. Sorption, ion exchange, and precipitation-dissolution, which act to take
chemicals out of solution -and retard contaminant movement.
b. Dispersion, which acts to reduce chemical concentrations as a result of
spreading,
c. Chemical diffusion into low-permeability zones, which acts to delay the
rate of migration and removal.
d. Migration of immiscible-phase contaminants.
These uncertainties tend to prolong ground-water remediation under the
current process. Initiating action is difficult because of the desire to reduce
uncertainties through additional site characterization and analysts. As tine next
section will discuss, it may be possible to initiate action earlier using the
observational method.
Application of Observational Method
The elements of the observational method should be considered at each
phase of the RI/FS/RD/RA process, and possibly several times within a
phase. New data at any phase may lead to a new conceptual model, requiring
changes in the expected conditions, deviations, and contingency plans. By
the nature of their activities, different phases of the process will tend to
emphasize different elements of the observational method, as shown in Fig.
3. For example, the remedial investigation should develop sufficient infor-
mation ta allow a determination of general response actions, probable con-
ditions, and deviations. Other investigations during the FS, RD, or RA,
however, should be followed by a reconsideration of the conceptual model,
probable conditions, response actions, and deviations.
467
-------
REMEDIAL INVESIIGATION
fl-ena-cl
Mcho-g
ffecogrtjg
impocf crt
mco»IOJr»ty
on Qreceu.
Sffs CftawfartirQlfan
IrWtotAly
Ci*e-j Ha altera
EiiabOtftaenefai
siieccroilcm
Ictern^ p-ot»Wi=»
c and tore end
rnjonotla dg
i
i
i
CetncftOftJOl response
based on DiobsOte
ccnrflocH.
Pi epf Q corcaptuc*
corVngenc* p
-------
DECISION FRAMEWORK
GENERAL RESPONSE
ACTION
YES Removal with/without
» Containment or In-Situ
Treatment
FIG. 4. Decision Framework (or Ground-Water Remediation and General Re-
sponse Action
As Fig. 4 illustrates, there are two basic questions that need to be an-
swered at this point. The First is whether or not ground-water contamination
at the site poses a significant risk. A preliminary risk assessment will be
required (o answer this question. If the answer is no, it may be appropriate
(o select natural attenuation as a general response action. This selection would,
of course, also depend upon other factors, such as local public perception
and state ground-water classification. If the answer is yes, then natural at-
tenuation is probably not a reasonable selection, and containment, in situ
treatment, and removal need to be considered.
The second question is whether the contaminated aquifer will yield suf-
ficient water to make removal feasible. If the answer is no, containment may
be an appropriate action to select, assuming containment is feasible on other
accounts (e.g., if the aquifer is shallow enough to allow for installation of
a slurry wall). In situ treatment may or may not be appropriate for low-yield
aquifers. If the aquifer has a high-yield, then removal, alone or in conjunc-
tion with containment or in situ treatment, is likely to be selected.
Thus, the data required to complete this step are those needed to estimate
the level of risk posed by ground-water contamination and the ability of the
aquifer to yield .ground water. To estimate the potential level of risk, the
following information will be required:
490
1. Rate and direction of ground-water movement.
2. Nature,of ground-water contamination (e.g., types of contaminants and ap-
proximate concentrations).
3. Location of potential receptors (e.g., drinking-water wells).
The potential yield of an aquifer can be determined based on pumping test
information obtained from new or existing wells in the vicinity of the site.
The test for completeness of this step is convergence on a solution. If, at
this point, a general response action cannot be selected based on the avail-
able information, more information needs to be gathered.
Gather Information to Establish General Site Conditions
The project team should gather information on site-specific hydrogeologic
conditions and the nature and extent of ground-water contamination, and
collect other information required to support remedial design.
This step is similar to what would normally be done for an Rl. The in-
formation is used to construct and test a conceptual model of the site (see
Fig. 5).
General categories of data to be collected would include:
1. Aquifer thickness and lateral extent.
2. Presence of confining units.
3. Material composition of aquifer and confining units.
4. Rate and direction of ground-water movement.
5. Long-term yield of aquifer.
6. Stratigraphy.
7. Recharge-discharge relationships.
8. Location of.pumping wells.
9. Lateral and vertical extent of c ntamination.
10. Type(s) of contaminants and concentrations.
11. Potential disposal/discharge options for extracted ground water.
Information-gathering is complete when it is possible to construct a sound
conceptual model. It is difficult to generalize as to the level of investigation
required to construct the conceptual model. The belief is that, in general,
less investigation is required than under the current process, because the
observational method acknowledges that there will be some level of uncer-
tainty. Under the observational method, there is no attempt to characterize
the site fully. When the assumptions underlying the conceptual model can
491
-------
Identify Probable
Condilions and
Reasonable Deviations
FIQ. 5. Iterative Process of Remedial Investigation
be confirmed, or when the residual uncertainties can be handled as reason-
able deviations in later steps of the process, the R1 can be considered com-
plete.
This step may involve a multiphased investigation, particularly to char-
acterize the lateral and vertical extent of contamination. Rarely will the first
set-of monitoring wells be sufficient to characterize Vhe extent of contami-
nation at a level of detail sufficient to initiate subsequent steps in the ob-
servational method. Even though the observational method has an anticipated
outcome of early action as compared to the current process, the intent is not
to nish out and start pumping ground water without a sound conceptual model.
There arc several distinct differences between the information gathered
under an observational, method R1 and that normally gathered under a current
process RI.
492
1. As was discussed earlier, the observational method explicitly acknowledges
that ground-water systems can rarely be characterized withl a high degree of
certainty. Thus, the RI is not directed at full characterization, but rather at con-
structing and confirming a sound conceptual model, including uncertainties.
2. The approximate lateral and vertical extent of contamination must be de-
termined as early as possible under the observational method. The earlier the
extent of contamination is characterized, the earlier remediation can begin. Be-
cause the observational method allows for some level of uncertainty, character-
ization of the nature and extent of contamination does not need to be precise.
A general characterization should be sufficient because the observational method
allows for plumes of larger or smaller extent to be handled as deviations.
3. Because the observational method moves directly to design, information
required to design the remedial action can be collected during the RI. This in-
formation is not always collected during a current RI, and is deferred to the RD.
For remedial actions involving removal through extraction wells, two types of
design information should be collected:
a. Material composition of aquifer (e.g., grain size).
b. Long-term yield of aquifer.
c. Chemical concentrations during extraction.
This information is heeded to size screens, casing, and pumps for the ex-
traction wells, and to size the treatment system.
Identify Most Probable Conditions and Reasonable Deviations
The project team should use the information collected during the RI to
construct a sound conceptual model of the site. This discussion would be
included at the end of the nature and extent sections in the RJ and FS. What
makes this activity different from development of the conceptual model un-
der the current process is that the model should not only establish probable
site conditions but also envision reasonable deviations. Although reasonable
deviations are probably identified implicitly during the current process, the
observational method requires that they be identified explicitly.
The most probable cgnditions and reasonable deviations should be devel-
oped with the consensus of all project team members. This should be done
before going on to later steps because all subsequent analysis will depend
on these views.
In applying the observational method to ground-water remediation, the
conceptual model should define the most probable:
1. Rate and direction of plume movement.
2. Extent of contamination.
3. Contaminants and concentration ranges.
4. Hydrogeologic conditions.
5. Aquifer response to pumping.
493
-------
6. Location of receptors.
7. Discharge/disposal options for extracted water.
8. • Presence of immiscible-phase liquids.
In constructing_lhe conceptual model, a number of assumptions will be
made. Information gathering during the Rl should not only support the de-
velopment of the model, but also confirm its underlying assumptions. As-
sumptions that cannot be confirmed establish the basis for identifying rea-
sonable deviations. The key to knowing when to stop the iterative process
of investigation and model development/testing, is finding that the remain-
ing uncertainties can be handled as reasonable deviations. If any of the re-
sidual uncertainties produce unreasonable deviations ("deal killers"), addi-
tional investigation is required. Initial thoughts on the types of deviations to
consider suggest they may divide themselves among chemical-specific, lo-
cation-specific, and action-specific categories. A long catalog of deviations
is not thought to be necessary. A focus on the uncertainties critical to the
remedial action decision is required.
Examples of deviations that are reasonable in the context of ground-water
remediation include:
1. A small deviation in the rate and direction of contaminant migration. A
large deviation would suggest that the conceptual model was not adequately tested
and the deviation could produce significant impacts if it were not detected.
2. A small deviation in the extent of contamination. A large deviation could
require a significant increase in the number of extraction wells and treatment
capacity that may not be easily accommodated. This particular deviation may be
even more important if containment is the general response action of choice.
3. The presence of another contaminant or higher-than-expected concentra-
tions. Depending upon the type of contaminant, this deviation may or may not
be reasonable. Finding dioxin in extraction well water may have significant, im-
pacts on the treatment system because of disposal problems. Finding a pesticide
where only volatiles were expected may simply require the addition of another
•unit process on the treatment system. Similar logic applies to whether finding
higher-than-expected concentrations can be considered a reasonable deviation. It
all depends on whether or not the treatment system or discharge/disposal option
is highly sensitive to higher, concentrations.
Deviations are not "worst-case" conditions. They are specific items de-
veloped in response to specific categories of uncertainties. As the foregoing
discussion implies, the conceptual model, by taking account of the expected
design and performance of the remedial action, should help to define rea-
sonable deviations. In the case of ground-water remediation through re-
moval, the conceptual model will define:
1. Number, location, and pumping rates for extraction wells.
2. Treatment plant influent flow rate and contaminant loadings.
3. Treatment plant unit operations and removal efficiencies.
494
4. Treatment plant effluent flow rate(s) and contaminant loadings.
5. Monitoring required to defect-deviations.
Feasibility Study
In this report, the project team should identify, at conceptual design level,
deviations, parameters for observation, and contingency plans. The con-
ceptual model developed previously will have defined a range of acceptable
performance for remedial actions. This will lead to the identification of a
series of remedial alternatives. These alternatives will be evaluated in this
step in a manner similar to the current feasibility study.
Each remedial action, including the elements of the observational method,
could be discussed as follows:
1. Identify specific expected conditions that form the basis of the remedial
action.
2. Describe the remedial alternative.
3. Describe the deviations, monitoring, and contingency plans.
4. Describe the potential impacts of the remedial alternatives, deviations, and
contingency plans.
For ground-water remediation through removal, several key parameters
should be observed at most sites:
1. Ground-water gradients in the vicinity of the extraction system. Gradients
should be observed to determine whether the extraction system is hydraulically
capturing and/or controlling the plume. Monitoring wells should be strategically
located to measure water levels prior to and during operation of the extraction
system. The number, location, and frequency of monitoring will depend upon
site conditions and the configuration of the plume.
2. Changes in the nature and extent of contamination. Observation of changes
in plume configuration and contaminant concentrations needs to be made to de-
termine if the extraction system is performing as expected. If aquifer remediation
is proceeding more slowly than estimated from the conceptual model, it may
indicate a deviation (see Fig. 6).
3. Chemical concentrations downgradient of the extraction system. Downgra-
dient concentrations should be observed to determine if the extraction system is
effective in capturing or containing the plume. Detection of downgradient con-
tamination when the action is directed at complete capture would definitely in-
dicate a deviation.
4. Chemical concentration in the influent to the treatment system. Observation
of influent concentrations is required to determine whether the treatment plant
can satisfy effluent discharge requirements and to detect higher-than-expected
concentrations or unexpected contaminants that may jeopardize treatment system
performance (see Fig. 7).
5. Chemical concentration in the effluent from the treatment system. Obser-
ation of effluent concentrations is needed to determine whether discharge cri-
195
-------
), ^
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 7
Process Controlling the Transport and Fate of VOCs in Soil
Neil J. Hutzler
Department of Civil and Environmental Engineering
Michigan Technological University-Houghton
January 12-14, 1993
Las Vegas, Nevada
-------
PROCESSES CONTROLLING THE TRANSPORT AND FATE
OF VOC'S IN SOIL
Neil J. Hutzler
Department of Civil and Environmental Engineering
Michigan Technological University
Houghton, Michigan 49931
Understanding the fate and transport of volatile organic chemicals (VOCs) through the
unsaturated zone in soil is important for a number of reasons. Chemicals released at
or near the surface ultimately may find their way to groundwater. One of the goals
of engineers and scientists is to prevent this migration. Once contaminated, however,
soil can be cleaned up by a number of means including soil washing, excavation, or
in situ biodegradation. If the chemical is volatile, one of the methods of choice is soil
vapor extraction. The ability to predict the behavior of VOCs can help to optimize
the design of soil clean-up systems and can be used to estimate the time for chemicals
to reach the groundwater.
Soil can be considered a two-phase (soil/water), a three-phase (soil/water/air) or a
four-phase (soil/water/air/free product) system, where chemicals move toward
equilibrium between and within all phases. This discussion will be limited to three-
phase systems as expected in the unsaturated zone of soil.
A number of mechanisms and factors affect VOC behavior in unsaturated soil —
advection with water and air, gas and liquid diffusion, hydrodynamic dispersion,
sorption, volatilization/dissolution, abiotic degradation, and biodegradation (EPA,
1991). The purpose of this presentation is to briefly discuss each of these
mechanisms with emphasis on how they affect VOC behavior.
Advection. Advection is the transport of chemical solutes with fluid flow.
Nonvolatile compounds move with water flow. Volatile organic chemicals, however,
may be advected with either water or air. In the case where there is little or no air
movement, most unsaturated zone models assume that surface water infiltrates and
percolates vertically with a uniform horizontal moisture distribution. Recent
observations, however, show that preferential flow paths develop, even in uniform,
homogeneous sand (Hillel, and Baker, 1988; Rice et al., 1986). Preferential*flow can
cause solutes to travel much faster than would be predicted by conventional advection
models, especially where the distance from the surface to the groundwater table is
small (Sophocleous et al:, 1990).
Air flow is commonly induced in soil vapor extraction systems. In this case, air
advection is a predominant transport mechanism (Gierke, 1990). Gierke (1990)
-------
showed that there exists an optimum air flow rate for removing volatile chemicals
from soil. At low flow rates, the effect of advection is counterbalanced by back-
diffusion of the VOC's, whereas, at very high flow rates, extraction is limited by
mass transfer out of water and immobile zones.
Diffusion. In systems where immobile zones exist, either air or water, diffusion is an
important mass transfer mechanism (Brusseau and Rao, 1989). Since gas diffusion is
usually four to five orders of magnitude greater than liquid diffusion, liquid diffusion
can greatly slow the movement of VOC's when a significant amount of chemical is
contained in immobile water. Gas diffusion can be the predominant transport
mechanism where there is little or no air flow. There have numerous instances where
gases like methane and radon have diffused great distances through soil.
Dispersion. Dispersion causes the mixing of contaminated fluid with that which is
not contaminated. Dispersion is caused by molecular diffusion, variable velocities
under laminar flow, and the randomness of flow pathways in porous media. For
transport in the air-phase, gas diffusion is the predominant cause of dispersion, while
hydrodynamic mixing is the major factor causing dispersion in soil water. Dispersion
is much more pronounced in unsaturated soil than in saturated, probably because of
the preferential flow paths that develop (Hutzler et al., 1989). DeSmedt and
Wierenga (1984) observed that liquid dispersion coefficients can be orders of
magnitude greater with unsaturated flow than it is with saturated flow.
Sorption. Sorption retards the movement of VOCs and includes both adsorption onto
soil surfaces and absorption into soil particles or organic carbon. In many cases, it is
not important or possible to distinguish between the two mechanisms. Chiou (1989)
summarizes much of the literature on sorption and proposes that the primary
mechanism for nonionic organic chemicals is partitioning into the soil's organic
matter. He proposes several empirical equations that relate the partition coefficient to
a chemical's water solubility or octanol-water partition coefficient as well as activity
coefficients and temperature.
Although adsorption of chemical vapors onto soil particles has be postulated, soil has
a much stronger affinity for water molecules. Therefore, if any water is present in
the unsaturated system, sorption can be described mathematically by a simple linear
isotherm. Research by Chiou and Shoup (1985) has shown that this is true when the
relative humidity in the soil vapor is greater than 50%. Gierke et al (1992) showed
that vapor sorption could be highly nonlinear, but only for oven-dried soils.
Volatilization. By definition, VOCs can be expected to volatilize in the unsaturated
zone. The partitioning of the chemical between air and water is. mathematically
described by Henry's Law. Henry's Law constants are available for a wide range of
VOCs.
-------
Biodegradation. The biological degradation of VOCs is well documented (Alexander
and Scow, 1989; EPA, 1991). Under the conditions of no air flow, anoxic or
anaerobic degradation is predominant, which promotes the dechlorination of
halogenated VOCs. In soil vapor extraction systems, however, the introduction of air
promotes aerobic activity. This promotes the degradation of hydrocarbons and some
classes of chlorinated organics. The Air Force currently has a program to promote
aerobic degradation of fuels through bioventing systems. While the time to clean up
is typically longer, the need for vapor treatment is greatly reduced.
Abiotic Degradation. A number of abiotic transformations such as hydrolysis,
oxidation-reduction, precipitation, and complexation may also be active in unsaturated
soils (EPA, 1991). Hydrolysis is the direct reaction between VOCs and water. The
rates of hydrolysis are typically slow, however, and, thus, hydrolysis may not play an
important role in the fate of VOCs. In fact, most abiotic processes are more likely to
affect the mobility of inorganic compounds.
Future Needs. While much is known about the behavior of VOCs in unsaturated
soil, there is still much that needs to be known. The impact of preferential flow paths
has only recently been investigated, and little has been done to model this behavior.
While laboratory and pilot experiments have revealed much about the mechanisms of
VOC fate and transport, more work is needed to translate this information to field-
scale problems such the design and monitoring of soil clean up systems. Certain
technologies such as air sparging and bioventing have been demonstrated at the field
scale, but little has been done to model such systems so that their operation can be
optimized. Finally, much more research is required to study the behavior of multi-
component mixtures of VOCs that are commonly found in practice.
-------
References.
Alexander, M., and K.M. Scow, 1989. "Kinetics of Biodegradation in Soil," in
Reactions and Movement of Organic Chemicals in Soils. SSSA Special Publication
No. 22, Soil Sci. Soc. Amer., Madison, WL, pp. 243-270.
Brusseau, M.L., and P.S.C. Rao, 1989. "Sorption Nonideality During Organic
Contaminant Transport in Porous Media," CRC Crit. Rev. Environ. Control.
19(l):33-99.
Chiou, C.T., 1989. "Theoretical Considerations of the Partition Uptake of Nonionic
Organic Compounds by Soil Organic Matter," in Reactions and Movement of Organic
Chemicals in Soils. SSSA Special Publication No. 22, Soil Sci. Soc. Amer., Madison,
WL, pp. 1-31.
Chiou, C.T., and T.D. Shoup, 1985. "Soil Sorption of Organic Vapors and Effects
of Humidity on Sorptive Mechanisms and Capacity," Environ. Sci & Tech..
19(12): 1196-1200.
DeSmedt, F., and P.J. Wierenga, 1984. "Solute Transfer through Columns of Glass
Beads," Water Resources Res.. 20(2):225-232.
Environmental Protection Agency, 1991. "Ground Water, Volume II: Methodology,"
EPA/625/6-90/016b, Office of Research and Development. Center for Environmental
Research Information, Cincinnati, OH, 141 pp.
Gierke, J.S., 1990. "Modeling the Transport of Volatile Organic Chemicals in
Unsaturated Soil and Their Removal by Soil Vapor Extraction," Ph.D. Dissertation,
Dept. of Civil and Env. Engrg., Michigan Technological University, Houghton, MI.
Gierke, J.S., N.J. Hutzler, and D.B. McKenzie, 1992. "Vapor Transport in
Unsaturated Soil Columns: Implications for Vapor Extraction," Water Resources
Res.. 28(2):323-335.
Hillel, D., and R.S. Baker, 1988. "A Descriptive Theory of Fingering During
Infiltration into Layered Soils," Soil Science. 146(l):51-56.
Hutzler, N.J., J.S. Gierke, and L.C. Krause, 1989. "Movement of Volatile Organic
Chemicals in Soils," in Reactions and Movement of Organic Chemicals in Soils.
SSSA Special Publication No. 22, Soil Sci. Soc. Amer., Madison, WL, pp. 373-403.
Rice, R.C., R.S. Bowman, and D.B. Jaynes, 1986. "Percolation of Water Below and
Irrigated Field," Soil Sci. Soc. Amer. L, 50(4):855-859.
Sophocleous, M., M.A. Townsend, and D.O. Whittemore, 1990. "Movement and
Fate of Atrazine and Bromide in Central Kansas Croplands," L Hvdrol.. 115:115-
137.
-------
Biography.
Dr. Hutzler is Professor of Civil and Environmental Engineering and Director of the
Environmental Engineering Center at Michigan Technological University. He
received his Ph.D. in environmental engineering from the University of "Wisconsin -
Madison in 1978. Dr. Hutzler is actively involved in teaching at the senior- and
graduate-level. He is also active in research with expertise in the fate of chemicals in
the subsurface, soil vapor extraction system design, and the beneficial utilization of
granular and fine particulate residuals. He is the author or coauthor of over 35
publications and has made over 35 presentations at national meetings. Dr. Hutzler is
a member of the American Geophysical Union, the American Society of Civil
Engineers, the Association of Environmental Engineering Professors, the Association
of Ground Water Scientists and Engineers, and the Water Environment Federation.
He is a registered professional engineer in Michigan and Wisconsin.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 8
January 12-14, 1993
Las Vegas, Nevada
-------
Soil Sampling Strategies and the Decision-making Process
Evan Englund, U.S.EPA
With the exception of occasional research and development studies, the need to make
decisions is tine driving force behind environmental soil sampling and measurement.
Typically, a set of sample measurements is used to estimate the concentration,
sometimes with confidence limits, of a particular area or volume of soil. At various stages
in a project, the decision to be made might be whether or not a potential problem exists;
whetherthe problem is bad enough to require remediation; determining exactly where the
backhoe operator should dig, or whether the remediation has been successful.
Because soil is spatially heterogeneous, the scale associated with any sample,
measurement, estimate, or decision is critical. The term "concentration" has no concrete
meaning except when referring to the mean concentration of a specified mass of material.
In geostatistics, we refer to the physical size, shape and orientation of such a mass as
the "support". In the analytical laboratory, ideally, each lot of material to be analyzed is
homogenized so that any aliquot will be nearly the same as the lot. The problem occurs
when we deal with soil in the field. A statement such as "remove all of the soil from the
site where the concentration of contaminant exceeds 1 mg/g" is ambiguous until the
decision support is specified. This often leads to misunderstanding as different parties
interpret it differently.
Only by explicitly and quantitatively defining the decision procedure, including the decision
support, can we begin to consider a sampling strategy to optimize the d.ecision process.
Optimization in the context of contaminated soils implies minimizing the overall loss to
society. This in turn requires an economic model which can be used to explore the trade-
off between the costs of sampling and the consequences of incorrect decisions. At least
in the case where the decision to be made is to remediate or not, and the remediation
costs are known, a simple, reasonable economic model can be generated (Englund et
al, 1992). Decision quality depends on estimation quality (Weber and Englund, 1992),
¦which depends not only on data quality, but on data quantity, and in a somewhat complex
manner, oh sample support. In the spatial context, it is quite possible for a large number
of biased, imprecise measurements to result in better decisions than a small number of
unbiased, precise values. In order to quantify the estimation and decision quality for any
proposed sampling approach, a model of spatial variability such as the variogram is
required.
In principle, if the economic model, the variogram model, and a parametric model for the
data distribution can be estimated, it should be possible to compute directly the expected
loss for any proposed sampling plan. Real world situations, however, are so complex that
the best approach seems to be the use of Monte Carlo simulation methods. Although
no comprehensive software to perform such an evaluation is yet available, preliminary
results with a simplified prototype using this approach are encouraging, and further
-------
development is in progress (Englund, and Heravi, 1992).
In general, measurements on larger sample supports will have lower variability. Thus,
it is generally advisable to make the sample support as large as possible, so long as it
does not exceed the decision support. The most practical method is to increase the
effective support through spatial compositing. Compositing; however increases the cost
of sampling and subsampling, may decrease subsampling precision, and in the case of
VOC's, will almost certainly increase bias. The detailed variogram model over the scale
of the possible sample supports can be used to estimate the potential decrease in
measurement variability which might be obtained by compositing. The overall effects on
cost and decision quality, including the potential effects of bias could then be evaluated
through a simulation.
NOTICE'
Although che research described in this article has been supported by the United States
Environmental Protection Agency, It has not been subjected to Agency review and therefore does
not necessarily reflect the views of the Agency and no official endorsement should be inferred.
-------
Technical References
Englund, E.J., Weber, D., and Leviant, N. (1992) The Effects of Sampling Design
Parameters on Block Selection, Mathematical Geology 24:3:329-343.
Weber, D. and Englund, E.J. (1992) Evaluation and Comparison of Spatial Interpolators,
Mathematical Geology 24:4:381-391
Englund, E.J. and Heravi, N. (1992) Conditional Simulation: Practical Application for
Sampling Design Optimization, in Proceedings of the Fourth Internationa! Geostatistics
Congress, Troia Portugal, Sept. 1992, A. Soares, ed., Kluwer Academic Publishers.
Refated Manuscripts
Isaaks, E., and Srivastava, R.M. (1989) Introduction to Applied Geostatistics. Oxford
University Press.
Deutsch, C., and Journel, A. (1992) GSLIB: Geostatistical Software Library and User's
Guide. Oxford University Press.
Englund, 'E.J. (1988) Spatial Autocorrelation: Implications for Sampling and Estimation,
in Proceedings of the ASA/EPA Conferences on Interpretation of Environmental Data, III
Sampling and Site Selection in Environmental Studies , W. Liggett, ed., EPA
230/8-88/035.
-------
Evan J. Englund
Mathematical Statistician
U.S. Environmental Protection Agency
Environmental Monitoring Systems Laboratory - Las Vegas
P.O. Box 93478
Las Vegas, NV 89193
(702) 798-2248
Evan is currently working on geostatistical methods for site characterization and
remediation, with emphasis on contaminated soils. He is currently doing research and
development on geostatistical simulation as a tool for integrated optimization of
remediation plans that is, finding the most cost effective combination of sampling
network, sampling and analytical methods, QA/QC, interpolation method, and decision
criteria for a specified remediation objective. In 1987-88, he was responsible for
developing the popular Geo-EAS software package.
Before joining the EPA in 1986, Evan worked for Exxon Minerals Co. in Houston as a
Senior Geologist specializing in geostatistical ore reserve estimation, and as a Technical
Systems Analyst developing computer systems for geostatistics and other mining
applications. Earlier, he worked for Phelps Dodge Corp. in Morenci, AZ, where he
acquired his interest in geostatistics as a Geologist conducting exploratory sampling
campaigns and estimating ore reserves, and as an Ore Control Engineer in charge of
sampling, estimation and ore selection during day-to-day mining operations.
He has a B.S. in Geology from the University of Wisconsin, an M.S. in Geology from the
University of Vermont, and a Ph.D. in Geology from Dartmouth.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 9
Data Quality Objectives and Statistical Treatment of Soil VOC Data
Alfred Haeberer
USEPA Office of Research and Development
Washington, DC
January 12-14, 1993
Las Vegas, Nevada
-------
Data Quality Objectives and Statistical Treatment of Soil VOC Data
Fred Haeberer and John Warren
Quality Assurance Management Staff
U. S. Environmental Protection Agency
The most neglected aspects of environmental data collection operations have been in the
areas of sampling design and sample collection. For years, most major environmental data
QA/QC efforts have focused on the analytical or measurement operations, with either the tacit
assumption that sampling activities had been properly designed and implemented, or with very
little attention given at all to the quality of field or sampling operations. Infrequent data quality
analyses did result in the recognition that major error sources were undoubtedly associated with
the sampling components of environmental data collection. However, little was generally done
to address or remedy these recognized data quality shortfalls. This paper will illustrate the
importance of adequate planning, implementation, and assessment of field sampling work and
laboratory/analytical results in the context of the Environmental Protection Agency's Superfund
program.
INTRODUCTION
Management of environmental data quality is crucial to making correct decisions about
Superfund remedial activities. Decision errors can be classified as either "false positive" or
"false negative." When the data erroneously fail to support the true hypothesis that a site does
not pose unacceptable risk, then a "false positive" decision can result. This could cause the
needless expenditure of millions of dollars to remediate a site that does not really require clean-
up. Conversely, when the environmental data erroneously support the false hypothesis that a
site does not pose unacceptable risk, then a "false negative" decision may result. As a
consequence, the site probably would not be remediated. This is clearly the more serious
situation in terms of protecting human health inasmuch as it would leave a potentially serious
health threat unresolved.
The traditional focus of environmental data quality has been on laboratory activities.
Laboratory measurement (analytical) methods have undergone intense scrutiny to assure that the
data are of the quality claimed and can be used in decision making. Millions of dollars of
research money have been spent to develop new and/or improved analytical procedures with
greater precision and accuracy to produce data that are perceived to be more defensible for
remedy selection decisions, litigation or negotiation with potentially responsible parties (PRPs),
and cost recovery actions. There has been a false sense of security in the quality of these data,
largely because until recently a similar effort had not been made to assure that the samples
collected were of the type and quality needed. Sadly, if "bad" samples are collected, even the
most precise and accurate analytical methods cannot rectify their quality. For example, a soil
sample improperly collected or composited may not represent accurately the actual distribution
of contaminants in the area of interest. Similarly, the improper collection of soil samples could
1 SLfcfM
-------
result in the loss of volatile constituents critical for risk or exposure determination. Even more
seriously, the analytical processes may unwittingly mask the sampling error from being detected
and result in the use of poor quality data.
A study of the Superfund RI/FS process identified the importance of sampling design and
field sampling as activities which contribute significantly to total data error. It emphasized the
need for adequate attention to planning, design, implementation, and assessment of environmen-
tal data collection programs. This study showed that focused planning, prior to the initiation of
any sampling, could improve the current RI/FS process by quantitatively establishing where and
how many samples should be collected, thereby reducing the need for unplanned follow-up
sampling episodes. In this way, RI/FS time and costs could be reduced and, at the same time,
the data user's decision error rates could be managed at acceptable levels. This approach
considers the activities associated with field sampling and laboratory analysis as components,
each contributing its own error portion to the environmental data collection process. It makes
the data user aware early in the planning process of the uncertainties associated with the
resulting data and provides design opportunities for addressing both sampling and analytical
concerns.
THE DATA QUALITY OBJECTIVES PROCESS
Effective field sampling is achieved by defining what is to be accomplished, planning
how to accomplish the defined goal, carrying out the documented plan, and assessing the
effectiveness or quality of the field work itself and the quality of the obtained results.
Unfortunately, in technocracies such as the EPA there predominate two types of individuals:
those that understand what they do not manage, and those that manage what they do not
understand (Putt's First Law). An effective planning effort is needed to draw these two types
of individuals into more productive communication in order to bridge the gap in their
understanding of the issues. In this way, results acceptable to all participants may be achieved.
Quality Assurance Management Staff (QAMS) advocates the Data Quality Objectives (DQOs)
process to accomplish this vital task. The quality of the results obtained from this planning
process in defining the data user's acceptable decision error rates (i.e., the error rates acceptable
in the utilized data) will directly determine the quality of the project's design. A responsive data
collection design can only be established after the decision, its needed inputs, and the decision
error rates acceptable to the data user have been adequately defined.
The DQO process is a course of action for planning environmental data collection operations that
helps the data user(s) decide what data quality (and quantity) will be adequate for decision
making and to direct the development of a statistical design to collect the data meeting those
needs. The DQO process emphasizes decision making and focuses on quantifying the levels of
uncertainty acceptable in data used in decisions. The DQO process provides a logical structure
that focuses data collection planning on the intended use of the data. There are seven steps to
the DQO process (see Figure 1). The output from each step is used in developing a statistical
data collection design.
-------
THE DATA QUALITY OBJECTIVES PROCESS
Figure 1
A brief overview of each step:
o State the problem: the planning team reviews existing information and any relevant
prior studies in order to describe the problem.
o Identify the decision: the planning team identifies the decision for which new data is
needed.
o Identify the inputs to the decision: they identify the variables to be measured (e.g., the
needed pollutant concentration).
o Define the study boundaries: the planning team specifies the spatial and temporal
boundaries on the decision (i.e., the area and time period to which the decision will apply).
o Develop a decision rule: the team incorporates the output from each of the previous steps
into a statement that describes how the data will be summarized and used in the decision.
-------
o Specify acceptable limits on uncertainty: the decision maker (data user) specifies the
decision error rates acceptable under a variety of circumstances (e.g., at different concentra-
tions).
o Optimize the design: a statistician takes the information from the previous steps and
generates alternative designs, including QC requirements. The decision maker picks the best
of the alternative designs, usually meaning the lowest cost design that meets all the DQOs,
including the limits on uncertainty (i.e., acceptable decision error rates).
The DQO process should be used at the planning stage of a data collection operation, before any
samples are taken. In general, QAMS's policy (and, by extension, EPA's) is to use the DQO
process to plan all data collection efforts that will require or result in a substantial commitment
of resources.
The planning team is established during the first step of the DQO process. It usually consists
of senior program staff (the decision maker/data user, i.e., the EPA Remedial Project Manager
[RPM] and/or the state equivalent), technical experts (field and laboratory personnel, engineers),
and needed specialists (risk assessor, hydrogeologist, statistician). It is important that all of
these people , including managers, participate (or, at a minimum, are kept informed) from the
beginning of the DQO process so that it can proceed efficiently. The planning team reviews all
available data and develops a Conceptual Site Model (CSM) (if this has not already been done).
The sampling and analysis budget and relevant Superfund deadlines are specified. The site
manager identifies the scoping team, including representatives from all data users, relevant
technical experts and a design statistician.
The planning team identifies the site decisions that will be based on analysis of collected data
(e.g., whether the site poses an unacceptable threat), as well as the actions that could result from
these decisions (e.g., whether or not remediation will be required).
In this step the planning team identifies the specific variables (i.e., the VOCs of concern) that
will be measured, along with any other information needed to make the decision. Action levels,
such as ARARs (Applicable or Relevant and Appropriate Standards, Limitations, Criteria, and
Requirements), or target risk levels for the site are also identified in this step.
The planning team identifies the spatial and temporal boundaries of the various media needing
to be addressed at the site (i.e., the area and time period to which the decision applies).
Additionally, the site manager needs to identify whether a single decision will be made about
the entire site or whether the decision will be applied to defined portions of the site. The later
case is most frequently appropriate for remediation of soil contamination.
In this step the planning team develops a statement, known as the decision rule, of how data will
be used in the decision process. This rule should include how data will be summarized and
compared to the "assumed" action level for the VOC of concern at the site. As stated earlier,
the action level may be based on an ARAR or a target risk level.
-------
In order to enable the statistician to establish the acceptable uncertainty in the data, the decision
maker must specify the decision error rates acceptable to him/her under various site conditions
(i.e., for different VQ.C concentrations). These rates are the acceptable probability of the results
causing the wrong decision to be made.
A knowledgeable environmental statistician should be able to tailor the data collection design and
the assumptions underpinning it so that the decision maker's uncertainty constraints can be met.
In this step the scoping team, along with the statistician, identifies the design option best suited
to the needs of the site (i.e., that option meeting all DQOs and budgetary constraints) and selects
the most cost effective one. The selected design is incorporated in the Sampling and Analysis
Plan (SAP) along with the QA/QC procedures needed for establishing the effectiveness of the
design and whether the design (i.e., the SAP) was properly implemented.
THE DATA QUALITY ASSESSMENT PROCESS
When environmental data are used to support a decision, the decision maker needs to understand
the quality of those data. If the data show a large amount of variation in the measurements, then
the decision maker may be facing some difficult questions. Do the data show a large amount
of variability in the characteristics of interest? How does this variability complicate the task of
making a sound decision (i.e., in managing the decision error rate within acceptable bounds)?
On the other hand, does the data variability indicate that the measurement system is unreliable?
These kinds of questions can be answered by applying the Data Quality Assessment (DQA)
process.
The DQA process is built on the fundamental premise that data quality is meaningful only in the
context of the intended use of those data. Data quality does not exist in a vacuum: if one does
not know how a data set is to be used, then the relevant criteria needed forjudging whether the
data are "good enough" can not be available. This is why the DQO process focuses on decisions
(i.e., on data applications) that will be supported by the environmental data. If the decision,
along with the acceptable uncertainty, is carefully defined, then the DQA process can be applied
to rationally and quantitatively evaluate the quality of the data.
The DQA process primarily involves the statistical analysis of data for decision-making with
respect to the planned or required levels of confidence. DQA is intended to detect inherent
flaws in the complete data collection activity by examining both measurement error (due to
unavoidable bias and imprecision of the measuring process or method) and sampling error (the
natural variability of VOC in soils, of the soils themselves, the effects of different sampling
protocols, the effect due to the limited numbers of samples being drawn, and other field
variabilities). The results of a DQA may be used to guide the planning and acquisition of
supplemental data, or be used in recommendations to the decision-maker to change parts of the
DQO (e.g. the required limits on uncertainty).
The view-points of the decision-maker and statistician converge when it is realized that the
structure of the process (The Scientific Method) is common, only the terminology differs (Figure
2). The actual collection of data occurs between steps three and four, and the heart of the DQA
-------
process lies in steps four and five. These two steps are mainly the purview of the statistician,
but as the contribution by the decision-maker and technical staff in experience in soil VOC
analysis is equally important, an understanding of what the statistician does in these steps is
essential.
TWO VIEWS OF DATA QUALITY ASSESSMENT
A DECISION MAKER'S VIEW A STATISTICIAN'S VIEW
1
DEFINE THE DECISION RULE
1
DEFINE THE
STATISTICAL HYPOTHESES
J I
2
SPECIFY ACCEPTABLE
LIMITS ON UNCERTAINTY
2
DETERMINE ACCEPTABLE
DECISION ERROR RATES
I
T r
3
SELECT METHOD FOR j
APPLYING DECISION RULE !
3
IDENTIFY STATISTICAL TEST
AND ASSUMPTIONS
1 i
4
ENSURE THAT METHOD |
IS DEFENSIBLE j
4
ASSESS VALIDITY OF
STATISTICAL TEST
t 1
1
5 APPLY THE DECISION RULE i
' 1
5
PERFORM STATISTICAL TEST
AND ASSESS DESIGN
Figure 2
Assessing the Validity of the Statistical Test
When the assumptions needed for the statistical test are valid (or approximately so), and the data
collection operation (including sampling and analysis) has been properly implemented, the DQO
projected error rates will be achieved. As the assumptions are weakened or violated, the error
rates change, often in unpredictable fashion.
Assessing the validity of assumptions really consists of two parts; 1.) an inspection of the actual
data values for anomalies and errors of transcription, together with a determination of degree
of adherence to SOPs and QA protocols, and, 2.) the more statistical investigation of the
assumptions used to establish the data quality parameters in the early stages of the DQA.
-------
The first part is usually covered by standard QC procedures but the importance of the experience
of the data user/decision-maker by inspecting the data for "reasonableness" cannot be
overemphasized as many anomalies undetectable by standard statistical methods may be
identified by a knowledgeable VOC analyst or data user/decision-maker.
The statistician makes use of both subjective judgement (graphical analysis for identification of
trends and anomalies), and statistical models and inference (outlier detection, autocorrelation
estimation) in the investigation of data for validity of assumptions needed to make a statistical
test. The process is outlined in Figure 3.
Figure 3
-------
For many situations in which the collection of VOC data is the primary focus of attention^ these
assumptions may be grouped as:
o The type of distribution of VOC concentration values
(for example, Normality or Lognormality)
o Independence of VOC concentration distribution values
(e.g. similarity in concentration of samples taken
closely together in time or space)
o Additivity and variability of measurement errors
(magnitude of measurement error varying directly
with the magnitude of VOC concentration)
When a reasonable number of data values are available, statistical methodologies may be
employed to explore the validity of these assumptions and determine the robustness of the
statistical tests under less than optimal circumstances. Typical tests employed by statisticians
to identify the reasonableness of assumptions include:
Type of Distribution
Independence
Additivity / V ariabilitv
Goodness-of-fit Tests
Shapiro-Wilk Statistics
Geary's Test
Autocorrelation Analysis
Time Series Analysis
Kriging Methods
Graphical Trends Plot
Coefficient of Variation Analysis
Box-Cox Transformations
The conclusions from these tests can only indicate when substantial departures from important
assumptions have occurred. It must be noted that in many cases, the relative paucity of
environmental data values often makes the investigation of assumptions difficult and the
collaboration between statistician and data user/decision-maker becomes of even greater
importance. Fortunately, many of the statistical tests used in environmental analysis are
relatively robust and the assigned error-rates remain approximately valid unless gross departures
from key assumptions occur.
Perform Statistical Test and Assess Design
Given the underlying assumptions are approximately valid, the data can be used in a statistical
test to distinguish between two distinct hypotheses, for example, the mean level of VOC
concentration is less that a certain value or, alternatively, the mean exceeds this value. These
hypotheses should be formulated under the DQO before the data are collected or analyzed.
Creation of these hypotheses based on the results of analysis of VOC concentrations violate the
concept of the DQO and indeed the Scientific Method. The statistician formulates these two
-------
hypotheses into a Null Hypothesis (a state of nature that describes what should occur), and an
Alternative Hypothesis (a credible alternative to the Null but which will be accepted only when
faced with overwhelming evidence that the Null is unlikely to be true). On the basis of a
statistical test, the statistician chooses one of these and is able to quantify the probabilities of
making a wrong choice. In the case of VOCs in soils;
Null Hypothesis: The mean concentration of VOC is less the Regulatory Threshold
Alternative Hypothesis: The mean concentration exceeds the Regulatory Threshold
The aim of the statistical test is choose between these hypotheses with the minimum possibility
of making an error of judgement by being led by the data to select the wrong hypothesis, and
consequently reach a false conclusion. Rejecting the Null Hypothesis when it is really true is
called a false-positive error, and accepting the Null Hypothesis when it really false is a false-
negative error. Although the possibility of either error is usually quite small, the limited
resources available to the decision-maker makes it impossible to make both small simultaneously;
one may be set at a small level but only at the expense of the other being relatively large.
It is the job of the statistician to work with the decision-maker to assess the importance of the
consequences of errors in decision-making (the Specify Limits of Uncertainty step of the DQO
Process) in order to achieve a balance between false positives and false negatives (related to the
Power of a statistical test). The procedure is illustrated in Figure 4.
5
PERFORM TEST AND ASSESS DESIGN
I
1
Calculate Power Curves
Based on Alternative Scenarios
Suggested by Data
(e.g., different variances)
Perform the
Hypothesis
Test
Recommend
Design Improvements
Perform Error Analysis; _ YES
Perform Error Analysis;
Recommend Solutions
(e.g., more data,
new methods, etc.)
NO
| Implement Solution
Figure 4
-------
The statistician can retrospectively assess how much power the statistical design actually
achieved and this may be used by the decision-maker to determine the degree to which the DQO
was met. Should further data be required to enhance the existing design in order to meet the
original DQO, this statistical analysis would form the basis for the supplemental DQO.
CONCLUSION
DQOs represent primarily the planning part of a data collection; DQAs represent primarily the
method by which an assessment of how well these DQOs were achieved. Although each can
stand independently of the other, it is by linking them together that progress towards the goal
of better quality environmental data can be achieved.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 10
Sampling and Analytical Determination of Soil VOCs
Michael J. Barcelona
Department of Chemistry - Institute for Water Sciences, Western
Michigan University, Kalamazoo
January 12-14, 1993
Las Vegas, Nevada
-------
SAMPLING AND ANALYTICAL DETERMINATIONS OF SOIL VOC'S
Michael J. Barcelona
Department of Chemistry - Institute for Water Sciences
Western Michigan University
Kalamazoo, MI 49008
Background
Volatile organic compounds (VOC's) poSe dual challenges to environmental
quality assessment and protection efforts. The first challenge presents itself through the
pervasive nature of soil (i.e. surface and subsurface solids) and water contamination in
our nation's population and industrial centers. (1,2) This aspect of VOC related
problems suggests that the contamination has occurred for some time and that land use
as well as environmental variables must be made part of any meaningful assessment
effort. The second challenge posed by VOC's is their persistence in the environment,
particularly when they are present with contaminated solids. (3) In this regard, we must
tune our approaches in environmental assessment (i.e. sampling and analysis) and
protection (i.e. interpretation and remediation) so as to address both short-term and long-
term risk management goals. It is clear that, in many of our sampling and analytical
activities, we have been assessing gaseous or aqueous symptoms of VOC contamination
rather than the long-term sources.
It is perfectly reasonable to utilize soil gas and ground-water investigations to
estimate the spatial extent of VOC contaminant influences on the subsurface environment.
However, in order to identify the VOC source distributions (i.e. the long-term problem),
it is essential that we focus on the free-product and sorbed VOC phases which will
continue to yield mobile contaminant for decades. (4, 5)
Soil permeability, moisture content, subsurface lithology, aquifer properties and
preferential pathways (e.g. sewer/pipeline excavations, foundations, etc.) for gas and
liquid migration should be determined early in the course of assessment activities. These
measurements provide the the initial basis for sampling and analytical program designs
and for data interpretation. The approach has the distinct advantage of identifying
impacted versus "background" areas in the context of physical, hydrogeologic and
geochemical conditions at investigation sites.
Program Objectives and Research Needs
The fates and transport properties of VOC's differ substantially based on their
vapor pressure, viscosity, aqueous solubility, sorptive characteristics, degradability and
density. The listing of compounds in Table 1 categorizes VOC's as DNAPLS (Dense
Non-Aqueous Phase Liquids) and LNAPLS (Less-Dense Non-Aqueous Phase Liquids)
and shows the range of properties they incorporate. In most investigations, one has little
-------
information on the original source, mass of the contaminant mixture released, or the time
frame over which transport may have occurred. Given the fact that non-equilibrium
transport and degradation conditions are common in the shallow subsurface, it is clear
that even with detailed source information we could not predict the spatial distribution
of contaminants. Sampling and analysis are therefore necessary components of
investigative programs which require substantial care in design, execution and
interpretation.
Objectives of specific programs range from determinations of the severity and
extent of contamination situations to the selection of remedial action options for surface
soils, subsurface solids and ground water control or management. Data quality objectives
during successive phases of an investigation may be expected to change as the dimensions
of the problem are. better defined. However, adequate spatial coverage of samples and
accurate (i.e. unbiased) analytical detail on major regulated contaminants are universal
concerns. (6) Identification of principal contaminant concentration distributions
(particularly free phase occurrences) in soil, aquifer materials and ground water is
desirable despite the fact that many risk assessment and management methodologies do'
not address free product adequately. (7)
The design and execution of sampling and analytical programs, nonetheless should
be approached with both short and long-term contaminant occurrence and fates in mind.
Determinations of both major contaminants and potential breakdown products over a wide
range of VOC concentrations must also be done within limited holding periods. With
respect to VOC's, this will require: comprehensive sample coverage and rapid-field
screening as well as laboratory analytical methods. To date, most sampling and
analytical methods which minimize sample transfer, handling and VOC losses have
focussed on soil gas and water sample screening. Much more attention needs to be
placed on contaminated solids since the bulk of the contaminant mass may often reside
in the solid and free-product phases. (5,8) The need is particularly acute for the field
preservation of solid samples after screening for more quantitative, complete laboratory
determinations of contaminant/breakdown product mixtures.
Sampling
Much has been written regarding the fundamentals of the design and execution
of environmental sampling efforts. (9,10) Cost, statistical rigor, and coping with the
uncertainties in relating sample populations to the universe of soil and aquifer
environments are real concerns to scientific professionals and the public. The ideal
sample population would pinpoint the primary media and timeframes of exposures as well
as provide data of known quality for risk assessment and management decision making.
For VOC's the ideal sampling design must include rapid turn-around from sampling to
analysis. In addition, sampling and preservation steps should enable rather than constrain
the analytical process.
-------
The compromise between the ideal and the practical has most often been wrought
on the anvil of expediency. I would submit that we must acknowledge uncertainties in
contaminant distributions (9,11) and environmental characteristics (12,13) from the outset
and plan further sampling and analytical efforts with the basis of initial field results,
hydrogeologic, and geostatistical interpretations. A proposed framework for applying
this approach to initial site assessment is shown in Figure 1. The emphasis of this
approach is on determination of the extent of contaminant influence based largely on field
screening, selected laboratory determinations, considerations of the contaminant's
physical state and controls on contaminant distributions. In this respect, proven field
screening methods can be used to their best advantage (i.e. semi-quantitative estimates
of contaminant concentrations in selected phases) to determine the scale over which more
statistically and analytically rigorous methods need to be employed. This approach relies
heavily on the use of existing regional data on geology, hydrology, land-use and site
operations to develop a skeleton sampling plan. The skeleton plan may be refined by
application of geophysical techniques and selected borings in background and potentially
impacted areas so that preliminary sampling for chemical analysis is done in formations
where contamination is likely. The use of minimally obtrusive "push" technology (e.g.
narrow diameter boring tools, cone-penetrometry, HydropunchR, etc.) and nested
piezometers provides sufficient sample for field screening. These techniques further lend
themselves to obtaining wider spatial coverage than conventional ground-water
monitoring well networks which often become obsolete or redundant early in remedial
investigations.
The results of the preliminary site assessment technique include:
1. A quantitative estimate of the land area involved in the_ overall
contamination situation with clear indications of: the vertical complexity
of site lithology and possible ground water involvement,
2. A reasonable estimate of the physical controls (e.g. clay or fine sediment
layers) on VOC movement in the gaseous or aqueous phase, the degree
of subsurface inhomogeneity and solid-associated VOC concentrations
which demand more/less intensive sampling and analysis to resolve 3-
dimensional distributions, and
3. A preliminary dataset with which an integrated statistical and
hydrogeologic sampling plan can be designed and implemented in the
refined site assessment phase.
This approach would be a reasonable starting point as the nation moves to
accelerate National Priority List site actions. Some identification is critical to cost-
effective remedial option design, execution and performance.
New Directions in Sampling
The techniques available for field sampling and analysis of gas, water and solids
have grown rapidly (14,15,16) and they may be expected to advance further in the future
(17)^ One of the major emerging research directions deals with immediate preservation
-------
of VOC samples with the extraction solvent used for more detailed laboratory analysis.
Bone (18) has published an application of the type of method which essentially begins the
analytical process in the field and affords the possibility of more meaningful VOC
analyses on heterogenous materials. Field decisions can be made to more/less intensively
process samples for the lab as core material of differing lithology or field screened VOC
content becomes available. There are also indications that the holding times for samples
preserved in this manner can be extended over those for bulk-samples. (19) The added
advantages for VOC determinations are that order of magnitude losses due to storage,
sub-sampling and transfer are minimized.
Analysis
Laboratory analysis techniques exist which can handle difficult matrices
adequately. (9,16) The critical considerations here are the difficulties associated with
spiking solid samples with standards (e.g. internal or calibration standards) for quality
control and the need for minimizing sample handling and transfers prior to VOC analyses
(e.g. for purge and trap methods). Headspace methods for initial separation have
promise in this regard. It is anticipated that ongoing research will result in the
acceptance of either static or dynamic headspace methods by regulatory agencies.
-------
Figure 1; Framework for Preliminary Site Assessment for VOC Contamination.
Zone/Project Activity Field Lab
Interim Outputs
Soil Unsaturated Zone
Land-Use/Regional Hydrogeologic Review
Geophysical Survey ~
Sampling
Interpretation of Subsurface —~ Geostatistical Picture of--—*
Conditions, Litbology Variability in Geology
t and Contaminant Influence
I
» Gas
Field Screening Fixed Gases/VOC's 1<
Solid Cores —~ Field Screening VOC's Lab Analysis VOC's (Selected Samples)
Organic Carbon
—» Permeability Testing —* soil moisture
mineralogy
grain size
Saturated Zone
Piezometer Array
(Primarily for water level measurement)
On-Site
Ground Water Flow
Direction/Magnitude
1 Flow-Net
Analysis
t
Permeability Testing
Sampling
•HjO-
-»Solid
Cores
¦ Field Screening VOC's —r » Lab Analysis VOC's (selected samples)
Organic Carbon
~mineralogy
grain size
-t
t
021b:symposiu.ppr
¦ Permeability Testing —
-------
Table 1. List of Common VQC's and Selected Physical Properties
Drinking Water/
Vapor Groundwater Standard
Density Viscosity Pressure Solubility or Guideline (mp/Lt
(g/cm3) (cP) (mm) (mg/L) U.S. EPA
DNAPLs
Methylene chloride
1.33
0.44
349
20,000
-
Chloroform
1.49
0.56
151
8 200
-
Carbon tetrachloride
1.59
0.97
90
785
0.005
Bromoform
2.89
2.07
5
3.010
-
Bromodichloromethane
1.97
1.71
50
4.500
-
1,2-Dichloroethane
1.26
0.84
61
8.690
0.005
1,1,1-Trichloroethane
1.35
0.84
100
720"
0.2
1,1,2-Trichloroethane
1.44
na
19
4.500
-
1,1,2,2-Tetrachloroethane
1.60
1.76
5
2.900
-
1,1-Dichloroethylene
1.22
0.36
590
400
0.007
Trans-1,2-dichloroethylene
1.26
0.4
326
600
0.1
Trichloroethylene
1.46
0.57
58
1.100
0.005
Tetrachloroethylene
1.63
0.90
14
200
0.005
1,2-Dichloropropane
1.16
na
42
2.700
0.005
Chlorobenzene
1.11
0.80
12
488a
0.1
1,2-Dichlorobenzene
1.31
1.41
1.0
100
0.6
1,3-Dichlorobenzene
1.29
1.08
2.3'
123"
LNAPLs
Benzene
0.88
0.60
76
820
0.005
Gasoline
0.72-0.78
0.5-0.7
-
150-300
Crude Oil
0.8-0.9
3-35.
-
5-25
Source: Adapted from Gillham and Rao (1990) and Havinga and Cotruvo (1990)
-------
REFERENCES
1. Havinga, A. and J. A. Cotruvo (1990) Statutory and Regulatory Basis for Control
of Volatile Organic Chemical in Drinking Water, Chapter 1, p. 3-13 in
Significance and Treatment of Volatile Organic Compounds in Water Supplies,
N.M Ram, R.F. Christman, and K.P. Cantor (ed's.) Lewis Publishers, Chelsea,
MI, 558 pp.
2. Westrick, J J. (1990) National Surveys of Volatile Organic Compounds in Ground
and Surface Waters, Chapter 7, p. 103-126 in Significance and Treatment of
Volatile Organic Compounds in Water Supplies, N.M. Ram, R.F. Christman and
K.P, Cantor (ed's.) Lewis Publishers, Chelsea, MI, 558 pp.
3. Mackay, D.M., P.V. Roberts and J.A. Cherry (1985) Transport of Organic
Contaminants in Ground Water, Environ. Sci. and Technol. 19, 5, 384-392.
4. Gillham, R.W. and P.S.C. Rao (1990) Transport, Distribution and Fate of
Volatile Organic Compounds in Ground Water, Chapter 9, p. 141-181 in
Significance and Treatment of Volatile Organic Compounds in Water Supplies,
N.M. Ram, R.F. Christman, and K.P. Cantor (ed!s.) Lewis Publishers, Chelsea,
MI, 558 pp.
5. Ball, J.W., G.P. Curtis and P.V. Roberts (1992) Physical-Chemical Interactions
with Subsurface Solids: The Role of Mass Transfer, p. 8-10 in Proceedings of
Subsurface Restoration Conference-Third International Conference on Ground
Water Quality Research, Dallas, TX, June 21-24, 1992, Rice University,
Houston, TX, 347 pp.
6. U.S.E.P.A. (1987) Data Quality Objectives for Remedial Response Activities
Development Process, U.S.E.P.A. 540/G-87/003, March 1987.
7. Calabrese, E.J. and P.T. Kostecki (1992) Risk Assessment and Environmental
Fate Methodologies, Lewis Publishers, Chelsea, MI, 150 pp.
8. National Research Council (1990) Ground Water and Soil Contamination
Remediation Toward Compatible Science, Policy and Public Perception, Water
Science and Technology Board, National Academy Press, Washington, D.C., 261
pp.
9. Keith, L.H. (ed.) (1988) Principles of Environmental Sampling ACS Professional
Reference Book, American Chemical Society, Washington, D.C., 458 pp.
10. Nielsen, D.M. and M.N. Sara (1992) Current Practices in Ground Water and
Vadose Zone Investigations ASTM 1118, American Society for Testing and
Materials, Philadelphia, PA, 431 pp.
-------
11. Borgman, C.E. and W.F. Quimby (1988) Sampling for Tests of Hypothesis When
Data Are Correlated in Space and Time, Chapter 2 in ref. 9.
12. Hoeksema, R.J. and P.K. Kitanidis (1985) Analysis of the Spatial Structure of
Properties of Selected Aquifers, Water Resources Research, 21, 4, 563-572.
13. Urban, J.B. and WJ. Gourek (1988) A Geologic and Flow-System Based
Rationale for Ground Water Sampling, p. 468-481 in Ground Water
Contamination: Field Methods, ASTM 963, A.G. Collins and A.I. Johnson
(ed's.) American Society for Testing and Materials, Philadelphia, PA, 1988, 491
pp.
14. First International Symposium on Field Screening Methods for Hazardous Waste
Site Investigations, U.S.E.P.A., U.S. Army Toxic and Hazardous Materials
Agency, Instrument Society of America, October, 1988.
15. Robbins, G. A., R.D. Bristol and V.D. Roe (1989) A Field Screening Method for
Gasoline Contamination Using a Polyethylene Bag Sampling System, Ground
Water Monit. Res. 2, 3, 87-97.
16. Simmons, M.S. (ed.) (1991) Hazardous Waste Measurements, Lewis Publishers,
Chelsea, MI 315 pp.
17. Barcelona, M.J. and J.A. Helfrich (1992) Realistic Expectations for Ground
Water Investigations in the 1990's, p.-3-23 in ref. 10.
18. Bone, L.I. (1988) Preservation Techniques for Samples of Solids, Sludges and
Non-Aqueous Liquids, Chapter 29 pp. 409-423 in ref. 9.
19. Maskarinec, M.P. and R.L. Moody (1988) Storage and Preservation of
Environmental Samples, Chapter 9 pp. 145-155 in ref. 9.
-------
Biography
Michael J. Barcelona is the Director of the Institute for Water Sciences, professor
of Chemistry and adjunct professor of Geology at Western Michigan University. He
received his B.S. and M.S. degrees in Chemistry from St. Mary's College (MN) and
Northeastern University (MA). He completed his Ph.D. in marine chemistry in 1976 at
the University of Puerto Rico (Mayaguez, PR). After three years as a postdoctoral
research fellow at the California Institute of Technology in environmental engineering
sciences, he joined the Illinois Water Survey in 1980. He has been in his current
position since 1989 where he continues to conduct research in environmental chemistry
and geochemistry with an emphasis on oxidation-reduction processes and the
transformations of organic compounds.
-------
Michael J. Barcelona, and John A. Helfrich
REALISTIC EXPECTATIONS FOR GROUND-WATER
INVESTIGATIONS IN THE 1990'S
REFERENCE: Barcelona, M. J. arid Helfrich, J. A., "Realistic
Expectations for Ground-Water Investigations in the 1990's."
Current Practices in Ground-Water and Vadose Zone Investigations.
ASTM STP 1118. David M. Nielsen and Martin N. Sara, Eds.,
American Society for Testing and Materials. Philadelphia. 1992.
ABSTRACT: Significant progress has been made in the design
and execution of ground-water investigations for site assess-
ments, the selection of remediation options and long-term
monitoring of corrective action operations. Improved relia-
bility and cost-effectiveness in ground-water studies can result
from the recognition of both the complexity of the subsurface
environment and the need for simple, integrated monitoring
approaches. Continued progress in the methodologies of
investigation will be enhanced by improvements in conceptualiz-
ation of subsurface conditions.
KEYWORDS: ground-water contamination, background levels,
error-control, monitoring network design
INTRODUCTION
There has been phenomenal growth in the number and complexity of ground-
water investigations in the last two decades. Large regulatory programs have
been established to monitor, assess and prevent or remediate ground-water
contamination. Advances in contaminant detection and assessment in support
Dr. Barcelona is Director of the Institute for Water Sciences and Professor of
Chemistry at Western Michigan University, Kalamazoo, MI 49008. Mr.
Helfrich is a Senior Environmental chemist with Waste Management of N. A.,
Inc., 3303 Butlcrficld Rond, Oak Brook, IL 60521.
3
-------
¦1 c.rii " ir VM
'w,: »;vrSKGMK>r.is
• ¦I f'i:nl:ip>rv ("ici'i.-iiiic Ii;u"c I'rrn rcmnil;il>)r Soil, soil pas. and pniiiml-nTrtc-i
snmplitrp .Hi"' ,iri.ily*i< Kvltniipies sir mw nmlinc'y applied lr> cr>JJTnmii»alinn
mw-siipnliniis Til die partpcrlnllioii level. These ci»ncenlrati:m level' ueic
im clr r>1 concern ?" renrs flp/> Ir ic com mm far Irc.iJ isaum surrmimYmp
7'Hiinp' derisions, property-transfers ami evaluations to liirpc on tIt: results of
«uhsi-i face in. cstip.il inns
Prrnlal'wv pimI .cpnl iivn:'i:- have bee i emboldened hr 1 u- ati.ances in
iincflipniional xcli:ii<|iics in nipccl I'.ipli level? of rrlifiHilily in all grourd-vialer
mvosi.paiii»ris. Ripiil guidelines fur sijdv desipns quality assuranrc/qualitv
ennirnl measures ami rnrilraci cninpliancc piciceduics have often lak.cn
jvrccilcnrr over technical experience, professional judgcrtienl and flcsihly
iuq k-ii ciilal prd ini'-wairi iiuxMipr.iiom (l(- Cipcc'.almns lor future prng.rcss
in simmicUvnicr investigations nm he tempered by *ltc current rcgulaiorv ami
11¦ jn;iL annosplicrc. Mnrlnnt in ncarlv nil ground-water int'f.stig.ilions. Naive
<-*p'.-(mii<i!it|s ClUSj. Sviitimti. pitiussiiiiji. Ii.irdnrss. ctiloridr. llunrUlc. sulfate, nitrate.
iu:mc:incsr rluninium. inm, props aljiiia. gross het.i, and (ol.il radium. Synthetic
I'ff.iiif'' <'nr(i["iur»ls ir t viilalilr t«i|k-rnt.«, fuels, pr*iidilrv etc.) on llsc oilier
1 11 ¦ I. -"ire ¦<' 11'"il 1 l.'.I*, wiiiv'o iii:l,» kre S.-ei'i Mitini'iieri' inlo Hie ^r^c.iril u'alcr
rnvif»iiirneiit as .1 irsull of limn,in niin ilics.
1 l:r" rli-duK 1 ion f-n tiiii: iiud anllirr^poccnic coiMHrHlnn:int.s
8A.RCFJOrj-, WD IIELFWCH <*.' 'MVESHGAllCWJf in Tir 5
is frtfrewhnt Hiirrcd in rlie e.isc of sonic chemical |»ar,-irnrier.s. for e\.unr"!e.
nitr.ilc antl cldorklc can He imnuluccd ftnm fcrlili7ers, snvrr lines, septic
wstews. animal surf r:i.u'.r name a fnr fururcrs A v.iricii of
natural anil man mtlticcd snmccs a' orpanir crmnpiMimU may nlsii mllticufe
': inlcrprctalion ol -sampling aivl r iaV. u-al
results. The mislal abundances llic csieiv lhal (lie solid mineral
phases equilibrate «'i1h prnuti snil
(< UK) ft;33m) samples from 4-1 siles. Focussing on constilncnts u'liic'i arc atc was dciceic:l ai
low average ccrncenlraJioi>s in 7(>, 100 and 92Tc ol these sample lypr1
respectively. Cyanide (CN ^ was delected at & and Wii- fTcqurTicics. rtspectiM'iy
in sand and clay fractions. alJicit a! pp!> levels. Virliolly all o( the average
concentrations o( inorganic chcinicnl con^itviciits v.ctc statistically dilfcicoi in
clay versus sancf samples. It is useful in note that these aullinrs observed
statistically significant, differences (i.e. at the 9st-dcp|. conirihutioiis of iialutal suspended tu colhtid.il
malcrials in grnmul-w-aicr pill nr .iiicrrclm.-irs hrutcrn vdinir-m roinpiniinn:
and Icachahlr/mnhilc metals, eic. fc.c, Toticilv C'liaiacicTMic Icnchmp
I'roccdtirc, (rcf. 11.72}} should nmsidcr hackpiinuid levels vrn rai< Inliy. '1 liese
-------
K rnOllNtl-WATER AND VftOOSt iC'NE INVESTIGATIONS
U £
<1
ii=8
i «5 — — C
C3
S,
H
<1
_ ~ 5 01 ir. r">
r _ r_ o <) m" ~
00 9-~ — O T O
f-
^
'to-®? n
1ft fl O -h
O
• C-l »
*n •
: g 1 cf ^ -r
-V
to
•W f'l o « Ci (5
* o •
e ® ri o
«I q 13 w ,D ^ Si* l! -ST-
?liSU
BARCELONA AND HELFB'.CH OH INVEST IGM IONS II: T'lf l°90<. 7
Mgumcnls apjil)'equally 1<«nrpjinicclifmi'cnI cnrisiiruri'.Lf llirm pll u-c ilrs. Jim brn f
llic f^ncfil of a 'nrg: cliil.W. -r)f Jvirkgrounct cnncnimiinns (1>.H.1S). Tlic
taprrlFilid* here is i;m1 ranjes of iiic-ffwcifir linrfcprii'.unl r^rV- [cri. Trir^i- ranees * if. h? .irr-r;
taocc frequently fsirikc klciilifcalinii f.{cnulnminilcd conditions,dcicmiiiiatinn
nf etc.in-iip levels. and eslim.il icir. of iftc pncciHf.il significance n! ir.vlinMc
.'prcici. ] i would he j-iaiikrutarlv useful if iMndrawry-Unsfd leariiijig pmcfiiurcs
•¦ere Tefirrd tr- bit most rcptcscT-JiUi1; nf acv.ial primnd-walc c nvliiinn* The
curremluiajc: nrBrtl,-aciu ciTacc ale hu Tiers mi = inly mc.iil kmil cmx'-iair
^cwnd ar.J 5Liibec-n:cnira!.iuns of otannic nikrncoiiinminanls. Cn »irluic w;ilc< i: -it V 11 n on
tin'.i>!ial. rep,--«i1-l [Nnrl'-icrn htliiwrM. Knr;h Ailarr.k CrV.rf Vls'a;- mil <:ai^
(WJ, KY, NT) scales. Tlic <.,irvrv mcludci'. d:iM frniri wxl'< Irp^ vhan 2^'Ti reel
(-76Jra) deep and cKCliKletl it;n~ from wcHs lli-l were i>npac(cd K lituwii pni'il
STiirrcf- nf rnnlanirnniiiin. Wmn tfttnl'Tj" tic ti. ilie
ciic-nsi't i.il:trr &F 11>r irnnrt pr.lv a hrk'f Sirrninry i-.r lhcir find up5 Cfli'i be
pre scaled here.
Based on Oic anatvsis nT more iHnn 122.(Ki(i wells in ihc WATSTORF-
tasc. ^hc r.EI tcf>ciTt cc*nclwdci.t (hat 51Tc of ttie nntiim's pmuntf «.ater rcwiurccs
in the upper 25C"'1 icer cxrccr) ;n Scnst one SUV/A MCI.or SMC'L. Watrr finm
drinking w.urr uc||«; w.m generally '>( hiptin vhan unter (th xclls u tn ,15 (vrcrn; An esjiwnr mure rcprcvtuniivc nf ilie
cti'.irr pauiv:! "n.':i1 sivpj'1!'." (i.e. Kii;liuliiit nn(i-
-------
P '"illOtir-rt-VV'TEP wn VA(10>v ZONE INVESTIGATIONS
Nationwide. I DS and total iron were the most common parameters tocxcccil
a dii/ikinp n.iicr si.vnhvd. II,wr) solely on data from drinking waicr wells the
rrr rcpori estimated thai about 3.V7r of drinking water supplies exceeded (he
SMO. for TllS nf <f diiiikiiip water wells. may Ik espcclcd in violate one nr mrtrc general water
¦(tialily standard.
In' tlic sltidies reviewed 11 ft), nitralc was llie parameter that exceeded an
MO. mini frequently. Sixty-eight percent of shallow wells less than KK) feet
deep c-irccded the MCI. of 1(1 F.ncluding wells <1(10 feet deep, on the
basis that Ihey arc most easily impacted by human activities, the frequency with
whirh nitrate exceeded Ihc MCI. was reduced In 7-1(1 pcrccnl of 1 lie nationwide
wells.
I lie I'll rrpurl (lf>) also Summari7cd data front a nationwide survey of
public water supplies{17| lo assess the distribution of radionuclides (fross-alpha,
firms-beta.lot,il radium, radium-22<> and radium-22R) in ground water supplies.
H.iscd on their analysis of the data ihc authors reported thai total radium
(radium-?^ ptiu mfiuin 22fl( twrcdcil the MCI--of ? (iCirt- in Z.ftfr of ground
water siipjiSies. pross-nlplia exceeded thr MCI. nf 15 pCi/1. in 2.2"7c, Ra-226 and
Ha-?..!R exceeded the MCI. of 4 mrcm/year in 1.4% and O.RCr, oT the samples,
respectively. Tlic pio.ss-hrin levels exceeded Ihc MH. of 1 mrcnA'car in only
a small number of the nation's ground water supplies. These data arc
considered conservative underestimates because among other teasons. private
wells and small public svslcins were not included in the survey. The radiological
data reflect geologic and hydrologic factors more so than (and use or sources of
anlhropnjecnic contamination.
Organic compounds: I hc Illinois rnvironmcnta' Protection Agency (IFtl'A)
has licrn rr inducting an onpoinp study ol public waicr supply ll'W.S) wells in Ihc
state since I'1R5 [ IK| More than 75Ti- nf Illinois' I'WS wells had been tested for
inoipanie parameters ami volatile organic compounds (VOCs)hy 1981. Statistics
en iimignnic parameters (summ;ui?cd above) were compiled (or more than H((l
wells and for VOC's mi nil 2ft h'l wells.
"I lie- IE-PA survey iiiclndc"; had V VI
o5
H
§g
o <
a. i£
°o
•y. >
¦ w =
¦J *
£
a
rJ „
¦;
uj :
$."
LL !
Uv
cf?
I
^ 9
a 2
OQ
ts
<
OQ
U.
O
>-
s
to
1
rt
jo
-O
'¦O-
J
f?
o
o
V|
o
^ -
Q S
"3^
O
Jb_
s
3
i 5
£ ^
ii
Si
z. ^
f !°
? !Q
f
r»>
tf
o
o
V
yf
v
J
Sr.
E
n
a.
0
o
II
6
j
&-
p.
B
-------
10 r.rsonrjti WATEn -aui i vvnOSE ZOt,r. IWESUGMIONS
detections revealed potential sources of comamiumion within 1(100 feel of
ihe well head.
The PWS gronndwaier quality situation is similar in California where
detection frequencies of VOC's (i.e., perchloioetliylenc, trichiorocthylcnc,
dihromurhlompropane. chloroform, etc.) ranged from - 2 to 3.5% of all wells
sampled |I9j The state had sampled 52% of the large (i.e., >20(1 service
connectors) PWS wells. A recent discussion of the relative frequencies of
detection of sclcclcil VOC's in PWS wells broaches some interesting insights for
the nailer |?ZJ. In this instance. the authors point out the low detection
Itrqirency of heiwene in Califoinia public water supply wells. neglecting the fact
thil feu- production wells aie screened at 1he water lable where fuel-related
contaminants ate more likely to he detected.
1 hese dam illustrate llwl Hie ambient quality of ground water is highly
variable and frequently fails to meet current dunking walcr quality criteria.
Some of the apparent contamination may lie attributed to human activities, as
in the ca^c of nilralc. VOC's and pesticides. However, many of the reported
S15WA standard cscccdaticcs ate due to natural conditions. As Ihe authors of
(lie t:R icprirl (ICS7J conclude: ".. (lie natural quality of nlmost ftof the
nation s ground ntitcc resources ahi'tvc depths of 25f the annual means. While it
out that quarterly sampling frequency is a good parting poini Inr monitoring
designs [33) for temporal trend analyses. I lie estimation of a cosi-effective,
rchahle frequency for trend detection or estimation of mean values remains
largely an empirical task. Working with an ovci-dcicvrr.med (i.e. biweekly
frequency) d.ilasct for ground-water quality nl three sites in a shallow sand and
gravel aquifer, it was possible to define an 'optima!'' sampling frequency which
resulted in maintaining a specific information content with minimal sampling
effort. The information content was defined as Ihe ratio of the effcciivc sample
(ncn) size to ihe loial sample size (r) for cacli of 24 water qunliiy parameters.
In most eases, (lie frequency maintaining an information content of ft.*) (i c less
1han 107c loss of informalinn due to redundant data collect inn) ranged from one
to five months at the three sites. Tigtirc 1 coniains plots ol the values of n,fIAi
versus ground-water path lenplh corresponding ifv the "optimal" sampling
frequency for selected water quality parameters at an uncoiiiniriimird location
-------
12 r.nruiNo Wat en aup vadost zonf investigations
Flow Pnlli (ml
0 0.2 0.4 0.8 O.B 1
neff/n
pll - fond. -O- Tfiap.
-1 CI- TOC 02
FIClUtF. 1
I'll.!1: nl (Mound-walei n|. It is recommended lhal one include
some target contaminant constituents (e.g. VOC's, total organic cavhon). major
ionic constituents (e.g. CI, SO/, Na. K, etc.) and gcochcmical indicators (eg.
CO,. CH„ Oj, Fell) in many initial ground-water investigations. Tfic
expectation here is tliftt. based on a sound hydrogcologic framework, and initial
flow or transport modeling results (i.e.calibratcd for CI', 0'. etc.). preferred
locations for chemical sampling could be idcnlificd. Then a reasoned, limiied'
suite of likely contaminants and breakdown products could be selected for long-
term monitoring if this purpose is relevant to the investigation. Redundant
sampling of specific organic compounds (at or near detection levels) for long
periods of time is simply too wasteful as compared to the improved information
return from a simpler suite of target analytcs determined at more sampling
points.
Well development: With the increasing emphasis on the collection of
improved geologic and hvdrologic inforrnalion in ground-water investigations lias
come the realization lhal the hydraulic connection between the sampling point
and the geologic formation is very important. Tumping monitoring wells for
development of the gravel pack, removal of fines or removal of stagnant water
prior 1o sampling have become as important as routine slug or pump-testing to
determine drawdown, yield and aquifer properties.
The removal of fines created during drilling has ial.cn on particular
importance given the potential for colloidal metal (e.g. A' or Fc) or organic
transport in gravels or fiacturcd materials (1 (1.37( and the confusion about the
need for sample filtration (?K| What is obvious is ihai c\tra can' is needed in
the construction, design and pumping of wells to minimi7c the creation and
cntrainmcnt of fine particulates which arc unlikely to be "mobile under natural
-------
1'1 C.FVXIfJO WMFn AND VADOSE ZONE INVESTIGATIONS
gradient conditions. Just as vigorous high-rate pumping and surging ol wells
timing development mot1ili7.es particulates, one should expect furllicr
development, lurliidity and potentially biased water samples to result from high
velocity pumping during purging or sampling (39). The expectation here will he
that future investigations will more carefully design, construct and develop wells
to minimize the collection of otherwise immobile fine materials. Also,
considerable care will he taken to prevent wclf/gravcf pack damage by high-rate
pumping after development.
.Well purging: The pumping strategy which is appropriate for individual
monitoring wells should include the development operations described above as
well as pumping during wctl purging and sampling. The primary consideration
in purging is the isolation (via packers) or (he removal of stagnant water in ihc
casing, screen and borehole adjacent to the screen (40,41,42) which has been
defined as the purge volume (43). The stored water is considered
unrepresentative of ground-water in the formation of interest. Criteria for
sample representativeness vary substantially. Tlicy may he characterized as: 1)
pumping a specific number of purge volumes, 2) pumping Ihc well until water
quality indicators (i.e. pU. conductance, temperature, dissolved oxygen, etc.)
sinhili/c satisfactorily, and, 3) pumping the well al a uniform rate for a liinc
period sufficient to permit .1 high percentage of formation water to enter the
well s screened interval (42.44) A number of workers have documented water
quality changes during purging in an attempt 1o arrive al a generalized,
consistent sampling recommendation |4?,44,45,46,47,48).
The results of these investigations have disclosed Ihree importnnl points.
71ie hydraulics of monitoring nells are incompletely understood (49). Order of
magnitude concentration changes in general water quality parameters, metals
and specific nrgsinie compounds occur in pumping idled monitoring wells
between 1 and 10 purge volumes. Well design (i.e. screen placement length,
slot s'i7*> construction, development, performance, rate of purge pumping and
placement of the pumping device can introduce order of magnitude level
changes In concentration during purging which may negate the impact or
sample location, frequency nnrf nnalyte selection ns major network design
variables. T he laltci theoretical and experimental gains have been made-by
Kobbiris ct al. in a series of papers |50.?1] which began with considerations of
weft-screen averaging and mass-continuity balances during purging (72).
I he expectation here is that carefully integrated hydrogeology-based
sampling protocols will be developed in 'lie future which incorporate more
vertically discrete well completions and Ihc use of models to improve our
concept of sample representativeness. In the meantime, empirical evidence
suggests thai low-rate pumping (so as not to cause substantial draw down) from
the mid-screen portion of a well and removing a consistent number of purge
volume* front cadi well prior to sampling is a sate approach. Clearlv. the
iicr"t-"i|\ n( purging has been demonstrated and purging procedures should be
documented (i c hydraulic basis, estimated purge volume, purging indicator
paramnrr measurements) lor each water sample.
BARCELONA AtJO HELFFltCH ON INVESTIGATIONS IN T1 ir iQuris 15
Sampling device selection: There have been a number of studies which have
cither discussed or experimentally evaluated the attributes of ground-water
sampling mechanisms 152-621. Many o( these evaluations have been based on
the precision and recovery of volatile organic compound determinations on
replicate samples collected in the field or laboratory. The volatile compounds
are particularly sensitive to losses by degassing at reduced pressures from suction
devices (e.g. peristaltic or centrifugal pumps, etc.) or turbulence in some
mechanical sampling devices (e.g. gas lift, bailers, etc.). In general, positive
displacement pumps (e.g. bladder pumps, etc.) which operate rcproducibly over
a range cf lift, hydraulic head ami depth conditions have been shown to provide
accurate reproducible sampling performance under testing condition-;.
In practical routine field usc.lhcre arc additional considerations wliich bear
on Ihc selection of a sampling device once sufficient accuracy nn<1 precision have
been demonstrated. Mainly these considerations have to do with the reduction
of operator dependent error (e.g. random error in bailing, excessive sample
handling/transfcrs, etc.), cost, ease of repair and cleaning. Dedicated sampling
systems suitable for both purging and sampling arc clearly preferred from the
points ol view of convenience, cost and consistency of results.
The expectation here is that along with Ihc use of more discrete well
completions with short screen lengths across a site that the use of dedicated,
reproducible, sampling devices for both purging and sampling will improve the
comparability and consistency of results. Since well design! pumping, and
purging effects can all lead to order of magnitude errors in the determination
of chemical constituents, there is much to be said for the selection of
dedicated, simple sampling devices which provide long-term control over human
errors.
Field determinations: Field operations at ground-water investigation sites
may inc'udc sample handling, measurement of wed purging parameters (i.e. pM,
conductance, temperature, dissolved oxygen) or alkalinity, sample filtraiion,
spiking samples with standards and decontamination procedures jfi2-fifij. Most
of these operations must indeed be accomplished at the well head for the
samples and subsequent measurements to be valid. Tlicy have been greatly
facilitated by in-line well-purging parameter measurement systems [31 ,fi3/>7|. in-
line filtration devices and increasing attention to the documentation of sampling
details. The rxpeclnlkin here is thai sampling and analysis steps which ijiclnde
sample pumping, collection, handling anil field determinations involve errors that
can and will be controlled through attention 10 ctror and simple sampling
protocols f.33j. In this way better estimates of spatial and temporal variability
can be obtained at known levels of confidence. This will greatly improve the
appro* nintion of actual contaminant distributions, transport and (ale in ground
water.
Dal a .?I,i,h,A's._r,n<,..in.terpirtaii(>n: Improved data qualiiv and confidence
come Ironi both the application ol informed technical judgement and normal
attention to quality assurance and quality control. The serious levels of error.
-------
i o r.fiocrm mrEH and whose" zowe imvestiganoms
nr apparent variability. which have been identified with well )ncation, design and
pumping clearly call fur t|iialily assured decision-making. As the radrc of
experienced ground-water investigators has grown, ihc inlcrprctalion of water
quality rlata has improved proportionally. Improved software and statistical
*nrls Ui* handling pi wind-wal cr quality da'will undoubtedly be developed
1n aid hi llic irMcgrnlinn cf -chemical and hydrogen logic information. The
expectation here is thai investigators will he given increased laliturte in ndworSt
design and operation to maximize Ihc information return from ground-water
investigations. Since the bulk of the scientific and engineering data thai will-be
c<-illcclcl»ci1 Hint flcxil»i1i»y will lie framed in these efforts.
conclusions
Our understanding of subsurface geochemistry and hydrogcoiogic controls
mi contaminant behavior has improved substantially in recent years, ll is
anticipated that, in the next decade, the purposes and designs (or ground-walcr
investigations will incorporalc mote in-depth considcralidns of natural variability
in chemical constituent concentrations and of the need for careful error
• identification and control. The increased use of integrated hvdrogcologic and
gcochcmicai data collection strategics should lead to more meaningful
performance criteria for valid monitoring efforts and results, lltnay be expected
that groiind-wairr investigations wilt expand in (he {uturc to adopt more Jong-
Icrm moniiniirg approaches to both contamination identification and
irmrdialinti.
ACKNOWl.P.nOUMF.NTS
'flic authors appreciate the help of Ms. Bonnie Duhc, Ms. Jancnc Hoover. Ms.
I.imla lours. and Ms. Kim Finkbcincr in ihc preparation of ihc manuscript-
nnrrnrNiTS
|1| Frcc/c. HA. and Cherry, 5,A. ftWI). What lias grinc wrong? firaanrf
VVal£t. 27. <1, .1SK-4M
|?) fiillimu. H.J.. Itirsch. H.M.. and Oilmv, P..J. (1(>M). Effect of Censoring
1 r.uc lxvcl Water Qtmlitv Data cm Trend Detection Capability,
f'nvinimmntal Science and Tcrtmnlciyv. ifi. 5W-W>.
BARCELONA AND HEt.FIUCH ON INVESTIGATIONS III THE 1?9Us 17
Gilbert, U.O. (1*>R7). Statistical Methods tor r.iivironmcntal Tolhmon
Monilorinp. Voj> Nnstiand Reinhold, N.Y.
Stumm, W. and Morgan. J J. < f5-171.
Dragun. ). (I^RSj. The Siii] Chemistry c>1 Ha7nrdons Materials.
Hazardous Matctials Cnntrol Rrscatch Institute. Silver Sfirinp. Ml).
-------
Grtatirm-vwiFn AtlD vadose ZONE INVESTIGATIONS
Stainc. !).n ami tlarkcr, J.F. (190(1). Tiic Detection of Naturally
¦Occurring Il l .X Ihuiiip a I lydrngcolngic Investigation. Ground VVaicr
Mori!(i>rmcJicvjcw, }tj. 2, K4-94
F'disnri Plcclric 1 usiiiulc (1987). Natural Quality of Groundwater in the
United-Stales.Compared tr>Drinking Water Standards. Prepared by Roy
r. Wcsion. Inc.. Scpicnnci 1987. U, p. Wcsirwi, Inc. Weston Way, West
CticMcr, rA WRti.
Uonon. T R. (1 '*P5 J t-Jationwidc Occurrence of Radon and other
N.-iliiraJ Radioactivity in Public Water Supplies. USEPA 520/5-85-008,
2<«p.
Illinois n.nvironmenial Protection Agency (1090). Illinois Groundwater
Protection Program: A Biennial Rcpt>rt Prepared by (he Interagency
-Ci'ordinaiing Committee on G round water, February, 1990.
California Dept. of Health Services (1986). Organic Chemical
CotiUmiuruinn of Larcc Public Water Systems in California, Calif. 1)1 IS,
Sacramento. ("A, April 1986.
(VS. f:nviriiiimcnial I'roteclion Agency < 19R6). Safe Drinking Water Act -
1«' Ainrndnicriis. USF.PA 57CI/9-86-d02. Office of Drinking Walci,
Washington. D C-
Miis. Jl.l.. artd Johnson. I..G. (IWif). PaKutinn Prohlems in Jo#a, in
I'J. Ilnricl;, led). Wnler Resources of Iowa. Iowa Acad. Sci. Univcrsily
Printing. Uwa Citv. I A, ppfllM)D.
Mndlcy, P.W. ami Armstrong. K. (IW). "Where's the Benzene?"
[-xnniiniiip Calilnriiia Ground-Water QuaJilv Surveys. Ground Water. 29.
I.
NicUco. II. (I:,(JI) let!.). Ground-Water Monitoring. l.cwis Publishers,
Chelsea, Ml.
American rctrolciiNi 1 nstitntc fl9fW). "A Guide lo the Assessment and
Remediation of Underground Prtrolcnm Releases." 2nd ed., A['l
I'nWivntinn American TYimlctim Institute, Washington. D.C.,
I Wt.
Keith. I II. M'tSfS) alitor. 'Piitniplcs of Flnvircinneriial SjinipJing"
Atiicticau Chemical Society !'mfcssi»nal Reference Book. ACS
W:isliin^imt, D.C.. 4SJ1 pp
BARCELONA AND HElFRICH ON INVf STIGATfONS IN 1HF. imiiV. 19
USEFA (1989). Cicostatistical Environmental Assessment Software (Geo
EAS), EPA-Environmental Monitoring Systems Laboratory. Office of
Rcr.carcli and Development. Las Vegas, NV.
Freeze. R.A.,Massman, J., Smith, L., Sperling, T , and laws. r>. (1WD).
Mydrogcnlogical Decision Analysis, 1 A. Framework. £jiroiind_Waie>. 2i<.
5, T3S-766.
Suchomel, K.H.. K reamer. F^.K. and l.ong, A (|9'NJ| Production and
Ttanspott e>5 Carbon Wioxuta in a Contaminated Vadosc 7.onc: A Stable
anc Radioactive Carbon Isotope Study. Environmental Science and
Technology 24, 12, 1824-1831.
Kaback, U.S., Bcrgren, C.L.. Carlson, C.A., Carlson, C1,. (IIW) Testing
A Ground Water Sampling Tool: Arc the Samples Representative?
Presented at National Water Well Assor.-Awr. of Grotnid Waicr
Scientists and Engineers Outdoor Action Conferences May 14-17, I'/<>(!.
Las Vegas, NV.
Barcelona, M.J., Holm, T.R., Scltock, M.R., George. G.K. ( 1QRO) Spatial
and Temporal Gradients in Aquifer Oxidation-Reduction Conditions
Water Resources Research. 25, 5, 991-100?.
Wallnn-Day, K., Macalady, D.L., Brooks. M II.. Tale..VT. (IWij. Field
Methods lor Measurement nf Ground Water Redos Chemical
Paiamcicrs. Ground Wafer Mcmit-orng Rcn'ru-. HJ. KI-S1.
Rohhins, G.A- (1989) Influence of Using Purged and Partially
PcnciTaltng Monitoring Wells on Contaminant Detection. Mapping and
Modeling. Ground Water. 27- 2, 155-U>2.
Barcelona, MJ-. Lciseiunaict, I).P. avid Schocfc, SLR (1W)) Ncnvork
Design Factors fnr Assessing Tcmpnnl Vatinl'i ilv «r G round-Hal or
CKtslitj'. f-nvimrmienlal Monitoring & Assessment 12. H'l-H1).
Pltmh, R.H. (I9K7). A comparison of Ground-Water Monitoring 1 j;v*s
from CF.RCl.A and KCRA Sites. Ground Water Monitoring Review 7,
4, 94-10(1.
Fx ri, 1.. Cortiicllv. J.I1, and l.imlnrff, IJ.M. (1^7). Groiinil U'^icr
Sampiinf: Addressing lite Turluilent Inconsistencirs p. 237-255 in
Proceedings of !.hc N'WWA Otiitloor Action C'unfcrcrcc. May l')S7, J.as
Vegas, NV. National Water Well Association. Dublin. Ohio.
-------
GFt'iUNP WAlFFt ANU VAOOSE ZONE INVESTIGATIONS
Plumb. k.ll. (I't'M). The Occurrence of Appendix IX Organic
Constituents in Disposal Site Ground Walcr, Groundwater Monitoring
Review. 11 (in press).
Gscbwcnd. P.M. ami Reynolds, M.D. (1987). Monodispcrsc ferrous
Phosphate Colloids in an Anoxic Ground-Water Plume. Journal of
C'ontamijiaiiMJj'^W'k'C;. 1- 30^-315.
I'iiI*. R.W.. ami Barcelona, M.J. (1989). Filtration of Ground-Water
Samples (or Metals Analysis. Hazardous Waste and Hazardous
Materials. 6. 4, 385-393.
t'uls, K.W., F.ychaner. I.II., and Powctl, R.M. (1991) Culloidal-
l:nrilitatcd T ransport of Inorganic Contaminants in Ground Walcr: I'art
1. Sampling Considerations (irt press).
Rnrcclona, M.J., (iihh. J.P.. Ilelfrich, J.A., and Garskc, F.F. (1985).
Practical f iiiiilc for Ground-Water Sampling. Illinois Slate Water Survey
Contract. Report 374. Champaign. 1L, 94pp.
Robin. M.I.I., and Gillham, R.W. (1987). Field Evaluation of VVell-
Putgiiig Proccdiircs. Ground Walcr Monitoring Review. 7, 3, 85-93.
Keely. J.F. ami Poateng. K.. (1987). Monitoring Well Installation Purging
and Sampling Tcclinit|ucs - Part 1: Conceptualizations: Ground Walcr.
.25, 3. 3DD-313.
Rul'hins. G.A.. Bristol. R.D.. Ilayden, J.M. and Stuart. J.D. (1989). Mass
Cnniimitiv and Distributions: Implications for Collection of
Representative Ciroiind Water Samples from Monitoring Wel.ls, p 125-
13". ] n I'roc, of Conf. on Petroleum Hydrocarbons and Organic
Chcniicals_in Ground Water: Prevention. Detection and Restoration.
Nnv. 15-17, IWJ. Houston. TX. National Water Well Assoc., Dublin,
Ohio.
Gibs. J. ami Imbripoita. T.F. (1990). Wcll-I'urging Criteria (or Sampling
Purgcablc Organic Compounds. Ground Water. 28, 1, 68-78.
Barcelona. M .I. and llclfiich. J.A. (1986). Well Construction and Purging
rflrcts on Ground-Water Samples. Linvironmcntal Science and
Jcchnology. 2(1. II. 1179-1184.
Perutiiv>. ID. (PiKJi| There's No Such Tiling as a Representative
Giniiml Water S. inplc (iriniiid .Wajcr_Monitorii,t^Rcvicw. S. 3. 4-9.
BARCELONA AND HELFfUCH ON INVESTIGATIONS IN THE 1990s
21
Smith, J.S.. Steele, D P. Mallcy, M. J. and Bryant. M.A. (1W). Chapter
17. p255-260 in Principles of F.nvironmcmal Sampling. 1..1! Keith (ed).
American Chemical Society Professional Reference Hook ACS,
Washington, D.C.'
Schullcr, R.M.. Gihb, J.J\, Griffin. R.A. (1981). Recommended Sampling
Procedures for Monitoring Wells. Ground Water Monitoring Review |
1, 42-46.
Unwin, J.P. and Malthy, V. (1988). Investigations of Technique's for
Purging Ground-Water Monitoring Wells and Sampling Ground Water
for Volatile Organic Compounds in "Groundwater Contamination Field
Methods". A.G. Collins and A t. Johnson (cds.), American Society (or
Testing and Materials, STP #963, pp. 240-252.
Rolibins, G.A.. Martin-!lavdcn, J.M. (1991). Mass Continuity Modeling
of Monitoring Well Purging. Journal of Contaminant Hydrology (in
press).
Robliins, G.A., Martiii-I laytlcn, J. M.. Bristol, R.D., and Stuart. J.IX
(1991). A Field Study of Mass Continuity Influences on the
Characterization of Subsurface Gasoline Contamination. Journal of
Contaminant Hydrology (in press).
Barcelona. M.J., Ilelfrich. J.A., Garskc. EE., and J.P. Gihb (l*>8t|. A
Laboratory Evaluation of Ground Walcr Sampling Mechanisms. Ground
Watei Monitoring Review. 4, 2, pp. 32-41.
Nielsen, D.M. and Yeatcs, G.L. (1985). A Comparison of Sampling
Mechanisms for Small-Diameter Ground Water Monitoring Wells.
Ground Water Monitoriup Review. 3. 2. pp. 83-94.
Sonnlag. W.I I. (1087). Comparative Tcsl of Two Sampling Devices (or
Obtaining I'urgeable Organic Compounds frum Ground Water Wells.
U.S. Geological Survey Open-File Report, 87-109, p. F.-7.
Pcarsall. K.A. and Eckhardt, D.A. (1987). Effects of Sclccicd Sampling
Equipment and Procedures on the Concentration of TricltloTOctfiylcne
and Related Compounds in Ground Walcr Samples Ground Water
Monitoring Review. 7. 2. pp. 64-73.
Pohlmann. K.F. and Hess, ) \V. (I98R). Generalized Ground Water
Sampling Device Malrix Ground Walcr Mfuiiltiriitp Review. 8. 4. pp.
R2-S4.
-------
GnOUfJP WATER AND VADOSE ZONE INVESTIGATIONS
Imlirigintin. T P.. Gibs, J.. Fusillo. T.V., Kisli, G.R. and llochrcilcr, J.J.
(1PSS). rjcUl Evaluation (if Sampling Devices for Purgeable Organic
compound in "Ground Water Ground-Water Contamination. Field
Methods'' ASTM SIT w>3, AG. Collins and A.J. Johnson, (eds.).
American Society for Tesiing and Materials, Philadelphia, PA, p. 258-273.
Barker. J.P., and Dicklwut. R. f 198P). An Evaluation of Some Systems
for Sampling Gac-Charged Ground Water for Volajile Organic Analysis.
Ground Water Monitoring Review g. 4. pp. 112-120.
Rchni. H.W., Sinluenhurg. T.R., and Nichols, U.G. (1985). Field
Mc.Tincmcnt Method". for T tydrogcologic Investigations. A Critical
Review of the Literature. FPR1 I'roj. Report EA-4301, Project #2-485-7,
October 1985 hy Residuals Management Technology, Inc., Electric
Power Research Institute, Palo Atlo. CA.
Lmviti. J.r. (IW4). A Laboratory Study of Four Methods of Sampling
Gioiiml Water for Volatile Organic Compounds. National Center for
Air and Stream Improvement, Technical Bulletin. No. 441, August 1984,
Ifipp.
Hlogen, R.I'.. I'ohlniann, K.F. and llcss, J.W. (1987). Bibliography of
Ground-Water Sampling Methods. F.I'A Report EPA/600/X-87/327,
September 1987, L'SEPA-EMSL, Las Vegas, NV.
Pankow. j r. (lWOj. Minimization of Volatilisation Losses During
Sampling and Analysis of Volatile Organic Compounds in Water.
Chapter 5, p. 73-86. In Significance and Treatment of Volatile Organic
Compounds in Water Supplies. N.M. Ram, R.F. Christman, and K.P.
Cantor leds.). Lewis Publishers. Chelsea, Ml, 558pp.
(iariter, S.. {I
-------
Chapter 1
Overview
of the Sampling Process
Michael J. Barcelona
The involvement of chemists, and analytical chemists in particular, can substantially
improve the result of sampling and analytical programs. The development of
meaningful sampling protocols demands careful planning of the actual procedures
used in sample collection, handling, and-transfer. Critical aspects of sampling protocol
development cannot be covered by traditional field and laboratory quality control
measures. Criteria for sample representativeness must be developed with careful
attention to the physical, chemicak and biological dynamics of the environment under
investigation. Preliminary sampling and a well-conceived sampling experiment can
provide the validation and experience necessary to design efficient sampling protocols
that will meet program needs.
Sampling in THE ENVIRONMENT for chemical analysis is a complex subject,
and can be as varied and complicated as the objects that must be sampled
to investigate the environmental effects and fates of chemical species on
the planet. The subject does not lend itself to textbook treatments for students
and has long been given diminished importance relative to the improvement
and verification of analytical methods for environmental applications. Indeed, in
reviewing- environmental opportunities in chemistry, a recent National Research
Council report (I) repeatedly cited the need for improved sensitiviry and
selectivity of analytical techniques and only in passing mentioned improved
sampling for chemical analysis. However, that which cannot be reliably sampled
is seldom worth the care and expense of analysis. Also, numerous analytical
problems exist with chemicals" in the environment at the parts-per-billion
(ng L~') and parts-per-trillion (rig L"1) concentration levels without delving
further into the frontiers of ultratrace analyses of air. water, or land environments.
U73-6/88/0003S06 25/0 ® 1988 American Chemical Socierv
-------
PRINCIPLE; Or ENMRCNME.ST.Ai. SaMPUS'C
Many scientists, however, recognise the need for accurate and precise
environmental sampling as well as analysis.
T'r.is chapter has two objectives for tmptoved experimental design for fieid
sampling problems. The main objective is to encourage the development of
accurate and precise sampling protocols that go beyond the confusion of general
methods, techniques, and procedures available at the present dme. Protocols are
thorough, winter. descriptions of the detailed steps and procedures involved in
the collection of samples. This sampling \-alication objective is analogous to chat
described by Taylor (2) for analytical protocols to avoid the serious conse-
quences of syswrrair:: error (Le bias or inaccuracy) on the results and
conclusions of erMtor.x.ir.uil stocks. These studies, may be J ride r taker, xi
investigate contaminant distributions or the potential for exposure to contami-
nants for a variety of purposes.
Field handling and procedural blank determinations that permit meaning-
ful evatuaiiOTis of conutiviuvmi. cj-posurci have been the focui of several rr-rPm
discussions. These determinations are particular!)' useful in interpreting the
effects of systematic errors on irii.erlabcrat.ory analytical comparisons (3-5). This
chapter supports the identification and control of sources of sampling error
when such errors exceed these inherent m analytical determinations.
The more subde objective of this chapter is to encourage a realistic
appraisal of the practical limits that systematic sampling errors and resultani b'as
place on the purpose, results, and conclusions of studies in environmental
chemistry. This objective has been included because although the actions of
sampling may be deceptively simple, they must be carefully planned, refined,
and documented if truly representative samples are to be provided for the
purposes of the investigation. Furthermore, the results of environmental studies
are frequently extended to purposes other than those of the original investiga-
tion. These objectives apply equally to specialised research as well as routine
regulatory or compliance monitoring efforts because the results of research
investigations are often generalised to a variety of environmental conditions.
Review of the Literature
Sampling for chemical analysts has been reviewed critically in the chemical
literature. The excellent works of Kratochvii and co-workers (6. 7) provide a
sound basis for chemists interested in recognizing and exploring chernica!
sampling problems. Their main emphasis is quite practical and the review
article (6) includes valuable citations of past work for a variety of environmental
matrices.. These authors carefully point out the roles that statistics (8) and
cbemorneves (9. fC) can pby in nesoinng sampling problems. They further
undersccie the vw.ee Cor the irxcU-eraeru of chem.sts. snd rr.oie specific-H;-
anaivsts. in the planning, execution, and interpretation of the results of sampling
and analytical efforts.
-------
BARCELONA Overview of the Sampling Process
5
An environmental scientist's view ot' the sampling process is often quite
different from that of j statistician. The scientist may be interested in
representative samples of water from the hypolimnion of a particular type of
lake. The statistician, on the other hand, may envision samples as a subset of
the universe of all reducing surface water samples. In the environment, collecting
samples (objects) from a largely uncharacterized universe of objects is often
required. Only through some prior sampling experience can a sample or sample
population be related to the universe or parent population that is the territory
of statistical theory. This prior sampling experience can also permit some
generalisation of the results if the experiment is properly designed. The
distinction between samples and parts of a parent population can be better
illustrated by a reexamination of the types of objects and samples that are
ultimately collected for analysis.
Kateman and Pijpers (II) categorized objects (e.g., a well-mixed fluid) from
which samples are derived on the basis of the degree of homogeneity and the
nature of the spatial change in a particular quality (e.g., dissolved lead content).
The aim of their categorization was to lead into the corresponding types of
samples and sampling strategies that are needed for analytical quality control.
Their scheme has been expanded in Figure 1 to include temporal as well as the
spatial change inherent in environmental sampling. Spatially heterogeneous
objects or sample origins present a much greater challenge to accurate sampling
than homogeneous objects that may only exist in the laboratory. This greater
challenge comes from the need for both more specific criteria for representative-
ness in heterogeneous populations and for more detailed characterization of the
conditions under which sampling takes place. For example, representative
sampling of dissolved lead in a well-mixed solution in a beaker is rather simple
compared to sampling dissolved lead in a variable mixture of reactive aqueous
effluents entering a treatment plant operation. In the treatment plant, the
effluent mixture is a much more complex object both in composition and extent
as well as in variability in space or time. Useful criteria for representative effluent
samples would necessarily include qualifiers of flow rate, process status, time,
and perhaps other physical and biological variables. Kateman and Pijpers (II)
provided a general review of the relationships between object types and
theoretically optimal sample sizes, numbers, and frequencies for a variety of
applications.
For each type of object or sample origin, corresponding types of samples
or subsamples can be found that result from sampling, pooling, or compositing,
as well as the reduction or preparation steps prior to analysis. Samples suitable
for analysis must be representative parts of the object. Figure 2 contains an
expanded overview of the nomenclature of sample types suggested by Kateman
and Pijpers (II).for field sampling applications. Increments (or grabs) issue from
the object or parent population. Thcv are parts of the object, but because they
are not representative parts of the object, they are not called samples. The gross
sample (.bulk sample) may be seen as a pool of two increments that is reduced
or prepared as subsamples for analysis. Field control samples and laboratory
-------
OBJECT POPULATION OBJECT
DEGREE Of HOMOGENEITY
OF OBJECTS
HOMOG
ENEOUS
NATURE OF CHANGE IN QUALITY
THROUGHOUT OBJECTS
HETEROGENEOUS
(
I
OlSCRETE
I
I
I
I
CONTINUOUS
I
I
(
I
I
I
I
1
I
DISCONTINUOUS
I
I
I
I
I
E)
(AMPLE
S
I
WELL-MIXED GASES
ORE PELLETS
REACTIVE
GAS MIXTURE
REACTIVE GAS PLUME
CM WIND HELD
WELL-MIXEO LIQUIDS
CRYSTALLINE
ROCKS
REACTIVE
SOLUTIONS
REACTIVE EFFLUENT MIXTURES
ENTERING TREATMENT .PLANT
PURE METALS
TRUE SOLUTIONS
SUSPENSIONS
SUSPENSIONS
DREDGED SEDIMENT
DISPOSAL OPERATION
MACROSCOPIC GRADIENTS
CHEMICAL
NO
NO
YES
YES
PHYSICAL
NO
NO
POSSIBLE
YES
Figurf I. Typrs of niiurnsfopif olijrcls or slr origins.
5
o
T|
m
2
/
2
pi
>
n
/
o
-------
9
3
-j.
3
&
3
I
ANALYSIS PROCEDURE
LA80«AT OflV
controls
I EXTERNAL RtFtflENCE STANOAftOS
2. Sample niMkirni'lcJfuj'f wet \>ievr.
-------
8'
Principles of Environmental Sampling
controls enter into the analysis stream with aliquots of the samples. Once these
controls and the subsampies have been prepared, negligible sampling error is
assumed to occur when aliquots are taken. For the purposes of the present work,
the division between sampling and analytical errors is placed after the reduction
step that is shaded in Figure 1
Representativeness in sampling, therefore, presumes thai the analysis of the
sample or duplicate.samples shows the same results as would the object itself.
In more practical terms, the characteristic analyte or quality of the sample must
be identical to. or minimally disturbed from, the objects quality. Sampling any
object will be subject to random variations. In representative sampling efforts,
however, we strive to identify and control systematic deviations, or determinate
error, caused by sampling.
Planning for representative sampling should be made an integral part of the
design of environmental studies. A number of references treat statistical sampling
designs within study designs that are worth careful attention. These references
include works that deal with biological (12-14), geochemical (f5), and water
quality monitoring studies. Among these, the methodology presented by Green
(!2) is particularly ¦well-suiied to environmental sampling For chemical analysis.
Green's methodology rests on a clear conception of the problem or question
that needs to be answered. His suggestions included the Following:
1. replicate samples within each combinadon of time, location, or
other controlled variable;
2. an equal number of randomly allocated replicate samples;
3. samples collected in the presence and absence of a condition in
order to test whether the condition has an effect;
4.. preliminary sampling to provide a basis for evaluation of
sampling design and statistical analysis options;
5. verification of the efficiency and adequacy of the sampling device
or method over the range of conditions' to be encountered;
6. proportional focusing on homogeneous subareas if the sampling
areas have large-scale environmental patterns;
7. verification of sample unit sire as appropriate for the sire,
densities, and spatial distributions of the objects that are being
sampled, and
8. testing of the data to establish the nature of error variation to
decide whether to transform the data, utilise distribution-free
statistical analysis procedures, or test against simulated null-
hypothesis data.
Green concluded his suggestions with seasoned advice: Once the best statistical
method has been chosen and has provided the test of the hypothesis, accepting
the result is preferable to rejecting it. and searching for a "better" method.
-------
!. ihhcell'sna (Jiwtow t'fibc iiimpting Prrass
Mora theoretical samp.ing design references (Tua\ also be -isefu;. and -£Meral
have been renewed in detail 1,7). TKe theoretical '-vork of Gy (19) (dealing with
solid sampling) is interesting, particularly one of his more recent publications
1.20). In the more recent work, he makes the distinction between sample
handling and sampling as an error-generating operation. This statement is an
acknowledgment that elements of a sampling operation may cause serious errors
and cannot be treated strictVy by statistics. Examples-of such error-prone
elements of the sampling operation are sampling locations and sampling
mechanisms or materials. Unlike sample size, number, or sampling frequency,
which are elements that can often be handled statistically, identifying and
controlling the previously mentioned sources of systematic error are verv
difficult without prior sampling experience (12, 20. 21)- Furthermore, in regard
to location; sampling devices: or mechanisms, materials, and handling
operations, the expertise of the chemist can be most Fruitfully employed.
This area has been recognized by the American Chemical Societv
Committee on Environmental Improvement (22), which suggested the following
minimum requirements for an acceptable sampling program:
k. a proper statistical design that takes '.ntc account the goals of the
study and its certainties and uncertainties:
2. instructions for sampie collection, labeling, preservation, and
transport to the analytical facility; and
3. training of personnel in the sampling techniques and procedures,
specified.
To bolster these suggestions, the committee emphasized that all sampling
procedures should be written into detailed protocols similar to the need for
analytical protocols in a quality assurance program. Furthermore, the committee
suggesied documentation af decisions as to what methods, techniques,
procedures, or materials are to be included in the protocols fcr sainplirg
particular matrices for partxt-lar cor-sctuents-
These suggestions for sampling. protocols are really expressions of
professional accountability analogous to those expected from the peer-reviewed
literature. Few chemical professionals would expect to publish a procedure for
an organic synthesis or for the analysis of an exotic chemical species without
carefully documenting the exact steps, preparations, and performance measures
for such a procedure. Yet. the environmental literature contains numerous
instances of striking phenomena "observed" in samples from one locale that
were collected by lately undocumented procedures. Many research and
monitoring efforts would materially benefit from improvement in the documen-
tation of sampling and analytical work.
tf an overall study program is viewed as a hypothesis to be rested by the
scientific method of observation (i.e.. sampling and analysis), which is followed
by interpretation and reevaluation of ihe hypothesis, the value of detailed
-------
Ip
Principles of Environmental Sampling
protocols can -be readily appreciated. This parallel development of program
purpose or hypothesis testing is depicted in Figure 3. The figure shows the
progression From mere methods, techniques, or procedures to detailed protocols
that irt turn will often need to be refined to adequately test the hypothesis or
achieve the purpose of a. program The value of experi men canon will be to
strengthen the results and conclusions of environmental sampling arid anah-sis
programs. This value is especially true for identification and control of
systematic error. The- literature of environmental chemistry also provides a
number of examples where recognition of systematic sampling problems has led
to significant advances in out understanding of environmental processes and
chemical Fates The framework of a sampling protocol provides a basis for the
application of growing sampling experience to a variety of present and Future
problems.
Elements of Environmental Sampling Protocols
There are far too many potential types and purposes of investigations in
environmental chemistry to present a generally applicable strategy or formula for
preparing sampling protocols. General guidance for combined sampling and
analysis quality assurance (QA) program planning is available for a number of
monitoring applications (23-28). However., guides that deal with protocols and
QA procedures for air. water, wastewater, seawater. and hazardous wastes
analyses are far more numerous than references that include the practical aspects
of sampling. Thus, the need is for successively refined sampling experiments in
specific sampling applications.
The historical strength of science as opposed to other fields of human
endeavor (e.g., politics and religion) is. the scientific or experimental method
bolstered by peer review. Experimentation and experimental design skills are not
widely taught in science or engineering curricula. For this reason, among others,
environmental chemistry is often viewed by Qur basic chemical colleagues as a
business of comparative, rather than absolute, measurements or observations.
Much of the recent literature in environmental chemistry shows that this view
is truly not the case, and examples of nearly absolute measurements of chemical
contaminants in admittedly nonequilibrium systems are not hard to find. The
fact that experimental design skills are necessary for reliable studies suggests that
the subject should be taught as are other details of experimental or analytical
work. A good source book for basic experimental design that is particularly
useful for students is that of Wilson (29), among others (30. 31) A generalised
sampling protocol for environmental applications presented here highlights the
results of successful sampling experiments.
Table I contains an outline for a sampling protocol that would have general
application in environmental research or monitoring. The protocol begins with
the purpose of the overall program and the specific purpose or purposes of the
-------
HYPOTHESIS
OBSERVATION
INTERPRETATION
PROGRAM
PURPOSE
SAMPLE
ANALYZE
INTERPRET
FORMULATE
QUESTIONS
AND DESIGN
SAMPLING
PROTOCOL
PROCEDURES
TECHNIQUES
METHODS
^ANALYTICAL.
PROTOCOL
PROCEDURES
TECHNIQUES
METHODS
RESULTS
RE-EVALUATE HYPOTHESIS/PURPOSE
Itjjuir Kiliiliomlii/) of //nigrum putjunv unj /muikuls to the u i,nlif
-------
12
Principles of Environmental Sampung
Table I. Outline of a Generalized Sampling Protocol
.Vluin Point
(Program Purpose)
Subeicments
Anaivtes of interest
Primary and secondary chemical constituents
and criteria for representativeness
Site, depth, and frequency
Design. construcnon,.and performance
evaluation
Mechanism, materials, and methodology
Preservation, filtration, and field control
samples
Unstable species and additional sampling
variables
Preservation of sample integrity-
Locations
Sampling points
Sample collection
Sample handling
Field determinations
Sample storage and transport
sampling effort. From this point, the specifics of what the samples are to be
analyzed for and die questions "how many", "where", "when", and "how" are
addressed in order.
A seemingly simple series of tasks in an initial sampling effort would be
to first estimate the variability and mean value(s) of the anaivtes in the samples
and apply statistics to estimate the number of samples and frequency of
sampling necessary'- to achieve the acceptable confidence levels and to fulfill the
program purposes. Then, the volumes or types of sample necessary for the
specific determinations would be identified. Finally, graduate students or
laboratory technicians would be sent out to the field armed with homemade or
commercial sampling gear and having a firm resolve to "drown a few worms"
and return with the needed samples.
This final step is often the weakest link in the sampling operations.
Although a sufficiendv large number of samples can normally account for
random errors (32), serious systematic or determinate sources of error may be
involved in the use of certain sampling devices. These problems mainly affect
sampling accuracy or the relation between the analytical result (presuming that
the analysis is perfecdy accurate) and the actual composition of the environmen-
tal medium being sampled. As Lodge (33) pointed out, a clear distinction exists
between sampling accuracy and representativeness. Sampling inaccuracy can
often be minimized by sampling experiments and technological refinements.
Representativeness is the correspondence between the analytical result and the
actual environmental quality or the condition experienced by a contaminant
receptor. Regardless of the purpose of the study or investigation, laboratory-
oriented QA measures can only account for errors that occur after sample
collection. The planning decisions, identification of sites, and procedures for
sample collection must also be subject to QA and peer review.
-------
3ARC£L0Na CKmir.v of ;he Sampling Prtxtss
'J
Certainly. scientists must exercise carc in the sample collection step of a
sampling protocol and ensure that the sampiing point and mechanism are not
subject to serious systematic error. Accepting manufacturers' claims of :he
performance of representative 1 sampiing deuces may be particularly impru-
dent .vuhout careful experimentation. Similar precautions may apply to che
other elements in a sampling protocol.
Program Purpose
The goals or purposes of an environmental program or study are implicit in the
task of sampling protocol preparation. The cost, time frame, and overall goals
of a particular srudv may override efforts to carefully plan and conduct an
adequate sampling operation. However, the long-term consequences of the
quick answer should be considered. Environmental scientists can also argue that
the cost of planning and conducting a limited sampling experiment can save
considerable expense as well as "face" in studies that deal with trace chemical
constituents of health concern. Ail individuals involved in the effort should
understand the overall and immediate purposes of the study and recognize that
the data must be well-documented as to quality.
Anafytes of'lnterest
The selection of chemical constituent(s) of interest to a particular study may be
categorized as primary and secondary. The primary chemical constituents may
represent known species that have been identified in the sampling matrix or are
required to be determined by regulation. Examples of these species include
those required by drinking water regulations, maximum contaminant level rules,
air quality regulations, or a variety of monitoring programs. Secondary chemical
constituents may include transformation products of primary chemical species,
environmental variables needed to characterise conditions or meet criteria for
representativeness, and other chemical species that may be indicators of sample
integrity. The secondary category is no less important than the primary
grouping.
Consider a study that required a survey of priority pollutant compounds
in groundwater. The concentrations of these trace compounds and species
provide little insight into the bulk chemical composition and geochemical
conditions in the environmental matrix. Many of the 129 priority polluiants
undergo substantial chemical or biochemical transformations in air, water, and
soils and result in the formation of other compounds 3nd products (34). A more
extreme example may result from restricting analytical determinations to the
parent compound (e.g., aldicarb) in a pesticide formulation for an aquatic fate
or transport study only to learn later that its transformation products are far
more mobile, persistent, and toxic (e.g., the aldicarb sulfoxide and sulfone) (35).
Allocating large amounts of funds and human resources for sampling and
-------
Prjsciples Of Evlhw.me.vtai. 5*mpuvg
tmaivsis of natural ivaters far selected ar s\axpect=d W-.i: corr.pautd: is
unrealistic without also orsicering pH, major c3Uons in rr.any cases
couid provide the means :o check the consistency of selected samples. fry niass
or change balaT.ce jnethods and may aid in determining coruofUng reacaors or
coneutiar.s impcram for aeismem cr ^jned'.aoar. e3c:s.
Ca-elui tor^cxr.Mr. saoaiii i'sc fct gives <3 the choice af analytical
methods fen- specific decercwnaliens. This eons-deration a ny< only important
for sprirr.caoon aC ¦sample rctte-Sor. rrancfirr^piaceciiires. but alac- 10 ioid
matrix interferences fof certain rypes of samples. Ar.aJvtical methods deve3oped
and validated for drinking water or industrial wastewaters may not be applicable
to samples thai represent mixtures of these matrices. Also, the aims of research
projects frequently require the use of more specific md strsnivE types of
analytical procedures than those in "standard1' references.
The minimum sample vaiuraos and types oi sample pronation and
handling procedures, also depend or chr derail and specificity of ihe proposed
analytical program. Sample'volumes necessary fer specific analyses must be
identified with ci.-efui consideration af representativeness. For example, both
orgaTik. (IS) ar\d ^.organic: :3"' chtmical spanies ir. dirking, water may vary
sigruticandy with the volume of die sarapj? depending co nhecher sampling
is done immediately an opening the tap or after allowing the water io flow
for j "D me.
In addition to the specifics of andyte and analytics! method selection,
-recagniiir.g j-.ai physical. Tieteorc4ogical, or bydrc'ogic variables rrw>" j'so r sec
to be ttitascjed -or ds.3erm.aed s irapcrtant. in rriar.y esses. "it usefulness cf
chemical results is fully reitisd only when the evidence of entfirar.cnerttal
cite sites I processes ca^ be listed ta other controlling variables. This data 15
therefore essential 10 the interpretation of [he chemical results ard sacu.i) be
iocstded irisarcaliTig cnmai! olaarung.
5ampto-.g Location and Frequency
Th-e sampling locitiQit Lr. space and rime can have a very real elTeci or the
qvalur and css/alriess af data in e^virartitten&l cberr-isirv Site seiec-ioc o(
course, should be maje p: .mari.y or. the basts of in; study joals and me nature
of the environmental phenomenon or process under consideration. The
optimum mimber. spacing, and frequency of samples at a coarse scale car, best
be estimated after a preliminary sampJifsgeKperiraeiti
-------
3.ASC3LCSA Omritiv >| (he iumpiinj process
obtaining representative solid samples over a iarje geographic region. Geochem-
ists have applied methods of hierarchical analvsis Ji' variance for preliminary
geochemtcal studies to optimize the ?ampling designs of large-scale efforts
For example, optimum numbers, discrete depths, jnd locations have
been chosen in lakes and within geographic zones of state or provincial regions
in selected studies. The power of these approaches is that they allow the scientist
jnd the program consumer (i.e., government agency or official) to qualitatively
discuss the trade-offs involved in changes of such designs.
From a practical point of view, formal design plans have real potential in
hazardous waste site assessments and in cleanup evaluations where the
challenge of inhomogeneous sample macrices may be further complicated by
spatial variability and the need to know suspect constituent concentrations at the
parts-per-million (mg L_l) level or be!ow (46. -IT).
Besides the numbers of sampling locations and samples to be collected, the
frequency of sampling is the most significant cost multiplier in a sampling
operation. Certain frequencies of sampling are set by regulation. More often in
research and other investigations, relatively short-term studies are undertaken
with little concern (or the dynamics of the environment. If the purposes of the
program include optimization of detection or sensitivity of the results of
sampling operations to trends (i.e., long-term changes tn environmental quality)
or periods (i.e.. short-term changes in environmental quality), this aspect of the
sampling protocol needs very careful attention.
The distinction between long- and short-term changes is usually relative to
the time period over which measurements are made. Periodic changes may be
of the order of seconds at a sampling frequency of minutes. Trends are normally
considered changes that may occur over several periods or sampling runs. If-the
environmental value or quality of interest varies with a certain frequency, the
sampling frequency must be at least twice the frequency of that variation.
The problem of optimizing sampling frequency within time, program
purpose, and cost constraints has been addressed for many environmental media
(18, 48-53). The results of a "one-shot" or single-period sampling operation
cannot be expected to represent average conditions in most instances.
Sampling Points. Samples may be collected from bridges, banks, stationary
towers, or platforms as well as ships, airplanes, or passive personnel monitors.
The important considerations here shouid be minimal disturbance of the
samples. The design and construction of sample access points should disturb
the local environment as little as possible or the collection of biased results may
be inevitable. The literature for specific applications should be considered in the
design of new operations.
Scale problems can also arise in the definition- of the sampling point.
Collection of a surface water sample from a lake or the ocearv seemed to be the
simplest operation one could perform with a bucket. In time, limnologists and
oceanographers recognized thai surface water samples collected with buckets
-------
1c PRINCIPLES Or ENVIRONMENTAL SAMPLING
were quite different in organic or trace metal content than samples collected by
screens (54). The surface microlayer. enriched from fivefold to several hundred-
fold in lipids. Pb. Ni. Cu. and bacterial content, provides very different resuits
from water samples taken just beiow the surface of water bodies (55-57). The
selection of the sampling point must be done with an appreciauon that certain
microenvironments mav either be created or traversed by sampling gear that mav
vield results quite different from the desired point in air. water, or soil
environments. In the case of the sea-surface microlayer. the discovery of this
microenvironment proved embarrassing to previous work: however, an exciting
new area of research was opened with implications for air-sea gas exchange,
bubble and aerosol formation, and chemical fractionation investigations (58). Too
often, the slow recognition of systematic sources of error, like picking up the
"slick" from the "surface" in water sample collection, may delay significant
advances in environmental science. The value of a critically evaluated sampling
experiment cannot be underestimated in such instances. Even though the
chemical or biochemical constituents of such microenvironments may not be the
object of specific studies or investigations, they may cause matrix effects or other
analytical interferences for the chemical constituents of interest. Field control
samples (i.e., blanks, spiked samples, and colocated samples) can be extremely
helpful in evaluating these analytical effects on chemical results.
Sample Collection. Sample collection involves contact of the sample with the
sampling device and its materials of construction. Often, the sample may contact
the sampling technician or adjacent materials as well. The entire path of sample
retrieval should be scrutinized to minimize systematic sources of error that
cannot be accounted for by conventional laboratory-oriented QA and quality
control (QC). The same care required for sample storage vessels should be
extended to the selection of materials, methods, and devices for sample
collection.
The history of trace metal sampling and analysis in the environment
provides classic examples of the value of recognition and control of systematic
sampling error. Basically, the modern environment has levels of Pb in the air.
dust, and soil that may bias improperly collected samples by orders of
magnitude. The pioneering work of Patterson and co-workers (59. 60) on
environmental Pb sampling and analvsis spurred an avalanche of research that
resulted in the well-known '"decrease" in levels of trace metals in the oceans over
the past 20 years (61-64). These references show that previous attempts to
unravel the geochemistry and environmental fates of trace metals were stymied
because of biased sample collection and handling techniques. The contamina-
tion of samples of air. water, and biota bv artifact trace metals had obscured the
true environmental distributions and controlling processes. Far worse, however,
was the delayed realization that not all environmental samples in geologic time
had equivalent industrial metal levels.
Even though sample collection and handling techniques .at the "clean-
room" level may not be necessarv for routine monitoring or investigative efforts.
-------
SascSU'N-* Oi-fnii.iv ,tf" ;iic iiimoixn; Process
very direful consideration must be jiven to (he choice of sampling mechanism
and methodology for chemical contaminants in the parts-per-b.iilion ijjg '
range. References 59-o-r represent careruijy documented attempts to control
systematic error.
Here again, as in the case of recognizing the impact of the sea-surface
mtcrolayer op surface samples, the control of industrial Pb contamination has
had a great effect on the results and interpretation of environmental chemistry
research. It is sobering to contrast the modern practice For sampling seafarer for
Pb in its diligence and detail with the thousands of somewhat hapha::ardlv
collected samples of groundwater delivered under strict chain of custody to a
variety of contract laboratories under various hazardous waste monitoring
programs We will undoubtedly discover subtle aspects of waste-constituent
geochemistry in the duiire one; sample collection details are as carefuity
documented as are laboratory QA and QC procedures (651.
Poor sample collection procedures can seriously bias chemical results.
Documentation of sampling procedures in protocols can help to identify and
control such errors. The core of the sampling operation is the mechanism or
device used to collect she sample. Because w« rarely know the true value for the
concentration of a particular chemical constituent in the environment at any
single time, we must either test sampling mechanisms under controlled
laboratory conditions or intercompare them simultaneously in the fieid. In the
laboratory, the concentration of the chemical consnruent can be controlled, but
the exact environmental conditions are difficult to simulate. In the field.-
depending upon whether loss (e.g., volatile compounds) or contamination (e.g.,
aitborne Pb) is a likely source of systematic mechanism-related error, the highest
or lowest result, respectively, is normally accepted as the most reliable for certain
applications. Device-related errors cannot be accounted for by blanks, standards,
or replicate control samples. The actual performance of sampling mechanisms
must be verified in critical sampling applications. Numerous examples in the
literature serve as a guide for future efforts. Many of these are provided in Table
tl. These examples demonstrate that sampling mechanisms and materials can
contribute relatively large errors in comparison to analytical errors in
environmental chemistry results, particularly for trace level concentrations (i.e..
< 1 mg L"1) of chemical constituents.
Poor recoveries and positive bias caused by contamination are not the only
problems that may irtse from the use of certain types of samplers. The efficiency
¦•vuh which a person can control the operation of (he sampler maintain stable,
reproducible operating conditions; and recognice a malfunction all play j
significant part in the actual performance of environmental sampling equip-
ment. Unforuinatelv, few manufacturers provide detailed operation or calibration
instructions. More frequently, the principle of operation is referenced to a
literarure citation and specific precautions for safe use may be provided.
Individual investigators, with the aid of the analytical staff, should plan to
evaluate sampling performance in order to at least estimate the degree of
systematic error as' well as the routine precision that mav be expected under
-------
IS Principles o? Environmental Sampling
Table II. Selected Sampling Mechanism and Materials Evaluations
Emironmmral
Marrix
Stimnlc
Reference
.Air
Aerosol metals
00
Aerosol SO.,-- and NO,"
95
Precipitation
67-6^
Water
Volatile organic compounds
70
Groundwater
Dissolved gases and volatile
organic compounds
Ferrous iron
Miscellaneous
73
Seawater
Trace metals
Surfaca microto^er
74 - 76
77
Soil
Pore water
73. SO
Pore water organic compounds
79
field conditions. A general validation scheme for sampling devices is shown in
the box on page 19. Although this scheme may be too involved for individual
projects, large field programs would benefit from the use of such validation
activities. If sampling errors are less than those that may be contributed by the
analytical operation, reducing them further is probably not worth the effort in
most cases.
Materials selection can be a potential pitfall for both sampling operations
and achieving the purpose of an investigation. In situations where the sample
remains in contact with a potentially reactive, sorptive. or leaching material, the
opportunities for gross systematic errors exist. Sampling tubing (70. SI. S2),
gaskets (36). plated surfaces (37. 83), and a variety of other siich exposures have
resulted in serious errors in sampling. The selection of a general sampling
mechanism or a specific device made of appropriate materials for the
application should be based on the most sensitive (i.e., labile, volatile, and
reactive) chemical constituents under investigation. Once the sampling
evaluation and experiment have been conducted, documenting the procedures
and format field books to record deviations from the sample collection part of
the sampling protocol is a relatively simple task (3->).
Sample Handling, Field Determinations. Storage, and Transport
Most of the recent QA and QC guidance manuals available for the national
environmental monitoring programs provide sound guidance for planning the
procedures for sample preservation and handling. Because manv sampling sites
are exposed to rain. wind, sunlight, and temperature extremes, field transfers and
-------
vaCELJN'A Overview iif (lit iampiin^ Pnxcss
Sampling De\ice Validation Scheme Protocol
Laboratory: Operating range *'ruggeanessi. recovery- (accuracy;, and
precision
I. Generate known atmosphere, solution, etc.
1 Pump or collect multiple samples v^nd backup if possible) with
device and collect controls by reference tdirect) method.
3. Repeat step 2 holding concentration fixed and varying humidity,
depth, lift, submergence, volume, or :low rate.
4. If possible, repeat step 3 over a range of concentrations for the
expected range of variables.
5. Repeat steps L-4 with interferences present, if possible.
Field: Estimation of recovery (accuracy) and precision
1. Use reference method, if possible, to establish the background,
stable concentration.
2. Pump or collect multiple samples with device: if necessary, use
backup'samples to check breakthrough.
3. Repeat steps 1 and 2 over the range of field variables.
manipulations should be kept to an absolute minimum. Filtration procedures
should be streamlined and protected^ so as to minimise sample transfers in the
open air. Filter media selection and prctreatment steps should be planned and
documented with the reduction of bias in mind (85, 86). A number of
researchers performed filtration experiments in order to identify and control
sample contamination or loss for trace metal determinations due to filtration
apparatus or procedures (87-90). The detailed publications of Hunt and
Gardner (90-92) are particularly useful toward minimising systematic errors in
filtration steps.
The sampling staff should be aware of routine laboratory procedures for
both personnel and sample integrity protection. Gloves and eye protection
provide valuable safeguards and can minimise artifact trace metals (93) or
organic .contamination of samples (94). Determinations of unstable chemical
species or those that are difficult to preserve and store should be completed as
soon as possible. A rapid field analytical method may be preferable to a
laboratory method that requires extraordinary or complex sampling and
handling procedures.
An effort should be made to handle and preserve field control samples (i.e..
blanks, spikes, and oolocated samples) in the same manner as the environmen-
tal samples. This precaution will allow more effective identification and control
of postsample collection errors. Although not always- possible, the experience ot
running the sampling device evaluation and sampling experiment should be
useful in planning the sample handling and preservation procedures.
-------
Principles Emwcsmestal Smipunc
Once the samples have beer delivered to the laboratory. the sampling
operation extend; ro the point of sample reduction prior to Jnaivsis. The anaiyst
who a consulted or, sarr.p'.Lng f ro'-cwi cyclop it. en car. materiallv amorove cine
overaJi rciiafc-iiity -if the nesults -and concision; cf die zrcersm. A wiJirtipess to
e.rperi.Tienr (i.e.. make jnd acknowledge errors] *ill provide the fcaiance
berween statistical arid experimental concerns needed to design and execute a
cost-eiTectivc sirr.piinc eiToyc.
Acknowledgments
I am thankful ic a number of mv past and present colleagues for their insight,
help, and advice in the preparation of this chapcer. The comments of the
re^ewers are also very much appreciated. Thanks also go to Pam Beavers who
typed the manuscript and to Lynn \Veis= who dafted the figures.
References
1. .Vaticna/ Research Council Opportunities 1)7 Chemistry; National Academy Press:
Washington. DC. 1981'. p 18, pp L95-2C8.
X Taylor, J. K. Anal CJiem. 1983. 55(6J. 6OOA-6OSA.
3, King. D. E. In National Bureau of Standards Special Publication ilZ: La Fleur,
Philip D.. Ei. National Bureau cf Stindaids: Gaithcrsburg. MD. 1975: pp Ml-153.
Ysudtr. W j. 5inii»l;wl Tr::hni
-------
iA&Ci-CNA i,V«tyi*w ¦'( the iiimpiing Pikcss
10. G*. P. M. ,;.ni;iusis 1983. i/i?V .-13---0.
2!. Brands. G. rrcSLTiiui Z. .Anai. Chem. 1983. J14. ?-o-o5i.
22. American Chemical Societv Committee on Environmental improvement. .Anui Ciiem.
1980. 32114'. 22+2-2249.
13. red. &sisl 1978. --it J322J. 29696-297-!.
24 Rijgan. Z M. Technical .'issisidncc Document ;br :amriing ,mj Anaiysis of Tone Organic
Cjmpounds :n .Arnbiem .Air-. 3atteile Laboratories. Columbus. OH: prepared 'or L'.S.
Environmental Protection Agency-Environmental Monitoring and Support Labora-
tory: Research Triangle Park. NC. 1983: EPA oC0.'J-83-027.
25. Sumpling -ind Anuivsu jj Ruin: Campoeil. S. A.. Ed.: ASTM Special Technical
Publication 323: American Society for Testing and Materials: Philadelphia. PA. 1983.
26. Peden. M. H. Development of Standard Methods for the Collection unci Anuivsis of
Precipitation: Illinois State Water Survey CR 381: U.S. Environmental Protection
Agency-Environmental Monitoring and Support Laboratory: Cincinnati. OH. 1986.
27 Peden. M. E In Sumpiing and Analysis of Rain: Campbell. S. .A.. Ed.: ASTM Special
Technical Publication 323: American Societv for Testing and Materials: Philadelphia.
PA. 1983: pp 72-83.
23. Topol. L. E.: Levon, M.: Flanagan. J.; Schwall. R. j.: Jackson. .A. E. Quality Assurance
Munuul for Precipitation Measurement Systems: U.S. Environmental Protection Agencv-
Ehvironmental Monitoring and Support Laboratory: Research Triangle Park. NC.
1985.
29. Wilson. E B. .An Introduction to Scientific Research: McGraw-Hill: New York. 1951
20. 8ox. G. E. P.: Hunter. W G.; Hunter. J. S. Statistics for Experimenters: Wiley: New York.
1973.
31. Mace. A. E. Sample-Sice Determination: Reinhold: New York. 1964.
31 janardan. K. D.: Shaeffer. D. J. .AnaL Chem. 1973. 51(7''. 1024-1026.
33. Lodge. J. P. In National Bureau of Standards Special Publication 422: La Fleur.
Philip D/. Ed.: National Bureau of Standards: Gaithersburg, MD. 1976: pp 311-320.
34 KSbayashi. H.: Ritimann. B. E. Environ. 5ci. TechnoL 1982. 16. I07A-I83A.
35. Cohen. 5. Z.: Creeger. S. M.; Carsel. R. F.: Enfield. C. G. In Treatment and Disposal
of Pesticide 'Aostes: Krueger. R. F.: Seiber. J. N.: Eds.: ACS Symposium Series 259:
American Chemical Societv: Washington. DC, 1984. pp 297-325.
36. LeBei. G. L.; Williams. D. T.J. Assoc. Off. .AnaL Chem. 1983. 660), 202-203.
37. N'elT. C. H.: Schock. VI. R.; Marden. J. I. Relationship Between Water Quality and
Corrosion of Plumbing Materials in Buildings: State Water Survey, prepared for U.S.
Environmental Protection Agency-Water Engineering Research Laboratory: Cincin-
nati. OH. 1987: EPA 600/S2-87-036.
38. Hughes. J. P.: Lettenmaier. D. P Water Resour. Res. 1981. 17(6). 1641-1650.
39. Flatman. G. I: Yfantis. A. A. Environ. MoniL .Assess. 1984. 4. 335-349.
-0. Gilbert. R. A.. Simpson. J. C. Environ. .Momt. Assess. 1985. 5. 113-135.
41. Brown. K. W; Mullins. J. W: Richitt. E. P.: Flatman. G. T.: Black. S. C.: Simon. S. J.
Environ. Menu. .Assess. 1985. i. 137-154.
-1 Grant. C. L.: Pclton. P. A. Role of Homogeneity in Powder Sampling-. Remedy, W. R.:
Woodruff. J. F.. Eds.: ASTM Special Technical Publication 540: American Society for
Testing :ind Materials: Philadelphia. PA. 1972: pp 16-29.
-3. Tourtelot. H. A.: Miesch. .V T. In Trace Elemcni Geochemistry in Health and Diseases:
Freeman. J.. Ed.: Geological Society of America Special Paper 155: Geological Society
of America: Denver. CO. 1975: pp i07-118.
44. Connor, j J.: Myers. A. T. Role oj Homogeneity in Powder Sampling; Remedy, W R.:
WoodrulT. J. F.. Eds.: .ASTM Special Technical Publication 540: American Society for
Testing and Materials: Philadelphia. PA. 1972: pp 32-36.
45. Garrett, R. G.. Goss, G. E. fn Proceedings of the rih International Ceochemicul
Evpioruuon Svmposiuin; W'atterson. j. R.: Theobald. P. K.. Eds.: .Association of
Exploration Geochemistry: Toronto. Ontario. Canada. !973: pp 371-383.
-------
PSISCP'JI- . r tNWASssiEST.Vi. SaMP'JNJ
-p. Perkct. C. L.. 3jr;oc;:. L. R. in Muinoraior\ Anaivsis i'/ Sin, :,v Lead: rc:ras. I K..
Lac-.'. ''A' j.: Conwav. R. A.. Eds.. A.STM Special Technical p-.-.bucation ?ftc: Amcr:c.:r.
Soc:etv for Testing and Materials: Philadelphia. ?A. '.9tio: pp iZI-lis
47. Loehr. R. C. Martin, j. Newhauscr. E. r. Ibic.. pp 2S5-29'
Madsen. 3. C Ainu's, Enurim. 1982. 16:10;. 2515-2319
Shaw. R. \V,; Smith. M. V.; Pour. R. J.J. Air Poiiut. Conrroi .-jsoc. 19tW. J-f'S .
50. Neison.J. D.: Ward. R. C. Ground Water 1981. 6. d 17—625.
51. Lottis. j. C.. Ward. R. C. !.\.iitcr Rcsour 3ui!. 1980. 16. 501-507
52. Sanders. T. G.. Adrian. D. D Water Resour Res. 197S. H. 5c"P-576.
53. Lcuenmaier. D. P. Water Resuur. Res. 1976. 12:5:. 1037-i046.
5-i. Garrett. W. D. in Proceeding of the S\mpostum on Organic Matter in S'aturai Waters:
Hood. D. \V. Ed.: institute of Marine Science Publication No. i: Lniversirv of Alaska.
AK. 1968.
55. Dtice. R. .A.: Quinn. J. G.: Olney. C. E.: Piotro«icz. S. J: Ray. 8. J.: U'ake. T. i. Science
'Washington. DC) 1972. 176. 161-163.
56. Zsolnay. A. Mar. Chem. 1977. 5. 465-475.
57. Morris, R. |:; Culkin F, Naiurf (Lt'nticn' 1974. 250(5463;. 640-642
58. Macinivre. F. In The Sea: Goldberg. E. D.. Ed.: Wiley-interscience: New York. 1974;
Vol. 5. Marine Chemistry: pp 245-299.
59. Patterson. C. C.: Settle. D. M. in National Bureau of Standards Special Publication
422: La Fleur. Philip D., Ed.: National Bureau of Standards: Gaithcrsburg. MD. 1976:
pp 321-351.
60. Patterson. C. C.: Settle. D. M.: Schaule. B. K.; Burnett. M. in Marino Pollutant Transfer:
Windom. H. L.: Duce. R. A_. Eds.: Lexington Books: Lexington. KT. 1976.
ol. Schaule. B. K.: Patterson. C. C. Earth Planet. Sci. Lett. 1980. 47. 176-198.
62. Burnett. M.: Patterson. C. C. In Proceedings of an International Experts Discussion on
Lead: Occurrence. Fate and Pollution in liie Ennronmnic Branica. M Ec.: Pergamon:
Oxford. England. 1978.
63. Bruland, K. W.: Franks. R. P.: Knauer. G. A.: Martin. J. H. Ann/. C/nm. Acta 1979.
105.233-245. •
64. W'indom. H. L.: Smith. R. G. Mar. Chem. 1979. 7. 157-163.
65. Quality Control in Remedial Site Investigations: Perket. C.. Ed.. ASTM Special Technical
Publication 925: American Society (or Testing and Materials: Philadelphia. PA. 1986.
66. Milford.J. B : Davidson. C. I.J. Air Poiiut. Conrroi Assoc. 1985. 35t 12'. 1249-1260.'
67. Slanina.J.: Van Raaphorst.J. G.: Zijp. W. L.: Vermuelen. A J.: Rolt. C. A. int. J. Environ.
Anal. Chem. 1979. 6. 67-SI.
68. Chan. W. H.: Lusis. M. A.: Stevens. R. D. S.. Vet. R. J. Water. Air. Soil Poiiut. 1983.
23. 1-13.
69. Schroder. L. j.; Linthurst. R. A.; Elison.J. E.; Voizo. S. F. Witter. Air. Soil Poiiut. 1985.
24. 177-187.
70. Ho.j. S -V.J.Am. Water Ubrla .Assoc. 1983. 12. 5S3-586.
71. Barcelona. M. J.: Helfrich. J. A.: Garske. E. E.: Gibb. J. P. Ground Winer Mom;. So.
1984. 4(2). 32-41.
72. Stoltzenburg, T. R.; Nichols. D. G. Preliminary Results on Chemical Changes in GrounJ
Wilier Samples Due to iumplinq Dtviccs: prepared for Electric Power Research institute:
Paio .Alto. CA. 1985: EA-411sr
73. Gibb. J. P.; Scnuller. R. M.: Griffin. R. A. Procedures for the Collection ofRepresentative
V^iittT (Jualin Data from Monitoring Wis: Coop. Ground Water Report 7. Illinois State
Water and Geological Survevs: Champaign. 1L. 1981.
"4. Scar. D A.: Berberian. G. A. in Analytical Methods in ckeanogranm: Gibb. 7 R.
Jr.. Ed.: Advances in Chemistrv 147. American Chemical Societv Washington. DC.
197?; Chapter 2. pp 9-15
75 Brewers, j. M. : W'mdom. H L. Mar Oieni. 19S2. II. T'.-tfo.
-------
L. na . .ViTvit-.»' ;if*:iit' P'reea
ipencer. M. j.: 2cn.r. ?. R.. i'lotrowic:. S. Mar. diem. 1982. .'/. -0.'— -!0
'•'an Vice:. iZ. Williams. ?. M. Limimi. Oavmnj*-. i980. 25. ~c--~7J.
3ro»'n. a. A', in Land Tivafminf. A Hazardous Waste Miiiitf^naN Alternative: 2eehr.
R. C.. Maiina. j. ?.. jr.. Eds.: Cantor :or Research in Water Resources. L'mversitv -:.i
Texas: Austin. TX. pp 171-135.
3arcee. G. L.: 3rown. j\. W 'AutiT. Air. ioii Poilut. 1986. 29. 321-331.
¦iO. Evbere:t. L. G.. Mc.Million. L. G. Ground Water Monit. 1985. 5(3'. 51-o0.
."fL. Sarcclona. M. j.; Heilrich. J. A.: Garske. E. E. Amii. Chem. 1985. 57¦ 2>. 460--O-
..Errata. Anai Chan. 1985. 5713. 2752.';
52. 3arceiona. M. J.: Giob. J. P.: Miller. 3. A. A Guide :o tiie Selection of Materials or
MoiiiIitih" Ut'il LJnsirucnon xind Ground "Aiucr iamvim^: Sute Water Survev
Publication 327: L'.S. Environmental Protection Agencv: Cincinnati. OH. 1983: EPA
600 S2-34-024.
33. Andersen. rC. E.: Nielsen. G.-D.: Flyv'nolm. M.: Fregert. S.: Gruvberge. B. Omnci
DirmuniLS 1983. 9. 140-143.
34. Sarceiona. M. j.: Giob. J. P.: Heifrich. J. A.: Garske. E. E. Practical Guide Jor Ground-
Water sampling: jute Water Survev Publication 374: L'.S. Environmental Protection
Agency: Cincinnati. OH. 1985: EPA oOO. S2-35-104.
>5. Quinn. J. G.. Meyers. P. .V LimnoL Oceanogr. 1971. !6il). 129-131.
36. Maienthai. E. j. Presented at the 157th National Meeting oi :he American Chemical
society. Minneapolis. MN. April 1969.
37. Wagemann. R.. Brunskill. G.J. Int. J. Emiron. Anul. Chem. 1975. 4. 75-84.
53. Truitt. R. E.: *Aeber. J. H. Anal. CI ion. 1979. 51(12!. 2057-2059.
39. Laxen. D. P. H.: Chandler. 1. M. Anni. uit7n. 1982. 5MS). 1350-1355.
90. Gardner. M. J.: Hunt. D. T. E. Anim-si 'Londoni 1981. 106. —71 - — 74.
91. Hunt. D. T. E. Filtration oiWdter Samples for Trace Metal Determinations: Technical
Report TR '.O-: Water Research Centre. Medmennam laboratory: Marlow. 3ucks.
Enaiand. 1970..
92. Gardner. M. J. Adsorption ol Trace Metais from Solution During •Filtration of Water
Samples: Technical Report TR 172: Water Research Centre. Medmenham Laboratory:
Mariow. Sucks. England. 1982.
93. 3erman. E. In National Bureau of Standards Special Publication 422: La Fleur.
Philip D.. Ed.: National Bureau ot Standards: Gaithersburg. MD. 1976: pp 715-719.
J4 Hamilton. P. B. Ann I. Chem. 1975. 47(9). 1713-1720.
P5. Milford. J. 3.: Davidson. C. 1. J. Air Poiluv Control Assoc. 37(2). 125-134. in press.
Receimd for rc\iew February 11. 1987. Accepted April 24. 1987.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 11
Field Screening and Soil Gas Measurement Techniques for VOCs
Thomas Spittler
Region 1 Laboratory, USEPA
Lexington, Massachusetts
January 12-14, 1993
Las Vegas, Nevada
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 12
Speaker to be announced
January 12-14, 1993
Las Vegas, Nevada
-------
National Symposium on Measuring and
interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 13
The Persistence of Several Volatile Organic Compounds in a Low
Organic Carbon Calcareous Soil From Southern Nevada
Spencer M. Steinberg, Department of Chemistry, University of
Nevada, Las Vegas; and David K. Kreamer, Department of
Geoscience and the Water Resource Management Program,
University of Nevada, Las Vegas
January 12-14, 1993
Las Vegas, Nevada
-------
The persistence of several volatile organic compounds in a Low Organic Carbon
Calcareous Soil from Southern Nevada
Spencer M. Steinberg. Department of Chemistry, University of Nevada, Las Vegas
David K. Kreamer, Department of Geoscience and the Water Resources Management
Program, University of Nevada, Las Vegas
The partitioning of volatile organic compounds (VOCs) between vapor and soil is
characterized by at least two types of interactions. Most VOC sorption and
desorption is kinetically rapid and can be described by a sorption isotherm. However,
recent field measurements have demonstrated that low molecular weight compounds
may persist in soils for decades after contamination [1,2,3,4]. This persistent sorptive
interaction has been attributed to rate limiting diffusion of volatile compounds from
soil micropores.
We have conducted laboratory studies aimed at characterizing both extremes in the
soil-vapor partitioning of VOCs for a calcareous soil that, is typical of Southern
Nevada. The compounds which were investigated include: 1,1,1- trichloroethane
(TCA), 1,2,2- trichloroethene (PCE), tetrachloroethene (PGE) and 1,1,2,2-
tetrachloroethane (TeCA), pentane, benzene, toluene, ethylbenzene, acetone,
acetonitrile and diethylether. The soil that was studied is a poorly sorted alluvial
deposit consisting chiefly of calcite, quartz and aragonite. Organic carbon contents
generally ranged from 0.1 to 0.5%. We have investigated two size fractions (medium
to fine sand and the silt and clay) as well as the composite soil.
Sorption isotherms of various volatile chemicals on the kinetically rapid sorption sites
have been determined using inverse gas chromatography (IGC). IGC involves using
the soil as a GC stationary phase and examining the elution behavior of the compound
[5,6]. Using this method we have characterized the initial or.rapid interactions with
the soil under dry conditions and in the presence of water (adding water to the GC
carrier).
IGC analysis involves interpreting GC retention times and peak shapes. For example,
in the case of a Langmuir or Freundlich isotherm when the sorption isotherm is
concave down, the front of the peak will be self sharpening and the rear of the peak
will form a tail or diffuse edge. If the sorption isotherms are linear, peaks will be
symmetrical and retention times will be independent of the mass injected [7].
We have examined the soil-vapor partitioning behavior of the VOCs listed above,
using this method, and have found that under dry conditions (no water added to
carrier) the chromatographic behaviors of chlorinated and non-chlorinated alkanes and
-------
alkenes and aromatic hydrocarbons are complex. The chromatographic peaks have
near vertical fronts and diffuse trailing edges, which would be predicted for non-linear
sorption isotherms, however, the shapes of the elution profile can not be completely
explained by the non-linear isotherms. Detailed analysis indicates that the kinetics of
desorption (and possible sorption) are slow enough to contribute the formation of the
diffuse edges of these chromatographic peaks.
A small addition of water to the carrier gas results in a substantial simplification of
the 1GC. Not only does the extent of sorption decrease (by orders of magnitude) but
peaks become symmetrical and retention on the soil column becomes independent of
concentration. These effects indicate that in the presence of water the sorption
isotherms are linear. The linear sorption coefficient (K (ml/g)) is a function of the
relative humidity (moisture content of the soil) and decreases with increasing
moisture. Evidently, water in the carrier gas competes with the VOC for sorption
sites on the surface of the soil.
The sorption of acetone, diethylether and acetonitrile, all of which are hydrogen
bonding compounds, were also investigated in the presence of water using this
method. The chromatographic peaks for these compounds had near vertical fronts
and long tails (diffuse edge) that and indicated that sorption isotherms were non-linear
and that desorption rates were also slow. We speculate that sorption of these
hydrogen bonding compounds may involve partitioning into soil water, where
diffusion may be slow.
We have investigated the sorption of the volatile aikanes, aikenes and aromatics,
which are listed above, onto the clay and silt fraction of this calcareous soil between
30-50°C at 52% relative humidity. Under these humidity conditions the soil
gravimetric moisture content remained constant (1.65% (w/w)). The enthalpies of
sorption were measured for several compounds and were generally close to the
enthalpies of condensation, which indicates that (as expected) sorption decreases with
increasing temperature when moisture is held constant. These results indicate that this
rapid sorption-desorption interaction decreases with increasing soil temperature and
increasing soil water content. For non-hydrogen bonding compounds at > 1.5% RH,
sorption is apparently reversible and rapid and sorption onto the rapid binding sites
can be described using a linear soiption isotherm.
As was indicated above, a second type of interaction (sorption) involves kinetically
slow processes which take on the order of days to years to reach equilibrium. This
type of sorption results in the persistence of VOCs in soil for long times after the
contamination event (leak or fumigation) and has led to persistent contamination of
aquifers in some areas. This slow interaction likely involves diffusion controlled
transfer of the VOCs from sorption sites within soil particles. The effects of
temperature and moisture on sorption at these two types of sites is very different.
-------
When VOCs are released to soil in large quantities by leaking storage tanks, spills or
fumigation, a kinetically slow.or recalcitrant VOC fraction may be formed in the soil.
For example, Steinberg et al., 1987 [1] demonstrated that ethylene bromide could be
retained in fumigated soil for decades following the last treatment.
We have studied this same phenomenon with TCA, benzene, toluene and ethylbenzene
in the present soil [2]. This soil was inoculated (2-50 /xl of VOC per gram of soil)
with a single compound. Samples were incubated (in sealed tubes) for 6-85 hours at 5
- 45°C. The VOC was then allowed to evaporate at an ambient temperature of
25 ±2°C in a laboratory fume hood for 24 hours. Extraction of the soil with methanol
[8] at 65-70°C demonstrated that a small fraction of the VOC remained persistently
sorbed by the soil. The persistent fraction increased in concentration with increasing
incubation time. This process could be modeled using first order kinetics. Sorption to
the recalcitrant sites increased with temperature and amount of inoculant. This
temperature effect is the opposite of the effect of temperature on the rapid interaction
that was described above, and has been interpreted to indicate that access to these
recalcitrant sites is controlled a highly hindered and temperature sensitive diffusion
through soil micropores.
The effect of moisture on this interaction was also investigated. While the formation
of the firmly bound VOC did decrease with increasing moisture, the effect was
smaller than that observed with the rapid interaction described above.
We have tested the effectiveness of several analytical methods including dynamic head
space concentration (DHC), a vigorous water homogenization procedure, and a hot
methanol extraction [8] for analyzing this firmly bound fraction. Our results indicate
that the hot methanol method was the most effective at extracting firmly bound
VOCs, followed by the water homogenization and DHC. The DHC procedure
method may underestimate volatile compounds in soil by 60-95 %. The water
homogenization procedure was tested against DHC as a field method at a
contaminated site in Southern California. Results from the two methods were highly
correlated, however, the DHC method gave consistently lower results, which
indicated that present analytical methods are not effective in analyzing this fraction.
In conclusion, at low concentrations VOC sorption is reversible and can be described
using simple equilibrium sorption isotherms. Upon prolonged exposure or during
large releases of VOCs, sorptive interactions are complicated by formation of
recalcitrant fraction, where sorption and desorption are diffusion controlled.
-------
References:
1. Steinberg, S.M.; Pignatello, J.J.; Sawhney, B. L. Environ. Sci.
Technol. 1987, 21, 1201-1208.
2. Steinberg, S. M. Chemosphere 1992, 24(9), 1301-1315.
3. Pignatello, J. J. In Reactions and Movement of Organic Chemicals in
Soil. SSSA Special Publication 22 1989, 45-80.
4. Paviostathis, S.G.; Jagial, K. Environmental Sci. Technol. 1987, 25,
274-279.
5. Goss, K-U, Environ. Sci. Technol. 1992, 26, 2287-2294.
6. Okamura, J. P.; Sawyer, D.T. Anal. Chem. 1974, 45(1), 80-84.
7. Paryjczak, T. "Gas Chromatography in Adsorption and Catalysis", Ellis
Horwood Limited, 1986, Chapter 4, 84-103.
8. Sahwney, B. L., Pignatello, J.J.; Steinberg, S. M. J. Environ. Oual.
1987, 17, 149-152.
-------
Key References:
Brusseau, M. L.; Rao, P.S.C. CRC Critical Reviews in Environmental Control 1989.
19, 22-99.
Goss, K-U, Environ. Sci. Technol. 1992, 26, 2287-2294.
Hewlett, A. D.; Mlyares; Leggett, D.C.; Jenkins, T.F. Environ. Sci. Technol. 1992,
26, 1932-1938.
Okamura, J. P.; Sawyer, D.T. Anal. Chem. 1974, 45(1), 80-84.
Paryjczak, T. "Gas Chromatography in Adsorption and Catalysis", Ellis Horwood
Limited, 1986, Chapter 4, 84-103.
Paviostathis, S.G.; Jagial, K. Environ. Sci. Technol. 1987, 25, 274-279.
Peterson, M. S.; Lion, L. W., Shoemaker, C. A. Environ. Sci. Technol. 1988, 22,
571-578.
Pignatello, J. J. In Reactions and Movement of Organic Chemicals in Soil. SSSA
Special Publication 22 1989, 45-80.
Sahwney, B. L., Pignatello, J.J.; Steinberg, S. M. J. Environ. Oual. 1987, 17, 149-
152.
Steinberg, S.M.; Pignatello, J.J.; Sawhney, B. L. Environ. Sci. Technol. 1987, 21,
1201-1208.
Steinberg, S. M. Chemosphere 1992, 24(9), 1301-1315.
Yaron, B.; Sutherland, P.; Galin, T.; Acher, A.J. J. Contaminant Hydrology 1989,
4, 347-358.
-------
David K. Kreamer
David K. Kreamer is presently the Director of the Water Resources
Management Graduate Program and an Associate Professor of Geoscience at the
University of Nevada, Las Vegas. He holds a B.S. in Microbiology and an M.S. and
Ph.D, in Hydrology from the University of Arizona. Dr. Kreamer's present teaching,
research and service involves the fate and transport of contaminants in the
environment. He is particularly interested in non-aqueous phase liquids, vadose zone
hydrology, radioactive waste disposal, groundwater hydrology, landfills, and
monitoring well design.
Spencer M. Steinberg
Spencer M. Steinberg is presently an Assistant Professor of Chemistry at the
University of Nevada, Las Vegas (UNLV). He received a Ph.D. in Marine Chemistry
from the Scripps Institution of Oceanography and a B.S. in Chemistry form the
University of California, San Diego. He currently serves as the Chemistry
Department's Graduate Coordinator and is advising graduate students in UNLVs
Environmental Analytical Chemistry and Water Resources Management programs.
Dr. Steinberg's research interests include analytical methods development, and the
fate, transport and reactions of organic and inorganic substances in water, soil and
sediment.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 14
Standard Model for Volatilization of Chemicals from Soil at Superfund
Sites
Janine Dinan, U.S. EPA, Office of Emergency and Remedial
Response, Washington, DC; and Craig Mann, Environmental Quality
Management, Inc., Durham, North Carolina
January 12-14, 1993
Las Vegas, Nevada
-------
Standard Model for Volatilization of Chemicals
from Soil at Superfund Sites'
Janine Dinan
U.S. EPA, Office of Emergency and Remedial Response
Washington, DC
Craig Mann
Environmental Quality Management, Inc.
Durham, North Carolina
Introduction
The Exposure Assessment chapter of the Risk Assessment
Guidance for Superfund, Human Health Evaluation Manual.
Part A (U.S. EPA, 1989) presents a set of equations where
site-specific data are used to estimate exposure to
contaminants on or migrating from a hazardous waste site.
Subsequently, the exposure estimates are combined with
chemical-specific toxicity criteria to determine the site
risk. The Risk Assessment Guidance for Superfund. Human
Health Evaluation Manual: Part B (U.S. EPA, Interim, 1991)
builds on the framework established in the Part A guidance
to derive risk-based Preliminary Remediation Goals (PRGs)
for Superfund sites. The goals are calculated by replacing
site-specific data in the Part A equations with default
values for the exposure and risk parameters. The resulting
PRGs are initial cleanup goals which are medium and land-use
specific, and correspond to an acceptable risk level of 10"6
for carcinogens or a Hazard Quotient of 1 for non-
carcinogens. The PRGs are intended to be used at Scoping,
and modified by applying site-specific data collected during
the Remedial Investigation/Feasibility Study (RI/FS).
To calculate PRGs for soil, Part B considers two major
pathways: 1) direct ingestion of contaminated soil; and,
2) inhalation of volatiles and fugitive dusts. The equation
for inhalation of volatiles consists of three parts: 1) a
volatilization model; 2) a dispersion model; and, 3) a human
exposure model. The volatilization component is based on
the Hwang and Falco Model (U.S. EPA, 1986) developed by
EPA's Office of Research and Development for estimating PCB
emissions. The model has several, limiting assumptions
including: no other vectors for loss of contaminants other
than vapor phase diffusion; a thin layer of uncontaminated
soil at the soil-air interface; and, under equilibrium
conditions contaminants are partitioned only between the
aqueous phase, interstitial vapors and the amount adsorbed
to soil. Most importantly, the model assumes dry soil to
-------
calculate the effective diffusion coefficient; thereby
maximizing diffusion.
Superfund management wants PRGs that are realistic and
useful to Remedial Project Managers and engineers in the
field. Thus, this study was designed to assess the relative
accuracy of the volatilization model by comparing the
modeled results to measured emission flux data.
Technical Approach
The following tasks comprise the technical approach of
the limited validation study performed by Environmental
Quality Management, Inc.:
o Review the theoretical basis and development of
the volatilization equation in Part B in order to
document key variables, assumptions and boundary
conditions of the model.
o Determine the availability of acceptable emission
flux data from experimental and field scale
studies of volatile emissions from soils.
o Screen the available emission flux studies to
determine the appropriateness of the study for
comparison to the model.
o Perform statistical analysis to determine the
correlation between the measured and modeled
results.
Results
The most significant soil parameter affecting the
steady-state flux through soil was found to be the air-
filled porosity of the soil. The water content of the soil
and its bulk density determine the air-filled porosity.
Thus, the assumption of dry soils in the Hwang and Falco
Model can lead to overestimates in emissions. It is
important to account for the effect of soil moisture on
effective diffusivity by including the Millington and Quirk
expression as suggested by Farmer, et al., (1980).
Other factors shown to have a great effect on emission
rates include: the initial concentration of contaminant;
the time from initial sampling; and, the soil-water
partition coefficient (Kd) .
-------
Conclusion
As applied in Part B, the Hwang and Falco Model employs
the conservative assumption of dry soil when calculating
effective diffusivity in soil. The results of this limited
validation study indicate that this assumption may
overpredict emissions by about one order of magnitude. On
the other hand, use of the Millington and Quirk expression
of effective diffusivity lead to no statistically
significant difference between modeled and measured emission
rates (under certain controlled conditions). It is unclear
how well the model would predict emissions under field
conditions, since the boundary conditions and environmental
factors such as wind speed, depth of contamination and water
movement cannot be accounted for as they are under
laboratory or pilot-scale conditions.
-------
References
Farmer,. W.J., M.S. Yang, J. Letey, and W.F. Spencer. 1980.
Land Disposal of Hexachlorobenzene Wastes. U.S. EPA,
Office of Research and Development. EPA-600/2-80/119.
Hwang, S.T., and J.W. Falco. 1986. Estimation of
Multimedia Exposure Related to Hazardous Waste
Facilities. Cohen, Y.(ed). Plenum Publishing Corp.
U.S. EPA. 1991. Risk Assessment Guidance for Superfund.
Human Health Evaluation Manual: Part B. Interim.
EPA/54 0/R-92/003.
U.S. EPA. 1989. Risk Assessment Guidance for Superfund.
Human Health Evaluation Manual: Part A. Interim Final.
EPA/540/1-89/002.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 15
VOC Contamination in Ground Water: Sources of Variability and
Comparison of Soil, Well and Hydropunch Results
Michael J. Barcelona, Institute for Water Sciences, Western Michigan
University-Kalamazoo; H. Allen Wehrmann, Illinois Water Survey,
Champaign; Jane Denne, US EPA-EMSL, Las Vegas, Nevada; and
Dannette Shaw, IWS-WMU, Kalamazoo, Michigan
January 12-14, 1993
Las Vegas, Nevada
-------
VOC CONTAMINATION IN GROUND WATER: SOURCES OF VARIABILITY
AND COMPARISON OF SOIL, WELL AND HYDRO PUNCHR RESULTS
Michael J. Barcelona, Institute for Water Sciences, Western Michigan University,
Kalamazoo, MI 49008
H. Allen Wehrmann, Illinois Water Survey, Champaign, IL
Jane Denne, USEPA-EMSL, Las Vegas, NV
Dannette Shaw, IWS-WMU, Kalamazoo, MI
Major concerns in the evaluation of contaminant distributions in the subsurface include
the spatial and temporal comparability of VOC concentrations determined in water and
aquifer solid samples. Spatial trend analysis and averaging techniques are quite sensitive
to the uncertainty in chemical concentrations determined on discrete samples (1,2).
Temporal trend analyses share this sensitivity and also suffer from the use of different
sampling: techniques/personnel, frequencies and analytical laboratories over the course
of an investigation (3). Clearly, we should strive to reduce variability from human and
methodologic sources and to identify processes which control contaminant distributions,
transformations and fates.
In order to reliably estimate actual spatial or temporal concentrations trends rather than
systematic errors or biases in sampling and analysis it is necessary to conduct controlled
field experiments. Temporal trend monitoring of ground-water quality and geochemical
constitutents can be done effectively without extraordinary efforts to minimize sampling
and analytical errors. (7) The use of dedicated bladder pumps, low flow rate purging,
and uniform water sample handling and analysis techniques reduces these sources of error
to less than 10% of observed variability. The application of this proven ground-water
sampling protocol to VOC contamination of a large (> 10/mi2) National Priority List site
in a shallow sand and gravel aquifer in Rockford, IL, has yielded similarly encouraging
results.
In this study, a network of more than 40 wells was constructed in an intensive study area
of — 4 mi2 to supplement several synoptic surveys of residential wells and the overall
NPL site investigation. The major objectives of the study were to evaluate levels of
spatial and temporal variabilty of VOC's in the ground water, and to intercalibrate
selected water and aquifer solid sampling/preservation methods for site characterization.
Quarterly sampling of the study wells for 21 months and several sampling intercalibration
experiments were conducted focusing on five principal VOC contaminants (i.e. 1,1-DCE,
1,1 dichloroethylene; 1,1-DCA, 1,1 dichloroethane; l,2,c-DCE, l,2cisdichloroethylene;
1,1,1-TCA, 1,1,1 trichloroethane; and TCE, trichloroethylene).
The outcomes of the work have significant implications on ground water and solid
sampling activities in support of site characterization and assessment efforts.
-------
Low flow-rate well purging with dedicated bladder pumps, while monitoring
conductance and dissolved oxygen levels as a function of volume pumped,
provides reproducible ground-water samples, [i.e. Sampling and analytical (static
headspace/GC with ECD and PID detection) errors can be controlled to within
±15% of overall variability.]
Temporal variability in VOC concentrations in water samples from wells with
short (i.e. < 2m) screened intervals was less than the levels of spatial variability
between wells located within 100m of each other within the dissolved contaminant
plume.
VOC concentrations in HydropunchR water samples taken in the vicinity of
existing monitoring well screens were statistically identical to those in pumped
samples; though some vertically discrete HydropunchR samples showed significant
vertical concentration differences on a scale of 2 to 3 meters.
Field preservation of aquifer solid samples from split-spoon samples with
methanol in static headspace analysis vials consistently yielded higher
concentrations of target compounds than did bulk solid samples which were
subsampled in the laboratory. 90 to 98% of total VOC's per unit aquifer volume,
reside in the solids, the remainder consisted of dissolved VOC's. (This holds for
properly preserved samples.)
Given the high degree of serial correlation between water samples taken from wells over
time leading to reduction in effective sample sizes, it appears that sampling intensity
should be focused on providing more spatially discrete aquifer solid data than repetitive
samples from wells. It would be particularly beneficial to pursue additional research in
this area since it may lead to better resolution of sources of VOC's and improved
methods for evaluating efficiencies of remedial actions.
Disclaimer
Although the information in this document has been funded in part by the U.S.
Environmental Protection Agency under Cooperative Agreement No. CR-815681 with
the Illinois State Water Survey, it has not been subject to Agency review and, therefore,
does not necessarily reflect the views of the Agency, and no official endorsement should
be inferred. Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
-------
References
1. Black, S.C. (1988) Defining Control Sites and Blank Sample Needs, Chapter 7, p.
107-117, in Principles of Environmental Sampling, L.H. Keith (ed.). ACS
Professional Reference Book, American Chemical Society, Washington, D.C.,
458pp.
2. Barth, D.S. and B.J. Mason (1984) Soil Sampling Quality Assurance and the
Importance of an Exploratory Study, Chapter 10, p. 97-98, in Environmental
Sampling for Hazardous Wastes. Schweitzer, G.E. and J.A. Santolucito ACS
Symposium Series #267 American Chemical Society, Washington, D.C.
3. Stoline, M.R., R.N. Passero and M.J. Barcelona (1993) Statistical Trends in Ground
Water Monitoring Data at a Landfill Superfund Site (Accepted for publication in
Environmental Monitoring and Assessment 9/92).
4. Barcelona, M.J., D.P. Lettenmaier and M.R. Schock (1989) Network Design
Factors for Assessing Temporal Variability in Ground Water Quality. Environ.
Monit. Assessment 12, 149-179.
5. Barcelona, M.J., H.A. Wehrmann and M.D. Varljen (1993) Reproducible Well
Purging Procedures and VOC Stabilization Criteria for Ground-Water Sampling
(manuscript in preparation for submission to Ground Water).
021b:vocconta.ppr
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 16
Statistical Simulation and Three-Dimerisional Visualization For
Analysis and Interpretation of Soil VOC Datasets
Toby J, Mitchell, Olivia R. West, and Robert L.
Siegrist
Oak Ridge National Library
January 12-14, 1993
Las Vegas, Nevada
-------
STATISTICAL SIMULATION AND THREE-DIMENSIONAL VISUALIZATION
FOR ANALYSIS AND INTERPRETATION OF SOIL VOC DATASETS
Toby J. Mitchell, Olivia R. West, Robert L. Siegrist
Oak Ridge National Laboratory 1
EXTENDED ABSTRACT
INTRODUCTION
The characterization of a VOC-contaminated soil region typically consists of two parts: (1)
measurement of soil VOC concentrations at discrete points within a three-dimensional soil
region, and (2) interpretation of the VOC distribution including the development of an
empirical spatial model for the three-dimensional soil VOC dataset. This empirical
modeling can be very simple, such as two-dimensional contouring of VOC concentrations
within fixed depth strata. Alternatively, empirical spatial modeling can involve a more
rigorous application of statistical and numerical procedures to determine a three-
dimensional interpolating or smoothing function that estimates the "true" VOC
concentration.as a function of location within three dimensional space. The appropriate
interpolating or smoothing function is chosen to be consistent with the VOC measurements
and with some prior assumptions about the smoothness of the "true" VOC concentration
function.
This presentation describes the application of various spatial modeling techniques to a
three-dimensional soil VOC data set generated through sampling and analysis of soil
underlying a land treatment facility in Ohio. Spatial modeling of this VOC data was
performed in order to: (1) predict and visualize the three-dimensional soil VOC distribution
within the sampled soil region, and (2) estimate the total mass of VOC within different
contaminated soil volumes. The results of spatial modeling were then used to help design a
full-scale remediation of the site.
DESCRIPTION OF THE VOC DATA SET
The site of interest was a land treatment unit, 121 wide by 266 ft long (-0.7 acre) that was
used for disposal of waste oils and solvents from ca. 1976 to 1983. The site is underlain
by silty clay deposits (90% of particles < 50 um; TOC ranging from 184 to 1190 mg/kg),
with ground water occurring at 12 to 14 ft below ground surface. VOCs, including
trichloroethylene (TCE), and 1,1,1-trichloroethane (TCA), were measured in the soil up to
depths of ca. 25 ft. The data set studied during this work consists of VOC concentrations
measured in 176 soil samples collected from 21 borings which were located on a grid with
nodes spaced at approximately 33 ft over the 0.7 acre site.
1 Oak Ridge National Laboratory is managed by Martin Marietta Energy Systems, Inc.
under contract DE-AC05-840R21400 with the U.S. Department of Energy. This
extended abstract was prepared for presentation at the National Symposium on Measuring
and Interpreting VOCs in Soils: State of the Art and Research Needs, Las Vegas, Nevada,
January 12-14, 1993.
-------
Within each boring, 100 g soil samples were collected at approximately 3 ft intervals to a
depth of 21 ft using a hydraulic probe. Duplicate soil samples were collected from 12
sampling locations. Soil subsamples (10 to 20 g) were analyzed for seven target
chlorinated aliphatic compounds, the major component being TCE. Analyses were made
onsite using a heated head space technique [1] and a Shimadzu 14A capillary column gas
chromatograph with an electron capture detector.
To date, spatial modeling has been performed on the sum of the soil concentrations of the
seven VOCs, expressed In ug of VOC/kg of field moist soil. Over the 176 samples, total
soil VOC concentrations ranged from 6 ug/kg to 154000 ug/kg. The data was
approximately log normally distributed with a median of 1400 ug/kg and a mean of 5700
ug/kg. Ninety percent of the sample VOC measurements were between 94 ug/kg and
20100 ug/kg. Within the 12 pairs of duplicates, the ratio of the largest to the smallest
concentration within each pair ranged from 1.1 to 8.71; the median of these ratios was 1.8.
These data indicate significant short- and long-range spatial variability in the VOC
concentrations.
The spatial modeling methods were also used to predict the VOC concentrations in a second
data set that was obtained four months after the first data set consisting of 176 soil samples
was collected. The second dataset consisted of 204 soil samples collected from at ca. 3 ft
depth intervals from ca. 42 probe locations within a rectangular area ca. 35 ft. wide and
120 ft long in the east central region of the overall site. Soil samples in this second data
set were collected and analyzed in the same manner as those in the first data seL
DESCRIPTION OF SPATIAL MODELING METHODS
Three different spatial modeling methods were applied to the first VOC data set described
above (176 points). These were:
(I) A three-dimensional commercially available interpolator based on the contouring
method of Briggs [2]. This method is a "strict interpolator", and searches for the
interpolating function by minimizing the "total curvature" over a grid that covers the
region of interest.
(II) A general thin-plate spline smoothing method developed by Wahba and Wendelberger
[3,4] and implemented in the public domain software RKPACK [5]. This method is
a "smoother" that seeks an interpolating function which compromises between the
criteria of minimizing total curvature and minimizing residual sum of squares. Such a
function is less sensitive to noise in the data and to the occurrence of random "hot" or
"cold" spots than is a strict interpolator.
(III) A Bayesian prediction method similar to 3-dimensional kriging [6]. This method is
based on viewing the VOC interpolating function as a sum of three components: (1)
a smooth, gradually varying function, intended to capture global behavior (much like
a linear or quadratic response surface), (2) a much rougher function, intended to
capture short-range variability, and (3) random "noise", which represents variability
between two samples at the same location. Each of these components is treated as a
realization of an appropriately chosen random function (random field, stochastic
process). Adjustable parameters of these functions were determined by the method of
maximum likelihood [7].
-------
COMPARISON OF SPATIAL MODELING RESULTS
Because of their range and skewness, the VOC concentration data were subjected to log
transformation prior to modeling. Justification for this transformation was provided by a
cross-validation study in which a subset of samples was omitted and the methods were
applied to predict the VOC values at the omitted locations. Two different scenarios were
considered in this cross-validation study. Scenario I included all 21 borings but all samples
at three specified depths were omitted. Scenario II included only 17 borings with all
samples from four specified borings omitted. Comparison of the predicted values to the
"true" (i.e., observed) values at the omitted locations using untransformed and transformed
data clearly supported making the log transformation.
The predictive performance of Methods I, II. and III was also evaluated using a cross-
validation study with the two scenarios that were used to evaluate log-transformation of
VOC data (Figure 1). There was little difference noted among these methods under
Scenario I; the majority of the ratios of predicted concentration to "true" concentration were
between 0.6 and 2.2. Under Scenario II, method III performed best; the majority of the
ratios of predicted concentration to "true" concentration were between 0.9 and 4.0 for this
method.
100
10
U
o
>
•a
u
4>
(A
4*
O
=5 0.1
44
W.
a.
0.01
i
ML
&
«
Sl-
M1
II
SI-
M2
i III I
1
I
SI -
M3
&
lif
I
III
1 1
i in i mm i
11
S2-
M1
II
S2-
M2
I
S2-
M3
Scenario (S) and Modeling Method (M)
Minimum E3 25th %
Median
"75ih %
Maximum
Figure 1. Comparison of predicted versus measured VOC concentrations, for three
prediction methods, in two cross-validation scenarios.
-------
Three-dimensional visualization of the VOC concentration functions produced by the three
methods was accomplished using the same commercial package that provided the
predictions in Method I.
The majority of the ratios of predicted concentration to "true" concentration were between
0.18 and 2.9 for Method II; Method I ratios were somewhat higher and Method III were
somewhat lower. The 95% Bayesian probability intervals of Method III failed to reach the
highest 13% of the observed VOC values. Moreover, it should be noted that these
probability intervals were quite wide, the upper bound on the predicted VOC concentration
being 200 to 300 times the lower bound (Figure 2).
Calculations of the mass of VOCs within the entire site and sub-regions within it were
made using the first data set with 176 soil samples. Because of the highly skewed nature
of the data, simple integration of the predicted VOC concentrations, which were based on
the log-transformation of the observed concentrations, would yield a severely biased (low)
estimate of the average concentration. To overcome this problem, stochastic simulations
were made based on a kriging approach (Method III). Essentially, one generates random
samples from the posterior (conditional on the observed data) distribution of concentration
functions, which expresses the uncertainty about the true concentration at all unsampled
locations in the domain of interest For practical reasons-, this is done only at the points of
a regular grid or at a set of points randomly selected from the plot. For each simulated
realization, the average concentration over this set of points is calculated. The population
of these averages over many simulations provides a collection of "feasible" values for the
true average VOC concentration. Assuming constant soil density over the plot, these
values can be converted directly to mass estimates. Various methods were used to
generate the stochastic simulations, including the "turning bands" method [6,7], as well as
a direct method that was derived for this implementation. Again, 95% probability intervals
for the VOC mass estimates were quite wide.
DISCUSSION
Three dimensional spatial modeling can be more complicated and/or cosdy to implement
with currendy available computing tools, but it can provide enhanced information about the
VOC distribution within a contaminated soil region. In this study, the different spatial
modeling methods resulted in relatively similar descriptions and predictions of soil VOC
concentrations. However, the kriging approach (Model III) is better able to provide
confidence limits, and is more useful for estimating total VOC mass within a contaminated
soil volume. Color graphic visualization of the three-dimensional dataset can aid in
communicating the spatial modeling results. This information can yield insight into the
VOC transport and fate properties and conceptualization of remedial action plans.
Given a three-dimensional dataset comprised of soil VOC measurements, prediction of the
median concentration within a region of interest can be achieved with a reasonable degree
of accuracy (e.g., predicted to observed = 1.0 to 2.0). However, accurate prediction of
soil VOC concentrations at discrete unobserved locations within the region can still be
exceedingly difficult In this study where a moderately high sampling density was
employed to characterize a 26,000 c.y. region (e.g., 1 sample per 100 c.y.), predictions
within this region were often high or low by one order of magnitude when compared to the
actual measured values. Difficulties in prediction of VOC concentrations at discrete
locations is due to spatial heterogeneities and to the non-uniform anthropogenic activities
-------
responsible for the release of VOCs to the subsurface. The multiphase behavior of VOCs
no doubt exacerbates the spatial variability normally observed at contaminated sites.
These results suggest that greater understanding of VOC concentration distributions within
a three-dimensional soil region require a high sampling density. The number of spatially
separate data points needs to be balanced with the accuracy and precision of the
measurement technique. Given the measurement error caused by conventional soil sample
handling and off-site laboratory analysis (up to 100% negative bias), support is given to
making more spatially separate measurements using less expensive, onsite analysis
techniques [8,9].
1 ¦ ¦ ' ' ¦ I i i I I ' ¦ ' I I I I I I I I L
Soil Sample (sorted by increasing predicted value)
— — — — upper
"""" "" Lower
Value
Predicted
Bound
Bound
Value
Figure 2. Comparison of predictions made using a spatial model developed from a site-
wide dataset (Method III Bayesian Approach) and subsequently applied to a
region of the site. (Note: upper and lower bounds are 95% Bayesian
probability limits)
-------
ACKNOWLEDGMENTS
This work was supported with funds from the U.S. Department of Energy, Office of
Environmental Restoration administered through the Portsmouth Gaseous Diffusion Plant.
Without this financial support and assistance, this work could not have been accomplished.
We would like to thank Envirosurv, Inc. (Arlington, VA) who conducted the soil sampling
and VOC analysis that provided the data set used in this study. We would also like to
thank Doug Pickering and Chris Muhr, both from the ORNL/Grand Junction office, for
supervising the field sampling and analysis.
REFERENCES
1. "Field Screening Methods Catalog, User's Guide", EPA/540/2-88/005.
Environmental Protection Agency, September 1988. (FM-5) Volatile Organic
Compound Analysis Using GC with Automated Headspace Sampler.
2. Briggs, I.C. (1974), "Machine Contouring Using Minimum Curvature"
Geophysics, 39, 39-48.
3. Wahba, G. and Wendelberger, J. (1980), "Some New Mathematical Methods for
Variational Objective Analysis Using Splines and Cross Validation", Monthly
Weather Review, 108,1122-1145.
4. Wahba, G. (1990), "Spline Models for Observational Data", CMBS-NSF Regional
Conference Series in Applied Mathematics, #59, Society for Industrial and Applied
Mathematics, Philadelphia.
5. Gu, C. (1989), "RKPACK and its Applications: Fitting Smoothing Spline
Models", Technical Report #857, Dept. of Statistics, University of Wisconsin-
Madison.
6. Journel, A.G. and Huijbregts, Ch. J. (1978), Mining Geostatistics, Academic
Press, London.
7. Kitanidis, P.K. and Lane (1985), R.W., "Maximum Likelihood Parameter
Estimation of Hydroloaic Spatial Processes by the Gauss-Newton Method, J.
Hydrol., 79, 53-71.
8. Mantoglou, A. and Wilson, J.L. (1982), "The Turning Bands Method for
Simulation of Random Fields Using Line Generation by a Spectral Method", Water
Resources Research. 18, 1379-1394.
9. Siegrist, R.L. and P.D. Jenssen. (1990). Evaluation of sampling method effects
on volatile organic compound measurements in contaminated soils. Environmental
Science & Technology. Vol.24, No.9, p. 1387-1392.
10. Siegrist, R.L. (1991), Volatile organic compounds in contaminated soils: The
nature and validity of the measurement process. J. Hazardous Materials. Vol. 29,
p. 3-15. September 1991.
-------
BIOGRAPHICAL SKETCHES
Toby J. Mitchell is a Senior Research Staff Member in the Statistics Group of the
Engineering Physics and Mathematics Division at Oak Ridge National Laboratory. After
receiving alPh.D.- in Statistics at the University of Wisconsin in 1966, he came to work at
ORNL, where he has been ever since, except for occasional leaves of absence to the
University of Wisconsin and the National Institute of Environmental Health Sciences. He
was elected Fellow of the American Statistical Association in 1979. His current research
interests include spatial statistics, design of experiments, and Bayesian statistical methods.
Mailing Address: Oak Ridee National Laboratory
P.O. Box 2008, MS 6367
Oak Ridge, TN 37831
Phone. No.: 615-574-3143 E-mail: mitchell@msr.epm.0ml.20v
Fax. No.: 615-574-0680
Olivia R, West is a Research Associate in the Soils and Environmental Engineering
Group, of the. Environmental Sciences Division at Oak Ridge National Laboratory. She
received a Ph.D. in Civil Engineering from the Massachusetts Institute of Technology in
1991, an S.M. in Civil Engineering in 1988, and a B.S. in Civil Engineering from the
University of the Philippines in 1985. Her research interests include three-dimensional
groundwater modelling, and in-situ remediation of VOC-contaminated soil.
Mailing Address: Oak Ridge National Laboratory
P.O. Box 2008, MS 6036
Oak Ridge, TN 37831
Phone. No.: 615-576-0505 E-mail: qm5@oml.gov
Fax. No.: 615-576-8543
Robert L. Siegrist is a Research Staff Member in the Soils and Environmental
Engineering Group of the Environmental Sciences Division at Oak Ridge National
Laboratory. He received a Ph.D. in Environmental Engineering at the University of
Wisconsin in 1986 after receiving an M.Sc. in Environmental Engineering in 1975 and a
B.Sc. in Civil Engineering in 1972. Before joining ORNL. Dr. Siegrist worked with the
University of Wisconsin, Ayres Associates, Inc., and the Norwegian Center for Soil and
Environmental Research. He is currently Adjunct Associate Professor at The University of
Tennessee. At ORNL he is responsible for research on environmental restoration. His
current interests include the behavior and measurement of organics in soils, in situ soil and
ground water treatment processes, and development of environmental quality criteria.
Mailing Address: Oak Ridge National Laboratory
P.O. Box 2008, MS 6038
Oak Ridge, TN 37831
Phone. No.: 615-574^7286 E-mail: bs7@oml.gov
Fax. No.: 615-576-8646
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 17
Comparison of Sample Collection and Handling Practices for the
Analysis of Volatile Organic Compounds in Soils
Alan D. Hewitt
U.S. Army Cold Regions Research and Engineering Laboratory,
Hanover, New Hampshire
January 12-14, 1993
Las Vegas, Nevada
-------
COMPARISON OF SAMPLE COLLECTION AND HANDLING PRACTICES FOR
THE ANALYSIS OF VOLATILE ORGANIC COMPOUNDS IN SOILS
Alan D. Hewitt
U.S. Army Cold Regions Research and Engineering Laboratory
72 Lyme Road, Hanover, N.H. 03755-1290, (603) 646-4388
This paper assesses the current approach (i.e., collection of subsamples for trans-
portation and storage followed by laboratory allocation) for obtaining test samples
from vadose zone soils for the determination of Volatile Organic Compounds (VOCs).
Arguments for new procedures are developed on the basis of the results from a field
study comparing the current method to a less disruptive, limited-exposure handling
method (LDE method), and are supported by other laboratory and field studies, includ-
ing those describing how VOCs exist in porous media.
During a recent site investigation for the U.S. Army Toxic and Hazardous Materi-
als Agency, a sample collection and handling comparison study was performed on
vadose zone soils contaminated with trichloroethylene (TCE). Soil samples from eight
boreholes were retrieved via split spoon samplers following hollow stem auguring. The
major variable investigated was the handling of the soil as it was transferred from the
split spoon to a collection vessel. The site investigators used a stainless steel serving
spoon and their hands to transfer subsamples to VOA vials for transportation and
storage. This standard protocol was compared to the LDE method, which uses a small
subcoring device to transfer subsamples to vials from which analysis was conducted
without additional exposure to the atmosphere. Thus, the soil sample collection and
handling procedure that limits soil structure disruption and exposure also acquires the
test samples during the field sampling exercise.
Current EPA field sampling guidelines (SW-846) for the determination of VOCs
in soil specify that subsamples be shipped to the laboratory in bottles that have been
filled to capacity (i.e., no headspace) and stored at 4°C for no more than 14 days, and
caution against contamination (clean plastic surfaces, plastic gloved hands, stainless
steel spoons or spatulas, glass bottles with Teflon closures).1 However, the above
document does not address the actual transfer process, nor, in this particular case, did
the site sampling plan, other than to put a time limit on the transfer of subsamples in
the field and to specify the use of 40-mL VOA vials as the transportation and storage
vessel.
For this site the investigators used stainless steel serving spoons with 3.5-cm o.d.
to fill 40-mL VOA vials with 1.8-cm i.d. openings. These utensils caused the threads
and sealing surfaces to become soiled during the collection of subsamples. Attempts to
-------
clean these surfaces not only increased the duration of exposure, but were marginally
successful, as evidenced by the sound of glass scoring when the vials were capped.
Following this procedure, the site investigators took approximately 1 minute to collect
subsamples. The contract laboratory analyzed soils obtained by the site investigators
by direct aqueous sparging of 5-g test samples.2 Thus, following the EPA guidelines,
the contract laboratory receiving the soil subsamples must transfer a known weight of
soil from the transportation and storage vessel to a sparge tube. Then the sparge tube is
connected to a purge-and-trap manifold of a GC/MS instrument for analysis. During
this test sample allocation in the laboratory, the soil is disaggregated and exposed to
the atmosphere for a second time.
In contrast, the new LDE method acquires test samples in the field while making
the initial transfer.3,4 A small coring device, such as a 10-cm3 syringe with the end
removed, was used to transfer soil plugs of discrete sizes from the split spoon to a
VOA vial. The coring device fit easily into the opening of the 40-mL VOA vessel.
This transfer process was often done in less than 10 seconds, and. wiping the syringe
barrel after obtaining a soil plug with a clean cloth ensured that the sample vessel
sealing surfaces remained clean. The collection of test samples during the field sam-
pling exercise eliminates additional laboratory transfer steps, other than removing
aliquots of MeOH, a solvent that retains VOCs. For this reason, subsamples collected
using the LDE method were placed into vials either containing 20 mL of MeOH, being
equipped with the modified purge-and-trap adapter for PT/GC/MS analysis, or contain-
ing 30 mL of water for HS/GC on-site screening analysis.2
Figure 1 compares the TCE concentrations obtained by the two sampling and
handling procedures, both analyzed by PT/GC/MS. This comparison is based on soil
subsamples taken from the same split spoon. Only those concentrations where TCE
was above the instrumental detection limits of PT/GC/MS are presented. In the ab-
sence of either random or systematic errors, all results would lie on the solid line
representing the theoretically expected 1 -to-1 correspondence of results. The actual
data indicate that TCE concentrations obtained by using the EPA sample collection and
handling protocol were one or more orders of magnitude lower that those obtained for
. the LDE method.
The above differences could be caused by either the sampling method or analysis.
To check this, samples collected by the same method (LDE method) from the same
split spoons were analyzed by both HS/GC and PT/GC/MS (Fig. 2). Here, the experi-
mental results cluster around the solid line, having a linear coefficient of determination
(r2) of 0.906. This agreement between these two methods of analysis is consistent with
an analogous study performed with laboratory treated soils, and thus provides confi-
dence in the LDE method of sample collection.5 Additional explanations for the
marked discrepancy between samples collected by the LDE method and the standard
EPA procedure might be interlaboratory instrumental calibration, biodegradation, hold-
ing time and analyte inhomogeneity. A careful analysis of all of these potential vari-
ables resulted in the conclusion that only the error introduced by sample collection and
handling could account for the magnitude of the differences between the laboratories2
-------
Figure 1. Log-log plot of TCE concentrations established by PT/GC/
MS for soil samples collected and handled by standard protocol and
LDE method.
This finding is supported by studies that have previously demonstrated that sample
disturbance can cause. VOC losses up to 80%.4'6 Additionally, both field and labora-
tory studies have shown large VOC losses when comparing the transfer of subsamples
to vessels with and without a solvent present.7'8 These studies, which included several
VOCs, also observed that losses tended to increase with analyte volatility,4,6,7 Along
with the studies that identify problems with sample transfer practices for several
VOCs, failure to completely remove soil from threads and sealing surfaces of sampling
bottles could contribute to VOC losses during storage. To test this hypothesis, MeOH
was stored in VOA vials, with and without grains of soil smeared over the bottles'
necks and threads. Soiled surfaces were wiped with a gloved hand before capping,
-------
Figure 2. Log-log plot of TCE concentrations established by PT/GC/
MS and HS/GC for samples collected by the LDE method.
which was the practice observed at this site. The MeOH immediately and continuously
was lost from the majority of soiled VOA vials, indicating that they could not be
properly sealed.2
The degree of sorption of nonionic VOCs in soil depends principally on water and
organic matter content.9 Typically, subsoils contain little organic matter, thus the com-
bination of moisture and soil texture-structure often control sorption for VOC vapors.
It is believed that VOCs sorb onto dry mineral surfaces by a physical process, perhaps
involving van der Waals forces.10 This concept of weak physical interactions has been
supported by sorption-desorption studies of VOC vapors on dried clay minerals, show-
ing an initial fast interaction followed by a slower one.10'11 The faster process is as-
sumed to be surface mineral release-sorption, whereas the slower process most likely
involves diffusion into or out of intra-aggregate micropores. Site investigators and
-------
laboratory technicians using the EPA guidelines disaggregate the porous soil to various
extents, depending on the structure and texture, while transferring it to and from the
storage and collection vessel. Under these circumstances, the disaggregation of native
soil would expose fresh grain surfaces, releasing physically trapped soil gases and
weakly sorbed molecules. Additionally, it is anticipated that rapid shifts in equilibrium
and partitioning occur when the isolated diffusion controlled microstructure of a po-
rous medium is disturbed. To avoid these potential loss mechanisms, the LDE method
transfers subsamples during field sampling exercises by using a sub-coring device,
such as described here and elsewhere,2,3 with little alteration to the native structure for
soils of small- and medium-sized grains and moisture contents ranging from moderate
(5%) to water saturated.
Collectively, these findings strongly suggest that proper sample collection and
handling procedures are necessary to obtain, accurate measurement of VOCs in soil.
The analysis of soil samples that have been disaggregated and exposed to the atmo-
sphere or stored in a vessel with soiled sealing surfaces will result in low VOC concen-
trations and false negatives. In contrast, methods that limit soil disruption and expo-
sure, and that require only a single transfer step in the acquisition of a test sample,
maximize the ability to obtain field samples representative of in-situ soil VOC concen-
trations.
Acknowledgments
Funding for this work was provided by the U.S. Army Toxic and Hazardous
Materials Agency, Marty Stutz, Project Monitor. The author thanks Dr. T.M. Spittler
for providing the information concerning headspace gas chromatography and the
method for collecting limited-disturbed soil subsamples, Barbara Topor and John Peck
from Ecology and Environment, Inc., and Dr. T.F. Jenkins and J. H. Cragin for critical
review of the text.
This publication reflects the views of the author and does not suggest or reflect
policy, practices, programs, or doctrine of the U.S. Army or of the Government of the
United States.
-------
References
1. Environmental Protection Agency (1986) Test methods for evaluating solid waste,
vol. IB. U.S. Department of Commerce, National Technical Information Service, U.S.
EPA, Washington, D.C.
2. A.D. Hewitt (in press) In: 16th Annual Army Environmental R&D Symp.,
Williamsburg, Virginia.
3. T.M. Spittler (1989) Personal communication. U.S. Environmental Protection
Agency, Environmental Services Division-Region 1, Lexington, Massachusetts.
4. T.E. Lewis, A.B. Crockett, R.L. Siegrist and K. Zarrabi (1991) EPA/590/4-91/001,
Technology Innovation Office, Office of Solid Waste and Emergency Response, U.S.
EPA, Washington, D.C.
5. A.D. Hewitt, P.H. Miyares, D.C. Leggett and T.F. Jenkins (1992) Environ. Sci.
Technol., 26(10): 1932-1938.
6. R.L. Siegrist and P.D. Jenssen (1990) Environ. Sci. Technol., 24: 1387.
7. T.F. Jenkins and P.W. Schumacher (1987) USA Cold Regions Research and Engi-
neering Laboratory, Hanover, N.H., Special Report, SR 87-22.
8. M.J. Urban, J.S. Smith, E.K. Schultz, R.K. Dickinson (1989) In: 5th Annual Waste
Testing & Quality Assurance Symp., U.S. Environmental Protection Agency, Washing-
ton, D.C., pp. H-87-II-101.
9. C.T. Chiou (1989) Reactions and Movement of Organic Chemical in Soils, B.L.
Sawhney and K. Brown, eds., Soil Sci. Soc. Amer. Special Publ. 22, Madison, Wiscon-
sin, pp. 1-29.
10. B.L. Sawhney and M.P.N. Gent (1990) Clays and Minerals, 38: 14.
11. S.G. Pavlostathis and G.N. Mathavan (1992) Environ. Sci. Technol., 26: 532.
-------
BIOGRAPHICAL SYNOPSIS
Alan D. Hewitl is a research physical scientist at the Cold Regions Research and
Engineering Laboratory (CRREL) (72 Lyme Rd., Hanover, NH 03755-1290). He has a
B.A. in chemistry from the University of New Hampshire and a M.S. in chemical
oceanography from the University of Connecticut. Present areas of interest are
identification of hazardous waste in environmental samples, ground water monitoring,
and the chemistry of snow and ice.
-------
REVIEW OF CURRENT AND POTENTIAL FUTURE SAMPLING PRACTICES
FOR VOLATILE ORGANIC COMPOUNDS IN SOIL
Alan D. Hewitt
U.S. Army Cold Regions Research and Engineering Laboratory
72 Lyme Road, Hanover, New Hampshire 03755-1290
603-646-4388
ABSTRACT
This study compares two sampling and handling
methods for the collection of soils to be analyzed for
volatile organic compounds (VOCs). One method,
which may be incorporated into future protocols, uses a
simple subcoring device that allows set volumes of soil
to be removed rapidly from the surrounding substrate
and transferred to a tared analysis vessel, that a) can be
analyzed via needle-septum puncture, b) attaches to a
purge-and-trap system, or c) contains methanol. This
less disruptive method not only limits mechanical frac-
turing during collection and lengthy sample exposure
while transferring, but avoids soiling of the collection
vessel seals. The findings show that, in order to- acquire
more accurate VOC concentrations in vadose zone
soils, there is a need for limited disruptive and exposure
practices.
INTRODUCTION
Recently, trichloroethylene (TCE) contamination
was discovered in production wells at a U.S. Army fa-
cility. During the subsequent site investigation, several
boreholes.and monitoring wells were installed to define
the extent of contamination. Under the supervision of
the U.S. Army Toxic and Hazardous Materials Agency
(USATHAMA), collocated soil samples were taken for
chemical analysis. Collocated samples should not be
confused with sample splits, the latter representing a
sample that has first been homogenized (mixed), then
sampled. The collocated subsamples were obtained by
following the commonly accepted practice of filling a
transfer bottle for shipping and subsequent contract la-
boratory analysis, and with a recently proposed limited-
disruptive and exposure method (T.M. Spittler, personal
communication; Lewis et al., 1991), for on-site analy-
sis.
Current soil sampling protocols require that sam-
ples for the analysis of volatile organic compounds
(VOCs) be shipped to the laboratory in bottles that have
been filled to capacity (i.e., no headspace). Both the
practice of filling bottles, and the removal of a portion
after shipping, cause the soil sample to be disturbed. In
addition, while filling the bottle the sealing surfaces are
often soiled, compromising leak-free containment.
Present guidelines limit storage time to 14 days and
temperature to 4°C, while cautioning against cross con-
tamination (materials; i.e., stainless steel spoons, glass
bottles with Teflon closures, plastic-gloved hands). The
guidelines, however, do not address the actual collec-
tion process, and thus fail to provide detailed de-
scriptions of how soil samples should be handled in the
field or transferred from the shipping vessel after being
returned to the laboratory.
The proposed method of T.M. Spittler (personal
communication) and Lewis et al., (1991), specifies a
procedure for transferring soil subsamples that elim-
inates unnecessary intergranular disruption and ex-
posure. In particular, their protocol states that dis-
turbance of the native soil structure should be
minimized, and that placement of subsamples should
only be into a vessel from which it can be analyzed, or
that contains a solvent. This can be achieved by using a
simple coring device that allows small "plugs" of soil to
be rapidly removed from the surrounding substrate and
quickly transferred to a tared vial, which a) is analyzed
by a needle-septum puncture, b) has an adapter al-
lowing for direct attachment to a purge-and-trap sys-
tem, or c) Contains an extraction solvent such as meth-
anol (MeOH). This method not only limits subsample
disruption and exposure but avoids soiling of the collec-
tion vessel sealing surfaces, ensuring a leak-free clo-
sure. The objective of this study was to determine the
effects that these two sampling and handling protocols
have on the measured concentrations of VOCs in soil.
EXPERIMENTAL
An outline of the two soil-sample collection and
handling methods used and their effect on the con-
centration of VOCs present is discussed here. Detailed
descriptions of the analysis procedures used on site and
by the contract laboratory can be found elsewhere (He-
witt et al., 1991; Hewitt et al., 1992; Hewitt et al., under
review).
I. Subsample Collection
A. Samples Collected for Contract Laboratory
Analysis
The site investigators tasked with the collection of
samples were requested to fill two 40-mL VOA vials
-------
for VOC analysis. These narrow-mouthed vials (1.8-cm
i.d.) were filled by transferring soil from the split-spoon
sampler using a stainless steel serving spoon (=3.5 cm
dia.). Using this collection method, it took about one
minute to fill each VOA vial, and left the vial's threads
and sealing surface covered with grains of soil. At-
tempts were made to wipe these surfaces prior to cap-
ping; however, this was effective only for small soil ag-
gregates; individual grains remained, as was apparent
by the sound of glass scoring, upon tightening the caps.
B. Samples Collected for Analysis On Site
The collection method used for samples analyzed
on-site followed the limit-disruptive and exposure tech-
nique of T.M. Spittler (personal communication) and
Lewis et ai. (1991). A 10-cc plastic syringe (1.6-cm
o.d.) with the needle end removed (Fig. 1) was used to
extract, isolate and transfer soil plugs (Fig. 2) from the
split spoon into the collection vessel. The plunger was
set so that between 2 and 3 cc of soil was retained in the
cylindrical barrel. After wiping the external surface of
the syringe, the soil plug was dispensed into the ap-
propriate VOA vial by depressing the plunger (Fig. 2).
A minimum of two soil plugs was extracted from each
split spoon sampled, one for headspace gas chrom-
Figure 1. Illustration of five subsample soil plugs
obtained with the coring device, for purge-and-trap
analysis.
atography (HS/GC) and the other(s) for purge-and-trap
gas chromatography mass spectrometry (PT/GC/MS)
analysis. Vials taken into the field were all tared and
contained either 30 mL of Type 1 water (HS/GC), 20
mL of MeOH, or were empty. The empty vials were
equipped with a modified purge-and-trap vial adapter
(Associated Design Model PT-6005-0002) which al-
lows for direct connection to a purge-and-trap system.
The decision on whether to place the soil plug into the
VOA vial containing 20 mL of MeOH or the empty vial
was made in the field based upon readings taken from a
PhotoVac (Photovac, Inc.) VOC hand-held probe.
When readings of greater than 200 parts per million
(ppm) total ionizable gases were observed, the vial con-
taining MeOH was used, whereas for readings less than
200 ppm the empty vial was selected. Measurement of
total ionizable gases was performed by sticking the sen-
sor tip into a freshly created finger depression in the
soil sample. All VOA vials were weighed after collec-
tion to determine the actual weight of soil (moisture in-
cluded). Each subsampling took approximately 15 sec-
onds, and, as can be seen in Figure 2, the majority of
the soil plugs remained intact, showing little granular
disruption, thereby minimizing exposure to the at-
mosphere prior to sealing the vial. Sealing surfaces re-
mained clean, allowing for leak-free closure.
II. Transportation and Storage
A. Contract Laboratory Samples
The VOA vials filled for VOC analysis were sur-
face wiped, recorded, custody sealed, stored and
shipped at 4°C, and analyzed within 14 days of collec-
tion.
B. Samples Analyzed On Sire
All soil samples were stored at 4°C. Aqueous ex-
traction HS/GC samples were analyzed within two days
and PT/GC/MS samples within 14 days of collection.
III. Laboratory Sample Handling and VOC Analysis
A. Contract Laboratory
Five grams of the soil were removed from the vial,
weighed, and transferred to a sparge tube. Once con-
nected to a sample purging system, 5mL of- water with
the appropriate surrogates were added and the sample
analyzed by USATHAMA-certified method LM16, a
PT/GC/MS analysis method that is similar to EPA SW-
846, Method 8240 (Hewitt et al„ 1991).
Figure 2. Illustration of 10-cc and 3-cc coring devices,
and a 10-cc corer inside a 40-mL VOA vial.
B. On Site
The subsample collection and handling method
used allows for direct analysis of the vessels into which
-------
Table 1. Trichloroethylene concentrations for collocated subsamples taken during the drilling of eight
boreholes.
Sample
On-site
~HS/GC
(mg/Kg)
•PT/GC/MS
Contract Lab
PT/GC/MS
(mg/Kg)
Borehole 1
A <0.003 NA
B <0.003 NA
C <0.003 <0.003
Borehole 2
A <0.003 NA
B <0.003 NA
C <0.003 <0.003
Borehole 3
A <0.003 NA
B <0.003 NA
C <0.003 <0.003
Borehole 4
A 0.20/0.11 0.14
B 9.6 13.0
C 130.0 33.0
D 0.40 0.63
Borehole 5
A 2.5 2.3
B 1.4 NA
C 15.0 0.10
Borehole 6
A <0.003 NA
B <0.003 <0.003
Borehole 7
A <0.003 NA
B <0.003 <0.003
B.orehole 8
A 1.6 3.6
B 0.30 0.53
C 0.58 0.82
* HS/GC Headspace gas chromatograph
•PT/GC/MS Purge-and-trap, gas chromatograph, mass spectrometer
NA Not analyzed
** Method reporting limit
**<0.0038
<0.0038
<0.0038
<0.0038
<0.0038
<0.0038
<0.0038
<0.0038
<0.0038
0:014
0.046
0.095
<0.0038
0.020
<0.0038/<0.0038
<0.0038
<0.0038
<0.0038
<0.0038
<0.0038
0.016
0:012
0.20/0.19
the soil plugs were placed, or the analytes of interest are
present as solutes in MeOH, a solvent from which
VOCs are not easily lost. Analysis was performed by ei-
ther HS/GC or by EPA SW-846, Method 8240 (Hewitt
et al., 1992).
RESULTS AND DISCUSSION
The TCE concentrations found in soils are pre-
sented in Table 1. Included are the results reported by
the contract laboratory (CL) and from on-site analysis.
None of the values were corrected for soil moisture
content, and those values supplied by US ATHAMA via
the contract laboratory will be processed further. The
effects of using preliminary data have little influence on
the interpretation presented.
Log-log plots of the soil TCE concentrations
shown in Figures 3-5 represent all collocated samples
where at least one method of analysis established a
value greater than 0.0038 mg TCE/Kg (reporting limit
-------
Figure 3. Log-log plot of collocated purge-and-trap
trichloroethylene soil concentrations (mg TCE/Kg), as
. established by the contract laboratory and on site.
Figure 4. Log-log plot of collocated purge-and-trap
and headspace-gas chromatography trichloroethylene
soil concentrations (mg TCE/Kg) as established by the
contract laboratory and on site, respectively.
Figure 5. Log-tog plot of collocated purge-and-trap
and headspace-gas chromatography trichloroethylene
soil concentrations (mg TCE/Kg) as established on site.
value). These figures include the axis of theoretical
agreement (solid line), and order of magnitude intervals
about this axis (dashed lines). By observing where the
points fall in relation to the dashed lines one can see rel-
ative disagreement between the two methods being
compared. Figures 3 and 4, both of which plot the re-
sults of the contract laboratory versus those obtained on
site using PT/GC/MS and HS/GC, respectively, show
that the majority of points differ by at least one order of
magnitude, and frequently by two or three orders of
magnitude. On an average basis, the disagreement ap-
proaches two orders of magnitude. Conversely, eight of
the nine points shown in Figure 5, a plot depicting soils
collected with the limited-disruptive and exposure
method and analyzed by two different analytical meth-
ods, are within one order of magnitude from the theo-
retical agreement axis. Overall, the variation present in
Figure 5 is well within the range of VOC analyte var-
iability often encountered with environmental soil sam-
ples (Hewitt et al., 1992).
Linear regression analysis comparing the TCE con-
centrations for these above background values (>0.0038
mg TCE/Kg) established the following correlation co-
efficients: CL-PT/GC/MS versus on-site-PT/GC/MS, r2
= 0.0454; CL-PT/GC/MS versus on-site-HS/GC, r2 =
0.0524; on-site-PT/GC/MS versus on-site-HS/GC, r2 =
0.878. Clearly, good agreement exists for collocated
-------
Table 2. Results for trichloroethylene in ground-
water samples.
On-Site Contract Lab
*HS/GC -PT/GC/MS
Well Sampled (mg/L) (mg/L)
1
<0.0001
<0.0005
2
1.0
0.53
3
<0.0001
<0.0005
4
0.0047
0.0060
5
0.040
0.050
6
46.0
14.0
7
0.011
0.0059
8
1.4/1.5
1.2/1.1
9
160.0
64.0
11
3.2
1.5
12
0.29
0.18
*HS/GC Headspace gas chromatograph
•PT/GC/MS Purge-and-trap, gas chromatograph,
mass spectrometer
subsamples when analyzed on site (valid at 99.9% con-
fidence level). Results for the collocated subsamples
collected by the site investigators using the currently ac-
cepted field sampling practices and analyzed at a con-
tract laboratory do not correlate with the results ob-
tained for those collected and handled by the newly
proposed less disruptive and limited exposure pro-
cedure, and analyzed on site (Fig. 3,4).
Two trends are apparent: 1) soil subsamples an-
alyzed by the contract laboratory resulted in lower TCE
concentrations than those analyzed on site; 2) soil sub-
samples collected with the limited-disruptive and ex-
posure method and analyzed by either HS/GC or PT/
GC/MS (EPA, Method 8240) provided very similar
TCE soil concentrations. The latter finding agrees with
our earlier studies that have been documented else-
where (Hewitt et al., 1991; Hewitt et al., 1992; Hewitt
et al., under review). Since the focus of this paper is
collection and handling practices, all possible explana-
tions for the marked discrepancy for the inter-
laboratory results will be addressed. These include in-
strument calibration, biode'gradation, analyte in-
homogeneity, and subsample collection and handling.
A. Calibration Error
The results 'obtained for a round of groundwater
samples analyzed by both the contract laboratory and
on site are shown in Table 2. These water samples were
collected by the site investigators two weeks after com-
pleting the installation of six monitoring wells (Table 3,
wells 7-12). The other wells presented in this table are
existing service wells. Here the method of analysis used
by the contract laboratory was USATHAMA-certified
•method UM17, a method based on PT/GC/MS; HS/GC
was performed on site. The results from the two la-
boratories are in much better agreement (r2 = 0.995,
Fig. 6) than was achieved for the soil samples (Fig. 3,
4). This water analysis comparison demonstrates that
the departure in values for the TCE soil determinations
between laboratories is most likely not a calibration
problem. Additionally, the discrepancy between la-
boratories for the TCE soil concentrations was random,
a pattern not attributable only to calibration offset (Fig.
3,4).
B. Biodegradation Losses
Since aqueous extraction HS/GC is not a de-
structive analysis method (subsamples are not sac-
rificed), a subset of these samples was monitored over a
Table 3. Holding time study of sample subset stored at room temperature and analyzed by headspace gas
chromatography.
*TCE mg/Kg
Holding Time
Sample
Id
4d
8d
lid
15d
21 d
25d
1
1.6
1.6
1.6
1.6
1.6
2
16.0
15.0
15.0
—
15.0
—
15.0
3
130.0
130.0
130.0
—
—
130.0
"4
0.29
0.30
0.28
0.26
—
0.24
—
5
0.25
—
0.26
—
0.24
—
0.23
6
0.15
—
0.15
—
0.14
—
0.13
*TCE.... Trichloroethylene
•4 Increase in cis-1,2-dichloroethylene observed as the concentration in TCE decreased!
-------
Figure 6. Log-log plot of purge-and-trap and head
space-gas chromatography trichloroethylene ground
water concentrations (mg TCEJmL) for first sampling
round.
period of 21 to 25 days (Table 4). The soil subsamples
selected were representative of the range of TCE con-
centrations observed, and include several taken near the
surface where higher biological activity is anticipated.
To stress potential effects, treatment between analyses
consisted of room temperature (22°C) storage. Between
analyses, vials were inverted so that the suspended soil
settled onto the Teflon-lined septa. Little or no decrease
in the TCE concentrations was measured as a result of
this treatment. The sample with the greatest relative
loss of TCE showed a corresponding increase in cis-
1,2-dichloroethylene. Although this test does not du-
plicate the handling and storage conditions subjected to
those samples sent for contract laboratory analysis, it
provides strong evidence that rapid biodegradation of
this highly chlorinated organic compound is unlikely,
even for an extended hplding period.
C. Sample Inhomogeneity
To address the problem of sample inhomogeneity,
the collection of multiple subsamples is usually per-
formed to assess analyte variability. The results in
Table 5, for subsamples taken during the drilling of the
groundwater monitoring wells, show that the in-
homogeneity within a split spoon was small (threefold)
Table 4. Subsamples taken are listed from top of
split spoon to bottom, bottom being the end of great-
est depth. Number of samples collected from a split
spoon dependent on the length of soil core retrieved.
*TCE (mg/Kg)
Individual Sample
Average and
*HS/GC
Std. Dev.
0.042
A.
0.035
0.038±0.004
0.037
0.38
B.
0.39
0.45±0.12
0.59
0.48
C.
0.50
0.76+0.46
1.3
0.29
0.27
D.
0.21
0.22±0.07
0.10
0.12
0.066
0.14
E.
0.16
0.14±0.05
0.17
0.18
25.0
F.
24.0
25.0±2.0
27.0
0.20
0.31
G.
0.18
0.19±0.07
0.13
0.14
*TCE...
. Trichloroethylene
compared to the disagreement found between the two
laboratories (tenfold to thousandfold). These sub-
samples were taken approximately every seven cm
along the length of the split-spoon core with no visible
discontinuities in the soil structure. This range of TCE
concentration would not necessarily be representative
of all split spoons, especially if a clay lens had been
-------
penetrated. Other soil features that may influence VOC
analyte distribution are grain-size distribution, soil type
and structure, mineral type, and organic and moisture
content (Chiou, 1989). However, since the boreholes
were often in the near vicinity of these groundwater
monitoring wells, and since analyte variability is often a
random feature, it is very unlikely that the soil sampled
by the site investigator always contained less TCE than
the subsamples taken for on-site analysis.
D. Collection and Handling Method
Investigators (Siegrist and Jenssen, 1990; Lewis et
al., 1991) have shown that sample disturbance can
cause VOC losses up to 80%, relative to undisturbed
controls. Laboratory evidence (significant at the 95%
confidence level) exists for VOC losses when com-
paring subsamples that were transferred from one ves-
sel to others that were either empty or contained a sol-
vent (Jenkins and Schumacher, 1987). In addition,
soiled threads and sealing surfaces of sampling bottles
will contribute to VOC losses. To test this hypothesis, a
simple test was performed looking at MeOH stored in
VOA vials with and without grains (<0.2 mm dia.) of
soil from this site smeared over the bottle's neck and
threads, then wiped clean with a gloved hand, as had
been observed in the field (soil grains = 0.06 g/bottle).
The results showed that seven of ten soiled VOA vials
suffered continuous weight loss while no weight loss
was found from ten unsoiled vials. Thus, VOCs may be
continuously lost from Vessels when casually filled to
capacity with soil.
Presented here is evidence supporting the concept
that soil sample collection and handling treatments
were the major factor responsible for the large dis-
crepancy between the TCE concentrations reported by
the two laboratories for the collocated soil subsamples.
A major objective of a site investigation is to establish
representative VOC contamination concentrations in
soil. This objective can not be met if native soils are
disaggregated causing fractures along grain boundaries,
releasing trapped gases of VOCs and exposing new sur-
faces. Because of the poor retentive properties of soils
in general for VOCs (Chiou, 1989), and these two re-
lease mechanisms (disaggregation and exposure), sam-
ple collection and handling should be performed with
limited disruption and exposure (T.M. Spittler, personal
communication; Lewis et al., 1991).
SUMMARY
The following general outline should be in-
corporated into site sampling protocols if soils are to be
collected for VOC analysis:
1. Soil samples should be retrieved from freshly
exposed surfaces:
2. Transfers should be made with a coring device
with an o.d. which fits into the i.d. of the sam-
ple container.
3. Soil subsamples should be put only into vials
that:
are three quarters full of Type 1 water (for
HS/GC analysis);
contain a sufficient amount of MeOH (for
high level "> lug/g" PT/GC/MS analysis);
are equipped with a purge-and-trap vial
adapter (for low level "< 1 ug/g" PT/GC/
MS analysis).
4. Soiling of the threads and sealing surface of
the collection vial should be avoided.
NOTE: Although not covered in this study, pre-
servatives often will be necessary to limit bio-
degradation of aerobically labile VOCs.
The advantages of using this sampling protocol are:
1. VOC losses are minimized by minimizing soil
surface area exposure.
2. Sampling time efficiency is maximized.
3. Closure leakage is eliminated.
4. The volume of contaminated soil that leaves
the site for chemical analysis is minimized.
5. Aqueous extraction HS/GC in a screening ap-
plication:
can be performed with the objective to re-
duce the number of background (blank)
analyses performed by PT/GC/MS;
can meet data quality objectives set during
various phases of a site investigation, al-
lowing samples to be processed on site,
decreasing the tum-around time and over-
all analytical costs;
provide an initial assessment allowing the
contract laboratory to establish which
sample (one extracted with MeOH or de-
signed to be purged directly) should be an-
alyzed by PT/GC/MS.
CONCLUSION
Disruptive soil collection and handling protocols
are prone to losing greater than 95% of the VOC re-
tained by native sandy-silt vadose zone soils. Inter-
granular disruption, unnecessary exposure, and soiled
closure surfaces all contribute to the losses of VOCs
that are weakly bound to soil particle mineral surfaces
or transiently exist among the soil voids.
-------
REFERENCES
Chiou, C.T. (1989) Theoretical considerations of the
partition uptake of non-ionic organic compounds
by soil organic matter: in Reactions and Movement
of Organic Chemicals in Soils, B.L. Sawhney and
K. Brown, eds., Soil Sci. Soc. Amer. Special Publ.
22, 1-29.
Hewitt, A.D., P.H. Miyares, D.C. Leggett and T.F. Jen-
kins (1991) Comparison of headspace gas chrom-
atography with EPA SW-846 method 8240 for de-
termination of volatile organic compounds in soil.
USA Cold Regions Research and Engineering La-
boratory, Special Report 91-4.
Hewitt, A.D., P.H. Miyares, D.C. Leggett and T.F. Jen-
kins (1991) An evaluation of headspace gas chrom-
atography for the determination of volatile organic
compounds in soil. Proceedings from the 15th An-
nual Army Environmental R&D Symposium, Wil-
liamsburg, Virginia, June, 137-142.
Hewitt, A.D., P.H. Miyares, D.C. Leggett and T.F. Jen-
kins (1992) Aqueous extraction-headspace gas
chromatographic method for determination of vol-
atile organic compounds in soil. USA Cold Re-
gions Research and Engineering Laboratory,
CRREL Report 92-6.
Hewitt, A.D., P.H. Miyares, D.C. Leggett and T.F. Jen-
kins (1992) Evaluation of aqueous extraction/
headspace gas chromatographic determination of
volatile organic compounds in soils- Environ. Sci.
Technol. (under review).
Jenkins, T.F. and P.W. Schumacher (1987) Compari-
son of methanol and tetraglyme as extraction sol-
vents for the determination of volatile organics in
soil. USA Cold Regions Research and Engineering
Laboratory, Special Report 87-22
Lewis, T.E., A.B. Crockett, R.L. Siegrist and K. Zarrabi
(1991) Soil sampling and analysis for volatile or-
ganic compounds, EPA/590/4-91/001, Technology
Innovation Office, Office of Solid Waste and
Emergency Response, U.S. EPA, Washington,
D.C.
Siegrist, R.L. and P.D. Jenssen (1990) Evaluation of
sampling method effects on volatile organic com-
pound measurements in contaminated soils. En-
. viron. Sci. Technol. 24: 1387-1392
Spittler, T.M. (1989) Personal communication. U.S. En-
vironmental Protection Agency, Environmental
Services Division-Region 1, Lexington, Mas-
sachusetts.
ACKNOWLEDGMENTS
Funding for this work was provided by the U.S.
Army Toxic and Hazardous Materials Agency, Marty
Stutz, Project Monitor. The author thanks Dr. T.M.
Spittler for providing the information concerning head-
space gas chromatography and the method for col-
lecting nondisturbed soil subsamples; Barbara Topor
and John Peck from Ecology and Environment, Inc.;
and Paul Miyares and Bruce Brocket! for critical review
of the text. Special gratitude is extended to Lawrence
Perry whose extensive soil gas "survey of this site pro-
vided guidance, contributing to the success of this
study.
This publication reflects the views of the authors
and does not suggest or reflect policy, practices, pro-
grams, or doctrine of the U.S. Army or of the Govern-
ment of the United States.
-------
Comparison of Analytical
Methods for Determination of
Volatile Organic Compounds in
Soils
Alan D. Hewitt, Paul H. Mlyares, Daniel C. Leggett, and
Thomas F. Jenkins
U.S. Army Cold Regions Research and Engineering
Laboratory, 72 Lyme Road, Hanover; New Hampshire
03755-1290
Reprinted from
ES&T, Volume 26, Number 10, Pages 1932-1938
Copyright© 1992 by the American Chemical Society
and reprinted by permission of the copyright owner
-------
Comparison of Analytical Methods for Determination of Vofatfle Organic
Compounds in Softs
Alan D. Hewitt," Pauf ft. Mtyarea, Daniel C. laggaH, and Thomas F. Jenkins
U.S. Army Cakf Begtors Research and Ensgi&eT&p Laborauny. 72 i.yrra Poed, Hanover Mai*1 Hampshire 03755-1290
¦ This study wapsFes aii^scua eHractioii headspaw/^sa
chromatography and, purge-eiid-trap ^sa chroraatogra-
pby/mtkB-sp^rrFBHietfy iEPA isW-EsiG, taetl»od BSWI Ere
Ibe determination cf fc*n vtArtile tssdapace GC using a portable instrument Can
be used to screen soils on site for VOCs. providing rapid
sanje-day results, that will consistently identify the pres-
ence of these analytes and provide quantitative resuits
which are generally not aigrcifLcaEUy different from slower
more expensive, ^abwaWTi-based tuivg-e-and-traj; analysts-
Tnf reduction
Volatile organic compounds {VOCs! ase the zaost fre-
quently encountered vontanunartta at hazardous waste sites
{I, SJ. Beca.ua® of tbeii pervasiveness and transience in
soils, VOCs have Jraw) considerable atttaUoru Sorption
a! nonpolar VOC vapors in th€ subsoil zone ba9 bf.eii
shown to depend on the availability of water, since these
two constituents often compete for the same sites 13,4}.
In the absence of cuoistnre, Ki/ittaJ sta&cea idaes!' fljad
organic matter provide sices for VOC sorption; ss'-T^&Kssit
increases less sites. a*e avails&ie, and vOCs p&rtiti&n s»ith
the water phase. Vadose zone soils typically feiv-e Itltie
organic mattej {<3% j, thus, a smnpr)? preparation method
that uses 'rater as an atractant and/or dispersion medium
could potentially displace the majority of VOCs. a-bich
exist weakly sorbed to days, as components of 3oil solution
or as transient va(Xirb ;u unoccupied pare spaces.
Currently, pr&toe&b specify the collection of bulk soil
samples from which sabsemptes axe removed in jbp laiuv
ralory foi VOC ewyie {SI. Many saopks lawlt in *b6lc»
detection' or background concentrations. Screening re-
locate sails on site or upon receipt by the contracted
laboratory would permit more efficient selection ci aaaptes
for sapeasive purge-and-tap gas chrcm&Uiprspb ciaaa
speeUtwaete* (PT/QC/MSI analysis. Also, the capability
of preforming on-site analysis would ail aw for timely de-
cisions to he made purge-and-trap gas chroiaatcgraphic (PI7CC) analysis
foe VOCaui aq-jeoea samples baw t«er. repeated 'J-O, i'j'i.
Method cc caparisons for soil samples, however, have suf-
fered froBi heterogeneity of field samples or fca'-& asen
liotcd x saife cired^.' Kith MeOFT-'"H and TCE ttv two field soils.
Procedures used in the ptep&rsltCB and handling of both
leboratoay-fortified and field samples were identical prior
to e*trectica and analysis. Spitting was accomplished by
a vapor fortification procedure that is analogous to the
espawufesf nP-saoirated soils to vapors originating' from
a separate conteminaat phase, ire ts-1,2-Dicril orw ti-.vi to t
(TDCE), trichforoethyfene (TCE}, benzeae (Ben), and
toliMne (Toi) wer? selected because they represent con-
tamination with industrial solvents and petfcUarn prod-
ucts and they are frequently found at hazardous waste sites
(13) Some relevant physical properties are given in Table
J. Preparing VCC-contaminated soil samples bv vapor
fortiflcatior. allows for a taore rigorous method e vaJueticn
loan iapaBsiM&iisir.g-solvent-sprking lechaiijues cjireatly
practitad ia eway QA 'QC arogrsuns; Sizes: no VOC
performance evaluation sarrvptes are avaiiaiple n: pressrl
for %iis, the assessmertl of sample determination accuracy
j*lies 00 soiution spifce. and recovery testa. The volatility
of VOCi has made matiii spiking a difficult taAh. G
-------
Table II. Characteristics of Soils
soils
Point
USATHAMA
Barrow,
std
AK
CRREL
Clarkson
1.45
6.69
0.08
0.13
53.6
20.1
<5
12
1.43
2.00
17
22
<0.1
<0.1
<0.1
10
characteristic
organic carbon (%)
clay (%)
% moisture" (%)
dipersion rate6
(min)
"ASTM D2216-66 (i.e.. weight percent relative to dried soil).
6 Time required to disperse 2 g of soil in 30 mL of water by hand
shaking.
. water extraction HS/GC/photoionization detection (PID)
and PT/GC/MS (5). Aqueous-HS/GC sample prepara-
tion and analysis were streamlined for field screening ap-
plications. Water was used as the extracting agent, hand
shaking/agitation to partition the VOCs into the aqueous
phase, and a portable GC for determination. Sample
preparation for PT/GC/MS analysis depended on the
anticipated soil VOC concentration. For anticipated
concentrations of greater than 1 Mg/g. soil samples are
extracted with methanol and an aliquot of the extract is
analyzed by PT/GC/MS. For samples expected to contain
less than 1 Mg/g, the sample is added directly to a special
vessel from which VOCs are purged directly after adding
water and heating the slurry to 40 °C.
Materials and Methods
Regardless of sample history, all subsamples were han-
dled with identical procedures prior to extraction and
analyzed within 8 h of collection, with the exception of the
second set of PT/GC/MS analyses of subsamples taken
from the Clarkson TCE-contaminated soil.
Vapor Fortification Treatment. Two soils (the U.S.
Army Toxic and Hazardous Materials Agency (USA-
THAMA) standard soil no. A046 and a soil obtained from
Point Barrow, AK) were fortified using a vapor treatment
method (12, 14). Characteristics of these soils are listed
in Table IL No VOCs were detectable in either soil prior
to fortification. For subsample treatment, soils were
weighed into 40-mL volatile organic analysis (VOA) vials
and positioned uncapped on a perforated aluminum plate
inside a large desiccator. These VOA vials served as the
subsample vessel during fortification treatment and ex-
traction and often for analysis (MeOH aliquots were re-
moved for the high-level PT/GC/MS analysis). An open
Petri dish containing the€ortification solution was placed
under the samples (Figure 1). To attain the appropriate
concentrations, high-level samples used 2.00 g of soil and
1.00 g was used for the low-level samples. Empty vials were
included to check for sorption onto vial walls.
The fortification solution was prepared by combining
reagent-grade Tol (1.21 g), TDCE (0.503 g), TCE (0.586
g), and Ben (0.351 g) in MeOH and diluting to 100 mL in
a volumetric flask. Concentrations of VOCs in the soils
ranging from 100 to 1000 ng/g were obtained after expo-
sure to the equilibrium vapor above a 50-mL aliquot of this
solution in the fortification chamber. A second high-level
fortified soil in the 1-100 ng/g concentration range was
obtained by exposing the soils to the vapor from a 50-mL
aliquot of a 1:1 dilution of this solution with tetraethylene
glycol dimethyl ether (tetraglyme). Low-level concentra-
tions (0.1-10 ng/g) were achieved by exposing the soils to
vapors from 10-mL aliquots of 1:10 and 1:20 dilutions of
this solution with tetraglyme. Vapor fortification treat-
ment periods were 4 days and between 39 and 46 days.
Desiccator
VOA V:als
wijh and *.-ncjt Soil
vopor
Performed
Plat e
Petri Dish with Exposure Solution
o' VOCs
Figure 1. Vapor fortification chamber.
Each fortification treatment started with 12 laboratory
air-dried subsamples of each soil and four empty vials so
that two complete method comparisons could be per-
formed. The comparison set comprised six subsamples of
each soil and two empty vials; thus, triplicate soil sub-
samples and a single empty vial (control) were analyzed
by each method of analysis. After removal from the de-
siccator, the vials were aspirated for 10 min by placing
them along the front edge of an exhaust hood. This step
was necessary because the amount of VOC vapor remain-
ing in the headspace of each vied was significant. Aspi-
ration for 10 min lowered the VOCs in empty vials below
detection. After aspiration, prealiquoted volumes of the
appropriate solvents were rapidly added if necessary, and
the vials were capped.
Field Samples. Several soil subsamples were collected
either 3 ft below the surface with a Veihmeyer tube or from
the surface with a shovel at the Cold Regions Research and
Engineering Laboratory (CRREL). Samples were taken
from several locations near known sources of contamina-
tion. A second field soil was obtained as a bulk sample
(~40 g) from Dr. S. G. Pavlostathis of Clarkson University.
This soil was selected because it is so poorly dispersed by
water (Table II), and in a previous study it had demon-
strated slow aqueous desorption of TCE (21). The
Clarkson soil was refrigerated after receipt and subsampled
after 2, 11, and 162 days of storage.
Field soils were rapidly subsampled once brought into
contact with the atmosphere by taking 1.5-mL soil plugs
with a tipless 10-mL plastic syringe (22). Depending on
the method, the soil plugs were placed randomly into
preweighed VOA vials that were empty or that contained
either water or MeOH. Six sets of quintuplicate subsam-
ples were collected from the CRREL site, and three sets
of triplicate subsamples were removed from the bulk
Clarkson soil.
Standards. The combined analyte solution prepared
as the fortification solution also served as the analytical
stock standard. Without further dilution, this stock
standard was used for the analyses in the 100-1000 ng/g
concentration range. Table HI lists the dilutions necessary
for the other concentration ranges. The stock solution was
refrigerated at 4 °C, and dilutions were prepared daily as
needed. A new stock was prepared monthly. Both in-
strumental methods used the same analytical standards
for calibration.
Extraction and Analysis of VOCs Present in Soils,
a. Aqueous Extraction HS/GC Analysis. Consistent
with previous studies (11, 22), samples and blanks were
extracted with 30 mL of deionized water (Type 1, Millipore
Environ. Scl. Technol.. Vol. 26, No. 10, 1992 1933
-------
Table III. Volumes of Stock Studied Used for the
Different Rtnm of Expected VOC Concentrations is the
Soil
vol. working vol. MeOH concn range
working std tor extract or headspace for soil
aui" calibration CuL) for anal. ^L) VOCs<(ig/g]
HS/GC/P1D6
1:100 stock
2.5-160
100
0.1-10
V.10 stock
10-80
25
1-100
a lock
50-200
2
100-1000
PT/GC/MS*
1:10 atocfc
10
100
1-100
stock
10
10
100-10C&
" Stock: 1.21 g of tolueaa. 0-503 g of erane-l,2-
Figure S. log-log pto! of mgeft concentrations
-------
Table IV. Intermethod Comparison of (raas-l,2-Dichloroethylene (TDCE), Trichloroethylene (TCE), Benzene (Ben), and
Toluene (Tol) for High-Level Fortified Soils
Mean Concns ± SD (jtf/g) for Vapor Treatment: Undiluted MeOH Stock
4-day exposure
39-day exposure
TDCE
Ben
TCE
Tol
TDCE
Ben
TCE
Tol
HS/GC PT/GC/MS
USATHAMA Standard Soil
72.8 ± 5.9" 66.0 ± 2.6"
117 ± 6.5 94.3 ± 2.2
214 ± 9.8 202 ± 16"
492 ± 21 529 ± 56°
Point Barrow, AK, Soil
148 ± 4.0 170 ± 9.7
198 ± 10 204 ± 5.0°
319 ± 27 444 ± 12
689 ± 76 1120 ± 5.8
HS/GC
135 ± 11
184 ± 4.0
372 ± 9.5
885 ± 50
225 ± 10
256 ± 14
416 ± 24
927 ± 63
PT/GC/MS
122 ± 19°
177 ± 34°
380 ± 105"
1660 ± 427
230 ± 12°
281 ± 32°
613 ± 57
2740 ± 61
Mean Concns ± SD (jtg/g) for Vapor Treatment: 50:50 Mixture of MeOH Stock and Tetraglyme
4-day exposure
39-day exposure
TDCE
Ben
TCE
Tol
TDCE
Ben
TCE
Tol
HS/GC PT/GC/MS
USATHAMA Standard Soil
1.63 ± 0.11* 1.84 ± 0.66°
8.75 ± 0.19 5.67 ± 1.30
11.7 ± 0.40 9.22 ± 1.68°
42.9 ± 1.50 33.3 ± 3.8
Point Barrow, AK, Soil
12.3 ± 0.40 12.5 ± 1.8°
29.1 ± 2.9 22.0 ± 3.0
34.1 ± 2.9 39.0 ± 1.8
87.8 ± 9.6 117 ± 8.3
HS/GC
1.71 ± 0.15
8.75 ± 0.09
11.6 ± 0.29
43.4 ± 2.2
11.7 ± 0.70
26.7 ± 0.29
34.2 ± 0.35
96.2 ± 1.2
PT/GC/MS
4.93 ± 1.34
7.33 ± 1.42°
15.1 ± 2.4°
41.4 ± 4.9°
19.1 ± 0.70
26.9 ± 1.9°
53.0 ± 2.3
134 ± 8.1
° HS/GC and PT/GC/MS analyses were not statistically different at the 95% confidence level. 'Mean and standard deviation (jxg/g) for
triplicate samples.
10
10 100 1,000 10.000
PT/GC/MS (>ig/g)
Figure 3. tog-tog plot of mean concentrations (pg/g) of al high-level
VOC determinations in the fortified Point Barrow soil.
3. Here the plot shows that the majority of points fall
below the unity axes, another means of demonstrating that
the MeOH-PT/GC/MS analysis generally quantified
greater soil VOC concentrations. For this soil the overall
average VOC concentration differences between these two
methods (19%) increased with time, from 11 to 26%.
When we group Ben and TDCE, and TCE and Tol, sep-
arating the compounds with the two lowest and two highest
K0/a (Table I), plots of the mean concentrations deter-
mined by the two methods show very different behavior
(Figures 4 and 5). The linear regression of the mean
TDCE and Ben concentrations had a correlation coefficient
~
o 100 200 300
PT/GC/MS ((ig/g)
Figure 4. Linear plot of mean (/ig/g) trans -1,2-dlchloroethy lene and
benzene (Ben) concentrations In fortified sofls. Slope and correlation
coefficient, 0.944 and 0.993.
of 0.993 and slope of 0.944, while the plot of the mean TCE
and Tol concentrations again shows the majority of points
below the unity axes. The fact that water was less able
to extract these hydrophobic VOCs is not surprising and
agrees with previous works (7,24-26) which have addressed
the influence of soil organic matter on partition coeffi-
cients. Thus the difference in method performance with
the two soils is probably due to the high (6.69%) organic
carbon in Point Barrow soil compared to 1.45% for the
USATHAMA soil.
300
200 —
0
C3
c/5
1
< 100
Environ. Scl. Technol.. Vol. 2fi, No. 10, 1992 1935
-------
Table V. IaU*mielhod Comparison of irsfls-l.JZ-Dichloroethylene (TDCE), Trichloroetkylene (TCE), Benzene (Ben}, Toluene
(Tol) (or Low-Level Fortified Soil*
Mean Concns ± SD (ws/g) for Vapor Treatment: 1:10 Mixture of MeOH Stock and TetragLyme
4-day exposure 46-day exposure
HS/GC PT/GC/MS HS/GC PT/GC/MS
USATHAMA Standard Soil
TDCE 0.195 ± 0.0206 0.383*0.062 0.154 ±0 .036 0.269 i 0.103°
Ben i.U * 0.040 1.21 ± 0.143" 1,02 ± 0.107 1.06 ± 0.214"
TCE 1.00 ± 0.040 1.35 * 0.107 1.04~± 0.124 1.15 ± 0.306°
Tol 6.43 ± 0.25 8.87 ± 0.93 7.60 ± 0.33 7.70 ± 1.85"
Point Barrow, AK, Soil
TDCE 0.939 * 0.113 1.11 ± 0.322° 0.550 ± 0.037 0.953 ± 0.247
Ben 1.67 ± 0.125 2.05 ± 9.340° 1.60 ± O.OSS 1.98 ± 0.246
TCE 2.B0 i 0.2SG 3.10 ± 0.485" 2.54 i 0.35 3.36 ± 0.422
Tol 9.97 i 0.67 11.5 ± 0.62 9.29 ± 0.32 12.4 ± 0.31
Mean Concna ± SD Gug/gJ for Vapor Treatment: 1:20 Mixture of MeOH Stock and Tetraglyme
4-day exposure 45-day exposure
HS/GC PT/GC/MS HS/GC PT/GC/MS
USATHAMA Standard Soil
TDCE 0.135 ± 0.017" 0.290 ± 0.080 0.084 ± 0.012 0.150 ± 0.055°
Ben 0.801 ± 0.034 0.915 ± 0.007° 0.715 ± 0.059 0.729 i 0.183°
TCE 0.733 ± 0.073 .0.873 ± 0.057* 0.599 ± 0.065 0.S97 ± 0.195"
Tol 5.10 ±0.31 4.47 ±0.34° 5.71 ± 0.38 4.78 ± 0.81°
Point Barrow, AK, Soil
TDCE 0.488 ± 0.007 0.690 ± 0.016 0.384 ± 0.029 NAC
Ben 1.01 ± 0.024 1.20 ± 0.038 1.03 ± 0.085 NA
TCE 1.39 ± 0.035 1.72 ± 0.064 1.48 ±0.11 NA
Tol 5.33 ± 0.069 6.16 s 0.33 5.86 =t 0.37 NA
"HS/GC and PT/GC/MS analyses were not statistically different at tie 95% confidence leveL 'Mean and standard deviation (pg/g) for
triplicate samples. ' Instrumental failure.
1000
§ 100
=L
t
10 100 1000 10.000
PT/GC/MS Oig'g)
Hgw# 5. loj-tog plot of mean trtcNoroeHiytene (TCEI and
totuene (T°Q coocantrattorts In lortffiod soils.
The method comparison showed similar trends with the
low-level samples; no consistent difference was found for
all analytea in the USATHAMA soil or for Ben and TDCE
in the Point Barrow soil, but differences were observed for
TCE and Tol in that soil Here the difference can be
attributed to the phyaicai process of extracting the VOCs,
showing that a dynamic process is more efficient than a
static one for mass transfer, since both methods use water.
This effect can be observed by spiking an organic matter
rich soil water slurry. VOCs, particularly those with higher
K0/w will be sorbed to a greater extent by the organic
matter, showing reduced static vapor-phase concentrations
when compared to an aqueous solution or to an aqueous
low organic matter aoil slurry (27).
Field Samples. Method comparison using the two
field-contaminated soils was performed for TCE on nine
subsample sets (Table VI). Both methods of analysis
showed no false negatives; however, different aualyte
variances for the soils were apparent The Large variations
for the CRREL soil demonstrates the problem of inter-
method comparisons using Held soils which are heteroge-
neous. No significant differences were found between die
two methods for CRREL soil because of the very poor
precision for both methods (Table VI). For the Clarkson
soil, however, good sample precision produced statistically
different means in every case with the mean concentrations
obtained by aqueous-HS/GC always less than the
MeOH-PT/GC/MS values. Here the difference is not as
likely due to the organic matter present in the CJarkson
soil (Table II) as to slow desorption kinetics (21, 28,29).
To test this hypothesis, a simple experiment was per-
formed on the third Clarkson soil subset After the initial
headspace analysis, two of the soil subsampJes were reei-
tracted twice with water. The soil and water phases were
separated by centrifuging the suspension at 2300 rpm for
10 min. Only 28 of the 30 mL of water was recovered and
replaced. Results were corrected for this small carryover.
As previously, analyses for this soil were performed after
10 min of sample agitation (immediate, additional agitation
showed no discernable increase in the headspace concen-
tration)." The third replicate never had the partition so-
lution changed, but experienced all of the physical agita-
tion received by the other two subsamples. A 6-h time
period lapsed between the initial and final analysis for all
of these subsamples. The results of this test are presented
in Table VH. Clearly, equilibrium had not been achieved
after 10 min of agitation and the low results are due to slow
t*M E.-r,rm SO. TschncL. K- 2C No. 0. ~BS2
-------
Table VI. Inter met hod Comparison for Trichloroethylene
(TCE) in Field-Contaminated Soil
TCE concns (pg/g)
HS/GC PT/GC/MS
High-Level Comparison: CREEL Soil
Subsample Set 1
18.3. 11.4, 6.47, 3.60, 10.7 83.5," 3.31, 28.7, 4.33, 2.51
10.1 ± 5.58® 9-71 ± 12.r
Subsample Set 2
14.3, 9.00, 4.86, 12.3, 16.8 4.40, 11.7, 36.0, 45.5, 36.9
11.4 * 4.66 26.9 ±17.8"
Subsample Set 3
4.42, 3.36, 2.87, 4.01,1.60 2.50, 1.23, 3.55, 80.7 * 3.69
3.26 ± 1 11 2.74 ±1.14=
Subsample Set 4
0.68. 1.14, 1.42, 1.57, 0.72 0.65, 0.70, 3.46," 0.70, 0.36
0.976 ±0.609 0.60 ± 0.1C
Subsample Set 5
1.42, 0.89, 13.8, 10.2, 4.39 0.44, 1.18, 2.07, 1.71, 7.83*
6.14 ± 5.66 1.35 ± 0.71r
High-Level Comparison: Clarkson Soil
3.63, 3.44, 4.17
3.81 ± 0.37
Subsample Set 1 (2 Days)
8.77.9.89, 11.5
10.0 ± 1.37
Subsample Set '2 (31 Days)
3.45, 3.54, 3.81 7.87, 7.71, 8.07
3.60 ±019 758 ±0.18
Subsample Set 3 <165 Days)
2.38, 3.16, 2.66 3.54, 4.36, 4.16
2.73 ± 0.40 4.02 ± 0.43
Low-Level Comparison: CRREL Soil
Sub&ies
0.172, 0.171, 0.132, 0.288, 0.133 0.188, 0.066, 0.261, 0.289, 0.274
0.179 ± 0.064 0.216 ± 0.092s
"Outlier as determined by Dixon's test (19) at the 95% confi-
dence level. 6Average and standard deviation (j*g/g). CHS/GC
and PT/GC/MS analyses were not statistically different at the
95% confidence leveL
Table VII. Concentrations (fig/g) of TCE in HS Samples
after Cumulative Agitation and/or Repeated Aqueous
Extraction of tbe Clarbioo Soil
extraction
subsample
1st
2nd
3rd
Wl"
2.38
1.24
0.62
W24
4.14
W3»
2.66
1.53
0.69
total
4.24
4.14
4.88
X 4.42 ± 0.40
"Analyzed after 10 min of sample agitation after each sequential
extraction. ' Analyzed after 30 min of cumulative sample agitation
performed in 10-min intervals over the course of 6 h.
desorption kinetics. In addition, comparison of the mean
of these three determinations (4.42 ± 0.40) with the mean
of the MeOH-PT/GC/MS (4.02 ± 0.43) shows na statis-
tical difference. This finding agrees with these earlier
studies {21, 27,28) and emphasizes that the extraction of
soil VOCs in some cases is sensitive to the degree of agi-
tation and length of equilibration.
Screening for VOCs in Soils. This evaluation of
VOCs in fortified and field soils probes beyond the ob-
jective of sample screening and reveals some of the limi-
tations and strengths of the aqueous-HS sample prepa-
ration and portable GC analysis when compared to a
laboratory-based purge-and-trap method. With regard to
screening, the water extraction HS sample preparation and
portable GC analysis technique always produced results
comparable to PT/GC/MS. The largest discrepancies in
concentrations for this intermethod comparison occurred
for a soil with unusually high organic carbon content and
for a soil that previously had demonstrated stow aqueous
desorption of VOCs {21}. Even in these two cases water
extraction HS/GC analysis provided concentration esti-
mates that were greater than 30% of those determined by
PT/GC/MS analysis following MeOH extraction.
Nonhomogeneity of VOCs in soils, as demonstrated by
the TCE levels in the CRREL soil, dictates that several
subs&mplea or composite samples be taken for proper site
assessment. Costs of performing PT/GC/MS analyses
may limit the number of samples collected for laboratory
analysis, thereby reducing the ability to assess analyte
variability at discrete locations. A simple procedure that
can be used on site, although providing somewhat less
accurate VOC concentrations, allows for more intensive
sampling, i.e., more representative evaluation of contam-
inant distribution. Analysis by.either aqueous-HS/GC or
PT/GC/MS may be of equal merit if individual subsam-
ples are taken for VOC concentrations in soils where spatial
variability exists (22).
Summary and Conclusions. Headspace sample
preparation and analysis with a portable gas chromato-
graph does not use hazardous chemicals and is ideally
suited for on-site screening. From its inception, this
procedure minimized sample handling and holding time,
thereby reducing the possibility of false negatives. In the
work reported here, no false negatives were obtained and
quantitative results comparable to laboratory PT/GC/MS
were obtained in all cases tested. Thus, once the VOCs
of concern have been identified, aqueous-HS/GC analysis
can fill a void between the quality of analysis necessary
for litigation purposes and preliminary soil gas monitoring.
Universal recognition of simple transportable site assess-
ment methods can expedite investigations and lower the
costs of soil VOC analysis.
Acknowledgments
We thank Dr. T. M. Spittler for providing the infor-
mation concerning headspace gas chromatography and
nondistruptive sample collection techniques, Dr. S. G.
Pavlostathis for providing a contaminated soil and in-
sightful comments, and Dr. C. L. Grant for critical review
of the text.
Registry No. TDCE, 156-60-5; TCE, 79-01-6; Ben, 71-43-2;
Tol, 108-68-3.
Literature Cited
(1) Plumb, R. H., Jr.; Pitchford, A. M. Presented at the Na-
tional Water Well Association/ American Petroleum In-
stitute Conference on Petroleum Hydrocarbons and Organic
Chemicals in Ground Water, Houston, TX, Nov 1985; pp
13-15.
(2) Zarrabi, K.; Cross-Smiecinski, A. J.; Star kg, T. Second
International Symposium, Field Screening Method a for
Hazardous Waste and Toxic Chemicals, 14, Las Vegas. NY,
Feb 1991; pp 235-252.
(3) Chiou, C. T.; Shoup, T. D. Enuiron. Sci. Technol. 1985,19,
1196.
(4) Chiou, C. T. Theoretical considerations of the partition
uptake of nonionic organic compounds by soil organic
matter. In Reactions and Movement of Organic Chemicals
in Soils', Sawhney, B. L., Brown, K., Eds. SSSA Spec. Pubt.
1989, No. 22, 1.
(5) Test Methods for Evaluating Solid Waste: 1986; U.S.
Environmental Protection Agency, U.S. Department of
Environ. Scl. Technol.. Vol. 26, No. 10, 1992 1937
-------
Commerce, National TecJinir-al In ferai s.t ion. ServirK
Washington, DC, 1986, Vol. IB.
(6) Spittler.T, M.; Clifford, W. S.; Fitch, L. G. National Water
Well Association Meeting, Nov, Denver, CO, 1985; pp
236-246.
<") Kiang, P. H.; Gi-ob, R. L. J. Environ. Set¦ Health 1986, 21,
71
(8) Griffith, T. J.; Robfains, G. A.; Spittier T, M. Proceedings,
FOCUS Conference on Eastern Regional Water Issues,
National Waler WeQ AsaoriatioE, Sept, Stamford, CT, 1988;
p-p 223-248.
<95 Bobbins, G. A.; Roe, V. D.; Stuart, J- D.; Griffiih, T. J.
Proceedings, KWWA/API Conference oil Petroleum Hy-
drocarbons and Organic Chemicals in Ground Water-Pre-
vention Detection and Restoration, Nov, Houston, TX,
1987; pp 307-315.
(10> Stuart, J. D.; Wang, S.; Robbina,G. A. Second Intemauonal
Symposium, Field Screening Methods for Hazardous Waste
and Toiie Chemicals, Feb, Las Vegas, NV, 1991; pp
407—414.
UU Dietz, E. A., Jr.; Singiey K. F. Artal. Chern. 1979,5 J, 1609.
(12) Hewitt, A, D.; Miyarea, P. H.; Leggett, D. C.; Jenkins, T.
F, SRS1-4;USA Cold Regions Ss-seaich and Engineering
Laboratory, HajKrer. NH, 1991.
(13) Lewis. T. E.; CrocSett, A. B.; Siegriat, R. L.; Zarrabi, X.
EPA/590/4-91/001; Technology Innovation Office, Office
of Solid Waste and Emergency Response, U.S. EPA,
Washington, DC, 1991.
(14) Jenkins, T. F.; Schumacher, P. W. SRS7-22; USA Cold
Regions Research arid Engineering Laboratory, .Hanover,
NH, 1987.
(15'i Verschueresi, K. Handbook of Environmental Data on
Organic Chemicals', Van Noatrand Reinhold: New York,
1383.
IX"I McGovern. E. W. ind. Eng. Cherrt. 1343. 3£. 1230.
till HanBch, L.; Leo, A-Sufrslitueni Constants for Correiattort
' Analysis in Chemistry and Biology, ElBevier Amsterdam,
1979.
(13) McDuiJie, B. Chemasphere 1981, JO, 73.
(191 Hine, J.; Moofcerjee, P. K. J. Org. Ckem. 1975, 40, 292.
(201 Roberto,P. V.; Dandliker, P. G. Environ. Sci. TechnoL 1983,
17,484.
(21) Pavlostathia, S. G.; Jaglal, K. Environ. Sci. Techno!, 1991,
25, 274.
(22) Spittier, T. M., personal communication, U5. Environ-
mental Protection Agency, Environmental Servicea Divi-
sion-Region 1, Lexington. MA, 1983,
<231 Diton, W. J. Biometrics 1953, March, 74.
i'24) Kecrickboff, S. W.; 3to»ti, D. S.; Scott, T. A. Water f?p?.
1979,13, 241.
(25) Chiou, C. T.- Porter, P. E.; Schmedding, D. W. Environ.
Sci. TechnoL 19S3, IT, 227.
(26) Boyd, S. A.; Sun, S. Environ. Sci- Tecftnoi. 1990,24,14E.
{'27) Hewitt, A. D.; Miyares, P. H.; Leggett, D. C.; Jenkins, T.
F. CR92-6; U.S. Cold Regions Research and Engineering
Laboratory, Hanover, NH, 1992.
(26) Smith, J. A.; Chiou, C. T_; Kammer. J. A.; Kile, D. E.
Eruiircm.. Sci. "TechnoL 1990 34, 67&.
[2S( Sawhuey, E. L., feat, M. P. "N. Ciayj Clay Miner. 19JB,
38, 14.
Received for review January 27,1992. Revised manuacript re•
ceiuedMay 13, 1992. Accepted June 16, 1992. Funding for this
merit cifzs provided by the U.S Army Toxic and Hazardous
Materials Agency, Duranl Graves, Project Monitor. This pub-
lication reflects the personal views of the authors and does not
suggest or reflect policy, practices, programs, or doctrine of the
U.S. Army or Government of the United States.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 18
Experimental Determination of Pre-analytical Holding Times For
Volatile Organics in Selected Soils
Roger A. Jenkins, Charles K. Bayne, Michael P. Maskarinec, Lynn H.
Johnson, Susan K. Holladay, and Bruce A. Tomkins
Analytical Chemistry Division, Computing Applications Division, Oak
Ridge National Laboratory, Oak Ridge, Tennessee
January 12-14, 1993
Las Vegas, Nevada
-------
BIOGRAPHY
Jim Pollard received his B.S. in marine biology and M.S. in fresh water ecology from
California State Polytechnic University. He was a research ecologist at the University
of Nevada, Las Vegas for 12 years. For five years he has been a principal
scientist/scientific supervisor at Lockheed Environmental Sciences and Technologies
Division. His primary areas of expertise are biological methods development, quality
assurance research, fate and transport modeling, and ecological assessment research.
E. Neal Amick received his B.S. (1973) in organic chemistry from Kansas State
College of Pittsburg. He has more than 18 years experience in instrumental analysis,
primarily in chromatography of trace organic environmental pollutants. He is
currently a principal scientist at Lockheed Environmental Systems and Technologies
Company. He has been involved in field methods development and evaluation since
1988.
-------
EXPERIMENTAL DETERMINATION OF PRE-ANALYTICAL HOLDING
TIMES FOR VOLATILE ORGAN ICS IN SELECTED SOILS
Roger A. Jenkins, Charles K. Bayne, Michael P. Maskarinec,
Lynn H. Johnson, Susan K. Holladay, and Bruce A- Tomkins
Analytical Chemistry Division
Computing Applications Division
Oak Ridge National Laboratory
Oak Ridge, Tennessee 37831-6120
Regulatory pre-analytical holding times (HTs) for volatile organic compounds (VOCs)
in soil stored at 4° C are 14 days. An increasing body of anecdotal evidence has
suggested that the actual stability of VOCs stored under such conditions and
processed and analyzed according to Contract Laboratory Program (CLP) procedures
may be considerably different. The purpose of this study was to experimentally
determine HTs for a variety of VOCs, soil types, and storage conditions. The
experimental factors and their levels are presented in Table 1.
Table 1. Experimental factors for the holding time study of VOCs in soils.
Factors
Factor Levels
VOC
19 Organic Compounds
Soil Type
USATHAMA, Tennessee, Mississippi
Storage Condition
-70°C, -20°C, 4°C
Storage Time (days)
0, 3, 7, 14, 28. 56, 111
The 19 VOCs used in this study were: bromomethane; chloroethane; methylene
chloride; 1,1-dichloroethene; 1,1-dichloroethane; chloroform; carbon tetrachloride;
1,2-dichIoropropane; trichloroethene; benzene; 1,1,2-trichloroethane; bromoform;
1,1,2,2-tetrachloroethane; tetrachloroethene; toluene; chlorobenzene; ethylbenzene;
styrene; and o-xylene. The three soil types used for this study were a U.S: Army
Toxic and Hazardous Materials Agency soil (USATHAMA), a Captina silt loam from
Roane County, Tennessee (Tennessee), and a McLaurin sandy loam from Stone
County, Mississippi (Mississippi). The Tennessee and USATHAMA soils have high
clay and silt content, while the Mississippi has a high sand content.
Soil samples were prepared by weighing 5 g aliquots of soil into 40 mL borosilicate
-------
glass vials with aluminum foil coated teflon faced silicone sepia and screw caps with
holes purchased from Shamrock Glass Company. These vials were received fully
assembled and precleaned according to EPA 40 CFR 136 and EPA 40 CFR 141
regulations. Three days prior to spiking with VOC stock solution, the soil samples
were wetted with 1.25 mL of reagent grade water (Burdick & Jackson) and agitated
with a vortex mixer for 30 s. The soil samples were then stored in the dark at room
temperature. This preparation step allowed bacterial growth to come to a steady
state.
In order to spike the activated soils, water was first introduced into a 1-liter Tedlar
gas sampling bag and allowed to degas for three days. This gas was then removed.
On the day that soil samples were to be spiked, excess air was removed from the
water filled TedJar bag. Thirty soil samples were spiked with 1.25 mL of water and
vortexed for about 30 seconds. These son samples served as quality control blanks.
Quality control blanks and soil samples were stored together in order to assess the
possibility of cross contamination.
Then, a VOC stock solution for each soil type was prepared by dispensing the 19
VCX? standards into the Tedlar bag with the remaining wa^er. Appropriate volumes
of each VOC standard solution were introduced through the septum port using gas
tight syringes. The contents of the Tedlar bag were mixed thoroughly by hand
agitation for three minutes, after which the bags were allowed to sit for thirty
minutes. The VOC stock solution was then aliquotted into the 40 mL vials containing
the test soil samples by gravity flow using 0.20-0.25 mL aqueous volatiles/g soil. At
this point, the soil samples were SO-100% saturated with water. Each soil sample viai
was sealed immediately with the Teflon-backed aluminum foil faced septum and
screw cap with hole and agitated with a vortex mixer for 30 seconds. Soil samples
were then analyzed or stored at the appropriate temperature ( -70'C, -20*C and
4-C).
For analysis, soil samples were purged directly from the 40 mL vials into the trapping
system directly. To accomplish this, a double threaded Teflon coupling machined in-
house with o-ring seals was used to replace the septum cap and septum. This was
done to allow an assessment of the actual content of the vials without the
complicating factor of sample weighing and transfer. The brief period of cap
changing was the only time from spiking to analysis in which the vial contents were
exposed to the ambient air. AH volatile organic analyses were performed by gas
chromatography with mass spectrometric detection (GC/MS) according to standard
EPA Contract Laboratory Program (CLP) methods, except for the use of daily
external standards (instead of internal -standards) to calculate results. Data were used
without recovery or blank correction, as is customary with this method.
-------
The total number of chemical analyses used to determine maximum holding times
were 3,382 analyses. Although 3,382 chemical analyses were performed, about 30%
(i.e., 1,016 analyses) of the data points were considered outliers and not used to
estimate the maximum holding time values. Potential outiiers were first identified by
comparing the changes in the standard deviations of neighboring time points for each
matrix type and storage condition. Additional potential outliers were also identified
by their large (e.g., > 2.5) studentized residuals for the zero-order and first-order
regressions of concentrations vs storage times. An identified outlier value was
marked in the data set not to.be used for estimating maximum holding times after
reexamining the raw data from the corresponding GC/MS chemical analysis.
Chemical judgement for marking an identified outlier was based on (1) an analysis
that resulted in an unusually low or high concentration due to contaminant peak
interference of poor separated peaks, or (2) an analysis corresponded to an incorrect
analysis of a reference standard, or (3) an analysis that had been compromised by
procedural problems (e.g., incorrect spiking concentration, instrument problems,
sample bottles not properly filled, data entry errors, etc.). A potential outlier found
by the statistical procedure was not necessarily set aside until after considering the
chemical analysis.
The results of this study were estimated maximum holding times (MHTs), which are
the maximum times a sample can be held prior to analysis before sufficient
degradation occurs to render the sample unreliable. It is the establishment of criteria
which defines "sufficient degradation" that can be a subjective process. Two statistical
definitions were used to determine MHT criteria. The first definition was specified
by the American Society for Testing and Materials, The second definition was
specified by Environmental Science and Engineering, Inc. for a holding time study
conducted in cooperation with U.S. Environmental Protection Agency. Both
definitions are based on an approximating model for predicting concentration with
time. The ASTM defines the MHT as the time the predicted concentration falls
below the lower two-sided 99% confidence interval on the initial concentration. The
ESE defines the MHT as the time the one-sided 90% confidence interval on the
predicted concentration falls below a 10% change in the initial concentration. The
main difference between the two definitions is the method of placing a lower bound
on the initial concentration.
Five approximating models were used to discribe the time course of the analyte
concentration in the test matrix: zero order, first order, log term, inverse term, and
cubic spline. The concentration/time data was fitted to the model which provided the
best fit, and the MHTs were calculated, using the criteria described above.
-------
In Tables 2 and 3 are presented the estimated MHTs determined using the two sets
of criteria. Maximum holding times depend on the VOC, storage temperature, and
soil matrix. Most MHTs results for -20 °C storage temperature have longer or
equivalent (e.g., ±2 days) MHTs than either -70 "C or 4°C storage temperatures.
Exceptions to this general rule are for bTomoform in USATHAMA soil stored at -
70" C; methylene chloride, 1,1,2-trichloroethane (i.e., ASTM MHT only), styrene, and
o-xylene in Mississippi soil stored at -70°C. A storage temperature of 4°C is
inadequate for all 19 VOCs in Tennessee and Mississippi soils -- except perhaps
methylene chloride (ASTM MHT = 14 days) in Tennessee soil. For bromomethane,
none of the three storage temperature are adequate for Tennessee and Mississippi
soils.
Table 2. ASTM MHTs in days for volatile organic compounds in three soils.
Volatile Organic
USATHAMA1
Tennessee2
Mississippi2
Nutn
Compounds
-70°C
-20°C.
-70°C
-20°C
4°C
-70°C -20°C
4°C
L
Bromomethane
2
108
0
1
0
0
3
0
2
Chloroethane
4
24
0
23
7
0
3
0
3
Methylene Chloride
24
lit
56
56
14
56
0
2
4
1,1-Dichloroethene
2
111
0
18
4
0
3
0
5
1,1-Dichloroethane
2
111
0
20
6
0
2
1
6
Chloroform
2
111
0
47
2
0
2
2
7
Carbon Tetrachloride
2
111
0
18
8
0
3
0
8
1,2-Dichloropropane
3
111
0
56
6
2
2
2
9
Trichloroethene
2
65
0
46
0
0
3
0
10
Benzene
2
81
0
28
0
0
3
2
11
1,1,2-Trichloroeihane
42
111
56
56
6
16
5
2
12
Bromoform
111
69
24
40
6
56
56
2
13
1,1,2,2-Tetrachloroethane
75
111
26
27
5
56
56
4
14
Tetrachloroethane
2
8
0
22
0
0
3
2
15
Toluene
2
76
1
44
0
0
16
15
16
Chlorobenzene
4
98
56
56
0
5
3
2
17
Ethylbenzene
8
28
0
37
2
3
4
2
18
Styrene
63
111
56
56
1
56
26
3
19
o-Xylene
2
47
56
56
2
19
11
2
1. Maximum experimental time was 111 days.
2. Maximum experimental time was 56 days.
-------
Table 3. ESE MHTs in days for volatile organic compounds in three soils.
Volatile Organic
USATHAMA1
Tennessee2
Mississippi2
Num
Compounds
-70°C
-20°C
-70°C
-20°C
4°C
-70°C -20°C 4°C
1
Bromomethane
1
111
1
1
1
1
2 1
2
Chloroethane
2
18
1
4
3
L
2 1
3
Methylene Chloride
21
111
56
56
1
56
2 1
4
1,1-Dichloroethene
1
111
1
3
2
1
2 1
5
1,1-Dichloroethane
1
111
1
4
3
1
1 1
6
Chloroform
1
111
1
25
1
1
1 1
7
Carbon Tetrachloride
1
111
1
3
3
1
2 1
8
1,2-Dichloropropane
2
111
1
43
3
1
1 1
9
Trichloroethene
1
52
1
39
1
1
2 1
10
Benzene
1
68
1
5
1
1
1 1
11
1,1,2-Trichloroethane
40
111
56
56
3
4
3 1
12
Bromoform
111
56
10
9
3
56
56 1
13
1,1,2,2-Tetrachloroethane
111
111
12
5
4
56
56 1
14
Tetrachloroethane
1
4
1
4
1
1
2 1
15
Toluene
2
46
1
41
1
3
2 2
16
Chlorobenzene
3
111
56
56
1
1
2 1
17
Ethylbenzene
3
16
1
8
1
1
1 1
18
Styrene
111
111
56
56
1
56
5 1
19
o-Xylene
2
62
56
56
1
13
2 1
1. Maximum experimental time was 111 days.
2. Maximum experimental time was 56 days.
Integrating the results of this study into actual field operations is not straightforward.
The fact that the pre-sterilized USATHAMA soil exhibited reasonably long MHT's
at one storage condition suggests that microbial action is extremely important to
contaminant stability in environmental samples. Of course, it is impractical to
sterilize real samples without loss of volatiles. The highly sandy soil (Mississippi) had
the shortest MHT's. For the two soils with higher organic, silt, and clay content, the
best overall storage conditions appear to be -20° C. The typical storage condition
of 4° C, commonly used in the United States, appears to be of minimal utility for
maintaining contaminant levels. A storage condition of -70 • C appears to be too low,
although there was insufficient data generated in this study to provide a definitive
-------
reason. Producing a -20° C condition under field conditions would probably most
easily be achieved using a salt brine and ice mixture. Maintenance of such a mixture
would seem more problematic during shipment. However, the data shows that -200
C is not the best storage condition for all contaminants and all soils. It would appear
that selection of storage condition will be dependent on soil and contaminant type.
These results add to the increasing body of evidence which indicates that current
sampling handling and storage practices result in unacceptably high losses of VOC's
prior to analysis. There is a growing sense that the sample must be acquired in a
manner which disturbs the soil to a minimum extent, that the sample must be
transferred to a container which is immediately sealed, and that the container must
not be opened or penetrated until after the analysis is underway. Recently, a team
of scientists assembled by the EPA has reported on apparati and methods for doing
just that. There is now commercially available sub-coring devices which permit the
removal of a small soil plug from a larger core, and deposition of the plug in a 40 mL
VOA vial, as well as a specialized cap for such vials, which permits the vial to remain
tightly sealed until a small Teflon ball seal is pushed out of the way as the vail is
loaded onto a purge and trap device. Alternatively, the sample can be immersed in
methanol. A recent study has shown that approach to be very promising, in terms
of minimizing losses of VOC's. However, analysis of an aliquot of the methanol
extract via purge and trap GC/MS has the advantage of inherently higher limits of
detection.
Research sponsored jointly by the U. S. Army Toxic and Hazardous Materials Agency,
DOE IAG No. 1769-1743-Al, the U. S. Environmental Protection Agency and the U. S. Air
Force, DOE IAG No. 1824-1744-A1, and the U. S. Navy, DOE IAG No. 1743-1743-A1,
under U. S. Department of Energy Contract DE-AC05-840R21400 with Martin Marietta
Energy Systems Inc.
-------
REFERENCES
Worthy, W. Chem. Eng. News. 1987, September 7, 33-40.
U. S. Environmental Protection Agency. Quality Assurance Newsletter, 1984, 6(3).
U. S. Army Toxic and Hazardous Materials Agency Report to U. S. Army Armament
Research and Development Command fUSAARRADCOM") Product Assurance
Directorate. University of Maryland College of Agriculture; College Park, MD, 1981.
B. T. Walton, T, A. Anderson, M. S. Hendricks, and S. S. Talmage, "Physicochemical
Properties as Predictors of Organic Chemical Effects on Soil Microbial Respiration,"
Environmental Toxicology and Chemistry. Vol. 8, pp. 53-63, 1989.
B. T. Walton and T. A. Anderson, "Structural Properties of Organic Chemicals as
Predictors of Biodegradatiori and Microbial Toxicity in Soils," Chemosphere. Vol.17,
No. 8, pp 1501-1507, 1988.
ASTM, 1991 Annual Book of ASTM Standards. Vol. 11.02 Water (II), pp 3-7,
ASTM, Philadelphia, Pa., 1991.
H. S. Prentice and D. F. Bender, Project Summary: Development of Preservation
Techniques and Establishment of Maximum Holding Times: Inorganic Constituents
of the National.Pollutant Discharge Elimination System and Safe Drinking Water Act.
Research and Development, EPA/600/S4-86/043, March 1987.
T. E. Lewis, A. B. Crockett, R. L. Siegrist, and K. Zarrabi, GrOund-Water Issue: Soil
Sampling and Analysis for Volatile Organic Compounds. U.S. EPA, EMSL-LV, Las
Vegas, Nevada, EPA/540/4-91/001, February 1991.
R. L. Siegrist, and P. D. Jenssen, "Evaluation of Sampling Method Effects on Volatile
Organic Compound Measurements in Contaminated Soils," Environ. Sci. Technol..
24, 1387-1392 (1990).
M. P. Maskarinec, C. K. Bayne, L. H. Johnson, S. K. Holladay, and R. A. Jenkins,
"Stability of Volatile Organic Compounds in Environmental Water Samples: Storage
and Preservation," Oak Ridge National Laboratory, ORNL/TM-11300, 1989.
-------
Biography
Roger A. Jenkins has a Ph.D. in Analytical Chemistry from the University of
Wisconsin - Madison. He is currently Group Leader of Special Projects, in the
Analytical Chemistry Division of Oak Ridge National Laboratory. He has had more
than 17 years experience in the research areas of the generation, sampling, and
chemical characterization of complex atmospheres, and field and environmental
analytical chemical methods development. He has been on a number of advisory
boards for the US Departments of Energy, Defense, and Health and Human
Services, as well as Health and Welfare/Canada.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 19
Evaluation of Sample Holding Times and Preservation Methods for
Gasoline in Fine-Grained Sand
Paul H. King
P&D Environmental, Oakland, California
January 12-14, 1993
Las Vegas, Nevada
-------
EVALUATION OF SAMPLE HOLDING TIMES
AND PRESERVATION METHODS
FOR GASOLINE IN FINE-GRAINED SAND
Paul H. King
P&D ENVIRONMENTAL
300 Monte Vista, #101
Oakland, CA 94611
An estimated 2 to 5 million underground storage tanks containing petroleum hydrocarbons
or chemicals are believed to be present in the United States. Over 100,000 spills or leaks
have been reported for petroleum hydrocarbon underground tank systems, and many more
reports of spills or leaks are expected within the next several years. To evaluate if a tank
has leaked, soil samples are collected during tank removal from beneath the tank. The
EPA has set forth a maximum holding time for such samples of 14 days when preserved
at four degrees Celsius (4°C), after which time the samples are no longer considered
valid. Furthermore, some California local regulatory agencies require that soil samples
be preserved with dry ice at the time of sample collection. The EPA arid local regulatory
agencies are unable to present any studies for soil samples which demonstrate the validity
of a 14 day holding time or the preservation of soil samples at 4°C or with dry ice as
providing adequate sample quality preservation.
To evaluate the 14 day holding time and the different prescribed preservation methods,
a mixing technique was developed for the homogenization of fine-grained sand, water and
gasoline. The resulting mixture had a gasoline concentration of approximately 100 to 200
parts per million (pp.m). After mixing, the sand-water-gasoline mixture was placed into
30 brass tubes. The mixing process was repeated so that a total of 5 mixes were
generated. The brass tubes for a given mix were preserved with a specific preservation
method. Mix A was preserved in brass tubes at room temperature, Mix B was preserved
in brass tubes at 4°C, Mix C was preserved in brass tubes in a cooler with dry ice, Mix
D was preserved in 40-milliliter (mL) volatile organic analysis containers (VOAs) with
5 mL of methanol at room temperature, and Mix E was preserved in VOAs with 5 mL
of methanol and refrigeration at 4°C. For each mix, six randomly selected samples- were
evaluated 0, 3, 6, 10, and 14 days after sample mixing to determine if the samples for a
given preservation method showed signs of deterioration with respect to time.
The samples were analyzed using a purge-and-trap (EPA Method .5030) and a packed
column gas chromatograph equipped with a flame ionization detector (FID) and recording
integrator (Modified EPA Method 8015). In addition to standard EPA quality assurance
procedures, a surrogate spike of bTomobenzene was analyzed with each sample as an
internal standard to evaluate sample recovery. Surrogate spike recovery tolerance limits
of -+/- 20 percent were used to determine sample validity Samples with surrogate spike
-------
Paul H. King
recoveries which exceeded the tolerance limits were reanalyzed until acceptable surrogate
spike recoveries were achieved.
The sample results showed that in Mix A (brass tubes with no refrigeration), gasoline
concentrations rapidly decreased from 91 ppm to less than 20 ppm within 6 days (Figure
1), and that in Mix B (brass tubes with refrigeration, in accordance with EPA protocol)
gasoline concentrations decreased from 120 ppm to 34 ppm withm 6 days (Figure 2).
The sample results for Mix C (brass tubes with dry ice. Figure 3), Mix D (VOAs with
5 mL methanol and no refrigeration, Figure 4), and Mix E (VOAs with 5 mL methanol
. and refrigeration at 4°C, Figure 5) did not show any sample deterioration exceeding the
precision of the analytical methods for the experiment during a 14 day holding time.
The results of this investigation indicate that conventional soil sample preservalion
techniques and holding times as set forth by the EPA for volatile organic compounds
(VOCs) are inadequate for maintaining sample quality integrity for samples of gasoline
in fine-grained sand. Sample preservation methods using dry ice or immediate extraction
into methanol are shown to provide adequate maintenance of sample quality integrity with
demonstrated holding times of up to 14 days for gasoline in Fine-grained sand. The data
trends of this investigation suggest that holding times longer than 14 days may be
acceptable with these alternate preservation methods.
The suggested application of these preservation methods consists of soil sample
preservation with dry ice, followed by sample extraction at the laboratory immedialely
after samples have thawed. Evaluation of the thawing time for a brass tube preserved
with dry ice and containing fine-grained sand, approximately 100 ppm gasoline and 10
percent water by weight showed that approximately 45 rmnuies were required for the
sample to thaw to above 0°C when placed on a counter top with an ambient air
temperature of approximately 20°C
-------
MIX A - NO REFRIGERATION
GASOLINE CONCENTRATION VS. TIME
CL
(X
O)
e
c
o
"to
w.
c
-------
MIX B - REFRIGERATION AT 4 C
GASOLINE CONCENTRATION VS. TIME
360 J
_ 340t
^ 320
£ 300]
ct 280j
^ 260 J
£ 240 H
c 220-
I 200-
£ 180;
¦ § 160-)
c 140-
6 120-
® 100:
=5 80=
8 . 601
O 40]
20 H
04—
-24
j DAYQ
*
*
DAY
li
¦«-
0AV6
DAY "SO
DAY 14
*
-¦*-
m
W-'
ir
24 72 120 166 216 264
HOURS
312 360
LEGEND | * SAMPLE RESULT -A- BATCH MEDIAN
FIGURE 2
-------
MIX C - DRY ICE
GASOLINE CONCENTRATION VS. TIME
360J
-
340 H
3201
cu
Q_
300J
rn
280 H
O)
£
260-|
240-1
c
2201
o
"5
200-
180-}
c
-------
MIX D - METHANOL - NO REFRIGERATION
GASOLINE CONCENTRATION VS. TIME
2
GL
QL.
U)
Ji
Ol
E
c
o
"3
u
c
cu
u
c
o
O
0)
_E
o
-------
CL
a
Di
-s:
E
c
.o
oa
c
c
o
O
a>
_c
o
w
co
O
~AY 0
360 j
340 H
320]
300 q
280 H
2601
240-
220-
200-
180-
160l
140-
120-
100-
80
60
40-
20 ;
0
-24
MIX E - METHANOL - REFRIGERATED
GASOLINE CONCENTRAION VS. TIME
DAY 3
*
DAY 6
DAY 10
DAY 14
*
*
*
M
1-
—•—m-—
* m
\
\
&
3K
m
*
24 ' 72 120 168 216 264 312 360
HOURS
LEGEND * SAMPLE RESULT BATCH MEDIAN
FIGURE 5
-------
LIST OF REFERENCES
American Institute of Chemical Engineers. 1961. Equipment Testing Procedure For
Solids Mixing Equipment. New York.
American Institute of Chemical Engineers. 1964. Equipment Testing Procedure For Paste
and Dough Mixing Equipment. New York.
Bone, L.I. 1988. Preservation Techniques for Samples of Solids, Sludges, and
Nonaqueous Liquids. In Principles of Environmental Sampling, ed. L.H. Keith, pp. 409-
413. American Chemical Society.
Bottrell, D.W. et. al. 1990. Holding Times: VOCs in Water Samples. Environmental
Lab, June/July: 29-31.
Chiou, C. T,, and Shoup, T. D. 1985. Soil Sorption of Organic Vapors and Effects of
Humidity on Sorptive Mechanism and Capacity. Environmental Science and Technology.
Vol.19, No.12, 1196-1200.
U. S. Environmental Protection Agency. 1986a. Underground Tank Leak Detection
Methods: A State-of-the-Art Review. EPA/600/2-86/001.
U. S. Environmental Protection Agency. 1986b. Test Methods for Evaluating Solid Waste
- Physical/Chemical Methods. SW-846.
U. S. Environmental Protection Agency. 1987a. Causes of Release From UST Systems:
Final Report to U.S. EPA/OUST. EPA Contract 68-01-7053, Subcontract 939-5, Work
Assignment 24. September 30, 1987.
U. S. Environmental Protection Agency. 1987b. Underground Storage Tank Corrective
Action Technologies. EPA/625/6-87/015.
U. S. Environmental Protection Agency. 1990. Musts for USTs. EPA/530/UST-88/008.
U. S. Environmental Protection Agency. 1992. Report to the Senate Committee on
Appropriations Regarding Underground Storage Tank Financial Responsibility and Related
Issues. EPA Office of Underground Storage Tanks. April, 1992.
Fairless, B.J., and Bates, D I. 1989 Estimating the Quality of Environmental Data.
Pollution Engineering. March.108-111.
Havlicek, S C. 1988. Characterizations of Fuels and Fuel Spills. Central Coast Analytical
Services, San Luis Obispo, California. August, 1988.
-------
Kreamer, D.K. et al. 1990. Development of a Standard, Pure-Compound Base Gasoline
Mixture for Use as a Reference in Field and Laboratory Experiments. Groundwater
Monitoring Review. Spring: 135-145
Lewis, T E., Crockett, A. B., and Zarrabi, K. 1991. Manual For Sampling Soils For
Analysis of Volatile Organic Compounds - Project Summary (DRAFT). EPA/600/X-
91/XXX.
Lewis et al 1993 Soil Sampling and Analysis for Volatile Organic Compounds.
EPA/540/4-91/001.
LUFT Task Force. 1988. Leaking Underground Fuel Tank Manual: Guidelines for Site
Assessment, Cleanup, and Underground Storage Tank Closure. May, 1988.
Maskarinec, M. P., Johnson, L. H, Holladay, S. K. 1988. Preanalytical Holding Times.
Paper presented at the Quality Assurance in Environmental Measurements Meeting, U:
S. Army Toxic and Hazardous Materials Agency, Baltimore, MD, May 25-26, 1988.
Maskarinec, £. P. and Moody, R. L. 1988. Storage and Preservation of Environmental
Samples In Principles of Environmental Sampling, ed. L.H. Keith, pp. 145-155.
American Chemical Society.
Siegrist, R.L. and Jennssen, P. D. 1990. Evaluation of Sampling Method Effects on
Volatile Organic Compound Measurements in Contaminated Soils. Environmental
Science & Technology 24:9, 1387-1392.
Siegrist, R. L. 1991. Measurement Error Potential and Control When Quantifying
Volatile Hydrocarbon Concentrations in Soils. In Hydrocarbon Contaminated Soils, ed.
E J. Calabrese and P.T. Kostecki, pp. 205-215. Lewis Publishers, Inc.
Slater, J. P. et al. 1983. Sampling and Analysis of Soil For Volatile Organic Compounds:
Methodology Development. In Symposium on "Characterization and Monitoring of the
Vadose (Unsaturated) Zone" December 8-10, 1983. National Water Weil Association.
Taylor, J.K. 1988. Defining the Accuracy, Precision, and Confidence Limits of Sample
Data. In Principles of Environmental Sampling, ed. L.H Keith, pp. 101-108. American
Chemical Society.
Tri-Regional Guidelines. 1990. Tri-Regional Board Staff Recommendations for
Preliminary Evaluation and Investigation of Underground Tank Sites. August 10, 1990.
Washall, J. W., and Wampler, T. P. 1990. Sources of Error in Purge-and-Trap Analysis
of Volatile Organic Compounds. American Laboratory. December38-44.
-------
Paul H. King
P&D ENVIRONMENTAL
300 Monte Vista, #101
Oakland, CA 94611
(510) 658-6916
Mr. King received a Bachelor of Arts degree in Geology from the State University of
New York at Binghamton in 1984 and a Master of Science degree in Environmental
Management from the University of San Francisco, California in 1992. He has performed
environmental investigations as a hydrogeologist in California since 1985. He presently
owns his own consulting firm (P&D Environmental) which is located in Oakland,
California. His professional interests include contaminant fate and transport, computer
modelling, vapor extraction remediation of VOCs and quality assurance issues related to
soil and groundwater sample collection and preservation.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 20
Development of an ASTM Standard For Sampling Soils and VOCs
Elly K. Triegel
Triegel & Associates, Inc., Berwyn, Pennsylvania
January 12-14, 1993
Las Vegas, Nevada
-------
DEVELOPMENT OF AN ASTM STANDARD FOR SAMPLING SOILS FOR VOCS
Elly K. Triegel
Triegel & Associates, Inc., Berwyn, PA
The American Society of Testing and Materials (ASTM) has
recently approved the "Standard Practice for Sampling Waste
and Soils for Volatile Organics" (D 4547-91). This practice
is intended to describe the field sampling procedures for
solid materials (i.e., solid wastes, soils and sediments)
for subsequent volatile organic carbon (VOC) analysis. The
standard was developed by a task group under the Waste
Management (D 34) Committee of the Society, and was approved
by Society ballot in 1991.
ASTM standards are approved only after a systematic process
of balloting and revisions at the various levels in the
organization. This standard was initially developed by the
task group, with informal input from a number of individuals
dealing with the problem of sampling soils or other solids
for VOC analysis. An effort was made to include information
from academic researchers, governmental agencies, industry
and consultants. The balloting proceeded from- the section
level (Waste Sampling and Monitoring, D 34.01.02), through
subcommittee (Sampling and Monitoring, D 34.01), committee
(Waste Management, D 34) and, finally to Society-wide
ballot. In general, this order of balloting corresponds to
movement of the standard from a small group of individuals
with a specific interest in the field, to larger groups with
generalized interest.
In each stage of the ASTM balloting process, members of the
relevant group. comment on the standard and can vote
negatively on its adoption. When a negative is received, it
is generally either voted non-persuasive or persuasive. In
the latter case, the standard is appropriately revised and
re-submitted to the beginning of the balloting process.
Although this process can be time-consuming, it provides for
technical input at all stages of the review process. It also
provides input from individuals with related interests
(e.g., member of the Committee on Soil and Rock, D 18), who
may not be members of the committee issuing the standard.
This process is particularly useful if the procedure
involves expertise from a number of different disciplines,
as is frequently the case in environmental work. In the
case of D 4547-91, the draft standard was revised a number
of times to incorporate comments found to be persuasive and
non-editorial in content.
The standard was designed to be as flexible as possible,
recognizing that' regulatory requirements, field conditions,
-------
and the nature of the wastes may vary considerably. The
standard was confined to the field portion of the
activities, and did not directly address related issues,
such as laboratory procedures or experimental design.
Principal areas of concern included the loss of volatiles,
the collection of a representative sample, ability to
correlate the results with other properties, and analytical
considerations.
Two methods of sampling loose granular materials (e.g.,
soils) were included in the final standard: (1) the
collection of samples in metals rings, typically inserted in
split barrels or similar devices, and (2) subsampling in the
field using a metal coring cylinder. Provisions were also
included for the sampling of cemented, non-granular solids.
When samples are collected in metal rings, the exposed ends
and adjacent samples are used to observe the properties of
the soils, and the ends are then sealed quickly using an
inert material. Extrusion or subsampling of the soil is
done by the analytical laboratory under presumably more
controlled conditions. This method-has the disadvantages of
not allowing the field personnel to use soil from the same
stratagraphic horizon for logging or field
screening/testing. In addition, it is not a recommended
method under conditions of poor soil recovery, where
variations in headspace may affect analytical results.
The second method involves the use of a small metal coring
cylinder to remove a sample from test pits, the ground
surface, larger blocks of soil, or samples that have been
removed from down-hole sampling devices. Immediately before
this sampling, the soil surface should be trimmed to remove
contaminants from, other soil strata or to remove surface
layers that may have already lost volatiles. The coring
device consists of a cleaned metal cylinder, open at both
ends, which is driven into the solid material. This method
allows the field sampling personnel to inspect the
surrounding soil, log its properties, and perform field
tests on . the excess soil. This method should not be used
for certain materials, such as stiff clays, which are not
easily cored or extruded in the field.
Three methods are provided to handle the sample once it is
collected. The first' involves shipping the sample to the
analytical laboratory within the metal ring, where the
subsample is extruded for analysis. The other two methods
deal with samples which are extruded in the field; these
methods are designated in the standard as Method 1 and
Method 2.
-------
Method 1 involves the addition of the soil or solid waste to
a wide-mouth jar containing 100 mL of methanol; that jar is
then shipped to the laboratory for analysis of the volatiles
in the methanol. Research comparing various methods of
sampling soils for VOCs have found that the methanol method
produces significant reductions in the loss of VOCs,
compared to samples jarred without methanol (Siegrist. and
Jenssen, 1989 and 1990). The addition of the soil in the
field allows a longer contact time between the methanol and
the soil, which improves the efficiency of transfer of the
volatile compounds from the solid matrix to the methanol.
Extraction of volatiles is performed with a larger subsample
of soil than is used in some other methods (e.g., 100 g
versus 5 g), which may produce data which are more
indicative of the overall environmental impact of the
contaminated soil. Sample size can be critical in obtaining
representative and reproducible results, particularly from
heterogeneous soils (Bone, 1988). In addition, this method
allows compositing of soils for VOC analysis, which can be a
critical component in the design and cost of soil sampling
plans.
At the time that this standard was prepared, the principal
disadvantages of Method 1 were considered to be: (1) a lack
of commercial laboratories providing this type of sample
preparation and analysis, (2) shipping of the methanol, (3)
the high method detection limits, (4) concerns related to
the ability of the methanol to extract higher molecular
weight compounds (e.g., those associated with hydrocarbon
spills), and (5) regulatory considerations regarding
required analytical procedures. Since that time, analytical
detection limits have improved, and some states (e.g.,
Minnesota and Wisconsin) have reportedly adopted this
approach .(personal communication, James Smith).
Method 2 consists of the placement of a subsample of the
solid in a glass container with a modified cap which allows
direct connection of the container with the purge and trap
device in the laboratory. The subsample is extruded into
the pre-cleaned jar from the coring device, ideally in one
operation, to minimize opening and closing of the jar and
potential losses from volatilization. This method does not
require removal of the soil from the jar for the analysis.
This method may be preferable to Method 1 if there are
restrictions on the analytical or shipping methods to be
used (either through availability or regulatory
requirements), or if only very small samples are available
(e.g., a thin horizon is to be sampled). Concerns related
to reliance on readily available analytical technology,
detection limits, • and regulatory guidelines are of
particular importance to the regulated community (API,
1987). However, this method has the disadvantage of
resulting in a greater loss of VOCs through volatilization,
as discussed above.
-------
As is demonstrated by other papers at this conference,
methodologies for the collection, preservation and analysis
of soils and other solids for VOC analysis is currently the
subject of research. The ASTM method of developing
standards requires that all standards be reballoted every
five years. Hence, when this standard is reconsidered in
1996, it is hoped that much additional research and field
information will be available to revise and improve the
methodologies described.
-------
References
API (1987), Pre-Meeting Technical Memorandum, API Workshop,
on Sampling and Analytical Methods for Determining Gasoline
in Soil, Vail, CO.
ASTM (1991), Standard Practice for Sampling Waste and Soils
for Volatile Organics, in the 1992 Annual Book of ASTM
Standards, Volume 11.04, p. 108-111.
Bone, L. (1988), Preservation Techniques for Samples of
Solids, Sludges, and Nonaqueous Liquids, in Principles of
Environmental Sampling. Am. Chem. Soc., p. 409-415.
Siegrist, R.L. and P.D. Jenssen (1989), Sampling Method
Effects on Volatile Organic Compound Measurements in Solvent
Contaminated Soil. Inst. for Georesources and Pollution
Research, Norway.
Siegrist, R.L. and P.D. Jenssen (1990), Evaluation of
Sampling Method Effects on Volatile Organic Compound
Measurements in Contaminated Soils, Environ. Sci. Technol.
24: 1387-1392.
-------
Presenters Biography
Elly K. Triegel, Ph.D., is president of Triegel &
Associates, Inc., an environmental consulting firm with
offices in Pennsylvania, Delaware, Louisiana and Florida.
She received her B.A. in earth science/chemistry from
Western Connecticut State College, an M.A. in geological
sciences from Harvard University, an M.S. in environmental
engineering from the University of Houston, and a Ph.D. in
soil science from the University of Tennessee. She is active
in ASTM, acting as chairperson of the section. D 34.01.01
(Environmental Sampling and Monitoring), and in the
Association of Ground Water Scientists and Engineers, where
she serves on the board of directors.
-------
<1
Designation- D 4547 - 91 AMERICAN society* for testing and materials
1916 Race St Philadelphia. Pa 19103
Reprinted from tfw Annual Boe* ot ASTM Standafda. Copyright ASTM
If not Ifcted in the current com&ined index. »<(' appear in Che ne*i edition.
Standard Practice for
Sampling Waste and Soils for Volatile Organics1
This standard is issued under the fixed designation D 4547; the number immediately Tallowing the designation indicates the year or
original adoption or. in the case of revision, the year of tait revision. A number in parentheses indicates the year oflast rupproval. A
superscript cpsilon (>) indicates an editorial change since the last revision or reapprovaJ.
1. Scope
1.1 This practice describes field sampling of solid wastes
for subsequent volatile organics analysis in the laboratory.
This practice is also intended to apply to soils and sediments
that may contain volatile waste constituents.
1.2 Both the collection of the sample and the method of
containing the sample for shipment to the laboratory are
considered.
1..3 This practice concerns only sampling methods to be
used in the field; it does not cover laboratory preparation of
containers or solutions or other laboratory techniques related
to processing or analysis of the samples.
1.4 It is recommended that this standard be used in
conjunction with Guide D 4687.
1.5 This standard does not purport to address all of the
safety problems, if any, associated with its use. It is the
responsibility of the user of this standard to establish appro-
priate safety and health practices and determine the applica-
bility of regulatory limitations prior to use. For specific
precautionary statements, see Section 6.
2. Referenced Documents
2.1 ASTM Standards:
D 3550 Practice for Ring-Lined Barrel Sampling of Soils2
D4687 Guide for General Planning of Waste Sampling3
3. Terminology
3.1 Description of Terms Specific to This Standard:
3.1.1 material—for the purposes of this practice, material
covers any soil or wastes that are solid.
3.1.2 sample—the portion of the waste or soil material
that is initially collected using the techniques described in
this practice; portions of the waste or soil, in generic terms.
3.1.3 subsample—the portion of the waste or soil that is
collected by subdividing or trimming of the initial sample.
3.1.4 waste—for the purposes of this practice, waste
covers any discarded material that is solid in form.
4. Summary of Practice
4.1 Samples of soils or wastes can be obtained with
minimal loss of volatile organic constituents. Materials may
be either sampled from ground surface or test pits or
obtained by using down-hole coring devices. These samples
may be shipped in metal rings (that is, hollow metal
1 This practice is under the jurisdiction of ASTM Committee D-34 on Waste
Disposal and is the direct responsibility of Subcommittee D34.0I on Sampling and
Monitoring.
Current edition approved Aug. U, 1991. Published October 1991,
2 Annual Book of ASTM Standards. Vol 04.08.
1 Annual Book of ASTM Standards. Vol 11.04.
cylinders) directly to the laboratory, or they may be
subsampled by trimming or by using a small coring cylinder.
With the coring method, the coring cylinder is driven into
the waste or soil surface to remove the solid material without
exposure to the air. The subsample is then extruded from the
cylinder directly into a sample container. This method does
not apply to cemented material or material with fragments
coarse enough to interfere with proper coring techniques;
such samples are trimmed before handling.
4.2 Subsamples obtained in the field are contained so as
to prevent loss of volatiles using one of the two following
methods: (1) the subsample is stored in a glass bottle with
methanol; or (2) the subsample is stored in a vial designed to
minimize loss of volatiles (for example, a specially adapted
VOA vial). Advantages and disadvantages to both methods
are discussed in Sections 7 and 8.
5. Significance and Use
5.1 The objective of this practice is to provide procedures
for obtaining samples which will result in analytical data
representative of the concentrations and compositions of
volatile compounds actually present in the waste or soil. The
procedure also allows for correlation of the analytical data
with other properties of the waste or soil materiaJs. Several
factors are identified that influence the objective of this
procedure.
5.1.1 Loss of Volatiles:
5.1.1.1 Loss of volatile organics during sample collection,
handling, and shipping affects the concentrations detected by
the laboratory. Comparison of the field testing of volatiles
(using a gas chromatograph) with subsequent laboratory
testing of the solids or ground water from the same zone
suggests that losses can be significant, but are not necessarily
due to one particular part of the sampling and analysis
process. The principal mechanisms of loss are volatilization
of the compounds and biodegradation. Susceptibility of the
various compounds to these losses varies. Both the actual
concentrations and the relative amounts of the compounds
detected can be afTected. In some cases, the loss of a
compound or the formation of other compounds not actu-
ally present in the waste can occur. Compound gain and loss
will result in analyses that are both unrepresentative of field
conditions and subject to ambiguities in interpretation.
5.1.2 Selection of Samples for Analysis—The choice of
representative samptes is of particular concern in waste
materials and soils in which heterogeneities are significant.
Interpretation of the analytical data is generally improved if
the individuals) most familiar with the site can describe and
select the sample(s) to be analyzed in the laboratory.
5.1.3 Analytical Considerations—The method of sample
handling and containment is dependent on the method to be
-------
4B* D 4547
used in the laboratory for anaiysis of the volatile compounds.
The laboratory methods are addressed here only insofar as
they affect the collection method and influence the objective
staled above.
5.1.4 Other factors affecting the interpretation of data are
as follows: sample size, sample matrix, whether the
subsample analyzed is representative of the entire sample,
potential losses during handling of the sample using the
laboratory procedure, and detection limit.
5.2 This practice should be used in conjunction with
Practice D 3550 and Guide D 4687.
6. Safety
6.1 Proper safety precautions must always be observed
when sampling solid waste or contaminated soil. For general
guidelines to safety precautions, refer to Guide D 4687 and
Practice D3350. These standards, however, should only
complement the judgement of an experienced professional.
7. Sampling
7.1 Introduction:
7.1.1 This section is intended to define general sampling
guidelines to be applied to a variety of possible materials and
conditions. Many of the specific materials and conditions,
however, are not addressed in this document. Specific
sampling methods are presented for granular materials (for
example, contaminated soils and non-cemented solid wastes)
and materials that are cemented or of sufficient cohesion to
make driven samplers impractical. The procedures are in-
tended to allow flexibility in the following:
• 7.1.1.1 The means of collection (for example, from test
pits, surface sampling, and sampling during drilling},
7.1.1.2 The selection of a method suited for the individual
requirements of a project or the conditions encountered at a
particular site, and
7.1. J.3 The design and dimensions of the actual sampling
equipment.
7.2 Genera/ Methods:
7:2.1 The sampling procedure should be completed in a
minimum amount of time, with the least possible handling
of the sample before it is sealed in a container.
7.2.2 Rough trimming of the sample in the field should be
considered if cross-contamination of the surface of the
sample from other waste or soil strata is likely to occur
during collection. Significant contamination of the sample
surfaces can lead to redistribution of the vol allies throughout
the sample during shipping and storage, which will result in
misleading analytical data. The reduction of surface contam-
ination errors by I rimming should be balanced with the
potential losses from volatilization during trimming opera-
tions.
7.2.3 IF possible, the sample should be inspected visually
and its characteristics logged. Adjacent samples that appear
to have similar physical properties should be retained for
testing to determine or verify the relevant properties of the
solid materials (for example, grain size distribution). Ideally,
the sample itself or another sample of the same material
should be available for inspection and notation of the
following: general appearance and color, presence of oils or
other visible signs of contamination, grain-size distribution,
volatile organics, and so forth. In the case of samples that are
collected directly into metal rings for shipment to the
laboratory such inspection may not be possible (see 7.3.3),
and alternative procedures (examination of the exposed ends
of the core or adjacent samples) should be used. Selection of
representative samples for volatile® analysis is aided greatly
by information obtained from field testing of other samples
from the same stratum.
7.3 Granular or Uncemenied Materials:
7.3.1 Granular materials may be collected from the
ground surface, the walls of test pits, blocks of wastes or soils,
or by using split barrel or other sampling devices during the
drilling of soil borings.
7.3.2 In the case of samples collected from test pits,
ground surface, or larger blocks of soils or waste, the surface
of the sample should be trimmed to remove contaminants
from other waste or soil strata or to remove surface layers
that may have already lost volatiles, This removal of surface
layers can be accomplished by scraping the surface using a
clean spatuJa or knife.
7.3.3 Collection of Samples in Mela1 Rings:
7.3.3.1 Samples from split barrels or similar devices may
be collected in precleaned metal rings inserted in the
sampling barrel, such as those described in Practice D 3550.
The exposed ends of the solid or waste in the ring or in
adjacent samples are used
-------
# D 4547
cylinder should be smaller than the inside diameter of the
mouth of the sampling container to avoid contamination of
the outside threads of the bottle, which may result in a bad
seal.
7.3.4.4 The coring cylinder containing the subsample of
material for analysis can be removed by excavation (surface
or test pit) or by cleaning away the excess sample with a
spatula or a clean, disposable towel. The solid materials
around the cylinder are used (7) to log the properties of the
sample, (2) to aid in determining whether the subsample is
representative of the horizon to be sampled, or (J) for
additional tests (for example, grain size analysis, field testing
of total volatiles, field gas chromatography). If the subsample
is not extruded from the cylinder immediately, it should be
sealed temporarily, by covering the ends with aluminum foil
and TFE-fluorocarbon tape, and stored on ice or similarly
cooled.
Note 1—Aluminum foil may be unsuitable in very alkaline environ-
ments.
7.3.4.5 The subsample is extruded using a cleaned rod to
push the subsample out of the cylinder. The subsample is
extruded directly into the sampling container. If the
subsample is to be placed in a 40-mL vial, ideally, the length
of subsample collected in the cylinder should be greater than
the height of the vial, so that the vial can be filled in one
operation. Extrusion should be performed rapidly and soon
after sampling to reduce volatilization and redistribution of
volatiles, which may result from contaminated subsample
ends.
Note 2—Extrusion of cooled subsamples under controlled condi-
tions is preferred if it can be performed on-site or within a short period
of time (that is, four to six h) after sample collection.
7.4 Cemented Solid Wastes:
7.4.1 Subsampling Cemented Material by Trimming—
The solid wastes or contaminated soils may be so hard that
the coring cylinder cannot be driven into the wastes.
Subsamples of such wastes and soils may be collected by
trimming the larger sample with a cleaned tool to a size that
can be placed in the sampling jar. Although some loss of
volatiles can be expected, the losses should be less than in
loose, granular solids due to the lower surface area exposed
to the atmosphere. Collection, trimming, and containment
of the subsample should be accomplished in the least
amount of time practical.
8. Handling
8.1 General:
8.1.1 In the case of materials not subsampled in the field
(that is, those collected in metal rings inserted in down-hole
sampling devices), the sample is retained in the metal ring
and sealed as described in 6.3.3. Both the sample and the
metal ring are then shipped to the laboratory, where the
subsample is extruded for analysis.
8.1.2 In the case of materials subsampled in the field, glass
containers with inert caps should be used for storage and
shipment. To retard volatilization and biodegradation,
subsamples should be placed on ice or similarly cooled as
soon as practical. Two alternative methods of containing
subsamples for shipment to the laboratory are outlined.
These methods have different applications, and advantages
and disadvantages.
8.2 Method 1—Methanol Container:
8.2.1 This method can be used for a wide range of cases,
but it is particularly useful for situations in which (1) larger
samples are desired to obtain a composite representation of
the volatiles concentration and composition, (2) high detec-
tion limits can be tolerated, or (J) biodegradation is a
concern.
8.2.2 Sample containers consist of wide mouth, 8-oz. glass
jars with TFE-fluorocarbon-lined lids. An aliquot of 100 mL
of an appropriate analytical grade of methanol is added to
the organic-free jar by the laboratory that supplies the jar, by
the sample collector, or by a third party. The solid waste or
soil is added to the jar containing the methanol to a specified
level in the jar. This level is defined by the party responsible
for preparing the sampling jars and is equivalent to the level
that would correspond to the addition of approximately 100
g of the soil or waste (at an assumed specific gravity). The
actual mass of material added to the jar is determined later
by comparison with a tare weight, by the analytical labora-
tory. The jar containing the methanol should not be left
open unnecessarily.
8.2.3 The volatile compounds are more soluble in the
methanol than in the soil water, which results in transfer of
the volatiles from the solid to the methanol for analysis.
Addition of methanol to the subsample in the field allows a
longer contact time with the subsample, which improves the
extraction efficiency. In addition, extraction of volatiles is
performed with a larger subsample than used in some
methods (100 g versus 5 g with a heated He purge), which
results in a more representative determination of the concen-
trations of the volatiles.
8.2.4 This method permits splitting of the sample into
several jars, or compositing by placing several aliquots of the
solid waste (from the coring cylinder described above) in one
sample jar. The methanol reduces volatilization during
repeated opening and closing of the jar for each subsample
and serves as a medium for extracting volatiles from each
subsample added to the jar. The physical mixing used in
other types of analyses, with its potential for volatilization
and incomplete mixing, is avoided with this method. This
method also allows for multiple laboratory analyses of the
same sample..
8.2.5 Since the partial pressure of the volatiles over the
methanol is very low, losses by volatilization are reduced.
The methanol also inhibits microbial activity, reducing losses
from biodegradation.
8.2.6 The primary disadvantages of this method are (/)
the need for laboratory cooperation in preparing tared
sample jars, (2) possible shipping restrictions (if the meth-
anol volume is sufficient to qualify the samples as flammable
materials), and (3) the reduction of sensitivity of the gas
chromatograph/Hall detector (if such detectors are used by
the method).
8.3 Method 2—Dry Container:
8.3.1 This method involves placement of a subsample of
the solid in a tared 40-mL glass container (or a size
compatible with the analytical instrumentation) for ship-
ment to the laboratory. This container is modified by the
addition of a cap that allows direct connection of the
3
-------
4§> D 4547
container with the purge and trap device in the laboratory, so
that removal of the subsample is not required for analysis.
Subsample weights are determined in the laboratory.
8.3.2 The samples are extruded into the clean, organic-
free jar from the coring cylinder, ideally in one operation, to
minimize opening and closing of the jar and the potential
loss from volatilization. This method is preferable for cases
in which the methanol method is not desired (due to
shipping limitations, detection limit requirements, or labora-
tory restrictions) or if only small samples are available for
analysis (for example, if only a thin horizon of contaminated
material exists or a limited zone of contaminated material is
the target for analysis).
8.3.3 The primary disadvantages of this method are (7)
the requirement for specialized containers, (2) the inability
to perform additional analyses of the same sample, and (J)
the small size of the sample, which can reduce the repre-
sentativeness of the sample.
9. Packaging and Package Marking
9.1 An indelible label identifying the sample should be
secured to the container. The label should contain or
reference the following information:
9.1.1 Name and location of site,
9.1.2 Date and time of sampling,
9.1.3 Location of sampling,
9.1.4 Sample number,
9.1.5 Description and disposition of sample,
9.1.6 Name of sampling personnel,
9.1.7 Type of preservative, and
9.1.8 Sampling conditions and analytical requirements.
9.2 Pack the sample container securely in a shipping
container. The sample container should be packed in ice and
cooled to 4*C. A min/max thermometer should be packed
with the samples.
9.3 Follow DOT (Department of Transportation)4 ship*
ping regulations.
9.4 Make arrangements for handling, logging in, adequate
storage and analysis of the sample or subsample at its
destination. If warranted, follow chain-of-custody protocol.
* Available from the Superintendent of Documents, U.S. Government Printing
Office, Washington, DC 20402.
The American Society for Tasting and Materials takes no position respecting the validity ol any patent rights assarted In connection
with any Hem mentioned In this standard. Users ot this standard are expressly advised that determination ot the validity ot any such
patent rights, and the risk ot Infringement ot such rights, are entirely their own responsibility.
This standard Is subject to revision at any time by the responsible technical committee and must be reviewed every live years and
U not revised, either reapproved or withdrawn. Your comments art Invited either for revision ot this standard or for additional standards
and should be addressed to AS TU Headquarters. Your comments will receive careful consideration at a meeting of the responsible
technical commtttee, which you may attend. It you feel that your comments have not received a fair hearing you should make your
views known to the ASTU Committee on Standards, 1916 Race St., Philadelphia, PA 19103.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 21
Soils, Synthetics and Screening: May the Odds Be With You
R. Rajagopal
Depts. of Geography and Civil and Env. Engr.
University of Iowa, Iowa City
January 12-14, 1993
Las Vegas, Nevada
-------
Soils, Synthetics and Screening: May the Odds Be with You
R. Rajagopal, Professor
Depts. of Geography and Civil and Env. Engg.
302 Jessup Hall
The University of Iowa
Iowa City, IA 52242
Extended Abstract
Numerous surveys across the nation have revealed the presence of organic compounds in
water resources and soils. The sums spent to monitor the 90,000 or more drinking water
supplies, the thousands of wells and hundreds of thousands of acres of soil at private
sector hazardous waste sites and federal facilities, and other data collection programs of
the U.S. and State Geological Surveys and state health and environmental agencies will
be at least in the range of a few to several billion dollars a year. Under these
circumstances, it is extremely important to effectively utilize the huge volume of
available field data in the development of regulatory and laboratory procedures.
Our ability to detect contaminants in water and soil is directly related to the strategy
employed in the diagnostic analysis of field data and the development of cost-effective
laboratory procedures. The central thesis of this presentation is to show that when the
likelihood of occurrence of a contaminant in a water or soil sample is small and distinct
patterns in co-occurrence of compounds exist, the use of certain field and laboratory
screening techniques for detecting contaminants will lead to considerable improvement in
our ability to identify and isolate problems. In this presentation, field evidence of more
than a million analytical measurements from governmental and private sector databases
will be used to illustrate and verify the utility of such techniques in ground-water quality
and soils monitoring. Recent results of implementing such models at actual field sites
will also be presented.
Several additional hypothetical and real world case studies will be used to substantiate
the central thesis put forth in the paper. The illustrations for a simple self-explanatory
case study is provided in the following pages.
-------
Key References:
R. Rajagopal and P. C. Li. Comparison of Two Screening Methods for the Detection of
VOCs in Ground Water. Journal of Chemometrics. 5(3): 321-331. 1991.
R. Rajagopal, Economics of Screening in the Detection of Organics in Ground Water,
Chemometrics and Intelligent Laboratory Systems. 9(1990): 261-272.
R. Rajagopal and L.R. Williams. Economics of Sample Compositing as a Screening
Tool in Ground-Water Quality Monitariae. Ground Water Monitoring Review. 9(1):
186-192. 1989, Discussioa: Ground Water.Moaitanng Eerievc. 9(3): S2-S8. .5S5
EL Rajagopal. Optimal Sampling Strategies for Source Identification in Environmental
Episodes, Environmental Monitoring and Assessment, 4(1): 1-14, 19S4.
R. Sajagcpal. Invited Guest Editorial.- Environmental Intelligence and
Regulations. American Laboratory. 24(183:6,8. December 1992.
-------
Number of Cells = 100 (100 spots per cell)
Contaminated Area = 1 Spot
Contaminated Spot
-------
Uniform Grid Sampling
Number of Samples = 100 Spots
Chance of Identifying
the Contaminated Spot = 1 %
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
/
o
o
o
o
o
o
o
o
° /
f
o
o
o
o
o
o
o
o
o
/o
o
o
o
o
o
o
o
o
—7—
Contaminated Spot
-------
Composite Sampling
Number of Samples = 43
Chance of Identifying
the Contaminated Area = 100 %
8"
29
^3
Contaminated Spot
-------
Composite Sampling
Number of Samples
25 Composites (400 Spots)
4 Composites (100 Spots)
2 Composites (50 spots & 50 spots)
2 Composites (25 spots & 25 spots)
2 Composites (12 spots & 13 spots)
2 Composites (6 spots & 7 spots)
2 Composites (3 spots & 4 spots)
4 Individual Samples ( 1 spot each)
43 Samples
Advantage: Built-in QA
Disadvantages: Loss of Resolution in DL &
Issues in Matrix Interference
-------
Case Study 1
Contaminated Area = 1 out of 10,000 spots
Composite
Sampling
Uniform Grid
Sampling
Max. Composite Size
400
1
Reporting Level
# of Samples
43
100
Cost of Sample Prep.
& Analysis
$ 2,000 / sample
$1,000 / sample
Total Cost ($)
86,000
100,000
Identification Prob.
iiiiim
ilPliiiliil ill.ii ii®iiiii!i!ii
-------
Prof. R. Rajagopal
The University of Iowa
A Biosketch
R. Rajagopal is a Professor of Geography and Civil and Environmental Engineering at the
University of Iowa. He has over twenty years of teaching, research and consulting experience
in the fields of ground water protection, environmental monitoring, impact assessment, and
management information systems.
He received a BS in Mathematics/Physics and two years of graduate-level education in
statistics from the University of Bombay; a master's degree in Operations Research from the
University of Florida; and a Ph. D. in Water Resources Management from the University of
Michigan.
He has supervised over 40 master's and Ph. D. students in environmental sciences, water
resources, forestry, geography, and engineering. Currently he advises two post-doctoral, two
pre-doctoral research fellows, and six doctoral and two master's candidates in geography and
environmental engineering.
He has organized and directed over fifty workshops and seminars on topics in
ground-water pollution, environmental protection, risk assessment, information integration,
problem-solving, creativity, and innovation to over 2,000 people in academe, government,
non-profit groups and industry.
During 1986-88, he served as a visiting scientist at the U.S. EPA's Environmental
Monitoring Systems Laboratory in Las Vegas. At the U. S. EPA, he was recognized for his
outstanding contribution to the Laboratory's innovative problem solving and creative thinking.
He is the founding editor of the quarterly journal The Environmental Professional. In
1985, he received the National Association of Environmental Professionals' Distinguished
Service Award.
Since its inception in 1985, Professor Rajagopal has been invited to serve as a nominator
for the annual Japan Prize (Japanese equivalent of the Nobel). He serves on the Health Effects
Committee of the Great Lakes Protection Fund, a $ 100-million endowment created to protect
the Great Lakes. He also serves on the Board of Trustees of ECO, the premier Environmental
Career Organization in the United States, a $ 5 million/year non-profit organization that assists
and counsels students and professionals with career choices.
He has authored, coauthored, or edited over 100 journal articles, research reports,
journals, and books in the fields of ground water protection, environmental monitoring and
information systems.
He is a recipient of several research grant and contract awards from the Department of
Energy, the National Geographic Society, the U.S. Environmental Protection Agency, the
U.S. Forest Service, the National Science Foundation, and other private foundations and
educational institutions.
-------
JOURNAL OI: CHEMOMETRICS, VOL. 5, 321-331 (1991 )
COMPARISON OF TWO SCREENING METHODS FOR THE
DETECTION OF VOLATILE ORGANIC COMPOUNDS IN
GROUND WATER
R. RAJAGOPAL AND PING-CHI LI
302 Jessup Hall. Department of Geography, The University of Iowa, Iowa City, IA 52242, U.S.A.
SUMMARY
It is shown that the presence of 31-35 commonly measured volatile organic compounds (VOCs) in
ground water can be detected with small error rates by using screening methods which analyze for a subset
of such VOCs. A study of selected data sets indicates that analytical determinations of only from two
to eight VOCs will suffice to detect 95% of all VOC hits. It is also shown that a serially optimal algorithm
for selecting the VOCs for screening is very nearly as accurate as a globally optimal algorithm and much
easier to implement. These conclusions are supported by empirical evidence from two drinking-water data
sets and one hazardous waste site data set. Additional research areas are also outlined.
key words Screening Ground-water quality Monitoring
Volatile organic compounds (VOCs) Optimization
INTRODUCTION
A National U.S. Environmental Protection Agency (EPA) survey and a state-wide survey in
Iowa have revealed,the presence of volatile organic compounds (VOCs) in ground-water based
community drinking-water supplies.With our increasing dependence on aquifers for
drinking water coupled with the threats of possible contamination of water resources, issues
related to ground-water protection have gained considerable importance in recent years.
There are no precise estimates of" national expenditures on ground-water quality monitoring.
Considerable sums are spent to monitor ground-water quality under the Sate Drinking Water
Act, RCRA. CERCLA and many other programs of federal and state agencies. The cost
effectiveness with which contamination problems are identified under these programs is directly
related to the strategy employed in monitoring. For example, when the likelihood of
occurrence of a contaminant in a sample is small and distinct patterns .in the occurrence and
co-occurrence of compounds exist, the use of certain screening methods for detecting
contaminants will lead to considerable improvement in our ability to identify and isolate
problems. The purpose of this paper is to illustrate the potential for cost reduction in detecting
VOCs in ground-water samples when screening methods are used.
0886-9383/91/030321 - I 1S05.50
1991 by John Wiley & Sons, Ltd.
Received May 1990
Accepted 10 December 1990
-------
322
R. RAJAGOPAL AND P.-C. LI
SCREENING METHODS IN GROUND-WATER QUALITY MONITORING
Screening, as defined in this paper, is an ordered approach for the analysis of samples based
on a ranking of total detections (occurrences and co-occurrences). Screening is especially
attractive in those situations where the probability of occurrence of any compound (from a
group such as VOCs) in a collection of samples is small and the relative probability of
occurrence of a few compounds is significantly larger than the probability of occurrence of the
remainder of the group. The derivation of the order of sequential selection of compounds in
a group such as VOCs so as to maximize a detection objective is an interesting matheniatical
problem. Another interesting variant of such a problem is to maximize the proportion of
contaminated wells detected.
Extensive analyses of ground-water quality data from ambient sources, drinking water and
hazardous waste sites have revealed patterns and relationships in the occurrence and co-
occurrence of compounds with the above-mentioned properties.1-3,4 Based on such findings,
two structured screening methods for VOCs in ground water are developed and their
performance efficiencies compared in this paper.
Screening methods attempt to develop efficient test procedures to determine and/or to
quantify the presence or absence of contaminants in samples. They are diagnostic tools for the
determination of the extent to which each sample should be analyzed. They fit logically "into
the realm of environmental monitoring. In screening, the results of the first stage help
determine the scope of analysis in the second stage; the results of the second stage help
determine the scope of analysis in the third stage; and so on. Such an approach uses
information from prior stages to guide decisions of future stages.5,5
SCREENING METHODS
As a result of previous studies,7-10 there is an extensive body of literature on sequential
implementation of sampling plans in industrial quality control. By adopting some of the basic
ideas offered in this literature, screening can be described as a decision-tree or feedback-based
approach to the analysis and interpretation of sampling results. Exhaustive analysis of all
compounds in a list (i.e. analyzing all samples for all constituents in a given list, such as thai
of EPA's Priority Pollutants listed under the Clean Water Act), on the other hand, is a case
of minimal or no screening with reference to the given list.
Throughout this paper the terms detections, occurrences and hits are used synonymously.
If several compounds of interest are detected in ground water, then how does one go about
selecting a few as screens without losing significant overall information? This problem is
simplified by framing the question in the context of a single group of priority pollutants, such
as VOCs in this paper. Similar problems can also be framed for other groups of compounds
such as acids, base neutrals and pesticides."
The application and the comparison of performance of the two screening methods for VOCs
are illustrated with data from the EPA's National Ground Water Supply Survey.' A summary
of performance comparisons of the two methods with reference to data from drinking-water
supplies in Iowa and data from hazardous waste sites across the U.S. is also included. The
mathematical forms and the algorithmic complexities of implementing the two methods arc not
discussed in this paper.
The utility of screening methods in regulatory applications is an extremely important and
complex subject. The eventual success of such methods in regulatory settings will require
further attention to analytical, economic, health and institutional considerations.
-------
SCREENING METHODS IN GROUND-W ATER MONITORING
323
METHODS
The national ground water supply survey (GWSS)
In 1981-1982 the U.S. EPA surveyed 945 of over 47 000 ground-water-based community
public water supplies of the U.S. for the presence of 35 VOCs.1 The geographic distribution
of the supplies, the sampling strategy and the rationale of the survey design are described in
Reference I. The 35 VOCs that were analyzed in ground-water samples are shown in Table I.
Table 1. Occurrence statistics of 35 volatile organic compounds analyzed in 945 samples obtained for the
National Ground Water Supply Survey (source of raw data: Reference I)
Notation
Compound
Frequency of
occurrence
Frequency of occurrence and
co-occurrence
A1
Bromodichloromethane
445
1742
A2
Chloroform
465
1712
A3
Dibromochloromeihane
405
1644
A4
Bromoform
249
J 098
A5
Trichloroelhylene
91
484
A6
1,1,1-Trichloroethane
78
395
AT
Tetrachloroethylene
79
334
A8
1,2(cisj fro/j5)-Dichloroethylene
54
323
A9
1,1-Dichloroethane
41
239
AI0
1,1 -Dichloroethylene
24
162
At]
Carbon tetrachloride
30
153
A12
o/p-Xylene
IS
109
A13
ffl-Xylene
16
94
A14
Dichioroiodomethane
IS
87
A 1 5
Toluene
11
59
A16
1,2-Dichloroethane
10
54
AI7
Vinyi chloride
7
53
A18
Benzene
n
5?
AI9
/7-Dichlorob enzene
9
40
A 20
1.2-Dk.hloropropane
13
38
A2!
Bromobenzene
ft
34
A22
Ethvlbenzene
f>
30
A23
Methylene Chloride
s
17
A 24
-------
324
R. RAJAGOPAL AND P.-C. LI
The problem
Given that only a few VOCs from any specified list are detected in a data set, how does one
form subsets of the list and use them as screens without losing significant overall information?
For a fixed number of VOCs in a screen, two methods to maximize the total expected
number of detections in ground-water samples are developed, compared and evaluated in this
paper. To simplify the problem, consider the case of VOC occurrences in samples from 945
community water supplies in the U.S. (Table 1). In all, a total of 33 075 (945 x 35) VOC
determinations were made for 35 compounds from 945 samples. As shown in Table 1, a total
of 2098 detections out of a possible 33 075 (6-34<7o) were noted.
Method 1. Serial optimization
Frequencies of occurrence and co-occurrence were sorted by compounds in descending order
of total number of occurrences as shown in Table 1. For example, bromodichloromethane
(Al), the first entry in Table 1, was detected in 445 samples. In those 445 samples, 1297 other
VOC detections also occurred, thus leading to a total of 1742 occurrences shown in Table 1.
If a complete VOC scan of the 35 compounds considered in this analysis had been carried out
for only the 445 samples in which bromodichloromethane was found, a total of 1742 hits
would have resulted (Table 1), or 839b of the total of 2098 detections would have been
captured (Table 2). The above process with chloroform as the first screening compound would
have resulted in a total of 1712 hits (Table 1). For a one-compound screening problem,
bromodichloromethane (Al) leads to a maximum number of total detections.
After removing the records of 445 samples in which bromodichloromethane occurred and
repeating the process described earlier, the choice of trichloroethylene (TCE) is found to yield
the most detections among the remaining 34 compounds. TCE occurred in 42 samples and had
121 co-occurrences resulting in a total of 163 detections (Table 2). Hence selecting
bromodichloromethane (Al) and TCE (A5) as the first two compounds in a two-compound
screening problem would have resulted in a total of 1905 detections (as shown in Table 2). This
in turn would add 7-8% to the total. Thus, if a complete VOC scan of the 35 compounds had
been done for the 445 bromodichloromethane samples, followed by the 42 TCE samples.
90-8% of total detections would have been captured. The per cent and cumulative per ceni
captures of total detections as a function of the number of VOCs included in the screen (up
to eight compounds) are shown in Table 2.
Essentially, instead of solving a single two-compound problem, we have solved two
successive one-compound problems. Similarly, we have extended the approach and solved a
series of n successive one-compound problems in place of a single /j-compound problem
(Table 2). This serial or incremental approach to identifying the compounds for a screen is
termed serial optimization. Such an approach does not guarantee global optimality except for
screen size one. In general, the solution obtained from n successive one-compound problems
can never be superior to the optimal solution of the single n-compound problem. Assessment
of the magnitude of this inequality is the central thrust of this paper.
Method 2. Global optimizaton
For one-compound problems the global and serial methods are identical and the choice of
bromodichloromethane (Al) is optimal in both cases (Tables 2 and 3). The choice of
bromodichloromethane (AI) and TCE (A5) as the first two compounds in a two-compound
-------
Table 2. Occurrence, co-occurrence and performance statistics (total, per cent and cumulative per cent detections) for a serially optimal screening
set of volatile organic compounds found at National Ground Water Supply Survey sites
Compound
1
2
3
4
5
6
7
8
Others
Total
hits
Cumulative
total hits
Cumulative
per cent hits
1. Bromodichloromethane
445
49
399
38
224
28
11
20
528
1742
1742
830
2. Trichloroethylene
42
11
21
3
25
2
11
48
163
1905
90-8
3. Chloroform
55
1
1
2
2
0
13
74
1979
94-3
,4. 1,1,1-Trichloroethane
18
4
6
0
5
8
41
2020
.96 3
5. Bromoform
17
1
0
0
10
28
2048
97-6
6. Tetrachloroethylene
17
0
1
2
20
2068
98-6
7. o/p-Xylenc
3
0
4
7
2075
98-9
8. 1,1-dichloroethane
4
2
6
2081
99-2
9. Others
17
17
2098
1000
Total detects
445
91
465
78
249
79
18
41
632
2098
Total non-detects
500
854
480
867
696
866
927
904
24883
30977
Total analyses
945
945
945
945
945
945
945
945
25515
33075
-------
Ihl
N1
0\
Table 3. Occurence, co-0 c currency and performance statistics (total, per cent detections) foi a globally optimal screening se! of volatile organic
compounds Found at Natrona! Ore and Water Suppfy Survey sites
Compound selected
Al
A2
A3 .
A4
A5
A6
A7
AS
A9
A10
A32
A20
Others
Cumulative
total hits
Cumulative
per cent hits
w
1. Al
445
199
391
224
49
3$
23
32
20
13
]1
4
88
1742
sac
73
>
t—
2. Ai. A5
44S
430
392
227
91
59
53
sa
31
20
tt
s
107
1905
90*8
>
3, A2, A4, A5
439
465
399
249
91
64
56
53
31
20
15
6
i2t
2007
95-7
o
4. A2, A4, AS, A7
43 9
4«5
399
249
91
70
79
54
34
21
\S
i>
120
2042
97 3
"V
>
5. Al, A3, AS, AS, A7
444
465
405
242
91
78
79
54
37
23
15
7
121
2061
93-2
r-
6. Al, A2, A4,
445
465
404
249
91
73
79
. 54
4!
24
15
1
122
2069
98-6
>
2
A5, A7, ,A!0
O
7. Al, A2, A4. A5,
445
465
4(14
249
91
73
79
54
41
24
18
7
3 25
2075
99-0
A7, A10, A] 2
n
3, Al, A2, A4, AS.
445
465
404
249
91
73
79
54
41
24
18
13
132
2082
99-2
c
A7, A9, A(2, A2G
Total detects
445
465
4fl5
249
91
78
79
54
41
24
18
13
137
2098
Ttiiat non-delects
500
480
541
696
854
86?
866
891
904
921
927
932
21598
30977
Total analyses
945
945
945
945
945
945
945
. 945
945
945
945
945
21735
33075
-------
SCREENING METHODS tN GROUND-WATER MONITORING
32?
screening problem has also, by chance, resulted in a globally optimal solution of 1905
deteciions (Tables- 1 and ?l- 'i 'he case of a three-compound problem the selection of
chloroform (A2), bromoform (A4) and TCE (A5) resulted in a globally optimal solution of
2007 detections (Table 3). In comparison, the selection of bromodichlorometbane (A)), TCE
(A5) and chloroform (A2) resulted in a serial optimization solution of only 1979 detections
{Table 2).
The mathematics of serial optimization and the development of algorithms to obtain
solutions for problems of various screen sizes are straightforward. The mathematics of global
optimization is also straightforward and is based on the application of classical techniques of
dynamic programming9''0 to the sequential selection of screening compounds. While the
mathematics of global problems is straightforward, the development of algorithms for such
problems increases in complexity with screen
RESULTS AND DISCUSSION
The cumulative per cent capture oi as a ((CKcm nf scraer. s:: '.iurJber if
compounds included in the screen) for the two screening methods is stiowri in Figure 1. The
difference in performance (cumulative per cent detections) between the two screening methods
is minimal and in no case is the serial optimization solution inferior to the global solution by
more than 1*4% (from Tables 2 and 3 or Figure 1). Of course, all our findings are based on
hindsight. In the future we plan lo do selected confirmatory studies. For example, for a given
list of compounds and a clearly defined objective, a comparison of optimal screens of various
sizes for similar but mutually exclusive environments can be carried out. Another comparable
study could be to use part of a data set (training data set) to develop a screening list and
evaluate the list's performance on ibe remaining part.
To obtain further experience in the application of serial and global methods to data from
varied ground-water environments, three more case studies, described below, were undertaken.
30 ' 1 ' » ' r—
<332345678
Screen Size
(Number of compounds in. the Screen}
Figure 1. Cumulative per cent capture of defections under serial and global opumal screening methods
(based on the analysis of 35 VOCs at 945 U.S. drinking-wajer sites)
-------
328
R.. RAJAGOPAL AND P.-C. LI
Case study I
The 35 VOCs analyzed as part of EPA's Ground Water Supply Survey included fcve
halomechanes (AI, A2, A3, A4 and AI4). They are known to be predominantly disinfection
byproducts and are not clearly linked to ubiquitous contaminant sources. They contributed to
1582 of a total of 2098 (7S-4%) detections. Therefore they were deleted from the data set and
a comparison of performance of the two screening methods was done with the remaining 30
VOCs which contributed to a total of 516 hhs.
Again, based on analyses similar to the one described earlier, the cumulative per cent capture
of detections as a function of screen size under the two screening methods is shown in Figure 2.
In this case the two methods performed identically, i.e. there was no difference in cumulative
per cent detection.
Case study 2
In 1986, the Iowa Department of Natural Resources undertook a one-time survey of all public
water supplies in Iowa. One of the major purposes of this survey was to discover the extent
of VOC contamination of drinking-water supplies in Iowa.2 For this case study, records of
analysis of samples from 755 ground-water-based drinking-water supplies for the presence of
38 synthetic organic compounds were obtained. In all, the resuJts of 28 690 (755x 38)
analytical determinations are available in this data set. A total of 1211 detections out of a
possible 28 690 (4-2%) were found. Again, as expected, a large number (1082 out of 1213) of
such detections were the byproducts of disinfection (trihaiomethanes).
The cumulative per cent capture of detections as a function of screen size under the two
screening methods is shown in Figure 3. Again in this case, there was no difference in
performance (difference in cumulative per cen detections) between the two methods.
100
V
a
u,
$
c
4f
y
w
V
&
V
0 1 2 3 A 5 6 7 8 9 10 11 12 13 14
Screen Size
(Number of compounds in the Screen)
Figure 2. Cumiitetm per cem capture of detections under serial and global optimal screening methods
(based on the analysis of 30 VOCs at 945 U.S. drinking-water sites)
-------
SCREENING METHODS IN OROUND-WATER MONITORING
329
88 -t— 1 ^1 1 > >
0 I 2 3 4 0 6
Screen Size
(Number of compounds in the Screen}
Figure 3. Cumulative per cent caplure of de.'eciion-S under serial and global optimal screening methods
(based or. the analysis of 38 VOO K 755 Iowa drinking-water sites!
50"! ¦ ' ~ 1 1 ' ' —' ' !
0123456769 10
Screen Size
[Number of compounds in tbe Screen)
Figure 4. Cumulative per cent capture of detections under serial and giobai optima! screening methods
(based on the analysis of 33 VOCs at 314 U»S, hazardous *asie sites)
Case study 3
The EPA's Environmental Monitoring Systems Laboratory in Las Vegas, based or. sue
investigations, lias organized a comprehensive data set.u It has analytical results for about
IS 000 ground-water samples fraits 4772 wdls at 314 sites. 1: is one of a few computer-coded
files that include substantial data on organic contaminants from a large number o( iveils at
hazardous waste sites. This data set was cdled to include 1565 wells in which ail of a group
of 31 VOCs (most priority pollutants listed under ihe Clean Water Act) were analyzed and
repotted. In all. the results of 4S 515 <1565 x 3!) analytical determinations are available in this
data set. Analysis of this data set indicated that only a fraction of organic* analyzed (2825 qui
-------
330
R. RAJAGOPAL AND P.-C. LI
of a possible 48 515 or about 5-8%) in ground-water samples from hazardous waste sites are
detected in any individual well.
The cumulative per cent capture of detections as a function of screen size under the two
screening methods is shown in Figure 4. Once again, the two methods performed identically.
CONCLUSIONS
Distinct patterns in occurrence and co-occurrence of a group of compounds such as VOCs in
ground-water samples are shown to be of much value in the development of screening
methods. Two screening methods, serial and global, that maximize the total expected number
of detections of VOCs in ground-water samples are developed, compared and evaluated. It is
shown that the presence of 31-35 commonly measured VOCs in ground water can be detected
with small error rates by using screening methods which look at a subset of such VOCs. In
all cases analyzed, except for one, there was no significant difference in performance
(cumulative per cent detections) between the two methods. Even in the one case where a
difference in performance was observed, the globally optimal solution was superior to the
serially optimal solution by no more than 1 -4%. Analysis of field data showed that analytical
determinations of only from two to eight VOCs from a list of 31-35 will suffice to detect 95%
of all VOC hits. The computationally simpler serial method for selecting VOCs for screening
is shown to be very nearly as accurate as the more complex global method and much easier
to implement. These findings are illustrated by the application of the methods to ground-water
quality data from drinking water and hazardous waste sites.
In general, the utility of screening methods presented in this paper is governed by many
factors. Rajagopal11 has shown that cost-effective screening is a function of patterns in the
occurrence and co-occurrence of contaminants and the price breaks offered by laboratories for
the analysis of a few compounds (screens) instead of a standard scan.13-14 In addition,
potential opportunities for achieving cost reduction in monitoring could also evolve from the
combination of screening methods with advances in fiber optics, X-ray fluorescence
spectrometers, GC/Fourier transform—infrared system with light pipe technology, and sample-
compositing techniques."'15-17
Finally, it is known that the sensitivities of commercially available mass spectrometers to
detect small concentrations of materials range from parts per million to parts per trillion.I!>
In a recent survey, concentrations of five selected herbicides (screens) were determined by a
sophisticated capillary GC/MS instrument with a stated capability of detecting concentrations
as low as 20 parts per trillion.19 Given such possibilities, it is prudent not to write off the utility
of screening techniques on the basis of the limitations of existing technology or standard
methods.
ACKNOWLEDGEMENTS
Joseph J. D'Lugosz of EMSL-LV, U.S. EPA is the Project Officer responsible for the
Cooperative Agreement under which the research reported in this paper was carried out.
Constructive comments of the reviewers, Prof. Charles Davis, and Dr Russell Plumb Jr., are
much appreciated.
Although the research described in this manuscript has been funded wholly or in part by the
United States Environmental Protection Agency through a Cooperative Agreement (CR
813552-01-0) with the University of Iowa, it has not been subjected to Agency review and
therefore does not necessarily reflect the views of the Agency and no official endorsement
should be inferred.
-------
SCREENING METHODS IN GROUND-WATER MONITORING
331
REFERENCES
1. J.J. Westrick, W. Mello and R. F. Thomas, 'The ground water supply survey', J. Am. Water Works
Assoc. 76, 52 (1984).
2. Iowa Department of Natural Resources, Pesticides and Synthetic Organic Compound Survey, Report
to the Iowa General Assembly, Des Moines, IA (1988).
3. W. M. Shackelford and D. M. Cline, 'Organic compounds in water'. Environ. Sci. Technol. 20, 652
(1986).
4. P. C. Li and R. Rajagopal, 'Modeling the occurrence of volatile organic compounds in ground
water', J. Am. Water Works Assoc., in review (1990).
5. L. G. Mitten, 'An analytical solution to the least cost testing sequence problem , J. Ind. Eng. 11,
17 (1960).
6. R. Rajagopal, 'Optimal sampling strategies for source identification in environmental episodes',
Environ. Monitor. Assess. 4, 1 (1983).
7. A. Wald, Sequential Analysis, Wiley, New York (1947).
8. W. E. Deming, Some Theory of Sampling, Wiley, New York (1950).
9. R. E. Bellman, Dynamic Programming, Princeton University Press, Princeton, NJ (1957).
10. R. E. Bellman and S. Dreyfus, Applied Dynamic Programming, Princeton University Press,
Princeton, NJ (1962).
11. R. Rajagopal, 'Economics of.screening in the detection of organics in ground water', Chemometrics
Intell. Lab. Syst. 9, 261(1990).
12. R. H. Plumb, Jr., 'A comparison of ground water monitoring data from CERCLA and RCRA sites',
Ground Water Monitor. Rev. 7, 94 (1987).
13. Aqua Tech Environmental Consultants, Fee Schedule, Marion, OH (1988).
14. Wilson Laboratories, 1988 Price List, Salina, KS (1988).
15. U.S. Environmental Protection Agency, Proc. First Int. Symp. on Field Screening Methods for
Hazardous Waste Site Investigations, Environmental Monitoring Systems Laboratory, U.S. EPA,
Las Vegas, NV (1988).
16. R. Rajagopal and L. R. Williams, 'Economics of sample compositing as a screening tool in ground
water quality monitoring', Ground Water Monitor. Rev. 9, 186 (1989).
17. U.S. Environmental Protection Agency, Analytical method sensitivity improvement', in Inside
Story, Newsletter, Environmental Monitoring Systems Laboratory, U.S. EPA, Las Vegas, NV, p. 24
(1988).
18. Finnigan MAT, Introduction to Mass Spectrometry and Glossary of Mass Spectrometrv Terms, San
Jose, CA (1989).
19. Monsanto, The National Alachlor Well Water Survey (NAWWS): Data Summary, Technical
Bulletin, St. Louis, MO (1990).
-------
¦ Original Research Paper
Chemnaiftrii-s iintl irtteHiRtnt Ln/wrmnry Syuems. 9 II9W) 261-272
Elsevier Science Publishers B.V. Amsterdam
Economics of screening in the detection
of organics in ground water
R. Rajagopal
Departments of Geography and Civil and Environmental Engineering. University of ionma,
Iowa City, IA 52:42 (U.S.A.)
(Received 23 June 1989; accepted 1 June 1990)
Abstract
Rajagopal. R.. 1990. Economies of screening in the detection of organic* in ground water. CVwiu-wcino ami InhHi^ent laboratory
Systems,^'. 261-272.
Screening may be denned as (he use of efficient lesi procedures to determine and/of lo quantifv the presence or absence of
contaminants in samples, fn suppori vf screening, many commercial laboratories offer price-breaks when analysis for only a few
compounds are sought instead of a siundard scan (such as those required under a governmental regulatory program* Screening is a
siep-wise or a decision tree approach to sampling and analysis. For a fixed number of compounds in a screen, a model to maximize
the total expected number of positive determinations (detections) of organic compounds in ground water samples is developed and
evaluated. This model is tested w«h eucnsi'vc data on organic measurements obtained from ground-wafer sampies ji hazardous waste
iites. The economic viability or (he modd is evaluated wiifiin ihe ioniext of pries-breaks in analviical services offered b»v v;»mmcri-iLl
laboratories. For a set of price^break assumptions., scieening is shown to be far superior in ihe detection oi pesticide. acid. ar.J hjse
neutral compounds than in the detection of volatile organic compounds in ground ^xiter at hazardous ^aste site's. By incorporating
relevant QA/QC costs, it is further shown that screening ts a cost-effective alternative for all four group* of organic compounds. A
few promising avenues for further research are also outlined.
INTRODUCTION
There are no precise estimates of lotal national
expenditures for ground-water quality monitoring
in the United Slates. The sums spent to monitor
the 50000 or more ground-water-based drinking
water systems and the many thousands of wells
monitored under the various state and federal
regulatory programs is estimated to be in (he
range of several hundred millions to one or two
billion dollars a year. Under these circumstances.
it is not feasible to analyze for all possible con-
taminants in ground-water samples at the same
frequency at all sites. Therefore, the development
of cost-effective ground-water quality monitoring
and screening strategies based on laboratory, field,
health effects, economic, and regulatory perspec-
tives is a valuable topic for further research.
Much has been w-ritten about the complexity of
designing monitoring strategies based on an un-
derstanding of the physical structure of watersheds
and the mathematics governing the fate and trans-
0169-1439/90/SG3.50 C 1990 - Elsevier Science Publishers B.V.
-------
Chtfrn^nieirics ami mtetlisterii Laboratory Systems
pon of cantammants in watersheds (Ij. In site-
specific instances, knowledge of spatial.and tem-
poral structure of ground-water contaminants
could lead to efficient nework designs (2-5). For
man-made contaminants, reduction in cost can
also be achieved by incorporating the release pat-
terns of effluent sources into monitoring plans
{6-8}. Recent advances in instrumentation, field
and laboratory screening strategies, and the devel-
opment of surrogate methods and sample com-
positing techniques offer seldom-tapped opportun-
ities for achieving cost-red action through sam-
pling, design 19-12], Over fifty papers on a wide
range of topics such as fiber optics. X-ray fluores-
cence spectrometers, soil gas analyzers, im-
munochemical methods, portable gas chtomaio-
graphs. and espert systems for fieid and labora-
tory screening were presented at a recent interna-
tional symposium sponsored by the U.S. EPA (11].
Screening procedure? are diagnostic tools used
to determine the extent to which each sample
should be analyzed. They fit logically into the
realm of environmental monitoring. A general def-
inition. statement of objectives, and description of
the utility of screening as given in tef. 12 is as
follows:
Screening may be defined as the use of
rapid and relatively low-cost test methods or
procedures to determine whether a char-
acteristic is present or absent (go/no go),
present at levels above some threshold value
("qualitative), or present in some concentra-
tion range (semi-quantitative) in a popula-
tion of values or individuals. Frequently,
screening, is done to identify where to place
the emphasis in subsequent sampling pro-
grams. i.e. :o tain a preliminary understand-
ing of the system to guide a cost-effective
follow-up required for confirmatory analysis.
Screening is being used increasingly in field
monitoring programs to minimize the cost of
"environmental zeros" i.e.. it helps con-
centrate on samples containing information
about exposure hazards, standard violations,
etc., based on the presence of toxicants. In
most coses, the cost of chemical analyses is
the same whether or not the compound of
interest is detected in the sample. If screen- -
ing procedures can suggest which samples
should be collected to find the parameters of
interest, more efficient sampling, designs
would result.
Screening is especially attractive in those situ;
(ion5 where the probability of occurrence of an
compound (from a group such as volatile organa
or pesticides) in a collection of samples is smai
and the relative probability of occurrence of a fe-
compounds is significantly larger than the prof
ability of occurrence of the remainder of the grouy
Extensive analysis of ground-water quality dai
from ambient, drinking water and hazardous wasi
sites have revealed patterns in t it occurrence an.
co-oocurrence of compounds with ihe above-men
tioned properties {13 — 17). By utilizing the co-oi
currerice patterns, a screening strategy for organic
in ground water is developed, evaluated, and pre
senled in this paper.
CO-OCCURRENCE PATTERNS AND SCRStVNtNt
METHODS
Many researchers have classified and rankc-i
organic compounds based on their physical am
chemical properties such as mobility, sorption
volatility, and degraduiion: environmental am
health hazards; production and distribution ii
commerce: and the frequency of detection in vari-
ous media [1.13.16-21], Screening, as described ii
this paper, is an ordered approach for the analyst
of samples based on a ranking of total detection-
(occurrences and co-occurrences*. The derivation
of the order of sequential selection of coiisiiiucutr
in a group, such us volatile organic compound.-
(VOCs). base neutrals, acids, and poticides, mj a.-
to maximize a detection objective, is an iulere.sl-
ing mathematical problem [22], In screening, the
results of the first stage help determine the scope
of analysis in the second stage: the results of the
second stage help determine the scope of analysis
in the third stage; and so on. Such an approach by
utilizing information from prior stages guides the
decisions of future stages [8.23].
-------
¦ Original Research Paper
263
Mathematics of screening
There is an extensive body of literature on
staged or sequential implementation of sampling
plans in industrial quality control and the related
field of dynamic programming [24-26]. Adopting
some of the basic ideas offered in such literature,
screening can be described as a decision-tree or
feedback-based approach to the analysis- and in-
terpretation of sampling results. Exhaustive analy-
sis (i.e. analyzing samples for all constituents in a
list, such as that of EPA's Priority Pollutants
listed under the Clean Water Act), on the other
hand, is a case of minimal or no screening.
If several compounds of interest are detected in
ground water, how does one go about selecting a
few of these compounds as screens to be analyzed
without significant loss of overall information?
This problem can be simplified by framing the
question within the context of four groups of
priority pollutants, listed under the Clean Water
Act, such as VOCs. base neutrals, acids, and
pesticides. A discussion of models of screening
and associated computational procedures to maxi-
mize the total expected number of positive detec-
tions, for different screen sizes, is presented by Li
and Rajagopal [22]. The performance of one such
model of screening for four groups of organic
compounds is evaluated in this paper with exten-
sive field data from hazardous waste sites. A com-
parative analysis of the performance of screening
methods in several ground-water environments is
presented by Rajagopal and Li [15].
Economics of screening
In support of screening, many commercial
laboratories offer price-breaks when limited analy-
sis of a few compounds (or screens) insiead of a
standard scan is requested. Using ground-water
quality data from 314 hazardous waste sites and a
set of price-break assumptions, the cost effective-
ness of the screening approach is compared and
evaluated with reference to the present practice of
exhaustive analysis.
Under the provisions of RCRA and Superfund
regulations, extensive ground-water quality data
have been collected from several hundred
hazardous waste sites across the United States.
Using data from site investigations conducted dur-
ing 1981-1984. the Environmental Monitoring
Systems Laboratory of the U.S. EPA in Las Vegas
has organized a comprehensive data base [19|. It
contains the results of analysis of more than 700
organic compounds. These results are based on
over 15000 ground-water samples from 4772 wells
at 314 sites (143 RCRA and 171 CERCLA sites).
It is one of a few readily available computer files
that includes significant data on organic contami-
nants from a large number of wells at hazardous
waste sites. The data base was culled to include
only those wells in which all of a group of 31
VOCs, 45 base neutrals, 11 acids, and 26 pesti-
cides (mostly priority pollutants listed under the
Clean Watei Act) were analyzed and reported.
Analysis of this data base indicates that only a
fraction of the organics analyzed in ground-water
samples from hazardous waste sites are ever de-
tected. Such results have also been observed and
reported in the literature by Plumb [19], Shackel-
ford and Cline [14], Gibbons [27], and many others.
Further analysis has also revealed an interesting
and economically important pattern which can be
summarized as follows: w hen a particular set of 15
to 20 organic substances are absent in a well,
another particular set of 120 to 130 other organic
substances are also absent from such a well. This
observation is also supported by measurements
taken from ambient and drinking water supplies
[16,17]. Such findings, based on non-occurrences
(or a function of occurrences, non-occurrences
and co-occurrences), can be useful for ranking the
occurrence of orstunics in ground water. The cost
effectiveness of screening approach using such a
ranking procedure is developed and evaluated
against a program of exhaustice testing for
organics in ground water in the next section. No
assumptions regarding the spatial or temporal
structure of ground-water quality measurements
are made in this paper. Lack of such assumptions
enables economic analyses of the worst case
scenario. Of course, if detailed data on spatial and'
temporal structure were to become available, the
economic framework presented in this paper can
be considerably strenghthened by the use of ap-
propriate geostatistical and time-series analysis
-------
Cliemomtirics unj IrHeHieciu Latvrai^rv S.v>t«;uv> ¦
¦vethcds (suck as those reponed bv EnaJund and
FLatman [5] and Bergman jnd Quimby (3]).
The problem: If only a few compounds are
delected in ground water at hazardous, wasie sites,
how does one go about ranking or ordering com-
pounds for use rrt screenm^?
Tfl simplify the problem, the qutMion cat. tie
{rimed u-i'ibin the corsiesr of four selened
>>r organic compounds - 31 VGC&. 45 ba^e neu-
iraK Jt adds, and 26 pesticides falinosi nil of
which are recognized as priority pofi'x/arrn under
she Clean Water Act). Let us first consider the
case of VOC occurrences, '.r, v.aitT samples, from
1565 wells at hazardous a an te sues. Frequencies
of occurrence and co-occurrence were computed,
sorted by compounds in descending order of iota!
nimiberof detections (hits or occurrences) and
presented in Table 1. For exampJe. iricfiloroelhyS-
t'oe 'TCEj. the first entry in Table 1, was detected
in 2S9 of 1565 wells. Among [hose 289 u-elli. 8?
Initd poi'!. ve for soiuene, 322 for mei'ny'iene chlo-
ride. 114 for 1,2-iraiis-dichloroettiytene, and so on
along 1 lie firil row of frequency or""co-occurrences
presented in Table 1. If a complete VOC scan ol
the 31 compourtd?. considered if! this .vrirvii- 'luif
been carried otrt only for samples from i'it 2K4
wells, a tola! of 1551 positive detections would
have occurred fas 5.ho>T. under ihe column ."
totals for TC£ in TaWe I\ or >?.93-- of ine wis! &•
2S25 positive dejsctiofl-s reported in da!a fcuse.
After removing ihe records of the 2SS wells in
w-hich TCE jC.-LTted. ri found tltut ioJuci-.l-
occurred in I>4 wells "Mi iirwl ca-cccu(Teus.cr-
totaling, 422 other positive determinations. Hence,
selecting toluene as the second screening com-
pound ti-ouJd hat e resulted in a to Ml cf 576 posi-
tive deletions (as shown in Table !)¦ TExis. in turn
would coEitribme to an additional 20.4% of ilw
fatal". Thus, if a complete VOC se;?n of the 3;
compounds had been done for the 289 TCE ivells.
followed by the 154 toiuene wells. 753% of zhc
total positive detections would have been cap-
tured, The percentage and cumulative percentage
capture of total detections a* a function of the
num'cer of V'OCs included in the screia at£ shtw,\
itt Fig. 1. The results of similar anafyscs are also
presented for base neutrals, acids, arid pesticides
in Tables 2. 3, and 4 and Figs 2, X and 4.
table i
Oveurfcnce. ji.>-ncL'oirt'r>t;£. jr.d performaiK-e- statistics t Joint. percentage, an.cS curvuTu'rivc > £v .jjj
^ Lot-'kbetJ Hnvirorm-i'jnl^ M.m;nvmi;!ji
and Services). La> Vegas. NV. 19S7.
V,^|uiiJc iargJir-it-' I
•v^irnpvUifui
3
l
4
5
tr
T
8
9
10
^ H\ts
ri Cumuljii
fins
1. TriehlrnviYiylene jTCE)
S"?
IU
132
M
*3
l2\
IU
i 55 i
y
2. "Folurnc
(S4
St:>
?\
CS
21
21
6S
1ST
57f.
^.4
?5..:
A Mcshykiu; v hlu; ic
[52
1
1
r
*>
b
%}
S.4
>15 -i
4, iJS-jr*wh-D<«AVTf.K-lhylcne
55
q
21
j
5 5
1,V.
'¦r" ,'r
5. l,i.{-TrkftUKtVth:nic
41
2
4
u
MX
c. t'h\*r'5
3. 3t\igeiw
16
0
0
)>
1.V1
9, Tctras:h?firoc[.hyFertf
2d
0
4
14
m.u
nviri-deltfwtS-
ftAl
vm
\m
13:4
I.Wi
1VTA
321^
4^69!J
T:»w3 -aaalv-'ifi 1.565
15c»5
lib5
15^5
15^5
1565
\CA5
*565
1565
1565
4W5J5
-------
Original Research Paper
265
Notations and explanations
/V = Total number of samples
n(i) = Number of samples in which the /th
screening compound is expected to
occur in the absence of the previous
(/— 1) compounds of the sequence.
For example, in the case of VOCs.
/i(1) = 289. /i(2) = 154. n(3) = 152.
and so on from Table 1
= Fraction of samples in which the / th
screening compound occurs in the
absence of the previous (/' - 1) com-
pounds of the sequence
= /!(/ )//V [289/1565 = 0.185 for TCE.
154/1565 = 0.098 for toluene, and
so on from Table 1]
= Cost of analyzing a standard scan
(such as VOCs. acids etc., required
under a regulatory program) in a
sample
= Cost of analyzing the first com-
pound in a screen
= Cost of analyzing each successive
compound following the first one in
a screen
= Cost of analyzing m compounds, in-
cluding the first one in a screen
= Cfl + (m - 1) X
= A fraction of additional cost per
sample due "to the collection and
handling of larger volumes of water
for use in screening (expressed as a
fraction of the cost of a scan)
CE = Total cost of exhaustive analysis by
standard scans
= NC^
CSEQ\(i) = Total cost of analysis by screening
when an ordered set of i compounds
are used in the screen
/
= /V[Cn+(,-i)*]+ £ Ky)Q.an
+ -'VCJC>tun
/?1(/) = Relative cost of a program using i
screening compounds vs. exhaustive
analyses
= CSEQ\{i)/CE
= {/V[Q1+(, -1)^]h- I n(j)C^m
+ 'VCJCscaJ{*QJrl
/
= {[cfl+(/-i)^]+ £ /(y)Q.an
+ CaQ,n){Cscan}-' (1)
CSEQ2(i) = Total cost of analysis by screening
using an ordered set of i compounds
and assuring all detections by at least
two analyses
= /V[Cn +
-------
TABU: 2
Occurrence, co-occurrcncc. and performance statistics (total, percentage, and cumulative percentage detections) for an ordered sequence of base neutral compound*
found at hazardous waste sites
Source of raw data: U.S. l.l'A. Environmental Monitoring Systems Laboratory (contract to Lockheed Environmental Management and Services), Las Vegas, NV. I9K7.
Bust: neutral
compound
1
2
3
4
5
6
7
8
9
10
11
12
i3
14
15
16
17
Total
% Hits
% Cun
hits
1. Bis( 2-elhylhexyl)
phlhulalc
167
26
7
7
1
1
5
26
10
14
1
6
4
0
2
1
26
304
55.4
55.4
2. Di-n-bulyl
phthalate
36
1
2
1
I)
1
0
11
2
1
1
0
0
0
0
6
68
12.4
67.8
3. Naphthalene
22
1
2
2
1
1
4
0
0
1
1
0
0
0
12
47
8.6
76.4
4. 1,2-Dichlorobenzene
15
0
0
1
0
0
0
0
0
0
0
0
0
18
34
6.2
82.6
5. Pyrene
5
0
0
1
0
0
0
0
0
0
0
0
21
27
4.9
87.5
6. Hexachloro-
henzene
5
0
0
0
0
0
0
0
0
0
0
7
12
2.2
89.7
7. Isophorone
10
0
0
0
0
0
0
0
0
0
1
11
2.0
91.7
8. Butyl benzyl
. phthalale
7
0
0
0
0
0
0
0
0
0
7
1.3
93.0
9. Diethyl
phthalale
6
0
0
0
0
0
0
0
0
6
I.I
94.1
10. Di-n-oclyl
phthalale
6
0
0
0
0
0
0
0
6
1.1
95.2
11. Acenaphthcne
5
0
0
0
0
0
1
6
1.1
96.3
12. Bis-(2-dichloro-
ethyl)ether
3
0
0
0
0
2
5
0.9
97.2
13. (Dimethyl
phthalale
2
0
0
0
1
3
0.5
97.7
14. 3.3-Dichloro-
hen/.idine
3
1)
0
0
3
0.5
9X.2
15. I)en/.idine
2
0
0
2
0.4
9K.6
16. N-Nitroso-
diphenylamine
2
0
2
0.4
99.0
17. Others
6
6
1.0
100.0
Total detects
l<)7
62
30
25
9
K
IK
41'
31
22
7
11
7
3
4
3
101
549
100.0
Total non-detects
952
1057
io«9
1094
1 110
1 111
1101
I07K
10SK
1097
1112
1 10X
1112
1 116
1115
1 116
32350
49X06
Total analyses
II 19
III')
1119
im
1 119
1 119
1119
1119
1 119
1119
1 1 19
1119
1119
1 1 19
1119
1119
32451
50355
-------
¦ Original Research Paper
267
s
s
3
to
3
Fig.
10 11 12 13 14 15 16
Number of Compounds In Screen
2. Percentage and cumulative percentage capture of hits (delects) vs. various screen sizes for base neutrals.
analyses for a two-fold QA case
= CSEQ2(i)/CE
= {N[Cn +(/ - \)X} +
i
2 £ n(j)Q.jn +
'VCaC_}{2,VQ,in}-;
I
= {[Cn+0-l)*] + 2i;/(y)Q,in
+ CaQ;in}{2Q.an}-''=l (2)
Cost-effectiveness comparisons
An explorative analysis of the cost components
in a sampling program showed that the relative
cost of a screeening strategy in comparison to
exhaustive analysis is dependent on the likelihood
of the presence of contaminants in sample "speci-
mens and the price-breaks offered by laboratories
for analytical services. To verify the strength of
such relationships, worked examples (one each for
VOCs. base neutrals, acids, and pesticides) are
presented and the results illustrated in Figs. 5 and
6.
Examples
The fractional /(/') values for a sequence of
screening compounds such as VOCs. base neu-
trals, acids, and pesticides are obtained from Ta-
bles 1-4, respectively. The /(<) values and selected
cost estimates are substituted in eq. (1) and (2) to
compute the Rl(i) and R2(i) values, for i ranging
from 1 to 10 for VOCs. 1 to 16 for base neutrals. 1
to 6 for acids, and 1 to 15 for pesticides.
Many laboratories (for example. Aqua Tech
Environmental Consultants [28] and Wilson
100
60-
40 ¦
20-
2 3 4 5
Number of compounds in the Screen
Fig. 3. Percentage and cumulative percentage capture hits (detects) vs. various screen sizes for acids.
-------
Clicmomelricx and I ntelligent Laboratory System-
cumulative percentage capture of detections
different screen sizes and the relative cost of
program at those sizes vary across the four cc
pound groups (VOCs. base neutrals, acids.
pesticides). For example, from Fig. 5. it can
seen that the use of TCE as a single screen
VOCs will lead to the capture of 54.9% of tc
detections at a fraction of 0.618 (or 61.8%) of
total cost of exhaustive analysis. On the ot:
hand, the use of phenol, pentachlorophenol. a
2.4-dichlorophenol as a three-compound scr<.
for acids will lead to the capture of 94.3% of to
detections at a fraction of 0.709 (or 70.9%) of i
total cost of exhaustive analysis.
As expected, the two factors that influer.
performance are the distribution of occurrenc
and co-occurrences of compounds within a gro
and the price-breaks offered by laboratories I
the analysis of a single or a partial list of cot
pounds in comparison to a full scan. For select
cost and distributional assumptions, screening
shown to be far more superior in the detection
pesticides, acids, and base neutrals than in t
detection of VOCs in ground water at hazardo
waste sites (Fig. 5).
Under the screening approach, a large numb
of positive determinations are verified by tv
analyses (once during screening and again from
standard scan). This provides considerable quali
TABLE 3
Occurrence, co-occurrence, and performance statistics (total, percentage, and cumulative percentage detections) for un order,
sequence of acid compounds found at hazardous waste sites
Source of raw data: U.S. EPA. Environmental Monitoring Systems Laboratory (contract to Lockheed Environmental Manngcnic
and Services). Las Vegas. NV. 19H7.
Acid
compound
1
2
.1
4
5
6
7
Total
T Hits
ri Cumulative
hits
1. Phenol
102
J
K
13
4
2
22
154
67.2
6 7.2
2. I'entachlorophenol
44
• 0
1
1
0
1
47
20.5
N7.7
.V 2.4- Dichlorophenol
y
0
3
0
3
15
fc.fr
94.3
4. 2.4-Dimethylphenol
5
0
0
0
5
2.2
96.5
5. 2.4.6-Trichlorophenol
4
0
0
4
1.7
98.2
6. 4-Nitrophenol
2
0
2
0.9
99.1
7. Others
2
2
0.9'
100.0
Total detects
102
47
17
19
12
4
28
229
100.0
Total non-detects
989
1044
1074
1072
1079
1087
5427
11772
Total analyses
1091
1091
1091
1091
1091
1091
5455
12001
Laboratories [29]) offer price-breaks based on the
number of organic compounds included in the
screening list and quantity discounts based on the
number of sample specimens submitted for analy-
sis. There is considerable variation in price quota-
tions. detection limits, and other services offered
by the different laboratories. To illustrate and
evaluate the efficiency of screening, typical prices
quoted by the laboratories are adapted and used
in the following cost estimates:
Cost estimates
VOCs :Q,,n = S180; Cri = $ 60:
*=510; C, = 0.1
Base neutrals : Cscan = $320; Cf, =$100;
JT = $10: Ca = 0.1
Acids :Cscan = SI50; Cn = $ 50;
A'= $10; Ca = 0.1
Pesticides :Csi..jn = S200; Cn = $ 50;
X = 510; C3 = 0.1
Discussion and analysis of results and limitations
Cumulative percentage capture of detections
for the four compound groups given in Tables 1-4
are plotted against the computed R\(i) values and
shown in Fig. 5. From Fig. 5. it is evident that the
-------
¦ Original Research Paper
269
100
80-
60"
40 *
20 *
a
a
>
s
9 10 11 12 13 14 15
Number of Compounds In the Screen
Fig. 4. Percentage and cumulative percentage capture of hits (detects) vs. various screen sizes for pesticides.
checks without any additional costs. This added
advantage of quality checks which is inherently
available in a screening strategy was not incorpo-
rated in either eq. (1) or Fig. 5. To account for
such quality checks, both the numerator and the
denominator of eq. (1) are modified so that each
detection is verified by at least one additional
analysis. The resulting expression R2(i), defined
by eq. (2), is plotted against the cumulative per-
centage capture of detections for the four com-
pound groups and shown in Fig. 6. In the context
of such an interpretation, screening for all four
compound groups is shown to fare extremely well
in cost comparisons (Fig. 6).
The exact distribution of occurrences and co-
occurrences of compounds within groups such as
VOCs, base neutrals, acids, or pesticides is not
known a priori. But there is overwhelming evi-
dence to show that only a few organic compounds
are ever detected in ground water; those detec-
tions occur in a small number of samples, and the
concentration of most of those detections are be-
low a few ng/1 [15-17,19,27], Such knowledge
alone is sufficient to seriously consider and
evaluate the utility of screening methods in
ground-water quality monitoring. Of course, the
lack of knowledge of spatial and temporal struc-
ture of ground-water contaminants can be consid-
Relative Cost of Screening
(Cost of Screening vs. Exhaustive Analysis)
Fig. 5. Cumulative percentage capture of detections as a function of screening efficiency at different screen sizes.
-------
TABLE 4
Occurrence, co-occurrence, and performance statistics (total, percentage, and cumulative percentage detections) for an ordered sequence of pesticide compounds found
at hazardous waste sites
Source of raw data: U.S. EPA, Environmental Monitoring Systems Laboratory (contract to Lockheed Environmental Management and Services), Las Vegas, NV, 1987.
Pesticide
compound
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Total
% Hits
% Cumu-
lative
hits
1. BHC-y
35
3
9
4
2
0
20
2
8
18
1
0
0
21
4
11
138
55.0
55.0
2. Endrin
10
1
2
0
2
0
1
2
1
2
0
0
1
2
3
27
10.7
65.7
J. Dieldrin
10
0
0
IJ
1
0
0
0
0
0
1
0
2
0
14
5.6
71.3
4. Chlordane
10
0
1
¦ 0
0 .
0
0
0
0
0
1
1
1
14
5.6
76.9
5. PCB-1260
11
0
0
0
0
0
0
0
0
0
0
0
11
4.4
81.3
6. Endosulfan sulfate
.9
0
0
0
0
0
0
0
0
0
1
10
4.0
85.3
7. BHC-/3
6
0
0
0
0
0
0
1
1
1
9
3.5
88.8
8. Endosulfan la
4
0
1
0
0
0
0
1
1
7
2.8
91.6
9. 4.4-DDE
' 3
0
0
0
0
0
0
2
5
2.0
93.6
JO. BHC-S
3
0
0
0
1
0
0
4
1.6
95.2
11. Hepiuchlor epoxide
3
0
0
0
0
0
3
1.2
96.4
12. PCB-1242
3
0
0
0
0
3
1.2
97.6
13. PCB-1254
2
0
0
0
2
0.8
98.4
14. BHC-o
1
0
0
1
04
98.8
15. Heptachlnr
1
0
1
0.4
99.2
16. Others
2
2
0.8
100.0
Total delects
35
13
20
16
13
12
27
7
13
23
6
3
3
26
12
22
251
100.0
Total non-detects
996
I01K
1011
1015
1018
1019
1004
1024
1018
1008
1025
1028
1028
1005
1019
11319
26555
Total analyses
1031
1031
1031
1031
1031
1031
1031
1031
1031
1031
1031
1031
1031
1031
1031
11341
26806
-------
¦ Original Research Paper
271
3
u
V
I
0
V
1
a
<3
100.0
90.0
70.0
60.0
50.0
Acids ^
Pesticides r r( /
VOCa
f/i /
Base/Neutrals
0.2
Fig. 6.
case.
0.4 0.6 0.8 1.0
Relative Cost of Screening
(Cost of Screening vs.. Exhaustive Analysis)
Cumulative percentage capture of detections as a function of screening efficiency at different screen sizes in two-fold Q/A
erably strengthened and fine-tuned by additional
periodic, site-specific, product-specific, and en-
vironment-specific data, if and when they become
available.
CONCLUSIONS AND NOTES ON FURTHER RE-
SEARCH
A framework for the economic analysis of
screening methods in ground-water quaiity moni-
toring is presented in this paper. Preliminary re-
sults indicate that screening is relatively more
cost-effective for the detection of pesticides, acids,
and base neutrals than for the detection of VOCs
in ground-water samples from hazardous waste
sites. The relative degree of effectiveness is shown
to be a function of observed patterns in occur-
rences and co-occurrences of contaminants and
the price-breaks offered by laboratories for the
analysis of contaminant groups. In screening, a
large number of positive determinations are veri-
fied by two analyses (once in the process of
screening and again by a standard scan). This
provides considerable quality checks without any
additional costs. If appropriate costs are allocated
to such added quality checks in exhaustive analy-
sis. the concept of screening in ground-water qual-
ity monitoring will fare extremely well in any cost
comparisons.
Characterizing the spatial and temporal distri-
butions of compounds (within groups such as
VOCs. acids, base neutrals, and pesticides) in spe-
cialized environments of 'ambient, drinking water
and RCRA/CERCLA sites and developing
screening approaches based on such knowledge is
a major area for further research. Such approaches
can be further strengthened by additional site-
specific and product-specific data and fate and
transport models. Developing mathematical mod-
els for screening and implementing computer al-
gorithms for their solution are also important areas
for further research [16.22].
Uncertainties in analytical results from both
screening and scanning approaches could lead to
false-negative or false-positive inferences. Incor-
poration of such uncertainties in models of screen-
ing is ah important area for further research.
ACKNOW LEDG KM F.N'TS
I wish to thank Ping-Chi Li of the Department
of Geography at the University of Iowa for pro-
viding considerable computational and graphical
assistance in the preparation of this manuscript.
Numerous constructive comments by the Editor
P.K. Hopke and three referees to this journal and
William H. Engelmann of the U.S. EPA are
sincerely appreciated. Joseph J. D'Lugosz of
-------
Chemoawrics and intelligent Laboraiorv Svsioms ¦
EMSL-LV, U.S. EPA is the Project Officer re-
sponsible for the Cooperative Agreemeni jrioer
which the research reported in this paper was
carried out.
NOTICE
Although the research described in this
manuscript has been funded wholly or in pan by
the United States Environmental Protection
Agency through a Cooperative Agreement (CR
813552-01-0) with the University of Iowa, it has
not been subjected to Agency review and therefore
does not necessarily reflect the views of the Agency
and no official endorsement should be inferred.
REFERENCES
1 P.M. Chapman. G.P. Romberg and G.A. Vieers. Design of
monitoring studies for priority pollutants. JournaI oj rhe
Water Pollution Control Federation, 54 (1982) 292-297.
2 K.W. Brown. J.W. Mullins, E.P. Richitl. G.T. Flatman.
S.C Black and S.J. Simon, Assessing soil lead contamina-
tion in Dallas. Te*as. Environmental Monitoring and .45-
sessment, 5 (1985) 137-154,
3 L.E, Borgman and W.F. Quimhv. Sampling for tests of
hypothesis when data are correlated in space and time, in
i_H. V.eiibi I El.ilCr). Prinapies of Enaror^v.snt^f
The American Chemical Society. Washington. DC. i988.
pp. 25-43.
4 T.H. Starks. A mode! approach for RCRA ground water
regulations. Chemometrks and Intelligent Laboratory Sr.v-
tems. 3 (198S) 15U157.
5 E.J. EngJund and G.T. Flaiman. Annua! Report of Research
it) Environmental Statistics. Geostaiistics. ami Chemotnef-
rws. U.S. Environmental Protection Agency.. EMSL-LV.
Las Vegas. NV. 1990, 60 pp.
6 S.G. Chainherlairu C.V. Beckers. G.P. Grimsrud and R.G.
Shulf. Quantitative method* for preliminary design of water
qualitv surveillance systems. Water Resources Bulletin. 10
0974)199-219.
1 A.l. Cohen. „Y. Bar-Shalum. W. Winkler and G.P. Grims-
rud. A quantitative method for effluence compliance moni-
toring resource allocation. Report So. EPA• 61)0/S--OJA
U.S. EPA. Washington. DC. 1975.
8 R. Rajagopal. Optima) sampling strategies for source iden-
tification in environmental episodes. Ewironmenioi Moni-
toring and Assessment* 4 (1983] 1-14.
9 C.E. Parker. Surrogate parameter analysis for organic prior-
ity pollutants. Journal of Water Pollution Control Federa-
ticn, 54 (1982) 77-86.
10 l.H. Souffet. J. Gibs. J.A. Coyle. R.S. Chorbak and T.L.
Yohe. Applying analytical techniques to salve groundwater
contamination problems. Journalo} American W/jter II arks
Association. 77 (1985> 65-72.
11 U.S. Environmental Protection Agency. Proceedings of the
First International Symposium on Field Screening Methods
for Hazardous Waste Site investigations. Environmental
Monitoring Systems Laboratory. U.S. EPA. Las Vegas. NV,
1988.
12 R. Rajagopal and L.R. Williams. Economics of sample
compositing as a screening tool in ground water quality
moniioring. Ground Witter Ifoniioriniz Ret9(!) (19S9)
186-192. "
13 J.J. Wcstrick. W. Mello and R.F. Thomas, The ground
water supply survey. Journal of American Water Warks
Association. 76(5] (.1984} 52-59.
[4 tV.M. Shacfce/ford and D.M. C/ine. Organic compounds in
water. Environmental Science and Technology, 20 (1986]
652-657.
15 R. Rajagopal and P C. Li. A comparison of two screening
methods for the detection of VOCs in ground water. Jour-
nal ^of Cheniometrics. in press.
16 P.-C. Li and R. Rajagopal. Modeling the occurrence of
volatile organic compounds in ground water. Journal of
American Water H7>r£j Assotiation. submitted for publica-
tion,
]7 U. Natarajan and R. Rajagopal. Determinants of pesticide
occurrence in ground water. Journal of American Water
Works Association* submitted for publication.
18 W.A. Pettyjohn and A.W. Hounslow. Organic compounds
and ground water pollution. Ground Water Monitoring Re-
view. 3(4) (1983) 41-47.
19 R.H. Plumb, Jr.. A comparison of ground water monitoring
aula from CERCLA jrid RCRA sites. Grctcrd Wafer
tonng Rtrcie**\ *J(.4) C1VS7"! 94-I0C-.
20 E. Haifon and M.G. Reggiani. On ranking chemicals for
environmental hazards. Environmental Science and Technol-
ogy. 20 0986) 1173-1179.
21 J.E. Lucius. Physical and Chemical Properties and Health-
Effects of Thirty-Three Toxic Organic Chemicals. U.S. Geo-
logical Survey. Open-File Repon 87-428. Denver, CO, 1987.
22 P-C. Li and R. Rajagopal. Mathematics of sequential
HTcenmg. Paper under internal review.
23 L.G. Miiien. An analytical solution to the least cost testing
• sequence problem. The Jvuruithd Industrial Fn'^trccrin^. II
(I960) 17.
24 A. Wald. Sequential Anulvsis. Wiley, New York. 1947.
25 R.E. Bclfnwn. Dimvnk• Pn>vrtinmiini:. Princeton Uimvr>ilv
Press. Princeton, NJ. 1^7.
26 R.E. Bellman and S. Dreyfus. Applied Dynamic Pm^ram-
min$. Princeton University Press. Princeton. NJ. 1962.
27 R.D. Gibbons. Statistical models for the analysis of volatile
organic compounds in waste disposal sties. Ground Water,
25 (1987) 572-580.
25 Aqua Tech Environmental Consultants. Fee Scbvd-jfe. Ma-
rion. OH, 1988.
29 Wilson Laboratories. J 988 Price List. Salina. KS. 1988.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 22
Hydraulic Probe Soil Sampling and Field VOC Analysis by Gas
Chromatography
G. Hunt Chapman and Jeffrey C. Tuttle, P.G.
ENVIRSURV, Inc., Fairfax, Virginia
January 12-14, 1993
Las Vegas, Nevada
-------
HYDRAULIC PROBE SOIL SAMPLING AND FIELD VOC ANALYSIS BY
GAS CHROMATOGRAPHY
G. Hunt Chapman and Jeffrey C. Tuttle, P.G.
ENVIROSURV, INC. - Fairfax, VA
ABSTRACT
The traditional approach to subsurface soil and groundwater sample acquisition and
analysis involves use of a drill rig and sample shipment to a fixed laboratory. Drill rig and
fixed lab costs however, often preclude collection of an adequate amount of data to fully
characterize site conditions. In addition, it generally takes weeks to months before Fixed
lab results are available with which to make important extent-of-contamination and risk
assessment decisions. An alternative site sampling and analysis approach is discussed in
this paper; namely the use of truck-mounted hydraulic probing equipment in combination
with on-site mobile laboratory analysis.
This approach was used recendy at a fonmer solvent disposal site at the Portsmouth
Gaseous Diffusion Plant in Piketon, Ohio. The work was completed under the direction of
Oak Ridge National Laboratory and Martin Marietta Energy Systems. Two sampling aiid
analysis phases were conducted. An initial site characterization phase included collection
and analysis of approximately 200 soil samples taken at multiple depths of up to 27 feet
These results were used to design a second phase-the purpose of which was to study the
destruction efficiencies of several in-situ soil treatment technologies. Approximately 500
pre- and -post-treatment soil samples were taken and analyzed on-site for seven chlorinated
contaminants. The results were used in the field to "fine-tune" the treatment processes and
measure the destruction efficiencies of each option..
This paper discusses field laboratory and hydraulic probe capabilities, advantages and
limitations, and how the probing/mobile lab concept was applied during the soil treatment
feasibility study recently performed at the Portsmouth Gaseous Diffusion Plant
INTRODUCTION
Innovative cost- and time-efficient approaches to environmental sampling and analysis have
evolved rapidly in recent years. The "traditional'' approach to subsurface soil sample
collection involves use of a drill rig. The traditional approach to sample analysis involves
shipment of samples to a fixed laboratory. It is suggested that the drill rig and fixed lab
approach to contamination assessment work is not always the most appropriate to meet
project objectives. Drill rig and fixed laboratory costs often preclude collection of an
adequate number of samples to fully characterize site conditions. In addition, fixed lab
analyses generally take weeks to months before data is available with which to make
important extent-of-contamination and risk assessment decisions.
At many sites, once an initial sampling investigation has been performed, a small number
of contaminants of concern can be identified. Subsequent site characterization efforts can
focus on these target contaminants without necessarily incurring the costs of full spectrum
analyses. Using gas chromatographs equipped with selective detectors (FID. PID and
ECD), samples can be quickly analyzed for target compounds. QA/QC requirements can
be established to meet site-specific Data Quality Objectives (DQOs) and to insure that each
data set is internally consistent and self-supporting.
-------
Recently developed subsurface hydraulic probing techniques allow for quick and relatively
inexpensive soil sampling. The equipment mounts in a 4-wheel drive pickup truck, and is
powered hydraulically from the vehicle's engine. Hardened steel probe rods (3/4" O.D.)
are driven to depths of up to 40-plus feet, depending on soil conditions. Soil samples can
be taken using a small diameter "piston-drive" shelby tube. Both the vertical and horizontal
extent of soil contamination can be readily determined. Fifteen to twenty-five samples can
be collected and analyzed in a single day, depending on the mobile lab method employed
and on subsurface stratigraphic and hydrogeologic conditions.
Truck-mounted hydraulic probing equipment in combination with on-site mobile laboratory
analysis was used during two studies at site X-231B at the Portsmouth Gaseous Diffusion
Plant in Piketon, Ohio. The initial soil sampling and field laboratory analysis work was
completed in January, 1992. The objectives of the January investigation were to determine
both the horizontal and vertical distribution of chlorinated solvent contamination in soil. A
total of 192 samples were collected at 3-foot intervals to depths of approximately 14 to 22
feet (4 to 6 per boring). The soils data collected was used for baseline characterization of
contaminant levels.
A second sampling episode took place in April/May, 1992. The project focused on pre-
treatment and post-treatment evaluation of several in situ soil remediation technologies;
namely in-place soil mixing coupled with either thermal extraction, chemical oxidation or
immobilization processes. The location of "treatment cells" for the various in-situ
technology evaluations were selected using the initial January baseline data as interpreted
through three-dimensional contouring and imaging software. A total of 467 pre- and post-
treatment field samples were collected and analyzed on-site in the mobile laboratory.
All soil samples in both investigations were analyzed for the following seven target
compounds using the headspace technique:
HYDRAULIC PROBING EQUIPMENT
Four-wheel drive truck-mounted hydraulic probing equipment has been developed in recent
years which permits "non-drilling" soil-gas, soil and shallow groundwater sample
acquisition. Advantages of this equipment over conventional drilling equipment include:
• Lower costs
• Improved site access
• Unobtrusive, low profile sampling
• Rapid sample collection
• No investigation-derived waste (e.g., drill cuttings and purge water)
• Collection of "undisturbed" in-situ soil and groundwater samples without
construction of permanent monitoring wells or piezometers
Target Analvte
Method Quantitation Limit
• 1,1-Dichloroethene
• Methylene Chloride
• trans-1,2-Dichloroethene
• 1,1-Dichloroethane
• cis-1,2-Dichloroethene
• 1,1,1-Trichloroethane
• Trichloroethene
1.0 Ug/Kg (ppb)
1.0 (Ig/Kg
1.0 |ig/Kg
1.0 jig/Kg
1.0 |ig/Kg
0.25 |ig/Kg
0.25 Hg/Kg
-------
Limitations, however, exist. The hydraulic probing equipment is only effective in
unconsolidated geologic materials. As a general rule-of-thumb, probing is possible under
conditions amenable to hollow stem auger drilling. Continuous core samples, however,
ant not reacLy abiahas-e. Bur^rten scrae subsurface.sirati.graphic information is already
available for a. .she, and ccnditian&ar: favorable rcrprobtr-g, Lie equipment aermits raaxi"
discrete-depth Scjn^.e col le^cn at 5iEci5.csx.ily lower costs =ax.pared k! conventions!
drilling.
Sampling tracks srz des.gr2 J for ;^s-sj~z.ecry Li Lie fide. Each itjcJc presides its own
power for the hydraulic probe emplacement system. Soil sampling equipment includes the
following items;
• Heavy duty 4-wheel drive pickup
• Rear mounted hydraulic probe driving nd removal system
• Rotary impact drill for asptialt/concrcie penetration
• Hardened steel sampling probes and accessories
• Probe/Accessory decontamination equipment
» Small-diameter shelby tube piston samplers for discrete depth soil sampling
• Extruder bracket and dowels for discharging soil directly into sample containers
PROBE DRIVING METHOD AND SOIL SAMPLE COLLECTION
Probe soil sampling requires the temporary installation of hollow steel rods into the
unsaturated soil zone. A plethora of equipment-exists for driving probe rods by hand using
a slide hammer, or a gas or electric powered rotary hammer. Probing using these manual
methods however, is generally limited to 10 to 15-foot depths and is time and labor
intensive.
Specially-designed hydraulic drive Jriits mounted in a vehicle are a mare efficient method
for probe installation, sample collection and probe removal. These hydraulically driven
units, generally equipped with percussion hammer attachments, afford a much greater
depth of penetration and the ability to hammer through "stiff" layers and drill surface
pavements, The result is that a greater number of probes and more samples can be obtained
ia a given amount of time, and for a given cost.
Undisturbed hydraulic probe soil samples are collected using a 1-foot long by 1-inch
outside diameter shelby tube sampler attached to the lead probe rod. Once the piston soil
sampler is driven to die top of the desired sampling depth, the piston is released via an
extensioniod inserted down ±e probe rod. With t.ne core bxtel free to move; the probe
rod is driven an additional 1-foot to collect approximately ISDgrams of soil. Probe rods
are then puLled and the core samples extruded, composited (with minima] disturbance),
containerized in pre-weighed 40 ml vials cleaned to EPA specifications, and immediately
delivered to the oil-site laboratory tor heacspa.ee sampfc preparation and subsequent
analysis.
ON-SITE MOBILE LABORATORY
EPA has, over the years, undertaken the development of a massive compendium of
analytical methods. The large nurabsrs cf contaminants regulated has necessitated
es tat I is inner J of generic neLKsds -^-li rac'ni-u-.aljie capaJbilJt.es-. Acd a. s truing en p h a^U
t*as been placed on .he jteai Jefctsilriity foe 5cperfund das.as.evidenced ic Lbs
4ocuTr\tftUKion and camoeraarca qca.' ~;," oar.irai require tntr-is of the Contract. La t> Pnrt-jtxi
-------
(CLP). This has resulted in long turnaround times (weeks to months) for sample data, and
significant costs. Mobile laboratory analysis offers significant advantages over the fixed
lab status quo.
Laboratory-grade gas chromatography equipment has been down-sized and ruggedized in
recent years to operate in confined areas and withstand field operating conditions. When
placed in a controlled environment, i.e., a mobile laboratory trailer with HVAC, standards
refrigeration and sample preparation equipment, you essentially have an on-site fixed
laboratory. Once target contaminants of concern or indicator compounds are known for a
given site, complete EPA laboratory analytical method protocols may be unnecessary.
Field analytical methods have been developed in recent years to permit rapid, relatively
low-cost screening for volatile and semi-volatile organic compounds. Screening results in
the field permit significant advantages over "delayed" fixed laboratory analyses:
• Larger numbers of relatively cheaper analyses can be used to compile data that is
more representative of site contamination.
• Real-time feedback of sample results allows for sample plan modification as
needed
• Extent-of-contamination can be defined without multiple rounds of sampling and
numerous return trips to a site.
• Cleanup/removal operations can proceed at a rapid rate, with limited down-time of
treatment/excavation contractors waiting for lab results.
• Sample integrity can be maintained without concerns regarding cross
contamination or degradation during sample shipment (a real concern with labile
volatile contaminants).
Mobile laboratory trailers are designed for self sufficiency in the field. Each trailer can run
on either generator-supplied or "shore" power. Power requirements for complete
laboratory operation are two 110V 15 amp circuits which is readily available on or near
most sites. Analytical instrumentation utilized is laboratory grade, and is supported by
computer driven chromatographic data collection and reporting systems. Mobile laboratory
features and instrumentation are listed below:
Mobile Laboratory Trailer
• Air suspension system and tie-downs for sensitive instrumentation
• HVAC for a controlled temperature environment
• Lab bench space for sample preparation
• Explosion-proof standards refrigerator
• Fume hood for solvent extraction procedures
Analytical Instrumentation:
• Laboratory-Grade Gas Chromatograph equipped with selective detectors (Flame
Ionization [FID], Electron Capture [ECD] and Photoionization [PID] Detectors)
and megabore capillary GC columns
• Chromatographic data system equipped with 20 MB hard disk and printer for data
storage and hard-copy output
• Safety Equipment: fire extinguisher, eye wash, first aid kit, etc.
-------
BIBLIOGRAPHY
1. Chapman, H. and Clay, P., Field Investigation Team (FIT) Screening Methods and
Mobile Laboratories Complimentary to Contract Laboratory Program (CLP), Ecology
and Environment, Inc., U.S. EPA Contract No. 68-01-7347, 1986.
2. Tuttle, J., "Site Inspections in Support of Superfund's Revised Pre-Remedial
Program", Proceedings of the 5th National Conference on Hazardous Wastes and
Hazardous Materials, Las Vegas, NV, HMCRI, Silver Spring, MD, pp. 273-277, 1988.
3. Tuttle, J. and Chapman, H., "Field Analytical Screening, Reconnaissance Geophysical
and Temporary Monitoring Well Techniques - An integrated Approach to Pre-
Remedial Site Characterization", Proceedings of the 5th National RCRA/Supetfund
Conference and Exhibition (New Orleans, LA), HMCRI, Silver Spring, MD, pp. 530-
537 1989.
4. Tuttle, J., Koglin, E., and Chapman, H., "Field Analysis and Site Characterization
Activities of the U.S. Environmental Protection Agency", Proceedings of the Third
International KfK/TNO Conference on Contaminated Soil, Karlsruhe, Federal
Republic of Germany, 1990;
5. Tuttle, J., "An Innovative Approach to Site Characterization - Combined Use of Field
Screening and Subsurface Probing Techniques", Ecology and Environment,
Inc.,WATTec '91, 18th Annual Technical Conference and Exhibition, Knoxville, TN,
1991.
6. Tuttle, J., Chapman, H., "Site Characterization Alternative: Hydraulic Probe Sampling
and Mobile Laboratory Analysis", Proceedings of the 13 th National RCRA/Superfund
Conference and Exibition (Washington, D.C.). HMCRI, Silver Spring, MD, pp. 228-
236,1992.
-------
Biographical Sketches
Gordon Hunt Chapman, Chemist
EnviroSurv, Inc.
Mr. Chapman is currently Vice President of Analytical Services at EnviroSurv, Inc. He
has had 20 years experience as an analytical chemist, specializing in environmental
analysis. He also interprets analytical data and advises clients as to the significance of
sample results and degradation by-products.
Formerly with Ecology and Environment (E&E), Inc., Mr. Chapman was the National
Project Manager for U.S. EPA-sponsored Field Analytical Support Project (FASP).
Before that, he worked for Frito-Lay, Inc., as National Quality Assurance Laboratory
Supervisor and at Normandeau Associates, Inc., as the Laboratory Director of the
environmental laboratory. He also worked at Texas Instruments as Supervisor of the
general laboratory section of Ecological Services, and at Biomedical Reference
Laboratories as Manager of Water Analysis Division.
B.S./B.S., Biology/Chemistry, Wofford College, University of North Carolina
M.S. Environmental Science, University of Texas
Jeffrey C. Tuttle, P.G., Hydrogeologist
EnviroSurv, Inc.
Mr. Tutde is currently Vice President of Hydrogeological Services at EnviroSurv, Inc.
He has had extensive experience in application of innovative sampling techniques for
soil-gas, subsurface soils, and groundwater. He also advises clients concerning sample
plan design, and interpretation of site geological conditions and fate and transport of
subsurface contaminants.
Mr Tuttle was formerly with Ecology and Environment, Inc. (E&E) as Technical
Operations Program Manager for E&E's Field Investigation Team (FIT) contract with
U.S. EPA. Before that he worked at the Nuclear Regulatory Commission as Project
Manager in the Low-Level Radioactive Waste Division. He has also worked at the
Phoenix Corporation as a Staff Geologist doing geophysical surveys for the mineral
exploration and hazardous waste industries.
B.S./M.S., Geology/Hydrogeology, West Virginia University
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
i1,^
Tab 23
An Evaluation of Four Field Screening Techniques For Measurement
of BTEX
E. N. Amick and J. E. Pollard
Lockheed Environmental Systems and technologies Company, Las
Vegas, Nevada
January 12-14, 1993
Las Vegas, Nevada
-------
AN EVALUATION OF FOUR FIELD
SCREENING TECHNIQUES
FOR MEASUREMENT OF BTEX
E.N. AMICK and J.E. POLLARD
LOCKHEED ENVIRONMENTAL SYSTEMS AND TECHNOLOGIES COMPANY
980 KELLY JOHNSON DRIVE
LAS VEGAS, NEVADA 89119
An investigation of field-portable technologies has demonstrated that analytical
methods for volatiie aromatic hydrocarbons are available which can provide an
investigator with data of'known quality in a timely, economical manner.
Volatile aromatic hydrocarbons are prevalent contaminants at hazardous waste sites
across the United States as a result of spillage of hydrocarbon fuels and leaking
underground storage tanks. Cleanup of these contaminated areas is a high priority
given current emphasis on environmental issues. Commonly used techniques for
measuring volatile compounds involve collecting field samples for shipment to a
laboratory for analysis. This process is time consuming, costly, and has a high
potential for error generation as a result of sample handling and transport. A faster
and potentially more efficient method for developing monitoring data for remediation
efforts is the use of field screening analytical methods. Using state-of-the-science
field analytical techniques, it should be possible to monitor a remediation effort for a
fraction of the cost of traditional laboratory-based analytical techniques.
Many field analytical instruments are commercially available. However, the lack of
reliable performance data is a problem when choosing an instrument. If performance
data is available, it is usually provided by the manufacturer and, as such, supports the
manufacturer's claims about the device. These claims are sometimes unrealistic.
Many of these instruments and the data they produce have not been thoroughly
examined by the scientific community.
The purpose of this study was to identify, select, and evaluate field methods for
analyzing benzene, toluene, ethylbenzene, and o-xylene (BTEX) in soil and water
samples. A literature search was made to identify current field analytical techniques
for the detection of BTEX in environmental samples. Four promising technologies
were selected to perform a laboratory evaluation for the purpose of providing
performance data on the quality of these technologies.
For each field method, accuracy, precision, and minimum detection levels were
compared to standard laboratory methods. Methods were also evaluated for ease of
-------
use, cost per sample and/or cost of equipment.
No single field methodology was found to be superior to others; each method has
advantages and disadvantages. Each technology and its assessment follow.
• The Antox™ immunoassay test is simple to perform and can be used as a
quick indicator of BTEX contamination in water. This test provides a reliable
qualitative indicator for BTEX for levels above 75 ppb.
• Detector tubes are simple to use and can provide a quantitative determination
for BTEX in water.
• The "Lab In A Bag™" sample extraction system provides a reliable method of
preparing water and soil samples for volatile hydrocarbon analysis. Low
detection limits (10 ppb) are achievable when used in conjunction with a
portable gas chromatography
• The MSI™ gas chromatograph provides accurate and precise quantitation for
BTEX. This instrument offers the advantage of chromatography: quantitation
of individual target analytes.
All the methods investigated can be used as presently available with little or no
modifications. This paper offers guidance to investigators choosing the proper
analytical methods to use for a particular problem. Method performance is presented
and advantages and limitations for the procedures are investigated.
-------
REFERENCES
Van Emon, J.M. and Mumma, R.O., eds., Immunochemical Methods for
Environmental Analysis.. ACS Symposium Series 442, American Chemical
Society, Washington, DC. 1990.
Amick, E.N. and Zimmerman, J.H., "Evaluation of Detector Tubes for
Determination of Volatile Organic Compounds in Water." Presented at Pittcon
1992, New Orleans, Louisiana, March 1992.
Mussmann, B., "Analysis for Environmental Protection - Measurement of
Chemical Contaminants in Soil and Water Samples by Means of Draeger
Tubes." Draeger Review, vol 51, pp. 17-19, April 1983.
Bather, W., "The Draeger Air Extraction Procedure - A Rapid Test to
Determine Pollutants in Water." Draeger Review, vol 61, pp. 9-17,
September 1988.
Robbins, G.A., Bristol, R.D., and Roe, V.D., "A Field Screening Method for
Gasoline Contamination Using a Polyethylene Bag Sampling System." Ground
Water Monitoring Review, pp. 87-97, Fall 1989.
U.S. Environmental Protection Agency, Office of Underground Storage Tanks,
Field Measurements - Dependable Data When You Need It. EPA/530/UST-
90-003, September 1990.
U.S. Environmental Protection Agency, Test Methods for Evaluating Solid
Waste. SW-846, 1986.
-------
• Miscellaneous Equipment: volumetric flasks, variable pipettes and disposable
glassware, etc.
• High-temperature drying oven for sample preparation and equipment
decontamination
• Electronic balance for weighing samples
• Heated water bath
• Personal computers for draft report preparation on-site
Laboratory Methods
The headspace technique is used to quickly analyze soil samples in the field. Headspace
testing is based on analyzing the air space above a containerized sample of soil or water for
partitioned VOCs. Soil and shallow groundwater headspace data is a better indicator of
actual soil or water contamination than soil-gas information in most cases, and with the
enhanced depth capability of the probing equipment described, soil and shallow
groundwater arc readily obtainable.
Diffusion is the principal driving force behind VOC vapor movement through the soil
matrix, and for partitioning of VOCs from soil into the void space of a partially filled
sample container. Compound-specific vapor pressures are what determine the ease of VOC
detection in headspace samples. VOC detection in soil is governed by compound-specific
air/soil partitioning coefficients (e.g., air concentration/soil concentration). The relative
ease with which a VOC partitions or moves out of the solid phase into the gaseous phase is
dependent on the compound's vapor pressure and affinity for soil organics, and is
described by its Henry's Law Constant (e.g., vapor pressure).
The methods commonly used in the field are standard EPA methods modified to shorten
analysis times and reduce lab generated wastes. Using existing data or knowledge of past
disposal activities, a list of several (between 5 and 15) target compounds is determined on a
site-specific basis. Method development techniques are used to optimize the analysis
method for these compounds. The result is faster analysis times and QA/QC customized to
meet specific site DQOs.
The EPA Superfund Branch has compiled a series of methods in the "Field Screening
Methods Catalog." This catalog is currently being updated with additional methods that
have been developed primarily under the Agency's Field Investigation Team (FIT)
Superfund contract. In the past five years, these methods have undergone method
validation, method quality control to identify appropriate end uses of the data and
performance testing under actual field conditions.
CASE STUDY- PORTSMOUTH GASEOUS DIFFUSION PLANT
Field-Scale Treatability Study Tor In Situ Soil Remediation
Site Historv/Background
The Portsmouth Gaseous Diffusion Plant is a U.S. Department of Energy (DOE)
production facility located near Piketon, Ohio. The area of interest for this study was the
X-231B Oil Biodegradation Unit located in the central portion of the plant site. The X-
23 IB Unit is approximately 1 acre in size and is covered by a "geomembrane" to reduce
volatilization and downward percolation of contaminants from rain water. It was
reportedly used from 1976 to 1983 for the treatment and disposal of waste oils ana
degreasing solvents.
-------
A previous site investigation was conducted by ORNL in December 1990. Three soil
borings were made in the northern portion of the X-231B Unit. Soil samples were
collected at multiple depths and analyzed for physical, chemical and bio/ogical properties.
Thirteen VOCs were identified in the soil samples collected, with TCE, TCA, 1,1-DCE and
Methylene Chloride being the most prevalent and at the highest concentration (e.g., several
hundred to several thousand ppb).
The purpose of subsequent sampling and analysis efforts was to obtain more complete
baseline data regarding the nature and extent of VOC contamination in the X-231B Unit.
Probe samples were taken from beneath the geomembrane and immediately analyzed onsite
in a mobile laboratory using headspace analytical protocols. Probe soil samples were
collected at three-foot intervals to a depth of twenty^four feet. Based on the initial January
results, "test cell" locations were selected for an in-situ soil mixing demonstration
performed in April/May. Mechanical mixing was combined with ambient air, heated air,
peroxide injection or grout to determine the most effective method for solvent reduction in
contaminated soils. Immediately following the various treatments in individual lest cells,
soil samples were re-collected from the exact same locations and depths to determine the
success of various treatment combinations. Soil samples were collected and analyzed in
real-time to limit many of the "sample degradation effects" associated with sample shipment
to and delayed analysis in an offsite laboratory.
Soil Sample Preparation
•The soil sample preparation for headspace analysis is summarized as follows:
• Approximately 20 grams of "wet" soil are immediately placed in a pre-weighed
40. ml VOA vial fitted with a septum cap following sample extrusion.
• The sample and vial are re-weighed in the mobile lab to the nearest 0.01 gram.
• The vial is placed in a 60° C water bath for 30 minutes.
• A subsample (usually 500|iL) of the headspace above the sample is
injected into die GC.
Gas Chromatography Analysis
Soil samples were analyzed by a GC equipped with an Electron Capture Detector and
Flame Ionization Detector (ECD/FID) in series. This configuration facilitated simultaneous
operation of both detectors. Each detector was calibrated over its entire linear range using
prepared standards containing each of the seven target compounds Using this arrangement,
the linear range of the analysis was gready increased. With a single injection, samples
could be quantitated from I ppb to approximately 500 ppm. This was necessary due to the
wide sample concentration ranges encountered and the need for rapid turnaround of sample
results.
In general, mobile lab results were consistently higher than "30-day" fixed lab results. The
positive bias was attributable to loss of VOCs during sample shipment and subsequent
sample handling at an offsite lab. The total price for the soil sampling and analysis portion
of the technology demonstration was approximately $90,000, or about $190 per sample for
sampling and analysis.
CONCLUSIONS
-------
Can acceptance be gained for the new hydraulic probe sampling capabilities and field
laboratory approach described? In many cases it already has, for example by EPA's
Superfund Site Assessment, Technical Assistance Team and Alternative Remedial Contract
Strategy Programs on hundreds of site inspections nationwide. As the probing equipment
and field laboratory methods become more frequently used by Federal, State and private
sector investigators, and data quality objectives are completely understood, widespread
acceptance will come. Rapid and cost-effective field sampling and analysis methods, in
conjunction with fixed laboratory confirmatory analyses and adequate quality control, will
undoubtedly result in growing support for this innovative and technically sound approach
to environmental site characterization work.
-------
National Symposium on Measuring and
interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 24
Application of Field VOC Data in Quantitative Risk Assessment at
CERCLA Sites
Ruth C. Kramel and Anthony Q. Armstrong
Risk Analysis Section, Health and Safety Research Division, Oak
Ridge National Library
January 12-14,1993
Las Vegas, Nevada
-------
Application of Field VOC Data in Quantitative Risk Assessment at CERCLA Sites
Ruth C. Kramel and Anthony Q. Armstrong
Risk Analysis. Section, Health and Safety Research Division
Oak Ridge National Laboratory
INTRODUCTION
Historically, field analytical data have been used primarily to guide on-going
sampling activities. Because of the uncertainty surrounding methods of field sample
collection and analysis (due both to a lack of experienced operators in the field, as well
as, to a lack of appropriate quality assurance/quality control (QA/QC) documentation),
the Environmental Protection Agency (EPA) has often considered these data unsuitable
for use in quantitative baseline risk assessments (BRA). However, recent advances in
technology allow on-site, real-time field analytical data, when accompanied by sufficient
and appropriate QA/QC measures, to be used in quantitative BRAs with more confidence
than in the past. Therefore, EPA (1990) has promulgated guidelines which permit this
use as long as proper procedures are followed. These procedures (including associated
documentation) should be established in the planning phase of any project to ensure that
data quality objectives (DQOs) are met for all data users involved in the project; thus
preventing costly and time-consuming additional sampling as the project proceeds. The
purpose of this paper is to describe appropriate procedures which must be followed if
field analytical data is to be used in quantitative BRAs which support risk management
decisions. Of necessity, discussion will be concerned with field analytical data in
general, in addition to some specific discussion relative to field VOC data.
Several benefits derive from using real-time field analytical data to characterize
VOC concentrations for risk assessment applications:
• Sample turnaround time is reduced, allowing real-time
decision-making in the field. Potential problems in either sampling
or analysis can be immediately identified, and the need for
corrective action can be assessed. For example, based on
preliminary data, decisions can be made regarding the need for
additional samples in order to provide further site information
and/or the suitability of sample collection and analytical methods
relative to providing information which is appropriate and
applicable to all aspects of the project.
-------
• Better site characterization can be achieved through the use of less
expensive field analyses, which allow the collection and analysis
of an increased number of samples, if these are necessary. Quite
often site conditions are not fully characterized, because fixed
laboratory costs preclude the analysis of an adequate number of
samples.
• Better site characterization can be achieved (especially for volatile
and semi-volatile organic compounds [VOCs, SVOCs]) by
immediate field analysis, which results in more accurate
measurements of contaminant concentrations than can be obtained
for those samples which must be packed, shipped and subsequently
handled by a laboratory.
• The potential for sample-to-sample contamination during shipping,
storage and subsequent laboratory handling is eliminated.
• More costly and time consuming CLP or other fixed laboratory
analyses are limited to a subset of confirmation samples, which
permits more cost-effective, time-effective completion of the
project for all concerned.
However, the extent to which these benefits are realized is directly related to the
extent to which the data collected and analyzed for a site support the needs of all data
users involved in the project. For this reason, a collaborative effort among all project
participants is necessary. This is especially true with respect to risk assessment due to
the nature of data usage in risk assessment and to the stringent data requirements
associated with that usage.
PROJECT SCOPING
Regardless of the nature of the project, the adequacy of the sampling and analysis
effort determines the subsequent quality of the BRA associated with that project;
therefore, it is imperative that the risk assessor be an active member of project planning
and that this involvement continue during the entire course of the project. The project
manager (PM) should facilitate the continuing, open and frequent communication among
all participants in a project which is so critical to success. The development of the BRA
within the framework of the project is an iterative process of action, feedback and
adjustment. -From the outset die PM, the risk assessor and other project participants
should work to identify a list of potential chemicals of concern (COCs) and to determine
data review procedures. A conceptual site model should be developed which depicts and
describes the current understanding of the extent of site VOC contamination, areas of
VOC contaminant source release to the environment (e.g., recreation areas; municipal
wells), applicable transport (e.g., migration in surface water/groundwater) and exposure
-------
pathways (e.g., dermal; incidental ingestion) and human and non-human populations
which are potentially at risk. Early identification of these sampling needs facilitates
decision-making relative to number, type and location of samples needed to quantitatively
assess exposure. All of these preliminary activities are predicated on a thorough review
of all existing data pertinent to the site.
Once these activities have been completedf the risk assessor will develop a final
data set summary for existing data which will provide the information used to design a
sampling plan which will best support BRA activities. This data summary should be
reviewed by the PM and other project participants, so that other data needs germane to
the project can be integrated and a preliminary sample-design based on ail project
requirements car, be formulated. The summary tables for existing, data should contain
(along with all associated assumptions, qualifications and limitations): 1) Site name and
existing sample locations (a site map with sample locations identified should be included);
2) Number of samples taken per medium (including sample depth); 3) Analyte-specific
sample quantitation limits; 4) Number of reported values which were above the sample
quantitation limits; 5) Measures of central tendency (95% upper confidence limit on the
arithmetic mean of the environmental concentration); 6) Treatment of qualified data; and
7) Ranges of reported concentrations.
QUALITY ASSURANCE/QUALITY CONTROL REQUIREMENTS
When trace-level VOC measurements are being performed using state-of-the-art
measurement technologies, a meaningful QA/QC program is a critical factor in the
process of data interpretation by the risk assessor. In the early project planning stages,
a determination of data quality objectives (DQOs) should be made by both management
and the technical staff. In support of the DQOs, performance criteria should be
established for specified sampling and analysis methods or techniques. Also, a review
of the qualifications of personnel proposed as operators of the instruments is warranted.
The application of appropriate QA/QC procedures is ultimately dependent on the
purposes for which the data will be used. For example, a sampling and analysis event
which is meant to determine compliance with an established regulatory standard or a
remediation limit requires a detailed QA/QC plan which will ensure data quality. This
QA/QC plan must be designed to produce data that can be used to verify whether an
enforceable violation has occurred (Brass and Kingsley 1989). On the other hand, a
qualitative determination may be sufficient to ascertain the absence or presence of a given
analyte after' a. soil treatment process; thus, reducing the need for exacting QA/QC
procedures and documentation. Without advance consideration of these factors, there is
a strong potential for the generation of useless data or for the inadvertent misuse of
results by the data user.
Specific documentation of data collection activities is requisite, so that the
resultant VOC data set can later be evaluated for completeness, comparability,
represetttativefl.'e&s, accuracy and precision. This documentation is also necessary so that
-------
the risk assessor will be aware of limitations associated with the data set. These
limitations must be delineated in the uncertainty section of the BRA so that accurate
interpretations of risk numbers can be made. The four major types of standard
documentation that are required are: 1) Sampling and Analysis Plan (SAP), including
a Quality Assurance Project Plan (QAPjP), 2) Standard Operating Procedures (SOPs),
3) field and analytical records and 4) chain-of-custody records. Once this last
preliminary planning activity has been completed, an optimally effective sampling plan
can be tailored to fit specific project needs.
PROJECT DESIGN
Based on the results of the scoping-phase (i.e., list of COCs, conceptual site
model, etc.), project participants then design a SAP to meet project specific DQO's and
QA/QC requirements. The choice of sampling and analysis methods to be employed
should be based on site-specific, project-specific data needs. It is desirable that data for
all COCs identified at the site be collected and analyzed by field methods. The
advantages garnered through the use of field data would be circumvented if samples must
still be sent to the laboratory for analysis of other chemicals. Additionally, it is
recommended that at least 10% of field analysis be confirmed by CLP or fixed laboratory
analysis (EPA 1990). This subset of confirmatory samples should be split in the field
and the field duplicate sent to an approved CLP or fixed-laboratory to be analyzed for
the same constituents as were analyzed for in the field. A comparison between duplicate
analysis results indicates whether or not COCs were appropriately identified by field
measurements and if measured concentrations are truly representative of site conditions.
The sampling strategies for a site must produce data which is appropriate for use
in a quantitative BRA; if the strategy is inappropriate, even the strictest QA/QC
procedures will not ensure the useability of sample results. From the risk assessment
perspective, the following are areas of major concern in sample design:
• Sample size based on the number of areas of concern and statistical
methods;
• Sampling locations based on the appropriate sampling strategy
(e.g., biased, random, systematic);
• Types of samples (e.g., point or composite);
• Temporal factors (e.g., potential reduction in VOC concentrations
over time);
Meteorological factors (e.g., VOCs should be measured during
both cool and warm seasons, as well as, during both dry and
humid conditions); and,
-------
• Existence of sufficient numbers of background samples so that
determination can be made regarding the validity of the assertion
that, there is no difference between site and background chemical
concentrations when measured at the desired level of confidence.
• Cost of sample collection and analysis.
The soil VOC sampling protocol most commonly used in BRAs is termed broad
spectrum analysis. This method provides a means for detection of an entire range of
compounds. There are a number of methods which may be used for broad spectrum
analysis; however, the method of analysis must be consistent with the stated objectives
of the study. Broad spectrum analysis is especially useful when the principal objective
is to determine which types of compounds are present (Hertz and Suffet 1990). VOC
results from broad spectrum analysis provide for rapid evaluation of exposure pathways
and selection of COCs by the risk assessor. With advances in analytical instrumentation
and techniques as well as the increased mobility of analytical equipment, broad spectrum
analysis can be used for determining the concentrations of individual VOCs in the field.
Subsequently, the potential risks from exposure to measured VOC concentrations may
be quantified to give a "snapshot" of risks at different locations at a given time.
SUMMARY
Time- and cost-effective field sampling and analysis methods implemented within
the framework of appropriate sample design and in conjunction with CLP or fixed
laboratory confirmatory analyses and adequate QC can produce data which is suitable for
use in quantitative BRAs. The advantages of this approach to environmental site
characterization are such that its implementation is likely to become the norm rather than
the exception. However, the current transition period between the fixed laboratory
paradigm and the field laboratory paradigm will not be an easy one unless the new
paradigm can be shown to produce consistently sound, useful data which can be validated
by accompanying QC documentation. Early and continuing communication among all
project participants is an indispensable component of the design process, which ensures
that the data needs of all participants are met in an efficient manner.
-------
REFERENCES
Brass, H.J. and B.A. Kingsley. 1986. Quality Assurance Programs in the Analysis of
Trace Organic Contaminants. In: Organic Carcinogens in Drinking Water.
(N.M. Ram, E.J. Calabrese and R.F. Christman, Eds.). New York: John Wiley
& Sons, Inc. pp. 173-195.
Environmental Protection Agency (EPA). 1990. Guidance for Data Useability in Risk
Assessment. Office of Emergency and Remedial Response. Washington, D.C.
EPA/540/G 90/008, Dir.: 9285.7-05.
Hertz, C.D. and I.H. Suffet. 1990. Research Methods for Determination of Volatile
Organic Compounds in Water. In: Significance and Treatment of Volatile-
Organic Compounds in Water Supplies. (N.M. Ram, R.F. Christman and K.P.
Cantor Eds.). Michigan: Lewis Publisher, Inc. pp. 39-56.
-------
Anthony Q. Armstrong, M.S. Mr. Armstrong received his M.S. in Microbiology from
the University of Georgia in 1989. While at the University of Georgia, he was a
research assistant at EPA's Environmental Research Laboratory in Athens. He was a
member of a team investigating the fate and transport of organic contaminants in soil and
groundwater at waste disposal sites. His research involved data collection and evaluation
and development, evaluation, and field testing of applicable bioremediation technologies.
In 1989, he obtained a position with the Risk Analysis Section, Health and Safety
Research Division, Oak Ridge National Laboratory. Since that time, Mr. Armstrong has
been involved with the development of sampling plans, data analysis, and risk
assessments for numerous hazardous waste sites at the United States Department of
Energy's Facilities throughout the United States. Currently, he is invotved with the
evaluation and implementation of remedial technologies for treatment of contaminated
soils'and groundwater at the Department of Energy's waste sites.
P O. Box 2008
105 MIT, MS 6492
Oak Ridge, TM 37831-6492
(615) 576-1555
Ruth C. Kramel has been with the Risk Analysis Section, Health and Safety Research
Division, Oak Ridge National Laboratory since September 1991. During this time she
has performed several human health risk assessments for Department of Energy (DOE)
facilities. Currently she is involved with methodology development for plant-wide
baseline ecological risk assessments (BERAs) at DOE facilities. In conjunction with this
work, she has been the lead author for ecological approach and strategy documents for
two DOE facilities and was task leader for development of a workplan for performance
of a plant-wide BERA at a DOE facility.
P.O. Box 2008
105 Mitchell Rd, MS 6492
Oak Ridge, TN 37831-6492
(615) 576-2732
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 25
Using Field Screening and Analytical Tools For Site Investigations
Fredrick W. Cornell
Environmental Liability Management, Inc., Princeton, New Jersey
January 12-14, 1993
Las Vegas, Nevada
-------
USING FIELD SCREENING AND ANALYTICAL TOOLS
FOR SITE INVESTIGATIONS
Frederick W. Cornell
Environmental Liability Management, Inc.
Princeton, New Jersey
SUMMARY
Proper management of a site investigation requires that the case team maintain a broad
perspective of the site characterization process during the planning and implementation
of each component of the field work. The case team must consider the purpose of the
investigation, the ultimate use of the data, and the relative uncertainties associated with
each component of the site characterization process, to ensure that all quality control
standards are appropriate for the investigation of interest. Prior to mobilization, a
thorough evaluation of the investigation process will ensure that the selected sampling
and analysis methods meet the investigation goals at the minimum cost possible.
Quality control criteria for the various components of a site investigation are generally
not developed by the case manager, instead they are developed by the individual(s)
responsible for, and most knowledgeable of, each component of the investigation. Thus,
the quality assurance and quality control (QA/QC) requirements for analytical data are
determined by laboratory groups and are based on the state-of-the-science for analytical
chemistry. The QA/QC requirements for sample collection and transport are generally
set up by the groups that perform sampling and are based on the state-of-the-science for
sampling technologies. And, the QA/QC requirements for health or ecological risk
evaluations are set up by risk assessors and are based on the state-of-the-science for
toxicology and exposure evaluation.
If the state-of-the-sciences for each component of the site characterization process were
similar, this method would be an efficient method of developing QA/QC goals for the
site characterization. However, the state-of-the-science of each component of the
investigation is dramatically different. For instance, when analytical chemists develop
QA/QC requirements for analysis methods, they expect uncertainties of about 5 % to 10%
and feel cumulative uncertainties greater than 100% are unacceptable. When
toxicologists develop a Reference Dose (RfD) for a chemical, they expect uncertainties
(or biases) of at least 10 times (1,000%) and feel cumulative uncertainties (or biases) of
1,000 times (100,000%) are acceptable. The reason is obvious, the state-of-the-science
for analytical chemistry allows for much greater accuracy and much better precision,
compared to toxicology. In fact, analytical chemistry is probably the most accurate
component of the site characterization process.
-------
Unlike the chemist or the toxicologist, the site investigation project/case team must focus
on the cumulative effect of analytical uncertainties, toxicological uncertainties and any
other uncertainties on the results of the site investigation. Thus, the case team must
recognize that a 100%, 200%, or maybe even 1,000% uncertainty in analytical data,
although unacceptable to an analytical chemist, may not only be of sufficient quality, but
may be of for greater quality than that which is necessary for a remedial investigation
where health-based cleanup objectives with uncertainties of 100,000% are to be used.
Consequently, the case team must recognize that the best analytical data is rarely, if ever
required. Instead data of sufficient quality for the job at hand are always required for
a site characterization.
Field analytical methods and instrumentation have become widely available in recent
years. Many of these field analytical methods/instruments were developed by making
laboratory equipment easy to use and easily transportable, i.e. field gas chromatography,
field x-ray fluorescence spectrometry, and field immuno-assay tests. When adapting the
laboratory instrumentation to field use, instrument performance and versatility were
diminished, resulting in poorer detection limits and accuracy. Additionally, due to the
size limitations, mass spectrometric detection systems are not routine on field gas
chromatographs, thus identification of unknowns is more difficult. Despite the
limitations of these methods, compared to their laboratory counterparts, field methods
are generally capable of providing data that is sufficient for use in most remedial
investigations.
This article discusses the uncertainties associated with sample collection, sample
transport, sample analysis (field and lab) and health-risk assessment, and applies the
concept of relative uncertainties to the selection of appropriate analytical methods for a
site investigation.
UNCERTAINTIES IN SITE CHARACTERIZATION
Sampling Design
The sampling plan is probably the most important component of the site characterization
process. A well designed sampling program will characterize the site without over
sampling the site. A poor sample design will introduce tremendous uncertainties into a
site evaluation and may be excessively costly, because if sample locations and/or depths
are not properly selected, contamination that is present at a site may not be detected
during the initial investigation phase.
-------
Sample placement must be based on: the source of the contamination, the characteristics
of the chemical and their effect on potential migration pathways, the physical setting of
the site and its effect on potential migration pathways, and the potential exposure points
for ecological or human receptors.
Two uncertainties are inherent in almost all sample designs:
¦ High Bias - because sampling is generally biased to areas of highest expected
contamination.
¦ Random Uncertainty - caused by the heterogeneity of sample matrices at a site.
Random uncertainty is most significant in soil samples where the adsorptive nature of the
soil material can significandy affect the amount of chemical present on that soil. The
rather large variances are observable in soil sample duplicates, where relative percent
differences of 100% are not unusual.
The high bias caused by sampling design, although frequently overlooked, is particularly
significant when site data are used in a health-risk evaluation (see Health Risk
Assessment section).
Sample Collection and Transport
The soil sample collection method can significantly affect the results of analysis. The
most accurate sampling procedures involve collection of a discrete, relatively undisturbed
soil sample, which can be prepared prior to analysis. Other methods include in situ
sampling methods, such as soil gas sampling and fiber optic sampling. Soil gas sampling
is a widely used screening tool that involves extracting air from the soil pore space and
analyzing the air sample. Fiber optic sampling involves-placing fiber optic probes into
the ground, and using spectrometry methods (which direct and collect light through the
optical fibers) to analyze the soil (or water) (Bernard and Walt, 1991; Lieberman et al.,
1991).
At sites where the soil is homogeneous and the contamination is evenly distributed in the
soil, soil gas sampling and fiber optic sampling can provide highly accurate results for
many analytes of concern. However, if highly heterogenous conditions exist, these
methods may produce erroneous results, which only loosely correspond to chemical
concentrations in the soil. Since field conditions at a site are generally somewhere in
between these two extremes, soil type and contaminant distribution must be carefully
evaluated.
-------
The effects of ex situ sampling methods, sample storage and preservation, and sample
transport were investigated by Siegrist and Jenssen (1990) and Maskarinec et al. (1990).
The results indicate that negative biases of up to 100% can be introduced if rigorous
procedures are not employed.
Samples Analysis Methods
The U.S. Environmental Protection Agency (EPA) has characterized analytical methods
according to accuracy, precision and QA/QC documentation (EPA, 1987a; EPA, 1987b;
EPA, 1990). Analytical methods, and associated data, were classified by the EPA as
detailed in Table I.
The EPA has based its classification on the ability of an analytical method to accurately
identify an analyte and quantify that analyte in a soil, water or air matrix. As discussed
earlier, field screening and analytical methods are less effective at identifying analytes
than their laboratory counterparts. However, if the site-specific chemicals of concern are
already known (i.e. lab data are available, or the source of the spill is known), field
analytical methods may be suitable for delineating the extent of contamination. To
facilitate selection of analytical methods the EPA published precision and accuracy data
for soil samples analyzed using level 1, 2, 3, and 4 methods (EPA, 1987b) (Table 2).
As can be seen, field analytical methods typically exhibit uncertainties of less than a
factor of 4, while laboratory methods exhibit uncertainties of less than 60%. Thus, in
situations were the analytes of concern have already been identified, field analytical
methods (level 1 and level 2 methods) can quantify chemicals of concern with
uncertainties typically less than 200% and almost always less than 360%.
Thus, by combining the uncertainties associated with sample collection, preservation,
transport, storage and analysis, the overall1 uncertainty of lab results is expected to be
approximately 220%. Similarly the overall uncertainty for field analysis methods can be
calculated; however, since field analytical methods are performed in the field, the
uncertainty associated with sample preservation, transport, and storage are eliminated.
Thus, the overall uncertainty in level I and level 2 results is approximately 360%.
Health-Risk Assessment
At most sites, cleanup goals are health-based and developed according to the risk
assessment protocol specified by the EPA (EPA, 1989; EPA, 1991a; EPA, 1991b).
These guidance documents detail the EPA's "policy" on how to deal with uncertainty in
a "scientific" manner. In the case of analytical uncertainty, the EPA has published a
document which describes the minimum useability requirements for analytical data in a
-------
baseline health risk assessment (EPA, 1990), approving the use of field screening (level
1 and level 2) methods, when, data quality objectives can be met and laboratory
confirmation samples axe collected.
The goal of a health risk assessment is to accurately estimate the risk of harm posed by
the chemical residuals present at a site. A health risk assessment generally includes four
components: data collection and evaluation, exposure assessment, toxicity assessment,
and risk characterization. Data collection will not be evaluated in this section, as it was
discussed in detail above.
The objectives of an exposure assessment are to identify potentially exposed individuals,
define the exposure routes, and estimate the magnitude, duration, and frequency of
exposure for each exposed individual. Rather than using site-specific values that are
appropriate for a site, the EPA has specified that default exposure values be used in
exposure assessments. These default values have been shown to over-estimate risk by
as much as three orders of magnitude or 100,000% (McKone and Bogen, 1991).
The objective of a toxicity assessment is to review available human and animal toxicity
data and determine an appropriate safe dose. The safe dose for non-carcinogens
(reference dose or RfD) defines a dose at which no significant adverse affects are
expected, and the safe dose for carcinogens (slope factor or SF) defines when the risk
of cancer is increased by a factor of one-in-one million. Generally the toxicity values
or safe doses (i.e. RfD and SF) are based on animal toxicity data; however, numerous
safety factors are used resulting in significantly under estimated RfDs and SFs (i.e. over-
estimates of risk posed by the chemical of concern).
For non-carcinogens safety factors of 100 to 1000 times (10,000% to 100,000%) are
typical, even though the use of multiplicative safety factors has been shown to produce
RfDs that obviously have no meaning. For instance, the EPA Integrated Risk
Information System (IRIS) database includes RfDs for zinc (5.0 x 10"2) and copper (5.0
x 10°) that represent the maximum amount of each chemical that a human may ingest
each day. "ITie RfD data were compared to the U.S. Food and Nutrition Board
Recommended Daily Allowances (RDAs) for zinc (12.5 mg/day); and the Safe and
Adequate Daily Intake for copper (1 to 3 mg/day) which were published in 1989. By
multiplying the average adult weight of 70 kg by the RfDs in mg/kg/day, the RfD was
converted to mg/day (Table 3).
Thus, if an individual eats the recommended daily allowance of zinc or copper, that
individual has already exceeded what the EPA has calculated to be the toxic dose of that
chemical (element). The cause of this obvious over-estimate of risk is the uncertainty
factors used for each of these chemicals, 1000 times (100,000%) and 500 times
-------
(50,000%) for copper and zinc, respectively. The over-estimates of risk caused by
multiplicative safety factors has been widely investigated (Brown, 1992; McKone and
Bogen, 1991; Burmaster and Lehr, 1991).
By combining the uncertainty associated with each component of a risk assessment, the
overall uncertainty in a health-based value can be as high a 100,000,000%, and the
uncertainty is almost always 100,000%. Furthermore, the uncertainty is designed to be
a conservative bias, which over-estimates risk.
Using Field Analysis Methods to Characterize a Site
A comparison of the overall uncertainty of the risk assessment derived value (at least
100,000%) to the overall uncertainty in the analytical data (200% to 500%) clearly shows
that the uncertainty associated with analytical data is almost always insignificant when
used in a health risk evaluation. Thus, field analysis methods should never be excluded
as an alternative based on accuracy or precision concerns. However, field analytical
methods do present some concerns.
As discussed, field analytical methods are not effective at identifying unknowns at a site.
Thus, in cases where the analytes of concern are unknowns, the worst case contamination
at a site should be screened with laboratory methods that provide high quality compound
identifications (i.e. GC/MS). Then, once the principle chemicals of concern have been
identified, the use of field analysis instruments may be appropriate for the remainder of
the site evaluation process.
In general the use of field analysis methods (FAM) is appropriate when: (1) a suitable
FAM is available, (2) the detection limit of the FAM is less than the concentration of
concern, (3) the uncertainty associated with that FAM accounts for less than 10% to 25%
of the overall multiplicative uncertainty of the site characterization process. When these
three criteria are met, FAMs should be calibrated to the site-specific chemicals of
concern, and the contamination at a site can be delineated using field analysis methods.
-------
TABLE 1
Analytical
Level
Analytical Method Type
Recommended Uses
(EPA, 1987a)
Level 1
Screening methods
Identifying sample locations
Level 2
Field adapted lab methods
Measuring presence or absence
Level 3
Lab methods with limited
QA/QC support
Conformational
Engineering
Level 4
Lab methods with full
QA/QC support
Confirmational
Toxicology
Level 5
Special services
Confirmational
TABLE 2
Analytical
Level
Analytical
Method Used
Range of Accuracy
Observed
Range of Precisions
Observed
Level 1
Field GC
-57% to +377%
Not reported
Level 2
Field GC
-99% to 354%*
Not reported
Level 3
HPLC
-18% to 0%
6% to 30%
Level 4
GC/MS
-59% to +14%
8% to 38%
* - one sample did exhibit a positive bias of 1,852 percent, but this sample was an
obvious outlier among the 23 samples and as such was neglected.
TABLE 3
Chemical
Compound
Uncertainty Factor in
the RfD
Reference Dose
(mg/day)
Recommended
Daily Allowance
(mg/day)
Copper
1000
0.35
1.0 to 3.0
Zinc
500
3.5
12.5
-------
REFERENCES
Barnard, Steven M.; Walt, David R.; "Fiber-Optic Organic Vapor Sensor", Environ. Sci.
Technol., Vol. 25, No. 7, 1991.
Brown, Stephen L.; "Harmonizing Chemical and Radiation Risk Management", Environ.
Sci. Technol., Vol. 26, No. 12, 1992.
Burmaster, David E.; Lehr, Jay H.; "It's Time to Make Risk Assessment a Science",
GMWR, Summer 1991.
EPA; "Data Quality Objectives for Remedial Response Activities Development Process",
EPA/540/G-87/003, March 1987a.
EPA; "Data Quality Objectives for Remedial Response Activities Example Scenario:
RI/FS Activities at a Site with Contaminated Soils and Ground Water", EPA/540/G-
87/004, March 1987b.
EPA; "Risk Assessment Guidance for Superfund Volume I Human Health Evaluation
Manual (Part A) Interim Final", EPA/540/1-89/002, December 1989.
EPA; "Guidance for Data Useability in Risk Assessment Interim Final", EPA/540/G-
90/008, October 1990.
EPA; "Risk Assessment Guidance for Superfund: Volume I - Human Health Evaluation
Manual (Part B, Development of Risk-based Preliminary Remediation Goals), Interim",-
Publication 9285.7-01B, December 1991a.
EPA; "Risk Assessment Guidance for Superfund: Volume I - Human Health Evaluation
Manual (Part C, Risk Evaluation of Remedial Alternatives), Interim", Publication
9285.7-01C, December 1991b.
Food and Nutrition Board; "Recommended Dietary Allowances", National Academy
Press, 1989.
Lieberman, S.H.; Theriault, G.A.; Cooper, S.S.; Malone, P.G.; Olsen, R.S.; Lurk,
P.W.; "Rapid, Subsurface, In Situ Field Screening of Petroleum Hydrocarbon
Contamination Using Laser Induced Fluorescence Over Optical Fibers", Symposium
Proceedings, Second International Symposium - Field Screening Methods for Hazardous
Wastes and Toxic Chemicals, 12-14 February 1991, Las Vegas, NV.
Maskarinec, Michael P.; Johnson, Lynne H., Holladay, Susan K.; Moody, Ronnie L.;
Bayne, Charles K., Jenkins, Roger A., "Stability of Volatile Organic Compounds in
Environmental Water Samples during Transport and Storage", Environ. Sci. Technol.,
Vol. 24, No. 11, 1990.
-------
McKone, Thomas E.; Bogen, Kenneth T., "Predicting The Uncertainties in Risk
Assessment", Environ. Sci. Technol., Vol. 25, No. 10, 1991.
Siegrist, Robert L.; Jenssen, Petter D., "Evaluation of Sampling Method Effects on
Volatile Organic Compound Measurements in Contaminated Soils", Environ. Sci.
Technol. Vol. 24, No. 9, 1990.
-------
BIOGRAPHY
Fred Cornell is a Project Manager with Environmental Liability Management (ELM) in
Princeton, New Jersey. He is currently a task group leader for an ASTM subcommittee
(Data Needs for Health Risk Assessment) and a sub-task group leader for the ASTM
committee developing field gas chromatography standards. Formerly Mr. Cornell was
a Technical Coordinator for the New Jersey Department of Environmental Protection and
Energy, where he drafted the New Jersey Field Analysis Guide. The Field Analysis
Guide details the appropriate uses of field analytical methods, and provides guidance for
method selection. Mr. Cornell has a M.S. in Analytical Chemistry from Penn State
University, and a B.S. in Chemistry from the State University of New York.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 26
Advances in on site and in situ VOC Measurement Techniques
Eric Koglin
USE PA Environmental Monitoring Systems Laboratory
Las Vegas, Nevada
January 12-14, 1993
Las Vegas, Nevada
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 27
Laboratory Analysis and Quality Assurance For Soil VOCs
R. J. Bentley, Michael J. Miille, Ph.D., Jerry L. Parr
Enseco
W. Sacramento, California
January 12-14, 1993
Las Vegas, Nevada
-------
LABORATORY ANALYSES AND QUALITY ASSURANCE FOR SOIL VOCs
R. J. Bentley, Michael J. Miille, Jerry L. Parr
Enseco
2544 Industrial Blvd., W. Sacramento, CA 95691
DEVELOPMENT OF VOLATILE ORGANIC METHODS
The June 7, 1976 court settlement involving the EPA and several
environmentally concerned plaintiffs brought the "EPA Consent Decree"
into being. A component of the Consent Decree was a list of "65
pollutants" which eventually formed the Toxic Pollutant List,.
Characterization sampling and analyses commenced shortly thereafter
prior to an in depth evaluation of minimum detection limits, sample
collection/preservation criteria, and available standard methods. Soon
afterwards, several additional organic pollutants were added to the
list, bringing the total to 129. This expanded list became known as
the Priority Pollutant List, and included compounds with a wide variety
of chemical properties, from inorganics to volatile and semi-volatile
organic pollutants.
The requirement for accurate screening of all volatile organic, compounds
in a single analysis prompted the use of GC/MS techniques combined with
the purge and trap method developed by Tom Bellar and Jim Lichtenberg at
USEPA Environmental Monitoring Systems Laboratory in Cincinnati, Ohio.
From these beginnings the various agencies developed similar but
different volatile organic methodologies incorporating independent
approaches for instrument calibration, linearity, quality check
standards, selection of surrogate standards, and so on. The Contract
Laboratory Program expanded the scope of analysis in the publication of
"Test Methods for Evaluating Solid Waste, Physical/Chemical Methods"5
where they applied the basic water and wastewater techniques to the
analysis of soils.
Over a decade later, the soil methods still have not been subjected to
in-depth method validation to develop method detection limits or
Practical Quantitation Limits (PQL's) for each compound. The method
simply lists the Estimated Quantitation Limits (EQL) that have been
extrapolated from the water and wastewater procedures.
SAMPLING TECHNIQUES
Although the analytical methods for volatiles in soil have not been
subjected to in-depth scientific scrutiny, the error imparted by the
method to the sample is probably far less than the volatiles lost in the
sampling procedures.
-------
The analytical results may give a fairly representative picture of the
volatiles remaining in the sample at the time of analysis, but how
representative are the results of the location it represents? The
studies conducted by Oak Ridge National Laboratory and presented at the
EPA "Symposiums on Solid Waste Testing and Quality Assurance" in both
1987 and 1988 indicated very short holding times for volatiles in soil.
These studies indicate, "volatile organics in soil are much less stable,
with primary losses occurring through volatilization and microbial
degradation. For these samples, rapid sample preparation appears to be
the only means of long term storage3". The Oak Ridge Study did indicate
an improved recovery for samples stored at -20°C compared with those
stored at 4°C.
This would seem to indicate one means to improve data from soil
volatiles would be a further integration of sampling techniques into the
methods and laboratory responsibilities. Flexibility should be
incorporated into the methods that could allow for different containers,
freezing samples upon collection, the spiking of surrogate compounds in
the field, rapid purging into a sorbent trap, or field fixation of the
volatile components in methanol as examples. Sample introduction
techniques other than purge and trap {i.e., vacuum distillation) should
also be allowed to accommodate new approaches4.
QUALITY ASSURANCE
It would be inappropriate not to include field sampling in a discussion
of quality assurance. Currently the reporting of volatile compounds for
Method 8240 includes the evaluation of the recoveries of surrogate
compounds. Recoveries falling within prescribed ranges by compound may
provide a false sense of security for the data user. The injection of
both internal standards and the surrogates into the purge and trap
vessel just prior to analysis assures reasonable recoveries and is an
indication that the system is .operating properly. It is important to
state that this may be far removed from the relationship between the
volatiles contained in the sample at both the time of collection and the
time of analysis and the distribution of volatiles on site. Current
thinking continues to push for method rigidity as a means to assure
quality and sampling comparability. This focus seems inappropriate when
difficulties with sample preservation and subsampling are primarily
responsible for the majority of errors in the analysis.
SUMMARY (CONCLUSION)
Methods for the analysis of volatile organics have evolved along
different paths in the past 15 years. Methodologies currently used for
waters, wastewater and soils contain method elements that are arbitrary
and cumbersome from a laboratories perspective. The soil methods in
particular focus on the method elements least likely to develop error,
and only briefly address the greatest sources of error, sampling and
delivery of a sample to the laboratory.
-------
The 1990's will see a shift away from the survey methods required for
many remedial investigations, toward corrective actions where method
flexibility will provide higher quality and more inexpensive approaches
to the accurate analysis of a smaller list of compounds. The work
currently being done by the EPA's Environmental Methods Management
Council (EMMC) is a vital first step in methods consolidation. It is
necessary to develop performance driven laboratory methods that will
allow the flexibility for improved sampling/subsampling and sample,
introduction techniques. These methods must then be validated
establishing detection limits, working range,, precision bias,
ruggedness, matrix and comparability. This should mitigate the current
bias toward a high number of false positives. Laboratories need to be a
partner in the Data Quality Objectives (DQO) process and help employ
techniques for improved data quality.
REFERENCES
.1. Keith, H., Telliard, W.A. "Priority Pollutants I-a Perspective
View", Environmental Science and Technology, Vol. 13, Number 4 (1979)
2. Wagner, R. E. Guide to Environmental Analytical Methods, Genium
Publishing Corporation, Schenectedy, N.Y. 1992
3. Maskartnec, M. P., Johnson, L. H., Holladay, S. K., Oak Ridge
National Laboratory. USEPA Symposium on Waste Testing and Quality
Assurance, Vol. II, July 11-15, 1988
4. Siegrist, R. L., Jensen, P. L. "Evaluation .of Sampling Method
Effects on Volatile Organic Compound Measurements in Contaminated
Soils", Environmental Science and Technology, Vol. 24, Number 9
(1990)
5. EPA, "Test Methods for Evaluating Solid Waste", SW-846, 3rd Edition,
September 1986.
BIOGRAPHY
Mr. Bentley is the Vice President and General Manager for the Western
Region of Enseco. Prior to 1986 he was the Vice President and co-founder
of Analytical Technologies, Inc. He has over twenty years of operating and
management experience since acquiring a degree in chemistry at Arizona
State University. He has presented papers on Chromatography, analyses of
incineration by-products by GC/MS techniques, and quality programs for
environmental laboratories.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 28
Interlaboratory Study of Analytical Methods For Petroleum
Hydrocarbons
Jerry L. Parr, Enseco-Rocky Mountain Analytical Laboratory, Denver,
Colorado; Roger Claff, American Petroleum Institute, Washington,
DC; Dianna Kocurek, Tischler/Kocurek, Round Rock, Texas, Jeff
Lowry, Environmental Resource Associates, Denver Colorado
January 12-14, 1993
Las Vegas, Nevada
-------
INTERLABORATORY STUDY OF ANALYTICAL METHODS
FOR PETROLEUM HYDROCARBONS
Jerry L. Parr
Enseco-Rocky Mountain Analytical Laboratory, Denver, Colorado
Roger Claff
American Petroleum Institute, Washington, DC
Dianna Kocurek
Tischler/Kocurek, Round Rock, Texas
Jeff Lowry
Environmental Resource Associates, Denver, Colorado
The American Petroleum Institute (API) has funded efforts to establish reliable, methods
for the measurement of a wide range of petroleum hydrocarbons in soil. The efforts
involved extensive single laboratory validation studies that incorporated the sampling and
sample handling processes along with the laboratory efforts, as well as interlaboratory
studies. As a result, the data that were obtained realistically assess the validity of the
method, encompassing both sampling and analytical variability. While API's work has
focused on a wide range of petroleum hydrocarbons, this presentation will focus only on
volatile petroleum hydrocarbons, e.g., gasoline range organics.
At the start of our work, we recognized that while the technology to measure gasoline
range petroleum hydrocarbons (e.g., GC/FID) was well known, there was virtually no
information relative to the reliability of this technology to measure environmental levels
of the analytes in soil samples and, furthermore, that few if any studies had addressed
the sampling and sample handling aspects. Accordingly, all of the validation work was
performed on soil matrices, and additional studies were conducted to evaluate sampling
techniques. These efforts, which have been presented elsewhere, demonstrated that the
analytical method was reliable for measuring gasoline range organics in soil and that
significant losses of volatile components can occur with conventional soil sampling and
sample handling practices. The method established by API recommends that soil samples
be preserved in methanol in the field to reduce these losses.
Following this single laboratory validation study, an interlaboratory study was performed.
The study was conducted in accordance with ASTM D2777-86 and involved Youden
pairs at three concentrations. Stable test materials were prepared as spiked soil samples.
The spike concentrations were verified. Following removal of outlying results, the
precision and bias were calculated using the techniques in ASTM D2777 and Youden's
-------
technique of calculating precision. The results from this study have been used to
document the performance of the method.
In summary, the key aspects of our work, which are lacking in many validation studies,
were:
o Validation in the matrix of concern, not extrapolation of reagent water data,
o Incorporation of sampling and sample handling techniques in the validation
study, and
o Interlaboratory studies that involve the analyses of authentic samples.
-------
REFERENCES
1. Sampling and Analysis of Gasoline Range Organics in Soil. Enseco-Rocky Mountain
Analytical Laboratory for the American Petroleum Institute, API Publication Number
4516, October 1991.
2. Jerry L. Parr et al., "Sampling and Analysis of Soils for Gasoline Range Organics,
Hydrocarbon Contaminated Soils and Groundwater. Lewis Publishers, Michigan,
1991, pp. 105-132.
-------
Jerry L. Parr
Enseco-Rocky Mountain Analytical Laboratory
Jerry L. Parr is the Director, Quality Assurance and Technology, with Enseco-Rocky
Mountain Analytical Laboratory, one of the largest environmental analytical laboratories
in the U.S. He is an organic chemist with 20 years of experience in the environmental
field. Mr. Parr has conducted numerous research projects relative to groundwater
contamination; was the primary author of an EPA report to the U.S. Congress on the
adequacy of EPA's methods; and was the principal investigator on numerous API
research projects including development of methods to measure petroleum hydrocarbons
in soil, use of the TCLP for RCRA regulations, and treatment technologies for petroleum
refinery wastes. He is currently working on efforts within EPA to develop performance-
driven consensus, methods and is a member of the LAETL Regulatory .Committee. Mr.
Parr received a B.S. degree in Chemistry from University of Texas at Austin.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 29
New SW-846 Methods For the Analysis of Conventional and Non-
conventional Volatile Organics in Solid Matrices
Barry Lesnik
Office of Solid Waste, Methods Section
U.S. EPA, Washington
January 12-14, 1993
Las Vegas, Nevada
-------
NEW SW-84 6 METHODS FOR THE ANALYSIS OF CONVENTIONAL AND
NON-CONVENTIONAL VOLATILE ORGANICS IN SOLID MATRICES
by Barry Lesnik, USEPA, Office of Solid Waste, Methods
Section (OS-331), 401 M St., SW, Washington, DC 20460
Background
Historically, the analysis of volatile organic compounds
(VOCs) in environmental samples has focused primarily on the
conventional analytes which are amenable to the purge-and-trap
technique in aqueous matrices. Attempts have been made, with
limited success, to adapt this technique for the analysis of these
analytes in soils and other solid matrices. Very, little attention
has been paid to other volatile analytes, the so-called non-
conventional analytes, which appear on various regulatory lists,
which for various reasons, e.g. water solubility, are not amenable
to purge-and-trap analysis. In this paper, I will focus on RCRA's
methods development program for these two areas.
Improved Purae-and-rTrap Methods for Conventional Analytes
For the past 15 to 20 years, the method of choice for sample
introduction for gas .chromatographic (GC) analysis of conventional
volatiles in aqueous matrices has been the purge-and-trap technique
(Method 503 0) at ambient temperature, developed at the
Environmental Monitoring Systems Laboratory in Cincinnati (EMSL-Ci)
during the 1970s. Determinations have been performed using a
variety of detectors, e.g. electroconductivity (E1CD) in Methods
8010 and 8021, photoionization (PID) in Methods 8020 and 8021, and
mass spectrometry (MS) in Methods 8240 and 8260. Acrolein and
acrylonitrile are analyzed by heated purge-and-trap (Method 503 0)
with either flame ionization (FID) (Method 8030) or MS (Method
8240/8260) detection.
Method 5030 has been adapted to address soil, sediment and
other solid samples using either water alone or a methanol
extraction followed by dilution in water as purging media. This
approach has historically resulted in low recoveries and erratic
precision, caused by several factors. The major causes of these
analytical problems with solid samples were loss of volatiles to
headspace and slow or incomplete release of volatiles from the
interstices of the solid particles.
Several approaches to minimize these losses and to increase
recoveries of volatile analytes from solid matrices have been used
by different groups, e.g. Oak Ridge National Laboratory, EPA Region
IV, Dynatech Precision, and EMSL-Ci, with varying degrees of
success. The most successful of these approaches is expected to be
included in the Third Update of the Third Edition of SW-84 6 as
-------
Method 5035. The method was developed by Dynatech Precision and
further refined by EMSL-Ci.
Method 5035 employs an automated purge-and-trap system which
agitates the sample during the purge step. The purge temperature
is 40°C, and the total sample introduction is performed in the same
pre-weighed vessel in which the sample was originally collected.
Recoveries of target analytes improved dramatically over those
obtained using Method 5030, with excellent precision. This
dramatic improvement in method performance was the reason for
writing up Method 503 5 as a separate method for solids. The Third
Update version of the original SW-846 purge-and-trap method (Method
5030B) will be written specifically for aqueous matrices.
Alternative Techniques for Conventional Analvtes
Purge-and-trap sample introduction may not always be the
optimum technique for dealing with some solid matrices and is
definitely inappropriate for dealing with oil matrices. Three
methods dealing with these "problem" matrices have been developed
and are expected to be included in the Third Update. These methods
are Method 5032-Volatile Organic Compounds by Closed System Vacuum
Distillation with Cryogenic Condensation, Method 5022-Volatile
Organic Compounds in Oil Matrices Using Heated Automated Headspace
Apparatus, and Method 3585-Volatile Organics in Oily Matrices by
Solvent Dilution/Direct Inject.
Vacuum distillation, Method 503 2 is a general volatile
organics method, developed by EMSL-LV, which has a potential
utility beyond the scope of the purge-and-trap methods, in that it
is also applicable to the alcohols, ketones and other non-purgeable
analytes. The method utilizes a vacuum manifold to perform a trap
to trap distillation, with the distillate collected cryogenically.
Method 5032 has shown applicability to a variety of matrices
including water, solids, and oils, and is currently being prepared
for evaluation in a multi-laboratory study.
Methods 5022 and 3585 were developed by OSW in support of the
Used Oil Listing effort, and are specific for volatiles in oil
matrices. Method 5022 is a heated automated static headspace
method for the preparation of volatiles in oils for analysis by
isotope dilution GC/MS. Since the target analytes are very soluble
in oils, the heated headspace conditions are severe with relatively
low actual recoveries, hence the need for an isotope dilution
determinative method. Method 3585 is a simple "dilute and shoot"
method utilizing hexadecane as the dilution solvent. The diluted
oil is then injected directly into a GC or GC/MS for analysis, with
the target analytes preceding the solvent front.
In addition to the ongoing vacuum distillation projects, work
-------
is progressing on a general automated static headspace method,
which should be completed in time for inclusion in the Third
Update. EMSL-Ci has completed its work on solids. OSW will
complete the project for aqueous matrices.
Methods for Non-Conventional Analvtes
Another group of volatile analytes which appear on many
regulatory lists, e.g. Appendix VIII, TC, etc. are the so-called
non-conventional analytes. These include many water soluble
compounds including alcohols, ketones, ethers, esters and the like.
This water solubility causes very low purging efficiencies which
causes these compounds to perform poorly compared to conventional
analytes in purge-and-trap methods. Third Update sample
preparation methods for these analytes are Method 5031-Non-
Purgeable Volatile Organic Compounds by Azeotropic Distillation and
Method 503 2-Volatile Organic Compounds by Closed System Vacuum
Distillation with Cryogenic Condensation. Method 5032 was
discussed in the preceding section.
Method 5031 was originally developed by EMSL-Ci. The
microdistillation technique that was added was submitted, by
Wadsworth-Alert Laboratories. The method is basically a
concentration technique for water-soluble, non-purgeable volatiles,
e.g. alcohols, ketones, esters, etc., in aqueous matrices using
azeotropic distillation to increase the sensitivity of the direct-
injection aqueous determinative methods (Methods 8015 and 8260).
A concentration factor of 50 to 70 can be readily achieved,
lowering detection limits to the low ppb range. Additional work is
continuing on sample preparation methods for these analytes in
solid matrices, and for the use of gas chromatography/Fourier
Transform infrared spectroscopy (GC/FT-IR) as a determinative
technique.
Another non-conventional volatile analyte for which we have
developed a method is formaldehyde. Method 8315, which will be
proposed in the Second Update, is a reverse phase high performance
liquid chromatography (HPLC) method utilizing pre-column
derivatization of formaldehyde to its 2,4-dinitrophenylhydrazone
with ultraviolet/visible (UV/Vis) detection. Formaldehyde can be
recovered from solid matrices by aqueous extraction, after which
the method for aqueous matrices can be followed.
Summary
We have recognized some of the analytical shortcomings of the
existing methods for volatiles in solid matrices. RCRA has spent
a great deal of time and effort in improving these existing
methods, as well as developing alternative analytical techniques
for both conventional and non-conventional analytes. These efforts
-------
to improve analytical methods will continue over the foreseeable
future, and along with ongoing concerted efforts to improve
sampling techniques, should result in dramatic improvements in the
quality and reliability of the analytical data for volatiles in
solid matrices.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 30
Purge-and-Trap GC/MS Method Modifications
S. Ward
Harry Reid Center for Environmental Studies
University of Nevada, Las Vegas
January 12-14, 1993
Las Vegas, Nevada
-------
BIOGRAPHICAL SKETCH
JACK C. DEMIRGIAN, Analytical Chemist, Analytical Chemistry Laboratory
Education: 1968 - B.A. (Chemistry)
Hunter College, Bronx, NY
(now Herbert H. Lehman College)
1976 - Ph.D. (Physical Analytical Chemistry)
State University of New York at Buffalo, Buffalo, NY
Dr. Demirgian was employed as an Associate Professor of analytical chemistry at Christopher
Newport College, formally part of the College of William and Mary, from 1976 until he came
to the EES Division of Argo'nne National Laboratory in'June 1981. He joined CMT in August
1982 as part of the ACL/OAG group. He was promoted to Chemist in 1986.
He is responsible for project development and management in environmental, engineering, and
special application areas. This work involves using FTIR, GC, GC/MS, and SFE to analyze
complex organic mixtures. New analytical methods are developed for specialized applications.
FTIR technology is being developed for the characterization of hazardous waste sites, incinerator
monitoring, and radioactive liquid flow monitoring. He is also involved in the development of
remote detection methods to monitor the environment. Extraction methods are being developed
for complex environmental samples. All analytical methods being developed are being automated
for computer-assisted rapid analysis.
He has approximately fourteen publications. He is a member of the American Chemical Society
&nd Technical Association of the Pulp and Paper Institute (TAPPI).
JCD/vts
3/31/90
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 31
Desorption Hysteresis in Five Ion Exchanged Montmorillonites
Jerry P. Fairley, Jr., Department of Geoscience, University of Nevada,
Las Vegas; and Spencer M. Steinberg, Department of Chemistry,
University of Nevada, Las Vegas
January 12-14, 1993
Las Vegas, Nevada
-------
Desorption Hysteresis in Five Ion
Exchanged Montmorillonites
Jerrv P. Fairlev. Jr.. Department of Geoscience, University of Nevada, Las Vegas
Spencer M. Steinberg, Department of Chemistry, University of Nevada, Las Vegas
When soils are exposed to volatile organic compounds (VOCs), whether accidentally,
as in leaking fuel tanks, or purposely, as in soil fumigation, a residual remains in the
soil long after the majority of contaminant has volatilized or decomposed. The
presence of these compounds may go undetected by standard methods of estimating
organic compounds in soils. Sawhney, et al. (1988) found that standard techniques
for estimating VOCs in soils (EPA Method 8240) underestimated quantities of 1,2
dibromoethane (EDB) in fumigated soils by up to 89%. This "persistent fraction"
may continue to desorb over a period of years, possibly providing a long term health
hazard. Clearly, knowledge of persistent contamination, is necescary to understand the
true impact of .our actions on the environment. This study undertakes to determine
where the persistent fraction resides within one component of soil (clay), and what
factors may effect formation of a fraction resistant to desorption.
Several studies support the belief that sorbate may be stored in physically inaccessible
sites within soil structure. Steinberg, et al. (1987) studied soils from Connecticut
tobacco fields which had been treated with EDB. Soils last fumigated with EDB up
to 19 years previously were found to retain concentrations of EDB at levels up to 200
ng/g. The presence of a recalcitrant fraction in these soils is particularly surprising
given the physical properties of EDB, which is highly volatile and has a low affinity
for soil. The investigators found that soils fumigated with EDB retained a fraction of
compound highly resistant to further desorption by volatilization (dry N2 gas passed
through soil samples at a rate of 30 volumes of gas per volume of soil per minute for
3.5 days only removed 8% of the recalcitrant fraction of one soil) and unavailable for
microbial degradation.
Another interesting aspect of the Steinberg, et al. (1987) experiments was the
response of the recalcitrant fraction to mechanical shock. When a ball mill was used
to physically break apart soil particles the percentage of recalcitrant EDB released to
a dry stream of N2 gas increased from 8% to 40%. This led the investigators to
postulate entrapment of EDB in soil "micropores" inaccessible to bulk fluids.
In clay minerals, researchers favor two probable areas for the storage of persistent
contamination: sites between the clay lamella, or sites within clay particle aggregates.
Pignatello (1989) suggests that soil particles can aggregate to form larger,
microaggregate particles with considerable porosity. These pores, with effective
diameters of 100 to 1000 nm; exclude advection of bulk fluids, limiting sorbate
transport within the pores to diffusion. This model of micropores agrees well with
-------
the release of EDB upon mechanical grinding of soils found by Steinberg, et al.
(1987), who concluded that "EDB in inaccessible regions becomes available as these
region are exposed by pulverization"
Another possible reservoir for a persistent fraction is the interlamellar space of clay
minerals. Clays are minerals composed of continuous two-dimensional tetrahedral
sheets based on the composition T205 (T=Si, Al, Fe3+,...). One or two tetrahedral
sheets are associated with one octahedral sheet, which co-ordinate to form a layer.
These layers can be electrically neutral, as is the case with pyrophyllite; in other
cases the layers carry charges due to substitutions in octahedral and/or tetrahedral
layers. The result is that interlayer cations are absorbed into the interlamellar region
to maintain neutrality (Brown, 1984). These interlayer cations can effect VOCs in a
variety of ways. Isaacson and Sawhney (1983) showed that the presence of transition
metal cations catalyzed the production of transformation products of parent phenols.
Soma, et al. (1984, 1985) found evidence for the polymerization of benzenes and
monosubstituted benzenes in-transition metal ion exchanged montmorillonites. In
addition, Aochi, et al. (1992) found evidence that, of the two EDB isomers (anti and
gauche) the anti conformer is more labile, resulting in an enrichment of the gauche
conformer in persistently sorbed compound. This behavior hints at the possibility of
dipole moment interactions with interlayer material or dielectric effects arising from
the clay structure itself.
To test the hypothesis that recalcitrant fractions( are formed in interlamellar spaces, a
Na/Ca Smectite (Montmorillonite No. 26, Clay Spur, Wyoming) was obtained from
Ward's Natural Science Establishment, Inc. and ion exchanged with five different ions
(K+, Na+, Ca2+, Mg2+, Fe3+) to form mineralogically similar clays with varying
interlamellar environments. These soils were inoculated with an organic compound
(toluene), incubated for 24 hours, and the recalcitrant fraction quantified for varying
desorption times. If the recalcitrant fraction resided in the interlayer region, one
would expect a strong correlation between the interlayer environment and the amount
of toluene sorbed persistently.
Two likely mechanisms for sorption of toluene to the interlayer region are ion induced
dipole interactions and dielectric field effects (see above). If the recalcitrant fraction
was related to ion induced dipole interaction in the interlamellar space, a strong
correlation between persistent toluene and ionic charge/radius2 (z/r2) would be
expected. In reality, correlations with z/r2 were indifferent to poor. Correlations
between toluene and interlamellar spacing which would be expected if dielectric field
effects held toluene in a persistently sorbed state were similarly lacking. The
apparent lack of relationship to interlayer environment argues for the existence of the
recalcitrant fraction in areas other than the interlamellar space.
Experiments were also performed by inundating toluene inoculated montmorillonites
-------
with ion exchange solutions in an effort to cause shock desorption of recalcitrant
toluene. Persistently sorbed toluene was released to solution; however, the amount of
toluene released was independent of the ion exchange solution used. This strengthens
the argument that toluene is not held in the interlamellar space by ion induced dipole
interactions: if toluene was being held by ion interactions, a variation in the amount
of toluene released would be expected based on the change in z/r2 ratio.
Furthermore, the amount of toluene released was directly proportionate to the amount
of agitation received by the sample. These results are consistent with the
microaggregate model of persistent fraction formation. In this model, residual toluene
is held within water stable and non-water stable microaggregates. The addition of
solution (ionic or non-ionic) causes decomposition of non-water stable microstructure
and subsequent release of toluene. Agitation results in further toluene desorption as
marginally water stable structures also collapse. These findings are remeniscent of
the Steinberg, et al. (1987) experiments, where the addition of mechanical energy to
the sample grinding resulted in increased EDB desorption. In both cases, the result
was a large increase'in the amount of residual compound released, apparently through
a loss of microstructure.
References:
Aochi Y.O. and Farmer W.J. 1992. In-situ Investigation of EDB sorption/desorption
processes on clay mineral surfaces by diffuse reflectance infrared spectroscopy. To be
published in Environ. Sci. Technol.
Brown G. 1984. Crystal structures of clay minerals and related phyllosilicates. Phil.
Trans. R. Soc. London A. 311:221-240
Isaacson P.J. and Sawhney B.L. 1983. Sorption and transformation of phenols on clay
surfaces: effect of exchangeable cations. Clay Miner. 18:253-265
Pignatello, J.J. 1989. Sorption dynamics of organic compounds in soils and
sediments, pp 45-80. In Reactions and Movement of Organic Chemicals in Soils, Soil
Science Society of America and American Society of Agronomy special publication
no. 22.
Sawhney, B.L., Pignatello, J.J. and Steinberg, S.M. 1988.Determination of 1,2-
Dibromoethane (EDB) in field soils: implications for volatile organic compounds. J.
Environ. Qual., 17(1): 149-152.
Soma Y. Soma M. and Harada I. 1984. The reaction of aromatic molecules in the
interlayer of transition-metal ion exchanged montmorillonite studied by resonance
raman spectroscopy. 1. benzene and prphenylenes. J. Phys. Chem. 88:3034-3038
-------
Soma Y. Soma M. and Harada I. 1985. Reactions of aromatic molecules in the
interlayer of transition-metal ion-exchanged montmorillonite studied by resonance
raman spectroscopy. 2. monosubstituted benzenes and 4,4'-disubstituted biphenyls. J.
Phys. Chem. 89:738-742
Steinberg, S.M., Pignatello, J.J. and Sawhney, B.L. 1987. Persistence of 1,2-
dibromoethane in soils: entrapment on intraparticle micropores. Environ. Sci.
Technol. 21:1201-1208.
-------
Key References:
Aochi Y.O. and Farmer W.J. 1992. In-situ Investigation of EDB sorption/desorption
processes on clay mineral surfaces by diffuse reflectance infrared spectroscopy. To be
published in Environ. Sci. Technol.
Brown G. 1984. Crystal structures of clay minerals and related phyllosilicates. Phil.
Trans. R. Soc. London A. 311:221-240
Isaacson P.J. and Sawhney B.L. 1983. Sorption and transformation of phenols on clay
surfaces: effect of exchangeable cations. Clay Miner. 18:253-265
Pignatello, J.J. 1989. Sorption dynamics of organic compounds in soils and
sediments, pp 45-80. In Reactions and Movement of Organic Chemicals in Soils, Soil
Science Society of America and American Society of Agronomy special publication
no. 22.
Sawhney, B.L., Pignatello, J.J. and Steinberg, S.M. 1988. Determination of 1,2-
Dibromoethane (EDB) in field soils: implications for volatile organic compounds. J.
Environ. Qual., 17(1): 149-152.
Soma Y. Soma M. and Harada I. 1984. The reaction of aromatic molecules in the
interlayer of transition-metal ion exchanged montmorillonite studied by resonance
raman spectroscopy. 1. benzene and p-phenylenes. J. Phys. Chem. 88:3034-3038
Soma Y. Soma M. and Harada I. 1985. Reactions of aromatic molecules in the
interlayer of transition-metal ion-exchanged montmorillonite studied by resonance
raman spectroscopy. 2. monosubstituted benzenes and 4,4'-disubstituted biphenyls. J.
Phys. Chem. 89:738-742
Steinberg, S.M., Pignatello, J.J. and Sawhney, B.L. 1987. Persistence of 1,2-
dibromoethane in soils: entrapment on intraparticle micropores. Environ. Sci.
Technol. 21:1201-1208
-------
Jerry P. Fairley, Jr.
Jerry P. Fairley, Jr., recently received his Master of Science degree in
Geology from the University of Nevada, Las Vegas. Mr. Fairley has worked for
consulting companies and State agencies as a geologist and hydrogeologist in New
York, Colorado, and Nevada since 1984, when he received his Bachelor of Science
degree in Geology from the State University of New York, College at Cortland.
Although he has a strong interest in working internationally, he is currently seeking
admission to a recognized Doctoral program.
Spencer M. Steinberg
Spencer M. Steinberg is presently an Assistant Professor of Chemistry at the
University of Nevada, Las Vegas (UNLV). He received a Ph.D. in Marine Chemistry
from the Scripps Institution of Oceanography and a B.S. in Chemistry form the
University of California, San Diego. He currently serves as the Chemistry
Department's Graduate Coordinator and is advising graduate students in UNLVs
Environmental Analytical Chemistry and Water Resources Management programs.
Dr. Steinberg's research interests include analytical methods development, and the
fate, transport and reactions of organic ar.d inorganic substances in water, soil and
sediment.
-------
Cbemospliere, Vol. 24, No-9, pp 1301 "131 5, 19^2
Printed in Great Britain
00£i5-6535/92 >3.00 + 0.00
Pprgaraon Press Ltd
PERSISTENCE OF SEVERAL VOLATILE AROKATIC AND EALOGEK'ATED
HYDROCARBONS IN A LOU ORGANIC CARBON CALCAREOUS SOIL
Spencer M. Steinberg
Department of Chemistry, University of Nevada Las Vegas
Las Vegas, Nevada B9154
ABSTRACT
Incubation of 'soil samples with volatile organic solvents at
part-per-thousand (PPT) concentrations leads to the formation of
a residual firmly bound fraction that resists evaporation. The
concentration of this firmly bound fraction increases with
temperature an
-------
fumigation treatments occurred more than twenty years earlier
(Steinberg et al., 1987; Pignatello, 1989). Furthermore, it was
discovered that the standard laboratory methods failed to extract
this recalcitrant fraction, and that the binding of the EDB by
soil prevented microbial decomposition. Recently, Pignatello and
Huang (1991) demonstrated that metolachlor and atrazine could
become firmly associated with soil and exhibit slow desorption
kinetics. Similarly Clay et al., (1988) noted considerable
difference between sorption and desorption isotherms for atrazine
and cyanazine from soils. Studies also indicate that the
association of soil chemicals with the intraparticle micropore
structure will prevent biodegradation as well as transport
(Rijnaarts,. et al., 1990). Thus, in some contaminated soils the
firmly bound fraction may become the most concentrated fraction
after leaching, volatilization and biodegradation have removed
the labile component.
In the vicinity of leaking fuel tanks or near accidental
spills, the initial hydrocarbon concentrations would be
comparable to the conditions experienced by soil during
.fumigation and should lead to firmly bound, hydrocarbon fuels and
industrial solvents in the contaminated soil. This hypothesis is
tested herein by incubating soils with various important organic
solvents at relatively high concentrations, and then quantifying
the fraction of organic compound that failed to volatilize after
24 hours at 32°C. The dynamic head space concentration method
has also been compared to a solvent extraction method (Sahvney et
al., 1987) and a water extraction method which was previously
used for field' screening of contaminated soils (Lufkin et. al.,
1991). Because, the firmly bound fraction may be the only
fraction of the volatile material from a spill that remains in
many contaminated sites after volatilization, leaching and
biodegradation have removed the bulk of the spilled material
(Pignatello, 1991), it is important to address the efficiency of
various methods for measuring the firmly bound fraction. This is
especially true for small to moderate spills in surface soils,
where microbial activity and evaporation rates are high, and
after dissipation and decomposition of the labile fraction, the
exposure of humans and other organisms to volatile compounds
would be determined by the rate of release of the firmly bound
compounds from the intraparticle micropores.
If the hypotheses that the firmly bound fraction is the most
important fraction of volatile organic compounds in older spills
is valid, it has important implications for the analysis of soils
for regulatory purposes as well as for theories of chemical
mobility and for the mechanism of groundwater contamination.
This paper reports the results of a preliminary
investigation of the formation of firmly bound 1,1,1-
trichloroethylene, benzene, toluene and ethylbenzene in a low
organic carbon calcarious soil, and the effects of soil moisture
and temperature on the formation of the firmly bound fraction.
MATERIALS AND METHODS
The soil used in this study was collected from a vacant
field in Clark County Nevada. The soil was. oven dried at 105 °C
and then large particles were removed with a standard 0.5 mm
sieve. All experiments were performed on this sieved soil.
All organic compounds were analytical reagent grade.
Aromatic standards were obtained fro- Alltech Associates and
1,1,1-TCA vis obtained frcn .-.idrich Chericsl.
-------
1303
Soil organic carbon was measured using the wet oxidation
method described in Black, (1979). Soil samples (3 grams) were
treated with 1. 2M H2S04 for 2 hours at room temperature to
liberate inorganic carbon from the sample. The samples were then
acidified with a 50:50 mixture of concentrated HjPO. and H2SOt.
KjCr20^ (2 grains) was added and the sample was then heated to
boiling while sparging with helium (20 ml/rain). Liberated CO,
was trapped on Ascarite (Arthur H. Thomas, Co.) and determined
gravimetrically. Particle size distribution for the soil was
determined by dry sieving for 8 minutes (ASTM D422). Soil
carbonate was determined gravimetrically by measuring weight loss
from acid treated soil (Black, 1979). Soil (1-2 grams) was
placed into a tared 50 ml erlenrr.eyer flask, with ground glass
stoppers, containing 20 ml of 3N HC1. After effervescence
slowed, the flasks were stoppered and allowed to stand at room
temperature. The flasks were vented and weighed every thirty
minutes until a stable weight was obtained.
The general procedure for the incubation experiments was as
follows. 'Oven' dried soil samples (1 gram) were placed in 40 ml
glass vials with teflon lined caps. For some experiments, a
known amount of water was added to the soil which was then
allowed to equilibrate for several hours. Neat organic compounds
were then added by opening the vial and directly pipeting 2 to 50
microliters (#jL) of the compound into the soil with a syringe.
The vial was quickly closed after addition of the organic
compounds. Recovery experiments for this "spiking" method
indicated that less than 10% of the volatile solvent was lost
during this addition procedure. In some of these experiments a
vial with a teflon slider valve (Pierce Chemical) was used. With
this type of vial the organic compound is added through a
silicone septum to prevent volatilization of the organic compound
during spiking of the soil samples. However, losses of volatile
organic compounds from sealed soil free vials was not
significantly reduced using the slider valve.-
The "contaminated" soils were allowed to incubate for 2 to
87 hours at known temperatures. After incubation, the sample
vials were opened and placed in a well ventilated oven at 32° C
for 24 hrs. The volatile organic compound which remained after
the 24 hour evaporation was extracted using the hot solvent
extraction method of Sawhney et al., (1987). Methanol was added
to the soil. The methanol soil slurry was heated for -24 hrs at
65° C. After heating, the soil slurry was allowed to settle and
the methanol supernatant was removed with a glass pipette and
filtered through a plug of pyrex wool. The various organic
compounds that were investigated in this study were analyzed as
follows.
TCA was measured by using Gas Chromatography (GC) . All gas
chromatography was carried out on a Hewlett Packard 5890 GC
equipped with an electron capture detector. The detector signal
was monitored and recorded using an Hewlett Packard 3390A data
system. All separation were performed on a 30 m x 0.53 ran inner
diameter Alltech At-5 column (1.2 micron phase). An aliquot (0.1
to 1 ml) of the methanol extract was diluted with 3 5 ml of water
and extracted with 2 ml of hexane. The 1-2 fiL of the extract was
analyzed without further treatment. GC conditions were as
follows. The column flow rate was .adjusted to 6 nl/min. The
make up gas used with the ECD detector was N2 at a flow rate of
30 ml/min. The ECD temperature vas adjusted to 300 °C. Sa-ples
were injected in the splitless rode. The injector tempersture was
250 °C. All separations vere carr:oi cv.'z iscth»rr ly with -ho GC
.colu-n tcrr.uc s-sa to H'C. The detector response vas
calibrated using external standards.
-------
Benzene, toluene, and ethylbenzene were measured in the
methanol extract using reversed phase HPLC with a UV/Vis detector
at 200 nm or 254 nm. All separations were performed
isocratically on an HPLC instrument that consisted of a Milton-
Roy minipump with an SSI membrane pulse dampener, a Rheodyne 710
injection valve and a Kratos Spectroflow 757 UV/Vis detector. An
Econosphere C)8 column (5jj particle size; 0.45 X 25 cm) purchased
from Alltech Associates was used for all separations. The eluant
was 70:30 acetonitrile:water. The column flow rate was 1 inl/min.
A fixed injection of 20 /iL, was used. The detector response was
calibrated using external standards.
Some of the soil samples that were incubated with TCA were
also .analyzed by using the sparging condition specified in EPA
method 8010 (SWA-846) and an extraction procedure designed for
field analysis. All analysis were performed using a Tekmar model
LSC-3 dynamic head space concentrator interfaced to the HP 5890
GC through the standard split/splitless injector. Dynamic
headspace concentration was performed as follows. 1 gram of soil
was loaded into a needle sparger and 5 ml of water was added.
The sparger was heated to 40 C and then purged for 11 min at 25
ml per minute. TCA liberated from the soil slurry was trapped on
the LSC-3 internal tenax/charcoal silica trap. After purging the
TCA was desorbed from the trap, by heating to 300° C, and
transferred to the GC. The GC conditions were as described above,
except that the injector was set for a 60:1 split ratio, and
during the desorption of the purged TCA from the internal trap of
the LSC-3, the first loop of the capillary column was immersed
in liquid N2 to cryogenically concentrate the analyte at the top
of the column. After desorption was complete the liquid N, was
removed and the chromatographic separation was performed at 35°C.
The field extraction method was as follows. Soil (1 gram)
was placed in a 40 ml v-ial (with teflon lined cap) , with 5 ml of
water and 2 ml of hexane. The soil slurry was vortexed for 30
seconds and the phases were allowed to separate. Two pL of the
hexane was injected into the gas chromatograph without any
additional processing. GC conditions were as stated above.
RESULTS AND DI8CUB8ION
The soil used for this study consisted of 53.0% medium to
fine grain sand (0.5mm to 0.1mm diameter) and 47.0% silt and clay
(< .1 mm diameter). The pebbles and cobbles were removed by
sieving. The soil organic matter was found to be 0.16%, while
soil carbonate was 9.8% as C02. This soil could be classified as
SP according to the unified soil classification (SWA-925).
After incubation of the spiked soil was complete, the
formation of the firmly bound fraction was assessed by allowing
the labile fraction to evaporate for 24 hours, prior to
extraction. This evaporation time was determined to be suitable
after measuring the evaporation of benzene spiked at three
concentrations into 1 gram of soil. The results of this
experiment for benzene at three concentrations are shown in
Figure 1. Similar results were obtained for ethylbenzene, which
is the least volatile compound used in this study, at a spiked
concentration of 24.2 mg/gram. These results indicate that
evaporation was complete after 12 hours and no further
concentration trends were apparent by 24 hours.
The experimental results for incubation experiments with
benzene, toluene, ethylbenzene ana TCA indicate that for all of
the volatile compounds tested, a sr.all fraction of the initial
spike became "firrdy bcurri" durir.- the inc'jbaticn. In Fin
-------
the extent of binding (residual concentration) of benzene is
shown to increase with increasing incubation time.
Several authors have considered the formation of firmly
bound volatile compounds to be a result of at least two types of
sorption sites in the soil matrix (Brusseau and Rao, 1989;
Pignatello, 1989). Thus, after initial sorption on to rapidly
equilibrating sorption sites, the compounds are then slowly
transferred to the less labile sites. Slow sorption and
desorption most likely result from the occurrence of rate
limiting diffusion, however, the initial process can often be
modeled as simple first order Kinetics where the rates constants
are a function of the diffusion coefficient in the soil matrix.
The formation of the firmly bound fraction can be envisioned to
occur after the formation of the labile component by slow
equilibration with inaccessible binding sites in the soil.
At the high concentrations of volatile organic compounds
used in these experiments, it is likely that there is also a
condensed organic phase present. The process of formation of a
firmly bound fraction should most likely consider transfer from
the condensed or labile sorbed phase to the slow equilibrating
sites. The data plotted in Figure 2& implies that the rate
formation of the firmly bound fraction may be proportional to the
concentration of the firmly bound fraction. Assuming a first
order dependence of the rate of formation of the firmly bound
fraction on the concentration of the firmly bound fraction, than
the first order differential equation describing the formation of
the firmly bound fraction is:
Where C1 is the concentration of the firmly bound fraction,
k represents a pseudo first order rate constant, and t is time.
The integrated form of this equation is:
In (Cf) =ln (C0)+>ct (2)
Where C represents-an initial rapidly formed component of the
firmly bound fraction.
The data from Figure 2a is plotted using equation (2) in
Figure 2b. Apparently, equation (2) is a reasonable description
of the initial kinetics of the formation of the firmly bound
benzene fraction. Pignatello (1990b) examined the binding of
several volatile halogenated hydrocarbons over longer incubation
periods (67 days) and also observed an increase in the
concentration of the firmly bound fraction with time. The
results from the present study are not directly comparable to
that of Pignatello, (1990b), where water extraction was used to
assess the formation of the firmly bound fraction. One possible
implication of Figure 2b is that the formation of this firmly
bound fraction represents sorption of benzene in clusters within
the soil matrix. In th.ese experiments, the initial concentration
of the labile, fraction are high enough to be considered constant
during the incubation period. This cooperative binding process
may be si-.ilar to capillary condensation on a micro-scale.
Alternatively, the results may imply that intraparticle diffusion
for benzene is a function of concentration. This type of behavior-
is observed for diffusion of organic compounds through orcanic
r.olyr.ers (Vc-ith, 199;).
-------
1 306
As noted above k is a pseudo first order constant. The
magnitude of k is dependent on the total amount of the volatile
organic compound added to the soil. If C0 in equation 2 is small
and the incubation time is constant, then relationship between
the spiked concentration and the firmly bound fraction can be
written as:
ln
-------
!30;
Figure 1: Evaporation of Benzene at 32°C from 1 grain of soil.
Spike concentrations are indicated.
-------
1308
Hours
Figure 2a: The formation of firmly bound benzene as a function
of incubation time at 25°C in dry soil.
Figure 2b: Data in 2a plotted in the form of equation (2). The
least squares line has an r2 =0.97.
-------
:• ¦
4.0
3.5
3.0
£ 2.5
2.0
1.5
.1.0
0.5
0.0
9 BENZENE
« TCA
10 15 20 -25 30
INCUBATION CONC.(MC/G)
35 40
Figure 3: The formation of firmly bound benzene and TCA as a
function of the incubation concentration. Incubation experiments
were at 65e C for 70 hours. Curves were drawn to aid in
visualization and have no other significants.
-------
Temperoture( C)
Figure 4: Formation of the firmly bound fraction for benzene,
toluene, ethylbenzene, and TCA at three temperatures. Initial
concentrations were 8.7, 8.7, 9.7 and 13.4 mg/gram respectively.
-------
Figure 5: Effect of water on the formation of firmly bound
benzene. The incubation time and temperature were 66 hours and
-------
much lower concentrations of volatile organic compound than in
this study. These observations imply that where water and solvent
concentrations are of comparable magnitude, the organic compound
can become firmly associated with the soil in environmentally
significant concentrations.
Because the firmly bound fraction of volatile organic
compounds is likely to form during leaking of storage tanks,
fumigation and accidental spills, it is of interest to determine
how effective standard analytical methods are in measuring the
firmly bound fraction. TCA contaminated soil which was prepared
by the incubation procedure described above (86 hrs incubation at
25°C) was analyzed using the dynamic head space concentration
procedure prescribed for EPA Method 801Q, a field extraction
procedure and the methanol extraction procedure. Results
indicate that the firmly bound TCA is only partially released by
the sparging method used in 8010. The various extraction methods
are compared in Figure. 6. These observations indicate that TCA
concentrations measured by the standard method may be biased low
by as much as 95%. The field extraction method was apparently
more efficient than the sparging procedure and released
approximately 40% of the firmly bound fraction.
CONCLUSIONS
In considering the environmental f^te of volatile organic
compounds, many workers have considered that the phase
association of the compound can be described using simple
equilibrium ot fugacity considerations. This picture is likely
to be accurate for much of the volatile material released into
the environment. However, the formation of the firmly bound
volatile organic fraction demonstrated in this study and the slow
equilibration rate demonstrated by others implies that PPM to
part-per-billion (PPB) concentrations of many volatile species
can persist f
-------
i: 13
Methanol CPA 6010 field extraction
Erlraclion
Figure 6: Comparison of the extraction of firmly bound TCA by
the three methods outlined in the text. The soil was incubated
for 87 hours at 25"C.
-------
u
REFERENCES
ASTM, Standard Method for Particle-Size Analysis of Soils.(D422-
63)
Ball, W.P and Roberts, P.V. (1991) Long-term Sorption of
Halogenated Organic Chemicals by Aquifer Material. 2.
Intraparticle Diffusion. Environ. Sci. Technol. 2_5, 1237-1249.
Black, C. A, Editor (1979) Methods of Soil Analysis. Society of
Agronomy.
Boesten, J.J.T.T.I. and van der Linden, A.M.A. (1991) Modeling
the Influence of Sorption and Transformation on Pesticide
Leaching and Persistence. J. Environ. Qual. £0, 425-435.
Brusseau, M.L. and Rao, P.S.C. (1989) Sorption nonideality during
organic contaminant transport in porous media. CRC Critical
Reviews in Ennvironmental Control. .19, 33-99.
Chou, C.T. and Shoup, T.D. (19S5) Soil Sorption of Organic Vapors
and Effects of Humidity on Sorptive Mechanism and Capacity.
Environ. Sci. Technol..19., 1196-1200.
Clay, S. A., Allmaras, R.R., Koskinen, W.C. and Kyse, D. L.
(1988) Desorption of Atrazine and Cyanazine from Soil. J.
Environ. Qual. 17, 719-723.
Di Toro, D. and Horzempa, L. (1982) Reversible and resistant
components of PCB Adsorption-Desorption: Isotherms. Environ.
Sci. Technol. 16, 594-602.
Jury, W. A. and Valentine, R. L. (1987) Transport Mechanisms and
Loss Pathways for Chemicals in Soil. In (S. Hern and S. M.
Melancon, eds.) "Vadose Zone Modeling of Organic Pollutants".
Lewis Publishers, 1987. .page: 37-60.
Lufkin, J., Steinberg, S., Parn^ll, C. (1991) Comparison of a
field screening method for volatile halocarbons in soil with EPA
method 8010. Presented at FACSS, Pacific Conference, 1991.
Ong, K. S. and Lion, L. (1991) Mechanism for Trichloroethylene
Vapor Sorption onto Soil Minerals. J. Environ. Qual 20, 180-188.
Peterson, M. S., Lion, L. W. and Shoemaker, C. A. (1988)
Influence of vapor-phase sorption and diffusion on the fate of
trichloroethylene in an unsaturated aquifer system. Environ. Sci.
Technol. 22, 571-578.
Pignatello, J. J. (1989) Sorption dynamics of organic compounds
in soils and sediments. In (B.L. Sawhney and K. Brown, Eds.)
"Reactions and Movement of Organic Chemicals in Soils",'SSSA
Special Publication no. 22. Soil Science Society of America,
Madison WI, pp 45-80.
-------
Pignatello, J.J. (1990a) Slowly reversible sorption of aliphatic
halocarbons in soils. I. Formation of residual fractions.
Environ. Toxicol. Chero. 9, 1107-1115.
Pignatello, J.J. (1990b) Slowly reversible sorption of aliphatic
halocarbons in soils. II. Mechanistic Aspects. Environ. Sci.
Toxicol. 9, 1117-1126.
Pignatello, J.J. and Huang, L. Q. (1991) Sorptive reversibility
of Atrazine and Metolachlor residues in field soil samples. J.
Environ. Qual. 20,, 222-228 .
Pignatello, J.J. (1991) Desorption of tetrachloroethene and 1,2-
dibromo-3-chloropropane from aquifer sediments. Environ. Toxicol.
Chem. 10, 1399-1404.
Rijnaarts, H.H.M.,. Bachmann, A., Jujnelet, J. C., Jehnder, J. B.
(1990) Effect of desorption and intraparticle mass transfer on'
the aerobic biomineralization of a-Hexachlorocyclohexene in a
contaminated calcareous soil. Environ. Sci. Technol. 24., 1349-
1354 .
Sahwney, B. L., Pignatello, J.J., Steinberg, S.M. (1987)
Determination of 1,2-Dibromoethane (EDB) in Field Soils:
Implications for Volatile Organic Compounds. J. Environ. Qual.
17, 149-152.
Steinberg, S. M., Pignatello, J.J. and Sahwney, B. L. (1987)
Persistence of 1,2-Dibromoethane in soils: Entrapment in
intraparticle micropores. Environ.' Sci. Technol. 21,, 1201-1208.
USEPA (1984) Test Methods for Evaluating Solid Waste:
Physical/Chemical Methods, United States Environmental Protection
Agency, Office of Solid Waste and Emergency Response, Washington,
DC, (SWA 846)
USEPA (1984) Soil Properties, Classification, and Hydraulic
Conductivity Testing, United States Environmental Protection
Agency, Office of Solid Waste and Emergency Response, Washington
DC. (SWA-925)
Veith, W.R. (1991) Diffusion in and through polymers: Principles
and Applications. Hanser Publishers, New York, Chapt. 4, 73-110.
Wu, S-C and Gschwend, P. M. (1986) Sorption kinetics of
hydrophobic organic compounds to natural sediments. Environ. Sci.
Technol. 20, 717-725.
(Received in Germany 1 February 1992; accepted 2 March 1992)
-------
Determination of 1,2-f)if.Homiie5hanc i}.H13) in Ku-td Soils:
Implications for Volatile Organic Compounds
B. L. SaWHNEY, J. J. PlGNATELLO,* AND S. SI. STtMS'BCRG
ABSTRACT
A method U d(*cdfc*d for dclffminlnj 1,2-dlbrorncxlh*(KI)8t (a
volU proloudy fumigiicd *lib KDB. Pv< p/ocrdurt Io*o1»« fvtric-
lloa of toil »Uh meihinol it 75 X fof 24 la i «»W %ltl, fotlo»csli of «hc (Unci. R<¦ pro-dueiblfI(> of lh<
rrnrlhod dtmoa\4rtl«J udflg rrplleited vimpltt by vpimlf
*nd ih« detection limit wu S I .8 ni, k|'\ Act lor* *nd icctonltnk rt
ij efficient. Olher published mcthodi for determining *ol*Jlk orjcnk
compounds iVOCi) i» will *nd vtdlmeelj ji>e *erj low rrsro^trirt
retail'c lo ib< those method, A pur^e led inp method I CP A Method
8240) ]<-i* ifcia lQri KDB, oea it higher lerofxolum ind
lonjcf ptargf ffnxi lfej« rt-commende-d. Tb*mstl doorptfoe (emporium. T*o otlrcr eorfhods.
v: aicsii-DD-ri.lrwlic-r ivdSoibid cilracion, *"klcSi ire commc rjj o>rd
Ic-r aoa*c>lilCk hhJ -sernSol* Hk compound.! I« vil, iJso ga*e low
nrt^rrto of CDS from lie fcrJ(.s.u*d fMd toils. »lj&rc-ua c-cedi-
tion =&«-d<-d ta.tilnd * l*i>!k- ccuipocrvt *!Lc tflB if* cohs-lsietsI n hi
ovt prtvlc-ta r ^lbp ili*l COB Is lie fckS unpta h (ripped Ic *oi£
tcJcroporw 'na irbki diffuse b
4dJfrianal Iftdtx P«JkrxSe rrwJu* injJ)-%h, Yotnil* acinic
comt>oucdt. Furfe lod irsp mrlhod, Sorpck»*<
We previously observed that 1,2-dfbromoethanc (EDB)—
a volatile, water soluble, biodegradable, and weakly sorb-
ing organic compound—persists in soils for a number of
years following soil application as a nematicide (Steinberg
et ah, 1987). This persistence was attributed to its entrap-
ment in soil micropores from which it diffuses only very
slowly and likely continues to slowly leach to groundwater,
over prolonged periods. Our experiments show that only
a fraction of EDB in the field samples was extracted by
the published methods for determining volatile organic
compounds (VOCs) in soils. Because no procedure has
been critically examined for the determination of residua]
EDB in soil, we tested a number oflikely procedures for
its determination, leading to the development of a
satisfactory method described here.
Methods for analyzing organic chemicals are general-
ly predicated on a high recovery of added compound
(spike-recovery) following a short equilibration lime witfT
the matrix, usually less than 24 h. Few studies have ex-
amined efficiencies of recovery in environmental samples
where equilibration limes can be years. A number of re-
cent investigations show that, with aging, hydrophobic
organic chemicals added to the soil or sediment become
B.L. Siwhney and J.J. PignaicHo, Dep. of Soil and Waier, The
Connecticut Agiic. EcTE, Glob&! Oo;!icmiy.ry Corp.,
6919 Eion Ave., Or.osii Pi/k, CA SI 303. Contribution Ticni the
Drp. oT S-cil and Wa'cr. The Ccnrrrticul Agric. Exp. Sin., New
Haven, CT. Rrcc^cd 20 Vu. J9'S-7. 'Couc-.^H&r.g I'jihor.
PcilisJ'ti s.i J. fcr.:."3n. Q'jaJ. J7::«9-I5J ii>r-S|.
increasingly difficult to desorb (Di Toro and Hor^err
39S2; Karickhoff, 19S0, 1984; Wu and Gschwend, !9S'
The kinetics of sorption/desorption has been inoc'cle-i
a Jiffu.ive transport imo.'out of intrapaniclc microtvj:
(Skopp, I9S6; W'u and Cschwcnd, 19S6; Steinberg ct 3
19S7). Thus, methods developed from short cquilibraii>
times of spikes must be considered suspect, and this r
also be iruc c^cn for organic chemicals that are weak
interacting (i.e., display a low soil-water partition co*-
fkient, Kp) with soils.
Volatile organic compounds represent an impo-ru.
category of pollutants, which includes industrial solver,:
low molecular weight petroleum products, 2nd sl
fjmiganis. fn ihc water qua iiy and ha^ardcus wss
manitaring programs of (he USEPA, (he reco.Trrtndi
procedure for aialyiing VOCs in water (EPA Mciiod 62
involves i'ic pu'ge and trap technique, where ihc ccn
pound ir sparged, from Lie system by a stream cf- int.-
gis and anal>zrd by gaj chromatography (GC) [Fedtr,
Register, 1979). AlthoLgh this meihod s satisfac'.oiy fc
determining VOCs in v,ater, it gives, low recoveries in so!;
and semisolid matrix samples (Hiatt, I9SI; Spraggins i
a!., I9SI).
The USEPA method for trace lev els of VOCs (bqilint
point <20Q'C) in solid matrices (Method 8240, revisei
Sept. 1986) requires purging a heated (4£>°C) suspcnsioi
of the sample in the purge apparatus for 11 min (USEPA
1982). Other methods pertaining to VOCs have beer
developed. Hiatt (1981) developed a vacuum vaporiza
tion method where VOCs from warmed (50°C) sedimen
suspensions were condensed in a trap cooled by liquic
N and subsequently analyzed by GC. Efficiency v.a:
based on short duration spike-recovery. Spraggins et al.
(1981) used a modification of the standard purge and trap
technique. They found a progressive increase in benzene,
toluene, ethyl benzene, and xylenes released from fielc
samples as the purge temperature was increased from 5C
to 1 J0°C. However, it is not known if complete remova,
of the compounds was attained even at the highest
temperature. Amin and Narang (1985) developed a
closed-loop stripping procedure using temperatures up to
120°C and trapping of the vaporized compounds on £
Poropak N column. Although recovery of various
organics from spiked clays was satisfactory, recoveries
from systems over long reaction limes were not deter-
mined except for frozen samples, which is unrealistic.
Initially, we obtained excellent recoveries (90 ± IO°To)
of EDB from surface soil that was spiked at 10 kg':
and allowed to equilibrate for 3 d at 3°C, by using a
modification of the hexane codistillation procedure
developed by Rains and. Holder (1981) for EDB in foo-
products. Ho'acier, this method was ineffective in
extrzJID? frcm fHd s?np!es. This obs
-------
\;.v i i.i" i - x. i i J i' <()>
1 he Mils
1 h -c.ls iVji KiJ ' 'c-iol .1; r_::i j-j'ei! *cre ..scd,
Cheshire Tint sand) 'cj~i i-n.irscloamy, mncd, imcmc fspc
D)strochrep(s) from ihe LixUood Farm of the Connecticut
Agricultural Experiment Station «3s fumigated once in I98J
at 6- kg ha". A cornel soil *js collected ffom Jn adjacent
.ututeatcd plot. T* i.'h .."cW. K >•: :. j> ". :: 'c,
Leiirc, or a j • er r Jl :ffere'st :.e-:i;catrs 'c
«.>il w.i« - . p-.".IcJ i.-.: -c ;j-:s wri n J e-J
screw vap . ul. » hit-h- * .is kept in an rrs*- .'r:cJ po*.t,ijn
(O aioid jnj s jpor loss. For "J 'C e*Katiions, ihe sjriiple «i:
r'.Kfd m a censtani temperature incubator. T7ie suspension w ai
ihen cemnfused. While the hoanc extract w as anal)7ed duecily
by GC.'EDB from other solvents -*as partitioned betsseen hex-
ane and voter and an aliquot of hexjne then analyzed for EDB
using GC. Concentrations thus obtained in the he sane layer were
corrected bs a factor to actouni for EDB left in the 3t)ueous
laser. The factor was determined by measuring the partition-
ing of 'C-ED8 (Amersham, 204 x 10' Bq mmoi' ) in the same
solsent mixture by counting both la>crs,
Sonication-extraction was carried out on 30 g soil in 100 mL
methjnol in i 400-mL beaker. The mixture was stirred briefly,
sonicated at r.onpulsed full power (3.85 W) with a Biosonik
sonicator (Bronwill Scientific Inst. Co., Shclton, CT) for 1.5
min, and then mixed briefly once again. A portion of the
supernatant was cenirifuged and then analysed byOC. So.xhlet
extraction was carried out on 30 g soi) and !25 mL methanol
using a standard Soxhlet apparatus.
RESULTS AND DISCUSSION
VV'C compared several methods for determination of
EDB in the field samples: purge and trap, thermal desorp-
tion, and various solvent extraction methods. The only'
satisfactory one was a solvent extraction technique, re-
ferred lo as the recommended method below, which gave
EDB concenlrations that are listed in Table 4. The other
methods are compared relative to the recommended one.
Purge end Trap
Thedata in Table I show that purging of the fumigated
soils for 11 min at 40°C, as prescribed by EPA Metho-
8240 for estimating VOCs in soils, removed only sma
amounts of EDB. For example, the treatment removed
only about I ng kg" EDB from the Cheshire fs! soil. A
second similar purging removed an additional I fig kg".
Purging at higher temperature and longer periods of time
increased the release of EDB; purging at 80CC for 30 min
released about 8.5 fig kg" as compared to 1 fig kg"1 at
40°C for 11 min and 3.2 ^g kg" at 40"C for 30 min.
Nonetheless, even under the most vigorous conditions,
the amount released was only a fraction ( < 11 ^o) of the
loial found by the recommended solvent extraction
method described below. Clearly, the purge and trap
method is not satisfactory for determining EDB in the
fumigated soils.
Thermal Desorptlon
Thermal desorplion of EDB in soil columns with a
stream of Ni at temperatures up to 200°C was in-
vestigated. Two soil samples were used in this study. The
Agawam fsl had a history of EDB applications and con-
tained 125 jig kg" EDB determined by the recommended
method. The residue-free control soil was spiked with "C-
EDB by injecting an aqueous solution into the center of
the column with a microliter syringe.
Es>scn:i?.ily no EDB waj desorbed from the field-
contr'r.ir.i'.cd soil c^c: the temperature range studied
150 J. F.nslron. Oc r.J., Vol. 17, no. 1, I92-*
-------
¦.'it I. Krir
nir:hr>d.
;U i 'I' f" • j :Ti r u rr. i ]
, h\ p-jr.
r nd trap
I a \> \ ( . Fhrr i:ii f d r s L>o i I j 15 f mj m t Kc \ ^ « a rr>,
¦ S •- r : • ¦ v - : f rr. r ' ? : u r r ^ f ."i r JO rn; n
* t.iil'.i >2
'""p
r-j.'^p rvo
Sv por>;t
c
nun
'K
> uf 101*1 *
CTw*hi#f fiJ-
*0
11
lit
L 1
1 $
10
11
*r>d tufcriiivc
0 1
1 0
to
10
3 2
&
30
lihi
« i
9 &
C[ 4|jf m!: gf V I )Ii '¦ vet T4bl<( 41 d etcr n-.tncsj bj* c rccoiTV
n*»cn^pd ."r.clhod.
(Tabic 2). When the thermally treated soil was subse-
quently extracted by the recommended method, it was
found thai a substantial percentage of the initial EDB re-
mained in (he soil after thermal treatment. This value
decreased with temperature from 79,"o ai 100 =C la 3.6ro
ai 200 °C. A likely explanation is thai EDB had decom-
posed. faster than it had desorbed into ihe stream of N|
$ai. In a previous study (Steinberg et a!., I987>, we
observed rates of diffusion of EDB into aqueous-sola-
lion from this soil that were consistent with highly
tortuous or sterically hindered diffusion paths within soil
particle micropores. Apparently, the residence time within
these pores is long enough to result in significant decom-
position, probably v ia hydrolysis with residual water or
surface-catalyzed reactions.
This conclusion is supported by results wit.h "C-EDB
spiked soils. A fresh spike was recovered quantitatively
at both 120 and 200°C; however, when the soil was
allowed lo "equilibrate" with the spike for IS h, only
wasdesorbed and only about 0.3°?a-was left behind
in the soil (Table 2). This suggests that the 49"7s collected
in the traps had been sorbed on exposed soil surfaces and
was readily released by the thermal treatment into the
stream of N,, whereas most of the remaining 5l^a had
entered less exposed regions of the soil during equilibra-
tion and decomposed in situ during heating. These results
clearly show that thermal desorption methods are not
suitable for determining EDB in fumigated soils because
of losses due to decomposition of EDB.
Solvent Extraction
The relative amounts of EDB extracted from the
Cheshire fsl Soil with different organic solvents are given
in Table 3. The EDB removed by two consecutive extrac-
tions with methanol for 24 h each ai 75 °C was taken as
1OOVt. Indeed, in most cases, over 95% was removed by
the first extraction alone. The amount obtained in the
second 24 h extraction period averaged 6^0 of total with
a ran^e of 2 to lO^a (nine determinations). The third ex-
traction always yielded <1^ of the local. Consequent-
ly, methanol extraction removes essentially all EDB in
the fumigated soil. Methanol extraction also removed
O&hfy steed "C-EDB qvauhsiii't'iy -fiia.'a so{ sh-c^rtl.
Tin S-Tiour.; of (IDB re:"C".~d ir.aenstd *»ih e.v.rrc-
nor. time sr.fi sr.d fcp-5o:-. o *
99
102
*9
S&
y}
:\
:j
1 6
tt J^cc Table H di '.f rrriir.«i bv the t
nit-ndcd .TicihoJ. f'cr iptVcd >4mplcs, t-*Ki+i *re bixxl on amount j
Tab3« 3, Extraction c| EDB fr&za QirsJainp /aJ set) bfd;7/erer
5o! V(Ql|.
T rt _
Ti me
fleliUve xtKo.verj
h
%
1. MVt-hjjaoL - 5 "C -
2 w 3|
l LOQl.
34
16
95
4
6«
2. Mk1:uioL 20 "C
a
64
24
60
4
24
3. Keiioe. 7S *C
2-4
S3
i. HeiAr*. 30*C
74
1
S. 1:1 KfiuieHjO, ofim
4
64
6. Acetotiilai*, 75 *C
24
114
7. Acetc-niLrile, 20*C
24
73
4
*\
&. Acetone. 75-C
24
1 j 3
S. Acoi-ort«. 20 *C
24
65
4
07
t Average of il tful ivo ^(lcrrm'a*Uoni b«$«d on recovery by Lvq 2A t
e*UricLiona wiih meihAfiol at "35 'C «i 100%.
in 16 to 24 h at 7S°C in the water miscible solvent-s.
Among the four solvents used, hexane was the least ef-
fective; in a 24-h extraction at 2Q°C, only 3^o EDB was
removed as compared to 60, 65, and 78^ removed by
methanol, acetone, and acetonitrile, respectively. Similar-
ly, removal by extraction with hexane at 75 °C approached
onfy S'S^'o, whffe near complete extraction was achieved
with the other solvents. Although acetone and acetonitrile
were as effective, or slightly more effective, than
methanol in extracting EDB from the fumigated soils, we
used methanol in later experiments because of its ap-
plicability to the widely used pufge and trap GC method
for analyses of vofatsle organics {FederoiRegister, 1979).
Thus, an aliquot of the methanol extract may be added
to water and purged. Also, it should be pointed out that,
although EDB hydrotyies iri water at elevated
temperatures (Vogel and Reinhard, 1986], il was essen-
tially stable under the experimental conditions used here;
.e., the extent of decomposition was only 3 ±
l< ft at 75" fn 9: t fv/v) metfisrcVwater,
The of the (scontir.cucfed method
h'.v-e arc gi^en be'ow:
per
J. EBLiran. Qu = l.. Vol: |1, ro, S. I9M 151
-------
Ti:.;c <. C'jr.-r * A \ n -.hr.-' *'.»! *•-, - tr'.rr-:. * -i
ty D.rihirjul fiL/icl»<)0 ti 7j C foe -4 S »o Lr^-jcti/ by l^o
>.,J \r.t!;. ii I 2
^ '
Ch.-sh.rt fO *.2 *3 96 » M
Agt»'«nn fil 121 • 12 1 JS » 21
KnfirW id 37 » 9 <3*6
Weigh 5 g soil in a JO mL glass scrcw cap vial, add 25
mL methanol, and cap firmly «,ith an unused
Teflon" backed siliconcrubbcr septum liner after en-
suring thai the septum face and vial lip arc free from par-
ticles. Invert the vial and mark the level of liquid as a
reference for determining any leak during heating. Place
the v ial inverted in an incubator at 75 'C for 2-) h. Gentle
orbital shaking (60 rpm) of the incubator, to facilitate
maintaining the soil in suspension, had no effect on EDB
release. As a precaution against breakage that could result
in injury, the vials should be placed inside another con-
tainer and not handled until cool. After cooling, check
the level of liquid and discard any vial where loss in
volume is indicated. Centrifuge (I500 rpm for 10 min)
and transfer'the supernatant into a 200-mL flask. Re-
suspend the soil in 25 mL -methanol, centrifuge, and
transfer the supernatant to the flask. To the combined
methanol extracts add 30 mL hexane and 100 mL distilled
water. Shake vigorously for 30 s and allow the phases
to separate. The volume of the upper hexane layer is not
appreciably different from the volume of hexane added.
Analyze .the hexane layer by GC and multiply the
calculated total EDB by 1. 13 to account for EDB left in
the water-methanol layer. We verified that EDB does not
partition into soil from methanol solutions within 3 d.
Thus, as an alternative to handling the entire superna-
tant and the rinse, an aliquot from the initial superna-
tant can be analyzed by GC.
The results in Table 4 show EDB concentration in three
fumigated soils determined by two analysts using the
above recommended procedure. Analyses were done in
triplicate. Agreement among replicates and between in-
dividual analysts show that the proposed method is
satisfactory for determination of residual EDB in soils.
Using a ratio of 5 g soil to 25 mL methanol, the detec-
tion limit is 1.8 ng kg'1; it should be noted, however, that
a 1:1 ratio has also proven satisfactory. The method may
be applicable to the determination of other organic com-
pounds of similar hydrophobicity.
We also tested two other solvent methods on the
Agawam fsl soil that are commonly used for extracting
nonvolatile and semivolatile compounds from soils and
sediments, to see how effective they-would be for EDB.
Sonication-extraction (EPA Method 3550, USEPA, 1982)
in methanol for 1.5 min total sonication time recovered
only 12°To, based on total EDB by the recommended
method above. Soxhlet extraction with methanol at 51
to 56°C (the temperature inside the thimble) gave
recoveries that reached a maximum of 61 ro after 24 h
but then declined slightly to 55°Tc after 48 h. The decline
rr.rv Kr.ve been due to evaporative losses. At best, Soxhlet
extraction j-o^r >iv'ds.
VM) CO NCI. I MOSS
Commonly uscJ mc'.hoJs for Jcienv.-.i VOCs
as direct puree and trap of jqucous suspensions of
even at high temperature, have proved to be ineffk
for the determination of EDB in previously fumigateo
soils. The use of high temperature thermal desorption
methods resulted in decomposition. Methods commonly
used to extract nonvolatile and semivolatile compounds,
such as sonication-extraction and Soxhlet extraction, were
also inefficient. Extraction with organic solvents such as
methanol, acetone, or acetonitrile proved satisfactory at
elevated temperatures.
As we have shown previously (Steinberg ct al., 1987),
the strong occlusion of EDB in soil is the result of ph>sical
entrapment in soil micropores and not specific interac-
tions such as chemisorption with soil components. A
model of diffusive release from soil panicles has been pro-
posed in which the rate of release into aqueous solution
is inversely related to the soil-water partition coefficient
(Kp) (Wu and Gschwend, 1986). However, despite its low
Kp (Steinberg"ct al., 1987), EDB requires vigorous con-
ditions for its extraction from field samples.
A C K N O W L EDC M F.NTS
The technical assistance of J. Brown and D. S'ellinger is great-
ly appreciated.
REFERENCES
Amin, T.A.. and R.S. Narang. 1985. Determination of volatile
organics in sediment at nanogram-per-gram concentrations by
gas chromatography. Anal. Chem. 57:648-651.
Di Toto, D.M., and L.M. Houcmpa. 1982. Reversible and rt
ant components of PCB adsorption-desorption: Isotherms,
viron. Sci. Technol. 16:594-602. '
Federal Register. 1979. Fed. Regist. 44<233):69532-69539.
Hiatt, M.H. 1981. Analysis of fish and sediment for volatile priority
pollutants. Anal. Chem. 53:1541 -1543.
Karickhoff, S.W. 1980. Sorption kinetics of hydrophobic pollutants
in natural sediments, p. 193-204. In R.A. Baker (ed.) Con-
taminants and sediments. Vol. 2. Ann Arbor Science Publ., Inc.,
Ann Arbor, Ml.
Karickhoff, S.W. 1984. Organic pollutant sorption in aquatic
systems. J. Hydraul. Div. Am. Soc. Chem. Eng. 110:707-735.
Rains, D.M., and J.W. Holder. 1981. Ethylene dibromide residue
in biscuits and commercial flour. J. Assoc. Off. Anal. Chem.
64:1252-1254.
Skopp, J. 1986. Analysis of time-dependent chemical processes in
soils. J. Environ. Qual. 15:205-213.
Spraggins, R.L., R.G. Oldham, C.L. Prescott, and K.J. Baughman.
1981. Organic analyses using high-temperature purge and trap
techniques, p. 747-761. In L.H. Keith (ed.) Advances in the
identification and analysis of organic-pollutants in water. Vol.
2. Ann Arbor Sci. Publ., Inc., Ann Arbor, Ml.
Steinberg, S.M., J.J. Pignatello, and B.L. Sawhney. 1987. Per-
sistence of 1,2-dibromoethane in soils: Entrapment in intraparticle
micropores. Environ. Sci. Technol. (in press).
U.S. Environmental Protection Agency. 1982. Test methods for
analysis of solid waste: Physical/chemical methods. SW-846 2nd
ed. Office of Solid Waste and Emergency Response, Washington,
DC.
Vogel, T.M., and M. Reinhard. I9S6. Reaction products and rates
of disappearance of simple bromoalVanrs. 1,2-uibromopropir.e,
and 1,2-dibromoethane in *zicr. Environ. Sci. Technol.
20:992-997.
Wu, S., and P..M. Gschvtnd. IS-6. Sorp'.ion l.inciics of
f'.v ire phobic orj?.r,ic compounds 'o r :: jr2.' sr:J:r-.::-.:j srd <"
T.: Cci. T>:' :ci ;0" I 7-715
152 J. Fnvlros. Qut)., YcA 17, no. I, 19£.S
-------
Envirsn 5c; Tecfwot 1937, ?i,
Literature Cited
ill Coihern. C. R.; CortigHo, W. .V: Marcus. W. L. TVchm'vues
for the Assessment of the Carcinogenic Risk to the U.S.
Population due fo Exposure from Selected lb!a':te Orqcnic
Compounds from Drinking Wcfer na the ingestion, fn-
halaticn and Dermai Routes: EPA Office of Drinking Waier
(WH-550): Washingxon, DC. 198-4: PB84-213941.
(2) Shehala, A. T. ToxicoL Ind. Health 19S5, 4, 277-298.
(3) AndeLrc&n, J. B. EHP, Environ. Health Perspect, 19&5, 62,
313—3 S3.
(4) Andelman, J. B. Sci. Total Environ. 1985, 47, 443—160.
(5) Decker, D.; DtMardi, S. R.; Calabrese, E. J. Med. Hy-
potheses 1934, 15, 119-124.
(6) Prichard, H. M.; Gesell, T. F. Health Phys. 1981, 41,
599-606.
(7) Hess, C. T.; Welffenbach, C. V.; Norton, S. A. Environ. Int.
1983, S, 59-66.
(S) Wallace, L. A.; Pelliizari, E,; Hartwell, T.; Rosenzweig, M.;
Erickson, M.; Sparacino, C.; Zelon, H. Environ. Res. 1984,
35, 293-319.
(9) Wallace, L.; Pe!])22arit E,; Sheldon, L.; HarlweU, T.;
Sparacino, C.; Zelon. H. In Pollutants in a Multimedia
Environment*, Cohen, Y., Ed.;*Plenuno: New York, 1986;
pp 239-315.
(10) UNSCEAR Sources and Biological Effects of Ionizing
Radiation^ United Nations Scientific Committee on the
Effects of Atomic Radiation 1977 Report to the General
Assembly, with annexes; United Nations: New York, 1977.
(11) UNSCEAR ionizing Radiation: Sources and Biological
Effects, United Nations Scientific Committee on the Ef-
fects of Atomic Radiation 1982 Report to the General
Assembly, with annexes; United Nations: New York, 1982,
(12) Mackay, D.; Paterson, S. Chemosphere 19S3,12, 143-154.
(13) Dixon, D. A.; Nacht, S. H.; Dijton, G. H,; Jennmgs, P.; Faha,
T. H. Methods for Assessing Exposure to Chemical Sub-
stances: EPA Office of Toxic Substance?: Wisfj.-rMn. L/C,
August 1955; Vol. 5, EPA 56Q/5-$5-005, PBio-'. i'loo.
(141 Foster. S, A,: Chrostowski. P. C. Presented at the 79th
Meeting of the Air Pollution Cor.:roi Association. Minne-
apolis, MN, 19S6; Paper S6-12.3.
(15) Bond. R. G.; Slraub. C. P.; Prober. R. Handbook of En-
vironmental Control, CRC: Cleveland, OH. 1973: Vol. HI.
p 155.
(161 Siegrist, R, L.J—Am. Wafer HV'ts Assoc. 1$$3. 75{7),
342-347.
(17) Edwards, D. K.; Denny, V. E,; Mills, A. F. Transfer Pro-
cesses: An /nfroducfion Co Diffusion, Convection, ond
Radiation; Holt. Rinehart and Winston: New York, 1973.
US) Lyman, W. J.; Reehl, \V. F.; Rosenblatt, D. H. Handbook
of Chemical Property Estimation Methods; McGraw-Hill-
New York, 19S2.
(19) Verschuoren, K. Handbook of Environmental Doto on
Organic Chemicals, 2nd e.4 barxed in 1953 by tr.t I'.S. Environment:*] Protection
Agency because of its potent Lai carcinogenicity and because
it was detected in groundwater supplies. EDB is volatile
(vapor pressure 13.8 mraHg and estimated Henry's law
constant 8.2 X lO-1 atm m3/mol at 25 °C) (J, 2) and
moderately water soluble (4250 mg/L at 2o °C) (J). It also
has a low affinity for soils as evidenced by low soil-water
partition coefficients ifp or carbon-referenced soil-water
partition coefficients Kk derived from sorption isotherms
(4, this study). In addition, when added at nanogram per
gram concentrations to surface soil suspensions, EDB is
degraded rapidly (within days) by soil microbes under both
aerobic (5) and anaerobic (S, 7) conditions. These obser-
vations suggest that EDB should disappear rapidly from
surface soils following application. We have, however,
found up to 200 ng/g EDB in the topsoil of tobacco fields
in Connecticut, as long as 19 years after its Ja;i known
application (sec below). Our objective was to understand
this unexpected persistence, because residual EDR could
be a continued source of groundwater oon'.a.Tvnation. Our
results indicate that it persists because ci Uv..»:i>j'.;3 sorp-
tion to the soil.
Neutral. organic compounds are =orbc-d bv and
sediments by partitioning into organic ph:.-es ami
adsorption on mineral surfaces. M-.-l cXL;.-:irr.e:it? in ihe-
lat.-->r.t:-ry has s beVn . r.rftime-. 'a:s- t\
sorption
-------
|it::oci of hours to schisve eq-j:libr:ur.» and rev^rsibie.
Msnv compau.nfs in soii.-sediment systems. however, ex-
hibi! hvstex=;ic sorption,'desorpcior: i&Lherrr.s, which 'lavs
prompted extra iui ."am of the ki.-mics and T*wsib:liiy
of these processes. l:i a study cf ?j3vchk-rot:phur.yl iriV-'.-i -ihezhe'e 1)4).
A po.-'.!i>n of eich soil -*¦?_< p.v-?ed. a; collected, ihr&jch
: 'j-rrm sieve and reiri^s-rated :c a used fci biode-::sda-
cic-n expiriraents. The remainder ȣ? a:.---;.-::: ird sieved
through a 2-rr.r.i sieve. To oatain *'e!-?ievtd size fractions.
;r.e air-cried suspended in w>;r.v allowed to
'J-Vi. 5o-a:-.d !; -'-lr j _-m. Trv, '.ice
:'r:.;e:.Hed wei ami uiec within 2 weeks. Total organic
car'aor .TOC) w3 = c- tc-rriin-e-d by CO: evolution fssrr.
comausieo soil that had been pclverijed ard usf-ted with
HC..
Determination of EDB in Soil. This meth«l has he en
described previously (/-#'. Briefly, the sell was emarled
with methanol in a glass screw-cap vial with a Teflon-lined
silicone rubber septum (Pierce Chemical Co.) at To ®C.
EDB was then transferred to hcxane after dilution of the
extract with water, and the hexane layer w3S-analyzed by
gas chromatography (GQ, EDB in some extracts was
verified by gas chrocoatography-mass spectrometry (7-0.
Volatilization, Soil (0.76 g) was placed loosely into a
Pasteur pipet in line with a column of Tenax GC adsorbent
(20/35 mesh, 0.2 g). A stream of dry N2 gas at 15 mL/min
was passed through for 3.5 days and the EDB collected on
the Tenax. The Tenax was then eluted with hexane (2 X
5 mL) and EDB determined by GC. The soil was then
extracted with methanol end analyzed for the remaining
EDB. .Control experiments showed that Tenax traps EDB
quantitatively from the vapor phase.
Kf, Determination. All experiments were carried out
in duplicate. Air-dried soil (5 g) and 10 mL of water
containing 0,01 M CaCl, (to aid in obtaining a clear su-
pernatant) and 200mg/L NaJ^j (to eliminate microbial
degradation) were placed in screw-cap seplum vials along
with a small volume of aqueous [1,2-HC]EDB (Mew Eng-
land Nuclear, Boston, MA; 98% purity, 23 mCi/mmol)
corresponding to 10,20, 50, or 100 ng/g of soil. The vials
were shaken gently at 25 ± 0.2 "C for 24 or 72 h. The vials
were then centrifugeci, and 2.00 mL of the supernatant was
counted in 15 mL of Opti-Fluor liquid scintillation fluid
(Packard Instrument Co., Downers Grove, IL). Counts'
were quench-corrected by the external standard ratio
method with quench curves. The concentration of sorbed
EDS was cafcufated from the difference between added
and supernatant counts, Kp was determined from the slope
of linear plots of sorbed vs aqueous EDB concentrations,
which were highly correlated in each case (r1 > 0.95).
The apparent Kt for native EDB was determined under
the same cc-nd-ti'ons except that tSe ("CjEDB amendment
was elknir-s.ted. After centrifugatioiv 5 mL of the aqueous
layer was extracted into 1 i&L of hexane ar-.d analyzed fc-r
EDB by GC. Kp was taker as the ratio of sorbed EDB
[determined by methajiol extraction aa above) aqueous
EDB concentrations.
Octanol-Water Partition Coefficient. Stock {,J|CJ-
EDB was partitioned between equal volumes of hexane
and water to remove any water-soluble impurities, and 20
/iL of the hexare layer containing 1 X 10s disintegrations
per minute (dpm) (157 ng) was added to 1 mJL of I-octajiol
and 12 mL of distilled water in 14-nnL screw-cap vials, in
triplicate. After being shaken vigorously for 30 s, the vials
were centrifuged (2000 rpm, 30 inin) in an inverted posi-
tion, and 1.00 mL of the aqueous layer was sampled by
syringe through the septum of tfie inverted vi'af. The n'ai
was then opened, and 0.50 roL of the octanol layer was
sampled. The samples were counted in 15 mL of Opti-
Fluor.
Release into Aqueous Solution: Purge Technique
!n 'onis !he rste of release of EOS from (he
soil tv'3? d';ts:--,::i=d by a purte technique ri:r.i!»r in puV.-
:d jr.ethc^is ¦Pu.-gi.15 wftc p:r:: /r::d in 250-ml
tv'.tk-s -M:hcn:lf::i;od il-V=i- r.< ririr-.-i;.
ustd for E".- di;ps:.»iyr.. Ct?.s «3i ir.tr into tr,«
-------
T&feie 1. Apparent EUii FAr;.'.j:ti Cc*[ficiecii aad Oihtr l>f.u Coc:
Limt dt^plirg
l?'i 'si'. 24,c-.. loul organic r^..i>£ t.C&, or, 5>,:^.:or. hiw c on cV?or-y. ,;o.
5oiI >iri-T tsrboa, ng.g of adtffd |;iCjE03 w m!w E.CS
I^ocki-jrad 0.9 1.11 |30 5.-t9±Q.&i' TiV
Warehouse F'-."'3 1.61 i-- 2.^3 ± 0,2 3v»
Brtitd Bloc's 13 I.S5 27 1.7O±0t.2 , 170
¦Siispc-nsL^g or 5 S pr joi! in .& pil of *aac, 25 ^C. 'The S5% *onriri*nc3 I-,rr»it5 oi cf sorted vj aqurcu? I4C i:onccnir0T;-OM.
' Esliprsied uncen&iniv. ±12%.
to two trapi placed in aeries containing 10 mL of hsxane
in g'asa i+ cvi. scr-ec-cap wpt-jfii ^sh. Gas .1nsistiei nei£
adjust? J by use of a Linde :"5cw meter aoth a Tuit meterir.j
¦*E>T-e.
To the bottles •¦».*££ added either 100 c:L of distilled water
or 10 g of soil and 100 mL of 200 tog/L NaN3 in distilled
water, An aiiquciL of i1JCi£DB was edd-to through a
Tenon tube, and tfi-e contents mew jifrred fit mwjb ieo-
perature, with a Teflon-coated magnetic stirring bit, for
5 min (distilled, water) oi 3 & (soil suspensions.} before
purging was iriti'stec!. PLrp'ng was performed for 1-Omi.n
periods, after ni::i tie,e %es Dc« wss bLsppec. ; ti:
'nexane traps wercchajig-ed. An aliquot of the h-esane was
adcfed 5-^0 (iie fticlion. WeI =a3
C«iaiva]«Ttt to 4 | drvj was =uspetid«i in aqueous medium
KnL] ccnt®iii| CM .V3 CtCL acd JtOct'LjuiijE
ecide ia li-triL E.;re%-rip i-ials. Aftercer.ln'5«g&tion,tlie
supernatant waa discarded snd isplaced «-ith tush me-
dium. Tlje Wais were iJssn raised an a hetnaic'.ogy miter
(Fisher) in an incubator at 2S'°C. Thii provided gentle
agitation to keep the soih in suspension. At each time
point, two of lh£ leplicttte lisJs wers ctntrihiged, ^nd 5
mL of the dear superra.Ujit waa extracted with 1 mL of
heiane and aualyv.ed by GC.
To determine EDB rehxm train puh'eri.zed JO)!, ib-dric-d
soil (1 gl was placed in the capsule of a mechanics! tall
mi!! (Wig-L-Eug, Crsscer;! Denial M.Tnufscturij'.b Co.,
QilivCj'O, iL) Tor varjinr periods. TJ)e espsuJe ft'ss Of.»rn«J.
quifsb" Croppni ir«io a -V.--".',. vial ccnuir.inj 25 mL of
*atsi, and shsk«n for Is rain. The *-ial wsi then ctntri-
fufed. ^nt for i^OB dE'.irmLr."(ion by GC.
NijSc on ^Lji;ts:Lr;:i!'r.- lo-sc-E cDS dcrir;g sa.-njili.-i;:
Th-• f-ienrys „-v,. • :':.- ^^"3 r.'r'J>:j :h-:: o:
flir/water mixture at 25 'C. Experimer.taily, using
J"CJEDB in distilled water, we IojjicI 3 i 0.5% in the
headspace at 20 'C- In the desoiptiart d«t; :,V>;d
«bai'e, vials contained rr.cch less thin '.C'^r by vclurie of
h?adspa-ce and were always coo3ed Id rooni temperature
before sArnplmg. Samples were obtained ailh a glass pipct,
taking caK not to agitate the solution o: allow bubbks of
sir U> run up the pipet Under these condi'icns, volati!-
izalion losses during sajnpling were unimportant,
. BiodcgrndBttqn Studies. ?crbiodegr»dation c-xperi-
Jrjf/jts, soil {1.70 i 0.01 g on b dry weight basis), autoc'.aved
iistC-'si "-a" • i <;..c it Li, =jjc 1"'C E03 ,n J 5C .Li! was allcmied so stand overnight
in t-j M HQ and dried at 12$ *C fur "1 h, anlh f jbsi.rface
dejjlb sad location at each site; for experiments teported
bere, or-.e atreplt from tiie Usp 5(1 co was choan from each
site, and the EDB conceritrationa ate given in Table 5.
Topsoil saxoples from a fourth airricullural site in Sljxis-
bury, CX which was last fumigated '.vith EDB 19 years
prior to sampling, gave EDB concentrations in the ranve
1-3-2 ng/g (unpublished daUS. Thes-t results indicate s
high decree of resistance of EDB to rr.chilUaiion a^d
depadation in the Held.
The native EDB is essentially nonvolatile. Air-rin ing
soil saraples in a Buchner lunne! under sspirater s-vscticn
iv - 2A h did not dec;-.--?? ?.DB concsau::-:rJ. P.-;=:rg
r. r ¦. c ,-y \ .= : : ht- > of .'-i"'.-";'. ."0 '.'C-i'jrTifs r-: ;¦ ¦. *
;¦: voiuir f tj vri! r>:r r'.;:ju:e to: X-5 c-.v-s ;ci" p. sv..: =
o* f--"ckv.--. tc-2 rectoi-ed enly about SSI KD3. Tr.is
cvr.t.-asts iKarply wiih ^viratioiv :v. ii-.:-.i \~.t
S ¦ !! F.PP- '¦ l! =:i!i-itrin a :V-v nour- at *C f:
-------
Figure 1. Removal ol EDQ from soil suspension by N, purging.
(Bottom curve) Removal ol nalive EDB from locXwood soil. (Mkidle
curve) Removal ol a {'"CjEDa spks (10 ng/g> Irom the same so3 after
a 3-h equS-itialion porkc (Top curves ftamoval ol |l4C}ECB £15
ng/mL) hom distilled walea
Ats-o, we found previously CM} that vigorous conditions—
me:!i =_ncl ac 75 "C lvr?-i h—were required to extract EDB
quantitatively from the environmental samples. By con-
trast, much milder conditions gave quantitative recovery
from spiked samples, even when they were allowed to stand
for several days before extraction.
Partition coefficients based on sorption isotherms of
[UC]EDB and on desorption of native EDB for the three
soils are given in Table I. Sorption of ("CJEDB appeared
close to -equilibrium" after 24 h, since the resulting Kp's
increased by an average of only 147u after a 72-h period.
The corresponding Kx values (in the range 103-134) are
in agreement with the values found by others (4). The
octanol-waler partition coefficient of EDB was found
to be 86 ± 6 (mean ± standard deviation of triplicates).
The Kk is somewhat higher (by a factor of 1.4-3.6) than
predicted by several empirical linear free energy rela-
tionships between Kx and K^ that have been developed
by others [13). These relationships assume that the dom-
inant sorbate-sorbent interaction is hydrophobic parti-
tioning of sorbate into the soil organic matter. However,
as pointed out before (J3), other, nonhydrophobic con-
tributions to the sorption process could increase the Kk
values. The important point to be made about these data
is that the Kf's based on desorption of the native EDB for
the same equilibration period were about 2 orders of
magnitude greater than the K„'s from sorption isotherms
{Table 1). Clearly, the native EDB is far from equilibrium
with the bulk aqueous phase.
Release of native EDB into aqueous solution from the
Lockwood soil suspensions by purging with N2 gas (Figure
13 showed that, whereas the freshly added EDB (3-h
equilibration time) waa rapidly removed from the soil,
native EDB was removed much more slowly. Purging for
a period of 100 ruin removed the freshly added EDB
completely while less than 5% of the native EDB was
removed.
Release of native EDB from the Lockwood soil was
highly tempcraiure dependent ^Figure 2], Apparent ar-
livatic-nparameters uereoV-iirc-db> neating EDBreiease
as a first-aider process. A. The effect o" hydrolysis would bo in --educe both
sqeeei:- ?.r,d solid-ph.r-e EDB concentrations. The per-
tinent rW.ivr.; are
S c
Time (h)
Figure 2. Eftect ol tefnperaiixo on nattve ED3 re'-sasa Irom LocJtwocd
soil Fraction released = |C]/[[S] + JC]) (see text).
where S = EDB associated with solid, C = EDB in aqueous
solution, k is the cesoipiior. rate constant, arid if, and
are the first-order hydiolj'sis race constants for the Mr hied
and aqueoOs states, Tes pec lively.
As the following discussion will show, hydrolysis can be
ignored if it is assumed that hydrolysis rate constants are
equal for the solid-phase ajid aqueous-phase states of EDB.
Macalady and Wolfe [20, 21) showed that alkaline hy-
drolysis is greatly slowed for sediment-associated molecules
compared to molecules in solution but that neutral (pH-
independent) hydrolysis was unaffected by sorption over
a broad spectrum of compound type and solvolytic
mechanism. EDB hydrolysis is pH-independent in the
range 4-9 [17, iff).
Under the assumption that the following
rate expressions for the above reactions hold:
d[C]/dt -MS] -MC] (t)
-^{SJ/df = A(SJ + *„[S| (2)
Integration of eq 2 yields
[S] = [S],(3)
where (S]0 is initial EDB concentration in the solid.
Substitution of eq 3 into eq 1 and solving the resulting
differentia] equation gives eq 4. Substitution of eq 3 into
eq 4 and rearranging terms gives eq 5.
JC) = [S]0(e-*' - e-^1) (4)
r [ci i
In (fraction remaining] = In | 1 - I = -kt
[_ Pj + [t'J J
(5)
Equation 5 now allows calculation of the first-order
desorption rale constants without the need for hydrolysis
rale constants. The data from Figure 2 at each tempera-
ture fit eq 5 well (r! > 0.91 in all cases). The Eyiing plot
shewn in Figure 3 yields the apparent aclivition enthalpy
Afl" = 56 ± li kj/raol C1S.3 ± 2.3 krsl/ra!].
The release of EDB fiam soil joiticlss intc-aqueous
solutior. may aisg-rn-fitr. sly bf t-esttd ss diffusion from a
pMC-us ec=orr>;-nt An anajyltiil solution is svai'ible i.2ij
for raia! diffus-x-n ftc-ta uniform s>r.er«i eysi. tadias
into a well-stirred i-ilu1..:t. of limit--:: v.iurr.e: the c-x^ct
eqtjjtion icq 6.33 o: rc-f '22), which is not reproduce: Iv.-re.
is ol' the form
CJC, = "• •?>
-------
• 15 i. i * < ' 1 < t—> t J
3.5 2.7 2.9 3-1 3.J 3.5
FJpuro 3* EyrirfS- ptol lc* natf*e EO0 desca-ption from Lix^wvcod whol*
so? using first-order rate constanis from Ihe (lata in Figure 2.
3 OC*
FVjufo 4. Release ot native EDB at 25 °C from Lockwood 2-53-^
pa/ticles and theoretical curves based on a diffusion equation fge, the
raean particle radius may be used to approximate the idea)
ca=e of uniform particle size (23). Wet-sieving techniques
were used to insure that wBter-stable particles we're being
represented in the experiments.
Figure 4 show; the fractional equilibrium vs (!/! for the
2-c:o-.i;rn size fraction, obtained from wet-sieved Lockwood
soil end ihe theoreiicalcuri-es forsetvrai D,* corresponding
to, the mean particle- radius of 13.S urn. The 2-53--m
i'r.-siti.v- of ihe \Vijehou=e Poi-t Foil gave s::v.i!ar re=x:!u.
The values of that b?;i fit ihe data tr-tethc-r with
t-i• rua'es of the time; r.irded to reach rurc equi 1 ibrium
are listed in Table 11 for '.he LocViwoo-d and Warehouse
Twna T >me ["i 'i
Figure 5, Release of naJive EDB at tva temperaiufes as a r^nciton
of panic to size Tor the Lock wood sofl: (circles) 2-53 pjr>, (squares)
5J-106 srxj (triangles) 106-250 ^m. Each point is the mean of
dupnea'e samples that agrood lo within ±0%. Line* represent the
feasf- sqosre fit of means.
11 1
1306 • T
<
_
-J
!
r*
"V
r
HATSVe EDB •
sol-
-
CD
O
*
'Vn
1-"c- = ca 5 =¦ IKs
E
- ©
—
NATIVE EDB
1 I
v 14c-fdb spike
i t i i
Po:.v: ?:¦?!.
The
ir.cirv.e ex'.rerulj
O 8 16 J4 32 <0
Titne (d)
Flflcrs fl. EDS degra<2afrxi in sci suspensions try rv5v>?"o 0.95). The
slopes corrpsptiBdinf to the two b.'cer fractions t.t>3-:0o
ar.c 10C-2.VD ural are iv:,i itpif.ca.v.ly different at the OW
ccrfic*:r.re leve! at either ter-fversty:'. -i-:t e ccrre- .
sportc\:k to the smallest ;r,:C".icn i2-53 ^.".u is si?
d::'frr-.-nt fr.;;- both larger fr.\;tLni at t5 "C. but :
the fraction at "3 °C. F .:rt;.v::n .rv. the
= :?r:l:trk ' C-^:n i frict'ivn rvit^;ei EDB at :nc.-t C'-fi.-'.i
-------
frora H'et-Sicverf Soils
3: -
1 ^
u
EDB soncn. r.?. ^
die.caec*r rar.^e. um Lini soil
0^ 4 « e 10 12
F'uIt 61"'I*1 "V" ^"r>5 (Mint
FTju-o 7. Percwil o( 8DB reteased ticm the Warehouse Point sofl Ho
water as a functton ol pfivariiBlicfl tiro in a ball mifi. The rsngo or
duplicate <5e terminations, whan larger lhan tha size cl trie s-ymSol is
Indicated by ercf N2 as rieit'ribed sb-rve.
Th- d'.-'.rib'ilion of riL3 a:r. - ,.v:--^r!i r.-.-iiz'.'i
LDS-250
5J-106
2-53
0-2
69 ± ia
1« t 5
III t I
21 ± I
KM ± 10
15J i 12
65 i D
34 ± t
* Range of dupticaU analyses.
avtemir.e':
LockwOod and Warehouse Point soils (Table III). The
results show that EDB was at maximum concentration in
the very fine sand fraction (53-106 (jm), declining with
larger or smaller size fractions, and at minimum concen-
tration in the clay fraction [<2 >im).
Discussion
EDB in the fumigated soils is extremely resistant to
volatilisation, release into aqueous solution, and degra-
dation by indigenous soil mltrulies, nliii.ii were aimulta
neously able to rapidly degrade freshly added [,'1C]EDB
at comparable concentrations. Pulverisation promoted
release, both to the aqueous and the gaseous phases. The
results suggest that EDB is entrapped in soil micropores.
Obviously, EDB in inaccessible regions becomes available
as these regions are exposed by pulverization. The in-
ertness to microbial degradation (Figure 6) suggests that
EDB 13 present at micropore sites that are sterically in-
accessible to bacteria and that do not equilibrate readily
K-ith bulk sir or l/quld phases. Although the high activa-
tion enthalpy [56 Jr 11 kJ/mol) may be consistent with
desoiptioc from a chemisorbed state, this is unlikely be-
cause EDB has no strongly interacting functionalgrcups
ind, in any case, can be ruled out on the basis -of the
piJ^eriialwii experiments.
The great difference in apparent 24-h Kf between added
and native EDB is Likely due to kinetics rather than
thermodynamics. From either a sorptive ordesorptive
direction, Kv is by definition only valid when true equi-
librium. has been reached. Karickhoff OS) sliowed that
sediments continue to sorb h>drophohic com pounds from
water indefinitely at a s:aw rate alter the initial rapid
uptake. Presumably, the native EDB is material from the
original fumigations that sorbed at the slow rate, and which
we new demonstrate is highly resistant to mobilization.
The nature of the soil micropores is at present obscure.
Organic compounds, particularly organocations, are known
to penetrate and bind at the expandable interlayer regions
of 2:1 layer silicate days of the smedite-verroioiJite group
[24). Sorption isotherms of EDB an Calt-: aturaled
montmoriUonite clay as a function of relative humidity
suggest that EDB enters interlayer regions at low, nonzero
humidities (25). Water appeared to displace EDB from
these regions. The dominant clay in the soils of this study
is illite, which does not have expandable inteilayers.
Nevertheless, even a trace of expandable clay conceivably
could be responsible for EDB occuision in our samp!es.
Tsblc ill, ho'-ve.ei. shows that ihe c'.ay fraction <<2-nn
effective part.clfcdiaireter) contained Ihe iavtit amourts
of EDB by f=r These rejuJis ce^oniimiB ih-: cia> ir-
tc-rliyer regions do not play a major role in eniraprrten'.
of F.DR.
of tDB in micropores suggf.-ts thai re!r&?e
inw auls: i^^tic.-; is diffuriqa-cotv.rcwed. By yje- of 2 raii.C
'difi'viirin m'.'-^.ol. :be «:o^la:ed rf:'ec:i-.e D.* > f
EDB v Lt"r.:r.' p;r:i:!e5 is V !?'L' c:-.- = 'T'-:'' i'
-------
nac'.:on ar-j iriiival uniform cisuibutior: \i
i.ndjvicuil P--' :- = -3? ni p:.r or.. n .;'.ui.::j'r:j dr fjiior:
coerilcie-nsof E:fSnic Tione'etlroKtes in wsLi-r h't
i.n tfe rings jiT'-lC'"3 cci'i's..
Tc jnceri-'.a.'ni' tlitse d: Tf usi. ¦«.£ consider tite
nodel c.r ttu sr.ri C-scSwend C.*21 f;r d.rfd-sic.r. of 'nvxr-zr
phobic organic cceipojnds from soil.'s—diiaerL particles.
This ir-odii assu-ics ikBr- '.r.dividu-il sU-bl-e particles are
aggregates of fins mineral grains and nature} c-rgsnic
matter. A compound diffuses in or on: of pore spaces
between the graLis, ftna this diffusion is. retarded by
takrcicile p-ir1.itioring of the to l.ipound between jure
fluids and pore solid surfaces or OTg^ric phases. The de-
lived equBiion relates to Kv:
D*rj[n,r)
¦rt= U-n)p^ + n [i)
where Da is the pore fluid diffusivity (cra'/s'J, rt is the
porosity of the sorbent particles, p, is the specific gravity
of the sorbent, Ke is the equilibrium partion coefficient,
acd f(n,T) L3 an undefined function of n and the tortuceity
of diffusion paths r.
A correlation of Dtfr with Kr using data from se. = .-al
system tscrption/de-sorption of various chlorcbei^erces
in sediments and soil, sorption of ketone on sediments;
FigtJie IlI of ietf 12} predicts a D,^ for a compound with
Kj = 1 [approxiniately equal to the Kp for EDB deter-
roin-ed from sorption isothertas,-Table If of about 10"'
CM®/3- This is at least 9 orders of magnitude greater than
the experimentally determined values (Table 11 >, which
am based on the release of EDB from the 2-53-jim size
fraction. Conversely, the cnodel would require KP for EX)B
of about ID3 rnl./g. The finding here thst release is rela-
tively insensitive to particle radius (Figure 5} may md.-zs ie
th.it release is mostly controlled by diffusion from micro-
pores of substructures within the particles that are com-
mon to all size fractions. Therefore, the determined values
of Dtf m&y be upper limits, making the discrepancy with
the predicted, values ever, more pronounced.
To further i'.lustisi« the contrasts, consider the de-
sorjj'.iua t.me of <2 days for 1,2,3,3-let rncb In rob&crceins
train sedjner.U U2) to that of decade-s lot EEE, d-ispite
tSe fact that tetrachlai obenren e is- much more hydropli cibic
than EDB. (The for tefrachJc-rcbenzene is about 3D1
times greater than that of EDB.) Furthermore, EDB re-
lease Ls strongly temperature dependent compared to
alight, if any, temperature dependence of 1,2,3,4-tetra-
chlarobeaiene sorption rates to sediment [72) The ex-
tremely small diffusivitiesand the large temperature de-
pendence, reflected in the high apparent dJiindicate
ertreciEly bortuoua oc aterically hindered diffusion patha
through microporous structures. It is known that d iffusisn
of !:-i«3tes-eti Jiin f»Biacs acJsorfcents csr be accompanied
by appreciable apparent activation energies (55).
An important implication of the diffusion model is- that
the-time needed to reach iorptive equilibrium for a givea
compound a-id soil wrD depend on the stirc end tortu-r^ity
fsctors assixiaUd ¦> :'.b the raicropjores. This was. pointed
oLit by V. j 2Ttd G^chw-end (j2). Presumably, joik preiess
t r:?.diti.ori cf pore sizes snd etfoctatfrd tor;ccbi;::-s. l.i-
c,- " :i ri .i'.c" ?i:r-5 -v.;! i - ;x p.ii '-1-" :r trr:0-
Hiihc: c: wf-r.traiw '.i of tl.i ! ir. /'u's ^r.
¦A-ii! i.'. ir.ore cb:anicj: d:lr"-.;-ir^ ic a p=r;icul?.r p;:e
in = win i' C-c si-tsi-j-'W-t =r-
I r. :¦:! '.he -'^rr:i--= ¦ ¦ c t!-.¦:- .
t-x::vi r;-fc Wa grw.pr Jcpeiiatncfr
of re.ta=e. Lt follows i-^r.i :h:5 thi: :r.e rapid =-a:t = =r;
Letr.pjr;.iur-= o>pendency ctierved by Wj and
Cirh-'Md U3I v,ay be r^prejenut^e c: relj:o ei;.- Ian It
¦¦''.is =-.nce"Kcse in'-^-tiga-ors visej short scrptior, pe,'icijs
and mor.itcred the bu?k cf added cherries'.
T'n-: crj1 due -..T EUB in '.he f jnv.gat^d =oi!i Lhj: ie v.r.i
msy be a srniij fTaction of the ^dded chrrr-ic.ii thjt diffuses
to rer.ic'.e aiicropore sites as a cor^eqvence c; '/Or.£ expo-
sure times, high initial field-apjlied conrentratior.s. or
effective penetration m the vapor state. The bu.k c-f th =
ir-itia'-lj' ?orbed material masS ;i^ely conformed to pre-
diction and. canssqueTitly, evaporated, degraded, or
.'eached to groundwater relatively qu irUy The ir-njiider
5 = = iatc-tsr:-: it can slowly leacb :.jt c-::
y--,ars to supply groundwater with con central ions thai may
be considered significant to hunaan health [0.1 ppb or less).
These findings show that direct application of a volatile,
aoftble, and weakly interacting molecule like EDB car
result in its tenacious binding to soil particles. Other
compounds, that come, into direct contact with soiis, need
to be investigated to see if they behave similarly. Clearly,
mathematical models descriSirg the fate and transport of
organic compounds in the environment kotM hi; to pre-
dict the behavior of EDB described Sere. Standard fcio-
dEgradatior tests are also inadequate. Persistence in the
field despite ready microbial transformation in the labo-
ratory has been ncrted for a number of pesticides {27).
Finally, these results may have bearing: on the availability
of toiic chemicals in soil during human ingestion or skin
contact, and in regard to soil remediation efforts.
Acs novfiedg merits
We thank D. Meliinger and J. Brown For their able
technical assistance.
Reg-istry No. B.-(CH,),Br, 106-93-4.
Literature Cited
(il S'juU. D. R. fnd. Eig. Che~. 1947, 39. SH.
(?) Rathtun, R. E.; Tai, D. Ewissn. SiTschnof. 35S&, 20,
94S
• :0 SoJviifui'es o{ .Wrgc--*: r.-.d O'gasvc Compound.s; Stepher..
H.. Su?her. T., iEids.; Macreillan: N'r Vo.-t, t363; Vol.
t. Fart T, p 3BI.
(tl Mir.gElpri.-i. U.; Gerstl, Z. J. cln -or Quof. 1933, 17, t.
<51 PienaUlio. S 3. AppL Erwiren. AficrDtfot. ISS.!., 5/. 588.
ISJ Pignaiello, J. J. freprinls of EitendcJ /Ibjl/Drti, 191st
National Meeting of the American Chemics! Society. Di-
vision of En vi roiuuealal Cheoiistry, Ktw York; American
Chemiee) Society: Washington, DC, 1SS6, paper 4, p 8.
Hi r-3T.-c. C. E.; Belser, N O. Snt-iron. 5c. Tec/jnol. 19GSr
2. 779
(3) Di Tore, D. K' . Hc,-rerr.ps. L. M. Snvirnn. Sci. Techs>cl.
11S2, J6.£9i.
f9> KariekbofT. S. ~Vf. lr. C^~:ar.<:^ ? - f.= i f r.
ft. A-, Kd.; Ann Arbor Science: Ann ,Arl»r, Ml. IS3C", p t93-
1103 Cur!, R. 1..; Kfoielsn. C- .Cr, l>o n S (:s,i r*. > ?. F.r. irryn .y:.: . -o!. I-.r.
"73 ~.
r. "i.:._:f 5. V.\ J. H-,.-.. it.-i HQ.jjl.
.-t-j 15 1. . .J . . r. -: -:. 5, M J
- :... .n .¦
.'»• hjrsti;.j. ?. z..ts.: : i:*s.
-------
Nifi. K. N„ £di.: .*C5 ^yinposiv:^. 3U: A:r.Jfi'S^ortv- DC ivi'i. ;*
U5& 5. A.; Cojien, a. 1. Pizp '•:*:* ..-'jfar.-Tcr-rs: ,-f ¦•¦
sirqcfs. 19 1-m National Meeting 01 iht American C twr^icaJ
Socieiy. Division or Envi'^nmenw? C.Sf-e»r»irv, Ameiican
Ctamicat Socteij": Washings, DC. 19SS: pap^r S. p ;2.
IW} Vo^ef, T- M.; Refrhafd, M. Em-iron. Sci. TechncL I9i&,
20, 953.
(£0f C. L.; WcJit, N. L, Ucafa3. 3-?, 15-3-5-
(24> BorchiH, S.. Hayes, M. H. B.* C J. tn TKf
of S*:>i7 ft oress?r; -Gr3c?n1s<-.d. D. J-. Haj ts. M.
H- B.. Eds.: Vi'iicy: S^w Var"k, 1951, p 221.
<251 CaJ!, F. J, 5c7. .FooeJ /^ic. L9S7, £, 630.
•{2St D- 0.; T^p-O^D 0. M. W. Bot-
Lcrworths: London, lttoi.
{?7> Al?*jrnJcr, M. Soil-StL. See, Am. Pr-
ethj-lene (1,1-DCEl a:id acetic ecrd, as shorn by oihen.
Tri:,i study torlirms thai TCA can be b'jtrELr-sfj-irnec ty
j-eductne deWajeiuitvon tal.l.-dic'hlorMt'darie U.l-DCA.h
and ch]oroelhane (CA) uridec met'nftrioger>ic coaditions-
Abo, reductive dehalogerstioo of 1,1-DCE to vinyl chloride
(VCj 13 cor.firirsd. Tbia study deroonslrates ttat these
trBJisfanuationi occur staichicmeiricaL-i-. In sddilion,
(l'C|TCA-, [14CJ-l,l-DCA, r"C)-l.l-DCE, ["CjCA, and
islClVC f-irt at least-pttrtiaJfy tnir.sra^rrod la 5>¦ z mined cni c.r. jg¦. ¦ir c.-r. ::- "-5
rC.-'l.-¦ ¦¦ --r- v'b.v-" ^ in an.:_:¦ - .c ./r - ¦¦ :i
eJ ijr-ii nji'-ior-i sr.d acetic acid aa a rejuit of hi'diotysi's \
The psevid<»-firit crder rate tonEtint for 1,1-DCE fonaa-
ticri frca TCA was lepcrted 6t 0 W yr'1 at 40 "C (Sj.
Vi'hile no rate toristant tc-r the tonr-.B.tion of ecetic acfd
hz-ca TCA has yet been documented, 'Haag el al (J Si have
shown thatacetii acid is produced 5 times as fast as 1,1-
DCE at 40 °C. Thetefore, the pseudo-first order rate
constant for the disappearanceuf TCA at20 °C could be
es t-igb as C.25 yr"' (SJ. Tbese values are consistent with
a TCA hydrolysis {to atetic acid} rate of aboii-t 0.2 yr
Both acetic acid and 1,1-DCE tan be fiirttiet biotrans-
forraed under tnelHanogenic funcitiocs. Acetic scid can
be ji-irerallied to CO] and CH^ {14), Traces of vinyl
chloride (VC) "r-ere observed after addisioi: :if 1,1-ECEuc
tnicrocosiris (J).
On tile basis of tfce efaove-, a possible scheme for tile fata
of TCA and iu stiioficaid biotic tr&jisfomia'.ifln products,
tncWdirvg that of 1,1-DCE, 1,1-DCA, feed CA, is in
'Figure 1. Research to date has not cltaily demonstrated
whether the indicated bloconversioo of TCA to 1,1-DCA
and CA nnd of i ,1-DCjE to VC n PAj. Osact ::ji .tb used v.eie
MeOft, acetone \9S.9" ; J. T. Baker C'ne.m:tstCc . Pj-;il-
rps'jJT?. E-irt 2-orojjaro! '99+^: A!drsc^> Ch?mi:?.l
Ca, M-'l-vevk-ee. V-p'C/TCA, t^Ci-i.l-OCA ['-CI
1 !-nrE. '-r.r. ;,%C]VC iSr-j- ?.r;! = ad NurlMr, B-s?c-n,
MA' •** r- :;-.d ir.i:ia"y ir. J- T.
Bafei-j Chemical Co., P.S'^i r- "-re. S'i to 5 1 x I¦;-
_L ~A ; ci' 7CA "L ol 'Mi-OHi. 2-S x IU"
d;:m , L [-1: r,: !.l-rw"..\ '~L of •!* x 10'
:r. _i. >c- ' -¦ -.:.!_ ' : V^jKl. sri 1.1 x
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 32
Review of VOC Sorption Behavior in Soils
Martha Minnich
Lockheed Environmental Systems and Technologies Co.
Las Vegas
January 12-14, 1993
Las Vegas, Nevada
-------
REVIEW OF VOC SORPTION BEHAVIOR IN SOILS
Martha Minnich
Lockheed Environmental Systems and Technologies Co.
Las Vegas, Nevada
There has been a surge in recent research on sorption of VOCs by soils and sediments.
Topics include prediction of equilibrium sorption coefficients, vapor sorption on dry soils
and sediments, modeling nonequilibrium sorption and transport phenomena, and rates and
mechanisms of sorption and desorption. Understanding of sorption processes is
necessary to improve sampling and analytical procedures for measuring soil VOCs.
Prediction of Equilibrium Sorption Coefficients
Predictions of sorption of nonionic compounds based on the concept of partitioning into
organic matter (K^ values) have been widely accepted (Chiou, 1989; Karickhoff, 1984).
The correlations are predicted from log-log relationships and the 95 % confidence interval
is about one logarithm unit, producing sorption estimates within approximately one order
of magnitude (Hassett and Banwart, 1989). Explanation of the variation between
measured and predicted values are generally based On qualitative distinctions of organic
matter or on the role of minerals in low-organic matter soils and sediments (Mingelgrin
and Gerstl, 1983; Gabarini and Lion, 1986). Variability in the composition of soil
organic matter produces differences in partition coefficients not greater than a factor of
3, provided that the soil or sediment organic matter content is >0.2% (Rutherford et al.,
1992). Allowing for these limitations, estimates of the sorbed fraction of weakly sorbed
VOCs in a soil that contains only 0.44% organic carbon range from 20% to 80% of the
total mass (Siegrist and Jenssen, 1990).
Increased polarity of VOCs (which occurs primarily if the compound possesses any
oxygen or nitrogen functional groups) increases the interaction with mineral matter
(Karickhoff, 1984). Sorption will be affected by additional bonding mechanisms (other
than hydrophobic) and may be related to the soil mineralogy.
The variable composition of organic matter can also be viewed in terms of polarity.
Fresh organic inputs (such as cellulose) are highly polar and poor at sorbing nonpolar
compounds; polarity decreases as the organic matter ages and becomes modified by soil
processes and increased sorption accompanies the change (Garbarini and Lion, 1986;
Rutherford et al., 1992). For applications other than first order approximations,
estimates of compound sorption by soil should be refined experimentally (Hassett and
Banwart, 1989).
-------
Vapor Sorption
Vapor sorption by dry and relatively dry soils (<90% relative humidity or less than 5
monolayers of water) has been shown to be greater than sorption by wet soils (Chiou and
Shoup, 1985; Poe et al., 1988; Ong and Lion, 1991a). Therefore, we can assume that
vapor sorption affects VOC retention and transport in any soil that is dry or that drys
periodically. Sampling when soils are dry may be more precise than sampling when soils
are.wet. The use of dry soils for creating soil performance evaluation materials by vapor
fortification has been shown to be more precise than the use of wet soils (Hewitt et al.,
1992).
Sorption of VOCs by dry soils and sediments is a physical adsorption process. The
specific surface area and internal pore structure have been identified as factors that
influence vapor sorption by soil minerals (Ong and Lion, 1991b). Predictions of vapor
sorption will depend on the moisture content, but the effect is not linear (Ong and Lion,
l'991a). Organic matter exhibits maximum sorption in a fully hydrated state, estimated
at approximately 80% relative humidity (Ong and Lion 1991b).
Nonequilibrium or Rate Limited Sorption
Both batch and column methods of studies of numerous soils and compounds have shown
that an rapid initial phase of sorption is followed by kinetically slow sorption processes
(Brusseau and Rao, 1989). Physical limitations of contact between soil and contaminants
generated by variable flow patterns (sometimes described as mobile/immobile regions of
soil water) and variable flow rates (limitations of contact time between soil and
contaminants) contribute to the phenomena of nonequilibrium sorption in field soils and
sediments. Solute transport models that address the kinetics of sorption and desorption
and/or use rate-limited mass transfer parameters to describe physical limitations during
solute transport have been advanced (Harmon et al., 1989; Brusseau and Rao, 1989).
Entrapped VOCs
Entrapped or slowly desorbing VOCs have been reported by several researchers
(Steinberg et al., 1987; Pignatello, 1990a,b; Pavlostathis and Jaglal, 1991; Pavlosthathis
and Mathavan, 1992; Ball and Roberts, 1991a,b). Sawhney et al., 1988 demonstrated
that standard USEPA purge and trap extraction specifications for soil (11 minutes at 40
°C) recovered less than 2% of the fumigant 1,2-dibromoethane (EDB) present in air-dried
field-contaminated soil. Their conclusions were based on a comparison with soil
extracted by a hot solvent technique (that is, in methanol at 75 °C for 24 hours). Release
of the some portion of the entrapped or residual fraction of VOCs can also be
accomplished by pulverizing soil in a ball mill (Steinberg et al., 1987; Ball and Roberts,
-------
1991a; Pignatello, 1991b). The data indicate that after an initial rapid sorption, VOCs
. continue to diffuse into soil organic matter and/or into soil micropores. Months or years
of contact between VOCs and soil may create an increased proportion of VOCs in the
entrapped state. After a contamination event, the concentration of VOCs that are
available for microbial degradation and migration in the macropores diminishes, while
the entrapped VOCs slowly increase in amount and proportion (Steinberg, 1992). This
entrapped VOC pool assumes significance only to the extent that it slowly leaches out to
supply groundwater with concentrations that may be considered significant to human
health (Steinberg et al., 1987).
Conclusions
Much information is needed to characterize the influence of soil, site, and environmental
parameters on VOC sorption. Both hydrologic and sorption factors are likely influence
vadose VOC concentrations. Whereas' partitioning into organic matter has provided a
gross descriptive and predictive tool, measured sorption coefficients on any given soil or
sediment may differ by an order of magnitude from predicted values.
Analytical methods that measure the total amount of soil VOCs should be reviewed and
appropriate procedures implemented. The potential negative bias being generated by
incomplete extraction of entrapped VOCs is estimated to be one to two orders of
magnitude (Travis and Maclnnis, 1992). The unextracted VOCs are not permanently
trapped. Entrapped VOCs diffuse back out of the soil matrix (Steinberg et al., 1987;
Sawhney et al., 1988). The soil, environmental, and compound-specific factors that
control the release of entrapped VOCs need to be elucidated and included in risk
assessments and designs for soil remediation.
Although the research described in this paper has been funded wholly by the U.S. Environmental Protection
Agency through contract 68-CO-0049 to Lockheed Engineering and Sciences Company, it has not been
subjected to Agency review. Therefore, it does not necessarily reflect the views of the Agency.
REFERENCES
Ball, W.P., and P.V. Roberts. 1991a. Long-term sorption of halogenated organic
chemicals by aquifer material. 1. Equilibrium. Environ. Sci. Technol. 25:1223-1236.
Ball, W.P., and P.V. Roberts. 1991b. Long-term sorption of halogenated organic
chemicals by aquifer material. 2. Intraparticle diffusion. Environ. Sci. Technol. 25:1237-
1249.
Brusseau, M.L., and P.S.C. Rao. 1989. Sorption nonideality during organic contaminant
transport in porous media. CRC Crit. Rev. Environ. Control 19:33-99.
-------
Chiou, C.T. 1989. Theoretical considerations of the partition uptake of nonionic organic
compounds by soil organic matter, pp. 1-29. In, Sawhney, B.L., and K. Brown, eds.,
Reactions and Movement of Organic Chemicals in Soils. SSSA Special Publication
Number 22. Soil Science Society of America, Inc. Madison, WI.
Chiou, C.T., and T.D. Shoup. 1985. Soil sorption of organic vapors and effects of
humidity on sorptive mechanism and capacity. Environ. Sci. Technol. 19:1196-1200.
Garbarini, D.R., and L.W. Lion. 1986. Influence of the nature of soil organics on the
sorption of toluene and trichloroethylene. Environ. Sci. Technol. 20:1263-1269.
Harmon, T.C., W.P. Ball, and P.V. Roberts. 1989. Nonequilibrium transport of organic
contaminants in groundwater, pp. 405-437. In, B.L. Sawhney and K. Brown, ed.,
Reactions and Movement of Organic Chemicals in Soils. SSSA Spec. Publ. 22. Soil
Science Society of America, Inc. Madison, WI.
Hassett, J.J. and W.L. Banwart. 1989. The sorption of nonpolar organics by soils and
sediments. In, Sawhney, B.L., and K. Brown, eds., Reactions and Movement of
Organic Chemicals in Soils. SSSA Special Publication Number 22. Soil Science Society
of America, Inc. Madison, WI.
Hewitt, A.D._, P.H. Miyares, D. C. Leggett, and T.F. Jenkins. 1992. Comparison of
analytical methods for determination of volatile organic compounds in soils. Environ.
Sci. Technol. 26:1932-1938.
Karickhoff, S.W. 1984. Organic pollutant sorption in aquatic systems. J. Hydraulic Eng.
110:707-735.
Mingelgrin, U., and Z. Gerstl. 1983. Reevaluation of partitioning as a mechanism of
nonionic chemicals adsorption in soils. J. Environ. Qual. 12:1-11.
Ong, S.K., and L.W. Lion. 1991a. Mechanisms for trichloroethylene vapor sorption onto
soil minerals. J. Environ. Qual. 20:180-188.
Ong, S.K., and L.W. Lion. 1991b. Trichloroethylene vapor sorption onto soil minerals.
Soil Sci. Soc. Am. J. 55:1559-1568.
Pavlostathis, S.G., and K. Jaglal. 1991. Desorptive behavior of trichloroethylene in
contaminated soil. Environ. Sci. Technol. 25:274-279.
Pavlostathis, S.G., and G.N. Mathavan. 1992. Desorption kinetics of selected volatile
organic compounds from field contaminated soils. Environ. Sci. Technol. 26:532-538.
-------
Pignatello, J.J. 1990a. Slowly reversible sorption of aliphatic halocarbons in soils. I.
Formation of residual fractions. Environ. Toxic. Chem. 9:1107-1115.
Pignatello, J.J. 1990b. Slowly reversible sorption of aliphatic halocarbons in soils. II.
Mechanistic aspects. Environ. Toxic. Chem. 9:1117-1126.
Poe, S.H., K.T. Valsaraj, L.J. Thibodeaux, and C. Springer. 1988. Equilibrium vapor
phase adsorption of volatile organic chemicals on dry soils. J. Hazardous Mater. 19:17-
32.
Rutherford, D.W., C.T. Chiou, and D.E. Kile. 1992. Influence of soil organic matter
composition on the partition of organic compounds. Environ; Sci. Technol. 26:336-340.
Sawhney, B.L., J.J. Pignatello, and S.M. Steinberg. 1988. Determination of 1,2-
dibromoethane (EDB) in field soils: Implications for volatile organic compounds. J.
Environ. Qual. 17:149-152.
Siegrist, R.L., and P.D. Jenssen. 1990. Evaluation of sampling method effects on
volatile organic compound measurements in contaminated soils. Environ. Sci. Technol.
24:1387-1392.
Steinberg, S.M. 1992. Persistence of several volatile aromatic and halogenated
hydrocarbons in a low organic carbon calcareous soil. Chemosphere 24:1301-1315.
Steinberg, S.M., J.J. Pignatello, and B.L. Sawhney. 1987. Persistence of 1,2-
dibromoethane in soils: Entrapment in intraparticle micropores. Environ. Sci. Technol.
21:1201-1208.
Travis, C.C. and J.M. Maclnnis. 1992. Vapor extraction of organics from subsurface
soils. Is it effective? Environ. Sci. Technol. 26:1885-1887.
-------
BIOGRAPHY
Martha Minnich has a B.A. in Chemistry from the University of California, Santa Cruz,
a M.S. in Soil Science from the University of California, Davis, and a Ph.D. in Soil
Science from Cornell University. Following her M.S. degree, she worked at the USDA
Beltsville Agricultural Research Center and with an environmental consulting firm
reviewing environmental fate data on pesticides for the FIFRA Registration Standards
Program. Upon completion of her Ph.D., she was an Assistant Professor in the
Department of Soil and Water Science at the University of Arizona. Her program
focused on the fate and movement of organic contaminants in soil. For the last five
years, she has been an environmental consultant. She used the Pesticide Root Zone
Model (PRZM) to estimate the fate and transport of pesticides to help develop a golf-
course management plan for a proposed development in New England. Currently, she
is employed by Lockheed Environmental Systems and Technologies Co. on contract with
work at EMSL-Las Vegas.' Her research pertains to the sampling and analysis of soil
VOCs.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 33
Laboratory Measurements of Non-steady-state Diffusion of O-xylene
and Naphthalene From a Sandy Soil
B. Lindhardt and T. H. Christensen
Department of Environmental Engineering/Groundwater Research
Centre, Technical University of Denmark
January 12-14, 1993
Las Vegas, Nevada
-------
Laboratory measurements of non-steady-state diffusion
of o-xylene and naphthalene from a sandy soil.
by
B. Lindb&rdt and T.H. Cbristenser
Department of Environmental Engineering/Ground water Research Centre, Building 115,
Technical University of Denmark, DK 2800 Lyngby, Denmark.
INTRODUCTION
Volatilization of VOCs from polluted soil may have a potential influence on outdoor
and indoor air quality, in particular because some of these organic compounds are
potential carcinogens. Unfortunately, very little is known as to the fluxes of organic
compounds from polluted soil. This paper presents laboratory measurements of non-
steady-state fluxes from soil surface. The soils were artificially contaminated with o-
xylene, as a volatile organic compound, or naphthalene as a semi-volatile compound.
The effect of the content of organic matter and water in the soils and the concentration
of the compound were investigated. The measured fluxes were compared to predictions
by a simple equilibrium phase-partitioning model.
MATERIALS AND METHODS
The volatilization from the soil was measured in a diffusion cell, a modified version of
the cell used by Farmer etal. (1980). The experimental set up is showed in Fig. 1. The
diffusion cell is made of stainless sieel, the soil surface is 9x9 cm and the depth of ihe
soil is 5 cm. The air void above the soil is 2 mm high and 10 cm wide and to allows
air spread out before reaching the soil surface. The flow rate through the air void is 0.9
1 per minute, corresponding, to a-wind-speed across the surface of 7.5 cm per second
Pump
Air inle
m
Carbon-filter
¦i Tenax
m
5 cm
Waler
Soil
9 cm
Figar 1: Diagram of the ex peri mental set up ror del er mining ihe nor - si e ad y-stale
volatilization from soil surface. -
-------
and a change of the air volume over the soil surface of 50 times per-minute. The inlet
air was cleaned by a carbon filter and humidified before entering' the cell. The
experiment took place in a climate box at constant temperature at 10.0°C ± 0.2°C.
The volatilization was measured for two sandy soils from an agricultural field,
with a content of organic matter of 1.06% (w/w) (Soil A) and 0.094% (w/w) (Soil B),
respectively. Water was added to obtania water content of either 5% (w/w) or 15%.
(w/w). The o-xylene was added to the soil 3 days before it was packed into the
diffusion cell and naphthalene 6 days before. To prevent biodegradation of the
compounds, 1 g HgCl2 was added per kg dry soil. After closing the packed cell, 24
hours elapsed before the experiment was started by passing air through the top camber
of the cell. The flux was quantified by trapping the organic compounds in the outlet air
on Tenax-Tx tubes 10 times during the experimental period at 24 hours. The tenax
samples were extracted with pentane. The concentrations of the organic compounds in
the soil were measured before and after the experimental period, by extraction with
dichloromcthan/methanol. Both extracts were analyzed by GC/FID.
PREDICTION OF DIFFUSIVE FLUXES OF VOLATILE COMPOUNDS FROM
A SOIL SURFACE
The model used to predicted the flux from the soil surface is based on expressions
developed by Mayer et al. (1974) and Jury et al. (1983). In this paper the following
assumptions have been made:
Diffusion in the air phase is the only transport process, assuming no water
or soil air pressure gradients are present and vertical transport by diffusion
in the water phase is negligible.
VOC is only remove from the soil surface.
The concentration of the VOC in the air void above the soil is approximately
zero.
Local equilibrium is assumed according to the partitioning between air
phase, water phase and soil organic carbon.
No degradation of the compounds takes place.
Describing the diffusion process according to Fick's Second Law, the one-dimensional
flux of a compound out of the contaminated soil can be approximated by:
Ct,o Ds
(1)
or in logarithmic form
log J = log CT o + log Ds + Vi log t (2)
where:
J The vapour flux [>tg/(cnr ¦ h)],
CT0 Concentration of compound per volume of soil [mg/cm3 soil] at time
zero (Cx o = pb -CI71 0, where pb is the bulk density [g/cm3] and Cm 0 is
the concentration on a weight basis [mg/kg])
Ds Specific diffusion coefficient
t Time
-------
Cm>0 must be determined analytically, while Ds must be estimated for each specific
situation with respect to compound, soil bulk density, soil organic carbon fraction,
water content and air content according to
where:
kh
K
h
foe
Pb-
Ds =
D0 • ^3'33
Kt
* +ea)" '[fob ' fcc " Koc +*w )+ KH "
Diffusion coefficient in air [cm:/h]
Henry's constant [-]'
.Organic carbon participation coefficient [-]
Volumetric water content f-]
Volumetric air content [-]
Fraction of organic carbon in the soil [gC/g soil]
Bulk density [g/cin3]
(3)
D0, Kh and Koc are compound specific parameters while $a, ffw, foc, and are soil
specific parameters.
RESULTS AND DISCUSSION
Both soils are characterized as sandy soils. The volumetric contents of soil-air, were
between 0.29 and 0.36 in the experiments with low content of water and between 0.06
and 0.14 in the experiments with high content of water, see Tabel 1. The concentration
of o-xylene in the soil at the start of the experiments, Cm0f varied between 4.7 pg/g
and 249 ng/g, and the concentration of naphthalene between 6.3 ^%fg and 39.9 /ig/g,
see Table 1. The concentration were different for each combination of organic matter
and water. Loss of compounds took place during transfer of the soil from the mixing
chamber to the diffusion cell, causing the in differences in concentrations level. The
relative vapour pressure in the soil-air varied between 0.02 and 0.30 in experiments
with o-xylene and between 0.05 and 1 in the experiments with naphthalene
Examples of the diffusive fluxes determined with the diffusion cell are shown in
Fig 2. The fluxes are plotted' versus time and in terms of the logarithm of the flux
versus the logarithm of the time. The log-log plots correspond to the linearized
expression of the used model. Measurements are given as solids symbols (¦) and the
corresponding regression lirtes are given as full lines. The estimated fluxes according
to the model are expressed as dot-and-dash lines.
The decrease in the logarithm of tile flux could be described as a straight line
versus the logarithm of time, with exception of two experiments where the fluxes in the
end of the experimental periods decreased faster. The slopes of the flux in the log-log
plot are between -0.27 and -1.11 in the experiments with o-xylene and between -0.39
and -0.48 in the experiments with naphthalene, see Table 2. From the theoretical model
it was expected that the siope sfioufd be -0.50. Particular the fluxes of o-xylene from
the soil with low content of organic matter and water, "XOOx", decreased faster than
the model predicted. An explanations for the deviation between the measured fluxes and
the predicted flux for this group of experiments is not available at this point. In the
other experiments the deviation between the measured decrease in the flux and the
predicted decrease attributed to experimental errors, There are too few data to decided
-------
Tabel 1: The content of organic matter (foc), water (w), volumetric content of air (6J,
the start concentration (Cm 0), the recovery of mass (Recov), and the relative vapour
pressur of the VOC in the soil air at the start (P/Pq).
Exper.
foc
w
Cm,o
Recov.
P/P0
% (w/w)
% (w/w)
mg/g
%
o-Xylene
"X000"
0.094
5.0
0.31
4.7
73
0.05
"X000"b
0.094
4.6
0.31
5.0
64
0.05
"X002"
0.094
5.0
0.31
22.3
58
0.22
"X02Q"
0.094
15.4
0.13
12.6
76
0.11
" X022"
0.094
15.0
0.14
34.0
71
0.30
"X200"
i.06
5.9
0.37
24.8
82
0.02
"X202"
! .06
7.4
0.35
249.0
72
0.23
"X220"
1.06
17.1
0.06
24.2
98
0.02
"X220"b
1.06
16.0
0.11
23.6
92
0.02
"X222"
1.06
18.9
0.07
280.6
104
0.26
Naphtha! en
'Z
o
o
o
0.094
4.7
0.32
6.3
104
0.27
"Noor
0.094
5.0
0.31
15.4
90
0.67
" N002"
0.094
4.3
0.32
24.6
90
1.07
"N020"
0.094
12.4
0.14
5.6
91
0.23
"N022"
0.094
14.5
0.12
26.0
90
1.05
"N200"
1.06
5.8
0.36
10.9
107
0.04
"N202"
1.06
5.7
0.36
37.9
95
0.15
"N220"
1.06
14.0
0.13
11.4
105
0.05
"N222"
1.06
14.7
0.11
39.9
103
0.16
if there is any structure in the deviation in the slope of the fluxes depending on the
combination of compound, content of organic matter and water.
Since the mean slopes of the regression lines of the observed fluxes (log) versus
time (log) was -0.43 in experiments with o-xylene, with the exception of the
experiments "XOOx", and -0.44 for the experiment with naphthalene, the following
comparison of measured and predicted fluxes is based on the assumption that the model
for the instationary flux is valid, which means that the change in flux over time can be
described in the linearized log-log plot by a slope of -0.50. The y-intercept of the
linearized plot assuming a slope of -0.50 allows for estimation of the observed specific
diffusion coefficient Ds when the concentration in the soil is known (CT0). The
observed specific diffusion coefficients are in Table 2 compared to the predicted specific
diffusions coefficient (D*) in terms of their ratio. Good agreement was found between
the ratios D*/Ds ^ in the experiments with naphthalene, the mean of the ratios was
1.15. In the experiments with o-xylene, the ratios of D*/Ds ^ varied between 0.93 and
3.2, with the exception of the experiments "XOOx". The data indicate that the deviations
-------
J [Hg/(cm" hour)]
8
7
6
5
4
10 . 20
Time [Hour]
J [|Igj'(cm hour)]
0.7 T
10 20
Time [Hour]
o-Xylene
J [ng/(cm hour)]
10
0,1
30 0,1 1 10
Time [Hour]
Naphthalene
J [;j.g/(cm2 hour)]
100
0,1
0,01
30 0/1
1 10
Time [Hour]
100
Figur 2:: Exampels of flux of o-xylene and naphthalene from soil with a content of
organic matter at 0,094% and a water content at about 15% (w/w).
form the predicted level of the flux depend on the combination of content of organic
matter and water in the soil. The results with o-xylene suggest that the model
overestimate the fluxes in soil with low content of organic matter and water, Only a few
studies have tried experimentally to prove the validity of the simple diffusion model.
Jury et al.(1983) reviewed a number of investigation and found a good agreement
bev-.veev, the predicted and the measured volatilization loss. The review dealt with
investigations of organic compounds with low values of Henry's constant, as low as
naphthalene or lower.
-------
Tabel 2: The slop, a, of the fluxes in a log-log plot with respect to time. And the
ration of perdicted and measured specific diffusions coefficents assuming the
established model for the instationay flux is valid.
o-Xylene
Naphthalene
H It
a
"XxxO" "Xxx2" "Xxx"b Avg
"a"
"NxxO" "Nxxl" "Nxx2" Avg
"XOOx" -0,82 -1.11 -0.65 -0.86
"X02x" -0.41 -0.42 -0.42
"X20x" -0.61 -0.56 -0.58
"X22x" -0.39 -0.37 -0.27 -0.34
"NOOx" -0.44 -0.45 -0.45 -0.45
"N02x" -0.39 -0.42 -0.40
"N20x" -0.47 -0.48 -0.48
"N22x" -0.40 -0.48 -0.44
D.VD?
"XxxO" "Xxx2" "Xxx"b Avg
d;/d*
"NxxO" "Nxxl" "Nxx2" Avg
"XOOx" 4.89 5.20 6.97 5.68
"X02x" 2.55 . 3.20 2.89
"X20x" 0.93 0.94 0.94
"X22x" 1.80 2.12 2.13 1.96
"NOOx" 0.89 0.87 0.99 0.92
"N02x" 1.28 1.16' 1.22
"N20x" 1.14 1.02 1.08
"N22x" 1.49 1.52 1.51
SUMMERY
This laboratory study of non-steady-state volatilization of o-xylene and naphthalene from
artificially contaminated sandy soil indicates that a simple diffusion model assuming
local equilibrium partitioning into soil air, soil water and soil organic carbon provides
a fair description of the flux of naphthalene, but overestimates the fluxes of o-xylene
up to a factor-3 depending on the content of organic matter and water in the soil. For
some combinations of organic matter and water content the model gave an inadequate
description of the trend of the fluxes. The study indicates that simple diffusion model
prediction may be a useful tool in preliminary assessment of the volatilization of VOCs
from the unsaturated soil, but futher studies are necessary to clarify the effect of the
content of organic matter and water on the flux of the most volatile compound.
REFERENCES
Farmer, W.J., Yang, M.S., Letey, J., and Spencer, W.F., (1980): Hexachlorobenzene: Its
Vapor Pressure and Vapor Phase Diffusion in Soil. Soil Sci. See. Anier. J.. 44, 676-680.
Jury, W.A., Spencer, W.F., and Farmer, W.J., (1983): Behavior Assessment Model for Trace
Organics in Soil: I. Model Description. J. Environ. Oual.. 12. (4), 558-564.
Jury, W.A., Spencer, W.F., and Farmer, W.J., (1984): Behavior Assessment Model for Trace
Organics in Soil: IV. Review of Experimental Evidence. J. Environ. Oual.. 13. (4),
580-586.
Mayer, R., Letey, J., and Farmer, W.J., (1974): Models for Predicting Volatilization of
Soil-Incorporated Pesticides. Soil Sci. See. Amer. Proc.. 38, 563-568.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 34
Performance of a New Soil Sampling Tool for Use with Methanol
Preservation of Samples Containing Volatile Organic Compounds
David E. Turriff, Ph.D.
En Chem, Inc.
Green Bay, Wisconsin
January 12-14, 1993
Las Vegas, Nevada
-------
Performance of a New Soil Sampling Tool for Use with
Methanol Preservation of Samples Containing Volatile
Organic Compounds
Author: David E. Turriff, Ph.D.
En Chem, Inc., Green Bay, WI
Wisconsin became the first State to implement a field methanol preservation method for
gasoline range organic analysis. The method calls for quickly collecting a 25 gm soil
sample. The sample is transferred to a 2 oz glass jar to which methanol is immediately
added.
To assist in the standardization of field sampling, En Chem, Inc. developed a stainless
steel sampling tool for collecting a volumetric plug of soil. The volume is calculated to
deliver an average of 25 gm of soil sample. The soil can be extruded by a pushrod
attached to a backplate which pushes the soil from the sampler. The sample can also be
stored in the sampler by attaching a stainless steel cap containing a Viton O-ring. The
cap encloses the soil in an airtight, headspace-free condition. Sealed samplers show
recoveries of BETX and Gasoline Range Organics(GRO) which are equivalent to
immediate preservation with methanol. Since the State of Wisconsin allows soils to be
stored without methanol for up to two hours on ice in a 2 oz glass container, we
compared BETX and GRO values of soils stored in the sampler vs 2 hour storage in 2
oz bottles. Ottawa sand prepared to 10% moisture was spiked with a gasoline
component spike. Figure 1 shows the recoveries under both conditions.
Figure 1.
Evaluation of Soil Sampler as a Storage Container for Gasoline Analysis
Method | GRO | Benzene | Ethyl Benz.
Toluene | Xylenes
Bottle Storage 61.5 mg/kg 1551 ug/kg 11,501 ug/kg
| 5,526 ug/kg 17,742 ug/kg
S. D. Bottle 4.4 19 93
146 24
Soil Sampler 65.9 mg/kg 1690 ug/kg 11,556 ug/kg
| 6,195 ug/kg 17,791 ug/kg
S.D. Sampler 4.9 26 111
175 36
n = 10, p<2.5% S.D. =Standard Deviation
The results show that the soil sampler is better at retaining GRO and BETX
concentrations especially of the lighter component compounds. Examination of the
GRO chromatogram confirms that lighter components show a selectively higher
-------
recovery. This is presumably due to the lower volatilization potential of the soil stored
in the soil sampler.
In order to evaluate the effectiveness of the sampler in field situations and, further, to
evaluate the potential of the soil sampler to store the soils for 6-24 hours. This would
allow the sample to be shipped to the laboratory in the sampler and would eliminate the
need for field preservation. This process would help to eliminate some of the field
inconsistencies in sampling that are seen with field methanol preservation.
Figure 2 shows the comparison of storing a variety of gasoline-contaminated soils in
the soil sampler for 6 hours and 24 hours versus storing the soils for 2 hours in a 2 oz
bottle on ice as is currently accepted in Wisconsin. Samples were immediately
preserved with methanol after the indicated storage times and thereafter each sample
was prepared identically for analysis.
Figure 2
Field Evaluation of the Soil Sampler versus the Standard
State of Wisconsin Method for Soil Sampling
Percent Comparison between Bottle Storage and
Soil Sampler Storage
Method
GRO
Benzene
Toluene
6 hour storage
128 %
176%
115%
n =
28
18
27
24 hour storage
157%
250%
161%
n =
21
11
21
As can be seen from the data, the soil sampler shows enhanced recoveries over the
Wisconsin approved method of storing the sample in a 2 oz bottle on ice. Statistical
analysis, however, shows equivalence between the two methods at the 95 % confidence
level for both 6 hours and 24 hours. More samples are required to determine whether
the soil sampler can statistically outperform the State method.
Based upon the data and upon experience of sampling in the field, we believe that this
method of sample collection is superior to other methods currently available for the
following reasons:
1. It is an extremely fast method for sampling soils and involves
minimal handling of sample during the sampling process.
2. It delivers a reasonably consistent sample weight based
upon a volumetric measure.
-------
3. It can be used as a storage container for collection and
transport to a clean environment or to the laboratory for
final sample preparation. Field methanol preservation can
be eliminated.
Field trials are currently being extended to determine more clearly the sources of
variance in the sampling process.
-------
References
1. Soil Sampling for Volatile Organic Analysis, Tim Lewis, EPA Superfund
Groundwater Issue, August, 1990.
2. Evaluation of Sampling Method Effects of Volatile Organic Compound
Measurements in Contaminated Soils, Robert Siegrist and Petter Jenssen,
Environmental Science and Technology, 24, 1990.
-------
Biography
Name: David Turriff, Ph.D.
Position: Director of Laboratory and Owner
En Chem, Inc.
An environmental laboratory specializing in UST analysis
Experience: Over 10 years experience in managing environmental laboratories.
Assistant Professor of Biochemistry at The Unversity of Illinois
from 1978-1984.
Research Associate at The University of Chicago from 1974-1978.
Ph.D. from The University of Rochester, Rochester, N.Y.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 35
Preanalytical Holding Times: Advanced Data Treatment and Analysis
Charles K. Bayne, Denise D. Schmoyer, and Roger A. Jenkins
Oak Ridge National Laboratory, Tennessee
January 12-14, 1993
Las Vegas, Nevada
-------
PREANALYTICAL HOLDING TIMES:
ADVANCED DATA TREATMENT AND ANALYSIS
Charles IC Bayne, Denise D. Schmoyer, and Roger A. Jenkins
Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6370
Practical Reporting Time Analyte concentrations are not always stable in
environmental samples. Analytical chemistry laboratories analyze environmental
samples as quickly as possible to insure accurate measurements of the concentration
at the time the analyte was sampled. With increasing number of environmental
samples, the time between collecting an environmental sample and its chemical
analysis (holding time) may be too long; during this time the analyte may biodegrade
{1,2,3) or may decompose. Regulatory agencies have specified holding times for
classes of compounds (volatile organics, semi-volatile organics, pesticides, explosives)
to standardize analytical laboratory procedures. For example, 40 CFR 136 (4)
requires that volatile organic compounds (VOCs) samples stored at 4°C must be
analyzed within 7 days of collection. This requirement is very stringent for most
analytical laboratories. Currently, the holding time has been extended to 14 days for
acid preserved samples (5). The U.S. Army Toxic and Hazardous Materials Agency
(USATHAMA, 6) recommends a 28 day holding time for VOCs preserved with
sodium bisulfate, and a 56 day (until extraction) holding time for explosives stored
at 4°C. These results are based on the Oak Ridge National Laboratory (ORNL)
holding time study (7,8,9,10). The purpose of this study is two fold. Firstly, a
statistical definition for a holding time is developed, and secondly, a method is
developed to predict the risks for analyzing an analyte beyond the holding time.
The basic concept of holding times is to specify how long a sample can be held
with reasonable assurance that the initial concentration has not changed significantly.
The definitions of "reasonable assurance" and "changed significantly" are key to
holding time determinations. This paper proposes a "Practical Reporting Time"
(PRT) based on statistical definitions of these terms.
A significant change in initial concentration is defined using statistical properties
of the measurement system. A critical concentration (CC) is determined on the first
day of the holding time study; it is the concentration below which there is only an a%
chance due to measurement error that a measured concentration would be observed.
A significant change has occurred when the concentration falls below this critical
concentration.
-------
Figure 1 illustrates.an analyte
concentration that is linearly
decreasing with time; and
measurement variation follows a
normal distribution. As the
concentration decreases, the
chance that an individual sample
will be below, the .critical
concentration increases. The
probability (y) that a sample
concentration is below the
critical concentration is used as a
measure of risk. The PRT is
defined as the day when there is
a risk of y% = 15% that the
measured aiialyte will be below
CC for an a% = 5%.
Fig. 1 Practical reporting time (PRT) for an analyte with
a linear decreasing concentration.
The PRT depends on the model used to approximate the degradation of an
analyte concentration and the precision (standard deviation) of the analytical
measurement variation. For a given anajyte, large measurement variation will give
longer PRT than those for smaller measurement variation. This result occurs because
it is more difficult to detect changes in the initial concentration with larger
measurement variation. The chance that a measured analyte is below CC will
increase the longer an environmental sample is held past the PRT. The rate of this
probability increase reflects the risk of missing the PRT.
ORNL Holding Time Study Practical Reporting Times have been calculated for
nineteen volatile organic compounds (VOCs) and four explosive for the holding time
studies conducted at ORNL (7,8,11) from August 1986 to September 1988. The
ORNL holding time study represents 476 experimental cases with 33,422
concentration measurements over long time periods under a variety of matrices and
storage temperatures.
Practical Reporting Times were calculated for the ORNL holding time study by
the following procedure: (a) an approximating model representing "concentration
versus time" is fitted to the data by the method of least squares (12); (b) the one-
sided 95% prediction limit for time zero is located for the approximating model (this
limit is the critical concentration); (c) a horizontal line is drawn from the critical
-------
concentration until it intersects with the one-sided 85% prediction limit; and (d) A
vertical line is drawn from the intersection described in (c) to the time-axis. This
intersection with the time-axis is the PRT.
Sigma-to-Slope Ratios Zero-order or first-order kinetic models [C = C0 + B
day, or ln(C) = Q, + B day], successfully approximated concentration data for 73%
of the ORNL holding time experimental cases. The holding time results for these
models can be related to the new PRT definition through the ratio of the single
measurement precision (standard deviation) to the absolute value of the slope of the
line that approximates the concentration change (i.e., sigma-to-slope ratios).
A quadratic polynomial fitted by the method of least-squares to PRTs versus
sigma-to-slope ratios represent 99.95% of the variation in the PRT values. Predicted
PRT values can be calculated by the quadratic approximation
For example, the ORNL holding time experiment estimated that (Sq/B) = -8 for
chloroethane at an initial concentration of 50 ng/L in Tennessee soil stored at 4°C.
First convert this number to sigma-to-slope ratio by taking the absolute value,
(So/1B |) = +8. The corresponding PRT value can be approximated by the quadratic
approximation as (rounded down to the next whole number of days)
Estimating PRT values based on the sigma-to-slope ratios is very useful for
determing holding times for compounds or storage conditions not included in the
ORNL study. If chemists are willing to assume a zero-order or first-order kinetics
model, a short term experiment could be conducted to estimate the slope of the
model and the standard deviation of a single analytical measurement. The best
procedure is to run replicate concentration measurements at the beginning and end
of their experimental time. Chemists may wish to run additional measurements on
5
PRT = -0.3051 + 0.6894 —2-
\\B\,
- 0.0134 xlO"2 3-
A*\.
PRT = -0.3051 + 0.6894(8) - 0.0134 xl0*2(8)2
PRT =52-5 days
-------
the third and seventh days to detect rapid degradation, and additional measurements
in the middle of their experimental time to detect lack-of-fit. From this experiment,
they can estimate the slope and the measurement standard deviation. Using the
sigma-to-slope ratio, a PRT value can be estimated for the compound using the
quadratic approximation.
Days Past PRT What happens to the risk (y) probability if samples are held past
the PRT? As the samples are held past the PRT, the risk probability will increase
that an analyte concentration measurement is less than the critical concentration.
The decision maker must decide if the increased risk is unacceptable. The rate of
increase of y will depend on the sigma-to-slope ratio. Figure 2 is a nomograph for
increasing y probabilities for days past the PRT value. This nomograph is based on
the sigma-to-slope ratios estimated for the ORNL holding time study. For example,
the PRT value is 5 days for chloroethane with an initial concentration of 50 /ig/L in
Tennessee soil stored at 4"C with a sigma-to-slope value of (S,/1B | ) = +8. Figure
2'shows that" at 1 days past the PRT value the y probability is a little less than 0.20.
This increase may be considered acceptable. But, 4 days past the PRT value, the y
probability increases to 0.30 and this increase may be considered unacceptable. This
probability means the analyte concentration has degraded so that there is a three in
ten chance an analysis of an environmental sample gives a concentration below the
critical concentration.
The nomograph illustrates the risk probabilities of concentration changes (below
the CC level) for individual analyte measurements rather than their actual
concentration changes. In addition to measurement precision and slope, the actual
concentration change would require the value for the initial concentration which may
vary for different holding time studies due to different spiking levels, environmental
matrices, storage conditions and measurement errors.
Conclusions Practical Reporting Time is a statistically defined holding time to
specify how long a sample can be held with reasonable assurance that the initial
concentration has not changed significantly. PRT values depend only on the
degradation kinetics and the analytical measurement variation. A quadratic
polynomial has been estimated to predict PRT values for a large class of analytes,
matrices, and storage temperatures with zero-order and first-order approximating
kinetic models. Future holding time studies with limited resources can calculate PRT
values using the sigma-to-slope ratio. The PRT method can indicate the risk of
making analytical measurements past the PRT. Finally, a nomograph is presented
for decision makers to evaluate the risk of analyzing a sample past the PRT value.
-------
Fig. 2. Holding time nomograph for days past the PRT. Contours are probabilities of an analyte
measurement being less than the critical concentration.
Acknowledgements Work sponsored by the U.S. Army Toxic and Hazardous
Materials Agency (USATHAMA) Project "Environmental Measurements Technology"
DOE Interagency Agreement No. 1769-A073-A1
-------
REFERENCES
1. Moore, A.T.; Vira, A.; Fogel.S. Environ. Set TechnoL 1989, 23, 403-406.
2. Walton, B.T.; Anderson, TA. Chemosphere 1988,17, 1501-1507.
3. Walton, B.T.; Anderson, T.A.; Hendricks, M.S.; Talmage, S.S. Environ. ToxicoL
Chem., 1989, 8, 53-63.
4. Federal Register 1979, 40 CFR Part 136, Proposed rules. 44, No. 233:69534. Dec.
3.
5. Bottrell, D.W.; Fisk, J.F.; Hiatt, M. Environ Lab 1990, 2, 29-31.
6. USATHAMA, U.S. Army Toxic and Hazardous Materials Agency Quality
Assurance Program, 1990, U.S. Army Toxic and Hazardous Materials Agency,
Aberdeen Proving Ground, MD 21010-5401.
7. Maskarinec, M.P.; Bayne, C.K.; Johnson, L.H.; Holladay, S.K.; Jenkins, R.A.
Stability of Volatile Organics in Environmental Water Samples: Storage and
Preservation, ORNL/TM-11300, August 1989, Oak Ridge National Laboratory,
Oak Ridge, TN 37831.
8. Maskarinec, M.P.; Bayne, C.K.; Johnson, L.H.; Holladay, S.K.; Jenkins, R.A.;
Tomkins, B.A. Stability of Explosives in Environmental Water and Soil Samples,
ORNL/TM-11770, January 1991, Oak Ridge National Laboratory, Oak Ridge,
TN 37831.
9. Maskarinec, M.P.; Johnson, L.H.; Holladay, S.K.; Moody, R.L.; Bayne, C.K.;
Jenkins, R.A. Environ. ScL TechnoL, 1990, 24, 1665-1670.
10. Maskarinec, M.P.; Johnson, L.H.; Bayne, C.K. J. Assoc. Off. Anal. Chem. 1989,
72, 823-827.
11. Maskarinec, M.P.; Bayne, C.K.; Jenkins, RA; Johnson, L.H.; Holladay, S.K.
Stability of Volatile Organics in Environmental Soil Samples, ORNL/TM-12128,
November 1992, Oak Ridge National Laboratory, Oak Ridge, TN 37831.
12. Draper, N.; Smith, H. Applied Regression Analysis, Second Edition, John Wiley
& Sons, New York, 1981.
13. ASTM, 1986 Annual Book of ASTM Standards, Vol. 11.02 Water (II), pp 21-27,
ASTM, Philadelphia, Pa., 1986.
-------
H.S. Prentice and D.F. Bender, Project Summary: Development of Preservation
Techniques and Establishment of Maximum Holding Times: Inorganic
Constituents of the National Pollutant Discharge Elimination System and Safe
Drinking Water Act, Research and Development, EPA/600/S4-86/043, March
1987.
-------
BIOGRAPHY
Charles K. Bayne, Ph.D. Statistics; Group leader of the Statistical Methodology
Group in the Computing Applications Division; 18 years experience consulting on
chemometrics and environmental analytical chemistry problems.
Denise D. Schmoyer, M.S. Statistics; Statistical consultant for the Statistical
Methodology Group in Computing Applications Division; 12 years experience
consulting on statistical process improvement and environmental modelling.
Roger A. Jenkins, Ph.D. Chemistry; Group leader of the Special Projects Group in
the Analytical Chemistry Division; 17 years experience on tobacco smoking chemistry
and environmental analytical measurement procedures.
-------
Reprinted from ENVIRONMENTAL SCIENCE & TECHNOLOGY. Vol. 24, 1990
Copyright © 1990 by the American Chemical Society and reprinted by permission of the copyright owner.
Stability of Volatile Organic Compounds in Environmental Water Samples
during Transport and Storage
Michael P. Maskarlnec,' Lynne H. Johnson, Susan K. Holladay, Ronnie L. Moody, Charles K. Bayne, and
Roger A. Jenkins
Analytical Chemistry Division, Oak Ridge National Laboratory, P.O. Box 2008, Oak Ridge, Tennessee 37831-6120
B The stability of volatile organic compounds in envi-
ronmental water samples has been studied, particularly
with respect to the establishment of preanalytical holding
times. Methods have been developed for the preparation
of standard samples containing known concentrations of
volatile organics. Three water samples were used: distilled
water, surface water, and groundwater. Samples were
stored at both room temperature and under refrigeration.
Data were collected over a 365-day period by gas chro-
matography/mass spectrometry. In water samples con-
taining low chloride content (distilled water), rapid deh-
ydrohaJogenation of tetrachloroethane to trichloroethylene
occurred. Such degradation was also evident in the surface
water and groundwater samples stored at room tempera-
ture. A less rapid conversion of trichloroethane to di-
chloroethylene occurred in distilled water samples stored
at 25 °C. Reduced concentrations of aromatic volatiles
were observed in both surface and groundwater matrices
after 28 days. Loss of carbon tetrachloride was also ap-
parent in surface water samples stored at room tempera-
ture. Subsequently, experiments were conducted to de-
termine the value of reduced pH in sample preservation.
It was shown that acidification with hydrochloric acid
effectively prevented degradation and allowed indefinite
storage. However, sampling and analytical considerations
make the use of HC1 impractical. Therefore, a study was
carried out using sodium bisulfate and ascorbic acid as
preservatives. Both substances effectively preserved the
samples, but sodium bisulfate proved to have several ad-
vantages over ascorbic acid. Samples preserved with either
acid were stable over the 112-day experimental period.
The implication is that with preservation the maximum
holding times for such samples will be limited only by the
need for sample turnaround.
Introduction
During the past two decades, there has been a dramatic
expansion of environmental legislation, including the
Comprehensive Environmental Response, Compensation,
and Liability Act; the Resource Conservation and Recovery
Act; the Toxic Substance Control Act; the Clean Water
Act; the Safe Drinking Water Act; the Marine Act; and,
most recently, the Superfund Amendment and Reau-
thorization Act. One result of these regulatory measures
has been a tremendous increase in the number of samples
collected and distributed for analysis. One estimate is that
federal, state, and local governments combined with pri-
vate industry accounted for 500000-700000 samples in
1986. Furthermore, this number is growing at a rate of
25-40% per year (1). Obviously, this has put tremendous
strain on the capacity of analytical laboratories. In many
cases, samples are collected at a particular site, shipped
to a central distribution point, and assigned to individual
laboratories on the basis of capacity. All of this is done
with relatively little knowledge of the stability of the
samples. For each analytical method, maximum preana-
lytical holding times have been established in a rather
arbitrary fashion. With a few exceptions (2), limited
systematic information on the long-term stability of vol-
atiles in water exists. This work focuses on the develop-
ment of a data base that allows documentation of the
stability of volatile organics in water, for purposes of in-
creasing the preanalytical holding times and therefore
reducing the cost associated with the analysis.
The generation of a data base establishing preanalytical
holding times presents formidable experimental difficul-
ties, including the need for a large number of identical
sample aliquots, the need for a variety of sample matrices,
and the desire for a large number of potential analytes to
be present. The high vapor pressure of these analytes
requires that precautions be taken to minimize losses
during sample aliquot preparation. In addition, since most
environmental samples contain only a few of the potential
analytes, a laboratory method for the preparation of sam-
ples containing all target compounds must be developed.
Fortunately, an analytical method (GC/MS) exists that
is capable of determining all of these analytes in a single
run. However, there are analytical problems related to the
long-term drift of the instrument, the stability of standard
compounds, and the use of a method that was originally
designed for screening purposes, not for highly accurate
quantitative determinations. In this work these limitations
have been largely overcome, and the data base reported
here can be used to "make an accurate assessment of the
stability of volatile organic compounds in environmental
water samples.
Experimental Section
This study was designed to take into account as many
variables as possible within the limitations of budget and
sample capacity. For the initial investigation, two con-
centration levels were used: 50 and 500 ng/L. For the
preservation studies, a single spike level, 100 Mg/L, was
employed. The exception to this was the spike level for
the ketones and carbon disulfide, which was 250 ^g/L.
Higher-levels were not considered since it was expected
that stability would improve with increasing concentration.
Three matrices were chosen in order to assess the effect
of varying water quality parameters on stability. The
storage conditions were chosen based on the possibility
that samples might not be continuously chilled during
collection and storage. Time intervals were chosen on the
basis of a logarithmic increase, but were also designed to
bracket the existing holding time of 10 days (3).
The sample storage vials used were 40-mL borosilicate
glass vials with Teflon-faced silicone septa and screw caps
with holes. These vials were received fully assembled and
precleaned according to EPA 40 CFR 136 and EPA 40
CFR 141 regulations. Three water sample matrices were
used for this study. The waters were reagent grade water
(water 1), groundwater (water 2), and surface water (water
3). The groundwater and surface water sample matrices
were obtained on site at Oak Ridge National Laboratory
(ORNL). The methanol used was distilled-in-glass grade.
Water quality parameters for the water matrices used in
this study are reported in Table I. All target compound
used either were obtained from the United States Envi-
ronmental Protection Agency (U.S. EPA) Quality Assur-
ance Materials Bank (ResearchTriangle Park, NC) (
-------
Table I. Water Quality Part-taetere for Groundwater and.
Surface Water* IW bPreac&blical Holding Time S
distilled
ground-
surface
chu arteritis
procedure0
alkalinity, mg of
<1
173.4
135-6
EPA 310-1 (12)
CaCCyt
BOD* mg/L
<1
<5
<5
SMEWW 507 (13)
COD, mg/Lc
<1
2.00
3.00
EPA
rhlcriii^ ^r/L
] =¦ In \ + j8i
where In is the natural logarithm (i.e., base e) and (3, the
slope, is now the change in the logarithm of the expected
concentration per day. The two unknown parameters A
and f? are estimated from the holding time data by the
method of least squares (5).
For those situations for which the data could not he
reasonably fitted to -either model, a third model, a cubic
spline, was used. Mathematically, the cubic spline has
been described in detail elsewhere (6) and approximates
the concentrations by a function of time, /(t) where ( is
time, in days as follows:
At) = C0 if t f,
where t0 and f! are an initial and final time at which the
concentrations are equal to C0 and C, The continuity
condition and initial and final concentration condition
places two restrictions on f{t): (1) f{t0) = C0 and /(ij) =
Cj. (2) /'(t0) = 0 and f'itj = 0, where f is the derivative
with respect to t0 and tx, respectively. By use of these two
restrictions for the cubicspline, the coefficients o, 6, c. and
e can be determined in terms of t0 and t,.
Each of two distinct definitions was employed to es-
tablish the criteria for an important change in concen-
tration. The American Society of Testing and Materials
(ASTM) definition (7) has been described. In this case,
the MHT was taken as the time at which the approxi-
mating model is equal to the value of the lower 997c
confidence limit on the intercept if the estimated slope is
negative. For positive estimated slopes, the MHT is the
time at which the approximating model is eijual to the
value of the upper 99% confidence limit on the intercept.
A second definition for MHT has been used in holding
time studies on inorganic analytes conducted by Envi-
ronmental Science and Engineering, Inc. (ESE) in coop-
eration with EPA's Environmental Monitoring and Sup-
port Laboratory (8). The ESE definition is based on in-
tersecting a 10% change in the intercept with a one-sided
90% confidence interval on the predicted concentration.
The application of these definitions to the models, as well
as use of the models to calculate the MHTs, has been
described in detail elsewhere (6)
Results end Discussion
Of prior ary importance to trie rn.d jc. igf this study-a is
the ability u> generate large numbers of identica! alkjuots
of the sample. It was expected that mixing and aliquoting
of the samples would take appreciable time and could
1668 Environ Sci. Technol., Vol 24, No. 11, 1990
-------
Table II. Stability of Volatile Chlorinated Hydrocarbons in
Water Stored in Tedlar Bags at 4 °C
concentration levels
day 0
day 3
mean,
%
mean.
%
compound
Mg/L
SD
eg/L
SD-
carbon tetrachloride
33.8
1.3
33.8
1.7
chlorobenzene
32.5
1.0
33.3
1.0
1,1-dichloroethane
34.3
1.5
30.0
1.7
chloroform
52.3
1.5
84.0
2.6
1,1-dichloroethane
29.8
2.2
26.3
2.0
1,2-dichloroethene
41.5
2.1
45.8
1.7
methylene chloride
45.5
1.3
57.0
1.4
chlorodibromomethane
42.3
1.5
47.8
2.2
trichloroethene
22.3
1.3
28.0
0.0
Concentration, ug/L
120
Time (Days)
j 0 Oiehloroetnylana 26C ^ T< icMoroetnane 2SC ?
I Trichloroelhylena «C TalrgchlOfOfllhane *C |
initial Spike- SO ug/L
Figure 1. Dehydrohalogenation reactions In distilled water.
create a bias in the concentration of the first aliquots
compared with the final aliquots. Therefore two possible
methods were tested. In the first method, the water sam-
ples were added to the vials and the stock solutions added
to each vial. In this case, variability in the concentrations
found in replicate vials was 10-20%. In the second me-
thod, the stock solutions were added to a Tedlar gas sam-
pling bag, mixed in the bag with no headspace, and added
to the vials {9, 10). In Table II are reported measured
concentration data for* solutions stored for up to 3 days
in Tedlar bags. The data, which are simultaneously a
measure of both analytical precision and aliquot to aliquot
variability, indicate that this procedure produced con-
centration variability of less than 5%. It was therefore
selected as the method of choice. Subsequent studies
showed that the concentrations of the compounds in the
bag did not change over a 24-h period.
In Tables III and IV are presented the maximum
holding times (MHT) calculated for each of the 17 target
constituents at both high- and low-spike levels. Those
MHTs for which a cubic spline calculation was required
are noted. For the high-level-spike conditions, the ASTM
criteria resulted in longer holding times in more than 70%
of the cases for which a difference was observed. For the
lower concentration study, the two sets of criteria produced
nearly equal fractions of longer MHTs. With the exception
of 1,1,2,2-tetrachloroethane, all of the compounds were
stable in distilled water for 24 days when samples were
stored under refrigerated conditions. This is interesting
in view of the limited official holding times allowed for this
type of sample. At room temperature, thetetrachloro-
ethane disappeared very quickly with a concomitant rise
in the concentration of trichloroethylene (Figure 1). This
reaction also occurred at 4 °C, although at a slower rate,.
Concentration, ug/L
70 e —
60 .
i I
20 u j
i
10 ^ '
0I 1 1 1 J
Day 0 Day 3 Day 7 Day 14 Day 28 Oay 56 Day 112 Day 370
Time
Mathylana CMorid* * Chloroform 0 Trlchioro*lhyian« |
TalracMoroathan* -**- Carbon TatracMoNd*
Initial Splka: 60 U0/L
Figure 2. Stability of chlorinated hydrocarbons In surface water stored
at 4 °C.
Concentration. ug/L
60*v
Day 0 Day 3 Oay 7 Day 14 Day 28 Oay 56 Day 112 Day 370
Time
| *— Toluena ° Elhylbenzene Styrene ¦
Initial SO ug/L
Figure 3. Stability of aromatic compounds in surface water stored at
4 °C.
Likewise, a decrease in levels of trichloroethane was ac-
companied by increased concentrations of dichloroethylene
in those samples stored at room temperature. These
phenomena are most likely due to dehydrohalogenation
reactions. Importantly, there seems to be no difference
in the stability of the compounds based on volatility, which
suggests that the current containers and storage conditions
are quite adequate for the elimination of losses due to
volatilization. There also does not appear to be any im-
portant differences in MHTs based on concentration, with
the high-level samples behaving in a manner similar to the
low-level samples. There were slightly more cases in which
the high-spike MHT was longer than that of the low-spike,
but the difference was not statistically significant.
In the case of the groundwater and surface water sam-
ples, both of which contained chloride, no dehydrohalo-
genation was noted under refrigerated conditions (Figure
2). However, degradation of the aromatic volatiles became
apparent within 28 days (Figure 3). It is not clear whether
this is the result of chemical or microbial action, although
both probably play a part. Since all samples were stored
in the dark, it is unlikely that photodegradation occurred.
The degradation was most pronounced with ethylbenzene
and styrene. The data for four EPA-CLP matrix spike
compounds in groundwater are shown in Figure 4 relative
to their contract-required recovery limits, and none fall
outside the mandated range, even though the calculated
MHTs for many of the component/criteria combinations
were less than 365 days. (Note that, in this case, the
narrowest matrix spike recovery window—76-125%—was
applied to ±10 ^g/L about the spike of 50 ^g/L-)
Environ. Scl. Technol.. Vol. 24, No. 11, 1990 1667
-------
Table III- Maximum Holding Time (Days) for Unpreservcd Water Samples (Spike Level 500 fg/L) Determined for Both
ASTM and ESE Definitions (See Text)0
distilled water grouiidwattr surface water
4 "C rm temp 4 "C rm temp 4 °C rm temp
compound
ASTM
ESE
ASTM
ESE
ASTM
ESE
ASTM
ESE
ASTM
ESE
ASTM.
ESE
methylene chloride
147
187
86
108
210
131
219
97
203
273
118
91
1,1-dichloroethylene
148
133
61
30
103
86
71
33
365
365
49
28
1,1- di chVoroethan e
153
184
365
365
183
170
220
154
365
365
365
365
chloroform
131
183
142
153
IS9
186
225
ISO
365
365
365
365
carbon tetrachloride
365
365
122
82
274
365
244
365
181
125
43
1
1,2-dichtoi cpropar, e
102
92
130
91
365
365
365
365
131
140
365
365
trichloioethyleJie
i24
28
51
21
202
166
98
57
365
365
314
33
benzene
m
14S
151
115
145
148
172
100
217
180
2S2
365
1,1,2-tricbioi-oetliane
183
171
14*
8*
365
355
34
no
231
365
365
365
bromoform
212
190
B8
61
365
385
365
365
238 .
¦ 365
143
68
1,1,2,2-tetrachloroe thane
?*
3*
1*
1*
265
3S5
7
3
330
365
36
5
tetrschloroethylene
126
92
22
3
78
58
40
U
86
92
34
7
toluene
288
365
72
41
168
152
145
62
75
86
120
41
chlorobenzene
307
365
46
20
365
365
125
62
365
365
134
39
ethylbejiaene
365
365
52
19
179
140
65
13
22*
11*
117
27
styrene
365
365
48
16
365
365
77
12
82
59
54
11
o-xylene
365
365
102
45
365
365
70
20
365
365
272
365
" Asterisk indicates data calculated by using a cubic spline carve fit.
Table IV. Maximum Holding Time (Days) for Unprescrved Water Samples (Spike Level SO **g/L) Determined for Both
ASTM and ESE Definitions (See Text)"
distilled water groundwater surface water
4 *C rm temp 4 "C rm temp 4»*,C rm femp
compound
ASTM
ESE
ASTM
ESE
ASTM
ESE
ASTM
ESE
ASTM
ESE
ASTM
ESE
methylene chloride
365
36a
199
197
189
252
365
365
63
126
64
125
1,1-dichloroethylene
365
365
93
58
225
365
151
99
70
81
61
67
1,1-dichloroetha/ie
252
365
237
365
365
365
365
365
48
106
74
152
chloroform
202
145
134
101
365
365
155
186
51
110
160
303
carbon tetrachloride
365
365
356
365
365
365
138
76
84
115
30*
15*
1,2-dichloropropane
365
365
365
365
86
66
130
78
75
243
365
365
trichloroethylene
133
24
180
162
301
365
365
365
47
132
161
138
benzene
365
365
324
365
97
32
169
¦ 101
55
103
76
100
1,1,2-tricJiJoroethane
278
365
7
4
42
105
1CK
144
77
214
U9
259
bromofcrm
283
365
132
49
81
99
86
51
45
129
23
37
1,1,2,2-te trechloroet hane
15'
7*
1*
0*
83
91
14
3
165
184
10
6
tetrachloroethylerse
365
365
119
34
365
365
203
87
49
84
74
66
toluene
365
365
218
112
62
10
49
20
50
104 ¦
26
32
chlorobenzene
358
365
197
90
3-55
365
134
65
57
132
104
129
ethylbemene
355
365
218
B0
26*
25*
32
S
14<
12*
46
16
styrene
248
365
155
52
6*
5*
3
0
11*
9*
2*
1*
o-iylene
310
365
217
91
74
12
19
3
118
108
29
25
• Asterisk indicates data caleJlated by using a cubic spline curve fit.
Due to the nature of the concentration changes that
occurred in the course of this study, it became apparent
that addition of hydrochloric acid to the samples, reducing
thepH below 2, might inhibit both dehydrohalogenation
and degradation of the aromalics. Therefore, a second set
of experiments was performed, using the same three water
samples stored under refrigerated conditions and analyzed
at intervals of 0, 34, 28, and 56 days. No deterioration was
rioted in any of the compounds except styrene. Moreover,
the stability of styrene was greatly improved, with almost
80% remaining after 56 days. This study indicated that
the maximum holding time of volatile organic compounds
in water could he increased to at [east 56 days if samples
are preserved with hydrochloric acid. The significance of
such an increase to the environmental analytical laboratory
cannot be overemphasized. Howeve*, preservation of water
samples with HC1 has several disadvantages. It is difficult
to ensure that the pH of the sample is reduced to 2 without
fiTSt.meas jringsarnpIe pH. It is else inconvenient to add
a corrosive liquid during Held sampling. Finally, HC1 does
have appreciable volatility and can be introduced into the
t DncenVaVicn. ujf'l
SO: . )
P
60 " ...
¦' - '
¦ * . - «
<0 - jr • ,
i
r *— ¦ 1
20 |
i '
0 Z T u 26 56 112 365
Time IDeys)
j ChtO'3b«ni«si» MS Recovarf Units
inriiai Concsnrj-atloh: 60 ug'L •/- i0\
Figure 4 Slab/lfty of matrix spike compounds in groundwater a! 4 cC.
Sold bars intfieale conserva've "navx spike ^ecows-y limits (see "texti.
instrumentation during purging. The possible detrimental
effect on the analytical equipment cannot be tolerated.
1668 Environ. Scl. Techno!., Vol. 24, No. 11. t990
-------
Table V. Calculated Maximum Holding Times (Days) tor Acid-Preserved Votatiies in Water Determined for Both ASTM and
ESE Definitions (See Text)"
distilled water groundwater surface water
sodium sodium sodium
ascorbic acid bisulfate ascorbic acid bisulfate ascorbic acid bisulfate
compound
ASTM
ESE
ASTM
ESE
ASTM
ESE
ASTM
ESE
ASTM
ESE
ASTM
ESE
chloromethane
112
112
112
112
112
112
If2
£12
112
112
U2
112
bremomethane
112
112
112
112
112
112
112
112
112
112
'96
112
chloroethane
112
112
112
112
112
112
112
112
112
112
61
12
methylene chloride
112
112
30
33
7
112
35
34
77
68
62
59
acetone
41
30*
112
112
112
112
112
112
112
112
112
112
carbon disulfide
97
112
49
9
106
112
61
12
23
10
54
10
1,1-dichloroethene
112
112
106
112
112
112
80
18
45
34
112
112
1,1-d ichloroethane
112
112
112
112
112
112
112
112
112
112
112
112
chloroform
112
112
112
112
105
112
112
112
112
112
112
112
2-butanone
112
112
112
112
112
112
112
112
112
112
112
112
carbon tetrachloride
<14
<7
91
75
112
112
112
112
112
112
112
112
1,2-dicUoropropane
112
112
64
62
112
112
112
112
112
112
70
73
trichloroethene
51
24
56
66
47
27
112
112
52
24
51
43
benzene
47
42
112
112
54
57
112
112
62
65
42
67
1,1,2-trichloroe thane
75
55
66
86
88
48
112
112
57
56
53
7B
bromoform
112
112
112
112
112
112
94
112
112
112
112
112
4-methyl-2-pentanone
68
29
112
112
67
21
112
112
31
23
112
112
2-hexanone
97
112
112
112
92
22
112
112
55
12
112
112
1,1,2,2-tetrachloroethane
61
44
112
112
58
32
112
112
33
34
85
112
tetrachloroethene
53
27
77
42
45
24
112
112
41
28
63
42
toluene
50
40
106
112
37
52
112
112
32
49
94
112
chlo'robenzene
48
36
112
112
47
55
112
112
36
49
86
90
ethylbenzene
51
34
67
65
44
51
112
112
34
44
112
112
styrene
25
a
54
56
84
47
112
112
81
46
112
112
o-xylene
96
112
42
58
85
67
112
112
69
55
112
112
"Asterisk indicates data calculated by using a cubic spline curve fit.
Concentration, ug/L
too i
60-
40 a
[ * c
20 *¦
I
0' i « : 1— A -ft .
Oay 0 Day 3 Day 7 Oay 14 Day 2B Day 50 Day 112 Day 3>o
Tim&
| • Etnyl&iiune —£1hy i b«n£of>0* N»HSO<
j o Styrsn* Sly*«r* • NtHSO*
Initial BDike {Unprescrw*dl: wc'L
Inilul spike (P«e«9fwed>: IDG uc'L
lnte*r.at irtnCtid auanthilion
Figure 5. Bisu'faie preservation of aromatics In surface waler stored
a1 4 °C.
Therefore, there has been a genera! reluctance to require
the use of HCI as a preservative for volatile organic com-
pounds in water.
Because of these problems, and also because the data
base suggested that pH reduction was the primary factor
involved in preservation, an attempt was made to identify
other acids that might have the preservative effect of HCl
without the attendant disadvantages. Two candidates were
identified: sodium bisulfate and ascorbic acid. Both are
noncorrosive (in the dry form), readily available, inex-
pensive, and nonvolatile. A 112-day study was carried out
using these acids as preservatives; the data generated were
compared to those obtained without preservation and with
HCl preservation. The MHTs calculated for the preserved
samples are reported in Table IV. In Figure 5 are por-
trayed the concentration vs time data for ethylbenzene and
styrene, two of the least stable aromatics. It is readily
apparent that sodium.bisulfate is as effective a preservative
Concentration, ug/L
500 -|
400 •
^00r
Day 1 Day 7 Day 14 Oay 20 Day 56 DAY 112
| ~ Action C«rbon Diswlffd» * 2-BulsriDn# |
! «'Mc-?»penic>on« -- 2-H«
-------
Concflnlralion, ug/l
500 i
400 -
Cay 1 Day 7 Day 14 Day IS Day 56 DAY 112
i —AcclDn* Carbon Ditu^'id* • 2 ¦ Oul* ~ ona j
A 4*M«-2-P«fttinen* — 2-H txMn9f I
L _ _ . I
FVgurs 8 BteuMate preservation of ketones in surface water stored
at 4 °C.
carbon disulfide. These compounds were not included in
the unpreserved study because of difficulty in obtaining
standard compounds at the time of the earlier study.
Gradual reductions in the levels of carbon disulfide were
evident during the 112-day study; the other four ketones
remained at or near their original concentrations. The bold
lines in Figure 7 again indicate EPA-CLP matrix spike
recovery limits. One would expect greater variability from
the more soluble ketones than from the more purgeable
matrix spike compounds. However, virtually all the ketone
data fall within matrix spike recovery limits.
Ascorbic acid was equally effective in preserving most
volatiles studied. However, it was not possible to acidify
the samples to pH = 2 with this acid, and solubility
problems were encountered before reaching pH = 3. Ad-
ditionally, the quantitation of bromoform proved difficult
in the pesence of ascorbic acid, with high standard devi-
ations between replicate samples.
Comparing the MHTs determined for both the unpre-
served and the preserved samples, there were 53 instances
(constituent, matrix, criteria combinations) in which the
refrigerated MHTs for the unpreserved samples were less
than 100 days. In almost all of the cases (49), preserving
the sample with one of the two dry acids resulted in a
longer MHT. In over 85% of those cases, the MHT was
longer by 21 days or more. Thus, for contaminants that
have holding times shorter than 100 days, preserving the
samples with dry acid enhanced sample stability.
Conclusions
From a regulatory point of view, extension of the holding
times without compromising data quality would reduce the
cost associated with waste site characterization and rem-
edial action by reducing the possibility that additional
sampling will be required due to the failure to meet the
holding times. This has an important economic effect on
investigations carried out under SARA. From the point
of view of RCRA, where quarterly groundwater monitoring
is carried out-, preservation of the samples would allow
direct comparison with the samples collected during the
subsequent quarter. Since regulatory decisions are made
based on changes in the groundwater concentrations of
contaminants, this would be important in reducing ana-
lytical variability. From the standpoint of the regulated
community, the ability to preserve and archive important
samples (J J) for later verification would greatly reduce the
possibility of error in regulatory decision making and would
:ertainly eliminate the need for resampling.
From the analytical standpoint, improvements in the
quality assurance process are expected. For the first, time,
stable, long-term performance evaluation materials can be
prepared and submitted in a truly blind fashion to the
laboratory. Controls can now be prepared fQr use in field
sampling. Finally, an estimate of the intralaboratory
variability over time of the analytical method is now
possible.
Registry No. Water, 7732-18-5; chloromethane, 74-87-3;
bromomethane, 74-83-9; chloroethane, 75-00-3; methylene chloride,
75-09-2; acetone, 67-64-1; carbon disulfide, 75-15-0; 1,1-di-
chloroethane, 75-34-3; 1,1-dichloroethene, 75-35-4; chloroform,
67-66-3; 2-butanone, 78-93-3; carbon tetrachloride. 56-23-5; 1,2-
dichloropropane, 78-S7-5; trichloroethene, 79-01-S; benzene, 71-
43-2; 1,1,2-trichlo roe thane, 79-00-5; bromoform, 75-25-2; 4-
methyl-2-pentanone, 108-10-1; 2-hexanone, 591-78-6; 1,1,2,2-
tetrachloroethone, 79-34-5; tetrachloroethene, 127-18-4; toluene,
108-88-3; chlorobenzene, 108-90-7; ethyibenzene, 100-41-4; styrene,
100-42-5; o-xylene, 95-47-6; sodium bisulfate, 7681-38-1; ascorbic
acid, 50-81-7; hydrochloric acid, 7647-01-0.
Literature Cited
(1) Worthy, W. Chem. Eng. News 1987, 65, (Sept 7), 33-40.
(2) Friedman, L. C.; Schroder, L. J.; Brooks, M. G. Environ.
Sci. Technol. 1986, 20, 826-829.
(3) U.S. Environmental Protection Agency. Statement of Work
for Organic Analysis, 1986.
(4) U.S. Environmental Protection Agency. Quality Assurance
Newsletter, 1984, 6(3).
(5:) Draper,.N. R.; Smith; H. Applied Regression Anulysis;
Wiley: New York, 1984.
(6) Maskarinec, M. P.; Bayne, C. K.; Johnson, L. H.; Holladay,
S. K.; Jenkins, R. A.; Stability of Volatile Organics in En-
vironmental Water Samples: Storage and Preservation.
ORNL/TM-11300; Final Report, August 1989.
(7) ASTM 1986 Annual Book of ASTM Standards; ASTM:
Philadelphia, PA, 1986; Vol. 11.02 Water (II), pp 21-27.
(8) Prentice, H. S.; Bender, D. F. Project Summary: Devel-
opment of Preservation Techniques and Establishment of
Maximum Holding Times: Inorganic Constituents of the
National Pollutant Discharge Elimination System and Safe
Drinking Water Act. Research and Development; EPA/
600S4-86/043; March 1987.
(9) Schuetzle, D.; Prater, T. J.; Roddell, S. R. J.—Air Pollut.
Control Fed. 1975, 24, 925-932.
(10) Francis, C. W.; Maskarinec, M. P.; Goyert, J. C. Mobility
of Toxic Compounds from Hazardous Wastes; Oak Ridge
National Laboratory: Oak Ridge, TN, 1984; pp 3-9.
(11) Becker, D. A. Trace Substances in Environmental Health.
Proceedings, University of Missouri's Tenth Annual
Conference on Trace Substances in Environmental Health;
Environmental Trace Substances Research Center, Univ-
ersity of Missouri: Columbia, MO, 1976; pp 353-359.
(12) U.S. Environmental Protection Agency Methods for
Chemical Analysis of Water and IVasfe.s; EPA-600/-1-79-
020; Methods 130.2, 150.1, 310.1 and 410.4: March 19S3.
(13) Clessceri, L. S., Greenberg, A. E., Trussel, R. R., Eds.
Standard Methods for the Examination of Water and
Wastewater, ]
-------
MAak-\R|SEC ET vLt 1 ASSOC OKF A\M. CHEM iVOU SO ? <23
REFERENCE STANDARDS
Preparation of Reference Water and Soil Samples for Performance Evaluation of Volatile
Organic Analysis
MICHAEL P. MASKARINEC, LYNNE H. JOHNSON, and CHARLES'K. BAYNE
Oak Ridge National Laboratory, Analytical Chemistry Division, PO Box 2008. Oak Ridge, 77V 37831 -6120
Methodology was developed to reproducibly prepare performance
evaluation materials for volatile organics analysis in soils and waters.
Tedlar gas sampling bags are used to prepare the volatile organics
spike solutions. The bags allow large volumes of sample or spike
solution to be prepared to a high degree of homogeneity while using
less methanol. Preparation of a large volume of sample or spike
solution allows for increased a ccuracy of fortification. The accuracy
is generally ±20% or better, and the precision is generally ±10% or
better for water samples. The precision for preparation of soil sam-
ples is also good, but tbe accuracy suffers from variable recovery
efficiencies from the soils. Most volatile organics were well preserved
in 3 water samples by storage at refrigerator temperature for 14 days.
Demand for analyses of environmental samples for volatile
organic compounds has dramatically increased in recent
years. This is due to increased public awareness of the grow-
ing problem of hazardous waste disposal and the abundance
of hazardous waste dump sites nationwide which require
remedial action. The need for hazardous waste cleanup and
the methodology required for establishing environmental
sample banks has brought about the simultaneous need for
analytical support. Some of the most commonly identified
contaminants include volatile organics.
To provide for consistent .results from analytical laborato-
ries nationwide, the U.S. Environmental Protection Agency
( EPA) has issued various analytical methods in the Federal
Regisier to standardize analyses. Among the quality assur-
ance needs in these methods is the need for reference sam-
ples. Such samples enable interlaboratorv comparisons to be
made. Because of difficulties encountered in the preparation
of reference samples for volatile organics. laboratories com-
mor.is add a methanolic solution of these compounds to blank
water samples immediately prior to performing an analysis.
This situation precludes the use of blind performance evalua-
tion samples. The objective of the present'work is to provide
methodology capable of producing performance evaluation
materials for volatile organics in water and soil that are truly
biind. a.ic therefore allow a more relevant assessment of the
precision and accuracy of the analytical methods.
Two criteria need to be met by performance evaluation
samples. These samples need to be "real", i.e.. they should
closeiv simulate the composition of actual samples. This is
cue to the need for reference samples to be blind samples.
Also. these samples need to be of defined stability. This is
necessary because of sample preparation and transfer time to
the analysis sites, and sample analysis requirements it the
a.-iaiyuca: laboratory. These 2 criteria often conflict. For
exampie. previous work by Friedman. Schroder, and Brooks
• i ». ir, which methanol was acced to samples spiked with
••oiatiie organic compounds to check the preservative effect
of such an addition, found no improvement in sample stabil-
Received Novemocr IS. 1983. Accepted Marcn 30. 1939
Research sponsored bv the U.S. Environmental Protection Agency. DOE
I \0 \j I tl-i-1 744. a I. and the L S. Department or" DeJ'ense. L S. Navy
jnU I S -\rm> Toiic and Hazardous Materials Agency. DOE IAG No
\ "
-------
M^SKARI'NEC ET M_.. J \SSOC OFF. ANAL. CHEM.'iNOL. SO. 5. :9S»>
Figure 1. Experimental purge and trap arrangement lor soil samples.
ride. ]. i -dichloroethene. !.-ciichioroethane. chloroform,
carbon tetrachloride. I ,2-dichloropropane. tricnloroeinene.
benzene. ].l.2-irichloroeihane. bromor'orm. I.I,2.2-ietra-
ch'ioroet'nane. tetrachloroet'nene. toluene. chlorobenzene.
ethylbenzene. styrene. anc o-\>iene.
Method
Water samples.—The desired water w as dispensed into a i
L Tedlar gas sampling baa (3. The water was aiiowed to
degas for 2 days, and the gas was removed from the bag.
Target compounds were received from '.he aforementioned
sources as meihanotic solutions of 1S00-2300 yg.voiaules
mL methanol. Appropriate volumes of each stock volatile
organic solution were introduced through the septum port via
gas-tight syringes. The contents of the Tedlar bag were
mixed thoroughly by hand agitation for 3 min after which the
bags were allowed to sit for 30 min. Alter mixing, aliquots of
the sample were transferred to the -iO mL vials by gravity
flow. Teflon tubing ('/a X 6 in.) was used to allow the vial to
be filled from the bottom upward, preventing mixing of the
water with air. Each bottle was completely filled with sample
so that no headspace would remain after the bottle uas
sealed. Each bottle was sealed immediately with a Teflon-
faced septum and screw cap with hole, and stored at the
appropriate temperature t J and 25° Ci.
Soil samples.— Five e soil was added to a 40 mL vial.
Twenty to 25% (volume/weight) reagent grade water was
added to partially wet each soil sample. Each vial was sealed
with a Teflon-faced septum and screw cap with hole. Each
vial was mined on a Voutx mixer for about 10 s and allowed
to sit at room temperature in the dark for 3 days. This was
done to inmate bacterial acttvn\ in the soil and a 1 low the
activity to come to equilibrium ipersonal communication. B.
T,"Walton. Environmental Science Division. ORNL. 1 9S61.
The spike solution of volatile* was prepared in a Tedlar bag
as described for water samples The spike solution w 3s added
:o each soil sample at the -j:e of 20-0 25 rr.L icueous
volatiles/g soil. At this point, the soil sample was 80-100%
saturated with water. Each vial was immediately capped.and
the contents were mixed on a Vortex mixer for about 30 s and
stored or analyzed as appropriate.
Analysis
Alt volatile organic analyses were performed by purge-
and-trap gas chromatography w ith mass spectroscopic detec-
tion (GC/MS) (i) approximately according to standard
EPA Contract Laboratory Program (CLP) methods (6). In
'.his procedure, an inert gas is bubbled through the sample,
removing volatile compounds from the aqueous phase and
carrying them to a sorbent column. Volatiles are trapped on
this column until it is heated and backflushed. desorbing the
purged compounds onto a gas chromatographic column.
Temperature-programming of the column oven permits sep-
aration of the volatiles mixture; individual components are
then detected and identified by mass spectroscopy.
Soil samples (with added reagent water! were purged di-
rectly from the 40 mL vials by use of Teflon couplings ma-
chined in-house for this study. This was done to allow an
assessment of the actual content of the vials without the
complicating factor of sample weighing and transfer. A dia-
gram of this experimental arrangement for soil samples is
show'n in Figure 1.
Results and Discussion
Precision and Accuracy
There are 2 possible approaches to the preparation of
multiple aliquots of performance evaluation materials. The
sample vials can be prepared and individually spiked, or a
large volume of sample can be prepared and aliquots can be
taken. In most cases, the second approach is preferred, since
this approach should provide for greater sample preparation
precision. However, for the volatile organic compounds, the
?ossibiiu> ni chancing sample composition curing disoensine
-------
M ^SKARINEC ET AL.: J *,SSOC. OFF CHEM. (VOL. NO. 5. I9S9)
s:s
Table 1. Repeatability 04 VOA spiking Into Individual VOA visit Table 2. Stability o< aqueous volatile* In Tedlar bags it 4cC*
(ng volatlles/L water)' 1 ~ ~ ~—'
Comoound
Beo 1
Seo 2
Seo 3
Carbon tetrachloride
21
Compound
Mean. ug. L
RSO.
Mean. uQ'L
SSD.
30
21
Chlorooenzene
:a
20
•9
Carson tetrachloride
33 8
1 3
33.8
' 7
' :-Dichloroethane
30
23
20
Chlorooenzene
32.5
1 0
33 3
' 0
1.".2-Trichloroemane
35
25
23
". l-Dicnioroetnane
34 3
1 5
30.0
* 1
Chloroform
^ i
-.8
-.9
1.1 2-Trichloroethane
45 8
1 5
52.0
: 0
\ '-Dichloroetnene
33
22
?2
Chlorotorm
52.3
1 5
c 4 0
2 5
t. 2-Dichloropropane
34
24
24
i. '-Oichioroethene
29 8
2.2
26.3
2 0
Methylene chloride
33
20
19
l.2-Dichloropropane
41.5
2.1
45 8
1.7
Chlorooibromomethane
£. 1
"3
•, 7
Methylene chloride
45.5
1 3
57 0
1.4
Trichloroethene
'< 7
12
12
Chlorodibromometnane
42 3
1 5
47 8
2.2
' laoea concentration: 30 ± 6 l.
Trichloroethene
22.3
"..3
2S.0
0 0
' *30eo concentration: 40 2: 8 ug I. n * i
of aliquots complicates the situation. A preliminary studv
uas performed to evaluate the differences between these
approaches for water sampies. A s>stematic error was imme-
diately noticed between vials, and the amount of methanol
had to be increased in proportion to the volatile organic
anah tes (Table 1). Large amounts of methanol create prob-
lems in the analysis. The amount must be kept below 50 yL,
40 mL.
The use of the Tedlar gas sampling bags provided several
advantages. First, large volumes of sample could be prepared
using minimal amounts of methanol. Second, vial-to-vial
variability was reduced. Third, the sample could be success-
fully stored in the bag (Table 2). Fourth, the accuracy of the
concentration of the analytes in the sample was improved due
to the relatively larger voiume of spiking solution. Finally,
the bag allowed the filling of the individual sample vials with
no introduction of headspace and minimal mixing of the
aliquots with air during filling of the vials.
Data from the analysis for 17 volatile organics in 3 water
samples on day zero are given in Table 3. With a few excep-
tions. the standard deviations from 4 replicates for each data
point are approximately lO^c or less of the mean value. It is
useful to look at each individual water sample at a given
concentration level to check the average of the data for ail of
the compounds. Table 3 also gives the average of the vaiues
for all compounds, for each water sampie and concentration
!evpl. The target concentrations were 50 ug/L- for the low-
concentration ievel and 500 ng/L for the high concentration
level. All of the averages of the values for a given sample ana
concentration level are within 15^ of the target concentra-
tion. with standard deviations of around 20?c or less.
Data from the analysis for the same volatile organics in 3
soil samples on day zero are given in Table 4. The target
concentration for these samples was 100 ng/kg. The data
indicated actual concentrations in the range of 14 to 118 Mg/
kg with standard deviations of less than 25°c. Although the
accuracy is not as good as with the water samples, there is the
additional complication of variable recovery of the analvies
using the analytical method. Such variability is evidenced by
the surrogate and matrix spike recovery limits specified in
the EPA CLP quality assurance requirements (6), which
have much wider limits for soil than for water. However, the
precision is acceptable for the purpose of performance evalu-
Table 3. Volatile organic compounds In water samples'
Water 1 Water 1 Water 2 Water 2 Water 3 Water 3
low level high level low level high level low level high ievel
Compound
Mean
5D
Mean
SD
Mean
SD
Mean
SD
Mean
SD
Mean
SD
Methylene chloride
53.1
3.5
464.1
31.3
66.4
1.5
396.1
111.3
69.9
3.7
544.0
- 38.8
1.1-Oichioroethene
57.6
3.2
472.9
17.5
61.3
1.7
383.4
121.1
53.2
46
408.9
61.3
1.1-Oichloroethane
63.9
2.7
514.7
18.1
66.2
• 6
388.6
111.3
64.7
3.4
593.8
59.5
Chlorotorm
59.6
1.8
488.8
25.0
61 0
1.4
385.5
114.3
69.5
4 1
570.3
46.7
Carbon tetrachloride
50.7
3.8
445.9'
27.2
56.2
1 0
551.1
39.4
46.8
2.6
681.8
62.4
1,2-Oichloropropane
49.7
3.0
421.0
13.2
34.4
0.3
589.7
64.5
55.1
2.3
553.7
79.2
Trichloroethene
56.7
3.9
287.0
1.4
107.9
2.9
570.3
63.2
60.0
3.6
505.7
61.0
Benzene
49.6
2.7
400.2
11.1
54 6
0 2
569.7
50.2
55.5
2.3
479.7
34.7
1.1.2-Trichloroethane
47 3
1.7
440.5
17 8
55.4
1.0
584.4
74.9
59.4
2.1
649.5
150.1
Bromolorm
33.9
3.5
619.3
47 2
49.6
0.1
583.3
77.9
54.8
2.2
766.1
178.2
1,1,2.2-Tetrachloroethane
418
1.3
531.3
49.1
48.5
6.0-
544.0
36.5
54.2
5.5
"97.3
127.1
T etrachloroethene
49.8
3.1
369.7
19.1
52.6
1.7
530.6
40.5
57 0
4.3
398.1
47.4
Toluene
49.7
3.1
383.0
19.4
54.1
0.8
550.0
34.4
54.3
3.3
462.3
35.0
Chlorooenzene
51.6
2.6
38 V7
21.7
55.4
1 1
547 4
36.5
55.9
3 6
476.9
33.9
Ethylbenzene
48 6
1 5
378.0
23.8
51.2
. 1.1
532.2
43.7
52.1
4 1
416.8
42 2
Styrene
52.5
2.4
394.1
24.4
63.3
1 8
535.2
41.2
57.7
3.5
486.1
28.7
o-Xylene
53.7
2.7
397.5
23.4
53.0
0.9
531.4
36.0
58.3
3.9
473.6
34 2
Mean
SO
RSO. %
51.2
6 6
12 8
434.7
74.5
17.1
58 3
14 4
24 8
516.1
73.1
14.2
57.6
5 8
10.0
545.0
115.7
21 2
4
-------
SZ6 MASfcAfUVEC ET >L.: J -kSSQC OFF ^S ^L CHEM i^OS. \Q /. : ="5o.
fabli 4. W »tll< wginlc ;arfrWru In i mollr* 1=bl« 5. 51tlljt ral cg» korr mi ycli c i 3uirIfcrnp!tt
Sail 1
Sjlr 2
Sa'i 3
Gonrounds1
sjia an
i?
¦R&D. V
^crriDOunas
SO
1 .J«ear
'¦ an
5D
E-rcmomeihap-e
66
« «i
-k. J
3 >
Ci(m cwifiarift
se
z
5: iTicna^A
Sri
V
VS
-
Me^ij'iene cnirx os
£6
22
Z. 1
C.u. ^cerare
£ '
:a
c3
6
;a
-1
1,1-OichlcroeirifjrK
:9
t
:
-.'•en-iyiere Ciilcnfle
cT
7 2
- -a
1. "\ -D'»chtarsseil-^.ie
= 3
;
¦0
1 :0«ci-ioroe-thene
£4
'0
Gc-
¦ ?
=3
3
C-iiOfciorrn
£ 4
£
¦0
'.'-Oic^lo'cetiare
:c
1 ;
91
' 5
132
3
Carbcr- teirgcbkiriae
c&
5
CMcrai-crm
*3
7
79
'2
\2
2
' .2-0 ic-nioroarcpan€
o2
-o
12
Carc-cm je^ach O'iOe
£4
3
65
: 5
*
T'k;hic»ro-e[he-ne
"3
A
VJ-Oichlorapr^pane
71
10
E2
*5
rS
5
Sensens
6-8
1
Tj-'cniwosine^e
72
10
59
"7
7 H
2
1 \2-Tric."»loroelharis
?4
: i
Serizsne
£-7
c
69
"5
c£
1
£
-O
0
^"
'. 1,2-TrichtaroeU".5.,"ie
53
s
63
c
£C
"
3,1.2^2-7e :r&ct 1 £>rceihs r &
f 3
6
D1
51 &FT1G tOrJT>
i |
5
23
«
50
4
7=e i rac-h! ore e neri-e
ES
. c
3
1-1.2 2-1 e-u5calcacettv?r e
42
5
*0
3
~6
2
Toluene
€1
a
14
Taifac"iiDfoemeris
64
9
5-c
- s
:s
5-
Ch&'cbenze/ie
-»6
S
17
^c-sjene
TC
3
T
iA
1
cth/Cs&nzene
55
10
T
Ch.o^ate'iJ&ne
52
S
35
*
5 J
1
Serene
55
'4
57
Etriytoerzerie
67
7
43
6
53
i
oXytene
54
11
20
Styrene
45
3
ie
3
-.4.
0 Xylene
as
5
3S
3
f 7
3
'I'rt&l 5p~+.0 -4v»l ICC j
' Cjfktefitrs- 00 aMact: iCC i ""O-
L-c'h- fctfK- -c
- ;^n * *
,
aLior. Recovery of "lit highly ixiaiile comtMjrds brcnna-
methanc and chlorost.iane indicates itiac analyse lesses '-c
vTclatiiiza'-iar Lktit 5 uccessfuih -estrkted fcysajnpie prepara-
tion and storage methods.
Reprcducibitity
Table 5 presents jTiean qiues cerivsd by a re ratine Jay
zero results fc-r a pattkuJar ;anpouiKi from ail 3 ro!sU-e
urEan;:- i_n water samplts with a gj'.'tn :a-gd concentrator.
Al: compounds sicepi iricJitoroeihers nad a stanaarc d-e'-'i-
sticm of >ess than 20?=. tnGeec.. for many compounds 2L 1o**-
c&n-ceniraiicn. the relative statndarc. deviation was less 1 ha.n
3'5d.Table 5 c.tariy indieitcs a hiehcitgr-ee ifrsp-icaucifcili-
tv cm be coiajned for volatile orjarii-cs in ¦water samples- by
Tat *5. S-la(t»l>eai data r-on analyst* of low and high.
«one«ntr.aJion leant 01 3 watm sainpiBS"*
lc* lew^i le^ei
HJm-
SC-
F.5C. %
Mean
S3
F3C ^
UetT)HB«W Iltorrjs-
f z
' l.=
4 50.1
50. i
-2.?
i.' -Oict fc roe .-isra
J' ¦:
a 3
S.E
1?:.'
^r.»
8 ?
3
C. H
.4
j;m
c«-_E
"G 5"
C i."a I'lo ti
£3 L
¦1 >
:.S
4E-.S
¦ !-r
IS.'
Cite '¦ tr.'act -: ide
51;
3 E
i.S
£=: S
¦::. ~
57 J
"-v
>JEI
BE
IE 9
S21 5
^:j£
IjJS
r-«r K-rMTr
: J.i
¦"3.4
31 3
>
"21 2
2E 7
;X2
?.6
-a ?
^B3 ;¦
53 2
¦ a 3
l Ji't ;ito>
= «.D
: 3
S,s
:"A1
:? 3-
5 5
= reflate rm
Jc i
E 3
>j
E:5J
= 3 1
-2 :¦
.1 2,j-~e-T»cr v-c-
efrA-iB
48.1
5.1
lO.i
£24.2
122.5
9.S
T etraciloroetfienB
531
2 3
5 e
432 8
70 1
I6.2
T-ciu«r>e
bin
.2-1 '
4 0
465.1
£B 2
14 1
Chlofobemena
5^ 3
3,9
3-5
4&a. t
67.-S
14.5
ttfi y! Mniers
SOS
IS
2.9
44J.3
6i.5
'f* £
Styrerw
S ? B
4 4
7.7
471 e
66 5
12.4
o-*yiene
55.0
2*
4 3
467 5
54 e
11 7
' coth"'d " ¦ 3i
the methodology of sampit preparation anc arslysis present-
ed in this paper.
Tafcie 6 presents similar cita tor the scil sanrplc-s. .^eain
3". p-;risicn iri accuracy a-c hoi as good, as far -he water
samples. but it is -difficuh to distinguish-sanpie »"ariabiliiy
from rrtthiod variability. Jr. fact li cailc ac asivmid that
>c s;* a jtisd si *At ic i sair.plts i~ a manner
wended to thai fojihe #acer samples, p-rrcssiora arc accjra-
ey oT "Jie sample preparation should t>e tougli:y eq.1iva.kn3.
The ncre-ased t-a nihility in ih-e results for sail samples ."s
probably due to method variability.
Stability
The stLdy bTtoe siabiJ:ty 0:' ihesc coirs poLncis is tine subject
of a. second repon. and tine results n-ilJ he summariiec iiere
or.Vy to the extent thai Lhey• .affc=i the production of perfor-
mance evahia(io/i samples. Vfost of tJte target compounds
remained ai or close to their day icrac-onceratraiicri on daw 1 J
fcrsiarageat boti 25 and 4"C. Exceptions tothisare I, i,2,;-
leirajciiIcrotitta-ie awi T -stse.cct.k. .ndi ctaiticed
jj,in:r=ar.l y on-: tig stofij; at 2isC. Tne ass of I i.iJ-
i:ira;ii arcetrarr ^as iccoenfin dc by a siu-.lLareoLS r-
crcaie :n 1 be C'3ticen:ra joricf .rkiloroe.Jrnr.ixlicilint t.^a:
ne decay sietbanissi of I.!-- 2-net*=c^lc"acihaie flay x a
d<-iyc.r3>al5geia:saB pees:- Z-ej:lcyki -cl ; -- <-e
arc:<3:-ly l: lc cins^ ccda: » TJa: i: = j3:s rticat:
.>= rrary af fie carirvic-J. fff.ci ^alalle argsnia are biii-
-ahitJ'ar jse :¦ psrfarirarpc Xf ir-1 ljlJ.'Ics l= nj ihcse
PK.tiWls.
£ _-nC -E ie r ¦
Tb: i::j ;t: ih i:tjc;¦ i;r- 11a: -aiat l-e jresiic fx-'z--
Tirce t:i :r. E-Lrr.^,^ -t?. tt ias 'ari
rr.aie concentratian wiLh reisonable precision ir. bc.h soil ana
water matriies. Funhermorc. -t i5 shown in wai-er ihai /r;Osi
of Lhe volatile orgarucs 0:" inieres: are stable at re/rigeralor
:emperaiure for a (ime sufficiem 10 allow distribution arrl
analysis. This should allow reference sarapJes :cr voJanle
crganics in water to be- prepared at a central liboraiory a.nd
distributed to participating-snaliiical laboratories. Studies
of nte-r!abora;or> ?er:cpmance cr Lh-.s rr.eir jc si.nc &ari-
-------
MASKARINEC ET AL.. J ASSOC. OFF. AL. CHEM. I VOL. NO. 5. 19891
pies cart now be performed for the first time. As the stability
of the samples is verified, in'tralaboratory performance over
long periods of time will also be possible for the first time.
References
ill Friedman. L. C.. Schroder. L. J.. & Brooks. M. G. t 9 S 6 >
Environ. Set. Terhnol. 20. S26-S-29
i 21 Becker. D. A. tl9?6) Proc.. L'liiversity of Missouri's Tenth
¦innuai Conference on Trace Substances :n Environmental
Health. Environmental Trace Substances Research Center.
University of Missouri: Columbia. MO. pp. 555-559
(5) Francis. C. V*'.. Maskarinec. M. P.. & Goyert. J. C. < ]0s-'
"Mobility of Toxic Compounds from Hazardous Wastes." O;*
Ridge National Laboratory. Oak Ridce. TN. rp. 5-c
i-U Schuetzle. D.. Prater. T. J.. £: Roddell. S. R. (i°"5) J. A;r
Pollui. Control. Fed. 14. 925-9.':
•5i Thomas O- v-- Stork. J. R.. A: Lammert. S L. i :?>0I J. Chr:-
matofr. Sci. 18. 583-595
¦ r ^ Siatement of Work for Organics Analysis. VOA D-1 2-VGA D-
56 i i0S61 U.S. Environmental Protection Agency. Washing-
ion. DC
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 36
Active Soil Gas Sampling - Collection by Air Withdrawal
S.W. Johnson
The Advent Group, Inc.
Brentwood, Tennessee
January 12-14, 1993
Las Vegas, Nevada
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 37
Identification and Quantification of Volatile Organics in Soils by Use of
Fourier Transform Infrared (FTIR) Spectroscopy
Jack C. Demirgian, Maureen Clapper-Gowdy, and M. Lyn Tober
Analytical Chemistry Laboratory, Chemical Technology Division,
Argonne National Laboratory, Argonne, Illinois; and G. Robitaille, U.S.
Army Toxic and Hazardous Materials Agency, Aberdeen Proving
Ground, Maryland
January 12-14, 1993
Las Vegas, Nevada
-------
Identification and Quantitation of Volatile Organics in Soils
by Use of Fourier Transform Infrared (FTIR) Spectroscopy
Jack C. Demirgian, Maureen Clapper-Gowdy, and M. Lyn Tober
Analytical Chemistry Laboratory
Chemical Technology Division
Argonne National Laboratory
9700 South Cass Avenue
Argonne, IL 60439
(708)252-6807
Fax (708)252-5655
G. Robitaille
U.S. Army Toxic and Hazardous Materials Agency
Aberdeen Proving Ground, MD
To be presented at
National Symposium on Measuring
and Interpreting VOC's in Soils:
State of the Art and Research Needs
January 12-14, 1993
Las Vegas, Nevada
The submitted manuscript has been authored
by a contractor of the U. S. Government
under contract No. W-31-109-ENG-38.
Accordingly, the U. S. Government retains a
nonexclusive, royalty-free license to publish
or reproduce the published form of this
contribution, or allow others to do so, for
U. S. Government purposes.
*Work supported by the U.S. Department of Energy under Contract W-31-109-Eng-38 and the
United States Army Toxic and Hazardous Materials Agency (USATHAMA).
-------
Identification and Quantitation of Volatile Organics in Soils
by Use of Fourier Transform Infrared (FTIR) Spectroscopy
Jack C. Demirgian, Maureen Clapper-Gowdy, and M. Lyn Tober
Argonne National Laboratory
G. Robitaille
U.S. Army Toxic and Hazardous Materials Agency
Aberdeen Proving Ground, MD
Contamination of soils with volatile organics is a serious environmental problem.
Analytical methods need to be developed which will identify these volatile organic
contaminated soils in the field. Fourier transform infrared (FTIR) spectroscopy is a
promising method for the on-site analysis of soils. We report here an FTIR method for
the determination of selected target volatile organic contaminants in untreated spiked soil
samples. The method should prove extremely useful for the rapid field characterization
of contaminated sites and evaluation of remediation status.
The mid- infrared spectral region is ideal for the determination of organic compounds in
soils. Organic compounds have unique absorption patterns in the 1200 - 700 wave
number region, which is commonly called the "fingerprint region." Because this spectral
region is free of significant absorbance from atmospheric gases, there are no background
interferences. Use of FTIR spectroscopy improves the data quality because of the high
throughput and low noise compared with older dispersive instruments. The spectroscopic
data are analyzed by classical least squares (CLS) and partial least squares (PLS)
methods. More sophisticated software can be used to digitally filter the data and enhance
the analyte absorbance prior to quantitative analysis. By applying this advanced
technology to soil samples from hazardous waste sites, field samples should be rapidly
and accurately characterized.
Our initial work focused on determining the feasibility of thermally desorbing soil
samples (spiked with volatile organics) into a specially designed long-path cell and then
analyzing them by FTIR spectroscopy. After spiked soil samples were successfully
analyzed, the problem of interfering absorbance was studied. The final phase, which is
ongoing, consists of determining the optimum desorption hardware for field use.
To demonstrate feasibility, this report will focus on with dichloro methane, which is
similar to the other volatiles studied. Interfering absorbances were studied with a mixture
of trichloroethylene, herachorobutadiene, and hexachlorocyclopentadiene.
The volatile organics studied include methane, methyl chloride, dichloromethane,
chloroform, carbon tetrachloride, vinyl chloride, vinylidene chloride, trichloroethylene,
perch loroethylene, chlorobenzene, hexachlorobutadiene, hexachloropentadiene, diisopropyl
methanephosphonate, and kerosene.
-------
Experimental. All data \Vere collected using a Nicolet 60SX FTIR spectrometer (Nicolet
Analytical Instruments, Madison, WI). Spectral resolution was 1 cm"1, and 128 scans
were coded before data processing. Data evaluation was performed using LabCalc
(Galactic Industries, Salem, NH) and Nicolet PCIR and Least Squares Fit (LSF) software.
A heated, long-path cell, model 4-22 (Infrared Analysis, Anaheim, CA), was interfaced
to the spectrometer. The cell pathlength was set at 13.2 m for data collection of pure
analytical standards. The mixture data were collected at 3 m. The cell was heated to
180°C. Cell volume was approximately 4.7 L.
Analysis of soil samples utilized a CDS 122 (CDS Instruments, Oxford, PA) pyroprobe
and interface which was connected to the top of the cell through a heated transfer line.
Spiked soil samples were weighed in a sampling tube and placed in the pyroprobe. The
pyroprobe was placed into the interface. A shut-off valve separating the interface from
the evacuated cell was opened, and the pyroprobe heated. The interface and the
pyroprobe are referred to as the thermal desorption unit (TDU).
The liquid standards and mixtures, were injected into the TDU or directly into the cell
through a septa in a 1/4-in. (0.6-cm Swagelok fitting.
For all samples, a gas inlet allowed nitrogen to flow through the thermal desorption unit
and into the cell. The cell was evacuated to less than 1 torr (133 Pa) before each
injection. In a typical experiment, the nitrogen inlet was closed, and a background
interferogram was obtained. Then, vacuum line was closed, and the sample was injected
directly into the cell. Collection of the sample spectrum was delayed 20 seconds to allow
entry and dispersion of the sample vapor throughout the cell. The data were transformed,
and a ratio of sample-to-background spectrum was obtained.
Demonstration of Feasibility. The objective of this work was to demonstrate that it is
possible to thermally disturb a soil sample contaminated with a volatile organic and to
detect a response at low ppm. The reproducibility, optimum conditions, and detector
curve linearity also had to be determined. Chlorinated hydrocarbons have very strong
absorbance in the mid-infrared region. Hence, it was necessary to dilute the standards in
a isooctane to verify the linearity of the response as a function of concentration.
Isooctane was chosen because contaminated soils often have hydrocarbon residues,and its
presence would provide an interference that would be seen in field samples.
1. Spectral Properties and Detection Level
The infrared spectaim of pure of dichloromethane (DCM), commonly called methylene
chloride, is shown in Figure 1. This organic has a unique absorption band at 790-710 cm
that can be used for primary peak identification. Most of the absorbance is in the 790 -
740 cm region. A smaller band at 1290 - 1240 cm-1 overlaps the isooctane band,
centered at 1287 cm. In Figure 2, the spectrum of DCM in isooctane is shown above the
spectrum of pure isooctane (B). The isooctane spectrum was spectrally subtracted and
the resultant spectrum is shown (C). The subtracted spectrum has no interference, and
-------
this region could be used for secondary peak identification.
The detection level of DCM is approximately 10 ug in the cell, which corresponds to 10
ppm (based upon a 1-g soil sample). This value was determined from the peak
absorbance of 0.208 absorbance units (AU) for the absorbance centered at 750 cm-1. The
average noise was low (0.00130 AU); however, we will be conservative and use a value
of 0.0015. Using a 3:1 signal-to-noise (S/N) ratio, detection would begin at 0.0045 AU;,
or approximately 10 ppm at a cell pathlength of 13.2 m.
2. Reproducibility
(a) Sample Reproducibility: TDU Injection
Twelve samples were run by injecting the standard into the TDU. Data collection was
initiated within a minute of injection for eleven of these samples. The average response
factor was 15.83 area counts/|ig and the mean error was 0.232, which is a 1.5% average
standard deviation.
(b) Sample Reproducibility: Direct Injection
Five samples were injected directly into the cell. The average response factor was 17.36
area counts/jig with a mean error of 0.268 which is a 1.6% average deviation and agrees
with the deviation obtained by the TDU injection method.
The response factor is approximately 9% greater than that obtained using the TDU. Less
than 1% of this value is due to the extra volume of the TDU. The remainder of the loss
is due to the time required to attain equilibrium, as will be discussed next.
(c) Equilibrium Time: Both Methods
Determination of the time required to attain equilibrium after the standard is injected
depends on the temperature of the TDU. In an initial test with the TDU at 180°C, a
standard was injected and data collected twice, immediately after injection and then again
after a three-minute wait. The response factor increased slightly, from 15.91 to 16.21.
However, when the TDU was cool, after a three- minute equilibrium time, the response
factor increased to 17.24 (9%), which corresponds well with the 17.36 average obtained
for a direct injection.
The three-minute equilibrium time was also tested with the direct injection method. The
response factor for a standard analyzed immediately after injection and then again after
a three-minute equilibrium time increased from 17.65 to 18.77, an increase of 6%.
Clearly, use of an equilibrium time is required.
3.
Soil Sample Analysis: Hot vs Cool TDU
-------
For soil samples, there is a distinct advantage to a cold TDU. It requires several seconds
to insert the solid sample probe into the TDU and then tighten the nut to make the unit
air tight. In a hot TDU, this time can result in a significant loss of analyte through
evaporation. The soil sample analyzed using a hot TDU had a response factor of 7.62,
while another spiked soil sample analyzed using a cold TDU had a response factor of
10.47. Based upon an average response factor of 15.83 for DCM, the recovery of analyte
for a hot TDU would be 48% and for the cold TDU 66%. We anticipate that recovery
would increase above the 66% observed after a three-minute equilibrium time. The
infrared spectrum of the two soil samples analyzed are shown in Figure 3.
4. Linearity of Detector Response
The MCT A detector used for this study gave a linear peak response over a very large
concentration range. Figure 4 shows a plot of concentration versus peak area for
standards collected using the hot TDU. The datum at 3975 jag injected represents a neat
injection. This datum fits on a linear curve along with the data obtained at concentrations
of 263 through 526 jag injected. For lower concentrations, the line crosses the (0.0) point
of the plot; hence, the detector response curve is linear up to 4000 tag injected, which
corresponds to 0 to 4000 ppm for a 1-g soil sample. The large range of linearity is
significant. It means that a calibration plot" is not required before sample analysis. A
calibration plot can be obtained in the laboratory prior to going into the field. In the
field, a neat sample can be run, and the response factor compared to that of the
calibration samples.
By verifying the detector response with a neat standard, there will be no need to carry and
analyze multiple dilute standards in the field.
5. Mixture Analysis
A mixture of trichloroethylene (TCE), hexachlorobutadiene (BUT), and
hexachlorocyclopentadiene (CPD) was analyzed by FTIR spectroscopy.
The three components have overlapping spectral signatures. The analysis region for
each is shown in Table 1. The effect of severe interference would be false negatives.
Less severe interference would produce low analyte concentrations for one target
and a corresponding high concentration for others. Figure 5 gives the spectrum of the
three component mixture and spectra of the three standards.
Table 1. Spectral Regions for TCE, BUT, and CPD
COMPONENT
SPECTRAL REGION
TCE
960 - 750 cm"1
BUT
870 - 770 cm"1
CPD
830 - 690 cm"1
-------
Even though TGE, CPD, and BUT have overlapping spectral regions, quantitation was successful.
Data for the analysis of all five mixtures are presented in Table 2. Mixtures consisting of 5 and
25-|aL injections were used to provide data over a wide concentration range. Three 10-|jL
injections were made to determine reproducibility.
All three components were identified in the 5-|aL injected sample. There were no false positive
or negatives. The error for TCE was only 3.5%, while BUT and CPD results were low by
approximately 20%. The TCE concentration was slightly high. We did not observe the expected
result for interference, which is a corresponding increase of one analyte at the expense of another.
The mean error was 14.7%.
Table 2. Results of Analysis of Three-Compone Mix Containing TCE, BUT, and CPD
Quantity
Quantity
Volume
Component
Expected
Observed
Error
Mean
Tile
(nt)
Name
(ng)
(ng)
(%)
Error
No.
59
5
TCE
140.7
145.5
3.5
BUT
276.5
222.3
19.6
14.7
CPD
289.3
228.4
21.0
57
10
TCE
281.3
300.8
6.9
BUT
553.0
645.0
•16.6
22.2
CPD
578.5
689.0
19.1
19
10
TCE
281.3
132.7
52.8
BUT
553.0
585.8
5.9
22.1
CPD
578.5
622.2
7.6
22
.10
TCE
281.3
217.0
22.9
BUT
553.0
828.0
49.7
43.4
CPD
578.5
911.0
57.5
58
25
TCE
703.3
713.0
1.4
BUT
1382.5
1221.0
11.7
6.3
CPD
1446.3
1362.0
5.8
As expected, the results for the 25-jj.L injection had a smaller error than that for the 5-jj.L
injection. The BUT had the largest error, which was less than 12%. The mean error was
only 6.3%. Again the values for BUT and CPD were low, and the corresponding
increase in TCE, which would indicate interference, was not observed.
The three 10- |aL samples used to determine reproducibility had wider, error ranges than
expected from the 5 and 25-|0.L data. The first mixture results were similar to those
obtained for the 5-|iL mixture. The average error was for all samples 14.2%. The
amount detected for BUT and CPD, were higher. The second mixture had a much larger
error for TCE. There was an increase in concentration in BUT and CPD, indicating a
-------
problem with interference. The errors for BUT ahd CPD were much lower than the
values observed for the previous mixtures. The errors for the last 10-(^L mixture were
high. Both CPD and BUT were over-recovered, and TCE was under-recovered.
The above quantitation was performed using older Nicolet LSF software. There have
been significant improvements in CLS software, and we will reanalyze these data to
determine if the deviations are due to the software or to sample handling.
Other volatile organic studied, have had results similar to those observed for DCM.
Conclusion. We have demonstrated that volatile organics can be detected in spiked soil
samples at low ppm concentrations, and that mixtures with overlapping absorbance can
be identified and quantified. The next phase is to develop a sample desorption system
which will allow us to reproducibly desorb one gram quantities of soil. Our initial
approach using the pyroprobe cannot be scaled up to gram quantities because water of
crystallization in clay causes the soil to blow out of the sample tube into the cell. We
are currently developing' a steel tube disturber that can be' snap-connected to the cell.
Initial results are favorable, and we intend to report on this work in the future.
-------
Figure 1. Infrared Spectrum of Dichloromethane..
Figure 2. Infrared Spectra of a Mixture of DCM and Isooctane, Pure Isooctane, and DCM
after Isooctane Was Subtracted.
-------
LINEARITY OF OICHLOROMETHANE RESPONSE
igure 4. Detector Linearity as a Function of Concentration for Dichloromethane.
:igure 5. Comparison of the Spectra of a Mixture with the Pure Analytes.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 38
Data Evaluation and Risk Assessment - Some Pitfalls When
Evaluating VOC Measurements
Nic Korte and Peter Kearl
Environmental Sciences Division, Oak Ridge National Laboratory,
Grand Junction Office, Grand Junction, Colorado
January 12-14, 1993
Las Vegas, Nevada
-------
DATA VALIDATION AND RISK ASSESSMENT - SOME PITFALLS WHEN
EVALUATING VOC MEASUREMENTS
by Nic Korte and Peter Kearl
Environmental Sciences Division
Oak Ridge National Laboratory
Grand Junction Office
Grand Junction, Colorado
Data validation, as described in Environmental Protection Agency (EPA)
protocols under the Contract Laboratory Program (CLP), yields false confidence
in the data and drives up costs while providing little benefit (Korte and Brown
1992). Commonly, these data are then used to perform a risk assessment. Much
of the published guidance for risk assessments in arid soils is inadequate because
it does not take into account vapor migration due to density-driven flow (Korte
and others 1992). Investigations into both of these problems have been
performed by personnel of Oak Ridge National Laboratory (ORNL) and are
described in this presentation.
DATA VALIDATION
Present guidance under the Comprehensive Environmental Response,
Compensation, and Liability Act is that analytical results for samples collected
from sites on the National Priorities List (NPL) must be subjected to a
comprehensive data validation. Unfortunately, the performance of a
comprehensive data validation is becoming the standard by which all quality
assurance/quality control (QA/QC) programs are judged. In the minds of many,
there is a belief that if the data are not validated according to the EPA manuals,
the data are not useful scientifically. This belief extends the application of data
validation beyond that for which it was designed.
EPA Data Validation Guidelines
A review of EPA guidelines demonstrates that data validation involves
checking whether specifications for sample handling, instrument tuning and
calibration, etc., were satisfied. Specifications for the analytical program are
provided in the CLP Statements of Work that have been published by EPA.
Thus, if CLP methods are used in the project, it is easy to establish a laboratory
contract that explicitly describes the analytical specifications. Once such a contract
is in place, data validation provides little more than protection against fraud.
Notwithstanding the fact that it has occurred (Zurer 1991), fraud is the
exceptional case and does not ordinarily have to be checked. In addition.
-------
performing data validation on one or two samples would generally be sufficient to
determine the adequacy of the documentation without reviewing every sample.
There are instances, however, when a determination of the technical validity
of reported data is needed. For example, EPA data validation manuals provide
guidance for disregarding certain concentrations of commonly reported laboratory
artifacts, such as acetone and methylene chloride. Laboratory artifacts such as
these are often reported in greater concentrations than can be disregarded based
on EPA guidelines. In such instances, particularly when faced with review by
inexperienced regulatory personnel or the public, such data can cause great
difficulty. Indeed, some type of technical data validation would be useful anytime
an anomalous result is reported.
EPA recognizes the limitations of data validation and the fact that it is
overused. The following quote is taken from a soil sampling QA/QC guide (EPA
1989a), "Examination of the results of a components of variance analysis
performed on soils data from an NPL site sampled for PCBs indicates that 92% of
the total variation came from the location of-the sample, while only 8% was
introduced when the sample was taken. Less than 1% of the total could be
attributed to the analytical process itself; yet, this latter area is where a majority of
the QA resources are normally focused." The EPA study suggests therefore, that
data validation is wasteful in many instances.
Cost of Data Validation
The cost of data validation is dependent on the number of parameters
analyzed for each sample, the level of data validation, the quality of the data
package from the laboratory, and the required deliverables for the client. Seven
subcontractors estimated that the cost of data validation is generally 10% to 18%
of the analytical cost (not including QA/QC analytical cost). Additional
requirements imposed in some regions have increased the cost to at least 25%
with some laboratories (R. Tarravechia, T. A. Gleason Associates, personal
communication to N. Korte, July 199.1).
Referee Laboratory Approach
Therefore, the question remains, Is there a scientifically meaningful method
for determining when data are valid for technical interpretation? The authors
believe that using a referee laboratory is such a meaningful approach. The
approach employed by the authors is to submit duplicates or replicates of 10% of
the samples collected to a second laboratory. The existence of the referee
laboratory is made known to the primary laboratory. This fact provides significant
assurance against fraud. A laboratory is unlikely to risk altering the specifications
when another laboratory is analyzing the same sample with the same procedure.
The overall premise is that if two laboratories can agree on the results for 10% of
the samples collected, then all of the results from the program are suitable for
technical interpretation.
-------
For more than five years, 10% of all groundwater monitoring samples from a
large industrial facility with more than 200 groundwater sampling locations have
been sent to a referee laboratory.
The program steps are as follows:
1. Data from the previous quarter are reviewed.
2. Wells to be sampled the next quarter are selected.
3. The sampling team is instructed to obtain replicate samples from wells for
which anomalous data had been reported from the previous quarter. The
replicates are sent to the referee laboratory.
4. If the number of samples specified in step 3 is less than 10% of the total being
collected, the sampling team is instructed to select the remainder, up to 10%.
The track record of this program over the five-year period has been impressive.
The same referee laboratory has been used since the program began, but the
primary laboratory has changed five times. Use of the referee laboratory has
been instrumental in providing continuity from one laboratory's results to another
and for resolving problems that inevitably arise when so much data are collected.
This general technique can also be used for soil samples. Because soil samples
cannot be true duplicates, more judgement must be exercised regarding the
comparison of the results between the laboratories. On the other hand, the
referee laboratory approach still provides both protection against fraud and a
means of screening laboratory artifacts and yields information regarding the spatial
heterogeneity of the contamination.
SUMMARY
Referee analysis, when implemented on a routine basis as described in this
paper, is an effective method of obtaining data validation in a technical sense.
When two laboratories agree' on an analytical result, there is substantial reason to
believe that the result is correct. Of greater value to a large monitoring program
is the fact that a routine program of referee analysis provides a built-in
mechanism for checking anomalies. A new trend in the data or an erroneous
value can be evaluated promptly without instituting a new or extra program. The
result is a self-correcting system that rapidly verifies or negates anomalous data.
This approach increases the cost of the basic analytical program by approximately
10% but is less expensive and more technically valuable than conventional data
validation.
-------
RISK ASSESSMENTS AT ARID SITES
Commonly used risk assessment guidance may not be adequate when applied
to solvent-contaminated soils in arid environments. The calculations
recommended for determining how such soils will affect groundwater assume that
leaching controls redistribution of the contaminants. If vapor-phase migration is
considered, convection and diffusion are generally assumed to be the dominant
processes (Roy and Griffin 1990). Another view, however, is that, under certain
conditions, gravity-induced vapor migration is the most important mechanism
when determining how certain volatile organic compounds (VOCs) redistribute
themselves from soil to groundwater (Falta and others 1989).
This paper presents an examination of tjiese issues that was performed in the
course of preparing a risk assessment for an industrial site in southern California.
The results of site-specific calculations demonstrate that leaching is not a
significant process and that gravity-induced vapor migration may be the dominant
mechanism.at the study site.
Risk assessment policy for Superfund is presented in the Superfund Public
Health Evaluation Manual (EPA 1986) and in Risk Assessment Guidance for
Superfund (EPA 1989b). Both manuals reference the Superfund Exposure
Assessment Manual (EPA 1988) for determining how to model the migration of
contaminants. This latter manual states that the most significant contaminant
movement in soils is a function of liquid movement.' "Dry, soluble contaminants
dissolved in precipitation, run-on, or human-applied water will migrate through
percolation in the soil" (p. 40). The only scenario addressed in reviewing the
effects of soii on groundwater is the percolation of rainwater flowing through the
vadose zone. No other mechanisms are mentioned.
In addition, EPA's Exposure Assessment Group (EPA 1987) prepared a
preliminary draft guidance document for determining clean-up levels in soils. A
lengthy section on soils contamination dealt only with the determination of
quantity of leachate. No other mechanisms are mentioned.
This approach with regard to soil contamination is shared by California. The
California Site Mitigation Decision Tree Manual (California Department of Health
Services 1986) is used to determine an appropriate course of action for
contaminated sites. This manual notes that the movement of fluids through the
unsaturated zone is due to gravity and/or pressure gradients resulting from
capillary forces. A two-page table is provided that lists the information needed to
determine how fast contaminants can move through the vadose zone. In each
case the table notes that the purpose of the monitoring is to "estimate percolation"
or to "infer the movement of liquid-borne pollutants". None of the monitoring
tasks address movement of vapor. Likewise, a risk assessment monograph for
trichloroethene (TCE) in California concluded that "TCE in the soil can be lost
through volatilization to the atmosphere, downward leaching, or sorption to
organic material" (Bogan and others 1988).
-------
In summary, none of these guidance documents addresses vapor-phase
redistribution of contaminants. Only migration in the liquid phase is addressed.
STUDY AREA
An industrial site near Los Angeles International Airport has been
contaminated by decades of solvent and fuel usage. The vadose zone is
dominantly sand. Soil and soil vapor sampling have demonstrated that much of
the soil above the water table is contaminated. Planned remedial action using
vapor extraction will remove contamination above 65 ft. The groundwater, which
is encountered at approximately 100 ft, is contaminated with more than 100 mg/L
of chlorinated solvents. The problem for this particular site was to determine the
health and environmental risk for soil contamination located from a depth of 65 ft
down to the water table.
ESTIMATING CONTAMINANT MIGRATION IN THE VADOSE ZONE
Liquid-Phase Transport
The principal contaminants at the study site, determined as described by EPA
(EPA 1988), are TCE and 1,1-dichloroethene (1,1-DCE). A review of risk
assessment guidance for determining groundwater contamination from soils, as
presented above, suggests that liquid-phase migration is the dominant mechanism.
Based on this assumption, data from the site demonstrate that there would be no
further movement of the contaminants. This conclusion is based on the fact that
the volumetric moisture content of the soil is approximately 0.02 to 0.04 cm3/cm3.
This moisture content is well below the residual saturation value for a sandy soil,
indicating that liquid-phase leaching will not occur.
Similarly, evidence from the soil moisture profile also demonstrates that
leaching is not occurring. A plot of moisture content versus depth showed that
moisture content is highest in the fine-grained zones of the aquifer. The curve
also showed that there was no perched water or increase in moisture above the
clay zone. If water were leaching downward, some should accumulate above the
clay. The moisture content of the clay is higher because of the greater capillary
pressure exerted by the small pores.
Thus, based on the risk assessment guidance and the low moisture content of
site subsoils, it could be concluded that the soil contamination would not affect the
groundwater. Reviewing mechanisms of vapor-phase transport, however, results in
a different conclusion.
Density-Driven Flow
Recent papers by Falta and others (1989) and Mendoza and Frind (1990) have
challenged the fundamental assumption's that either leaching or diffusion dominate
vapor migration in the unsaturated zone. These papers suggest that in the
-------
absence of leaching, dense organic compounds migrate downward under the force
of gravity at a rate that is significantly greater than transport by diffusion.
For example, they reported that for a sandy soil with an intrinsic permeability
of 10"7 cm2, the flux of TCE is 3 x 10"2 g/cm2/day — a result triple that for simple
diffusion. At the Los Angeles study site, the intrinsic permeability, as measured
with an air pumping test, is 10 s cm2 — a significantly higher result than that
employed by Falta and others (1989). These results are consistent with the high
concentrations of contaminants (>100 ppm in some locations) ipund in the
groundwater. The results prove that an assumption of density-driven transport as
the primary mechanism responsible for redistribution of contaminants from the
soils to the underlying groundwater is the most conservative approach for
estimating the risk of soils contamination remaining at this site.
SUMMARY
Risk assessment guidance documents do not adequately treat solvent-
contaminated soils when estimating their effect on groundwater. In particular,'
arid sites with sandy soils contaminated by residual volumes of chlorinated solvents
are best examined by assuming that density-driven vapor transport is the primary
process affecting contamination of the groundwater. This mechanism is neglected
in commonly used risk assessment guidance. Moreover, much of the technical
literature indicates that diffusion is the primary mechanism affecting vapor
transport; however, density-driven transport may be orders-of-magnitude more
significant, and, therefore, more appropriate for use when performing a risk
assessment.
BIOGRAPHY
N. E. Korte received a B.S. in chemistry from the University of Illinois in
Champaign-Urbana and an M.S. in analytical chemistry from the University of
Arizona, Tucson, Arizona. Currently he is chemical projects manager at Oak
Ridge National laboratory (Environmental Sciences Division, Grand Junction
Office, Grand Junction, CO 81502). Korte's primary research interest is studying
the fate and effect of trace species in the environment. He is a Certified
Hazardous Materials Manager and a Certified Ground Water Professional.
P. M. Kearl received a B.S. in geology from Mesa State College in Grand
Junction, Colorado, and an M.S. in hydrology/hydrogeology from the University of
Nevada, Reno, Nevada. Currently he is a research scientist at Oak Ridge
National Laboratory (Environmental Sciences Division, Grand Junction Office,
Grand Junction, CO 81502). Kearl's primary research interest is the accurate
quantification of groundwater velocity measurements. He is a Certified Ground
Water Professional.
-------
REFERENCES
Bogan, K. T., L. C. Hall, L. Perry, R. Fish, T. E. Kone, P. Dowd, S. E. Patton, and B.
Mallon. Health Risk Assessment of Trichloroethylene (TCE) in California Drinking
Water. UCRL-21007, DE88 005364. Lawrence Livermore National Laboratory.
California Department of Health Services. 19S6. The California Site Mitigation Decision
Tree Manual.
EPA. 1986. Superfund Public Health Evaluation Manual. EPA/540/1-86/060. U.S.
Environmental Protection Agency.
EPA. 1987. Guidance for Establishing Target Cleanup Levels for Soils at Hazardous
Waste Sites. 204060, U.S. Environmental Protection Agency, Office of Health and
Environmental Assessment, Washington, D.C.
EPA. 198S. Superfund Exposure Assessment Manual. EPA/540/1-88/001. U.S.
Environmental Protection Agency.
EPA. 1989a. Soil Sampling Quality Assurance User's Guide. EPA/600/8-89/046. U.S.
Environmental Protection Agency.
EPA. 1989b. Risk Assessment Guidance for Superfund, Environmental Evaluation Manual,
Volume II. EPA/540/1-89/001. U.S. Environmental Protection Agency.
Falta, R. W., I. Javandel, K. Pruess, and P.A. Witherspoon. 1989. Density-driven
flow of gas in the unsaturated zone due to evaporation of volatile organic compounds.
Water Resources Research, 25:2159-2169.
Korte, N. E., P. M. Kearl, T. A. Gleason, and J. S. Beale. 1992. The inadequacy of
commonly used risk assessment guidance for determining whether solvent-
contaminated soils can affect groundwater at arid sites. Journal of Environmental
Science and Health, A27(8):2251-2261.
Korte, N. E., and D. E. Brown. 1992. Referee analyses — A better approach than
data validation. Environmental Management. In press.
Mendoza, C. A. and E. O. Frind. 1990. Advective-dispersive transport of dense organic
vapors in the unsaturated zone. Water Resources Research, 26:379-387.
Roy, W. R. and R. A. Griffin. 1990. Vapor-phase interactions and diffusion of organic
solvents in the unsaturated zone. Environmental Geology and Water Science, .15:101-
110.
Zurer, P.S. 1991. Contract labs charged with fraud in analyses of Superfund samples.
Chemical and Engineering News, 69(8):14-16.
-------
REFEREE ANALYSES - A BETTER APPROACH THAN DATA VALIDATION
N.E. Korte1, Environmental Sciences Division2, Oak Ridge National Laboratory3, Grand
Junction Office, Grand Junction, Colorado
D.E. Brown, Environmental Safety and Health Department, AJlied-Signal Aerospace,
Kansas City Division, Kansas City, Missouri
1 Author to whom correspondence shouia De addressed.
2 Publication No. 3S74, Environmental Science Division, ORNL.
J Managed by Martin-Marietta Energy Systems. Inc. for the U.S.
Department of Energy under Contract No. DE-AC05-S40R21400.
Accepted for publication - Environmental Management
-------
ABSTRACT
Data validation, as prescribed in guidance documents provided by the United
States Environmental Protection Agency (EPA), yields legal evidence that an analytical
laboratory has performed analyses according to pre-deter'mined specifications. This
validation process, because it involves only the checking of procedural documentation,
provides minimal information concerning the technical validity of the data. Unfortunately,
in the minds of many, the performance of such a data validation has become synonymous
with technical validity. A better approach is to implement a routine program of referee
analyses. That is, submit a pre-determined number of samples from the project to a
second laboratory. A routine program of this type provides a self-checking and self-
correcting mechanism for determining whether data are technically valid.
-------
REFEREE ANALYSES
- A BETTER APPROACH THAN DATA VALIDATION
Present guidance under the Comprehensive Environmental Response
Compensation and Liability Act (DOE 19SS; DOE 1990) is that analytical results for
samples collected from sites on the National Priorities List must be subjected to a
comprehensive data validation (EPA 19S8a; 19SSb). Unfortunately, the performance of a
comprehensive data evaluation is becoming the standard by which all Quality
Assurance/Quality Control programs are-judged. Io the mind.s of many, there is a belief
that if the data are not validated according to the EPA manuals, the data are not useful
scientifically. This belief extends the application of data validation beyond that for which it
was designed. Data validation serves to determine whether the laboratory performed their
work according to predetermined specifications. It is an expensive practice that provides
support for legal proceedings but may add little scientific value to the project. This paper
describes the present circumstances and presents a more' cost-effective and scientifically
based approach to the problem of determining whether data are technically valid.
EPA Data Validation Guidelines
Table 1 presents a brief list of the primary components of data validation as
specified in the EPA manuals. A review of Table 1 demonstrates that data validation
involves checking whether specifications for sample handling, instrument tuning and
calibration, etc. were satisfied. Specifications for the analytical program are provided in
the Contract Laboratory Program (CLP) Statements of Work (EPA 19S9a; 19S9b). Thus,
if CLP methods are used in the project, it is easy to establish a laboratory contract that
explicitly, describes the analytical specifications. Once such a contract is in place, data
validation provides little more than protection against fraud. Notwithstanding the fact that
it has occurred (Zurer 1991), fraud is the exceptional case and does not ordinarily have to
-------
be checked. Most laboratories are honest and perform their work according to their
contracts. In addition, performing data validation on one or two samples would generally
be sufficient to determine the adequacy of the documentation without reviewing every
sample.
There are instances, however, when a determination of the technical validity of
reported data is needed. For example, the EPA manuals (EPA 19SSa; 19SSb) provide
guidance for disregarding certain concentrations of commonly reported laboratory artifacts
such as acetone and methylene chloride. In the authors' experience, however, greater
concentrations of these laboratory artifacts, than can be disregarded based on EPA
guidelines, are frequently reported. In such instances, particularly when faced with review
by inexperienced regulatory personnel or the public, such data can cause great difficulty.
Indeed, some type of technical data validation would be useful anytime an anomalous
result is reported.
Another situation requiring a different approach to data validation occurs when
legal staff request validation for data previously obtained for which CLP methods were not
used. A recent example, involving one of the authors, involved a series of domestic wells
contaminated with chlorinated solvents. The support laboratory and the consultant,
realizing that the wells had been in use as drinking water supplies for a long time, had
selected EPA 500 series analytical methods for the analyses. The principal, functional
difference between the 500 series methods and the CLP methods is that the latter were
designed for sites with unknown characteristics where a large range of unknown
interferences might be present. In contrast, drinking water supplies are unlikely to contain
large quantities of unexpected interferences. The 500 series methods, therefore, have less
rigid documentation and analytical requirements, preventing the performance of the full
EPA-CLP validation. The validation that was performed in this instance consisted of
comparing the published procedural requirements to the specifications documented by the
laboratory. This limited validation proved only that the laboratory employed the analytical
methods that were requested. Until that step was completed, however, the legal staff did
not consider the data valid for technical interpretation and would not permit the
consultant to report the data ~ even though the validation process provided little
information regarding the accuracy or precision of the reported results.
-------
The preceding example demonstrates that data validation is being requested in
circumstances where it is inappropriate. Checking for fraud is an important aspect of an
enforcement case or a site where PRPs (potentially responsible parties) will litigate their
relative liability. But, under many circumstances, data validation does little more than
increase the cost of the project.
The EPA recognizes the limitations of data validation and the fact that it is
overused. The following quote is taken from a soil sampling quality assurance/quality
control guide (EPA 19S9c): "Examination of the results of a components of variance
analysis performed on soils data from an NPL site sampled for PCBs indicates that 92% of
the total variation came from the location of the sample, while only S% was introduced
when the sample was taken. Less than 1% of the total could be attributed to the
analytical process itself; yet, this latter area is where a majority of the QA resources are
normally focused." Data validation checks only the analytical process. This EPA study
suggests, therefore, that data validation is wasteful in many instances. Similarly, another
opinion offered by EPA personnel has. suggested that the increasing use of risk analyses
will demonstrate that the greatest uncertainty in data collection is the nonanalytical
aspects of the program (Fairless and Bates 1989).
Cost of Data Validation
The cost of data validation is dependent on a number of factors, including the
number of parameters analyzed for each sample, the level of data validation, the quality of
the data package from the laboratory, and the required deliverables for the client. The
Department of Energy's Hazardous Waste Remedial Action Program has estimated that 1
to 1.5 hours per sample per parameter are required for a full CLP validation
(memorandum from C. W. Stanley to M. Nickelson, Martin-Marietta Energy Systems,
Hazardous Waste Remedial Action Program, Oak Ridge, Tennessee, April 1991). Seven
subcontractors estimated that the cost of data validation is generally 10% to 1S% of the
analytical cost (not including QA/QC analytical cost). Additional requirements imposed in
some regions (EPA 1990) has increased the cost to at least 25% with some laboratories
-------
(personal communication, R. Tarravechia, T.A. Gleason Associates with N. Korte, Oak
Ridge National Laboratory, July 1991).
Referee Laboratory Approach
Therefore, the question remains: is there a scientifically meaningful meihod for
determining when data are valid for technical interpretation? The authors believe that
using a referee laboratory is such a meaningful approach. The approach employed by the
authors is to submit duplicates of 10% of the samples collected to a second laboratory.
The existence of the referee laboratory is made known to the primary laboratory. This
fact provides significant assurance against fraud. A laboratory is unlikely to risk altering
the specifications when another laboratory is analyzing the same sample with the same
procedure. The overall premise is that if two laboratories can agree on the results for
10% of the samples collected, that all of the results from the program are suitable for
technical interpretation.
For more than five years, 10% of all groundwater monitoring samples from a large
industrial facility with more than 200 groundwater sampling locations have been sent to a
referee laboratory.
The program steps are as follows:
1. Data from the previous quarter are reviewed.
2. Wells to be sampled the next quarter are selected.
3. The sampling team is instructed to obtain duplicate samples from wells for which
anomalous data had been reported from the previous quarter. The duplicates are
sent to the referee laboratory.
4. If the number of samples specified in step 3 is less than 10% of the total being
collected, the sampling team is instructed to select the remainder, up to 10%.
The track record of this program over the five year period has been impressive.
The same referee laboratory has been used since the program began but the primary
laboratory has changed five times. Use of the referee laboratory has been instrumental in
-------
providing continuity from one laboratory's results to another and for resolving problems
that inevitably arise when so much data are collected. Following are some examples that
demonstrate the advantage of this approach.
Evaluation of Laboratory Performance
Every one to two years, the contract for the primary support laboratory is awarded
to the lowest bidder that satisfies all of the specifications in the bid package. On one
occasion, results from a new support laboratory indicated significant deviations with
historical data. This is shown in Table 2 for Well 69L for which the new laboratory
reported an approximate 6-fold increase in contamination. The use of the referee samples
showed that historically reported concentrations were correct and that the results from the
new laboratory (160,000 /ig/L) were erroneous. A conventional EPA data validation
would not have uncovered any problem with the new laboratory. A frequent lack of
agreement with historical data persisted and was clearly documented by the referee
analyses. Because this was a procurement of a government agency, terminating a contract
would have been difficult and-is typically met with great resistance by procurement
officers. In this case, the documentation from the referee analyses enabled a rapid
resolution to the problem with neither procurement officials nor the laboratory seriously
contesting the Endings. A new contract was let for another primary laboratory and the
ensuing results agreed with the historical data.
Rapid Identification of New Data Trends
The most beneficial routine usefulness of referee analyses is for evaluating the
accuracy of sudden changes in data trends. Table 2 demonstrates how this is
accomplished. Well 102U was not expected to contain trichloroethene or dichloroethene
contamination when 40-50 ^g/ml were reported. Other monitoring wells between this well
and potential source areas were clean, as expected. Because of this anomalous data, a
referee sample was requested for the next quarter. Both laboratories agreed that the data
were correct. Similarly, the sudden decrease in trichloroethene and 1,1-dichloroethene in
-------
Well 106 was unexpected because the well is located in a central portion of a contaminant
plume. In this case the referee analysis, in conjunction with the new analysis by the
primary laboratory, pro%i'ded evidence that the decreases reported the previous quarter
signified an actual change in the data trends for that well.
SUMMARY
Referee analysis, when implemented on a routine basis as described in this paper,
is-an effective method of obtaining data validation in a technical sense. When two
laboratories agree on an analytical result, there is substantial reason to believe that the
result is correct. Of greater value to a large monitoring program is the fact that a routine
program of referee analysis provides a built-in mechanism for checking anomalies. A new
trend in the data or an erroneous value can be evaluated promptly without instituting a
new or extra program. The result is a self-correcting system that rapidly verifies or
negates anomalous data. This approach increases the cost of the basic analytical program
approximately ten percent but is less expensive and more technically valuable than
conventional data validation.
-------
Table 1. Primary Components of Data Validation
Volatile &
Semivolatile
Component of Data Validation Organics1 Pesticides1 Inorganics2
holding times
GC/MS tuning
calibration
blanks
surrogate recovery
matrix spike/matrix spike duplicate
field duplicates
internal standards performance
TCL compound identification
compound quantitation and reported
detection limits
tentatively identified compounds
system performance
overall assessment of data for a case
pesticides instrument performance
compound identification
ICP interference check sample (ICS)
laboratory control sample (LCS)
duplicate sample analysis
furnace atomic absorption QC
ICP serial dilution
sample result verification
XXX
X
XXX
XXX
X X
XXX
X X X
X
X
X X
X
X
XXX
X
X
X
X
X
X
X
X
'EPA. 19SSa.
2EPA. 19SSb.
GCA'IS: gas chromatograph/mass spectrometer.
ICP: inductively coupled plasma.
QC: quality control.
-------
Table 2. Examples of the Use of Referee Analyses
Well
number
Historical average"
TCE DCE
/vg/L /vg/L
Questionable primary
lab result
TCE DCE
/ig/L /vg/L
Primary lab result - next
sampling period
TCE DCE
VfJL //g/L
Referee lab result -
next sampling period
TCE DCE
/ig/L /ig/L
1. Detecting Analytical Errors
69 L
22,600 12,750
160,000 42,000
43,000"
39,000
26,000 24,000
2. Confirmation of Contamination
102U
New well in an area
presumed uncontam-
inatcd
20 25
18
22
16 19
3
. Confirmation of Change of Contaminant Trend
106
29 161
15 10
13
62
13 59
' Based on at least 4 successive, previous analyses.
For each case presented, represents results from a different primary laboratory.
TCli = tiichloroelhene
DCE = 1,2-dicliloroclhcne
-------
References
DOE. 19S8. Hazardous Waste Remedial Actions Program Requirements for Quality
Control of Analytical Data, HZ/RAP-102-1, U.S. Department of Energy.
DOE. 1990. Requirements for Quality Control of Analytical Data, DOE/HWP-65/R1,
Hazardous Waste Remedial Actions Program, U.S. Department of Energy.
EPA. 198Sa. Laboratory Data Validation Functional Guidelines for Evaluating Organics
Analyses, U.S. Environmental Protection Agency, Washington, D.C.
EPA. 198Sb. Laboratory Data Validation Functional Guideline for Evaluating Inorganics
Analyses, U.S. Environmental Protection Agency, Washington, D.C.
EPA 1989a. USEPA Contract Laboratory Program Statement of Work for Organics
Analysis, U.S. Environmental Protection Agency, Washington, D.C.
EPA 19S9b. USEPA Contract Laboratory Program Statement of Work for Inorganics
Analysis, U.S. Environmental Protection Agency, Washington, D.C.
EPA 19S9c. Soil Sampling Quality Assurance User's Guide, EPA/600/S-89/046, U.S.
Environmental Protection Agency, Washington, D.C.
EPA 1990. Laboratory Documentation Requirements for Data Validation, 9QA-07-90,
U.S. Environmental Protection Agency, San Francisco, Calif.
Fairless, B.J. and D.I. Bates. 19S9. Estimation of the Quality of Environmental Data,
Pollution Engineering, March, p. 108-111.
Zurer, P.S. 1991. Contract Labs Charged with Fraud in Analyses of Superfund Samples,
Chemical and Engineering News, Washington, D.C., February, p. 14-16.
-------
Part A
Environmental Science
and
Engineering
including
Toxic and Hazardous Substances Control
Part B
Pesticides, Food Contaminants
and
Agricultural Waste
Part C
Environmental Carcinogenesis Reviews
-------
.'curnc! of Eir,i.'OMMniol Science ojoaf Health
Part A
ENVIRONMENTAL science and engineering
[nd'jJing, Tosfc a.«d Hu ai Jous Subilsnces Control
ExKu:5ve. Editor
JaMES w.roeinsom
Department of Chemistry
Louisiana Stcte University
Baton Rouge. Louisiana 70803
P»Tt B
PESTICIDES, FOOD CONTAMINANTS,
AND AGRICULTURAL WASTES
Editci: SH A.KAMUT D. KHAN
Land Resource Centre
-'JejfarrA Branch. Apiculture Caned?, X. W. Neaiby Building
Oi w n;sr, Qr,tano, Citadi #L"jX Ki
Pan C
ENVIRONMENTAL CARCINOGENESIS REVIEWS
Editors
JOSEPH C_ aRCCS
MARVF ARGUS
YIK-lAX woo
P.O. Box 2*25
¦Tp-ri.->-;?(•* J1 f-rri-L" a 22J 52
For Subscription information wrire jo;
Promotion. Department
Marcel DeJJter. lric,
^DMadiKiH A«-et jc
Neu-Ycrl, N.Y ]D0I5
-------
J. ENVIRON. SCI. HEALTH, A27(8), 2251-2261 (1992)
The Inadequacy of Commonly Used Risk
Assessment Guidance for Determining
Whether Solvent-Contaminated Soils Can
Affect Groundwater at Arid Sites*
N. E. Korte P. M. Kearl
¦Oak Ridge National Laboratory^
Grand Junction Office
Grand Junction, CO 81502
J. A. Gleason
TA Gleason and Associates
Cincinnati, OH 45226
J. S. Beale
Allied-Signal Aerospace Company
Torrance, CA 90509
Abstract
Commonly used risk assessment guidance is not adequate when ap-
plied to solveni-comaminated soils in arid environments. The equations
that are recommended for calculating how such soils will affect ground-
water assume thst liquid phase leachingcontrols contaminant migration.
If vapor phase migration is considered at all, diffusion is assumed to be
"Publicaiion No. 38^8, Environmental Sciences Division, ORNL-
t Managed by Martin-Marietta Energy Systems, Inc. for the U.S. Department of Energy under
Co.uraci No DE-ACOJ-S40R2140Q.
2251
Copyright © 1992 by Marcel Dekker, Inc.
-------
2252
KORTE ET AL.
the dominant process. In contrast, a field study performed at an indus-
trial site in Southern California demonstrated that leaching could not
account for the transport of contaminants to the water table and the re-
cent technical literature suggests that gravity-induced vapor migration
may be the principal mechanism for vapor phase migration. Thus, regu-
latory guidance provided by the United States Environmental Protection
Agency and the Slate of California could result in a significant underes-
timate of the amount of chlorinated solvent that could contaminate the
groundwater at arid sites.
Introduction
Commonly used risk assessment guidance may not be adequate when applied
to solvent-contaminated soils in arid environments. The calculations recom-
mended for determining how such soils will affect groundwater assume that
leaching controls redistribution of the contaminants. Convection and diffu-
sion are generally assumed to be the dominant processes in vapor phase mi-
gration [1], Under certain conditions, however, gravity-induced vapor migra-
tion is the most important mechanism when determining how certain volatile
organic compounds redistribute themselves from soil to groundwater [2].
This paper presents an examination of these issues that was performed in
the course of preparing a risk assessment for an industrial site in Southern
California. The results of site-specific calculations demonstrate that leaching
is not a significant process and that gravity-induced vapor migration may be
the dominant mechanism at the study site. Indeed, it appears that regula-
tory guidance provided by both the United States Environmental Protection
Agency and the State of California could result in significant underestimates
of the amount of chlorinated solvent that could contaminate the groundwater
at certain arid sites.
Review of Existing Guidance
Risk assessments are performed by using existing data and assumptions based
on site conditions to calculate exposures to individuals and the environment.
The typical risk assessment approach is to assume worst-case conditions.
Such conservatism is appropriate because risk assessments are often used as
a justification for a less conservative remedial action, possibly no action. Sev-
eral guidance documents discuss the calculations for determining the effect
of soils on groundwater.
-------
SOLVENT-CONTAMINATED SOILS
2253
3n 19B5, in a nine-volume repon, the United States Environmental Pro-
tection Agency [3] described computer models that could be used 10 deier-
mine the effect of soil contamination on groundwater. That report, though it
mentioned ihe pariiiionirig of contaminants in soil vapor, described leachate
production as the means by which soil contaminants reached groundwater
from rtie vadose zone. Subsequently, EPA provided more specific guidance
for the performance of risk assessments under the Comprehensive Environ-
mental Response, Compensation, and Liability Act (CERCLA or Super-
fund).
Risk assessment policy for Superfvnd is presented in the Superfund Pub-
lic Health'Evaluation Manual |4] and in Risk Assessment Guidance for Super-
fund [5], Both of these manuals reference the. Superfund Exposure Assessment
A/cnua/[6]fordetermininghowia model the migration of contaminants. This
manual states that the kioji significant contaminant movement in soils is a
function of liquid movement. "Dry, soluble contaminants dissolved in pre-
cipitation, mn-cn, or human-applied water will migrate through percolation
in the soil" [6], The only scenario addressed in reviewing the effects of soil on
groundwater is the percolation of rainwater flowing through the vadose zone.
>'oortiet mechanisms are mentioned.
Ir. adJiikin, EPA's Exposure AMHsnnen: Group |7] prepared a prelim-
inary draft guidance document for determining cleanup levels in soils. A
lengthy section on soils contamination dealt only with the determination of
the quantity of leachate. No other mechanisms are mentioned.
This approach with regard to soil con lamination is shared by California -
a state with a vast amount of arid land affected by chlorinated solvents. The
California Siie Mitigation Decision Tree Manual [8] is used to determine an
appropriate course of action for contaminated sites. This manuai notes that
the movement of fluids through the unsaturated zone is due to gravity and/or
pressure gradients resulting from capillary forces. A two-page table is pro-
vided that lists the information needed to determine how fast contaminants
can move through the vadose zone. In each case the table notes that the pur-
pose of the monitoring is to "estimate percolation" or to "infer the movement
of liquid-borne pollutants". None of the monitoring tasks address movement
of vapor. Likewise, a risk assessment monograph for trichloroethene (TCE)
in California concluded that "TCE Ln the soil canbe lost through volatilization
to the atmosphere, downward leachirig. or sorption to organic materia]" [9],
Jn summary, none of rhe guidance addresses vapor phase redistribution
of contaminants. Only migration in the liquid phase is addressed.
-------
2254
KORTE ET AL.
GROUND SURFACE
GEOLOGIC UNITS
OLDER OUNE Sand
UpDer C'oyey-Silt
Upper Sond
M.odie Cloy
Lo»er Sond
L&KCWGOD FORMATION
K'.cnhcUcn Seech Cloy
Coce Aquifer
El Stgundo Aquilcrd
SAM FCORO rCR^iTlON
£ii-.-erodo Aquifer
Figure 1: Lithology of the study area.
-------
SOLVENT-CONTAMINATED SOILS
2255
Study Area
The study area is an industrial sire near Los Angeles International Airport.
Soils in the unsaturated zone have been contaminated by decades of solvent
and fuel usage. As shown in Figure lf thevadose zone is domir.amlysand. Soil
and soi] vapor sampling have demonstrated that much of the soil above the
waier table is contaminated. Planned remedial action using vapcr extraction
will remove contamination above 65 ft. The groundwater, a: approximately
100-ft depth, is contaminated with more than 100 mg/L of chlorinated sol-
vents. The problem for this particular site was to determine the health and
environmental risk for soi] contamination located from a depih of 65 ft down
to the water table. Only the incremental effect of these soili on ihe ground-
water needed to be determined. Air pollution and dema! effects could be
neglected because of the depth of the contamination.
Estimating Contaminant Migration in the Yadose
Zone
Liquid Phase Transport
The principal contaminants at the study site, determined as described by
EPA [4], are trichloroethene (TCE) and 1,1-dichloroethene (1,1-DCE). A
review o? risk assessment guidance for determining groundwater contamina-
tion from soils, as presented above, suggests ihat liquid phase migration is the
dominant mechanism. Eased on this assuxnpiicn, data from the she demon-
siiates that there would be no further movement of the contaminants. This
conclusion is based on the fact that ihevolumetric moisture content of the soil
is approximately 0.02-0.04 cm3/cm3. This moisture content is well below the
residual saturation value for a sand}'soil indicating that liquid phase leaching
will not occur.
Evidence from the soil moisture profile similarly demonstrates that leach-
ing is not occurring. A plot of moisture content versus depth (Figure 2) shows
that the moisture content is highest in the fine-grained zones of the aquifer.
The curve also shows that there is no perched water or increase in moisiure
above the clay zone. If water \*-as leaching downward, seme should have ac-
cumulate d above the clay. The moisture comeni af the clay is higher because
of the greater capillary pressure exerted by the small pores.
Thus, based on the risk assessment guidance and the low moisture con-
tent of site subsoils, it could be concluded that the soil contamination would
-------
2256
KORTE ET AL.
UTKXOGY DESCRIPTION
SM S'LlY San'O: tc medium groined
wiih scotlercc g'cveJ.
CL SANOY CLAY: 5'«gMly 5'. o':\«
g'oding to yellowish brown 01 iz f:
s:c:te?e
-------
SOLVENT-CONTAMINATED SOILS
2257
not affect the groundwater. Reviewing mechanisms of vapor phase transport,
however, points to a different conclusion.
Vapor-Phase Transport
Convection
Convective transport, which is induced by temperature and pressure gradi-
ents, is not a factor when considering contamination at a depth of 65 ft. Weeks
and others (10] concluded that the effects of temperature and pressure gradi-
ents are only important in the first meter or two of soil. Likewise, Kearl and
others [11] showed that surface induced temperature gradients were negligi-
ble 20 ft below the land surface. The latter study also showed that pressure
gradients would not have an effect deep in the subsurface.
Diffusion
For one-dimensional transport, the concentration distribution of an organic
vapor in a porous media at time t can be estimated using the following equa-
tion [12]:
(1)
where
C is the concentration of vapor at a distance z from the source,
Co is the initial concentration of vapor at the source (saturated vapor con-
centration),
z is the distance from the source,
D is the diffusion coefficient of the organic vapor in air,
t is time, and
erfc is the complementary error function.
Using trichloroethene as a sample compound and a diffusion coefficient
of 0.039 cm2;s [13], the concentration gradient (dC/dz) is calculated. In order
to calculate the flux of TCE migrating from the source area, Fick's first law is
used to calculate diffusive transport:
-------
2258
KORTE ET AL.
F = D
(2)
where F is the flux of vapor transported by diffusion.
The rate of transport calculated by this method (i.e., the rate at a distance
1 m from a saturated vapor source) is approximately 10~2 g/cm2/day. This
value can be compared to the result for density-driven flow calculated in the
next section, especially in light of the fact that the groundwater contamination
exceeds hundreds of parts per million.
Density-Driven Flow
Recent papers by Falta and others [2] and Mendoza and Frind [14] have chal-
lenged the fundamental assumptions that either leaching or diffusion domi-
nate vapor migration in the unsaturated zone. These papers suggest that in
the absence of leaching, dense organic compounds migrate downward under
the force of gravity at a rate that is significantly greater than transport by dif-
fusion.
Falta and others [2] present an equation for calculating the downward flux
of dense gases in air as follows:
where
Vd is Darcian Velocity,
kkrg is the gas phase relative permeability,
g is the gravitational constant,
is the saturated vapor pressure of the liquid phase,
fa is the gas viscosity,
R is the universal gas constant,
T is the absolute temperature,
M is the molecular weight of the compound under study, and
Mair is the molecular weight of the air mixture.
Falta and others [2] reported that for a sandy soil with zn intrinsic per-
meability of 10-7 cm'2, the flux of TCE is 3 x 10-7 g/cm2/day, a result triple
that for simple diffusion. At the study site, the intrinsic permeability, as mea-
kkrtgP°(M - Mgjr)
fa RT
(3)
-------
SGLVENT-C0NTAM3NATED SOILS
2259
5jred with an air pjmprig lcot,L; ]G~5 cm!,j S]gnificariLy higher result than
that employed by Fah.i and others £2}. Jbtte rashes atc consistent vilti the
high concentrations of con Lam in ants {greater ihait 103 pprri in some loca-
tions] found in :3ie groundwater. The lesulrs prow :hzi ar» assumption of
density-driven transport, as ihe primary mechanism responsible for redistri-
bution of com aim in ants from the soils :o the underlying groundwater, is the
most conservative approach for estimating iheris!vrc = idua3vc u rr es of c h I o - _na ted
-------
2260
KORTE ET AL.
solvents are best treated by assuming that density-driven vapor transport is
the primary process affecting contamination of the groundwater. This mech-
anism is neglected in commonly used risk assessment guidance. Moreover,
much of the technical literature indicates that diffusion is the primary mech-
anism affecting vapor transport, but density-driven transport may be orders-
of-magnitude more significant, and therefore more appropriate for use when
performing a risk assessment.
Notation
CERCLA (Superfund) Comprehensive Environmental Response, Com-
pensation, and Liability Act
TCE trichloroethene
1,1-DCE 1,1-dichloroethene
References
[1] W.R. Roy and R.A. Griffin. Vapor-phase interactions and diffusion of
organic solvents in the unsaturated zone. Environmental Geology and
Water Science, 15:101-110,1990.
[2] R.W Falta, I. Javandel, K. Pruess.and P.A. Witherspoon. Density-driven
flow of gas in the unsaturated zone due to evaporation of volatile organic
compounds. Water Resources Research, 25:2159-2169,1989.
[3] Environmental Protection Agency. Methods for assessing exposure to
chemical substances. In Methods for Assessing Exposure from Disposal of
Chemical Substances. Volume 3, EPA 560/5-85-003, 1985.
[4] Environmental Protection Agency. Superfund Public Health Evaluation
Manual. EPA/540/1-86/060, 1986.
[5j Environmental Protection Agency. Risk assessment guidance for Su-
perfund. In Environmental Evaluation Manual. Volume II, EPA/540/1-
89/001, 1989.
[6] Environmental Protection Agency, Superfund. Exposure Assessment
Manual. EPA/540/1-88/001, 1988.
[7] Environmental Protection Agency. Guidance for Establishing Target
Cleanup Levels for Soils at Hazardous Waste Sites. 204060, Washing-
ton, DC, 19S7. Preliminary Draft prepared by the Exposure Assessment
Group, Office of Health and Environmental Assessment.
-------
SOLVENT-CONTAMINATED SOILS
2261
[8] The California Site Mitigation Decision Tree Manual. California Depart-
ment of Health Services, 1986.
[9] K.T Bogan, L.C. Hall, L. Perry, R. Fish, T.E. Kone, P. Dowd, S. E. Pat-
ton, and B. Mallon. Health RiskAssessmentofTrichlorocihylene (TCE) in
California Drinking Water. UCRL-21007, DE8S 005364, Lawrence Liv-
ermore National Laboratory, 1988. Prepared for the California Public
Health Foundation.
[10] E.P. Weeks, D.E. Earp, and G.M. Thompson. Use of atmospheric fluo-
rocarbons F-ll and F-112 to determine the diffusion parameters of the
unsaturated zone in southern high plains of Texas. Water Resources Re-
search, 18:1365 1378, 1982.
[11] P.M. Kearl, J.J. Dexter, and M. Kautsky. Vadose Zone Characterization
of Technical Area 54, Waste Disposal Areas G and L, Los Alamos Na-
tional Laboratory, New Mexico, Report 3: Preliminary Assessment of the
Hydrologic System. DOE/GJ-44,Bendix Fieid Engineering Corporation,
Grand Junction, CO, 19S6.
[ 12} R.A. Freeze and J.A. Cherry. Groundwater, chapter Groundwater Con-
tamination. Prentice-Hall, Englewood Cliffs, NJ, 1979.
[13] W.J. Lyman, WF. Reehl, and D.H. Rosenblat, editors. Handbook of
Chemical Property Estimation Methods - Environmental Behavior of Or-
ganic Compounds. McGraw-Hill, Toronto, Canada, 1982.
[14] C.A. Mendoza and E.O. Frind. Advective-dispersive transport of dense
organic vapors in the unsaturated zone. Water Resources Research,
26:379-387,1990.
Received: October 23,1991
Accepted: January 29,1992
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 39
Vapor Fortification: A Method to Prepare Quality Assurance Soil
Samples for the Analysis of Volatile Organic Compounds
Alan D. Hewitt
U.S. Army Cold Regions Research and Engineering Laboratory,
Hanover, New Hampshire
January 12-14, 1993
Las Vegas, Nevada
-------
VAPOR FORTIFICATION: A METHOD TO PREPARE
QUALITY ASSURANCE SOIL SAMPLES FOR THE ANALYSIS OF
VOLATILE ORGANIC COMPOUNDS
Alan D. Hewitt
U.S. Army Cold Regions Research and Engineering Laboratory
72 Lyme Road, Hanover, N.H. 03755r1290, (603) 646-4388
The wide use, and subsequent improper disposal of volatile organic compounds
(VOCs) has made this group of chemicals our most common environmental hazardous
waste problem.1 Even so, there is no readily available source of quality assurance (QA)
VOC standards for soils assessed during site investigations.2 Attempts to spike, ho-
mogenize and transfer soils have proven unsatisfactory due to the inability to control
volatilization losses.3 Currently, estimation of the accuracy of VOC soil sample analy-
sis relies on the results from sample spike and recovery tests. One common practice is
to add dilute methanol (MeOH) solutions of analytes to an aqueous suspension of the
soil in the purge vessel of a purge-and-trap system. This method of evaluating perfor-
mance focuses only the determinative step, allows no time for analyte/soil interaction
to occur, and does not simulate the manner in which soils become contaminated in the
field.
Vapor fortification offers an alternative means of treating test soils with VOCs for
performance evaluation studies.4,5 This laboratory treatment method of spiking is
analogous to the process by which vadose zone soils are contaminated by vapors from
liquid pools of organic solutions, and avoids introducing either large quantities of
water or MeOH to the test matrix. Moreover, soil or sediment matrices prepared by
vapor fortification can be evaluated for both extraction efficiency and determinative
accuracy, and thus provide a more comprehensive means of assessing a laboratory's
capability.
Recently, the initial phase of a feasibility study to evaluate vapor fortification as a
means of preparing VOC-contaminated soils was performed.6 Soil fortification was
accomplished by exposing individually prepared soil subsamples, in a closed desicca-
tor, to the equilibrium vapor from a solution of the analytes of interest in an organic
solvent.4,5 During exposure both the test matrix and fortification solution were con-
tained in separate open vessels. In this study, trans-1,2-dichloroethylene (TDCE),
trichloroethylene (TCE), benzene (Ben), and toluene (Tol) were the test analytes
studied.5
The objective of this study was to assess the various parameters, applications and
handling protocols requiring consideration when choosing optimal guidelines for the
-------
vapor fortification of soils. Overall the results showed that when vapor fortification
exposure solutions contained tetraethylene glycol dimethyl ether (tetraglyme) in
addition to MeOH, analyte sorption reached maximum concentrations after relatively
short treatment periods (4—5 days). Concentrations on the soil showed no significant
change thereafter for an extended treatment period (>35 days), and were not signifi-
cantly influenced by small temperature fluctuations.6 Additionally, the potential
influence of laboratory relative humidity was removed by conditioning an air dried,
sieved, and thoroughly mixed soil by CaS04 desiccation.
The procedure chosen resulted in both predictable and stable concentrations (for at
least 14 days) covering a concentration range that includes the action levels used for
regulatory purposes (0.1-10 |lg/g). To determine precision among and within identical
treatments, 2-g quantities of a U.S. Army Toxic and Hazardous Materials Agency
(USATHAMA) standard soil was used, and analysis was by headspace gas chromatog-
raphy (HS/GC).5 Average analyte concentrations for triplicate test samples were not
significantly different among three separate identical fortification treatments, and the
relative standard deviations within treatments was <9% for the three least volatile
analytes (Table 1). Packaging of the performance evaluation soil standard for distribu-
tion or mass production was not addressed.
The second phase of the feasibility study involves designing ways to reliably mass
produce and distribute these performance evaluation standards, and to establish their
shelf life. Initially, heat sealable 1-mL pre-scored glass ampules will be tested as
exposure vessels. The following general guidelines were used to prepare a preliminary
batch of soil QA standards in this sealable'vessel.
Table 1. Concentrations (|ig/g) established for three separate
vapor fortifications of soil contained in aluminum foil cups.
Compound ABC
TDCE
5.87±1.34*
(22.9%)t
5.77±1.29
(22.3%)
5.84±0.83
(14.2%)
Ben
6.83±0.61
(8.90%)
7.06±0.48
(6.78%)
6.96±0.44
(6.35%)
TCE
8.17+0.66
(8.10%)
8.66+0.40
(4.62%)
8.41+0.61
(7.31%)
Tol
9.82±0.44
(4.47%)
10.5±0.21
(1.98%)
10.2±0.34
(3.31%)
* mean and standard deviation
t relative standard deviation
-------
Soil Preparation
The test soil was air dried, passed through a 30 mesh screen, mixed thor-
oughly, desiccated with CaS04 by placing 2.00±0.01 g subsamples into 1.0-mL
glass ampules and transferring to the treatment desiccator.
Fortification Exposure
Vessels containing soil subsamples were exposed in a closed desiccator to
the equilibrium vapor above a 50-mL 50:50 solution of MeOH (100-mL stock
with approximately 0.490 g TDCE, 0.369 g Ben, 0.597 g TCE, and 0.600 g-
Tol) and tetraglyme, with the analytes of interest being minor constituents in
the former solvent.
Collection
After exposure the chamber was opened, a 5-mm-diam. glass ball was
placed in the opening of each ampule, and the vials were rapidly heat sealed
with a propane torch.
To assess the precision of this procedure, the
same parameters were used as for the fortification re-
sults reported in Table 1. Analysis of fortified soil
sealed in the glass ampules was performed 1 day after
heat sealing. Ampules were placed in 40-mL VOA
vials containing 30 mL of water and were opened
(broken) by vigorously shaking the sealed vials by
hand. To facilitate the opening and dispersion of the
contents, the ampules were inverted; thus the weakest
point (sealed tapered end of ampule) was subjected to
greatest impact upon striking the glass VOA vial's
bottom. Due to material strengths, only high quality
VOA vials (I-Chem) should be used, and gloves
should be worn while breaking the ampule and dis-
persing the soil. Analyte concentrations for triplicate
test samples analyzed by HS/GC are shown in Table
2. These results suggest that fortification precision
may be further improved by using glass ampules.
Acknowledgments
Funding for this work was provided by the U.S. Army Toxic and Hazardous
Materials Agency, Marty Stutz, Project Monitor. The author thanks Dr. T.F. Jenkins
and J.H. Cragin for critical review of the text.
This publication reflects the views of the author and does not suggest or reflect
policy, practices, programs, or doctrine of the U.S. Army or of the Government of the
United States.
Table 2. Concentrations (|ig/g)
established for preliminary
analysis of soils treated in
sealable glass ampules.
Compound
TDCE 7.97±0.223*
(2.79%)t
Ben 8.83±0.250
(2.83%)
TCE 10.3±0.306
(2.97%)
Tol 12.2±0.379
(3.11%)
* mean and standard deviation
t relative standard deviation
-------
References
1. Plumb, R.H., Jr. and A.M. Pitchford (1985) Presented at the National Water Well
Association/American Petroleum Institute Conference on Petroleum Hydrocarbons and
Organic Chemicals in Ground Water, November 13-15, Houston, Texas.
2. Zarrabi, K. A.J. Cross-Smiecinski and T. Starks (1991) In: Second International
Symposium, Field Screening Methods for Hazardous Waste and Toxic Chemicals,
February 12—14, Las Vegas, Nevada, p. 235-252.
3. Maskarinec, M.P., L.H. Johnson and C.K. Bayne (1989) Journal for the Association of
Official Analytical Chemists, 72: 823-827.
4. Jenkins, T.F. and P.W. Schumacher (1987) USA Cold Regions Research and Engi-
neering Laboratory, Special Report 87-22, Hanover, N.H.
5. Hewitt, A.D., P.H. Miyares, D.C. Leggett and T.F. Jenkins (1992) Environmental
Science and Technology, 26:1932.
6. Hewitt, A.D. (in press) USA Cold Regions Research and Engineering Laboratory,
Special Report, Hanover, N.H.
-------
BIOGRAPHICAL SYNOPSIS
Alan D. Hewitt is a research physical scientist at the Cold Regions Research and
Engineering Laboratory (CRREL) (72 Lyme Rd., Hanover, NH 03755-1290). He has a
B.A. in chemistry from the University of New Hampshire and a M.S. in chemical
oceanography from the University of Connecticut. Present areas of interest are
identification of hazardous waste in environmental samples, ground water monitoring,
and the chemistry of snow and ice.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 40
Field Observations of Variability of Soil Gas Measurements
J. D. Fancher
Westinghouse Hanford Company, Richland, Washington
January 12-14, 1993
Las Vegas, Nevada
-------
WHC-SA-1763-A
Field Observations of
Variability of Soil Gas
Measurements
J. D. Faneher
Dale Published
December 1-992 ¦
To Be Presented at the
Naiional Symposium on Measuring and
Interpreting VOCs in Soils: State of the
Art and Research Needs
Las Vegas, Nevada
January 12-14,19S3
Prepared for the U.S. Department of Energy
Office of Environmental Restoration
and Waste Management
rally Richland, Washington 99352
Hanford Operations and Engineering Contractor for the
U.S. Department of Energy under Contract DE-ACOS-S7RL10930
Copyright License By acceptance cl this artids-, th&putthVi& aftdrtiT iecipieni acAJiwtesi^&s 'J.S.Cov^wv^-I'e right -/i
ie;ain a nsnaydjsiva. ipyaity-1ree license jh and ic a.iycopyi"sgt"J covering th'is paper.
P.O. Box 1970
Approved for Public Release
-------
LEGAL DISCLAIMER
This report was prepared as an account of work sponsored by
an agency of Ihe United States Government. Neither the
United States Government nor any agency thereof, nor any of
their employees, nor any of their contractors, subcontractors
or their employees, makes any warranty, express or implied,
or assumes any legal liability or responsibility for the
accuracy, completeness, or any third party's use or the results
of such use of any information, apparatus, product, or process
disclosed, or represents that its use would not infringe
privately owned rights. Reference herein to an'y specific
commercial product, process, or service by trade name,
trademark, manufacturer, or otherwise, does not necessarily
constitute or imply its endorsement, recommendation, or
favoring by the United States Government or any agency
thereof or its contractors or subcontractors. The views and
opinions of authors expressed herein do not necessarily state
or reflect those of the United States Government or any
agency thereof.
This report has been reproduced from the best available copy.
Printed in the United States of America
DISCLM-2.CHP (1-91)
-------
WHC-SA-1763-A
FIELD OBSERVATIONS OF VARIABILITY OF SOIL GAS MEASUREMENTS
J. D. Fancher
Westinghouse Hanford Company
P. 0. Box 1970 MS IN N3-05
Richland, Washington 99352
INTRODUCTION
A baseline monitoring survey is being performed at the U. S.
Department of Energy's Hanford Site located in southeast Washington
State. Monitoring is in support of the carbon tetrachloride Expedited
Response Action (ERA) vapor extraction system (VES) operations. Since
late 1991, soil-gas probes and wellheads have been routinely monitored
for volatile organic concentrations. The. monitoring network now "
encompasses 59 locations. These include 46 wellhead locations,
11 shallow soil-gas probes [1.2 m (4 ft) deep], and 2 deep soil-gas
probes [11 and 22 m (37 and 73 ft) deep].
The project site is an area where carbon tetrachloride (£C1J and
co-contaminants were discharged to the soil between 1955 and 1973.
There are three separate CC14 disposal areas at the project site. This
contamination is linked to past liquid waste disposal practices
resulting from operation of the Plutonium Finishing Plant at the Hanford
Site.' The contamination caused an extensive vapor plume in the vadose
zone and a groundwater contamination plume that covers over 12 km2
The following are the objectives of this baseline monitoring
survey: 1) to measure the existing concentrations of CC1, in the
subsurface prior to initiation of the vacuum extraction; 2) to
investigate how the existing concentrations of CC1A vary with time; 3)
to evaluate the impact of vapor extraction on the distribution and
concentrations of CC14 in the subsurface; and 4) to provide data to help
maintain a safe working environment.
METHODS
Baseline monitoring is performed at the Environmental Protection
Agency analytical level I. Field screening is conducted using an
Organic Vapor Monitor (OVM) outfitted with an 11.8 eV limp. Discrete
samples are collected twice a week and the data is entered on a data
sheet and logged internally within the instrument for later retrieval.
One round of sampling usually takes less than four hours to complete.
-------
WHC-SA-1763-A
All monitoring instruments are calibrated daily. Isobutylene
calibration standards are certified and traceable to a national or
industry recognized standard. For consistency in monitoring with the
OVM, the response factor was set at 1.0.
To begin the sampling routine, the sampler monitors around the
edge of the well cap, then lifts the well cap and inserts the OVM probe
into the well. The sampler then draws air through the instrument while
monitoring the real time readings; this continues while the readings
increase. Once the reading has reached a maximum peak, the maximum
value is recorded on a data sheet and internally within the instrument.
At each station, air is drawn through the instrument for at least ten
seconds.
If the sample reading is not reproducible or sample readings vary
for no apparent reason, th& instrument is challenged with, calibration
gas. If the challenge is not within 10% of the original calibration,
the instrument is recalibrated. At the end of each day's sampling the
instrument is challenged with a calibration standard.
RESULTS'
Results of the monitoring indicate concentrations of volatile
organic compounds (VOC) vary widely over time and space. Previous
analysis with a gas chromatograph indicates that the-.majority of VOCs
present in the wellheads and soil-gas probes is CC14. During the course
of the monitoring, sample readings with an OVM reached a wellhead high
of 10,704 ppm, a shallow soil-gas probe high of 97.4 ppm, and a deep
soil-gas probe high of 10,400 ppm. The assumption is that the VOCs
measured with the OVM are CC1A.
SHALLOW SOIL-GAS PROBE DATA
The shallow probe data is fairly consistent and uniformly lower
than the wellhead data. Of the 11 stations routinely monitored, the
highest average VOC value was 7.3 ppm. The lowest average value was 1.5
ppm. The maximum probe value detected was 97.4 ppm. Results from
shallow soil gas probes show less variability than data from well head
measurements. This is caused primarily to the construction and
installation method used. The shallow soil gas probes were installed in
a uniform method, to a uniform depth of 1.2 m (4 ft).- The probes range
in distance from 1 to 58 m (3 to 190 ft) to a CC14 disposal site.
-------
WHC-SA-1763-A
DEEP SOIL-GAS PROBE DATA
Two deep soil-gas probes were emplaced at the 216-Z-9 Trench area.
The deepest probe [22 m (73 ft)] consistently yielded high organic vapor
measurements. Results ranged from 0 to 10,400 ppm of VOCs; the average
value was 769.5 ppm.
The other deep probe [12 m (40 ft)] station was emplaced in August
1992. The limited data set has lower values (0 to 30.6 ppm), with.an
average value of 7.8 ppm.
WELLHEAD SAMPLING DATA
The 46 wells routinely monitored now.were installed over a 38 year
period (1954 to 1992) using different drilling and completion methods.
Depths range from 23 to 182 m (76 to 600 ft) below surface. Screen
lengths vary and screened intervals lithologies include sands, gravels,
cobbled, boulders, and clay. Depth to groundwater ranges from 58 to 65
m (190 to 214 ft). The wells are located at distances from 0 to 90 m (0
to 296 ft) from CC14 disposal.sites.
Because of the differences of well construction, location, and
sampling times, wellhead concentrations varied widely. The VOCs
measured with an 0VM had a high of 10,704'ppm. While the data itself is
valid, comparisons between different wellheads are difficult.
IMPACT OF VAPOR EXTRACTION ON SOIL-GAS CONCENTRATIONS
The VOCs in shallow soil-gas probes emplaced near the ERA site may
be influenced by the VES operation. Throughout spring and early summer,
shallow soil-gas probe concentrations varied from barometric pressure
and other natural factors. On July 10, 1992, vapor extraction with a
granular activated carbon removal system began from two wells and on
July 23, 1992, two additional wells (a total of four wells) were added
to the VES. This continued until September 23, 1992, when extraction
was reduced to only three wells.
The data indicates a significant decrease in detectable volatile
organiO:s in shallow soil gas probes up to 35 m (115 ft) west and
southwest of the ERA site. Four probes show very low or no detections
through the end of the reporting period (July 20 to September 30, 1992).
While other factors may contribute to the very low detections,
there is enough evidence to suggest operation of the VES may influence
the shallow soil-gas probe concentrations.
-------
WHC-SA-1763-A
IMPACT OF VAPOR EXTRACTION ON WELLHEAD CONCENTRATIONS
A comparison of the wellhead VOC concentrations and the VES
extraction schedule indicate no similarities. While the VES may
influence wellhead concentrations, currently there is no evidence in the
baseline monitoring data to support this.
NATURAL FACTORS AFFECTING DETECTIONS OF VOLATILES
Barometric Pressure
Barometric pressure appears to be the dominant natural factor
affecting baseline monitoring data. There is evidence in the data to
confirm this correlation. The low pressure and increased wellhead
concentrations appear to .be-fairly consistent.' On high'pressure days
[usually above 74 cm (29.2 in.) of mercury] during May through
September, there are rarely any detectable VOCs at wellheads. During
December through April, measurements with pressures above 74 cm
(29.5 iri.) of mercury yielded few detectable volatiles.
The effect of pressure on VOCs in wellheads is usually greatest
when there first is a prolonged high pressure period (three or more
days), then a sharp drop in pressure. On these days VOC concentrations
are usually much higher.
The detection of VOCs in shallow soil-gas probes is not as
consistent. Sometimes VOCs are present on low pressure days, and on
other occasions also present on high pressure days. This trend of VOC
detection in shallow soil-gas probes on high pressure days [above 74 cm
(29.2 in.) of mercury] was unexpected. Often these detections on high
pressure days would occur during a period of rapidly rising or falling
barometric pressure. Often VOCs would not be detectable in wellheads
but would be present in shallow soil-gas probes. A possible cause of
this occurrence is higher pressure concentrating VOCs normally in the
upper meter of soil at a level accessible by the shallow soil-gas probe.
No correlation can be made at this time between VOC levels in deep
soil-gas probes and barometric pressure.
Temperature
Correlation of temperature effects and baseline monitoring of VOC
concentrations is difficult. There may be a relationship between
temperature and shallow soil-gas probe VOC concentrations. During
December 1991 through early February and late July through September
-------
WHC-SA-1763-A
1992, there were few detections of VOCs in shallow soil gas probes.
Temperature may have been a factor in these low detections.
Precipitation
A theory exists that precipitation may act as a "cap" to retain
VOCs beneath the surface. This theory may be correct, but because most
rainfall is associated with low pressure systems, and low pressure does
increase detectable volatiles, a direct correlation is difficult to
prove. The possibility exists that seasonal variations that cause more
rainfall in winter (hence a more moist soil) will cause a general
increase in VOC detections during the winter and spring.
Earth Tide Potential
There appears to be a correlation between earth tide potential, and
wellhead VOC concentrations. In the winter months (December through
mid-March) the low point of maximum tide potential appears significant.
During these periods there are frequently detectable VOCs. Because of
the seasonal nature of this trend, it is possible there are actually
other causes or multiple causes that work to enhance the effect of earth
tide potential. If this tidal period coincides with low barometric
pressure, the concentration of VOCs is even greater.
CONCLUSIONS
There is evidence that operation of the VES may cause a decrease
in detectable VOCs in shallow soil-gas probes.
There appears to be a correlation between increased VOCs detected
in shallow soil-gas probes and high barometric pressure. There is ample
evidence to support previous theories that wellhead concentrations of
VOCs increase with decreasing barometric pressure.
Wellhead concentrations may be affected by tidal forces. During
the winter (December through mid-March), increased wellhead
concentrations can be correlated with many lows of the maximum earth
tide potential. It is difficult at this point to make this correlation
to soil-gas probes, or during any other season. Additional work to
clarify the affect of tidal forces is warranted.
Real time VOC emission wellhead sampling is a method useful in
helping determine subsurface VOC concentrations and their changes caused
by natural factors. Comparison of sampling results from wellhead to
wellhead can only be done in gross terms. The different sample
collection times, various well depths, screening intervals, and
completion methods make strict comparisons dubious.
-------
WHC-SA-1763—A
Comparison of soil-gas probe data and wellhead monitoring data is
difficult at best. As discussed earlier, differences in emplacement
method and depth make comparisons of questionable significance.
-------
WHC-SA—1763-A
REFERENCES
Devitt, D. A., Evans, R. B., Jury, W. A., Starks, T. R., Eklund, B.,
Gnolson, A., Van Ee, J. J., 1987, Soil Gas Sensing for Detection
and Mapping of Volatile Organics, National Water Well Association,
Dublin, Ohio.
DOE/RL, 1991, Expedited Response Action Proposal for the 200 West Area
Carbon Tetrachloride Plume, DOE/RL-91-32, Draft A, U.S. Department
of Energy Richland Field Office, Richland, Washington.
DOE/RL, 1992, 200 West Area Groundwater Aggregate Area Management Study
Report, DOE/RL-92-16, Decisional Draft, U.S. Department of Energy
Richland Field Office, Richland, Washington.
Green, J. W., 1992, Design, Operation, and Monitoring of the Vapor
Extraction System at the 216-Z-1A Tile Field, WHC-SD-EN-TI-010,
Rev. 0, Westinghouse Hanford Company, Richland, Washington.
Rohay, V. J., 1992, FY92 Site Characterization Status Report and Data
Package for the Carbon Tetrachloride Site, WHC-SD-EN-TI-063,
Rev. 0, Westinghouse Hanford Company, Richland, Washington.
WHC-CM-7-7, Environmental Investigation and Site Characterization
Manual, Westinghouse Hanford Company, Richland, Washington.
-------
WHC-SA-1763-A
AUTHOR'S BIOGRAPHY
J. D. Fancher is an Advanced Scientist with the Environmental
Field Services Function at Westinghouse Hanford Company located in
Richland, Washington. He received his Bachelors Degree in Geology from
Eastern Washington University. He is currently a Masters candidate in
Environmental Science at Washington State University. His main work is
site investigation and characterization. He has worked at many
hazardous and mixed waste sites on the Hanford Site as a leader of
sampling activities. Currently he is working at the carbon
tetrachloride Expedited Response Action implementing baseline monitoring
systems and determining the effects of vapor extraction system
operations on emissions of volatiles from the subsurface. Previous work
includes minerals exploration in the western United States and Alaska.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 41
Estimation of Potential VOC Emissions During Trial Excavation
Activities Via Flux Chamber and Fourier Transform Infrared Open
Path Transform
Michelle A. Simon, U.S. EPA, Risk Reduction Engineering Laboratory;
and Bart M. Eklund, Radian Corporation, Austin, Texas
January 12-14, 1993
Las Vegas, Nevada
-------
ESTIMATION OF POTENTIAL VOC EMISSIONS DURING TRIAL EXCAVATION ACTIVITIES
VIA FLUX CHAMBER AND FOURIER TRANSFORM INFRARED OPEN PATH TRANSFORM
Michelle A. Simon, U.S. EPA, Risk Reduction Engineering Laboratory
Bart M. Eklund, Radian Corporation, Austin, TX
Volatile organic compound (VOC) emissions were measured via a flux
chamber and Fourier Transform Infrared Open Path Monitor (FTIR) during
trial excavations at a former oilfield service company. The site
contained two open pits, containing contaminated sludge and soil; the
washout pit and the west pit, as shown in Figure 1. The site also
contained a former, backfilled, disposal pit, called the former west
pit. The site contained relatively high levels of benzene, toluene,
ethyl benzene and xylenes (BTEX); concentrations are listed in Table 1.
Ultimately, the site will be excavated and the contaminated soil and
sludges will be incinerated on site. Although the site is located in a
rural area, there are several residents on the surrounding properties.
It was decided to measure the air emissions during trial excavation
activities in order to estimate safety and health consequences for both
on site and off site personnel during full-scale excavation of the site.
The level of VOC emissions from excavation are temperature dependent,
with the highest emission rates expected to occur during the summer
months. The trial excavation took place during the week of August 19-
24, 1991 - during the hottest, most humid conditions possible for this
southeastern site. Personnel were outfitted to Level C - coveralls and
respirator. Temperatures and humidity were an uncomfortable 95 °F. and
95%, respectively. Shade was not available during the actual sampling
activities. Heat exhaustion was quite possible and was the number one
safety concern. Due to careful planning, no problems occurred. EPA's
Emergency Response Team and Radian supervised the field activities, Roy
F. Weston performed most of the sampling and gas chromatograph/mass
spectrometer (GC/MS) analyses, and Blasland Bouck & Lee operated the
FTIR. Radian performed the mathematical calculations and provided the
final report.3
The pits were dewatered before any excavation took place. Six sampling
runs were performed. The test matrix is shown in Table 1, the sampling
and analysis methods in Table 2. Three methods for measuring VOC
emissions were utilized during this test; flux chambers, FTIR, and
mathematical modeling. Typically, the flux chamber was placed on
undisturbed soil and then later on the same location after it was
disturbed. Air samples were collected from the flux chambers via tedlar
bag and analyzed on site by a portable GC/MS. Stainless steel canisters
were also collected and analyzed by an off site laboratory.
Excavation was performed via a trackhoe. Concurrently to the flux
chamber measurements, the air quality was measured via FTIR open path
monitor. The air was evaluated by FTIR before the diesel powered
trackhoe's engine was started, measured again while the trackhoe's
engine was running but no pit material was disturbed and then during
-------
digging operations. A portable three-meter meteorological tower was
erected at the site. The tower was equipped with Climatronics sensors
to measure wind speed, wind direction, ambient temperature, relative
humidity, and barometric pressure. Thirty second averages of all
measurements were stored within a data logger.
The use of optical remote sensing (ORS) techniques for assessing
emissions from Superfund sites is receiving widespread interest . The
open-path FTIR monitoring performed at this site involved the generation
of an infrared (IR) light beam over a range of wavelengths. The beam
was directed along the monitoring path of interest and any chemical
species in the path absorbed IR radiation at their characteristic
wavelengths. A monostatic configuration was employed, i.e., the
receiver and transmitter were at the same end of the path. A mirrored
corner cube array (retroreflector) or a flat mirror were used to direct
the beam back on itself. A conceptual drawing of the monostatic
configuration is shown as Figure 2. The spectrum received by the system
was compared- to a library spectrum for the chemical compounds of
interest and Fourier-transformed to aid in the identification and
quantification of the compounds present. A component-specific analysis
and comparison was performed via a least-squares-fit spectral software
package to the reference spectra library.
An emission model5 based on diffusion and puff releases was used to
estimate the emissions from specific materials handling activities at
the site such as excavation. The modeT estimates were compared 'to the
field data to validate the models, i.e. to attempt to develop a site-
specific correction factor for the emission models. Both the
measurement and modeling results were used to develop site-specific
estimates of VOC emissions from full-scale remediation activities.
The equations used are shown below. The average emission rate (g/sec)
from excavation, ER, is equal to the sum of emission rates from the soil
pore space, (ERps), and from diffusion, (ERpIFF):
ER = ERPS + ERdiff (Eq. 1)
ER
PS
ER,
PMW 10s Ea Q ExC
RT
(C)(10,000)(SA)
¦DIFF
K.„ k.
TCt
°e Keq,
(Eq. 2)
(Eq. 3)
Equation 2 is based on the assumption that the soil pore gas is
saturated with the compound of interest. If this is not the case, then
-------
Equation 2 may overpredict the emission rate.
The output from Equation 2 should be multiplied by the duration of
excavation and compared to the total mass of contaminants present in the
soil:
3
M = C * Sv * 106
m3
(Eq. 4)
where: M = Total mass of contaminant in a given volume of
soil (g).
If Equation 2 gives a value that exceeds one-third of CT0T, then the
following equation should be substituted for Equation 2:
„ ^ 0.3 3 (Eq- 5)
ERPS = M * —
SV
where: t = Time to excavate a given volume of soil (sec).
SV * x '
The FTIR instrument was unable to detect the target analytes (BTEX) at
the ambient concentrations present at the site. No BTEX was detected
before, during, or after excavation activities. Emissions of methane
and aliphatic hydrocarbons were detected. The total straight-chain
hydrocarbon concentrations were reported as octane and the total
branched-chain hydrocarbon concentrations are presented as iso-octane.
Emission rates for selected species for each run were estimated using a
predictive model. Most of the required inputs for the emission models
were measured in the field and average or representative values were
developed from the data set. The emission rates measured in the field
using FTIR and flux chamber are presented in Table 4 and compared to the
model results. To develop the values given in Table 4, it was necessary
to convert any emission flux data to emission rates based on the amount
of exposed, emitting surface area present during each run.
CONCLUSIONS
Pilot-scale excavation activities at the site did not result in
high levels of contaminants in the ambient air. The best, conservative
estimate of emissions from excavation at the site based on predictive
models are:
• Benzene - 0.01 ug/m2-sec;
• Toluene - 0.005 ug/m2-sec;
• Xylenes - 0.003 ug/m2-sec; and
• Ethylbenzene - 0.002 jig/m2-sec.
-------
These emission flux values can be multiplied by estimates of the total
exposed surface area in the excavation pit and associated short-term
storage piles to yield estimates of the maximum short-term emission rate
to be expected for various excavation scenarios. The actual emission
rates encountered in the field may be significantly lower.
ocpers/veelemewt y
DETECTOR COMPUTER
SOL RCE: Rffei-fnre^
-------
MWe i.
Sampling and Analysis Matrix
Soil
Eipcwd
Number tCSimpte AMlyitd i
PU
Run
m
of
ffcujj
Volume
Surf icc
A««* fJ»J)
Dsl<
OO'jitS I
4c/i»iiy |s»mc»ls Tyxft
BTX
YOCt
SVOCi
frjjfrrifi
YTMhoitf
fS* foiiifff
i
n
*3
ujivh
S&ill
Flu* Cfcirn.
3
0
a
0
0
l }
Q i
Jl'ic
m
a
0
4
PoSl-di*
Soils
Flvi Cham.
4
0
0
1 1
0
Wa^hnui
|SE Cc-neil'
j
*5
'¦"
T
Pdsc-d"(
Soils
Flux Chin.
I
3
j
1
•n
1
¦0
¦fra^houl
!
1
i
15
l.T
3.>
01/10:11
Fre-Mii.
Soil
FTIFL
3
s a
0
c
1
0
J-CK
I.T
: i
W ii in|.
FTIFL
- - a
i q 1
a
&
30
1.7
3 3
PoJt-
Mi*
Soah
FTIR
3
• -j
0
a
c
4
0
fftM
J5F jTnrn** (
1?
:i
ns
OS/]1/91
fre-dU
So ill
f t«* Cham-
I
i
0
0
I
....
Dift
FTIH
"Z
0
c
a
Po5l-die
Soil!
Flux Cham.
t
0
i
• 2
2
J
a
S
'0
41
0I/2W9;
Dig
rrm
— i
4
D
a
iSF. Crwfter)
Pail-d\g
Soils
Flux Cham.
3
0
0
0
1
0
i
0
Ccnowt
Pit
ft
15
26
9(c
QI/Uf*L
Dig
FTIft.
Palh-Avg,
-I
0
0
0
0
0
0
Sails
Flux Cham.
]
1
1
2
1
2
I
0
a - fTK «re ttwrttittutflr trery tva Ithnrlci,
b - The «*c*virion oil Hid ¦ w*f*et arc* of I M mJ and the plte of joii Kid > turfjet area of 100 to 2C0 rn .
e - Ttw MtjvJiion r,j htd i turfKt irtt of *4 m1 «wl t(w pil? of toil hid a lurftcc irei of 41 m!.
NOTE: SVOC - S«n>i-VolxliIe Omiilc Compound
Tabla J.
Sa.sc t£jic jr.d A-ialy£ic3~ Be triads
Simple
Matrix
Ameses
Sample CoHectian
MsthotJ
Afiali'tic.it Ketwd
Soi I
B7EX
voc
Crwscte
PC?
Hand Au^er
Hind Auger
tfSiTrf
Hand Au
Ostector
-------
Sesuits of 3TEX Analysis for Soil Sa-pie;'
'
Average Concentration (iia/q)
Total
Run No.
Pit
Activity
Benzene
Toluene
Xylenes
Ethylbenzsne
1
Washout
Pre
2.15
6.7
23
--
Post
19
43
91
22
2
Washout
Pre
--
--
--
;
Post
4.6
10
54
7.3 |
3
Box
Pre
4.4
10
33
--
Post
6.0
11
34
i
4
West
Pre
1.7
5.0
25
--
Post
11
22
70
--
5
West
Pre
--
--
--
-•
Post
10
19
62
¦15
6
Former
Pre
--
--
--
west
Post
2.0
7.0
42
9.0 |
* Average of combined on-site and off-site analytical results.
Table 4.
Comparison of Emission Rate Estimates
Compound
Emission
Estimation
Approach
Emission Rate (g/sec) j
Run 1
Run 2
Run 3
Run 4
Run 5
Run 6
Benzene
Model
Flux Chamber
FTIR
0.47
3.2E-04
0.067
3.0E-06
0.035
<6.2
1.3
<7.0
1.5
1.5E-04
•1.4
2.5E-05
<0.24
Toluene
Model
Flux Chamber
FTIR
0.18
3.8E-04
0.026
4.0E-06
0.014
<10.4
0.64
<22.8
0.91
3.0E-04
0.58
2.6E-04
<0.79
Xylenes
Model
Flux Chamber
FTIR
0.064
1.2E-03
0.0084
9.7E-06
0.0052
<4.7
0.46
<12.9
0.76
1.4E-04
0.36
1.2E-03
<0.37
Ethylbenz
ene
Model
Flux Chamber
FTIR
0.074
0.010
1.1E-06
0.0055
<0.81
0.34
<4.3
0.51
3.2E-Q4
0.28
2.4E-04
<0.13
-------
REFERENCES
1. Roy F; Weston. Gulf Coast Vacuum Report - Draft. January 1992.
2. Roy F. Weston and Blasland Bouck & Lee. VOC Emission Rates
Derived From FTIR Measurement Data During Pilot Scale Remediation
Activities -Vol. I and II. December 1991.
3. Eklund, B., Estimation of VOC Emissions From Excavation
Activites, Report to Ms. Joan Colson, U.S. EPA, Cincinnati, OH,
June 1992.
4. Draves, J. and B. Eklund. Applicability of Optical Remote Sensing
For Superfund Sites, EPA/45l/R-92/001, May 1992.
5. Eklund, B., S. Smith, and A. Hendler. Estimation of VOC Emissions
From Excavation Activities at Superfund Sites. EPA-450/1-92/004
(NTIS. PB92-171925). March 1992.
-------
AUTHOR'S BIOGRAPHICAL INFORMATION
Michelle M. Simon is a Chemical Engineer with the Risk Reduction
Engineering Laboratory in Cincinnati, Ohio where she is a member of the
Superfund Technical Assistance Response Team (START). She earned her
B.S. in Chemical Engineering from the University of Notre Dame in South
Bend, Indiana and a M.S. in Chemical Engineering at the Colorado School
of Mines in Golden, Colorado. At EPA, her areas of interest are Soil
Vapor Extraction, Nonaqueous Phase Liquids, and air emissions from
hazardous wastes.
Bart M. Eklund is a Senior Scientist with Radian Corporation in Austin,
Texas where he manages the Atmospheric Chemistry Group. He earned his
B.S. in Chemistry from the University of Illinois at Champaign-Urbana
and has over thirteen years of environmental consulting experience. At
Radian, Mr. Eklund's primary duties are directing contract research
programs. His ares of expertise include monitoring, modeling, control
of air emissions from hazardous wastes.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 42
The Effect of Barometric Pumping on the Migration of Volatile Organic
Compounds From the Vadose Zone Into the Atmosphere
Robert J. Pirkle, Microseeps, Pittsburgh, Pennsylvania; and Douglas
E. Wyatt, Van Price and Brian B. Looney, Westinghouse Savannah
Rivar Company, Aiken, South Carolina
January 12-14, 1993
Las Vegas, Nevada
-------
THE EFFECT OF BAROMETRIC PUMPING ON THE
MIGRATION OF VOLATILE ORGANIC COMPOUNDS
FROM THE VADOSE ZONE INTO THE ATMOSPHERE
Robert J. Pirkle
Microseeps, Pittsburgh, PA
and
Douglas E. Wyatt, Van Price and Brian B. Looney
Westinghouse Savannah River Company, Aiken, SC
Several studies reported in the literature have suggested that barometric pumping
is effective in causing the vertical migration of soil gases in the vadose zone. McNerney
and Buseck(1) in a survey of mercury in soil gases in Arizona reported that mercury
emission was sensitive to changes in barometric pressure. Barometric highs were found
to correspond to low mercury emissions and barometric lows to high mercury emissions.
Denton® reported that "an increase of 8 millibars in atmospheric pressure between
23 and 24 October, 1975 appears to have caused a decrease of nearly 50 ppb in the
average abundance of helium in samples" from a survey near Roosevelt Hot Springs,
Utah.
In 1978, Weeks® studied the vertical permeability of sediments in the unsaturated
zone using piezometers to record pneumatic head versus time at depths to 189 feet.
These data were acquired to support tests of the feasibility of storing radioactive gases
in unsaturated basalts underlying the sediments at Birch Creek Playa, Idaho. Even at 189
feet, variations in pressure correlated with barometric pressure changes, suggesting the
movement of soil gases. At 10 feet below the surface, the pressure was found to be
virtually identical to barometric pressure suggesting that soil gases at this depth are
moved with great efficiency by changes in the barometric pressure. The average
permeability for the upper 10 feet of these playa sediments was reported to be about 6
darcies.
In the early 1980's at Gulf Research, interest in the migration of soil gases led
to experiments designed to determine if light hydrocarbons found in the vadose zone
were moved vertically through the soil/atmospheric boundary. One of these
experiments(4) was carried out at the site of a propane gas leak where soil gas propane
concentrations in the vadose zone were far above background level. The results, shown
on Figure 1, reveal a large increase in propane concentration under a groundsheet during
periods of low barometric pressure. The data clearly showed that barometric pumping
was an effective force in moving vadose zone soil gases through the soil/atmospheric
boundary.
-------
A recent study at a shallow landfill at the Savannah River Site near Aiken, SC
extends these observations. The landfill was established in 1974 and is still active. It
covers an area of about 70 acres. Burial has been in excavated earthen trenches about
20 feet deep. The known disposal includes paper, plastic, construction debris, solvent
rags, metal debris and carcasses.
A soil gas survey was carried out at the landfill in December, 1990. The
objective of the survey was to determine the presence and extent of volatile contaminants
in near surface soil gases. A sample grid on 100 foot centers consisting of 288 sample
locations was established as shown on Figure 2. Samples were taken from a nominal
depth of three feet. Species monitored were the light hydrocarbons, Q-C4; gasoline
range normal paraffins and aromatics, C5-C10; and selected chlorinated organics. Low
levels of volatile organic compounds including trichloroethylene, tetrachloroethylene and
1,1,1-trichloroethane had been reported in groundwater monitoring wells at the site.
The results of the soil gas survey confirmed that a variety of common petroleum
based fluids and chlorinated solvents had been a part of the materials buried at the
landfill. Methane, generated from the biological degradation of cellulose and other
organic materials, was found in concentrations ranging up to 63% by volume.
The observation of large concentrations of volatile species, including methane, in
the burial trenches of the landfill and our previous experiences, led to experiments to
determine if these gases could move through the nominal 3 foot soil cover over the
trenches during periods of falling barometric pressure. Four groundsheets (8x8 feet)
were placed at selected locations at the landfill based on methane concentrations
determined in the soil gas survey. Samples were taken from underneath each
groundsheet every four hours for a period of two weeks. Each sample was analyzed
using gas chromatography for the light hydrocarbons methane, ethane, propane, i-butane,
n-butane, ethylene and propylene. Selected samples were analyzed for the volatile
organics found in the soil gas survey.
Methane concentrations from all samples taken from under each of the four
groundsheets are shown on Figure 3. Concentrations under each groundsheet were
observed to vary greatly over the two weeks of observation. Concentrations observed
under groundsheet #2 are much larger than under groundsheets 1, 3, or 4. Careful study
of Figure 3 reveals that despite the differences in magnitude, the variation of methane
concentrations under all four groundsheets are approximately in phase. This would be
expected if these variations in concentration resulted from a single driving force such as
barometric pumping.
Methane under groundsheet #2 and barometric pressure are shown as a function
of time on Figure 4. Methane concentrations under this groundsheet increase rapidly
during periods of falling barometric pressure. Close study of the data reveals that even
-------
short periods of stable pressure in an overall falling pressure regime causes cessation of
the methane flux.
Some variations in the rate of methane flux over the course of the experiment may
be attributable to changes in soil moisture content as a result of periodic heavy rains
during the observation period similar to suggestions by McNerney and Buseck*1*
Increases in soil moisture are thought to decrease the permeability of the soil and impede
the pressure equilibration which causes the flow of soil air through the soil/atmosphere
boundary.
The other light hydrocarbons, C2-C4, exhibited variations similar to methane. The
only significant difference was their much smaller concentration. One sample,which
corresponded to a methane maximum under groundsheet #2, was further analyzed for
volatile organics. A chromatogram of that sample, shown in Figure 5, reveals that many
if not all of the complex volatile organics buried at the landfill migrate vertically through
the soil/atmosphere boundary into the atmosphere. The flux is maximized during periods
of falling barometric pressure and is minimized during periods of rising or stable
barometric pressure.
The effect of this phenomenon is to cause volatile species in the vadose zone to
be distributed vertically to the surface and into the atmosphere. It is largely this
phenomenon which allows shallow soil gas measurements to detect and delineate the
distribution of volatile species in or above the water table. The efficiency with which
this can be accomplished is mediated by the porosity and permeability of the vadose
zone.
Through barometric pumping, the volatile contaminants in the vadose zone
represent a source of these species in ambient air. Attempts to model the magnitude of
this source, which do not include the barometric effect, will undoubtedly misrepresent
its significance.
REFERENCES
1. McNerney, J.J. and Buseck, P.R., Economic Geology 68, 1313 - 1320, 1973.
2. Denton, E.H., USGS Open File Report #77-606, 1977.
3. Weeks, Edwin P., Geological Society Professional Paper #1051, 1978.
4. Pirkle, R. J. and Price, Van, "Leak Detection and Monitoring", Paper presented
at the Conference on Hazardous Materials in the Environment, Atlantic City, NJ,
June, 1986.
-------
*"**«
Si
•nmg Qf
Propan<>
*• /. ^'Va^
man
e 'Noex'^ s'Te
River?
^Ap
8
MORTH AREA
QOqpOOOOOO
ooqicooooo
OO^kOOOCOO
Cod! o o oo o o
o a o o o o
o o o
on£^» o o o
0 0 0 3 0 3
4cd)OOCOQO
occjiooooao
«*2S» ,
Gro,
SOUTH 1 AREA
o o a a o ot
oooQ
o o o W
O O <3 O O C<
O o o o o oc
o 0 a o o o <
<3 o o o a o
o a o o o ®
o o o o o c
o a o o o o
,000000
o o o o 5 0
0SB0O oooo O 9 O O 0
est "^eattons
f «sa*
the Lsnemi s0i/
Gas Sw
Grfa
-------
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
DATE SAMPLED (JANUARY 1991)
Figure 3. Methane Concentrations Under Four Landfill Groundsheets vs Time
Figure 4. Methane Concentrations Under Groundsheet #2 and Barometric Pressure
vs Time
-------
FUe«C:\C?\dS\uSA.36a Oat# printed » 03-27-1991 Haw a 16:34:22
3.70 to 29.92 ain. Low T * -0.05649 iav High T * 0.47969 ay Span * 0.S3618 1w
'* . i '( ' 1 '1 ta ii 11 li a is ic ij i» n 20 ti 12 a :* zs zt z; zs zi
Figure 5. VOC Hydrocarbon Chromatogram of Sample from Groundsheet #2 During
a Period of Falling Barometric Pressure
-------
Robert J. Pirkle received a B.S. and M.S. in chemistry from Auburn University
and a Ph.D. in chemistry from The University of Western Ontario. With three
other colleagues from Gulf Research and Development Company, he formed
Microseeps in 1985 and has served as president since 1990. His interests include
geochemical and environmental analyses and their application to petroleum
exploration and environmental assessment.
Douglas E. Wyatt received B.S. degrees in zoology and geography in 1979 and
1980 from the University of Tennessee and an M.S. in geology from Vanderbilt
University in 1984. He is currently a doctoral candidate in geology at the
University of South Carolina. After eight years as an exploration geologist with
TransAtlantic Exploration and Eastern Natural Gas, he joined the Westinghouse
Savannah River Company as a Senior Scientist and Program Manager for
RCRA/CERCLA investigations in 1990. His interests include the integration of
characterization technologies and the application of high resolution techniques to
deep investigations.
Van Price received a B.S. in geology from the University of South Carolina and
the M.S. and Ph.D. in geology from the University of North Carolina. He has
been at the Savannah River Site for over 15 years and, since 1990, has been
Manager of Ground Water Monitoring in the Westinghouse Savannah River
Company's Environmental Protection Department. His interests include
application of geology, geochemistry and geophysics to environmental problems.
Brian B. Looney received a B.S. in environmental sciences from Texas Christian
University and a Ph.D. in environmental engineering form the University of
Minnesota. Since 1983 he has been a research environmental engineer in the
Environmental Sciences Section at the Savannah River Laboratory. His interests
include development of innovative methods for environmental remediation,
modeling, and risk assessment.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 43
Use of Risk Assessment Groundwater Model in Installation
Restoration Program (IRP) Site Decisions
David K. Goldblum, Ph.D., P.E. and John M. Clegg, P.E., Sverdrup
Environmental, San Antonio, Texas; and John D. Erving, University of
Michigan, Ann Arbor
January 12-14, 1993
Las Vegas, Nevada
-------
USE OF RISK ASSESSMENT GROUNDWATER MODEL IN INSTALLATION
RESTORATION PROGRAM (IRP) SITE DECISIONS
David K. Goldblum, PhD, PE
John M. Clegg, PE
Sverdrup Environmental, San Antonio, TX 78216
John D. Erving
University of Michigan, Ann Arbor, MI 48109
INTRODUCTION
Management of hazardous chemicals and their disposal involve both assessment of the
exposure risks and regulations to control these risks. The Installation Restoration Program
(IRP) is a multidisciplinary Department of Defense (DOD) program to assess and remediate
hazardous waste problems on DOD installations (7).
Quality data is" necessary to assess the risk associated with toxic chemicals situated and
disposed of at a site; quality data is also required to set priorities for clean-ups at such
hazardous waste sites. This risk assessment process estimates total carcinogenic and
noncarcinogenic risk at each site (1,2,3,4,5), and decisions on remediations are based on
its outcome.
The critical review of a contractor's risk assessment for a drum storage area at a military
installation prevented a costly and unnecessary remedial action. This review shows how
faulty mathematical calculations and the questionable inclusion of a chemical as a
compound of concern led to an apparent need for clean-up. When the risk was recalculated
using valid data, correcting the faulty mathematical calculations and using appropriate
chemicals of concern, no clean-up was required.
The three chemicals involved were: (a) trans- 1,2-dichloroethene (b) trichloroethylene
(TCE); and (c) poly-chlorinated biphenyl - 1260 arochlor (PCB-1260). However, the
concentration of the trans-1,2-dichloroethene (noncarcinogenic compound) was negligible
in all environmental sectors so that in this risk assessment only TCE and PCB-1260 were
relevant. For TCE and PCBs, risks are essentially carcinogenic, so noncarcinogenic risks
are not pertinent. Thus, only chronic intake doses based on average concentration are
relevant. The primary errors in the original risk assessment were that:
• The average TCE concentration was based on only part of the data resulting in an
apparent average concentration almost double the actual average over all data points.
• A major part of the risk value resulted from the PCB-1260, which was based on
an average PCB-1260 concentration over 3 monitoring wells consisting of 2 non-
detect values and one estimated value below the method detection limit.
Other considerations in the risk assessment include:
• The use of nearby bodies of water e.g., are they drinking water supplies or not
• The surface water analytical results
• The rate of migration for the distance between the contaminant source and the
point of exposure
ANALYTICAL MODELS/METHODOLOGY
Initially, overall chronic intake dosages of the TCE and PCBs are estimated, which tire
multiplied by the slope factors to compute the risk component for that given chemical from
that specific exposure pathway. The risk assessment is completed when the individual risk
-------
components are summed into the total site risk. The three potential exposure pathways
involved in this risk assessment were:
• ingestion of water via drinking
• inhalation from showering
• dermal exposure from bathing
The concentrations of these chemicals were found to be negligible in the air, surface water,
and both surface and subsurface soils. Hence, only exposures for inhalation results from
inhaling volatiles during showering, dermal contact from bathing, and ingestion from
drinking; using this contaminated groundwater. It is worth noting that there are presently
no wells drawing water from the zone of groundwater contamination (the contaminated
plume). Therefore, the calculated ingestion exposure due to drinking this contaminated
groundwater overestimates actual exposure from drinking.
For all three exposure pathways, BW is the average adult body weight of 70 kg (1,5),
and for a conservative upper bound estimate on intake, it is assumed that EF*ED/AT = 1,
i.e. EF = 365 days/yr, ED = 70 years, and AT = 70*365 days = 25,550.days. In addition,
the contractor applied a factor of 58/68 to the chronic intakes to give a weighted chronic
intake for adults, where 58 is the number of years as an adult out of-a 68-year lifetime.
Due to large uncertainty in expected exposures for the 0-2 age category in most cases, the
lifetime adjustment takes away these two years and the factor applied is 58/68 instead of
58/70, in order "to lean on the conservative side of the doubt"
Calculation of Exposure from Ingestion:
(1) Intake (mg/kg-day) = (CW«IR*EF«ED)/(BW>AT)
Studies have shown that on the average 75% of water consumption is for home use (8). At
this site, actual exposure is far less, since it is strictly a work area with no residential area
nearby. Thus, equation (1) simplifies to equation (2). IR = 2 liters/day (1,5).
(2) Intake (mg/kg-day) = (CW*IR*0.75)/BW
Calculation of Exposure from Inhalation:
(3) Intake (mg/kg-day) = (CA«IR»ET*EF*ED)/(BW*AT)
(4) Intake (mg/kg-day) = (CW*IR*VWsh)[SD/SV + BD/BV]/BW
CW*VWsh/SV and C\V*VWsh/BV represent CA for the shower process and after-shower
bathroom exposure, respectively. Furthermore, for shower ET = SD and ET = BD for
after-shower bathroom exposure time, so that equation (3) becomes equation (4). In
addition, VWsh = 100 liters and is average amount of water used during a shower (8),
IR = average adult value of 20 m^/day (1). Other values pertinent to equations (3) and (4)
include: SD = BD = 10 minutes (1/6 hours); and SV = 3 m^; and BV = 10 m^ (8).
Calculation of Exposure from Dermal Contact:
(5) Intake (mg/kg-day) = (CW*SA'PC'ET'EF«ED*CF)/(BW«AT)
MF (mass flux) is taken as 0.5 mg/cm^-hr, since this was found to be 0.4336 mg/cm^-hr
in an absorption study (9). This corresponds to a PC value of 5.0 x 10~4 cm/hr, since TCE
is at sufficiently low concentrations, where it can be deemed as a dilute aqueous medium.
Thus, equation (5) simplifies to equation (6).
(6) Intake (mg/kg-day) = (CW*SA»MF«ET«CF)/BW
SA = 18,150 cm2> the average adult surface area of exposure (1,5). As shown in the
results, exposure from dermal contact is very small compared with that from ingestion and
inhalation (approximately 3 orders of magnitude smaller).
-------
Route/Chemical Specific Risk:
(7) Rij = SFij • (Intake)ij
Risk numbers (dimensionless) denote frequency of excess cancer cases above natural
background (250,000 cases per million people, about a quarter, due to natural
environmental factors). If the risk is 2 x 10~4, then there would be 2 excess cancer cases in
10,000 people over those expected under environmental background levels (1). Slope
factors (SFs) are obtained from the EPA IRIS (Integrated Risk Information System). For
TCE, the most recent SF is from 7/1/89, and for PCBs, the most recent SF is from 5/1/89.
Total (Overall) Risk:
(8) Total Risk = R = Sum of all Risks from equation (7)
3 2 3 2
(9) R = X X Rij = X X SFij. (Intake)ij
i=l j=l i=l j=l
Note that total risk is obtained by summing over both chemicals and over the three
exposure pathways. Since PCB is not volatilized at all, only 5 chemical/exposure pathway
specific risks are added into the total risk calculation: ingestion of PCB and TCE; dermal-
contact of PCB and TCE; and inhalation of TCE.
RESULTS
Initially average concentrations of specific analytes in medium applicable to given
exposure pathway are determined. For all three exposure pathways only the groundwater
is of concern. Therefore, the average concentrations of TCE and PCB-1260 in the
contaminated groundwater must be determined.
For the TCE, there were a total of 6 measurements from 4 monitoring wells (MWs):
MW-1: 0.0027 mg/l
MW-2: 0.0081 mg/l
MW-3: 0.08, 0.016, and 0.17 mg/l
MW-4: 0.026 mg/l
The average of these 6 readings is 0.05 mg/l, but it may be unclear as to which numbers
should be entered into this average, since 3 of these readings came from MW-3 and these 6
readings were not all taken at the same time. Therefore, the average of the 4 wells was
taken, where only the highest reading from MW-3 was retained. This average is 0.0517
mg/l, which is slightly above the 0.05 mg/l value. Moreover, if a weighed average at MW-
3 were taken, and this were averaged with the other 3 wells, then the average would be
even lower at 0.0314 mg/l. Hence, the 0.05 mg/l average TCE concentration serves as a
solid conservative upper bound estimate. However, the contractor used an average
concentration of 0.0895 mg/l, which seemed to average the MW-2 and highest MW-3
readings, and ignored the other 4 measurements, in effect, doubling the apparent TCE-risk.
For the PCB-1260, there were a total of 3 measurements from 3 monitoring wells
(MWs):
MW-1: 0.00081 mg/l (estimated)
MW-2: none detected
MW-3: none detected
For "none detected" measurements, 0.0005 mg/l, half the value of the method detection
limit, 0.001 mg/l (7), is entered into the average to be conservative. Thus, average PCB-
1260 concentration is 0.0006 mg/l, considerably below the detection limit. Hence, it is
questionable whether PCB-1260 should have been deemed a chemical of concern.
-------
A comparison of the risk assessment performed for average TCE concentrations of .0.05
mg/1 and 0.0895 "mg/1 follows as shown in Tables 1 and 2, with the PCB-1260 risks
included. The intakes are computed using equations (2), (4), or (6), depending on the
exposure pathway calculated. The factor of 58/68 is then applied to the intake to give the
weighted intake, which in turn is multiplied by the slope factor (SF) to give the specific
risks for the particular exposure pathway and given chemical. Then total site risk is the
sum of all of specific risks.
TABLE 1: Risk Assessment (TCE Concentration = 0.05 mg/1)
Chemical
TCE
PCB-1260
TCE
PCB-1260
TCE
PCB-1260
CW (mg/D
0.05
0.0006
0.05
0.0006
0.05
0.0006
Pathway/Equation
Ingestion/(2)
Ingestion/(2)
Inhalation/(4)
Inhalation/(4)
Dermal Contact/(6)
Dermal Contact/(6)
Intake (mg/kg-dav)
1.071 x 10"3
1.286 x 10"5
4.299 x 10-3
non-volatile
1.080 x 10-6
1.296 x 10"8
Weighted intake fmg/kg-davl
9.139 x 10"4
1.097 x lO-5
3.667 x lO"3
non-volatile
9.215 x 10"7
1.106 x lO'8
SF (mg/kg-dav)
0.011 (ingest)
7.7 (ingest)
0.013 (inhale)
7.7 (inhale)
0.011 (dermal)
7.7 (dermal)
-1
Specific Risk
1.005 x 10"5
8.444 x 10"5
4.767 x 10"5
non-volatile
1.014 x 10"8
8.514 x 10"8
Total risk sum of specific risks is approximately 1.4 x 10"4, upon rounding to one
significant figure in accordance with USEPA instructions, gives a risk of 1 x 10"4. This is
within the acceptable range for USEPA (1,4,5).
TABLE 2: Risk Assessment (TCE Concentration = 0.0895 mg/1)
Chemical
TCE
PCB-1260
TCE
PCB-1260
TCE
PCB-1260
CW fmg/1')
0.0895
0.0006
0.0895
0.0006
0.0895
0.0006
Pathway/Equation
Ingestion/(2)
Ingestion/(2)
Inhalation/(4)
Inhalation/(4)
Dermal Contact/(6)
Dermal Contact/(6)
Intake (mg/kg-dav)
1.918 x lO"3
1.286 x 10"5
7.695 x lO"3
non-volatile
1.934 x lO"6
1.296 x lO*8
Weighted intake (mg/kg-dav-)
1.636 x lO"3
1.097 x 10"5
6.563 x lO"3
non-volatile
1.649 x lO'6
1.106 x 10'8
SF (mg/kg-dav")
0.011 (ingest)
7.7 (ingest) .
0.013 (inhale)
7.7 (inhale)
0.011 (dermal).
7.7 (dermal)
1
Specific Risk
1.799 x lO"5
8.444 x 10'5
8.533 x lO"5
non-volatile
1.814 x 10'8
8.514 x 10"8
-------
Total risk sum of specific risks is approximately 1.9 x 10~4, upon rounding, gives a risk of
2 x 10~4. This falls in the high risk category according to USEPA guidelines (1,4,5).
Figure 1 illustrates total risk for both TCE concentrations with the PCB-1260 risks
discarded as well as that with PCB-1260 risks included. Note that in case "a", where both
TCE = 0.05 mg/1 and the PCB-1260 risks are discarded, total risk is well below the EPA-
threshold of 1 x 10~4.
R = 2 x 10-*
1 excess cancer
5,000 people
R = 2 I 10 '
EPA
THRESHOLD
R = 1 X 10-4
1 nTrw ib's
B i 10"®
7 ,10-»
6 * 10"4
5 x 10"®
4 t 10"5
3 x 10"®
2 i 10"°
: 1 * 10
R = 6 x 10 "®
-5 1 excess cancer
100,000 people
CL
L
c-
ct
Effect of Conditions on Carcinogenic Risk Estimate
R = Carcinogenic risk (
tf ot excess cancer casea
population
)
Figure 1: Effect of Conditions on Carcinogenic Risk Estimate
a) PCB-risk component discarded and correct average
TCE-concentration at 0.05 mg/1 (Total risk is 6 x 10"^)
b) PCB-risk component discarded and erroneous higher average
TCE-concentration at 0.0895 mg/1 (Total risk is 1 x 10"4)
c) PCB-risk component included and correct average
TCE-concentration at 0.05 mg/1 (Total risk is 1 x 10"4)
d) PCB-risk component included and erroneous higher average
TCE-concentration at 0.0895 mg/1 (Total risk is 2 x 10"4)
Clearly, the quality of the risk assessment can make a crucial difference in a remedial action
decision.
CONCLUSIONS
The calculated risk reduction which resulted from using the corrected average TCE
concentration allowed the Air Force to place this site in the No Further Action (NFA)
status. Moreover, since the PCB-1260 concentrations for all data points were below the
method detection limit, while a major part of the calculated risk value was due to the PCB-
1260 component in the ingestion of drinking the groundwater supplies, further questions
arose concerning the wisdom of pursuing a costly remedial action at this site. The Air
-------
Force was considering an expensive pump and treat remedial action, which would have
been of little value at this site.
Other factors considered in arriving at the NFA decision include the absence of both
TCE and PCBs in all other environmental media, namely air, soils, and nearby surface
water and sediments. Furthermore, the groundwater present at the site is not used as an
active drinking water source, since drinking water is provided to all users both on and near
this installation by the local township from an alternate water source. In addition, the
closest municipal well drawing from the alternate water source is approximately three miles
upstream from this installation. Since there was no PCBs nor TCE detected in the alternate
water source nor in surface water, no contamination problem of domestic groundwater
supplies is apparent. In addition, no contaminants were identified in the surface water. In
summary, when performing a risk assessment, all factors must be considered, as well as
proper quality assurance and quality control of data and mathematical operations to insure a
valid portrayal of the risk.
NOTATION
AT = time interval over which the exposure is averaged, days
BD = average after-shower exposure time in bathroom , hours
B V = volume of bathroom, rrP
CA = concentration'of chemical in air, mg/rr?
CF = conversion factor (1 liter/1000 cm^ or 1 liter/10^ mg)
CW = concentration of chemical in water, mg/1
ED = exposure duration, years
EF = exposure frequency, days/year
ET = exposure time, hours/day
Intake = dose intake over the lifetime interval, mg/(kg-day)
IR = intake rate (2 1/day for drinking water; 20 m^/day for breathing air)
PCB = poly-chlorinated biphenyl
R = total risk
Rij = the exposure route/chemical specific risk
SD = duration of shower, hours
SV = volume of the shower stall,
SF = slope factor (carcinogenic potency factor), kg-day/mg. SFs are obtained from EPA's
lRIS(Integrated Risk Information System)
SFij = slope factor for exposure pathway i and chemical j
TCE = trichloroethylene
Subscripts:
i = 1, 2, and 3 for the exposure pathways (ingestion of drinking water, inhalation during a
shower, and dermal contact from taking a bath, respectively)
j = 1, and 2 for the chemicals (TCE, and PCB-1260, respectively)
-------
REFERENCES
1. USEPA. 1989b. Risk Assessment Guidance for Superfund - Volume I - Human Health
Evaluation Manual (Part A) - Interim Final. U.S. Environmental Protection Agency - Office
of Emergency and Remedial Response. EPA/540/1-89/002, December, 1989.
2. USEPA. 1989b. Risk Assessment Guidance for Superfund - Volume II - Environmental
Evaluation Manual- Interim Final. U.S. Environmental Protection Agency - Office of
Emergency and Remedial Response. EPA/540/1-89/001, March, 1989.
3. USEPA. 1989. Ecological Assessment of Hazardous Waste Sites: A Field and
Laboratory Reference. U.S. Environmental Protection Agency - Environmental Research
Laboratory. EPA/600/3-89/013, March, 1989.
4. USEPA. 1986a. Superfund Public Health Evaluations Manual. U.S. Environmental
Protection Agency - Office of Waste Programs Enforcement. ICF Incorporated. October,
1986.
5. USEPA. 1988. Superfund Exposure Assessment Manual. U.S. Environmental
Protection Agency - Office of Remedial Response. EPA/540/1-88/001, April, 1988.
6. USEPA. 1988b. Guidance for Conducting Remedial Investigations and Feasibility
Studies under CERCLA. U.S. Environmental Protection Agency - Office of Solid Waste
Emergency Response. October, 1988.
7. Handbook to Support the Installation Restoration Program (IRP) Statements of Work
for Remedial Investigation/Feasibility Studies (RI/FS). May, 1989 (Version 3.0).
8. Symms, K.G., "An Approximation of the Inhalation Exposure to Volatile Organic
Synthetic Organic Compounds from Showering with Contaminated Household Water",
paper presented at the Symposium of American College of Toxicologists, 15 November
1985.
9. Tsuruta, H., Ind. Health 16: 145-148 (1978) as cited in Health
and Safety Executive Monograph: Trichloroethylene #6 p. 3 (1982).
-------
BIOGRAPHIC SYNOPSIS
David K. Goldblum. PhD. PE
David Goldblum received a B.S. in Mathematics from Camegie-Mellon University in
1975, an M.S. in Chemical Engineering from Rice University in 1980, and a Ph.D. in
Chemical Engineering and Environmental Health Science from University of Michigan in
1988. Early experience at PPG Industries Chemicals Technical Center at Corpus Christi,
TX involved chemical engineering in a technical support group and cost studies. Recent
tenure as a bioenvironmental engineer consisted of environmental consulting in water
quality, hazardous waste management, as well as environmental chemistry in the
Installation Restoration Program. Most recently, served as the Chief Consultant for the
Environmental Chemistry Function at the Air Force Center for Environmental Excellence,
and is currendy acting as a consultant to Sverdrup Corporation in risk assessment.
John M. Clegg. PE
John Clegg received a B.S. in Civil and Environmental Engineering from University of
Houston in 1971, and an M.S. in Environmental Engineering from Tulane University in
1975. Premilitary experience consisted of design of a lagoon and an oxygen enrichment
system at a Petrotex chemical plant for a consulting firm, W.B. Joseph's, as well as design
& construction in hydraulics and hydrology for the Corps of Engineers. Military
experience as a bioenvironmental engineer involved numerous base level chief
bioenvironmental engineer assignments both nationwide and worlwide. Most recently,
served as the Chief of Program Execution and Base Closure Operations at the Air Force
Center for Environmental Excellence, and is currently the Southwest Regional Manager for
Sverdrup Corporation directing the Sverdrup Environmental Office in San Antonio.
John D. Erving
John Erving received a B.S. in Biochemistry from Eastern Michigan University in 1978
and an M.S. in Chemistry from University of Michigan in 1987. Previous assignments in
the A*ir Force consisted of program management and acquisition of technology and
involved extensive worldwide travel. Formerly, with the Air Force Center for
Environmental Excellence, John Erving is now pursuing graduate studies at the University
of Michigan.
-------
A REPRINT FROM THE MAY 1992 ISSUE OF
SMT/SL
Progress
USE OF RISK ASSESSMENT GROUNDWATER
MODEL IN INSTALLATION RESTORATION
PROGRAM (IRP) SITE DECISIONS
David K. Goldblum and John M. Clegg
Air Force Center for Environmental Excellence
Environmental Services Office
Environmental Restoration Division
Bldg. 624W
Brooks AFB, TX 78235-5000
John D. Erving
Michigan State University
Det. 380 -.AF ROTC
122 Bessey Hall
E. Lansing, Ml 48824-1033
-------
Calculation of Exposure from Ingestion
Intake
J-
mg
V kg - day
(CW)(IR){EF ){ED)
(BW)(AT)
(H
Intake
J-
V kg - day
0.75 (CW){IR)
(BW)
(2)
The IR value is laken as 2 liters/day [7, J], The BW is the
average adult body weight of 70 kg [/, 5].
Calculation of Exposure from Inhalation
Intake
\k%~ day
{CA){}R){ET){EF)(ED)
(.BW){AT)
(3)
Similar to ingestion, EF*ED/AT = 1, that is, EF = 365
days/yr, ED = 70 years, and AT = 70*365 days = 25,550
days.
Intake —
\kg
m3 \ = {CW){IR){VWjf SD + fl£A
-day) (BW) IsV BV J
Note that CW*VW;h/SV and CW+VWJh/BV represent
CA for the shower process and additional bathroom expo-
sure after showering, respectively. Furthermore, for shower
ET = SD and ET = BD for additional bathroom exposure
time after showering. Thus. Eq. (3) becomes Eq. (4). The
IR value is taken as the average adult value of 20 m3/day,
see pp. 6-44 in Risk Assessment Guidance for Superfund -
Vol. I - Human Health Evaluation Manual (/]. The BW is
the average adult body weight of 70 kg [/, 5], The other
values pertinent to Eq. (3) and [4) include: SD = BD = fO
minutes (1/6 hours); and SY = 3 m3; and BV = 10 m' {
The SA value is taken as the average adult value of 1 8,150
cm2!/, 5], and BW as the average adult body weight of 70
kg [/, 5J. The MF is taken as 0.5 mg/cm--hr, since this was
found to be 0.4336 mg/enr-hr in an absorption study [9],
This corresponds to a PC value of 5.0 x 10J cm/hr, since
the TCE is at sufficiently low concentrations such that it
can be deemed as a dilute aqueous medium. As will be
illustrated in the Results, the risk associated with derma)
contact is very small compared with that from ingestion
and inhalation, such that if a PC value of 8.4 x 10"4 cm/hr
[/] were used, the risks from dermal exposure would still
be more than 2 orders of magnitude less than the specific
risks from the ingestion and inhalation exposure pathways.
For all of the exposure pathways, the contractor applied
a factor of 58/68 to the chronic intakes to give a weighted
chronic intake for adults, where 58 is the number of years
as an adult out of a 68-year lifetime. Due to targe uncer-
tainty in expected exposures for the 0-2 age category in
most cases, the lifetime adjustment takes away these two
years and the factor applied is 58/68 instead of 58/70, in
order "to lean on the conservative side of the doubt". The
pre-teenage (less than 12) group could not possibly be pre-
sent at this site. In addition, assuming that the adult period
starts at age 12 is extremely strict, since a work zone would
not likely have anyone present under the age of 18. and
certainly not under the age of 16. Perhaps, inherent in this
assumption, is that the 12-18 group may be present in spo-
radic amounts by visits, so this age group is counted as if
they were present on a regular basis. Therefore, a factor of
52/68 or 54/68 would have been more reasonable, but to
err on the side of caution, the stricter factor of 58/68 was
retained in This review of the contractor's risk assessment.
Route/Chemical Specific Risk
(SFJ
dimensionless
Kg - day
mi;
(intake)^
mg
Kg - day
(7)
Hole that risk values are dimensionless. since slope fac-
tors have units reciprocal to (hose of intake. Risk numbers
denote frequency of excess cancer cases above what is to
be expected for environmental backgrounds (250,000 cases
per million people, about a quarter, due 10 natural environ-
mental factors). If the risk is 2 x 10'4, then there would be 2
excess cancer cases in 10.000 people over those expected
Environmental Progress (Vol. 11, No. 2)
May 1992 93
-------
under environmental background levels f/). The slopefac- ¦
tors (SFs) were obtained from ihe EPA iRIS (Integrated
Risk. Informal ion System). For TCE, the most recent SF is
from 7/1^89, and for PCBs Ehe most recent SF is from
5/1/89-
These SFs can be obtained from six sources with the fol-
fcowing priority (see pp. 7-13 -7-15 from [./].
(i) IRIS
(ii) HEAST - Health Effects Assessment Stmmary
Tables, which are updated quarterly.
(lii)EPA Criteri a Docume nis,
(iv) Agency for Toxic Substances and Disease
Registry (ATSDR) toxkologicaJ profiles.
{v) EPA's Environmental Criteria and Assessment
Office.
(vi) Open literature.
Totaf Risk
Total Risk = R = Sum of all Risks from Eq.f7) fS)
R = i i R;,; = i i (S F,Xfxtake)h. (9)
1 <=>)=>
Note that total risk is obtained by summing over both
chemicals and over the three exposure pathways. Since
PCB ts not volatilized at all, only 5 chemical/exposure path-
way specific risks are added into the total risk calculation:
ingestion of PCB and TCE; dermal contact of PCB and
TCE; and inhalation of TCE.
ResuFts
The first step in analyzing the risk for any given expo-
sure pathway is lo tie-.ermine average cone en ra lie is of the
specific analyies in that medium, applicable to the exposure
pathway. The three above noted exposure pathways indicate
iriat the only environmental medium of concern is the
groundwater. Therefore! the average concentrations of the
TCE and PCB must be determined. For these iwo chemi-
cals. the results of measurement are normally reported in
ug/1. However, these are shown in mg/l in order to obtain
the chronic intakes in (mg/kg-day) since the slope factors
are in reciprocal (mg/kg-day) and the risk is dimensionless.
For the TCE, there were a total of 6 measurements from
4 monitoring wells (MWs):
MW-1: 0.0027 ng/1
MVV-2: 0 0081 mg/l
MW-3: 0.08, 0.01 6, and D.I7 nig/;
MW-4: 0.026 mg/1
The average of these 6 readings is 0.05 mg/1, but it is
unclear as to which numbers should be entered into this
average, since 3 of these readings came from MW-3 and
these 6 readings were not all taken at the same time.
Therefore, the average of the 4 wells was taken, where only
the highest reading from jMW-3 wa$ retained, This average
is 0 05 17 mg/1, which is slightly above the 0.05 mg/l value.
Moreover, if a weighed average at MW-3 were taken, and
this uari averaged with the other 3 wells, then the average
would be even lower at 0.0314 mg/1, Hence, the 0.05 mg/1
average TCE concentration serves as a solid conservative
upper bound estimate. However, in this risk assessment, the
contractor used an average concentration of 0.0895 mg/1,
which seemed to average the NW-2 and highest MW-3
readings, and ignored the other 4 measurements. This
resulted in an almost double value for the TCE risks. A13
TCE-readings were welt above the method detection limit
of 0.0006 mg/1 [7J.
For the PCB, there were a total of 3 measurements from
3 monitoring welis (MWs):
MW-I; 0.00081 mg/l (estimated]
MW-2: none detected
MW -3: none delected
Fot" the "none detected" measurements, 0.0005 mg/l, haif
the valje of the 5.0C1 rrg/1 [7] metlioc dececlioa lirru:, is
entered into the average to assure thai a cor.servative,esti-
mate is taken. Thus," the average concentration for PCB is
0.0006 mg/l, considerably below the detection limit.
Consequently, the PCB-risk is negligible, since the concen-
trations are below the method detection limit. Thus, it is
highly questionable whether the PCB should have been
deemed a chemical of concern at this site. However, the
PCB-risk component is added here to show an upper bound
risk for these site conditions.
A comparison of the risk assessment performed for aver-
age TCE concentrations of 0.05 mg/l and 0.0395 mg/3 fol-
lows-as sbem-rt .ri Tables ] and 2 Ths nflibc: ars-cornrpined
using Eq (2]. (4), or (6), depending on the exposure path-
way calculated. Trie faaor of 58/6S is then, applied to the
intake to give the weighted intake, which in turn is multi-
plied by the slope factor (SF) to give the specific risks for
Ihe particular exposure pathway and given chemical. The
sum of all of the specific risks then gives the total risk for
the site, which in this case is a carcinogenic risk.
From Figure !, it is evident thai ihe quality of the risk
assessment can make a crucial difference in the outcome in
a remedial action decision. In case "a", where the PCB-risk
component was discarded and the correct average TCE-con-
cenlration at 0.05 mg/l was used, the total risk is 6 x i0"5,
well below the EPA threshold for acceptable cancer risk In
case "b". the total risk is 1 X 10"', when the PCB-risk com-
ponent was discarded and the erroneous higher average
TCE-concentration at 0.0895 mg/l was used. In case "c",
where the PCB-risk component was included and the cor-
rect average TCE-concentration at 0.05 mg/l was used, the
total risk is 1 x. 10^. In case "d", the total risk is 2 X 10""*,
when ihe PCB-risk component was included and the erro-
neous higher average TCE-concentration at 0.0895 mg/l
was used.
94
May 1992
Environment! Progress (Vol. 11, No. 2)
-------
TABLE 1 • Risk Assessment (TCE Concentration = 0.05 mg/l)
Chemical
TCE
PCB-1260
TCE
PCB-1260
TCE
PCB-1260
CW (mg/l)
0.05
0.0006
0.05
0.0006
0.05
0.0006
Pathway/Equation
Ingestion/(2)
Ingestion/(2)
Inhalation/(4)
Inhalation/(4)
Dermal Contact/(6)
Dermal Contact/(6)
Intake (mg/kg-day)
1.071 x 10"3
1.286 x lO"5
4.299 x 10"3
rion-volatile
1.080 x 10"6
1.296 x 10"8
Weighted intake (mg/kg-day)
9.139 x 10"4
1.097 x 10"5
3.667 x 10"3
non-volatile
9.215 x 10"7
1.106 x 10"8
SF (mg/kg-day)1
0.011 (ingest)
7.7 (ingest)
0.013 (inhale)
7.7 (inhale)
0.011 (dermal)
7.7 (dermal)
Specific Risk
1.005 x lO'5
8.444 x 10"5
4.767 x 10"5
non-volatile
1.014 x 10"8
8.514 x 10"8
The total risk sum of the specific risks is approximately 1.4 x
10"4, which, rounding to one significant figure in accordance with
USEPA instructions, gives a risk of 1 x 10"4. This is within the
acceptable range for USEPA (1,4, 5).
TABLE 2 - Risk Assessment (TCE Concentration = 0.0895 mg/l)
Chemical
TCE
PCB-1260
TCE
PCB-1260
TCE
PCB-1260
CW (mg/l)
0.0895
0.0006
0.0895
0.0006
0.0895
0.0006
Pathway/Equation
Ingestion/(2)
Ingestion/(2)
Inhalation/(4)
Inhalation/(4)
Dermal Contact/(6)
Dermal Contact/(6)
Intake (mg/kg-day)
1.918 x 10"3
1.286 x 10"5
7.695 x 10"3
non-volatile
1.934 x 10"6
1.296 x lO"8
Weighted intake (mg/kg-day)
1.636 x 10"3
1.097 x 10 s
6.563 x 10"3
non-volatile
1.649 x 10"6
1.106 x 10-8
SF (mg/kg-day)-1
0.011 (ingest)
7.7 (ingest)
0.013 (inhale)
7.7 (inhale)
0.011 (dermal)
7.7 (dermal)
Specific Risk
1.799 x 10"5'
8.444 x I0"5
8.533 x lO"5
non-volatile
1.814 x 10'8
8.514 x lO"8
The total risk sum of the specific risks is approximately 1.9 x
10"4, which, rounding, gives a risk of 2 x 10"4. This falls in the
high risk category according to USEPA guidelines {1,4, J). Note
the difference in the specific risks for the TCE exposures, for all
three exposure pathways.
Environmental Progress (Vol. 11, No. 2)
May 1992 95
-------
r> n . 1 »iceiicanc*r
' 3.666 p»opU
- - EPA
THRESH OCP
fl » 1 i 10
—n- » i io4
~R.I aior*
—K.T a \9*
— R.I ¦ 10*
— ft . » ¦ 10*
—«. ~ ¦ 10*
— R.J • (8*
_Jt.» BUT4
1
Elfect of Conditions on Carcinogenic Risk Estimate
R = Carcinogenic risk ( )
Figure 1. Effect of conditions on carcinogenic risk
estimate, a) PCB-risk component discarded and
correct average TCE-concentration at 0.05 mg/l;
b).PCB-risk component discarded and erroneous
higher average TCE-concentration at 0.0895 mg/l;
c) PCB-risk component included and correct
average TCE-concentration at 0.05 mg/l; d) PCB-
risk component included and erroneous higher
average TCE-concentration at 0.0895 mg/l.
Conclusion
The calculated risk reduction which resulted from using
the corrected average TCE concentration allowed the Air
Force to place this site in the No Further Action (NFA)
status. Moreover, since the PCB concentrations for all data
points were below the method detection limit, while a
major part of the calculated risk value was due to the PCB
component in the ingestion of drinking the groundwater
supplies, further questions arose concerning the wisdom
pursuing a costly remedial action at this site. The Air
Force was considering an expensive pump and treat reme-
dial action, which would have been of little value at this
site.
Other factors considered in arriving ai the NFA deci-
sion include the absence of both TCE and PCB in all of
the other environmental media, namely air, soils, and near-
by surface water and sediments. Furthermore, the ground-
water present at the site is not used as an active drinking
water source, since the drinking water is provided to all
users both on and near this installation by the local town-
ship from an alternate water source. In addition, the clos-
est municipal well drawing from the alternate water source
is approximately three miles upstream from this installa-
tion. Since there was no PCB nor TCE detected in the
alternate water source nor in surface water, no contamina-
tion problem of domestic groundwater supplies is appar-
ent. In addition, although surface water receives discharge
from the contaminated groundwater, no contaminants were
identified in the surface water.
In summary, when performing a risk assessment, par-
ticular attention must be paid to the media of the pathways
analyzed, methods of analysis, fate and transport of the
environmental contaminants, as well as proper quality
assurance and quality control of data and mathematical
operations to insure a valid portrayal of risk. Figure 2
illustrates data quality from the perspective of the macro-
scopic view of numerous engineering firms.
Acknowledgements
The authors would like to acknowledge the invaluable
efforts of Ms. Barbara Smith-Townsend and Marilynn
Talal (currently a Doctoral Candidate at the University of
Houston) for their tremendous input during their most dili-
gent review of this paper.
Notation
AT = time interval over-which the exposure is aver
aged, days
BD = average additional exposure time in bathroom
after shower, hours
BV = volume of bathroom, m3
BW = body weight (70 kg for average adult)
CA = concentration of chemical in air, mg/m3
CF = conversion factor (I liter/1000 cm3 or I
liter/106 mg)
CW = concentration of chemical in water, mg/l
ED = exposure.duration. years
EF = exposure frequency, days/year
ET = exposure time, hours/day
Intake = dose intake over the lifetime interval, mg/(kg-
day)
fR = intake rate (2 l/day in drinking water; 20
m3/day in breathing air)
MF = mass flux, mg/hr-cnr
"Worldly"
Macroscopic View of
Engineering Firm
Biack-Box~
Microscopic View
FIGURE 2. Quality in perspective.
96
May 1992
Environmental Progress (Vol. 11, No. 2)
-------
PC . = chemical-specific dermal permeability con
stant, cm/hr
PCB = polychlorinated biphenyl
R = total risk
Rj. = the exposure route/chemical specific risk
SA = skin surface area of exposure, cm2
SD = duration of shower, hours
SV = volume of the shower stall, m3
SF = slope factor (carcinogenic potency factor), kg-
day/mg. SFs are obtained from EPA's
IRIS(Integrated Risk Information System)
SF|j = slope factor for exposure pathway i and chemi
cal j
TCE = trichloroethylene
VW. = average amount of water used during a shower
(100 liters)
Subscripts
i =1,2, and 3 for the exposure pathways (ingestion of
drinking water, inhalation during-a shower, and der
mal contact from taking a bath, respectively)
j = 1, and 2 for the chemicals (TCE, and PCB, respec
tively)
sh = during the shower process
Literature Cited
1. USEPA. Risk Assessment Guidance for
Superfund - Volume I - Human Health
Evaluation Manual (Part A) - Interim Final. U.S.
Environmental Protection Agency - Office of
Emergency and Remedial Response. EPA/540/1-
89/002 (December, 1989).
2. USEPA. Risk Assessment Guidance for
Superfund - Volume II - Environmental Evaluation
Manual- Interim Final. U.S. Environmental Protection
Agency - Office of Emergency and Remedial
Response. -EP A/540/1 -89/001 (March. 1989).
3. USEPA. Ecological Assessment of Hazardous
Waste Sites: A Field and Laboratory Reference. U.S.
Environmental Protection Agency - Environmental
Research Laboratory. EPA/600/3-89/013 (March. 19S9).
4. USEPA. Superfund Public Health Evaluations
Manual. U.S. Environmental Protection Agency -
Office of Waste Programs Enforcement. ICF
Incorporated (October, 1986).
5. USEPA. Superfund Exposure Assessment
Manual. U.S. Environmental Protection Agency -
Office of Remedial Response EPA/540/1-88/001
(April, 1988).
6. USEPA. Guidance for Conducting Remedial
Investigations and Feasibility Studies under CERCLA.
U.S. Environmental Protection Agency - Office of
• Solid Waste Emergency Response. October, 1-988.
7. Handbook to Support the Installation Restoration
Program (IRP) Statements of Work for Remedial
Investigation/Feasibility Studies (RI/FS) (Version
3.0) (May, 1989).
8. Symms, K.G., "An Approximation of the Inhalation
Exposure to Volatile Organic Synthetic Organic
Compounds from Showering with Contaminated
Household Water", paper presented at the Symposium
of American College of Toxicologists. 15 November.
1985.
9. Tsuruta, H., Ind. Health 16: pp. 145-148 (1978) as cited
in Health and Safety-Executive Monograph:
Trichloroethylene #6 p. 3 (1982).
Environmental Progress (Vol. 11, No. 2)
May 1992 97
-------
A REPRINT FROM THE JANUARY 1990 ISSUE OF
AIChE JOURNAL
OXYGEN TRANSPORT IN BIOFILM ELECTRODES
FOR SCREENING OF TOXIC CHEMICALS
David K. Goldblum
Steven E. Holodnick
Khalif H. Mancy
School of Public Health
Univers ty of Michigan
Ann Arbor, Ml 48109
Dale E. Briggs
Department o! Chemical Engineering
Uriversity of Michigan
Ann Arbor, Ml 48105
-------
Oxygen Transport in Biofilm Electrodes for
Screening of Toxic Chemicals
A biosensor electrode system with unique configuration and a thin
layer of immobilized yeast cells, set on the surface of an amperometric
oxygen membrane electrode, was developed for rapid screening of
toxic chemicals in a variety of pollution and process control applica-
tions. Measurement is based on the instantaneous detection of changes
in oxygen respiratory activity of biofilm of yeast cells upon exposure to
toxic chemicals..
The design of this electrode system, referred to as biofilm elec-
trodes. was based on a mathematical model of oxygen transport in the
biofilm and the electrochemical current response. The biofilm, which
consists of three sublayers—boundary layer, filter pad, and yeast cell
layer—was modeled as a one composite diffusion layer, or three sepa-
rate layers in series. While the three layer model is more theoretically
complete, the one layer model was more reliable and simpler to use.
David K. Goldblum
Steven E. Uolodnick
Khalil H. Mancy
School of Public Health
University of Michigan
Ann Arbor, Ml 48109
Dale E. Briggs
Department of Chemical Engineering
University of Michigan
Ann Arbor. Ml 48105
Introduction
Biological oxygen uptake rates of suspensions of microor-
ganisms have been conveniently determined by using closed-cell
(reactor) respirometers, as shown in Figure la. This Is based on
in-siiu monitoring of dissolved oxygen by amperometric mem-
brane electrodes. Since this is a closed system from the atmo-
sphere, the test period is limited by the amount of oxygen in test
medium. Once the dissolved oxygen is consumed, respiration
rate monitoring will terminate. In certain applications, it is
desirable to monitor respiration rates over extended periods of
times. This can be achieved by using a reactor open to the atmo-
sphere. as shown in Figure lb. Under these conditions, respira-
tion rate monitoring can be extended by replenishing oxygen
from the atmosphere (Goldblum, 1988). Further optimization
of this system led to the development of the biofilm electrode,
shown in Figure Ic, where the microorganisms form a thin layer
separating the oxygen electrode from the test solution. The oxy-
gen concentration in the test solution is maintained at equilib-
rium with the oxygen atmospheric partial pressure. With this
configuration, the electrode response indicates the oxygen flux
across the biofilm, which under controlled experimental condi-
tions, is solely dependent on the biological oxygen uptake rate.
Electrode systems of this type are referred to in this article as
Current address D. K.Goldblum USAF. Occupational and Environmental Health Labora-
tory Brooks AFB. TX 78235.
Correspondence concerning this paper should be addressed to D f*. Rriggs.
biofilm electrodes (BFE). The biofilm can be made of enzymes
(Davis, 1986; Lowe, 1985; Mancy, 1984), organelles such as
mitochondria (Haubcnstrickcr, 1984), bacteria (Holodnick,
1988), yeast, and mammalian cells (Goldblum and Holodnick.
1988).
The BFE system under investigation consists of an ampero-
metric oxygen membrane electrode, also known as a dissolved
oxygen electrode (DOE), and a biofilm of immobilized yeast
cells. This electrode system shows great commercial potential in
both industrial and regulatory applications, largely based on its
ability for rapid screening of toxic substances at a great saving
in time and expenses. Details of the BFE are illustrated in Fig-
ure 2. Oxygen transport in such BFE systems is the physical
phenomenon that renders it advantageous as a biosensor. Expo-
sure of the BFE to a toxic chemical changes the respiration rule
of the biofilm and consequently the oxygen concentration at the
DOE. In this article, mathematical models for oxygen transport
through the immobilized biofilm and the BFE difTusion current
equation arc presented. These models serve as the bases for the
optimal design of BFE systems.
Model Description
The composite biofilm in a BFE is physically viewed as three
distinct layers: I. boundary layer; 2. filter pad (polysulfune
pad): and 3. yeast cell layer. The composite biofilm can be mod-
eled as a one-layer volume element, or the volume element can
AlC'hE Journal
January 1990 Vol. 36, No. 1
19
-------
(a)
Cells
+
Support Medium
(b)
Gases
Exchange
If
Celts
+
Support Medium
(c)
Gases
Exchange
g ¦ a a
Rinfilm \s///sS/s
Support Medium
Figure 1. Biosensor systems.
be modeled, as. three scparalc layers in series with respiration
confined to the cell layer.
Tlic: oxygen mass balance on a. volume element is used to
derive a concenlralion profile (C) as a function of the diffusion
pa*h distance (:] Trom (he bulk, solution towards I he oxygen sen-
sor, where r = 0 allhe boundary layer /bulk solution intcrfi.ee
and - « .b at the plastic mem brsrey yeast cell layer interface.
From the concenlralion |C], the concenlralion at the plastic
membrane/yeast cell layer interface (Q.]. i.e., C at z = b, is
obtained, and cellular respiration rale per unit volume, k is
determined.
Mode) Formulations
The DOl: current output (;) is proptjrtionaI to the oxygen
concentration iCt) at the plastic membrane interface (2 - ft)
( Mancy el at.. 196 2).
r = i.'.Ci
where rp is ihc electrode sensitivity coefficient
is
4> ¦¦
(i)
(2)
Cathode
////////////
Electrolyte
ii.
S
0>
-o
o
-fc <
u
.22
UJ
|
«+-
o
m
Membrane
—X
T
H
-Hp-
yeost
(~ 5fi. thick)
Filter Pad
T
150,11
Boundary Layer
50a
1
and
Figure 2. Composite structure of DOE biofilm.
"FAPm
(3)
An oxygen mass balance for any volume element contains
some or alJ oT the following:
A-cctimulalion(/ICCI = influx(7)
- effluxtfl + gcneration(G]
Since lhe BFE responds rapidly and is essentially in steady stale,
accu rim la Lie n vanishes, i.e., ACC - 0. The influx represent
oxygen -diffusion inio (he volume element al : = whereas
efflux represents oxygen diffusion out of the volume element at
z — 2 +¦ Az, where I he volume's cross-sectional area is (hat of
the working eleclrode surrace [A). Diffusion is governed by
Pick's First Law and is assumed 10 be the sole mechanism or
oxygen Iransport.
The gencralio-n lertn rcipreseau the wygen
-------
Three-layer model
The cell layer in ihe three-difTusion-layer model is compara-
ble to the composite biofilm in the one-layer model. The concen-
tration profiles in the boundary layer and filler pad are linear,
since the generation term in the oxygen mass balance is zero.
Steady state is assumed so that the accumulation term is zero in
all three layers. Hence, the differential equations from re-
spective oxygen mass balances are:
Boundary Layer:
d:CSl
dz1
(7)
(c) Cell Layer
p
c - ct - cs - -=
cA— + + c,
Dw D
- - a - V,
D< .
~[Li -<=-'
Interface Concentrations:
(a) Boundary Layer/Filter Pad
P C
Cj\ = Cs — ~7~7~r ^
(15)
(16)
Filler Pad:
Cell Laver:
d_%
dz1
d-C, _ kN_
~¥.~TC
(8)
(9)
Boundary Conditions:
At r = 0. C = CBl =¦ Cs
At r - 5. the continuity of concentration and flux at boundary
layer/filter pad interface results in C = C3L = Cf = C[u so that
-D,
dC,
dz
= -D„
dCffi
dz
(10)
(b) Filter Pad/Ceil Layer
Cc-
PmCt fj_
D.
(c) Cell Layer/DOE Plastic Membrane
¦_ PmCk I 6
C. — — +
D»- £>,
kNL;
i +
P„L-
LPD,
Solving for k, Eq. 19 is obtained
20,
NL]
Cs
PmC„ I 6
Lp \Dh
Ct 1 +
(17)
(18)
(19)
At r = 6 + Lf. the continuity of concentration and flux at fil-
ter pad/ccll layer interface results in C = Cf = Cc = Cp so that
-O,
dCf
dz
:-i+L,
(M)
Al r - b. the fl ux at cell laycr/plastic membrane interface
results in
dCr
°'if
lead inn to
- iI(nFA ) = tpCb/{nFA)
nFAP„
u
C\
nF A
dSi
dz
PJ.\
i-JX:
C9L - Cs
P„CV
PjCh
(12)
where h = o + Lf + Lc = thickness of composite biofilm
Concentration Profiles:
(a) Boundary Layer
(13)
From the three-layer model, k can alternatively be estimated
from the boundary condition stated in Eq. I I. i.e.. continuity of
flux at the filter pad/ccll layer interface. Hence, from Eqs. I I.
14 and 15, k can be represented as
P„ (Ck - Q
L, \ :\L.
n FA'VL,
(20)
where
r* - steady-state current: for the control runs and ?> for the
test runs. A or ^A
Derivation of k from (he boundary condition stated in l:q. I 1 is
discussed by Goldblum (1988).
One-layer nmdel
In the one-layer model. Eqs. 9 and 12. apply to the entire bio-
lilm. which is considered as one homogenous diffusion layer. In
these equations, Cr is replaced by Cand Dr by O to give Eqs. 21
and 22. The mass balance thus becomes:
d:C
dz-
k:V
T
(21)
(b) Filter Pad
c ~ CI -
Cs -
PmCt( 6
L. \D
(14)
Equation 21 is a simple second-order dilTerential equation sub-
ject to the following boundary conditions:
Al.-~0.C-G
Al ~ h. Flux - i/{tiFA) - -DdCjdzl.,
AlC'hF. Journal
January 1990 Vol. 36, No. I
21
-------
From Eqs.l and 3.
Solvinu for C:
—
d: L
P£*
L„D
C- Cs -
At - - b, C = C» and
PmCt
L„D
kiVz
TD
(2b - :)
(22)
(23)
Cc-
C„
kNb
~2D
PJb
L,D
(24)
The oxygen concentration C4 is observed from the steady-state
current so that the cellular respiration rale, k can be computed
from (£q. 24,
2D
M3
Cs - c
pmb'
M>/J
(25)
Determination of Parameters for Models
The oxygen concentration in the bulk solution at saturation,
Cs. is obtained from Oj solubility charts at a given temperature
and pressure (Hitchman, 1978), and the cell number, N, is
determined with a hemocytometer. In the diffusion models pre-
sented, only the ratio PmjLp needs to be evaluated. From is and
7S. the electrode sensitivity coefficient is computed, using Eq. 2
md P„/Lp is obtained from Eq. 3. The parameters b and D are
evaluated experimentally via diffusion studies. The cellular res-
piration rate per unit volume, k must be evaluated, but is
strongly coupled with the respiration experiments so that it must
be handled differently from the other parameters.
Permeability of plastic membrane
The thickness of the plastic membrane Lp was 0.00109 cm
and was measured using a micrometer under a light microscope.
The plastic membrane thickness showed a CO V of 4.19% over
20 measurements. The measured surface area of the electrode
was 1.327 cm2. Over 37 data runs, the average ratio P„/Lp was
H.615 x I0"5 cm/s, (COV ~ 19.1%, 95% C/ -- 12.8% mean PJ
Lp). The observed permeability coefficient (Pm) for this plastic
membrane (Reynold's plastic polyvinyl chloride film) was
9.357 x I0"8cnr/s.
Thickness of composite bioflm
The thickness of the filter pad was 150 >tm. The mean yeast
cell dimensions were obtained from 50 measurements using ah
optical digitizer under oil immersion microscopy. The short and
long dimensions (mean 11 std. dev.) were 4.56 ± 0.71 and
5.49 - 0.64 ^m. respectively, which agrees well with results
obtained by Bencfield and Molz (1985). Based on the number
(~3 x I06) and estimated volume (6.56 x I0~"cmj/ecll) of in-
dividual cells, the yeast should be deposited on the filter pad in a
single layer with a voidagc of approximately 10%. Virtually all
filtered cells are retained as the filler pad has an effective pore
size of 0.45 jim.
The boundary layer thickness was estimated by comparing
the current readout of the DOE and BFE with no cells at stir
rates of 340 rpm and 440 rpm. The current changed with rpm.
but leveled off by 440 rpm. It is assumed that the boundary layer
is negligible at the higher stir rate so that the current readout is
indicative of the saturation current. The lower stir rate yields
less current than the higher stir rate, due to an increased
boundary layer thickness. Thus, the oxygen concentration at the
plastic membrane interface with the filter pad is going to be
slightly less at 340 rpm. Using Fick's First Law and equating the
oxygen diffusivity in the boundary layer to that in water, the
boundary layer thickness can be estimated from
Flux - —-— - —Dw —
nFA dz
-D,
&CaL
'¦ &
(26)
where
i - steady-state current at low stir rate, A
- oxygen concentration difference across the boundary
layer (proportional to the difference in current between
the high and low stir rates), mol/cm3
Dw -= diffusivity of Oj in aqueous media, cm2/s
Dw is taken to be 2.84 x 10~scm2/s, in which Dw at 25°C is
2.5 x I0"!cmj/s (Perry and Chilton, 1973) so that the Wilke-
Chang empirical relation, a modification of the Stoke-Einstcin
relation (Bird et al„ I960; Perry and Chilton, 1973; Reid et al.,
1977) is used to correct for the temperature from 25°C to 30°C
(the test media temperature). Ten estimates of-the boundary
layer thickness resulted in a mean i - 50 Mm with a COV of
35%. Estimates of the boundary layer thickness ranged from 27
to 83 A»rn, the 95% CI being about '/«the magnitude of the mean
5. Thus, the thickness of the composite biofilm can be considered
to be 205 Mtn.
Oxygen diffusivity
The overall 02 diffusivity for the composite biofilm layer was
measured at the end of each run by completely inhibiting the
yeast respiration (killing) with 100 ppm KCN. Using Fick's
First Law for the composite layer.
Flux
nFA
dC
I —
dz
¦O-
AC
(27)
where
AC' • C\ ¦ Cs O; concentration gradient across the com-
posite biolilm after kill-off
Since
A - 0,/is )CS (28)
and Ck - Ct when k - 0, i.e., no respiration, then
C' " I + (P„b/LrD) (29)
From an average of 40 data runs, a mean diffusivity of 3.17 x
12
January 1990 Vol. 36, No. 1
AlC'hE Journal
-------
10 'cnr/s was obtained in which the COY was 39.7^, and the
95" CI about I i^c of the mean value of D.
The O; difTusivity in the filter pad is measured independently
by comparing the current output of a DOE with and without a
filter pad at the high stir rate. From the measured flux, concen-
tration difference, and the filter pad thickness {Lf), one can
solve for the O, difTusivity in the filter pad (D/) by Fick's First
Law (analogous to Eqs. 26 and 27).
Flux
i
nFA
dC
¦D'*
-D,
AC
interface
(30)
where
i - steady-state current with the filter pad at high stir rate,
A
AC - oxygen concentration difference from the bulk solution to
the plastic membrane interface, i.e., across the filter pad,
mol/cm1
Lf - 0.015 err (150 /im) and Dr - 2.879 x ICT'cm'/s (n - 7,
22.7% coy).
There was no independent means of measuring the02 difTu-
sivity of the cell layer (Df) or its thickness (£.,). Thus, it was
obtained by considering the difference between Lhe total diffu-
sion (mass transfer) resistance (b/D) and the sum of the boun-
dary layer and filter pad diffusion resistances (if Dw + LsjD;).
Since the three layers are diffusion resistances in series so that
the individual resistances add up to the total diffusion resistance
of the composite biofilm (bj D).
L,/Dc - b/D - 5/Dw - LjJDf
where LrjDr was computed for each run by Eq. 31.
Thus.
(31)
D
k + k + JL
O + D,~ Dw
(32)
so thai
Or -
Or
_b_
K-
(33)
DilTusivities obtained for L, - 5 ium were: D 3.170 x
10 'enr'/s and D, - 4.626 x 10 'cnr/s. The COV for LJDe
was 135'". The cell layer diffusion resistance can be neglected,
.since u is small compared to the filter pad diffusion resistance.
Thus, uncertainly in the filter pad diffusion resistance can easily
overshadow the cell layer resistance. Taking the limit as LrjD,
goes to 0. Eq. 32 reduces to
D -
(34)
Or n.
The overall difTusivity, D is calculated to be 3.698 x I0~*cm'/s,
which agrees with previous three-layer calculations. Hence, D
can be taken as 3.698 x 10 "em'/s. since there is no indepen-
dent means for obtaining a more definite assessment of Lt or Dt.
From this difTusivity. an estimate for Dr was obtained for Lc - 5
*im; using Eq. 33, and Dr vas calculated as being 3.173 x
10~'cmJ/s.
The overall difTusivity for the three-layer model with cell
layer diffusion resistance neglected appears to yield a more rea-
sonable estimate of the cell layer diffusivity, since the magni-
tude of Dc more closely agrees with D#.and Df. However, for the
one-layer model, D is computed for each run, and for the three
layer model Dc is computed for each run. Li et al. (1988) report
an overall difTusivity of 2.18 x 10"scm!/s in a biofilm of kappa-
carragecnam mounted on a DOE. Presumably, the fibers in the
filter pad offer more resistance to 02 mass transfer than the car-
rageenam gel containing the cells.
Oxygen concentration at plastic membrane interface
The C„ is computed from the is value and the steady-state
current as
Ct - )C,
(35)
This is done for both the control and test data in each run, it
being computed from Eqs. 19, 20 and 25.
Experimental Methods
Wild strain bakers' yeast (Saccharomyces Cerevisiae, C276
a/a) is inoculated in acetate growth media and mixed in a 30°C
incubator for 36-40 hours. Haubensticker (1984) shows that
yeast cells show the same metabolic activity for incubation peri-
ods of 12 to 48 hours, as long as the yeast population is in the
exponential (tog) growth phase. Therefore, calibration curves,
obtained for a given biosensor over time and with different yeast
loadings, arc the same. Hence, a sensor need not be calibrated
w ith each unknown sample.
Cell growth is halted by placing the culture .in an ice-watcr
bath, at which time a cell density is assessed with a hcmocytom-
eter (Spencer Bright-line 0.1 mm, American Optical Co.) cou-
pled with a Zeiss microscope (10 x oc/10 x obj). The growth
medium is made up by mixing 18.2-g Na acetate, I.I 5-mL gla-
cial acetic acid, and 14-g of yeast Nj base (Difco Laboratories.
Detroit) with sufficient distilled water to make 1 L of such a
growth buffer-solution at a final pH - 5.8. The cell culture is
diluted with pH 5.8 acetate buffer (growth media minus jeast
N. base), and 3-5 x 10* cells (-3 ml.) are suction-filtered over
the central I .l-cnr area of a 0.45-^im GA-6S polysulfone filter
membrane having a diameter of 25 mm (Gclman Science. Ann
Arbor). After filtration, the damp filler is inverted onto the head
of a galvanic DOE and retained with a cap assembly, yielding a
BFE. The BFE is inserted in a chamber containing 55-mL air-
saturated acetate assay medium (buffer) stirring at 340 rpirt
and being maintained at 30°C. The BFE current output attains
/„ in 45-50 min. This i„ current datum is then used to compute
an oxygen level al the plastic membrane interface (Ct in Eq. 35)
and a respiration rate per cell (k in Eqs. 19. 20 or 25).
The two chemicals studied were potassium cyanide (KCN), a
strong respiratory inhibitor, and 2.4-dinitrophcnol (2,4-DNP), a
strong respiratory uncoupler (Lehninger, 1975). One mL of a
standard KCN or 2,4-DNP solution (prepared 55x strength) is
added to lhe assay medium, for a second sleady-statc current. if.
Afier this steady slate is attained, C4from Eqs. I and 2 and k for
AlChE Journal
January 1990 Vol. 36, No. 1
23
-------
the te*t are «jV>i-JIncd- At ilis""¦-¦ini. 10C-p?m KCN is added if
cumplctcls tnhi'jit respiration, thus giving the kiU-. -he i la,.r need (.Ejq. 1S3. arise
boundary coradriion at ttrc jilicr pad/cdf layer interface f&q.
-0). thec.tprc^sijn for RCR simplifies to
RCR
I,'
I.
o
U&3
Thus, RCR is the same regardless of which model is used it
compute k- For further elaboration of this, sec Goldblunn
I IWS).
furthermore, RCR is independent of all parameters other
than respiration rate: i.e.. plastic membrane properties (Pm and
f-,1. composite oiofilm properties (D and b), and cell layer prop-
erties l/J.and l.e). This is readily seen when E<1- I is used to rep-
resent RCR as
RCR
0
S0.9 ^ 2.J
13 t z 2.8
c ov
¦OS
l.jO
a.j.v
Tbc RCR values are fcpojlcd as a mean ; 1 SD. wilh ihe COV expressed 1'
percentage. Square brackets denote chctntcal conccriratton
So that the LlflD, will cancel out in the test la control ratio for
is]liC5:jri£*(br cJic-con-c1 ^rc1 (est iai .ai
firc«"iously mentioned.
KCN Results
Three KC'N concentrations tOQI. 0.1 and 1.0 pnsn) were
tested in tjuadruplicate using cells from J single culture. Tpe
results are shuwn ir Table,I.
There was li high degree of correlation between RCR and
KCN concentration, tne correlation coefficient, / » - 0 9S-iS
(Figure J). The linear regression of the dose response curve was
obtained, using the mean RCR. The best fit curve can be
c.ipres.sed as
RCR » 93.05 - 60.46jK.CNJ
This relation indicates a 60'^ reduction in respiration per ppn)
(mg/Lj chance in KCN concern rat ion. The limit of detection is
between 9.01 andO ( ppnv (rru./LI KCN (sec Figcrc }|.
since r, = ifiC, (Eq. 28). ij - (iCJtest. and i„ - [control.
Note. [C\ -¦ f;.lc„ - C\ - Giles; aid lt'; - f.l,,,^ - -
C^lconvrol, since C, is at the sumc osygon level ji ihe cell layer-
plastic membrane in'*.er!'acc after the bioliliit is killed oil with
MXJppin KCV
f:or the one-Jayer model, il J'ollO"'.* from and 29 that
IC, CO
s,w
T(T
(3a j
su that the !r/2f> will cancel out in the ratio of Ihe test value for
'Cj'- Ct) to the control value Tor |C't - CB). For the llircc-
i.ner model it follows from Fijs. IX. 29 and 31
-------
Table 2. Effect of 2,4-DNP: n - 4
[2.4-DPN], ppm
RCR
COV
7
102.2 ± 2.2
2.17
14
102.1 ± 1.0
0.99
28
103.5 ± 1.8
1.69
42
104.9 ± 0.7
0.69
The RCR values arc reported as a mean ± I SD, with the COV being expressed
as a percentage. Square brackets denote chemical concentration.
2,4-DNP Results
2,4-DNP was evaluated at four concentrations (7, 14, 28 and
42 ppm) in quadruplicate using cells from a single culture. The
results are presented in Table 2.
The RCR correlated well with the concentration of 2,4-DNP,
r - 0.9727 (Figure 4). A linear least squares regression analysis
yields the following expression:
RCR - 101.28 + 0.08354[2,4-DNP]
For the. 2,4-DN P analysis, there was a 3-4% increase in respi-
ration with approximately 40 ppm (mg/L) 2,4-C>NP. Thus, 2,4-
DNP stimulates respiration at the tested concentrations greater
than 1 5 ppm.
One-layer model
Limits of detection were further analyzed for each chemical
discussed according to paired /-tests between k values for the
control and test sets. Paired /-tests on k and kN were carried out
where the KCN sets were individually compared with the
respective control set (3 degrees of freedom, since n - 4).
Control vs. Test />
0.01 ppm KCN 1.7057 (/> >0.1)
0.10 ppm KCN 3.4486 (p < 0.05)
1.00 ppm KCN 4.6034 {p < 0.02)
Thus, the limit of detection was approximately 0.05-0.08
ppm KCN, since the 0.01 ppm KCN set was slightly below the
significance cutoff, and 0.1 and 1.0 ppm KCN were clearly
above the limit of detection. Furthermore, paired /-tests were
done comparing the control sets for all three KCN concentra-
tions. There were no significant differences among any of the
control sets indicating that statistical differences can be attrib-
uted to the efTect of KCN on yeast cell respiration.
Paired /-tests were done comparing the individual 2,4-DNP
test and control sets, (3 degrees of freedom, since n - 4).
Control vs. Test
7 ppm 2,4-DNP
- 3.0928 (p > 0.05, borderline)
14 ppm 2.4-DNP
- 1.1526 (p > 0.2)
28 ppm 2,4-DNP
-5.9295 (p < 0.01)
42 ppm 2,4-DNP
-4.8315 (/>< 0.02)
The limit of detection was approximately 10-20 ppm 2,4-
DNP, since both 28 ppm and 42 ppm were clearly above the
limit of detection, whereas 14 ppm was below the limit of detec-
tion and 7 ppm was borderline. Paired /-tests among the dif-
ferent conirol sets showed no significant differences so that all
the statistical differences can be attributed to the different 2,4-
DNP concentrations.
Three-layer model
' Paired /-test scores for the k values will now be shown for the
three-layer model (k computed by Eq. 19) as well as for the
analysis based on the boundary condition at the filter pad/cell
layer interface (k computed by Eq. 20).
For the KCN series, the following /-scores were obtained,
where the KCN sets were individually paired with their
respective control set (3 degrees of freedom, n = 4). The paired
/-test on k computed from Eq. 19 (three-layer model) shows the
following:
Control vs. Test
0.01 ppm KCN 0.2588 {p > 0.8)
0.10 ppm KCN 1.1387 (p > 0.2)
1.00 ppm KCN 1.3373 {p > 0.2)
The paired /-lest on k computed from Eq. 20 (filter pad/cell
layer interface boundary condition) shows the following:
Control vs. Test
0.01 ppm KCN 1.8568 (/>> 0.1)
0.10 ppm KCN 5.5599 (p 0.02)
1.00 ppm KCN 4.8305 {p < 0.02>
900 5 10 15. 20 25 30 35 40 45
ppm, 2,4-Dinitrophenol
Figure 4. RCR (% of control) vs. 2,4-DNP concentration.
Based on the Eq. 20 analysis, the limit of detection is 0.05-0.OS
ppm KCN. From the F.q. 19 analysis, the limit of detection is
uncertain.
For the 2.4-DN'P series, the following /-scores were obtained,
with the 2,4-DNP sets being paired individually with the
respective control set (3 degrees of freedom, n = 4).
The paired /-test on k computed from Eq. 19 (three-layer mod-
el) shows the following:
Control vs. Test /,
7 ppm 2,4-DNP
14 ppm 2,4-DNP
28 ppm 2,4-DNP
42 ppm 2.4-DNP
-3.4672 (p < 0.05)
0.8432 {p > 0.4)
-3.5865 (p< 0.05)
- 0.4247 (p > 0.6)
The paired /-test on k computed from Eq. 20 (filter pad/cell
AlChE Journal
January 1990 Vol. 36, No. 1
25
-------
Table 3. Variation inthe BFE Parameters
Parameter
'Mean
Std.
COV.%
95 0.2)
-6.9192 (/> < O.OI)
-6.0462 (p < O.OIJ
Based on Ihc Eq. 20 analysis the limit of detection is 10-20 ppm
2.4-DNP.The limit of detection based on the three-layer model
is ill defined, ranging from 7-50 ppm.
The f-scores or the paired r-tesis are different for the three
different analyses on k. This becomes clear when the ratios or k
computed among the three analyses are considered.
ir
k0 - k computed from Eq. 20, the boundary condition concern-
ing flux continuity at the filter pad/ccll (aver interface in
Eq. I I
fci - k computed from Eq. 25, using the one-layer model to esti-
mate the respiration per cell
k3 - k computed from Eq. 19, using the three-layer model lo
estimate the respiration per cell
then the following ratio expressions are obtained:
h 2Cst*
K- ' *
-------
Diffusion Path
Diffusion Path
Figure 5. Concentration profiles of the one-(top) and
three (bottom)-layer diffusion models.
BL - boundary layer: FP - filler pad; CL - cell layer
el. C is linear tvilh respect loz with a uniform slope when ihere is
no respiration (k - 0), and quadratic when there is respiration
(A- > 0). However, for the three-layer model, C is linear with
respect to j in both the boundary layer and filler pad regardless
of respiratory state, the slopes differing because of different dif-
fusivilies in the two layers. In the cell layer of the three-layer
model. C is linear when there is no respiration and quadratic
when there is respiration.
The mathematical models presented in this article can be
used for the optimal design of BFE systems. In this study, exper-
iments were conducted with single toxic substances. In many
upplicutions, ihetesl sample may include mixtures of chemicals.
Future studies should be directed towards modeling of BFE
responses to mixtures of toxic substances of different potencies.
C's - O; concentration in the biofilm under air-salu^aiiun condi-
tions. mg/L or mol/em'
Ct - confidence interval, usually 95% confidence interval
C£."-celt layer
COV - coefficient of variation - ratio of the standard deviation to the
mean, %
D - diffusivity of 02 in the composite biofilm, em'/s
Dr - diffusiviiy of O, in the cell layer. cm!/s
Df - diffusivity of O, in the filter pad, em'/s
Dr - diffusivity of 02 in the boundary layer, taken to be that of
aqueous media at the appropriate temperature, cm!/s
F - Faraday's constant - 96,490 C/equivalent
FP - filler pad
i - current readout from the electrode, A or jiA
it - current readout from the electrode at steady slate which is i„
for the control datum and lyfor the lest datum in Eq. 20, A or
•mA
if - current readout from the electrode after the steady state is
auained after a specific dose of a toxic chemical is added lo
the test medium, A or ft A
i, - current readout from the electrode after steady state is
attained with a nonviable biofilm, A or ^A
ij - current readout from the electrode when the aqueous solution
is air-saiurated, A or mA
i'jj - current readout from the electrode after steady state is
auained before toxic chemical is added to the test medium, A
Or.nA
k - maximum specific or cellular 02 utilization (consumption)
rate via cell respiration pcf unit volume, mg/(L • s • cell)
Jt„ - k computed from Eq. 20 (flux continuity at the filler pad/cell
layer interface in Eq. 11), mg/(L - s • cell)
i, - A computed from Eq. 25 (one-layer model), mg/(L • s
eel!)
ky - k computed from Eq. 19 (three-layer model), mg/(L ¦ s
cell)
Ks - Monod constant - concentration at which respiration rate
per unit volume is half the maximum O? consumption rate
(A/V), mg/L or mol/em'
S - thickness of boundary layer, cm or um
Lc - thickness of cell layer, cm or
Lf - thickness of filler pad. cm or jjm
Lp - thickness of plastic membrane, cmoriiiri
n - number of equivalents per mole of 0_, (a - 4); sample size in
the various analyses
jV - number of cells in the biofilm
P„ - permeability of plastic membrane, cm:/s
R - generation rate of 0; per unit volume, mg/(L - nlin) or
moles/(cm' ¦ s)
RCR - respiratory control ralio - ratio of a respiration parameter for
the lest to the control run, %
RCRl - respiratory control ratio based on the parameter k
RCRliV - respiratory control ratio based on the parameter A;V
SD - standard deviation
i » lime, s or min
/i < /-score in the paired r-lest with 3 degrees of freedom [n -• 41
r - axial distance into biofilm awav frnm the aqueous bulk solu-
tion (diffusion path), cm or win
Notation
A ~ cross-sectional area of the biofilm. ctn:
b - biofilm thickness, cm or urn
BL - boundary layer
C - O* concentration in the biofilm. mg/L or mol/cm'
C4 - Oj concentration in the biofilm at the plaslic membrane/cell
layer interface. mg/L or mol/cm1
£~sl - O; concent ration in the boundary layer, mg/l.or mol/cm'
C, - 0;concentration in the cell layer, mg/L or mol/cm1
Cf - O, concentration in the filler pad, mg/L or mol/cm'
Q, - O, concentration at the boundary layer/filler pad interface,
mg/L or mol/cm' .
Cf, - O, concentration at the filter pad/cell layer interface, mg/L
or mol/cm3
Ci - O, conceniralion in the non-viable biofilm at the plastic mem-
brane/cell layer interface, mg/L or mol/cm'
Greek tellers
!> - thickness of boundary layer, em or urn
ib •- electrode sensitivity coefficient ^ /.y/C.r. ^A/(mg/L)
Subscripts
h - cell layer/plaslic membrane interface
BL - boundary layer
c - cell layer
f - lilter pad; final currenl reading after addition of toxic chemi-
cal
/I - boundary layer/filler pad interface
fX - filter pad/cell layer interface
k - conditions after total inhibition of cell respiration
m - plastic membrane
M « Monod equation
¦J - air-saturation conditions
AlChE Journal
January 1990 Vol. 36, No. 1
27
-------
ss - final steady-state current before toxic chemical is added
W - -water (aqueous conditions apply in the boundary layer)
0 - based on boundary condition concerning flux continuity at the
filter pad/cell layer interface
1 - based on one layer model
3 - based on three layer model
Literature Cited
Bcnefield, L., and F. Molz, "Mathematical Simulation of a Biofilm Pro-
cess," Biotechnol. and Bioeng.. 27, 921 (1985).
Bird. R. B.. W. E. Stewart, and E. N. Lightfoot, Transport Phenomena,
Wiley, New York, 515 (I960).
Davis, G., "Advances in Biomedical Sensor Technology: a Review of the
1985 Patent Literature," Biosensors. 2, 101 (1986).
Goldblum, D. K., "Biofilm Electrode for Screening of Toxic Chemicals:
Electrode System Characterization," PhD Thesis, Chemical Engi-
neering and Environmental Health Science, University of Michigan
(1988).
Haubenstricker, M. E., "Development of a Toxicity Biosensor Based on
Changes in Mitochondrial Respiration Rates," PhD Thesis, Environ-
mental Health Science, University of Michigan (1984).
Hitchman, M. L., "Measurement of Dissolved Oxygen," Chemical
Analysis. 49, P. J. Elving, J. D. Winefordner, and I. M. Kolthoff, eds.,
Wiley, New York (1978).
Holodnick, S. E., "The Biofilm Electrode Sensor System for Acute Tox-
icity and Viral Screening," PhD Thesis, Environmental Health
Science, University of Michigan (1988).
Kornegay, B. H., and J. F. Andrews, "Kinetics of Fixed-Film Biological
Reactors," J. Water Pollution Control Federation. 40(1 I). Pan 2.
R460 (1968).
Lehninger, A. L., Biochemistry. 2nd ed„ Worth Publishers. New York,
519,611(1975).
Li, X. M.. B. S. Liang, and H. Y. Wang, "Computer Aided Analysis for
Biosensing and Screening," Biotechnol. and Bioeng.. 31, 250 (1988).
Liang, B. S., X. M. Li, and H. Y. Wang, "Cellular Electrode for Antitu-
mor Drug Screening," Biotechnol. Prog., 2(4), 187 (1986).
Lowe, C. R., "An Introduction to the Concepts and Technology of Bio-
sensors," Biosensors, 1, 3 (1985).
Mancy, K. H., "Development and Application of Biosensors in Pollution
Control Programs," Int. Symp. Electrochemical Sensors. Rome,
Italy (June 12-14, 1984).
Mancy, K. H„ D. A. Okun, and C. N. Reilley, "A Galvanic Cell Oxygen
Analyzer," J. Electroanal. Chem.. 4, 65 (1962).
Parulekar, S. J., G. B. Semones. M. J. Rolf, J. C. Lievense, and H. C.
Lim, "Induction and Elimination of Oscillations in Continuous Cul-
tures of Saccharomyces Cerevisiae." Biotechnol. and Bioeng.. 28, 700
(1986).
Perry, R. H., and C. H. Chilton, Chemical Engineers Handbook. 5th
ed., McGraw-Hill, New York, 3-225 (1973).
Reid, R. C., J. M. Prausnitz. and T. K. Sherwood, The Properties of
Gases and Liquids. 3rd ed., McGraw-Hill, New York, 567 (1977).
Rittman, B. E., and P. L. McCarty, "Model of Steady-State-Biofilm
Kinetics," Biotechnol. and Bioeng.. 22, 2343 (1980).
, "Evaluation of Steady-State-Biofilm Kinetics," Biotechnol.
and Bioeng.. 22, 2359 (1980).
Manuscript received Nov. 7. 1988. and revision received Nov. 2. 1989.
28
January 1990 Vol. 36, No. I
AIChE Journal
-------
A REPFffNT FROM THE FEBRUARY 1991 ISSUE OF
Wi-VB'jJl'i.'JHIl II I I' UMIMIII II
OXYGEN MEMBRANE ELECTRODE USED
AS A TOXICITY BIOSENSOR
David K. Goldblum
Steven E. Holodnick
Khalil H. Mancy
School of Public health
The University of Michigan
Anr Arbor, Ml 43109
and
Dale E. Br'iggs
College of Engineering
The University of Michigan
Ann Arbor, Ml 46109
-------
Oxygen Membrane Electrode Used as a
Toxicity Biosensor
David K. Goldblum, Steven E. Holodnick, Khalil H. Mancy
School of Public Health, The University of Michigan, Ann Arbor, Ml 48109
and
Dale E. Briggs
College of Engineering, The University of Michigan, Ann Arbor, Ml 48109
Three biosensor configurations thai included a dissolved oxygen electrode and
yeast cells {Saccharomyces cerevisiae) were evaluated as a means to assess the -
environmental effects of toxic chemicals. The configurations were: closed
suspension—assay medium closed to air and cells suspended in the medium; open
suspension—assay medium open to air and cells suspended in medium; and
biofilm electrode (BFE)—assay medium open to air but cells immobilized on the
surface of the electrode. The BFE was the most advantageous configuration
based on assay times and detection limits.
INTRODUCTION
Ttwic che-'iiicals in the environment have prompted increased
public awareness of environmental concerns. Two main issues
are environmental effects (damage), and control and manage-
ment of the risks associated with toxic chemicals. When act-
dressing environmental effects, the human health, ecology, and
abiotic sectors neecE 10 be considered. For contra! aspects, risk
of exposure and regjlaxty aspect; r.eed to ijc addressed.
T nereis an fed lo-de^dofc si ropier and more rapid procedures
farassfssi-HS environ mental effects of tonic chemicais |j| A
biofilm electrode consisting of a dissolved oxygen (DO) mem-
brane-covered electrode coupled with immobilized yeast (Soc-
charomyces cerevisiae) cells was developed in the present work.
This biosensor constitutes a simple and rapid procedure for
screening toxic chemicals in the environment based on changes
in respiratory activity of the lest cells.
The DO membrane-covered electrode under consideration
is amperornetric, i.e., it generates a current which is propor-
tional to the oxygen concentration on the plastic (polyvinyl
chloride) membrane surface. The base electrode uses a silver
cathode and a lead anode. At t.he cathode, the oxygen is reduced
David Galdfciunt is ¦Cbjrr«m(y 1 Captain, USAF, Occ-jpaliona( and
Environmental Health Laboratory Eroais AFB. TX 782.35.
to hydroxide ion, whereas at the anode, the lead is oxidized.
The DO electrode is also said to be Galvanic, since it does not
require an external voltage for Us operation. This DO mem-
brane-covered electrode was first described by Mancy, et al.
[21. In the present work, this DO etecirode will be referred to
as a DOE.
In the three configurations considered in Figure 1, the DOE
is used as the osygen sensor and indicates the oxygen level on
the surface of the plastic membrane or the DOE. The Jim
con figuration is (he dosed suspension, '.vfiere the a.=say medium
.5 closed ro the atmosphere, and the yeast «lis are suspended
.n the assay medium. The second con figuration is the open
suspension, where the assay medium is open rathe atmosphere,
and the yeast cells are suspended in the assay medium. The
third configuration is the bioflrr electrode (BFE), where the
assay medium is open to the atmosphere, and the yeast cells
are immobilized on the surface of the DOE. A detailed sche-
matic of the BFE is illustrated in Figure 2.
BACKGROUND
There has been a wi-desprea-d use of membrane-covered elec-
trode systems to.monitor ¦ar.d mst-abolites in bioiogjcal
24 February. 1S91
Environnnar;!al Pio
-------
(a)
SJESLQ.
Celts
+
Support Medium
(b)
Gases
Exchange
Ceils
+
Support Medium
(C)
Gases
Exchange
H C9 SI U
r///////i
Biofilm
Support Medium
on the DOE and ihe rale of respiration of the biofilm is de-
termined by monitoring changes in the oxygen concentration
within the biofilm, or at the interface between the biofilm and
the membrane of the DOE. Recent applications o£ the BFE
are the cellular elect rode for antitumor drug screening by Liang
and Wang {JO] as well as ihe present BFE'suse in viral screening
by Holodnick [31]. Another BFE example is the bacterial elec-
trode for determination of monomethyl sulfate [32], which
consists of the physical entrapment of the monomethyl sulfate
degrading bacteria, Hyphomicrobium. A multispecies biofihn
model was developed for computer simulation of symbiosis
and/or competition for space and common substrates between
several microbial species [33]. The concentrations of microbial
cells in suspension were determined with a graphite electrode
1*1.
THEORY
The assumptions, mass balance, boundary and/or initial
conditions, and final solution for the oxygen concentration
are considered for three configurations illustrated in Figure 1.
Oxygen excess(C»KJ) in the Monod Equation 135-37] (Equa-
tion 1) is assumed for all configurations.
kNC
RESPIRA TION RATE= -~—p, = kN. (1)
Aj + C
Moreover, the yeast cells are in the exponential growth phase
during the test runs and their growth rate follows 1st order
kinetics (Equation 2).
N=N^''
(2)
The mathematical analysis begins with an overall mass bal-
ance on. oxygen (Equation 3).
ACCUMULATION (ACC}=IN~ OUT
+ GENERA TJON (GEN) (3)
The GEN term is the negative of the Monod Respiration Rale
FIGURE 1. Biosensor Systems.
as well as environmental media, e.g., enzyme electrodes, and
biofilm electrodes [3]. Membrane-covered electrodes are those
with a selectively permeable membrane, e.g., polymeric mem-
branes, which separates the electrode itself from the test so-
lution. In the galvanic cell oxygen analyzer the membrane is
permeable to gases such as oxygen. Surface active and elec-
troactive substances, which may interfere with the electrode
reactions, are screened out. Several authors [4,J] have specif-
ically addressed the issue of selectively permeable membranes.
Oxygen has been the gas most extensively measured with mem-
brane covered electrodes [2,6-9], Other gases so measured
include carbon dioxide [6,7,10], ozone [II], and chlorine diox-
ide [12].
Enzyme electrodes are membrane-covered electrodes such
that the membrane is an enzyme layer on the surface of the
•base electrode and monitor a metabolite (substrate) through
that particular enzymatic reaction. These highly selective bio-
electrode systems have been used to monitor glucose [13-18],
urea [70], lactate [20-24], pyruvate [25], cholesterol [26], cre-
atinine [27], salicylate [.25], and 1-alanine 129].
The BFE is an extension of the membrane-covered enzyme
electrode. In the BFE. a film of microorganisms is immobilized
UJ
L_
£
CD
TJ
S<
u
£
LU
J
«+-
O
~0
® LU
0> O
° •
*"8'
> w
O O
©
'>>
O
-1
o
CD <
03
"<75
o
Q.
e
o
O
Cothode
////////////
Electrolyte
Membrana
—X
yeosf
thick)
Filter Pad
T
s
—LLm
150jj.
Boundary Layer
50u
i
FIGURE 2. Composite Structure of DOE Biofilm.
Environmental Progress (Vol. 10, No. 1)
February, 1991 25
-------
s
u
T5
o.o a i
0.2 0.3 0.4
ppcr, KCN
0.6
^f=-kN= -kNy
at
larger than the respiratory consumption by the yeast cells. The
determination of k for the open suspension, is discussed in
detail by Goldblum [3S\.
In the BFE configuration, steady state is attained Quite rap-
idly, so that ACC in the mass balance vanishes (i.e., dC/dt = 0)
and the number of cells can be taken as constant during the
test run. Thus, the oxygen mass balance becomes:
d2C
0 = D~rr-kN
dz2
00)
FIGURE X Open Suspension Response to Potassium
Cyanide (KCN),
from equation I multiplied by the volume of the system, i.e.,
GEN= -kNV. The ACC term is VdC/dt.
In the closed suspension, IN = OUT = 0, so that Equation 3
reduces to:
<4>
The initial condition is that the assay medium is air-saturated
at ( = 0, i.e., C= C, ® t = 0. If the number of cells is taken as
constant during the test run, i.e., the yeast cell growth is neg-
ligible, then the oxygen concentration as a function of time is:
The boundary conditions are that C~Ct @ z=0, and dC/
rfil{.6 = -(P^C^/iLpD), where C4=C at z = b. Note that
z = 0 at the biofilm-assay medium interface, and z= b at the
biofilm-pJastic membrane interface, and the thickness of the
bio/ilm is b. The assay medium is maintained at air-saturation
during the test run by vigorously stirring, so the outer surface
of the biofilm is assumed to be air-saturated. The second
boundary condition results from the oxygen flux at the plastic
membrane interface, which is proportional to the curTem pro-
duced by the DOE, or.
i dC
flux = —— = -D—
nFA dz
t - 6
(ii)
using Fick's Jaw of diffusion [39| along with the electrode
sensitivity coefficient () from Mancy's analysis (21.
= ~
nFAPm
(12)
C=C,~kNt
(5)
If the growth of the yeast cells is considered, then the oxygen
concentration is represented by:
It is important to note that t/> is the proportionality constant
between the current produced by the DOE and Cj,(i= $ * Cy).
Equation 10 can be solved foT C. Then, if z = &, the solution
or Cj, is:
C=C,-j^ 1)
<«>
For the closed suspension, the oxygen in the assay medium is
depleted rapidly, so that Equation 5 would apply.
In the open suspension IN-OUT=*^£(0,-0, which is
the oxygen mass transfer from the air atmosphere into the
assay medium. Thus, the oxygen mass balance becomes:
^ = *yklC,-C)-kN=^iC,-C)-kN*t'' (7)
The assay medium is air-saturated at r = 0, as is the case with
the closed suspension, i.e., C=C, @ / = Q. If the number of
cells is taken as constant during the test run, i.e., the yeast cell
growth is taken as negligible, then the oxygen concentration
as a function of time is:
" )
(8)
If the growth of the yeast cells is considered, then the oxygen
concentration becomes:
C=CS —
~jta:{e' )
*'+
(9)
There is the potential of losing sensitivity in the open suspen-
sion analysis if the oxygen mass transfer from the air is much
Ct-
C6=-
kNb*
2D
1 +
P«b
LpD
(13)
In the biofilm, the steady state condition simplifies the analysis
considerably. From Equation 13, k is computed from the ob-
served C4l which leads to:
k~-
2D
Mb2
c,-c„
LpDJ
(14)
For the open suspension and the biofilm electrode, results are
expressed as a respiratory control ratio (RCR):
150
2 125
"5 100
75
*—4 RCRfc
o—oRCRkN
b
0.0
J L__L.
25.0
50.0
J I I L
75.0
ppm, 2,4-Dinttrophflnol
FIGURE 4. Open Suspension Response to 2,4-
Dinltrophenol.
26 February, 1991
Enrironmenial Progress (Vol. 10, No. 1)
-------
RCR = RCRk = -~- (15)
or
kN
RCR = RCRkN= "" (16)
kNnnlrel
It is noteworthy that RCR is based on the cellular respiration
rate (Equation 15) or on the respiration rate of the entire
biomass (Equation 16).
For open suspension analyses, the control and test data on
respiration rates are obtained from separate runs. Although,
the experimental design is to have the same number of cells
for each run in a given series, there is variation in cell counting
as well as in the cell-assay medium suspension so that the
number of cells in the control run need not be exactly the same
as the number of cells in the test run. Thus, RCR based on k
will differ slightly from RCR based on kN (see Figures 3 and
4). For the biofilm electrode analyses, the control and test data
on respiration rates are obtained from the same run, so that
the number of cells in the test and control data is the same.
Thus, RCR based on k will be equal to RCR based on kN.
METHODS
Two test chemicals used in the present work were: (a) po-
tassium cyanide (KCN), which is a strong respiratory inhibitor;
and (b) 2,4-dinitrophenol (2,4 DNP), which is a strong respi-
ratory uncoupler [40\, i.e., it stimulates respiration. The yeast
cells are incubated [41] at 30°C in acetate buffer growth me-
dium with shaking. The acetate growth media consists of 0.0182
kg sodium acetate + 1.15 mL glacial acetic acid + 0.014 kg
yeast nitrogen base (Difco Laboratories, Detroit) in 1 liter
volume of aqueous solution (pH = 5.8). The acetate buffer
assay medium is the growth medium less the 0.014 kg yeast
nitrogen base. An inoculum of yeast is added to 75 mL of
growth medium in an Erlenmeyer flask, which is put on a
shaker for 36-48 hours for the suspensions and 24-36 hours
for the biofilm electrode. After incubation, cell counts are
taken using a hemocytometer (Spencer Brightline 0.1 mm,
American Optical Co.) coupled with a Zeiss microscope
(lOxoc/lOxobj). Standards for the two chemicals were pre-
pared using distilled water. The DOE current was converted
to voltage and recorded with a strip chart recorder.
The system was built in house for the most part. The stir
plate assembly was constructed from a turntable (Bang and
Olufson model 2400) with two preset stir rates (340 and 440
rpm). A Teflon coated stir bar was rigidly held in place at the
bottom of the center of the assay chamber, about 1.8 cm from
the magnet in the stir plate assembly. The assay chamber was
set in an isothermal bath at 30°C, using a constant temperature
circulator (Polyscience Corporation, Evanston, IL).
For the suspension test runs, it is necessary to incubate for
a somewhat longer (36-48 hours) period to assure that the cell
count has reached approximately 3 x 106 cells/mL. Without
adequate biomass, the atmospheric aeration will drive the cur-
rent to near saturation values, i.e., atmosphere reaeration would
dominate cellular respiration. If the cells grow unusually fast,
it becomes necessary to dilute the cell suspension with acetate
buffer assay media (1:3 to 1:5) to get the final cell count near
3 million cells per mL, since this condition was observed to be
experimentally optimal. In these open suspension respiration
test runs, 55 mL of yeast cell suspension was monitored for
three hours. For a given yeast cell suspension the first test run
is the control, and subsequent test runs involve progressively
higher levels of KCN or 2,4 DNP.
In the closed suspensions the oxygen in the assay medium
is depleted within approximately 20 minutes, so that the closed
suspension is of limited value, albeit its theoretical analysis is
quite simple [42].
For the biofilm electrode, it is necessary to incubate for 24-
36 hours to produce yeast cells in the exponential growth phase.
For these test runs, 55 mL of acetate buffer assay medium is
used and the yeast cells are center-filtered onto a microporous.
filter pad (Gelman Sciences, Ann Arbor, MI). The present
configuration responds well when approximately 3 million cells
are immobilized. This cell mass consumes dissolved oxygen at
a rate which yields approximately 40% of the current of the
base DOE, a condition which allows for the measurement of
respiratory inhibition as well as respiratory uncoupling. During
these test runs, the control datum is obtained after steady state
is obtained then a dose of either KCN or 2,4 DNP is added
to the assay medium. After steady state is again achieved, the
test datum for the run is obtained. At this point, 100 ppm of
KCN is added to totally inhibit biofilm respiration. In these
runs, Q is obtained for both the control and test data to give
the cellular respiration rate k for both data. After 100 ppm
KCN is added, Ck is then obtained, which is CA when Ar = 0.
RESULTS
For' the open suspension, the process for obtaining k was
complex, due to the changing response for the initial (1-5 min)
and later (130-180 min) time regimes (38). These time regimes-
result from the analysis of the time derivative of C from Equa-
tion 9.
The assay chamber surface area available for oxygen mass
transfer from the air was approximately 14.6 cm2. This was
based on the electrode having a 1-inch (2.54 cm) diameter and
the assay chamber having a 2-inch (5.08 cm) diameter. From
reaeration experiments, the oxygen mass transfer coefficient,
kL was about 0.19 cm/min. The growth rate constant, k', was
found to be 0.002 min"' in a separate incubation experiment.
From studies of control cell respiration, k was obtained for
both the initial and later time regimes. It was concluded that
k from the later time regime was more reliable, due to the
uncertainty in the initial slope and the high variability in the
k value obtained from the initial time regime. The best re-
producibility was obtained using the k' and k values in the
later time regime, and is the reason the respiration run lasts
for three hours. Furthermore, when doing a series of increasing
doses, the runs are done sequentially. Therefore, evaluating
one concentration and a control takes about 5 hours.
Open suspension tests with the chemicals resulted in a de-
tection limit of 0.2-0.4 ppm for KCN (correlation coefficient
between RCR and [KCN] > 0.99 up to approximately 0.5 ppm
KCN, see Figure 3). Respiratory uncoupling by 2,4, DNP was
not detected at concentrations less than or equal to 73 ppm
2,4 DNP, see Figure 4. Thus, the open suspension detected
the effect of the respiratory inhibition of KCN, but not the
respiratory stimulatory uncoupling of 2,4 DNP.
For the biofilm electrode, k was obtained from Equation
14. The working surface area of the electrode was 1.33 cm2,
the number of cells was about 3-4 million, and the biofilm
thickness was 205 /im. The biofilm consists of the yeast cell
layer (5 /im), the micropore filter pad (150 ^m), and the bound-
ary layer (50 fim). The biofilm is taken as one composite dif-
fusion layer, but can be taken as three diffusion layers in series.
The one diffusion layer biofilm model and the three diffusion
layer biofilm model are compared by Goldblum [43]. The
overall effective 02 diffusivity is about 3-4 x 10'^crrr/sec,
the thickness of the plastic membrane is 0.428 mil (10.86 (im),
and the permeability of the plastic membrane, Pm, is approx-
imately 10~1cm2/sec. Average values for the oxygen concen-
trations are: Cs = l AOmg/L, Ck = 4.66mg/L, and C6 = 2.79
mg/L [JS], The assay time in this test is about 1.5 hours, which,
allows for the acquisition of both the control and test data.
Biofilm electrode tests with the chemicals resulted in a de-
Envlrorimental Progress (Vol. 10, No. 1)
February, 1991 27
-------
ppm, KCN
FIGURE 5. Biofilm Electrode (BFE) Response to
Potassium Cyanide (KCN).
tection limit of 0.05-0.2 ppm for KCN (correlation coefficient
between RCR and [KCN] > 0.99 up to I ppm KCN, see Figure
5), and a detection limit of 10-20 ppm for 2,4 DNP (correlation
coefficient between RCR and [2,4 DNP] is approximately 0.97
up to 42 ppm 2,4 DNP, see Figure 6). Thus, the biofilm elec-
trode detected the effect of both the respiratory inhibition of
KCN and the respiratory stimulatory uii£6tipllng of 2,4 DNP.
activity. This configuration can be used most feasibly when
in-situ measurements are required. It is possible to have this
toxicity biosensor serve as an effective screening tool for toxic
chemicals in environmental samples, based on changes in the
respiration rate of the immobilized cells.
NOTATION
Open Suspension
AL = area available for 02 mass transfer from the atmos-
phere into the aqueous solution
C = Oi concentration in the aqueous solution
C$ = O2 concentration in the aqueous solution under air-
saturation conditions
k — maximum specific or cellular O2 utilization (con-
sumption) rate via respiration per unit volume .
kL = O] mass transfer coefficient in the liquid part of the
boundary layer at the gas-liquid interface
k' = first order growth rate constant in Equation 2
N = total number of cells in the aqueous solution
Nq = total number of cells in the aqueous solution initially
(1 = 0)
RCR = respiratory control ratio = ratio of a respiration pa-
rameter for the test run to the same respiration pa-
ramgtsr for the control r»n rvpr^sml as a percentage
CONCLUSIONS
In comparing the open suspension with the biofilm electrode
configuration, the open suspension test has advantages in its
relative simplicity in experimental set up and mathematical
analysis. The open suspension oxygen concentration is a func-
tion of time but independent of the spatial coordinate in the
assay chamber. In open suspension respiration test runs, there
is little vulnerability to variation in experimental technique.
The open suspension would have much smaller assay times if
the initial time regime could be made to show more repro-
ducibility, otherwise its assay times are considerably longer
using the later time regime. However, the biofilm electrode
has advantages in its shorter assay times and lower detection
limit capabilities, but the mathematical analysis would be rather
difficult if the steady state condition were not justified. The
oxygen concentration is a function of the distance from the
assay medium-biofilm interface, and is independent of time in
the steady state condition.
The biofilm electrode is the most advantageous configura-
tion evaluated. The BFE allows for the rapid and reproducible
measurement in inhibition and uncoupling of yeast respiratory
RCRk=RCRkN
_ 105
O
0 100
O
3*
95
QQ ft t 1 1 t 1 1 . rt . . 1 1 t , I ¦ . I . 1 t I I t t 1 ¦ t . 1 ¦ I f . 1 ¦ ¦ t l 1 . !
O 5.10 15 20 25 30 35 40 45
ppm, 2,4-Oinitrophenol
FIGURE 6. Biofilm Electrode (BFE) Response to
2, 4-Dlnitrophenol.
Biofilm Electrode (BFE)
A
C
Cb =
C* =
Cs =
D
F
i =
k
N -=
n =
Pm =
RCR =
cross-sectional area of the biofilm (actually area of
the working electrode surface)
01 concentration in the biofilm
Oj concentration in the biofilm at the plastic
membrane/cell layer interface
02 concentration in the biofilm at the plastic
membrane/cell layer interface after the biota are
killed off by 100 ppm KCN
O2 concentration in the biofilm under air-saturation
conditions
diffusivity of 02 in the biofilm
Faraday's Constant
current readout from the electrode
maximum specific or cellular O2 utilization
(consumption) rate via respiration per unit volume
number of cells filtered onto the filter pad
number of equivalents per mole of O2 reduced to
hydroxide ion (n = 4)
permeability of plastic membrane
respiratory control ratio = ratio of a respiration
parameter for the test run to the same respiration
parameter for the control run expressed as a
percentage
axial distance into biofilm away from the aqueous
bulk solution (diffusion path)
LITERATURE CITED
1. Mancy, K. H., "Development and Application of Bio-
sensors in Pollution Control Programs," International
Symposium on Electrochemical Sensors, June 12-14,1984,
Rome, Italy.
2. Mancy, K. H., D. A. Okun, and C. N. Reilley, "A Gal-
vanic Cell Oxygen Analyzer," J. Electroanal. Chem., 4,
65 (1962).
3. Davis, G., "Advances in Biomedical Sensor Technology:
a Review of the 1985 Patent Literature," Biosensors, 2,
101 (1986).
4. Kunze, K., and D. W. Lubbers, "Absolute Poj-Measure-
ments with Pt-Electrodes Applying Polarizing Voltage
28 February, 1991
Environmental Progress (Vol. 10, No. 1)
-------
Pulsing,1' Oxygen Transport to Tissue, Editors, H. I.
Bichefand D. F. Bruley, Plenum, New York,. 35 (1973).
5. Tang, T. E., V. G. Murphy, A. W. Hahn, and R. E. Barr,
"The Operation of Platinum Oxygen Sensing Microeilec-
trodes," J. of Bioengineering, 2, 381 (1978).
6. Hahn, C. E. W., "Techniques for Measuring the Partial
Pressures of Gases in Blood. Part I - In Vitro Measure-
ments," J. Phys. ESci Instrum., 13, 470 (1980).
7. Hahn, C. E. W,, "Techniques for Measuring the Partial
Pressures of Gases in Blood. Part II - In Vivo Measure-
ments," J. Phys. £.: Sci Inst rum., 14, 783 (1981).
8. Hitchman, M. L., "Measurement of Dissolved Oxygen,"
Chemical Analysis, Editors, P. J. Alving, 1. D. Wine-
fordner, and I. M. Kolthoff, 49,. John WiJey & Sons, New
York, (1978).
9. Lilley, M. D., J. B. Story, and R. W. Raible, "The
Chronoamperometric Determination of Dissolved Oxygen
Using Membrane Electrodes," J. Electroanal. Chem., 23,
425 (1969).
10. Opdycke, W. N., and M. E., Meyerhoff, "Development
and Analytical Performance of Tubular Polymer Mem-
brane Electrode Based Carbon Dioxide Catheters," An-
alytical Chemistry, 58, 950 (1986).
11. Smart, R. B., R. D. Herrera, and K. H. Mancy, "In Situ
Voltammetric Membrane Ozone Electrode," Analytical
Chemistry, 51, 2315 (1979).
12. Herrera, R. D., and K. H. Mancy, "Voltammetric Mem-
brane Chlorine Dioxide Electrode," Analytical Letters,
13(A7), 561 (1980).
13. Clark, L. C-, and C. A. Duggan, "Implanted Electroen-
zymatic Glucose Sensors," Diabetes Care, 5, No. 3, 174
(1982).
14. Cleland, N., andS.O. Enfors, "Monitoring Glucose Con-
sumption in Escherichia Coli Cultivation with an Enzyme
Electrode," Analytica Chimica Acta, 163, 281 (1984).
15. Gough, D. A., J. Y. Lucisano, and P. H. S. Tse, "Two-
Dimensional Enzyme Electrode Sensor for Glucose," An-
alytical Chemistry, 57, 2351 (1985).
16. Olsson, B., H. Lundbock, and G. Johanson, "Theory and
Application of Diffusion-Limited Amperometric Enzyme
Electrode Detection in Flow Injection Analysis of Glu-
cose," Analytical Chemistry, 58, 1046 (1986).
17. Sonawat, H. M., S. P. Ratna, and G. Govil, "Covalent
Immobilization of FAD and Glucose Oxidase on Carbon
Electrodes," Biotechnology and Bioengineering, 26, 1066
(1984).
18. Wiech, H. J., G. H. Heider, and A. M. Yacynych, "Chem-
ically Modified Reticulated Vitreous Carbon Electrode with
Immobilized Enzyme as a Detector in Flow-Injection De-
termination of Glucose," Analytica Chimica Acta, 158,
137 (1984).
19. Szuminsky, N. J., A. K. Chen, and C. C. Liu, "A Min-
iature Palladium-Palladium Oxide Enzyme Electrode for
Urea Determination," Biotechnology and Bioengineering,
26, 642 1984).
20. Cenas, N., J. Rozgaite, and J. Kulys, "Lactate, Pyruvate,
Ethanol, and Glucose-6-Phosphate Determination by En-
zyme Electrode," Biotechnology and Bioengineering, 26,
551 (1984).
21. Durliat,-H., M. Comtat, and A. Baudras, "Spectropho-
tometric and Electromechanical Determination of L( + )-
Lactate in Blood by Use of Lactate Dehydrogenase from
Yeast," Clinical Chemistry, 22, No. 11, 1802 (1976).
22. Durliat, H., M. Comtat, and J. Mahnec, "A Device for
the Continuous Assay of Lactate," Analytica Chimica
Acta, 106, 131 (1979).
23. Mascfni, M., D. Moscone, and G. Palleschi, "A Lactate
Electrode with Lactate Oxidase Immobilized on Nylon Net
for Blood Serum Samples in Flow Systems," Analytica
Chimica Acta, 163 , 45 (1984).
24. Soutter, W. P., F. Sharp, and D. M. Clark, "Bedside
Estimation of Whole Blood Lactate," Br. J. Anaesth.,
50, 445 (1978).
25. Suand-Chagny, M. F., and F. G. Gonon, "Immobilization
of Lactate Dyhydrogenase on a Pyrolytic Carbon Fitter
Microelectrode," Analytical Chemistry, 58, 412 (1986).
26. Satoh, I., I. Karube, and S. Suzuki, "Enzyme Electrode
for Free Cholesterol," Biotechnology and Bioengineering,
19, 1095 (1977).
27. Meyerhoff, M. E., and G. A. Rechnitz, "An Activated
Enzyme Electrode for Creatinine," Analytica Chimica
Acta, 85, 277 (1976).
28. Fonong, T., and G. A. Rechnitz, "Enzyme Electrode for
the Determination of Salicylate," Analytica Chimica Acta,
158, 357 (1984).
29. Pau, C. P., and G. A. Rechnitz, "Bound Cofactor/Dual
Enzyme Electrode System for L-Alanine," Analytica
Chimica Acta, 160, 141 (1984).
30. Liang, B. S., X. M. Li, and H. Y. Wang, "Cellular Elec-
trode for Antitumor Drug Screening," Biotechnology
Progress, 2, No. 4, 187 (1986).
31. Holodnick, S. E., "The Bio film Electrode Sensor System
for Acute Toxicity, and Viral Screening," PhD Thesis in
Environmental Health Science, University of Michigan,
(1988).
32. Schar, H. P., and O. Ghisalba, "Hyphomicrobium Bac-
terial Electrode for Determination of Monomethyl Sul-
fate," Biotechnology and Bioengineering, 27, 897 (1985).
33. Wanner, O., and W. Gujer, "A Multispecies Biofilm
Model," Biotechnology and Bioengineering, 28, 314
(1986).
34. Matsunaga, T., and Y. Namba, "Selective Determination
of Microbial Cells by Graphite Electrode Modified with
Adsorbed 4,4'- Bipyridine," Analytica Chimica Acta, 159,
87 (1984).
35. Benefield, L., and F. Molz, "Mathematical Simulation of'
a Biofilm Process," Biotechnology and Bioengineering,
27, 921 (1985).
36. Rittman, B. E., and P. L. McCarty, "Model of Steady-
State-Biofilm Kinetics," Biotechnology.and Bioengineer-
ing, 22, 2343 (1980).
37. Rittman, B. E., and P. L. McCarty, "Evaluation of
Steady-State-Biofilm Kinetics," Biotechnology and Bioen-
gineering, 22, 2359 (1980).
38. Goldblum, D. K., "Biofilm Electrode for Screening of
Toxic Chemicals: Electrode System Characterization,"
PhD Thesis in Chemical Engineering and Environmental
Health Science, University of Michigan, (1988).
39. Fahien, R. W., Fundamentals of Transport Phenomena,
McGraw-Hill, New York, 60 (19S3).
40. Lehninger, A. L., Biochemistry, 2nd Ed., Worth Pub-
lishers, New York, 519 and 611 (1975).
41. Parulekar, S. J., G. B. Semones, M. J. Rolf, J. C. Liev-
ense, and H. C. Lim, "Induction and Elimination of Os-
cillations in Continuous Cultures of Saccharomyces
Cerevisiae," Biotechnology and Bioengineering, 28, 700
(1986).
42. Haubenstricker, M. E., "Development of a Toxicity Bio-
sensor Based on Changes in Mitochondrial Respiration
Rates," PhD Thesis in Environmental Health Science,
University of Michigan, (1984).
43. Goldblum, D. K., S. E. Holodnick, K. H. Mancy, and
D. E. Briggs, "Oxygen Transport in Biofilm Electrodes
for Screening of Toxic Chemicals," AlChE Journal, 36,
19 (1990).
Environmsnial Progress (Vol. 10, No. 1)
February, 1991 29
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 44
A Fiber Optic Chemical Sensor for the Measurement of TCE
Stanley Klainer
FiberChem, Inc.
Las Vegas, Nevada
January 12-14, 1993
Las Vegas, Nevada
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 45
Calibration of the Michoud Aquifer Hydraulic Model
Martin A. Rowland, P.E.
Rowland & Associates
New Orleans, Louisiana
January 12-14, 1993
Las Vegas, Nevada
-------
CALIBRATION OF THE MICHOUD AQUIFER HYDRAULIC MODEL
Martin A. Rowland, P.E.
Principal, Rowland & Associates
New Orleans, Louisiana
There is an increasing need for groundwater hydraulic models which accurately
characterize aquifers with concentrations of contaminants. This is especially true when
used to justify a risk-based closure option for a contaminated site—an option to be
available soon for sites undergoing RCRA remediation. The current method in use by
the U.S. Environmental Protection Agency (US EPA) and the Louisiana Department of
Environmental Quality (LDEQ) to determine when a site is successfully remediated, is
based on the non-exceedance of drinking water quality standards for groundwater. The
same standards apply whether or not the groundwater has ever or is likely ever to be used
as such. Effecting a more societal-efficient allocation of resources for the remediation of
a contaminated, aquifer (as represented by the risk-based closure approach) requires, a
more complete, demonstrable definition of site conditions. This paper contributes to this
process by proposing a generalized empirical methodogy for assigning hydraulic
conductivity values for individual elements of a model grid, by quantifying the reduction
of hydraulic conductivities for segments of the aquifer affected by bacteriological
processes, and by quantifying the degree of correlation in the calibration of a
groundwater computer model for an aquifer within the Mississippi River Deltaic Plain.
This" generalized methodology simplifies the more complex methods cited in the
literature.
Classical procedures for modeling groundwater flow begin with field procedures
involving one or more groundwater wells. Data generated in this way are graphically
manipulated to derive aquifer parameters, including hydraulic conductivity, typically
assumed to be regionally homogeneous and isotropic. In most situations, and definitely
in the one presented herein, the hydraulic conductivities of an aquifer vary considerably.
Even after these field tests are conducted, the estimated hydraulic conductivities are
applicable only for the aquifer segment between the wells which were measured.
Additional tests for all combinations of wells could be performed and results analyzed to
derive and estimate the variable hydraulic conductivities over the entire area of
investigation. However, this could be time-consuming, labor-intensive, and
inconclusive, as only 2 wells are typically evaluated at a time. Also, in aquifers with
high dissolved metal content such as the one discussed herein, the introduction of oxygen
(through water table depression) to iron bacteria within aquifer soils can cause iron and
manganese oxides to settle out of solution, producing hydraulic losses (reductions in
hydraulic conductivity) due to the plugging of the well screen and adjacent soils with the
precipitate. The combined effect of growing organisms and precipitating minerals can
significantly plug a well within a short time. Thus, the act of conducting pump-
-------
drawdown tests may adversely affect the derivation of the desired parameter because of
the investigator's impact on the system he is measuring.
The literature in the field of groundwater model calibration describes the use of complex
computer programs and non-linear regression analyses to arrive at a satisfactory
convergence between observed well heads and simulated heads. Using this procedure,
discretized model grid elements are defined over a project area's ground surface for 2-
dimensional definition, with aquifer thickness as the third dimension. These elements are
then assigned initial values for various aquifer parameters, usually based on classical
procedures. Several iterations of calculations, using the governing partial differential
equations which define groundwater flow in an aquifer (subject to assigned Conditions
for selected parameters), are performed until the above-described convergence of well
heads is observed. This investigation performed the same type of analyses as the ones
cited in the literature, except that an empirical method was employed to perform a non-
linear regression-type analysis of the output (simulated well heads). A linear regression
analysis was then used to determine which iteration results had the best correlation with
observed values. The advantage with the methodology described herein is that it is
relatively easy to understand and is user-friendly in its application.
This investigation involves the development of an empirical method for determining
aquifer hydraulic conductivities for a shallow aquifer in the Mississippi River Deltaic
Plain, with limited pump test and observation well data. This method solves the inverse
problem of parameter identification in groundwater modeling studies, using an indirect
output error approach and a linear statistical method for testing validity, assumptions, and
model fit. The REM (Rowland Empirical Method) for groundwater hydraulic model
calibration is a directed-trial and error process involving four steps; each step builds upon
knowledge gained in a previous step. The beginning steps utilize results from
homogeneous, isotropic aquifer assumptions. The latter steps make use of knowledge of
natural soil deposition processes in the region, including beach trend, delta, and river
levee formation. The method is general enough to be of value for geographic regions
other than the Mississippi River Deltaic Plain.
Two new terms are defined. The aquifer hydraulic conductivity reduction factor
(AHCRF) describes an estimated decrease in the hydraulic conductivty of a segment of
an aquifer, due to the presence of an anaerobic biofilm (a consortia of anaerobic bacteria)
that is actively degrading chlorinated organic contaminants. This biofilm is estimated to
have effected approximately 1.6 acres of a 45 acre shallow aquifer segment. The 45 acre
area selected for this study is part of a 400 acre isolated sand formation (Michoud
Aquifer) which is hydraulically connected to the Gulf of Mexico. The other term,
calibration correlation quotient (CCQ), is a suggested measure of accuracy of the
simulation of field-observed well heads by use of a groundwater hydraulic flow model.
-------
The SUTRA (Saturated-Unsaturated TRAnsport) computer program simulates fluid
movement in a subsurface-environment by employing a 2-dimensional hybrid finite-
element and integrated-finite-difference method to approximate the governing equations
mat describe groundwater movement. Aitnougn tne program nas capabilities Tor
modeling the transport of either energy or dissolved substances, it was used only for fluid
movement in this investigation. Until there is a thorough understanding of aquifer
hydraulics, contaminant transport modeling is premature. Knowledge of chemical spills,
the location of the contaminant plume, evidence of anaerobic degradation, and an
understanding of the theory behind the hydraulic conductivity reduction characteristics of
a biologically active aquifer allowed the Michoud Aquifer Model to be calibrated.
Several conclusions and recommendations are offered.
1. The REM is a useful technique with possible wide application. Once an aquifer's
soil depositional processes are known and experience is gained from running computer
model simulations with homogeneous, isotropic assumptions, a directed-trial and error
process can be used to calibrate the hydraulic flow of an aquifer.'
2. If anaerobic biological degradation of organic contaminants is suspected, reducing
hydraulic conductivities by 90% may generate the desired correlation between field-
observed and computer-derived drawdowns.
3. Reductions of hydraulic conductivity at the pumping well from aerobic iron bacteria
(as well as turbulent flow at ibe outside edge of tine well screen) are also indicated by the
calibration results; reductions of at least 50% are suggested
4. One benefit of using the SUTRA program over programs cited in the literature for
groundwater flow modeling, is that the Michoud Aquifer can now be calibrated for
contaminant transport without extensive change to the definition of the model. SUTRA
has this built-in capability. Most other programs are designed strictly for groundwater
flow.
5. The use of the CCQ for comparing the success of model calibration is recommended
to all groundwater modelers. This or some other similar yardstick would allow
investigators (who are studying site with similar conditions and using similar
assumptions) to assess how well their results compare.
6. Conditions at the site have changed considerably since the 1985 measurement of
well drawdowns used in this study. New wells have been installed and old wells have
been plugged and abandoned. It is also likely that aquifer hydraulic conductivity
reduction from biodegradation of organic solvents is a transient phenomena. As opposed
to aerobic biological degradation, anaerobic degradation is not known to generate a
biomass residue that could permanently, physically alter an aquifer's hydraulic
-------
conductivity. The anaerobic process generates energy (for the degradation process) and
waste gases. It is recommended that additional analysis of existing conditions be
performed. During the site remedial measures investigation, scheduled to resume in
early 1993, verification and analysis of this aquifer impact may be an appropriate pursuit.
7. Several REM calibration exercises using the same wells, but at different pumping
rates, and for different grid sizes are recommended. Also, taking measurements at
different times of the year may show evidence of seasonal variations. It is important that
these additional tests are also performed following periods, of steady-state pumping, as
was done in the one discussed herein. The test results could then be mathematically
reduced to arrive at an even more reliable estimate of hydraulic conductivity for the
various elements.
8. The flow in the pumpingwell should be reduced to minimize the exposure of the
aquifer soil and screen to the atmosphere. This should reduce the rate of growth of iron
bacteria and the precipitation of aquifer-plugging metal oxides:
9. A useful area of further investigation would be the quantification of the aquifer
hydraulic reduction factor over time for varying concentrations of contaminant and for
varying concentrations of electron acceptors. Also, future studies may try optimizing this
in situ degradation process for a cost-effective component of an integrated approach to
the reduction of TCE contamination in groundwater. As TCE is resistant to
biodegradation under aerobic conditions, current developments in the scientific
understanding of its metabolism under under anaerobic conditions is forming a promising
scientific foundation upon which to build novel bioremediation strategies.
-------
52
REFERENCES
( 1 ) Bouwer, E. J., P. L. McCarty, 1984, Modeling of Trace Organics
Biotransformation in the Subsurface, Ground Water Magazine, v. 22, no. 4,
pp. 433-440.
( 2) Driscoll, F. G., 1986, Groundwater and Wells, Johnson Division, St. Paul,
Minnesota, 98 p.
(3) Morgan, C. J., 1989, Modeling Groundwater Flow and Trichloroethylene
Transport in a Shallow Aquifer, University of New Orleans M.S. Thesis, pp.
5-46.
(4) Golden Software, Inc., 1987, SURFER Contour Software, Golden, Colorado.
( 5 ) Todd, D. K., 1980, Groundwater Hydrology, John Wiley and Sons, 334 p.
( 6) Cooley, R. L., 1977, A Method of Estimating Parameters and Assessing Reliability
for Models of Steady State Groundwater Flow, Part 1., Water Resources
Research, Vol. 13, No. 2, pp.318-324.
(7) Nueman, S. P., 1973, Calibration of Distributed Parameter Groundwater Flow
Models Viewed As a Multiple-Objective Decision Process Under
Uncertainty, Water Resources Review, Vol. 9, No. 4, pp.'1006-1021.
(8) Yeh, W. W., 1986, Review of Parameter Identification Procedures in
Groundwater Hydrology: The Inverse Problem, Water Resources Research,
Vol. 22, No. 2, pp.95-108.
( 9 ) Tassin, R. A., 1983. Interoffice Memo on Environmental Liability, Reference No.
31 00-83-1 21.
(10) Snowden, J. O., W. C. Ward, J. R. J. Studlick, 1980, Geology of Greater New
Orleans: Its Relationship to Land Subsidence and Flooding, New Orleans
Geological Society, LA, pp. 1-13.
(11) Sax, N. I., R. J. Lewis, 1989, Dangerous Properties of Industrial Materials, Van
Nostrand Reinhold, N.Y., Volume III, Seventh Edition, TIO750 p.
(12) Sowers, B. S., G. F. Sowers, 1970, Introductory Soil Mechanics and Foundations,
The Macmillan Company, 46p.
(13) Marrotte, R., 1990. Supplemental Closure Investigation—Rinsewater
Impoundment and Appurtenant Structures for NASA Michoud Assembly
Facility, CH2M Hill Environmental Consultants, pp. 6-1 through 6-4.
(14) Voss, C. I.,1984, SUTRA, U.S.G.S. Government Document, pp. 1-217.
-------
53
REFERENCES (continued)
(15) Rowland, M. A., T. N. Eisenberg, 1989, Anaerobic Biodegradation of
Trichloroethylene in a Shallow Aquifer, Proceedings of the Third National
Outdoor Action Conference on Aquifer Restoration, Ground
Water Monitoring, and Geophysical Methods, National Water Well
Association, pp. 551-561.
(16) Suflita, J. M., G. W. Sewell, 1991. Anaerobic Biotransformation of Contaminants
in the Subsurface, U.S.E.P.A. Government Document--EPA/600/M-
90/024, pp. 1-8.
(17) Envirocorp Services & Technology, 1990, Recovery Wells RW-1, RW-2, and
RW-3 Servicing/Workover, Baton Rouge, Louisiana
(18) Ang, A. H., W. H. Tang, 1975, Probability Concepts in Engineering Planning and
• Design, John Wiley & Sons, pp. 286-296.
(19) Chilcote, D. D., 1989, Laboratory Feasibility Study of Biological Treatment of
Groundwater Contaminated with Chlorinated Solvents, Biotrol, Inc.
Consultants, 2 p.
-------
219
Marty Rowland 4678 Arts St. New Orleans, LA 70122
(504) 283-1711
Mr. Rowland is a registered professional environmental engineer in the States
of Michigan and Louisiana, currently directing the contaminant remediation
investigation and corrective measures study at the Michoud Assembly Facility in
New Orleans, Louisiana.
After receiving his Bachelor of Environmental Engineering degree from the
University of Michigan in 1975, Mr. Rowland has held positions as a State of
Michigan environmental enforcement official, site manager for a secure
chemical waste landfill facility, consulting engineer responsible for waste site
closures, and senior environmental engineer for a major aerospace facility. In
1992, Mr. Rowland received a Master of Environmental Engineering from the
University of New Orleans.
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 46
Field Analysis of VOCs by Photocaustic Detection
J. F. McClelland
Ames Laboratory
Iowa State Univeristy-Ames
January 12-14, 1993
Las Vegas, Nevada
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 47
Evaluation of Headspace Method for Volatile Constituents in Soils and
Sediments
B. B. Looney, C.A. Eddy and W. R. Sims
Westinghouse Savannah River Company, Aiken, South Carolina
January 12-14, 1993
Las Vegas, Nevada
-------
EVALUATION OF HEADSPACE METHOD FOR VOLATILE
CONSTITUENTS IN SOILS AND SEDIMENTS
B. B. Looney, C. A. Eddy, and W. R. Sims
Westinghouse Savannah River Company
Aiken, SC 29808
Summary
Detection and delineation of volatile organic contaminants (VOCs) in sediments and soils
underlying a hazardous waste site is often a complex problem. The number and quality of
analyses used in characterization studies can be compromised by the difficulties and costs
associated with quantitative analysis of volatile analytes. A headspace analysis method was
developed to facilitate the accurate and rapid delineation of the vertical and horizontal"
distribution of VOCs in the subsurface, and to reduce the sample handling, laboratory
preparation, arid analytical complexity associated with most existent sampling and analysis
schemes. The headspace method consists of the following four steps:
• Subsample the core immediately after retrieval using a small tube/plunger system.
• Place the Subsample into a 22.5 mL headspace vial.
• Add 5 mL of suspending solution and cap with a teflon lined septum.
• Analyze an aliquot of the headspace using a gas chromatograph.
This method was developed and modified as a result of multiple site investigations and has
been applied to over 2000 samples from both saturated and unsaturated sediments. Data
collected during these studies indicated that distilled water suspending solution is similar to
an ionic (Na2S04/H2PC>4) solution and that sonication of the samples does not enhance the
recovery of VOCs. Sealed samples exhibited stable concentrations for more than 20 days.
To further evaluate the headspace method, 92 pairs of samples were analyzed to allow
direct comparison of the headspace method to a modified EPA solvent extraction method
typically applied to environmental samples. Despite the precautions incorporated into the
solvent extraction method, the results indicated that sample transfers in the field and
laboratory resulted in substantial volatilization of VOCs. The headspace method minimized
these losses and generated results rapidly, facilitating informed decision making during site
characterization.
Background
Barcelona (19.89) suggests that sample collection and handling activities can contribute to
systematic errors in environmental data. These errors are often relatively large compared to
the random and systematic errors typically associated with the instrumental analysis.
Perhaps the most difficult sample collection and handling error to delineate and control is
negative bias (measured value less than true value). In the case of VOCs, this error is
principally caused by volatilization of the analyte during sample collection, storage, and
handling. In samples from the saturated zone, displacement of pore water by drilling fluids
-------
or drainage of the core can contribute to negative bias. Recent research (Seigrist and
Jenssen, 1990; Urban et al., 1989) indicates that the typical methods (containerization of
disturbed samples, followed by refrigerated storage and solvent extraction) may lead to
substantial volatilization loss; the investigators recommended controlled research and
development of alternate procedures.
Analytical methods for VOCs in water samples can logically be grouped into the following
three categories:
• solvent extraction
• static headspace methods
• dynamic headspace methods (purge and trap)
Each of these categories relies on a partitioning of the contaminant from the water into an
alternate phase prior to instrumental analysis. In the case of solvent extraction, the alternate
phase is typically a liquid organic solvent, while in the headspace categories, the alternate
phase is a volume of gas. The success of an analytical method depends ori the relative
affinity of the VOC for the alternate phase, the compatibility of the extracting phase with the
analytical instrumentation, and the ability to reproducibly contact the phases and handle the
extract For solvent extraction and static headspace methods, an aliquot of the extract is
generally introduced into a gas chromatograph (GC) .equipped with an appropriate detector
or mass spectrometer. In purge and trap analysis, gas is bubbled through the sample at a
constant rate for a specified time. Contaminant vapors are collected on an adsorbent trap;
following the purge, the contaminants are thermally desorbed into the GC.
Existing analysis methods for soil and sediments are predominantly in the solvent
extraction class; however, both static headspace and purge and trap methods are
documented in the literature (McNally and Grob, 1985). Successful application of the gas
phase extraction methods to soils and sediments relies on effective, and reproducible
partitioning from the solid to the gas phase for dry samples, and from the solid and liquid
phases to the gas phase for wet samples.
The relative simplicity and minimal sample handling suggest that the static headspace
method may be a relatively attractive technique for analysis of volatile constituents in soil
and sediment samples. Static headspace methods are most applicable to samples with
mineral or low organic matrices. Additionally, these methods require relatively constant
conditions for reliable proportionality between original sample concentration and the mass
of each VOC introduced into the GC. In particular, the properties of any suspending
solution (e.g., ionic strength, pH, etc.) and the physical conditions in the vial (e.g.,
temperature, pressure, phase volumes, etc.) should be selected to minimize sorption and
maximize the conditional Henry's Law partitioning from the solution to the gas phase.
Reports in the literature document successful application of static headspace methods to
water, wastewater, industrial effluents, soil, sediments, and sewage (McNally and Grob,
1985). In cases where reproducible results are documented, headspace methods are often
preferred because they are simpler and faster, and therefore less expensive than either
solvent extraction or purge and trap methods. Since sample handling can be minimized and
analyses are generated rapidly, results of this screening approach can be incorporated
during the site characterization process.
-------
Methods and Study Design
The sediments for the headspace sampling and analysis studies were collected from borings
at sites in the vicinity of the metallurgical manufacturing facility in M-Area at the Savannah
River Site (SRS). Solvents ~ trichloroethylene (TCE) and tetrachloroethylene (PCE) ~
were used in this facility during the late 1950s to the early 1980s to degrease the fuel and
target tubes prior to use in other facilities at SRS. Concentrations of VOCs in the partially
saturated and saturated sediments vary vertically and horizontally beneath the site. Proper
site characterization and long term remediation system design requires adequate delineation
of this variation.
The boring locations in M Area were selected based on results from a shallow soil gas
survey, combined with process records, groundwater data and past core data. The overall
method development study consisted of two parts. First, following an initial period of
method development, a series of samples was collected and analyzed to optimize the
conditions for sampling and analysis. Second, a series of paired samples was collected to
compare the headspace method to a solvent extraction method typically used for soil-
sediment analysis. A brief discussion of the headspace method is provided below,
followed by specific details associated with each phase of the study.
Drilling and Coring
Continuous borings were drilled to an approximate depth of 200 feet at each location using
two drilling methods. Within the vadose zone (130 -140 foot depth), samples were
collected using 4.25-inch inside diameter hollow-stem augers and a standard 2-inch inside
diameter, split-spoon sampler. Below the water table, the boreholes were stabilized using a
(beritonite-based) mud rotary system, and continuous samples were collected ahead of the
borehole using a CP wireline system and Christensen Core Barrel.' All subsamples for
VOC analysis were collected as quickly as possible after the core was retrieved.
Headspace Sampling-Analysis Procedure
The headspace sediment subsample (3-5 grams) was collected immediately from the open
split-spoon using an open-ended plastic disposable syringe and extruded into a 22.5 mL
borosilicate vial. Using as pipet, 5 mL of suspending solution were added to the subsample
and the vial was sealed by crimping an aluminum cap around a teflon-lined butyl rubber
septum. The sample vial was labeled and placed in an ice chest cooled to approximately 4°
C for later analysis at an onsite laboratory. The subsample corer (syringe) was
decontaminated between sampling events by brushing and rinsing with isopropanol
followed by a distilled water wash.
The onsite laboratory consisted of a headspace analyzer connected to a Hewlett Packard
(HP) 5890A Gas Chromatograph (GC). Details of the headspace analysis procedure used
are given below. Prior to field sampling, we determined the average weight of a sealed
headspace vial containing five milliliters of suspending solution. Upon receipt of the
headspace sample vials from the field, the capped vials containing the sediment samples
were weighed. The amount of sediment in each vial was determined by subtracting the
average weight from the sample weight. Each vial with the sediment subsample was
analyzed using the HP GC equipped with an electron capture detector, an HP 19395
headspace sampler, an HP 3392 networking integrator, and a 60 m widebore capillary
-------
column coated with a nonpolar silicone phase. The flow and oven temperature conditions
recommended by the manufacturers were used. The instrument was calibrated using vials
containing known quantities of VOCs, suspending solution, and (in some, cases) clean
representative sediments. The conditions in the vials (headspace volume, suspending
solution volume, and temperature) were standardized as much as possible to maintain the
proportionality between the sample concentration and VOC mass in the headspace aliquot.
The heated (70 °C) bath in the headspace sampler maximizes the transfer into the vapor
phase. The data for each peak was entered into a spreadsheet and the concentration of
contaminant in the original sample was estimated using the response factors from the
calibration. All values were reported in units of micrograms of VOC per gram of bulk
sediment (ng/g). Approximately 30 to 50 samples were analyzed each day.
Headspace Method Optimization Study
The purpose of this phase of the project was to identify the most effective operating
conditions for the three phase static headspace procedure. Specifically, the nature of the
suspending solution and the need for physical agitation were analyzed. In each case, a
reference condition was identified and the relationship between the reference and alternate
conditions were evaluated by comparing the relative recoveries for a large number of
sample pairs. Because of documented superiority of adding salt when analyzing water
samples (Gottauf, 1966), a Na2SC>4/H2P04 buffer solution was selected for the reference
suspending solution (200 mL distilled water, 10 g sodium sulfate, and 0.3 mL concentrated
phosphoric acid) and was compared with the distilled water. The reference physical
agitation method was sonication, which was compared with the alternate method of no
agitation (i.e., the vials were placed directly into heated headspace bath). Additionally, a
time-series study was performed to determine the stability of the sediment samples sealed in
headspace vials.
Comparison of Headspace Method to Solvent Extraction
The two separate laboratories utilized in the study to analyze the sediment subsamples were
an onsite laboratory operated by Savannah River Laboratory personnel and a close support
laboratory (CSL) operated by independent subcontract personnel. The onsite laboratory
analyzed the headspace sediment subsamples and the CSL analyzed the sediment
subsamples collected by the independent subcontractor. Both laboratories used standard
chain-of-custody procedures and collected quality assurance/quality control (QA/QC)
sediment subsamples to comply with the SRS QA requirements. These requirements
included the analysis of duplicate samples, matrix spikes, and trip blanks. All analyses
were performed within the required holding time. TTie method selected by the
subcontractor for the CSL was typical of those applied at waste sites in the United States
(EPA method 3550). The method generally consisted of containerization of disturbed
samples followed by refrigerated storage, sample transfer, and solvent extraction.
During this study, water was used for the headspace suspending solution and the
headspace samples were not sonicated. The results from the headspace analysis and CLS
were used to determine screen intervals for vapor extraction wells installed as part of a
vadose zone remediation program.
-------
Results and Discussion
The data suggest that a headspace analysis approach provides rapid and reproducible
analytical results for analysis of VOCs in many common soils and sediments. The
parameter optimization phase of the study indicated that a distilled water suspending
solution is similar to a Na2S04/H2PC>4 suspending solution and that sonication of the sample
does not improve the transfer of contaminant into the headspace from the solid/liquid
phases. The time-series data suggested that samples are relatively stable following
collection; the replicate vials generated similar concentrations for the entire time-series
period of 20 days. Elimination of the buffer solution and sonication step, based on the
parameter optimization phase, yields a sampling/analytical scheme that is rapid and simple
to implement
In a second phase of the study, the headspace method was directly compared to a modified
EPA solvent extraction method. Despite the precautions incorporated into the solvent
extraction method, the analytical results indicated that sample transfers in the field and
laboratory resulted in significant volatilization of VOCs from the sediment samples prior to
analysis. The headspace method appears to provide more representative data on the
samples. The headspace analysis method generally resulted in a higher value for the
measured concentration of both TCE and PCE. The two primary exceptions to this general
trend are samples with very high concentrations of contaminants and samples where both
methods were below detection limits. For example, in the samples from one of the cores,
there are five examples where the two methods are the same for TCE. All of these
examples are found where the analytical results are below detection limits for both
methods. Similarly, in this core, the results from the solvent extraction method are greater
in only 4 of 33 examples. All 4 examples result from overloading of the GC during the
headspace analysis (the samples can not be diluted). These same trends may be observed
in all of the other cores. In die comparison study, the headspace method indicated the
presence of contamination in each of the silty, clayey, and poorly graded layers throughout
the vadose zone. The solvent extraction method generated below detection results for most
of these zones. Additionally, the headspace method indicated low (but measurable)
concentrations in the well-graded sands, while the solvent extraction method indicates
below detection results in almost all of these layers.
The paired data were ranked and ordered for statistical analysis. In this form, a Wilcoxon
Signed Rank Test was applied to determine if the two methods yielded statistically similar
results. This hypothesis was rejected at greater than the 99% confidence level, signifying
that the two populations are different. Thus, the statistical test indicated that there is greater
than a 99% probability that the two methods are statistically different (i.e., the headspace
method generates higher values).
As discussed above, one limitation of the headspace method is that the sample can not be
diluted; thus, very high concentrations are truncated by an upper limit of detection. In most
cases, this truncation may not be of practical significance because it occurs at relatively high
concentrations (e.g., 100,000 ng/g). This truncation can be essentially eliminated by
splitting the column effluent to a flame ionization detector (FID) in parallel with a halogen
specific detector. In this configuration, the less sensitive FID extends the range of the
analysis by several orders of magnitude.
-------
Conclusions
The results indicate that the headspace method minimized loss of volatiles associated with
sample handling and provided large amounts of closely spaced data. From an analytical
standpoint, at sites with low sediment organic carbon and relatively volatile constituents,
there are several advantages of the headspace method over solvent extraction methods.
Some of these advantages include the following:
• reduced sample handling effort and time in the field
• no solvent extraction required (the Henry's Law mass transfer in the headspace
vial requires no operator effort)
• elimination of multiple sample transfers and minimization of the opportunities for
volatilization of analyte
The headspace sediment sample is sealed in its final form ready for analysis within a few
seconds of collection and is never directly handled again during weighings or transfers.
Once in the laboratory, approximately 50 samples can be analyzed in a normal working day
on a single instrument. Headspace analysis is cost effective; we have calculated the fully
loaded costs of the analysis to be $50 - $100 per sample. In addition, the headspace results
can be generated rapidly and transferred to the field so that informed decisions can be made
during site characterization.
The information contained in this article was developed during the course of work under
Contract No. DE-AC09-89SR18035 with the U. S. Department of Energy. By acceptance
of this paper, the publisher and/or recipient acknowledges the U. S. Governments right to
retain a nonexclusive, royalty-free license in and to any copyright covering this paper along
with the right to reproduce, and to authorize others to reproduce all or part of the
copyrighted paper.
-------
REFERENCES
Barcelona, M. J., 1989. Report in Principles of Environmental Sampling. L. H. Kieth
editor, American Chemical Society, Washington DC, 3-23.
Gottauf, M., 1966. Verbesserte quantitative gaschromatographische Spurenanalyse
fliichtiger organischer verbindungen in wasser. Fresenius' Zeitschriftfiir Analytische
Chemie, 218, 175-184.
Hachenberg, H and A. P Schmidt, 1977. Gas Chromatograsphic Headspace Analysis,
Heyden Publishers, Philidelphia PA.
Ioffe, B. V., and A. G. Vitenberg, 1984. Headspace Analysis and Related Methods in
Gas Chromatography. John Wiley and Sons, New York.
Kerfoot, H. B., 1991. Groundwater, 29, pp 678-684.
Kolb, B., 1980. Applied Headspace Gas Chromatography. Heyden Publishers,
Philidelphia PA.
McNally, M. E. and R. L. Grob, 1985. Current applications of static and dynamic
headspace analysis: part 1: environmental applications. American Laboratory, January
1985, 20-33.
Siegrist, R. L. and P. D. Jenssen, 1990. Evaluation of sampling method effects on volatile
organic compound measurement in contaminated soils. Environmental Science and
Technology, 24:9, 1387-1392.
Urban, M. J., J. S. Smith, E. K. Schultz and R. K. Dickson, 1989. Report in Fifth
Annual Waste Testing and Quality Assurance Symposium; U. S. Environmental Protection
Agency, Washington DC, II87-II101.
EPA, 1986. Test Methods for Evaluating Solid Waste, Physical/Chemical Methods. SW-
846. U. S. Environmental Protection Agency, Office of Solid Waste and Emergency
Response, Washington, D. C.
-------
BIOGRAPHICAL SKETCHES
B. B. Looney received his PhD in Environmental Engineering from the University of
Minnesota in 1983. He is currently employed as a research engineer in the Savannah River
Laboratory and is an adjunct professor in the Environmental Systems Engineering Program
at Clemson University. His interests and responsibilities include, developing and testing
new methods for environmental characterization and remediation, risk assessment, and
modeling. Westinghouse Savannah River Company, 773-42A, Aiken SC 29808.
C. A. Eddy is completing her PhD in geology at University of California, Davis. She is
currently employed as a research geologist at the Savannah River Laboratory. Her interests
include geology and analytical geochemistry. Westinghouse Savannah River Company,
773-42A, Aiken SC 29808.
W. R. Sims received his MS degree in Geology from the University of Akron in 1987. He
is currently a hydrogeologist in the Environmental Restoration Department at the Savannah
River Site. His responsibilities and interests include RCRA/CERCLA waste site
assessments, vadose zone and groundwater characterization and remediation, and
regulatory support for groundwater corrective actions. Westinghouse Savannah River
Company, Environmental, Aiken SC 29808
-------
National Symposium on Measuring and
Interpreting VOCs in Soils: State of the Art and
Research Needs
Tab 48
On - Site Analysis of VOCs in Soils by Transportable GC/MS
Jeff Christenson and Dave Quinn
Viking Instruments
Reston, Virginia
January 12-14, 1993
Las Vegas, Nevada
------- |