xvEPA
United States
Enironmental Protection
Agency
Office of Research and
Development
Washington, DC 20460
EPA/600/R-93/140
May 1993
Behavior and
Determination of
Volatile Organic
Compounds in
A Literature Review
-------
EPA 600/R-93/140
May 1993
BEHAVIOR AND DETERMINATION OF
VOLATILE ORGANIC COMPOUNDS IN SOIL:
A LITERATURE REVIEW
by
Marti Minnich
Lockheed Environmental Systems & Technologies Company
980 Kelly Johnson Drive
Las Vegas, Nevada 89119
Contract No 68-CO-0049
Work Assignment Manager
Brian Schumacher
Exposure Assessment Research Division
Environmental Monitoring Systems Laboratory
Las Vegas, Nevada 89193-3478
ENVIRONMENTAL MONITORING SYSTEMS LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
LAS VEGAS, NEVADA 89193-3478
Printed on Recycled Paper
-------
NOTICE
The development of this document has been funded by the U.S. Environmental
Protection Agency under Contract No. 68-CO-0049 to Lockheed Environmental Systems
and Technologies Co. as part of an ongoing research effort in support of the Agency's
Superfund and Hazardous Waste programs. It has been subject to the Agency's peer and
administrative review, and has been approved for publication as an EPA document.
Mention of trade names or commercial products does not constitute endorsement
or recommendation for use.
-------
EXECUTIVE SUMMARY
Accurate measurement of soil volatile organic compound (VOC)
concentrations is crucial to site investigation, evaluation, and remediation efforts at
Superfund sites contaminated by VOCs Soils that are contaminated with VOCs
are potential reservoirs of long-term ground water contamination. This report
summarizes literature pertaining to (1) the fate and transport of soil VOCs and, (2)
the sampling and analysis of soil VOCs by SW-846 Methods 8240/8260 using purge-
and-trap/gas chromatography/mass spectrometry (PT/GC/MS).
FATE AND TRANSPORT
Nonpolar VOCs are sorbed predominately by soil organic matter in moist or
wet soil. Soil sorption exhibits an initial phase of fast uptake, followed by slow
continued sorption or diffusion of VOCs into soil microsites. Resorption studies
show a similar rapid resorption phase preceding an extended slow release phase.
Soil water retains VOCs in proportion to compound-specific Henry's Law constants.
VOC vapors are adsorbed by soil minerals in dry soil and the quantities adsorbed
are 2 to 4 orders of magnitude greater than sorption by wet soil. Contamination by
nonaqueous-phase liquids (NAPLs) results in a residual saturation fraction,
described as tiny portions of NAPL held by capillary forces in soil pores, which
changes in composition over time by physiochemical weathering. The size of the
residual NAPL fraction is related to the soil porosity.
Biodegradation of naturally occurring VOCs (such as petroleum products)
readily occurs under aerobic conditions. Microorganisms also degrade halogenated
aromatics (such as chlorobenzene) aerobically, but more slowly than the naturally
occurring VOCs Halogenated aliphanes (such as chloroform and TCE) are
degraded far more slowly than the other compounds, by microorganisms or abiotic
processes, and mainly under anaerobic conditions. Degradation of halogenated
aliphanes, however, has been observed in soils containing substantial amounts of
biodegradable carbon compounds, presumably by co-metabolism.
VOCs move in soils by diffusion and advection. Vapor diffusion, density-
driven NAPL vapor advection, and gravity-driven NAPL advection are the most
important mechanisms for movement. The movement of two fluorocarbons by
diffusion in deep sediments in Texas progressed approximately 44 m vertical in 40
years (time since manufactured). Movement of carbon tetrachloride (a dense
solvent) 177 m to ground water at a site in Idaho (time of travel unknown)
in
-------
is believed to be caused by density-driven vapor advection. Movement of benzene,
toluene, and xylene (solvents less dense than water) 24.4 m vertically in less than 7 years
at a California site has been attributed to gravity-driven NAPL advection.
SAMPLING AND ANALYSIS
Substantial volatile and degradative losses of soil VOCs have been documented to
occur from sample preservation and subsampling steps of SW-846 Methods 8240/8260.
Soil samples stored cold (4°C) have maximum holding times of less than 3 days before
the concentration falls below the 90% confidence limit of the initial value. Laboratory
soil transfers create VOC losses that widely vary by compound and soil, but losses
average approximately 60%.
Immersion of soil samples in methanol has been shown to reduce VOC losses
during sample storage and preparation for analysis. Although the analytical sensitivity of
methanol-preserved samples is less than that of soil/water samples analyzed by purge-
and-trap (PT) preparation, soil-VOC concentrations in methanol-preserved samples were
1 to 3 orders of magnitude greater than soil-VOC concentrations in collocated samples
analyzed by low level PT/GC/MS. This implies that much of the existing data of soil
VOCs analyzed by SW-846 Method 8240 could be 1 to 3 orders of magnitude below
values obtained in properly preserved samples or obtained by field analysis.
To a large extent erratic recovery of same-day spikes and loss of analyte during
storage has impeded the accurate assessment of soil-VOC measurement errors. Quality
control samples or performance evaluation materials (PEMs) are not available for soil
VOCs Recently, vapor fortification of small (2 to 3 g), dry soil samples (four
compounds spiked onto two soils) has established low relative standard deviations among
samples and storage of at least 3 weeks without measurable sample loss. The technique
does not calculate spike recoveries but creates stable and reproducible concentrations of
VOC-contaminated soils. It is limited to small aliquots of dry soil. Another option for
PEMs might be samples immersed in methanol.
Current analytical methods that utilize PT techniques to remove soil VOCs are
not sufficient to extract entrapped VOCs (also referred to as residual, nonequilibrium, or
slowly desorbing VOCs).
Field (static) headspace techniques offer a rapid means of quantifying soil VOCs
with some restrictions. First, the detection limit is not as low as can be achieved with a
PT preconcentration step. After the compounds of interest are identified, however,
iv
-------
options for detectors other than the mass spectrometer allow for extremely low detection
limits. Second, in soils that are high in organic matter or soils that have a large fraction
of slowly desorbing VOCs, PT extraction may be more thorough than soil headspace,
thus necessitating laboratory corroboration of field data.
-------
CONTENTS
Executive Summary iii
Figures viii
Tables ix
Abbreviations and Acronyms x
Acknowledgment xi
Introduction 1
Objective and Scope 1
Background 2
Definition of a Soil VOC 2
Occurrence and Ranking of VOCs 3
Statement of the Problem 7
Interphase Transfers 9
Aqueous-Sorbed Distribution 9
Nonequilibrium Sorption 13
Vapor-Sorbed Distribution 16
Vapor-Aqueous Distribution 20
NAPL-Aqueous/Vapor Distributions 21
Summary of Interphase Transfers 22
Degradation 23
Microbiological Degradation 23
Abiotic Degradation 26
Factors Affecting Degradation Rates 27
Summary of Degradation 29
Movement of VOCs in theVadose Zone 31
Vapor Diffusion 31
Vapor Advection 32
Aqueous Convection 33
Volatilization-Gaseous Diffusion and Aqueous Convection Combined 33
Field Studies of VOC Fate and Movement 34
Field Experiments 34
Field Investigations 36
Summary of VOC Movement 38
Modeling the Movement of Soil VOCs 40
Screening/Management Models 43
Laboratory Soil Column Simulations 44
Field-Scale Simulations 46
Summary of Models 48
vi
-------
Obtaining and Maintaining VOC Samples 50
Current Sampling Methods 50
Laboratory Sampling Studies 52
Field Samplings Studies 54
Spatial Variability 57
Sample Storage and Preservation 58
Constraints on the Container Material 58
Studies of VOC-Spiked Soil Storage Times 59
Summary of Sampling and Preservation methods 61
Analytical Methodology 64
SW-846 Method 8240 and Related Methods 64
Modifications Offered to Improve Soil Purge-and-Trap Analysis 65
Analytical Sensitivity of Solvent Extracts 67
Exhaustive Extractions to Recover Sorbed VOCs 68
Summary of Analytical Methodology 71
Field Methods for Determining Soil Gas and Soil VOCs 73
Justification for Field Methods 73
Soil-Gas Measurements 74
Soil Headspace Methods 77
Advanced Field Extraction and Analysis Methods 79
Summary of Field Methods 80
Conclusions 82
Fate and Transport of Soil VOCs 82
Sample Size 85
Sample Preservation and Analysis 85
Field Methods 87
References 89
vn
-------
FIGURES
Number Page
1 Log Koc-Log Kowrelationship 11
2 Vapor-phase sorption as a function of moisture content 19
3 Collocated soils samples analyzed by conventional PT/GC/MS, limited disruption
(LD)PT/GC/MS, and headspace (HS)GC 55
4 Fate of soil VOCs 83
Vlll
-------
TABLES
Number
1 VOCs on the Revised Priority Hazardous Substances List (HSL) 4
2 Laboratory Dissipation Half-Lives of Some Organic Compounds 29
3 Comparative Features of Some Vapor-Transport Models 41
4 Comparison of Purge-and-Trap Versus Solvent Extraction for Analysis of Aged,
EDB-Contaminated Soil 70
IX
-------
ABBREVIATIONS AND ACRONYMS
ASTM American Society of Testing and Materials
BET Brunauer-Emmett-Teller
BTEX benzene, toluene, ethylbenzene, and xylene(s)
BTX benzene, toluene, and xylene(s)
CLP Contract Laboratory Program, administered under the EPA
Superfund Program
EDB ethylene dibrornide, 1,2-dibromoethane
EPA U.S. Environmental Protection Agency
FID flame ionization detector
GC gas chromatography
HS headspace
INEL Idaho National Engineering Laboratory
LD limited disruption
MHT maximum holding time
MS mass spectrometry
NAPL nonaqueous-phase liquid
NOC nonpolar, nonionic organic compound
PCB polychlorinated biphenyl
PCE perchloroethylene, tetrachloroethene
PEM performance evaluation material
PID photoionization detector
PT purge and trap
PTFE polytetrafluoroethylene, Teflon®
PVC polyvinylchloride
QA/QC quality assurance/quality control
RCRA Resource Conservation and Recovery Act
RH relative humidity
RSD relative standard deviation
SFE supercritical fluid extraction
SOW Statement of Work, laboratory procedures for the CLP
TCLP Toxicity Characteristic Leaching Procedure
TCE trichloroethylene, trichloroethene
TPH total petroleum hydrocarbons
VOA volatile organics analysis
VOCs volatile organic compounds
-------
ACKNOWLEDGMENT
The author is grateful for the helpful conversations and reviews furnished by
colleagues at Lockheed, especially Neal Amick, Roy Cameron, Patty Fitzpatrick, Tim
Lewis, Cindy Mayer, Beth Moore, and Jim Pollard. Formal reviews were provided by
Brian Schumacher, Robert Siegrist, Steve Ward, Jeff van Ee, Katrina Varner, and
William Spencer, all of whom helped revise and clarify this document. Many thanks to
the editors Domenic Fuccillo, Marianne Faber, and Bobbie Stephens who endured and
managed to concentrate while reading this manuscript.
XI
-------
SECTION 1
INTRODUCTION
Volatile organic compounds (VOCs) are the most common and the most mobile
subsurface contaminants encountered at Superfund and other hazardous waste sites.
VOCs can be toxic, mutagenic, or carcinogenic. Soil VOCs are of concern primarily as a
potential source of ground-water contamination. They may contribute to inhalation
exposure, which can result when volatile emissions emanate from the soil surface. Soil
VOCs also may be associated with ingestion exposure, which can occur when children
play in contaminated soil or when the compounds are absorbed into the edible portion of
agricultural plants. Accurate soil and sediment VOC determinations are needed to
assess the extent of contamination to make decisions on appropriate cleanup activities,
and to verify remediation efforts.
OBJECTIVE AND SCOPE
At the request of the U.S. EPA a literature review was conducted to present and
assess literature research results pertaining to the problems and inconsistencies observed
in the sampling and analysis of soil VOCs by SW-846 Methods 8240/8260. SW-846
Methods 8240/8260 are the primary soil-VOC laboratory methods, intended to provide
the most definitive compound identification and the lowest detection limits. These
methods entail a purge-and-trap (PT) preparation/extraction step SW-846 Method 5030
and gas chromatography/mass spectrometry (GC/MS) analysis procedures (USEPA,
1986, 1990). SW-846 Method 8240 uses a packed GC column and Method 8260 uses a
capillary GC column. Field sampling procedures are largely unspecified. The results
and discussion presented here are intended to be used by the U.S. EPA to evaluate
problems with the current SW-846 methods and to be the basis for future potential
research needed that will increase the precision and accuracy of soil vadose zone VOC
measurements.
The scope of this project included vadose zone soil and sediments only, even
though the SW-846 methods are designed to be applied to any solid matrix samples.
Literature on vapor-phase and other field measurement techniques was included insofar
as data comparisons with laboratory purge-and-trap (PT) techniques were given or when
data were supplied that help define the representativeness, precision, and accuracy of the
total soil-VOC measurements.
-------
The term "soil" in this report refers generally to any unconsolidated surficial
geologic sediments and associated organic matter, 2 mm or less in size, irrespective
of pedogenic processes. This definition derives from common usage in the fields of
engineering (Holtz and Kovacs, 1981) Geologists and soil scientists would consider
this definition to represent "soils and sediments."
The literature search for this project was conducted in three modes: (1) tree-
searching, starting with an initial body of literature obtained from Lockheed
researchers, (2) scanning Current Contents: Agriculture, Biology, and
Environmental Sciences (Institute for Scientific Information, Inc., Philadelphia, PA)
from 1990 through mid-1992 for relevant titles, and (3) personal communications
with researchers currently studying soil-VOC measurement procedures. On-line
data bases were searched by using different strategies and key words. On-line
searches provided some, but not sufficient, references to support this project.
Literature was collected for this review through mid-1992, with the exception of the
inclusion of abstracts from the January 1993 National Symposium on "Measuring
and Interpreting VOCs in Soils State of the Art and Research Needs."
BACKGROUND
Definition of a Soil VOC
For soil and water samples, the prevailing definition of VOCs is associated
with the PT/GC/MS analytical methods (USEPA, 1986, 1990, Lesage and Jackson,
1992). These are broad-based methods, i.e. designed to measure as many compounds
as possible with a single procedure, comprised of compounds that are relatively
insoluble in water and that have boiling points below 200°C (USEPA, 1986, 1990) or
below 150°C (Lesage and Jackson, 1992). The PT preparation technique promotes
low detection limits (parts-per-billion range) and the MS detector provides positive
compound identification. Volatile compounds that contain polar functional groups
(such as low molecular weight ketones, alcohols, aldehydes, nitriles, and ethers) are
generally soluble in water, do not purge well, and produce broad, tailing GC peaks
that give poor quantitative estimates and are often difficult to identify by MS
(Swallow, 1992). Some polar compounds are included in SW-846 Methods 8240/8260
(USEPA, 1986, 1990), but the recovery of polar compounds is often less than 20%
(Swallow, 1992). New preparation methods for some of the nonconventional
analytes (water soluble analytes such as alcohols, ketones, ethers, and esters) are
included in the Third Update to SW-846 (Lesnik, 1993). These methods are
Azeotropic Distillation (SW-846 Method 5031) and Closed System Vacuum
Distillation with Cryogenic Condensation (SW-846 Method 5032).
-------
While the efficiency of broad-based methods is positive, the use of broad-
based methods as definitions has tended to induce an obtuse view promoting the
likeness of VOCs and the aggregate similarities of compound behavior in soil
(Siegrist, 1993). In fact, the physicochemical properties of various VOCs vary over
orders of magnitude. Differences in vapor pressure, water volubility, and octanol
water partition coefficient impart even larger differences in air-water partitioning
and soil sorption coefficients. Additionally, divergent appraisals of a compound's
physiochemical parameters are the rule, as several plausible measurement and
estimation techniques exist. The reader is referred to Lewis et al. (1991), Devitt et
al. (1987), or recent handbooks (Howard, 1990; Lyman et al., 1990) for listings that
describe the physiochemical properties of VOCs
Occurrence and Ranking of VOCs
The presence of VOCs in ground water is well documented A study of 479
waste disposal sites throughout the United States (Plumb, 1991) reported that VOCs
accounted for 84% of all the detectable events in the composite data set of the
Resource Conservation and Recovery Act (RCRA) Appendix IX organic
constituents (52 FR 25942, July 9, 1987). VOCs were also the most prevalent subsets
when organic compounds found in ground water were ranked by number of sites
and regions (Plumb, 1991).
The Priority List of Hazardous Substances is revised annually as mandated
by the Comprehensive Environmental Response, Compensation, and Liability Act
as amended by the Superfund Amendments and Reauthorization Act. Two
agencies, the U.S. Department of Health Services and the U.S. Environmental
Protection Agency (EPA), are required to produce a list of substances most
commonly found at facilities on the National Priorities List and which, at the
discretion of these agencies, pose the most significant potential threat to human
health (see 52 FR 12866, April 17, 1987). The 1991 list (56 FR 52169, October 17, 1991)
ranked substances with a formula that included three factors (1) frequency of
detection in all media at sites on the National Priority List, (2) toxicity, and (3)
potential for human exposure.
Table 1 lists the volatile chemicals of concern, as ranked on the Revised
Priority List of Hazardous Substances (56 FR 52169, October 17, 1991). The list
includes all compounds that appear in EPA methods for analysis of volatile
organics in soil (SW-846 Methods 8240/8260; USEPA, 1990) and compounds that could
be termed "nonconventional" VOCs (Lesnik, 1993). For comparison, the compounds
are also ranked by frequency of detection in ground water as the only criteria
-------
TABLE 1. VOCs ON THE REVISED PRIORITY HAZARDOUS
SUBSTANCES LIST (HSL)*.
HSL
Rank
4
5
8
10
22
33
35
36
44
49
52
59
60
61
62
63
64
65
66
68
80
96
Ground-water
Frequency
Rank+
15
11
5
2
3
16
9
38
76
161
1
NL
18
10
98
7
8
14
12
NL
6
4
Contaminant
vinyl chloride (chloroethene)
benzene
chloroform (trichloromethane)
trichloroethylene (trichloroethene, TCE)
tetrachloroethylene (perchloroethylene, PCE)
carbon tctrachloride (tetrachloromethane)
toluene (methyl benzene)
hexachlorobutadiene
dibromochloropropane (DBCP)
1,2-dibromoethane (ethylene dibromide, EDB)
methylene chloride (dichloromethane)
methane
naphthalene
1.2dichloroethane
2-hexanone (methyl butyl ketone)
1,1-dicholroethane
1,1,1-trichloroethane
chlorobenzene
ethyl benzene
total xylene
1,1-dicholoroethane
1,2-dichloroethane, trans
SW-846
Analysis
Methods**
A
A
A
A
A
A
A
B
A,C
A,C
A
D
B
A
A3
A
A
A
A
A
A
A
(continued)
-------
TABLE 1. VOCs ON THE REVISED PRIORITY HAZARDOUS
SUBSTANCES LIST (HSL)*.
HSL
Rank
99
101
111
120
126
129
131
132
137
138
139
145
150
175
182
183
187
189
190
191
194
197
201
Ground-water
Frequency
Rank+
21
95
NL
36
22
102
NL
19
26
25
20
27
NL
NL
69
NL
43
NL
75
NL
82
42
NL
Contaminant
acetone (2-propanone)
acrolein (propenal)
l,2-dibromo-3-chloropropane (DBCP)
1 , 1 ,2,2-tetrachloroethane
1 ,2-dichlorobenzene
carbon disulfide
trichloroethane
1 , 1 ,2-trichloroethane
2-butanone (methyl ethyl ketone)
1 ,4-dichlorobenzene
chloroethane
1 ,2,4-trichlorobenzene
hexane
dichlorobenzene
chlorodibromomethane
bromodichloroethane
1,3-dichlorobenzene
1,2-dichloroethane
chloromcthane (methyl chloride)
ethyl ether
bromoform (tribromomethane)
o-xylene
dichloroethane
SW-846
Analysis
Methods**
A,E
A,F
A,C
A
B
A
A
A
A,E
B
A
B
D
B
A
D
B
A
A
E
A
A
A
(continued)
-------
TABLE 1. VOCs ON THE REVISED PRIORITY HAZARDOUS
SUBSTANCES LIST (HSL)*.
HSL
Rank
209
211
212
228
229
234
235
237
247
249
250
253
254
259
262
264
Ground-water
Frequency
Rank+
67
NL
153
NL
96
68
24
29
NL
NL
NL
NL
NL
NL
NL
NL
Contaminant
methyl isobutyl ketone (4-methyl 2-pentanone)
trichlorofluoroethane
pentachloroethane
formaldehyde
1,3-dichloropropane, cis
styrene (vinyl benzene)
trichlorofluoromethane
1 ,2-dichloropropane
m-xylene
p-xylcne
isopropanol
1,2-dichloroethane, cis
dichloroethane
1,3 -butadiene
isopropyl ether
bromodichloromethane
SW-846
Analysis
Methods**
A,E
D
A,G
D
A
A
A
A
A
A
D
A
A
D
D
A
Abbreviation: HSL = Hazardous Substances List; NL = not listed
* 56 FR 52169, October 17, 1991
"Frequency of detection in disposal site ground water (Plumb, 1991).
** SW-846 methods of analysis:
A - Methods 8240/8260
B - Methods 8260 and 8250/8270
C - Method 8011
D - No SW-846 method
E - Method 8015
F - Method 8030
G - Method 8240 notes poor chromatographic behavior for direct injection and it is inappropriate
to use purge and trap for this analyte
-------
(Plumb, 1991). Within the context of organic and inorganic hazardous substances in
all media, four VOCs are included in the uppermost ten priority substances (56 FR
52169, October 17, 1991). On the basis of potential ground- water contaminants,
VOCs comprise the entire list of the top ten organic substances occurring in
ground water (Plumb, 1991).
Statement of the Problem
Soil VOCs are particularly difficult to describe because they occur in several
phases (gas, aqueous solution, sorbed, and nonaqueous-phase liquid [NAPL]) within
heterogeneous media that often must be drilled to obtain samples. VOC collection
and quantification are confounded by the relative mobility of the vapor phase.
The vapor fraction at the time of sampling will depend primarily on the
physiochemical properties of the compound; total concentration of the compound
temperature; soil organic matter content; soil water potentiat; and the
amount, character, and distribution of soil pores. Estimates of the phase
distribution of VOCs in soil are generally based on equilibrium calculations. The
distributions, however, are simplifications of complex media and are difficult to
verify because of the practical limitations of studying multiphasic, multicomponent
soil systems.
Difficulties in measuring soil VOCs occur in sampling, storage, subsampling,
and analysis steps. First, mixed and variable sources of contamination
superimposed on a naturally heterogeneous medium aggravate sampling difficulties
for all soil contaminants, including VOCs however, for VOCs rapid sample
collection without any homogenization steps or compositing is generally necessary
to minimize volatilization losses. This requirement imposes short-range variability
that greatly aggravates the problem of representing field VOC concentrations.
Second, storage of samples prior to analysis has been associated with large losses of
VOCs Rapid and severe loss of soil VOCs occurs during storage in sealed vials at 4
°C (Jenkins et al., 1993 King, 1993). Third, the current laboratory subsampling step
causes losses averaging 60% (Maskarinec et al., 1988). Finally, laboratory analytical
procedures contribute to large data variances. No adequate performance
evaluation materials exist to assist in quality assurance/quality control (QA/QC) of
soil volatiles analyses (Zarrabi et al., 1991). Analytical accuracy as measured by
matrix spike recovery is generally 40% to 120%; however, the recoveries reported
are achieved only when the matrix spike is purged seconds to minutes after
addition of the spike. The use of PT sparging to extract volatiles from soils may be
inadequate when the compounds have been in contact with a particular soil for
months or years (Sawhney et al., 1988; Pignatello, 1990a; Pavlostathis and Jaglal, 1991).
-------
Considered together, the difficulties mean that data obtained by following SW-846
Methods 8240/8260 are prone to poor field representativeness and a large negative
bias.
A symposium was held in January 1993 in Las Vegas, NV, on "Measuring and
Interpreting VOCs in Soils State of the Art and Research Needs." The symposium
served as a national forum for soil-VOC data users and generators of that data to
(1) explore the foundation of the conventional VOC measurement and
interpretation process, (2) examine results from research and practice that have
advanced the understanding of this process, and (3) attempt to develop consensus on
current practices, recommendations for alternative procedures, and critical research
needs. Many abstracts from that symposium have been included in this review.
When available, the symposium proceedings will offer additional perspective and
insight beyond the scope of this literature review.
-------
SECTION 2
INTERPHASE TRANSFERS
Soil VOCs can exist in four distinct phases aqueous, gaseous, sorbed, and
NAPL. This section provides a review of studies on the equilibrium distribution of
soil VOCs among phases aqueous-sorbed, vapor-sorbed, aqueous-vapor, and NAPL-
aqueous/vapor distributions. Studies of kinetically slow or nonequilibrium
distributions of sorbed VOCs are presented in a separate subsection.
AQUEOUS-SORBED DISTRIBUTION
Most VOCs analyzed by SW-846 Methods 8240/8260 are relatively nonpolar,
nonionic organic contaminants (NOCs) that partition into soil organic matter
because of the hydrophobic nature of the compounds. Here the term "partition" is
used to describe a model in which the sorbed material permeates or dissolves
(absorbs) into an organic phase (Chiou, 1989). "Adsorption" refers to the
condensation of vapor or solute on the surfaces of a solid by physical forces or
chemical bonding. The term "sorption" is used to denote uptake of a vapor or
solute without reference to a specific mechanism (Chiou, 1989). These terms are not
consistent in the literature. "Partition" or "distribution" coefficients may be used
more generally to denote the equilibrium ratios of a compound between any two
phases, viz., air-water, soil-water, or soil-air. The term "sorption coefficient" carries
the implicit assumption of a reversible process at equilibrium, that is, a state in
which sorption and resorption are occurring at the same rate.
The sorption of NOCs in saturated porous media, as measured in the
laboratory, can be predicted within an order of magnitude based on properties of
the pollutant, using either the water volubility or the octanol-water partition
coefficient (Kow) and the weight fraction of soil organic carbon (Chiou et al., 1979;
Karickhoff, 1984; and Karickhoff et al., 1979). Sorption is described by a linear
equation of the form S = KdC, where S is the NOC concentration in soil, Kd(also
denoted as Kp) is the sorption coefficient, and C is the equilibrium NOC
concentration in solution. The sorption coefficient, Kd, can be normalized by
dividing it by the fractional organic carbon content of soil, focto give the relatively
invariable organic carbon partitioning coefficient, Koc(Kd/foc= Koc) This linear
partitioning is bounded at the low end by some minimal value of soil organic
matter (e.g., organic carbon fraction exceeding 0.1%; Schwarzenbach and Westall,
1981) and bounded at the high end by some fraction of sorbate (NOC) volubility
-------
(Piwoni and Banerjee, 1989). Brusseau and Rao (1989) place the upper boundary in
terms of the sorbate activity coefficient; a concentration threshold of 0.056 M is
established, below which the sorbate activity coefficient is constant and linear
isotherms are likely.
The order-of-magnitude error in estimates of sorption based on Koc occurs
because there are inherent limitations to the empirical correlations. Figure 1 shows
the 95% confidence limits for Kocas estimated by an empirical correlation with the
KoWfor a diverse set of 34 NOCs, including pesticides, polycyclic aromatic
compounds, and 7 VOCs (Hassett et al., 1983). Graphical depiction of the log-log
correlation demonstrates the approximate 6 orders of magnitude over which the
relationship was generated and the order of magnitude error for any particular Koc
value within the 95% confidence limits of the correlation. The situation is similar
for the other empirical correlations with Kow or water volubility (Hassett and
Banwart, 1989).
Other reasons for differences between estimated and measured Kdvalues
generally fall into two categories the contribution of soil mineral matter, or
differences in the chemical nature of organic matter. Mingelgrin and Gerstl (1983)
reviewed the literature on soil sorption of nonionic compounds and discussed the
many limitations of Kocestimates. They cited work by some researchers which
showed that removal of organic matter from soils and sediments had relatively
little effect on, or actually increased, sorption of nonionic compounds. In part, this
phenomenon may be an artifact caused by the difficulty in measuring small quan-
tities of organic matter and small values of sorbate uptake (Rutherford et al., 1992)
Garbarini and Lion (1986) studied sorption of toluene and TCE by several
organic sorbents (viz., whole soil; humic acid, fulvic acid, and humin extracts of soil;
tannic acid, lignin, zein, cellulose; ethyl ether extracted soil, and the resulting
extracts, which contained soil fats-waxes-resins). They found widely varying
affinities for the chemicals that could not be explained by the organic carbon
content of the sorbent. Lignin followed by zein had the highest sorption
coefficients of all sorbents studied. On an organic carbon basis (Koc, however, the
fats-waxes-resins sorbed the largest amount. Observations on the relative degree of
decomposition, the direct or indirect effects of the inorganic matrix, and the
contribution of relatively hydrophilic oxygen-containing functional groups in
organic matter were explored by multivariate regression analyses. Results showed
that using the oxygen and carbon content of a sorbent yielded more accurate
predictions of the sorption coefficient than did carbon content alone.
10
-------
PREDICTION OF EQUILIBRIUM SORPTION COEFFICIENTS
7
6
5
o
3 3
2
1
95% CONFIDENCE
INTERVAL
= 0.088 + 0.909(±0.002) log Kow
f2 = 0.93
01 2345678
LOG Kow
Figure 1. Log Koc- Log Kowrelationship (after Hasset and Banwart, 1989.)
11
-------
Similarly, Rutherford et al. (1992) found that peat sorbs relatively more
NOCs than does cellulose. They propose that differences correlate with the polar
to nonpolar group ratio [(O+N)/C]of the organic matter. They suggest, however,
that variability in soil organic C content of organic matter falls within 53% to 63%
for most soils. On the basis of this assumption, and confining the prediction to soils
that contain at least 0.2% organic matter (or approximately 0.1% organic carbon),
the variation in partitioning coefficients attributable to the nature of the organic
matter is within a factor of 3.
In addition to indigenous soil organic carbon, anthropogenic sources of
organic carbon (such as residual petroleum) act as a highly effective partition
medium for organic contaminants. Boyd and Sun (1990) showed that residual
petroleum was approximately 10 times more effective than soil organic matter as a
partition medium for pentachlorophenol or toluene. Soil distribution coefficients
were predicted as the sum of the partitioning into natural organic matter and
partitioning into the residual oil or polychlorinated biphenyls (PCB) phase. Oil-
water distribution coefficients were evaluated in a manner similar to that of
octanol-water distribution coefficients. In soils that contained residual petroleum
or PCBS, the magnitude of the oil-water coefficient greatly enhanced the NOC
uptake and was believed to limit the effectiveness of some remediation efforts
(Boyd and Sun, 1990). Bouchard et al. (1990) studied the same phenomena on soils
that were treated with unleaded gasoline in the laboratory to form a residual
hydrocarbon fraction. By comparing sorption of benzene and naphthalene on
treated and control soils, they showed that the most profound effect occurred on
those sorbents low in natural organic carbon.
Estimation of VOC sorption based on Koc values is recommended only for
soils that contain more than O.l% organic carbon (or more than O.2% organic
matter). No procedure exists for the estimation of VOC sorption in soil that is
very low in organic matter. Mineralogical effects related to the surface charge of
individual mineral species will affect sorption when organic matter is low. In soils
that contain mixed wastes, such as the soils in landfills, the nature of the organic
matter will have the greatest effect where the organic input is extremely fresh
(virtually undegraded) or where organic carbon is exceptionally aged (coal is an
extreme example). The presence of anthropogenic, nonpolar organic liquid wastes
will further increase sorption of NOCs.
12
-------
Soil sorption of polar VOCs (e.geketones, aldehydes, nitriles) is generally
nonlinear and varies with the type and quantity of clay minerals in the soil.
Estimates of sorption for polar compounds may be based on empirical relationships
generated by chemical class (Karickhoff, 1984).
NONEQUILIBRIUM SORPTION
True sorption equilibrium may require weeks to months to achieve
(Karickhoff, 1984). A growing body of literature is addressing the impact of
nonequilibrium or rate-limited sorption-desorption on estimates of organic chemical
distribution coefficients and on solute transport modeling (Brusseau and Rao, 1989).
Equilibrium sorption occurs rapidly (2 to 48 h), and little change is observed when
the experimental time is doubled, as represented by sorption coefficients (Kdor Koc)
"Sorption nonequilibrium" denotes slow sorption-desorption processes and
nonreversible sorption. The term "chemical nonequilibrium" refers to rate-limited
interactions between the sorbate and sorbent (e.g. chemisorption). Conventional
VOCs are largely devoid of functional groups that participate in chemisorption,
thus irreversible sorption is unlikely. Intraaggregate or intraorganic matter
diffusion in soil is believed to be the rate-limiting process for NOC sorption
(Steinberg et al., 1987; Hamaker and Thompson, 1972; Pignatello, 1990b; Brusseau et
al, 1991, Ball and Roberts, 1991b).
The term "nonequilibrium" refers more generally to the concept that
numerous physical and chemical factors preclude equilibrium in field
environments. The existence of secondary soil structures, including aggregates,
fractures, and bedding, creates what are termed transport or physical
nonequilibrium effects (Brusseau and Rao, 1989). Transport-related nonequilibrium,
resulting from the existence of a heterogeneous flow domain, is discussed in the
section on modeling in this review. Nonequilibrium sorption is a term that spans a
scale ranging from microscopic effects on soil surfaces to macroscopic diffusion or
"mass transfer" effects within pores of varying sizes.
Laboratory sorption data have been found to exhibit a two-stage approach to
equilibrium: a short initial phase of fast uptake, followed by an extended period of
much slower uptake (Brusseau and Rao, 1989; Harmon et al., 1989). Sorption is
viewed as a rapid partitioning of NOCs into organic matter (within hours or days),
followed by a much slower uptake phenomenon involving intraaggregate or
intraorganic matter diffusion. Batch resorption studies similarly show a rapid
release phase followed by an extended slow resorption phase (Pavlostathis and
Mathavan, 1992). Brusseau and Rao (1989) estimated that, following the initial rapid
13
-------
phase of sorption, approximately 100% more sorption will occur by slow sorption
processes. Measurement techniques do not distinguish the scale of the continued
sorption, and therefore slow diffusion into microsites is incorporated into the term
nonequilibrium sorption.
Evidence for nonequilibrium sorption in a field soil was first reported by
Steinberg et al. (1987). The soil fumigant 1,2-dibromoethane (EDB) was being
detected in ground water, but extensive searching had not identified the source.
Hot methanol (75 °C for 24 h) was used to extract the overlying agricultural soils.
Some of the soil had received no known applications of EDB for as long as 19 years
before the hot solvent extraction. EDB residues in the 1OO ng/g range were found
in the soils. The residue was resistant to volatilization and microbial degradation,
and the release into aqueous solution was extremely slow at 25 °C. Increasing the
temperature to 75 °C released greater than 25% of the EDB from a fine sandy loam
soil in less than 3 h. Release of residual EDB to aqueous extracts from pulverized
soil increased with degree of pulverizing (time in ball mill), leading the authors to
conclude that the EDB residues were occluded in soil micropores.
Pignatello (1990a) studied the potential of several halogenated hydrocarbons
(all VOCs) to form slowly reversible or "residual" VOC fractions in soils. Nine
halogenated aliphanes were added to soil. The residual fraction was defined as the
proportion of VOCs that remained after repeated washing with water residual
VOCs were subsequently extracted with hot acetone. Results demonstrated the
trend toward formation of a residual sorbed fraction of VOCs that increases with
equilibration time. The residual VOC concentrations varied among the compounds
and by treatment. In general, the alkenes (TCE, tetrachloroethene, and 1,3-
dichloropropene) produced greater residual concentrations than the alkanes (carbon
tetrachloride, 1,1,1 -trichloroethane, 1,2-dibromopropane, 1,2-dibromo-3 chloropropane,
and 1,2-dibromoethane).
Further studies (Pignatello, 1990b) presented additional data, demonstrating
the effects of (1) incubation time, (2) initial concentration, and (3) different soil
pretreatments on the formation of soil residual VOC concentrations). Again,
residual concentrations increased with incubation time, but TCE residual
concentrations appeared to level off after 8 days at a high loading rate (TCE-spike
addition of 104mg/kg). Tetrachloroethene was extremely fast at forming a large
residual concentration; an order-of-magnitude increase in residual concentration
after 7 days incubation was observed for tetrachloroethene (0.41 mg/kg after 1 day
and 5.6 mg/kg after 7 days). The residual concentrations increased with increasing
concentrations of chemical present during the sorption period. The residual
14
-------
concentrations showed a log linear increase with the final solution concentration.
The slope of the concentration dependence varied by compound for example,
tetrachloroethene, with a slope of 0.90, had a nearly linear response, and
tetrachloroethene had a slope of 0.73. Residual TCE concentrations were close to
half-order in TCE concentrations in the medium (slope of 0.49), whether soil was
treated with pure compound or aqueous solution.
Ball and Roberts (199la, 1991b) investigated the long-term sorption equilibria
of 14C-labeled tetrachloroethene and 1,2,4,5-tetrachlorobenzene with low-carbon
aquifer material (organic carbon content of less than 0.021% in bulk sample, less
than O.l% in any specific particle-size fraction). Samples were sterilized and sealed
in glass ampules to avoid losses from biodegradation and volatilization. Sorption
was calculated by difference from initial solution concentration after accounting
for other losses (e.g., headspace). Compound recovery from blanks was greater than
90% during the 100-day study. Rate studies showed that tetrachloroethene sorption
by the bulk material reached equilibrium in 30 days but that pulverized material
reached sorption equilibrium within 1 day. The estimated rate constants for
pulverized material were between 40 and 80 times higher than rate constants for
unaltered solids (Ball and Roberts, 199 Ib). The sorption capacities were essentially
the same for bulk and pulverized material, indicating that rapid sorption
experiments could be accomplished using pulverized samples. Pulverized samples
had the added advantage of lower relative errors in sorption estimates (27.6%
relative error for bulk samples versus 8% relative error for pulverized samples; Ball
and Roberts, 199la). Results were interpreted by a physical diffusion model (Ball
and Roberts, 199 Ib). The effective pore diffusion coefficients were estimated to be
roughly 2 to 3 orders of magnitude lower than bulk aqueous diffusivities in the
aquifer material studied. Diffusive length was dramatically reduced by pulverizing,
consistent with the proposed mechanism of intragranular diffusion (Ball and
Roberts, 1991b).
Sorption by specific particle-size fractions revealed that sorption was greatest
by the largest size fractions (Ball and Roberts, 199la). The larger size fractions had
the greatest organic matter contents and the greatest surface areas on a weight
basis. (The larger particle-size fractions evidently consisted of soil aggregates.)
Sorption coefficients exceeded values predicted on the basis of organic partitioning
by an order of magnitude or more. Average measured log Kocvalues were 5.1 and
3.6, and average calculated log Kocvalues from the literature were 4.0 for
tetrachlorobenzene and 2.4 for tetrachloroethene. Ball and Roberts (1991a)
15
-------
suggested that either an exceptionally adsorptive organic phase existed in the
aquifer material or that mineral matter was partly responsible for the observed
sorption.
Pavlostathis and Mathavan (1992) studied (1) the resorption kinetics of five
field-contaminated soils and (2) the effect of residence time (up to 15 months) on a
laboratory-contaminated soil. Resorption of TCE, tetrachloroethylene, toluene, and
xylene in soil-water mixtures was biphasic. A fast resorption phase was complete
within 24 h, followed by a very slow resorption phase. Methanol extractions of the
soil pellet following centrifugation of the soil-water samples (13 h at 20 °C, and 30%
soil by weight) showed that a substantial portion of the sorbed contaminant mass
(48 to 94%) resisted resorption in deionized water after 7 days. The rate and extent
of resorption did not correlate with the soil properties surveyed (organic carbon
content, cation exchange capacity, or specific surface area) or with the sorbate
water solubility. In a separate study, soil spiked with TCE was treated with sodium
azide to reduce biological activity and was stored in the dark at 4 °C. Samples were
analyzed at 2.5, 5.5, and 15.5 months by six successive washings in deionized water
followed by extraction of the soil pellet with methanol. The TCE that resisted
resorption in water was 10% of the total amount that sorbed at 2.5 months but
increased to 45% of the total amount that sorbed at 15.5 months. The partition
coefficient, as observed by the successive washings, also increased with time; Kpwas
0.4 mL/g at 2.5 months and 1.5 mL/g at 15.5 months. The amount of TCE that
resisted resorption might be shown to be even greater if a hot methanol extraction
were used (demonstrated by Sawhney et al. [1988] for EDB-contaminated soil).
VAPOR-SORBED DISTRIBUTION
Although sorption of chemicals on saturated soil has been studied
extensively, reports on the sorption of VOCs on unsaturated or dry soils are
relatively few. Of these studies, those that compare vapor uptake on dry versus
wet soil show that dry soil vapor uptake is greater than that of wet soils, is
nonlinear, and is suppressed by the presence of water in a nonlinear manner (Chiou
and Shoup, 1985; Poe et al., 1988; Chiou, 1989; Ong and Lion, 1991a, 1991b; Rhue et alv
1988). Chiou and Shoup (1985) explain soil as a dual sorbent in which the mineral
matter functions as a conventional adsorbent (physically covering soil surfaces) and
organic matter functions as a partition medium. Polar water molecules are
strongly adsorbed on mineral surfaces and effectively displace organic compounds
as the soil water content or relative humidity increases. In the absence of water
vapor, strong mineral adsorption of organic vapors exceeds the effect of
partitioning with organic matter.
16
-------
Chiou and Shoup (1985) studied the vapor sorption of water and five organics
(benzene, chlorobenzene, p-dichlorobenzene, m-dichlorobenzene, and 1,2,4-
trichlorobenzene) on an oven-dried (140 °C) soil. Gravimetric determinationsof
VOC sorption were performed using a dynamic, temperature-controlled, vapor-
sorption apparatus. Results of soil uptake were plotted against relative vapor
concentrations (equilibrium partial pressure divided by saturation vapor pressure of
the compound) to normalize the activity of each compound with respect to its own
pure state and to allow for the comparison of vapor uptake between different
compounds and between vapor and normalized liquid uptake (equilibrium liquid
concentration divided by the volubility of the compound) of the same compound.
Sorption isotherms for all compounds on dry soil were distinctly nonlinear. The
capacity of the soil for sorption was greatest for water and the presence of water
vapor sharply reduced the soil sorption capacity for the organic compounds
Water-saturated sorption was about 2 orders of magnitude less than dry-soil
sorption of the VOCs Chiou and Shoup (1985) observed that at 90% relative
humidity the vapor sorption isotherms for m-dichlorobenzene and 1,2,4-
trichlorobenzene fall close to the corresponding isotherms for aqueous solution.
For benzene, however, the 90% relative humidity isotherm deviated from the
aqueous isotherm (positively) by more than a factor of 5. The authors suggested
that error in the measurement of the relative humidity may have caused this
deviation.
Poe et al. (1988) looked at vapor phase sorption of five VOCs (benzene,
dichloropropane, methylcyclohexane, ethyl ether, and methanol) on four air-dry
soils and found consistent adsorption capacities among them. The relative order of
adsorption in the four soils remained the same regardless of the chemical
compound. The soil that had the highest clay content and largest surface area
adsorbed the greatest amount of each VOC. For the two soils that had similar clay
content and surface area, however, the soil that sorbed the greatest amount of
VOCs had a lower organic carbon content, suggesting that organic matter may
block mineral surface sites. Both Poe et al. (1988) and Chiou and Shoup (1985) found
that sorption increases as compound polarity increases.
Ong and Lion (199la) used a headspace technique to study the sorption of
TCE over a range of moisture contents by soil components, viz., alumina, hydrated
ferric oxide, montmorillonite, kaolinite, and humic-coated alumina. They found
that sorption by oven-dry solids was 2 to 4 orders of magnitude greater than
sorption by wet solids. A description of vapor sorption as a function of moisture
content was presented (Figure 2). When moisture contents ranged from oven-dry to
water sufficient to provide monolayer coverage of solid surfaces, vapor sorption
17
-------
decreased linearly with moisture content. Water sufficient to form 1 to 5
monolayer on solid surfaces exhibited complex behavior, attributed to interactions
between TCE vapor and surface-bound water. At approximately five monolayer
of water a sorption minimum was observed, and above five monolayer sorption
increased slightly with moisture content. Ong and Lion (199la) showed that the
gradual increase in sorption at moisture contents greater than five monolayer
could be accounted for by vapor dissolution, as predicted by Henry's Law constant
(see Vapor-Aqueous Distribution Section below). Five monolayer of water
corresponds to 25% moisture by weight on the high-surface area alumina and to
2.9% moisture by weight on the low-surface area kaolinite in this study.
Gravimetric studies of TCE and water vapor sorption on soil minerals
(montmorillonite, kaolinite, iron oxide, silica, and alumina) and on humic-coated
alumina and humic acid showed that surface area serves as a good measure of the
adsorptive capacity of dry solids (Ong and Lion, 199 Ib). However, variability in
relative sorption isotherms among the sorbents (calculated as the ratio of the
sorbed quantity to the monolayer capacity) demonstrated that other factors are also
involved. Specific sorbate-sorbent interactions and vapor condensation in
micropores are cited to explain the variability among sorbents. Sorption of TCE
onto the mineral solids in the presence of water at several levels of relative
humidity (RH) showed that the polar water molecules are sorbed preferentially
over the nonpolar VOCs Addition of water to humic acid resulted in a large
increase in the amount of TCE sorbed; TCE sorption by humic acid at 80% RH was
much greater than at O% RH and much greater than saturated aqueous sorption.
Expansion of the oven-dry humic acid due to hydration and exposure of internal
surfaces for sorption/condensation as the RH increases was proposed. Although
montmorillonite also expands with the addition of water, its sorption capacity for
TCE decreased, indicating that the interlamellar pores were not readily available
for the uptake of TCE.
Equilibrium vapor-phase adsorption of VOCs is described by the classical
Brunauer-Emmett-Teller (BET) isotherm (Jurinak and Volman, 1957; Chiou and
Shoup, 1985 Poe et al., 1988; Rhue et al., 1988), indicating that multimolecular layer
adsorption occurs. At very high vapor pressures, increased uptake of VOCs
through vapor condensation is regulated by available pore space (Ong and Lion,
199 Ib). This effect is most likely to occur close to field sources of contamination,
such as nonaqueous-phase organic liquids.
Vapor sorption of VOCs has not been shown in the field, but Ong and Lion
(199la) suggest that field conditions dry enough to permit vapor sorption exist.
18
-------
A
o>
o
£
0)
o
o
c
«.
h.
o
in
1
<^
t
V
\
\
\
\
\
2
<^
\
\
\
V
3
1. OVEN-DRY TO SINGLE
>,
SURFACE MONOLAYER Ol
2. 1-5 SURFACE WATER I
3. >5 MONO-LAYERS OF S
WATER
«"j
.AY
URI
— — •
Moisture Content (%)
Figure 2. Vapor phase sorption as a function of moisture content
(after Ong and Lion, 199la.)
19
-------
Vapor extraction systems generally require moist air to improve the efficiency of
vapor removal (V. Fong, personal communication). Smith et al. (1990) found that
vapor sorption would not be important at the Picatinny Arsenal in New Jersey. In
the laboratory, they measured the vapor-phase sorption of TCE and water using
the experimental apparatus of Chiou and Shoup (1985). As an example, a surface
soil sample of 4% organic carbon and 13.5% clay reached water vapor saturation at
6% moisture on a weight basis. The vapor saturation soil-moisture content is
defined as the mass of water sorbed by the soil in equilibrium with water vapor at
its saturation vapor pressure. The field soil moisture concentrations were all much
greater than the vapor saturation moisture contents (on the day these were
sampled) and thus they concluded that vapor-phase sorption was unimportant at
this site.
VAPOR-AQUEOUS DISTRIBUTION
Vapor-aqueous equilibrium distributions for dilute solutions at or below one
atmosphere pressure are directly proportional. Henry's Law states that the
equilibrium distribution of a compound, i, between gas and liquid phases is linearly
related
P= H.X,
where Ptis the partial pressure of compound i, H^s Henry's Law constant for
compound i at a given temperature, and Xi is the mole fraction of compound i.
The distribution can be expressed in terms of the concentration of a compound in a
liquid, yielding a proportionality constant, or Henry's constant, that has units of
kPa nr'molXor arm rn'mol"1). Also, a dimensionless form of Henry's constant (gas
phase molarity/liquid phase molarity) is common; this form is related to the other
two constants by expressing the gaseous partial pressure in terms of moles through
the ideal gas law.
Although strictly applicable only for dilute solutions, Henry's Law has been
found to persist to the point of saturation for many chemicals (Spencer and Cliath,
1970). Hence, the dimensionless Henry's Law constant, KH, may be calculated as the
ratio of saturated vapor density, C0* (g/m3), to water solubility, CL*(g/m3)
K = r */C *
rvH *^a >^L •
Further evaluation of the use of vapor pressure and volubility data to estimate
Henry's constants was reported by Munz and Roberts (1987) who found good
20
-------
agreement between experimental and predicted values for seven halocarbons
(bromoform, hexachloroethane, chloroform, TCE, 1,1,1-trichloroethane, carbon
tetrachloride, and dichlorodifluoromethane) and poor agreement for one compound
(tetrachloroethene). They suggested that errors in the estimation of Henry's Law
constant occur due to the wide range of volubility values reported in the literature
for many compounds. No effect of solute concentration on the Henry's Law
constant for the solute was observed for solute-liquid mole fractions as large as 103
(Munz and Roberts, 1987).
For soil that is saturated or unsaturated, but not dry, Henry's Law constants
can be used to estimate the distribution of a VOC between the liquid and gaseous
phases The effect of temperature on Henry's Law constant must be considered
Henry's Law constant increases by a factor of approximately 1.6 for every 10 °C rise
in temperature (Munz and Roberts, 1987). Data to demonstrate the effect of
temperature (10 °C and 30 °C) on the Henry's Law constant of five VOCs can be
found in Munz and Roberts (1987).
NAPL-AQUEOUS/VAPOR DISTRIBUTIONS
Soil contamination by a NAPL (e.g., gasoline, chlorinated solvents) produces a
residual soil-NAPL fraction. Pools or "gaglia" of pure phase liquid are retained in
pore spaces by capillary forces. The residual NAPL saturation, or amount of
NAPL that will be held against gravity, depends on the soil pore structure but is
estimated to range from 5 to 40% of the pore space of most soils. With time, the
residual fraction dissipates by volatilization and solubilization. Complex NAPL
mixtures, such as petroleum derived fuels and lubricants, change in composition as
the volatile and soluble components weather in chemical sequence from the
residual fraction (Bouchard et al., 1990).
The release of a VOC from residual NAPL will be governed by (1) the VOC
volubility in combination with the rate of water flowing through the soil and (2)
the VOC volatility in combination with the vapor concentration gradients and
pressure gradients acting on the vapor phase. The aqueous volubility of some
chlorinated solvents alone and in binary mixtures was reported by Broholm et al.
(1992). Solubilities at 23 °C to 24 °C ranged from 242 mg/L for tetrachloroethene to
8,668 mg/L for chloroform. The chlorinated solvents exhibit ideal behavior in
mixtures. That is, the aqueous concentration of a compound can be predicted by
the mole fraction of the compound in the organic phase.
21
-------
SUMMARY OF INTERPHASE TRANSFERS
Rapid sorption of VOCs by wet or moist soils is generally predicted from the
organic-carbon partitioning coefficient (Koc) for the compound and from the soil
organic carbon content. The Kocis estimated through empirical log-log correlations
with the octanol-water partitioning coefficient (Kow) or water volubility of the
compound. An order-of-magnitude error is associated with the 95% confidence
limits of these correlations (Hassett and Banwart, 1989). Deviations from predicted
values by a factor of 3 may be attributed to qualitative differences in soil organic
matter (Rutherford et al., 1992). Sorption is controlled by mechanisms other than
organic partitioning and Koc predictions are indefensible if (1) the soil contains less
than 0.1% organic carbon (or 0.2% organic matter) or (2) the compounds contain
polar functional groups (e.g, ketones or nitriles).
Slow sorption processes, termed "nonequilibrium sorption" are believed to be
controlled by diffusion within soil organic matter (intraorganic diffusion) or by
diffusion into soil micropores. Following the initial rapid "equilibrium" sorption
phase, an estimated 100% greater sorption may occur given sufficient time
(Brusseau and Rao, 1989). With time, the soluble, volatile, and easily desorbed
phases dissipate, and the nonequilibrium fraction becomes the dominant form of
soil contamination (Steinberg et al., 1987). The significance of the nonequilibrium
fraction will be defined by the prevalence of this phenomenon and the resulting
rates of resorption. No data are available for identifying rates and conditions
under which nonequilibrium VOCs are released.
Dry soil sorbs 2 to 4 orders of magnitude more volatile compounds than the
same soil sorbs when wet. Vapor sorption proceeds by physical adsorption, rather
than by organic partitioning. Polar water molecules rapidly and effectively
displace surface-adsorbed VOCs The significance of vapor sorption under field
conditions has not been documented, but cannot be dismissed without investigation.
Vapor-aqueous distributions of VOCs are predicted using Henry's Law con-
stants. The values are often estimated by the ratio of the saturated vapor density
to the water volubility of a compound. Henry's Law constants increase by a factor
of approximately 1.6 for every 10 °C rise in temperature (Munz and Roberts, 1987).
Residual NAPL saturation in soil is a function of soil porosity and is not
predicted by Koc sorption. Contamination by NAPLs results in a NAPL residual
saturation fraction held by capillary forces (against gravity) in soil pores (Bouchard
etal., 1990).
22
-------
SECTION 3
DEGRADATION
VOCs may be lost from the soil or a soil sample by either microbiological or
abiotic degradation. A comparison of sterilized and viable samples is generally
used to distinguish between the degradation mechanisms. Field studies, and
occasionally laboratory studies, commonly report dissipation rates because
additional pathways of chemical losses (volatilization, sorption, or movement with
soil water) are not measured during the study. This section discusses reports on
microbiological and abiotic degradation, and on the factors affecting degradation
rates.
MICROBIOLOGICAL DEGRADATION
Although microorganisms are sharply reduced in number and kind with soil
depth, significant microbial degradation occurs for some chemicals throughout the
vadose zone and into saturated substrata (e.g. Wilson et al., 1983; Barker et al., 1987;
Sulfita, 1989). Data on the biodegradation of many organic compounds in soil is
summarized by Dragun (1988). It is evident that the experimental conditions
influence results, and that published values used to estimate biodegradation rates
must be carefully evaluated before equating results to field or sample storage
conditions.
Soil microbiological transformation of low-molecular-weight aliphatic and
aromatic hydrocarbons figures prominently in the environmental fate of these
compounds. Generally, the shorter the hydrocarbon chain, the more rapid the
oxidative biodegradation (Alexander, 1977). For example, bacteria able to oxidize
volatile hydrocarbons proliferate in the vicinity of natural gas leaks, consuming
available 02and creating locally O2-deficit regions (Alexander, 1977). Benzene,
toluene, xylene, ethylbenzene, and naphthalene exist in a dynamic state in soils, and
they are both synthesized by and destroyed by microorganisms (Alexander, 1977;
Dragun, 1988). The total biological oxygen demand for degrading these aromatics is
very large (values such as 9 moles of 02per mole of toluene), which implies that
the subsurface degradation will be limited frequently by 02 diffusion.
Barker et al. (1987) studied the biodegradation of benzene, toluene, and m-, p-,
o-xylene injected into the water table in a shallow sand aquifer in Borden, Ontario.
Essentially complete removal of a field-injected pulse was reported within about 1.2
23
-------
years; benzene was the most persistently retained compound in the system.
Horizontal layers of near-zero dissolved oxygen in the saturated strata
corresponded to contaminant persistence 32 days after injection. This finding
confirmed that diffusion of oxygen was controlling the field biodegradation rates.
Laboratory experiments with anaerobic microcosms and sealed vials demonstrated
that the losses could be attributed to biodegradation under aerobic conditions.
Phenolic and acidic breakdown products were detected in anaerobic microcosms.
Results in septum-sealed vials snowed zero-order kinetics and lack of anaerobic
transformation products, indicating that oxygen leaked into the vials. Laboratory
zero-order rate constants were compared with field ground-water degradation rate
constants, and the laboratory rate was about 1.5 times the field rate for benzene and
toluene. Field rates were higher than laboratory rates for m- and p-xylene.
Halogenated aromatics degrade only under aerobic conditions (Kobayashi
and Rittman, 1982). Wilson et al. (1983) found chlorobenzene to biodegrade at a rate
of approximately 5% per week under aerobic conditions, although there was no
significant degradation under anaerobic conditions. Vadose material collected from
depths of 1.2 and 3.0 m showed degradation of chlorobenzene, but no degradation
was observed in aquifer materials (5-m depth) or in autoclaved (sterilized) samples
(Wilson et al., 1983). Wilson and McNabb (1983) found 1,2-dichlorobenzene and 1,4-
dichlorobenzene much more likely to degrade than 1,3-dichlorobenzene in aquifer
material.
In contrast to the compounds discussed above, microbial degradation of
volatile aliphatic chlorinated hydrocarbons occurs primarily by reductive
dehalogenation, that is, the replacement of chlorine by hydrogen under anaerobic
conditions (Smith and Dragun, 1984). Reductive dehalogenation of
tetrachloroethene, 1,1,1-trichloroethane, TCE, and tetrachloromethane has been
demonstrated at Eh values below 300 mV and pH 6.8 to 7.0, although carbon-
halogen bonds do not rupture in microcosms devoid of viable microorganisms
(Parsons et al., 1985). Such highly reducing conditions do not commonly occur in
most soils but could occur in water-logged soils, under landfills, or in highly
contaminated soils exposed to fluctuating water tables.
Cline and Viste (1985) have shown that anaerobic degradation of chlorinated
solvents occurs under landfills and under solvent recovery plants. Three landfills
used for disposal of municipal and industrial wastes received the parent compounds
(solvents) 1,1,1-trichloroethane, TCE, and tetrachloroethene. Breakdown products
included dichloroethanes, chloroethane, dichloroethenes (cis- and trans-), and vinyl
chloride. The breakdown products dominated in ground water downgradient from
24
-------
the waste disposal boundaries. Data from solvent recovery sites and an industrial
site showed that locations receiving both chlorinated and nonchlorinated solvents
had much higher percentages of breakdown products downgradient than sites
receiving only chlorinated solvents. The authors concluded that the nonchlorinated
carbon source promotes rapid co-metabolism of the chlorinated solvents by
microorganisms.
The reaction of vinyl chloride is a notable exception to anaerobic
degradation among the halogenated aliphanes. It is often a breakdown product of
other, more highly chlorinated compounds and is highly persistent under anaerobic
conditions. In contrast, vinyl chloride degrades rapidly under aerobic conditions
(Sims, 1990; Hartmans and de Bent, 1992). Dragun (1988) reports chloroethane as a
degradation product, and Hartmans and de Bent (1992) observed an intermediate
epoxide chlorooxirane degradation product (a reactive species), but subsequent
degradation products were not identified.
No detectable degradation of 1,2-dichloroethane, l,l,2~trichloroethane, TCE, or
tetrachloroethene was found in soil and sediment samples collected near Lula,
Oklahoma, from vadose depths of 1.2 and 3 m and from below the water table at 5
m (Wilson et al., 1983). These compounds were highly refractory in a related
laboratory study that found no degradation when samples were incubated for 16
weeks in a N2 atmosphere. The same study found that bromodichloromethane was
degraded in sediment from the saturated zone, chlorobenzene was degraded in
vadose soil, and toluene was rapidly degraded by microorganisms in soil collected at
all depths. There was no detectable degradation of any of the chemicals tested
after the soils were autoclave.
Under certain circumstances, such as in the presence of natural gas,
oxidation and dechlorination of halogenated aliphanes may occur. A study by
Wilson and Wilson (1985) showed that TCE was rapidly and effectively removed
from a soil column exposed to a stream of natural gas (0.6%). Soil columns poisoned
with sodium azide, or in the absence of natural gas, allowed significantly more TCE
to pass through. Enzymes produced by methanotrophs were credited with the
degradation of TCE in the soil exposed to natural gas. The authors concluded that
other chlorinated aliphanes will also undergo oxidation and dechlorination in the
presence of natural gas.
25
-------
ABIOTIC DEGRADATION
Mineral surfaces often serve as catalysts for abiotic organic reactions such as
hydrolysis, elimination, substitution, redox, and polymerization (e.g. a review by
Voudrias and Reinhard, 1986) Transition metal cations on or in clays can act as
Lewis acids by accepting electrons from organic compounds. The dissociation of
water coordinated to exchangeable cations of clays results in Bronsted acidity. At
low water content, the Bronsted sites may produce extreme acidities at the clay
surface. Dragun (1988) lists benzene, ethylbenzene, naphthalene, and toluene as
chemicals that may undergo free-radical oxidation in soil. Reactivity on clay
surfaces is highly specific and is most commonly observed with compounds that
have polar functional groups. For example, acetonitrile undergoes hydrolysis to
form acetamide in the presence of a Wyoming bentonite, but not in the presence of
a Montana vermiculite (Dragun, 1988). Although abiotic degradation of some
compounds on some soils may occur, the extent of such reactions cannot be
estimated from available data.
Chemicals that may rapidly polymerize in the presence of water include at
least three VOCs: acrolein, acrylonitrile, and vinyl acetate (Dragun, 1988). Vinyl
acetate has been removed from the Revised Priority List of Hazardous Substances
(56 FR 52169,17 October 1991), and information on the environmental significance of
the polymerization reaction for the other two compounds was not found.
Abiotic reactions of chlorinated aliphatic VOCs in water have been
documented. Hydrolysis and oxidation, the principal chemical reactions, yield
products that tend to be water-soluble intermediates which are often difficult to
detect at trace levels. Evidence for these reactions is, therefore, the disappearance
of the chemical under sterile conditions at carefully controlled temperatures.
Tetrachloroethene and TCE have been reported to degrade in water, with half-lives
of about 0.75 and 0.9 years, respectively, at room temperature (Dilling et al., 1975).
The hydrolysis of 1,1,1-trichloroethane has been reported with a half-life of 0.5 to 0.8
years at 25 °C (Dilling et al., 1975).
Hydrolysis rates reported for a given chemical can vary over a few orders of
magnitude depending on initial concentration or on procedural differences. The
hydrolysis half-life of trichloromethane has been reported as 1.25 years Dilling et
al., 1975) and 3,500 years (Mabey and Mill, 1978). Similarly, the hydrolysis half-life
for dichloromethane is reported as 1.5 years Dilling et al., 1975) and 700 years
(Mabey and Mill, 1978). Smith and Dragun (1984) suggested that if the longer half-
lives (calculated from kinetic studies carried out at elevated temperatures) are
26
-------
reliable, the shorter experimental values must result from processes other than
hydrolysis. However, variable hydrolysis half-lives reported for
tetrachloromethane, i.e., 7 years at 1,000 mg/L and 7,000 years at 1 mg/L, have been
attributed to a hydrolysis reaction that is second order in tetrachloromethane
(Mabey and Mill, 1978).
Recognizing variation in the data, these values all suggest that hydrolysis
should not interfere with typical laboratory procedures to measure halogenated
alkanes in water over holding times of a few days. Hydrolysis rates in the presence
of soil may be slower or faster than those reported in water, depending on the soil
and compound-specific effects of pH, Eh, sorption, or surface-catalyzed reactions_
Abiotic soil degradation studies are inherently less conclusive than studies in water
as exemplified in the study of Anderson et al. (1991) reviewed below.
FACTORS AFFECTING DEGRADATION RATES
When biotransformation does occur, it will usually be more rapid than
abiotic transformation (Vogel et al., 1981 Dragun, 1988). Environmental half-lives
from abiotic processes for halogenated aliphatic compounds are generally on the
order of years to hundreds of years (Vogel et al., 1987), although the same
compounds exhibit half-lives of days to weeks in the presence of microbially active
soil or static-flask culture conditions (Dragun, 1988). Unfortunately, however, soil
biodegradation rates elude quantitative description. Dragun (1988) discusses some of
the reasons that estimation techniques have not been devised, including the many
soil and environmental factors that influence biodegradation rates and the widely
varying protocols and lack of studies to relate protocols used in biodegradation
studies. Some general guidelines are given, such as: (1) increasing the number of
chlorine atoms within the molecule decreases the biodegradation rate; (2) water
soluble chemicals are usually degraded faster than less water soluble chemicals; (3)
unsaturated aliphatic organics have faster biodegradation rates than corresponding
saturated aliphatic organics; and (4) n-alkanes, n-alkylaromatics, and aromatic
hydrocarbons in the C5to C9 range are biodegradable, but in most environments
volatilization competes very effectively with biodegradation as a fate process.
Smith and Dragun (1984) suggested that laboratory-derived half-lives should
be taken as lower limits (the most rapid that could be expected), and not necessarily
the half-lives expected to be found under field conditions. In contrast, Barker et al.
(1987) demonstrated that laboratory-derived rates were slower than field- measured
degradation rates for m- and p-xylene (in water-saturated strata). The initial
concentration of a substrate can change the kinetics of degradation, as shown by
27
-------
Boethling and Alexander (1979) and by Mabey and Mill (1978). Trace levels of
organic compounds will degrade at much slower rates than at higher
concentrations. Also, it is likely that many laboratory studies involving VOCs have
unaccounted volatile losses or sorption "losses" that increase the apparent
degradation half-life (e.g. Anderson et al., 1991; discussed below).
Pavlostathis and Jaglal (1991) have argued that organic pollutants buried
within micropores of soil aggregates are, for the most part, inaccessible to
microorganisms. Many soil bacteria range in size from 0.5 to 0.8 [am, and more than
half the pore volume in a silt loam soil may be represented by pores of radii less
than 1 [jm. Therefore, the contaminant in solution generally constitutes the
readily bioavailable fraction, and resorption rates will greatly influence the
biodegradation rate of sorbed compounds. Evidence for the bioavailability of
chemicals in solution has been demonstrated by Steinberg et al. (1987). They found
that EDB residues in old tobacco field soil had resisted degradation even though
fresh 14C EDB additions degraded relatively rapidly.
One recent study attempted a mass balance of VOCs incubated in soil in the
laboratory (Anderson et al., 1991). The study followed the disappearance rate of 15
volatile and semivolatile organic compounds in two soils using experimental
procedures to distinguish between biodegradation and abiotic losses (including
volatilization). Chemicals included in the study and reported half-lives are shown
in Table 2. All samples were incubated in the dark at 20 °C. Losses were
attributed to "abiotic processes" because differences in disappearance of organic
compounds between sterile (autoclave 1 h on 3 consecutive days) and nonsterile
samples were not significant. Sterilization was checked by measuring CO2 efflux
for the experimental period of 7 days and by incubation of soil extracts on nutrient
agar plates.
In this study, dissipation of 14C toluene was traced in one of the soils
(Anderson et al., 1991). After 7 days, 20% of the recoverable radioactivity was in the
charcoal traps (volatile losses), 4% was recovered in soil methanol extracts (cold),
and 65% was associated with the soil organic matter (extracted twice with NaOH),
leaving at least 10% unaccounted for. The authors stated that short-term spike and
recovery analyses of individual compounds yielded consistent recoveries, which
were used as correction factors. Mass balances in sterile soils were not obtained.
The authors suggested that problems with storage and holding times or possibly
nonreversible sorption contributed to losses. Spike levels were a nominal 100 mg
for each chemical/kg soil dry weight. To minimize volatilization, however,
chemicals were not mixed when added to the soil. This procedure may have caused
28
-------
TABLE 2 LABORATORY DISSIPATION HALF-LIVES OF SOME ORGANIC
COMPOUNDS*
Compound Half-life (days)
benzene
ethylene dibromide
toluene
cis- 1 ,4-dichloro-2-butene
chlorobenzene
pxylene
1 ,2,3 -trichloropropane
1,2-dichlorobenzene
chloroform
methyl ethyl ketone
carbon tetrachloride
tetrahydrofuran
nitrobenzene
2-chloronaphthalene
hexachlorobenzene
<2.0
<2.0
<2.0
2.0
2.1
2.2
2.7
4.0
4.1
4.9
5.0
5.7
9.1
11.3
11.3
Data from Anderson et al., 1991
locally toxic levels of some compounds in the soils and may have limited the
opportunity for the compounds to contact soil microbes. Organic partitioning
appeared to be the major sink for toluene (soil had 1.5% organic carbon), and this
toluene was not recovered by soil extraction with cold methanol.
SUMMARY OF DEGRADATION
Degradation of naturally occurring aromatics (e.g., BTEX compounds) is
likely to be caused by microorganisms in aerobic soils, but abiotic degradation is
also plausible (Dragun, 1988). The biodegradation rate of such compounds has been
shown to be limited by oxygen diffusion (Barker et al., 1987). Degradation of
29
-------
halogenated aromatics is similar to that of the unhalogenated aromatics in that it is
predominantly aerobic biodegradation. The degradation rate of halogenated
aromatics, however, is generally slower than that of the unhalogenated species
(Dragun, 1988).
Halogenated aliphatic compounds (e.g., chloroform, TCE) degrade by
reductive dehalogenation, that is, the replacement of the halogen by hydrogen
under anaerobic conditions. Biomediated dehalogenation occurs as a consequence
of microbial degradation of other organic carbon sources, or co-metabolism (Cline
and Viste, 1985; Wilson and Wilson, 1985). Abiotic dehalogenation has also been
reported, but estimates of abiotic half-lives in water vary widely (e.galess than 2
years Billing et al., 1975] and more than 700 years [Mabey and Mill, 1978] for
dichloromethane).
The kinetics of degradation are affected by many factors, including the
substrate concentration (Boethling and Alexander, 1979; Mabey and Mill, 1978).
Laboratory estimates of degradation rates cannot be applied directly to field
situations, and comparisons among laboratory studies are subject to artifacts of
differences among procedures. Sterile control conditions are difficult to verify and
volatile losses during experimental procedures are difficult to avoid (Anderson et
al, 1991).
30
-------
SECTION 4
MOVEMENT OF VOCS IN THE VADOSE ZONE
VOCs may move through soil by diffusion of the vapor or aqueous phases, or
by advection or convection1 of the vapor, aqueous, or NAPL phases. This section
first provides general discussions of studies on vapor diffusion, vapor advection,
aqueous convection, and volatilization. Field studies are then reviewed separated
into (1) studies that had a priori data concerning the time and amount of soil
contamination and (2) studies where the objective of the field investigation was to
describe the concentration or extent of soil contamination.
VAPOR DIFFUSION
In general, vapor diffusion will dominate the soil movement of compounds
with high vapor pressures (Taylor and Ashcroft, 1972; Kreamer et al., 1988). The
rates of vapor diffusion in soil are obviously slower than in free air. "Effective"
soil diffusion coefficients are influenced by the tortuosity of the channels, which is
estimated by such parameters as overall porosity and the volumetric air and water
saturation levels. The Millington-Quirk tortuosity formula (Millington and Quirk,
1961)2is frequently used to estimate soil vapor diffusion coefficients (e.g., Jury et
al., 1983 Sleep and Sykes, 1989). A summary of formulas for estimating soil
diffusion coefficients and values calculated for many VOCs using the Millington-
Quirk equation is given by Roy and Griffin (1990). In general, effective diffusion
coefficients for VOCs are estimated at 0.1 to 0.4 mVday, depending on porosity
variables, temperature, and compound.
'The terms "advection" and "convection" refer to movement due to bulk flow of a phase. Here, the
term "advection" is used for bulk flow of the gas and NAPL phases and "convection" for bulk flow of the
aqueous phase.
2Millington and Quirk (1961) reported the empirically derived formula for effective gas diffusion
coefficients, Dge, in unsaturated soil as:
D,, = (a1(V3«t.-2)Dg*
where a is the gas-filled porosity, <|> is the total soil porosity, and Dg* is the diffusion coefficient in air.
The factors that compose the coefficient modifying Dg* are often termed the "tortuosity" of the medium.
It should be noted that the tortuosity varies with soil moisture content.
31
-------
Use of the term "diffusion coefficient" varies among groups of researchers.
Studies may define a lumped effective diffusion coefficient or unapparent
diffusion coefficient to refer to the observation of a combined diffusion/retardation
coefficient.
VAPOR ADVECTION
At the soil surface, gaseous advection due to changes in barometric pressure,
temperature gradients, rainfall or irrigation, or wind are generally said to
penetrates to depths of many centimeters. For example, a 2-mbar change in
barometric pressure is estimated to cause air replacement to a depth of 8 cm in 12 h
(Taylor and Ashcroft, 1972, p. 367). At the soil surface, temperature gradients will
generally move vapors downward during the day and upward during the night,
moving from warm to cold areas (Hillel, 1971, pg. 118).
The effect of barometric fluctuations below 1 m is generally small (Kreamer
et al., 1988). Massmann and Farrier (1992), however examined situations in which
the barometric fluctuations might be significant. They argued that some storm
systems can produce barometric pressure changes of 20 to 30 mbar during a 24-h
period and that these storm circumstances can occur several times a year or more,
depending on the geographical location. Model calculations showed that "fresh air
may migrate several meters into a highly permeable subsurface during such large
barometric pressure cycles and the depth of penetration increases as the thickness
and permeability of the vadose zone increase. Massmann and Farrier (1992) thus
suggested that the concentration of gaseous VOCs may be lower when barometric
pressures are high and that soil gas measurements will show the largest fluctuations
during times of rapidly rising or falling barometric pressures. Their analysis,
however, omitted the effects of soil water on gas permeability and on diffusion
coefficients.
Falta et al. (1989) have suggested that gas phase advection may dominate the
transport of VOCs that originate from a NAPL in soils of high permeability. As
organic liquids that have high vapor pressures and low molecular weights
evaporate, the density of the gas in contact with the liquid changes with respect to
the ambient soil gas. This density contrast results in an advective gas flow.
Organic hydrocarbons and solvents have vapor densities that are greater than air,
so the resulting density-driven flows will be downward. Falta et al. (1989) cited the
contamination of ground water beneath waste management facilities at Idaho
National Engineering Laboratory (INEL) as a likely example of this phenomenon.
At INEL, the water table is 177 m deep and yet carbon tetrachloride, present in the
32
-------
waste, has been found in the ground water. The magnitude of density-driven flows
is a function of the saturated vapor pressure of the organic liquid, the gas-phase
permeability, and the gas-phase retardation coefficient. Contaminants that are
likely to be affected by density-driven flow include TCE, chloroform, 1,1,1-
trichloroethane, methylene chloride, 1,2-dichloroethene, 1,2-dichloroethane, 1,1-
dichloroethane, carbon tetrachloride, Freon 113, and possibly benzene.
Contaminants that are not likely to be affected by density-driven flow include
toluene, ethylbenzene, xylene, chlorobenzene, naphthalene, and phenols (Falta et al.,
1989).
AQUEOUS CONVECTION
Aqueous convection of VOCs refers to the movement of the VOCs dissolved
in soil solution. Gravity flow predominates, but evapotranspiration moves water
and dissolved species upward in the soil surface. Lappala and Thompson (1983)
suggested that ground-water convection can also affect VOC movement. They
postulated that the frequency and magnitude of ground-water level fluctuations
may provide the driving force for moving ground-water contaminants into the
vapor phase. If the capillary fringe is lowered into contaminated ground water, the
previously clean capillary water becomes increasingly mixed with contaminated
water during recurring fluctuations. VOCs can then volatilize into air-filled pores,
thus establishing a vertical concentration gradient in the soil gas phase.
VOLATILIZATION-GASEOUS DIFFUSION AND AQUEOUS CONVECTION
COMBINED
Volatilization refers to the gaseous loss of chemicals to the atmosphere from
the soil surface. The potential volatility of a chemical is related to its inherent
vapor pressure. Actual volatilization from soil depends on interphase transfers,
movement to the soil surface, and vaporization into the atmosphere.
The rate at which a chemical moves away from the surface is controlled
primarily by diffusion. There is relatively little air movement close to the soil
surface consequently, a vaporized substance is transported from the soil surface
through this stagnant air boundary layer only by molecular diffusion. The rate of
movement away from the surface will be proportional to the diffusion coefficient
and the vapor density of the chemical at the evaporating surface. Factors such as
wind velocity and surface cover (plants) alter volatilization through their effect on
the thickness of the stagnant air layer.
33
-------
Jury et al. (1990) simulated volatilization of chemicals that resided in
subsurface soils. They amended the input data for some of the chemicals with
results reported in Jury et al. (1992). They calculated relative volatile losses for 35
organic chemicals placed in a 30-cm-thick layer 100 cm below a soil surface. Two
uniform soil conditions were simulated in soils of sandy or clayey texture and (1)
with no water evaporation or (2) at a low, steady 0.1 cm/day water evaporation rate.
Results showed that when no water evaporation occurred, four compounds would
lose more than 50% of the initial concentration by surface volatilization in the
sandy soil over the first year. These four compounds were dichlorodifluoro-
methane, chloromethane, 1,1,1-trichloroethane, and bromoethane. Only
dichlorodifluoromethane lost more than 50% by volatilization under analogous
conditions in the clay soil. When water evaporation was simulated, seven chemicals
underwent cumulative volatilization losses of 50% or more in 1 year in the sandy
soil (including the four compounds listed above that exhibited >50% volatilization
without water evaporation and TCE, ethylene dibromide, and dichloromethane).
Again, only dichlorodifluoromethane volatilized 50% or more in the clay soil.
Although there were no field data to validate these model results, the data imply
that appreciable quantities of the most volatile compounds will escape readily from
buried sources. Volatilization from the soil surface can be an important
mechanism for the removal of the above-mentioned VOCs from the vadose zone.
FIELD STUDIES OF VOC FATE AND MOVEMENT
Field studies attempt to define rates of dissipation and movement of VOCs
in soil. These studies generally measure the distribution pattern of VOCs and
relate the results to prominent factors postulated to affect the dissipation and
movement of soil VOCs The studies reviewed here have been termed either "field
experiments" in which the researchers were able to control or had some a priori
knowledge of the contamination event, or "field investigations" in which case the
researchers were studying contamination without a priori knowledge of the
characteristics of the contamination event. In general, the studies have assumed
that VOCs move by vapor diffusion, tempered by the ability of the soil to transmit
gas.
Field Experiments
Weeks et al. (1982) measured gaseous concentrations of two fluorocarbons
(released to the atmosphere during the 40 years prior to this study) as deep as 44 m
in unconsolidated sedimentary deposits in the Southern High Plains of Texas. They
estimated lumped gaseous-diffusion coefficients for the combined effects of
34
-------
tortuosity, sorption, and solubility. Assuming that gaseous diffusion was the
primary mechanism for transport of these highly refractory compounds, the
lumped effective diffusion coefficients of 0.04 mVday and 0.09 mVday for
fluorocarbons F-ll and F-12, respectively, occurred in the sediments as opposed to
theoretical values of 0.78 mVday and 0.86 mVday in free air. Convective transport
resulting from movement of air to fill the voids caused by a declining water table
(estimated to have declined at 0.3 to 0.6 m/year since the early 1950s) was considered,
but the contribution of convective transport was deemed small enough to be
ignored.
In a similar study, Kreamer et al. (1988) used a fluorocarbon tracer,
bromochlorodifluoromethane, to observe gaseous diffusion in a shallow (35 m)
deposit of aeolian sand near Barnwell, South Carolina. The objective was to
quantify the in situ gaseous tortuosity at the existing moisture content (a lumped
gaseous-diffusion coefficient as in Weeks et al., 1982). Kreamer et al. (1988)
performed independent measurements of the free-air diffusion coefficient of the
tracer, air-water, and soil-water tracer distribution coefficients and of the soil
porosity and moisture content in appropriate subsamples. A network of
piezometers distributed radially and vertically from the buried tracer source was
sampled and analyzed continuously for 3 days before the fluorocarbon source was
introduced and for 7 days following the placement of the source. Kreamer et al.
(1988) computed a "sorption-affected" porosity value with which to compare the
field data because the field-measured concentrations are modified by soil-water and
air-water partitioning. Variability in the data (attributed to analytical error) and
uncertainties in the porosity and volumetric moisture content of the sand (only one
undisturbed core was obtained) greatly weakened the study. The tortuosity of the
geologic unit was determined to be about 0.4 (dimensionless) and the sorption-
affected porosity was reported to be 0.22 (also dimensionless).
A study by Poulsen and Kueper (1992) iooked at the advective movement of
a NAPL. The authors released tetrachloroethene to cleared areas 5 cm below the
soil surface. A nonvolatile hydrophobic dye was mixed with the tetrachloroethene.
Two 6-L releases were followed, one consisting of a slow drip (100 rein) over
approximately 1-cm2 surface area, the other consisting of a rapid (90 s) ponded
episode over approximately 1,000-cm2 surface area. Excavation of each area began a
day after the release. The distribution pattern of the dye was observed and a 2-mL
piston subsampler was used to collect soil for tetrachloroethene analysis. The
movement observed in the stratified sandy subsurface indicated that capillary
forces dominated. Tetrachloroethene was distributed in distinct stringers occupying
sand laminations separated by a few to several centimeters and characterized by
35
-------
subtle variations in texture, color, and grain size. The bedding in the upper 1.86 m
dipped to the northeast and the slow-drip release closely followed the bedding. The
ponded release produced a greater concentration of tetrachloroethene in the upper
portion overriding the bedding in the upper 0.5m. The depths of migration were
approximately 3.2 m for the slow release and 2.0 m for the rapid release. Average
residual concentrations were 0.49% and 1.26% for the slow and rapid releases,
respectively.
Field Investigations
The diffusion of TCE from contaminated ground water at a depth of
approximately 42 m was measured by analysis of data on shallow (<2 m deep) soil
gas (Marrin and Thompson, 1987). At a nearby unspecified distance, unlined solvent
evaporation ponds had been used to dispose of halocarbon solvents from 1951 to
1977. In 1984, the gaseous TCE plume spanned 3 orders of magnitude (<0.001 to 2
^g/L) on a 0.5-km2area of adjacent property. Soil gas TCE values within a 10-m
radius of each of five ground-water monitoring wells were shown to correlate with
ground-water TCE concentrations. A limited number of vertical soil gas profiles
showed that TCE concentrations generally increased with depth. Caliche zones did
not affect TCE gas concentration gradients, but a sharp decrease in the
concentration gradient occurred across clay lenses. Nearly saturated strata showed
anomalously low TCE concentrations. Overall, the TCE diffused approximately 40
m in this arid environment in much less than 30 years, assuming that transport to
ground water from the pond and subsequent lateral transport by ground-water flow
would have required more time than the vertical gaseous diffusion.
Kuhlmeier and Sunderland (1985) reported on the distribution of petroleum
hydrocarbons from leaking buried storage tanks at a site near Livermore,
California. The leaks occurred over a period of 6 months and permeated
approximately 24.4 m of unsaturated lacustrine and fluvial sediments before
reaching ground water. Soil borings were sampled from a split-spoon sampler at 1.5-
m intervals; stainless steel liners were covered with aluminum foil, sealed with duct
tape, and placed on dry ice after sampling and during transport to a laboratory for
subsequent analysis by SW-846 Method 8020 (volatile aromatics). Model predictions
of the movement of benzene, toluene, and xylene (BTX) were compared with the
laboratory data. Field data showed a marked increase of BTX in clay zones as
compared to sandy units (e.g., 3,000 to 4,800 mg/kg BTX were found in a clay layer,
overlying 1,138 mg/kg in the sandy unit). This attenuation by clay indicates an
increased adsorption coefficient in the finer textured deposits. It was also reported
36
-------
that organic carbon had no effect on sorption (total organic carbon varied between
0 [sic.] and 3%). Unexpectedly, little or no lateral dispersion of BTX was observed.
This is particularly surprising in light of theory and numerous observations that
horizontal flow occurs along the interface between coarse and fine lenses.
Kuhlmeier and Sunderland (1985) suggested that the volume of gasoline released
was large enough to create advective NAPL flow gradients.
Johnson and Perrott (1991) studied the vapor transport of gasoline in a soil
that had a high water content. Soil vapor concentrations of butane, pentane,
hexane, benzene, toluene, oxygen, methane, and carbon dioxide were measured
periodically for a year at fixed sampling ports near some underground storage
tanks in Portland, Oregon. Data indicated relatively constant gasoline-component
concentrations at sampling points near the center of the vapor plume. Lower
gasoline concentrations that increased during the year were detected a few meters
away. The different components of gasoline vary in their physiochemical
properties such that calculated retardation coefficients vary widely however,
isoconcentration contours showed that the pattern and extent of contamination
were very similar for all components. Environmental factors of barometric-
pressure fluctuations and water-level fluctuations did not show significant effects
on vapor concentrations. Vapor concentrations did appear to be directly affected
by soil temperature. The soil temperature dropped approximately 10 °C during the
winter and plots of winter vapor concentrations with time showed a decrease; the
curve shape was similar to that of the temperature curve. In contrast to vapor
movement around tanks that are placed in engineered backfills (e.g., pea gravel),
vapor movement in high-water content soils is very slow. Model predictions
estimated that less than 1% of the source concentration of a nonsorbing compound
(unit retardation factor) would travel 7 m after 8 years. Methane, resulting from
anaerobic degradation, was a good indicator of the contaminated zone. However,
the slow diffusion in this soil and general persistence of gasoline components led
the authors to conclude that once the vapors entered the soil, future leak detection
using vapor sampling may not be possible.
Smith et al. (1990) presented data on soil gas TCE concentrations, TCE-water
vapor sorption isotherms, and concentrations of TCE sorbed by soil at a site above
a contaminated aquifer at Picatinny Arsenal in New Jersey. TCE-containing
wastewater had been discharged into lagoons and into a nearby unlined, overflow
dry well from 1960 to 1981. Soil gas data closely paralleled changes in the
concentration of TCE in the shallow ground water. Temporal effects in vertical
gas concentration profiles were influenced by the ground-water temperature.
37
-------
Increased ground-water temperatures in July and October (as opposed to the
temperatures in December and February) generally resulted in increased
concentrations of TCE in the soil gas.
SUMMARY OF VOC MOVEMENT
Movement of VOCs in soil results from diffusion and advection. Diffusion is
driven by concentration gradients. Advection can be driven by pressure, density,
gravity, or thermal gradients. Gaseous molecular diffusion coefficients exceed
those of liquid coefficients by 4 to 5 orders of magnitude and therefore, gaseous
diffusion will dominate over liquid diffusion (Sleep and Sykes, 1989). Density-
driven NAPL advection is estimated to dominate VOC movement in highly
permeable soils when the NAPL source is present in excess of the soil residual
saturation level and the NAPL relative vapor density is greater than that of air
(Falta et al., 1989). When present, large-scale heterogeneities-such as bedding,
textural discontinuities, structural changes, anthropogenic materials and other
debris, channels, cracks, and fissures-can create distinct flow paths for liquid and
gaseous movement (Falta et al, 1989; Kreamer et al., 1988; Poulsen and Kueper, 1992).
Gaseous diffusion measured in situ was assumed to have prompted the
movement of two fluorocarbons in deep deposits in Texas (Weeks et al., 1982). The
fluorocarbons were found to have moved 44 m in approximately 40 years.
Movement of TCE upward through a deep deposit in Arizona, presumed to be
diffusing from contaminated ground water at approximately 42 m, was observed at
a depth of 2 m (Marrin and Thompson, 1987). The diffusion time was shorter than
that in Texas, less than 30 years, in response to a greater concentration gradient. In
Oregon, which has a more humid environment, and in a soil said to be of high
moisture content, gaseous diffusion of gasoline components was estimated at
approximately 7 m lateral in 8 years (Johnson and Perrott, 1991). The appearance of
carbon tetrachloride in ground water at 177 m suggests a mechanism other than
diffusion moving the VOC at a site in Idaho (Falta et al., 1989).
Calculations based on physiochemical properties of soils and NAPLs and on
environmental parameters have demonstrated the possible importance of gas
advection in VOC movement. First, large barometric pressure fluctuations can
cause air to migrate several meters in a highly permeable soil. This movement may
be reflected in soil gas measurements read before and after large storms pass
(Massmann and Farrier, 1992). Second, gas advection can be driven by density
gradients By examining the saturated vapor pressure and sorption coefficients
(Koc) of many VOCs as well as their soil permeability characteristics, Falta et al.
38
-------
(1989) showed that density-driven gaseous advection could dominate flow of many
organic solvents, including TCE, chloroform, 1,1,1-trichloroethane, methylene
chloride, 1,2-dichloroethene, 1,2-dichloroethane, 1,1-dichloroethane, carbon
tetrachloride, Freon 113, and possibly benzene. Contaminants that are not likely to
be affected by density-driven flow include toluene, ethylbenzene, xylene,
chlorobenzene, naphthalene, and phenols (Falta et al., 1989).
Field studies have shown that NAPL advection is sensitive to small
variations in soil permeability and capillary characteristics (Poulsen and Kueper,
1992). The migration pattern of tetrachloroethene followed subtle changes in
bedding of a stratified sand deposit. Characteristics of source release also affected
the NAPL migration. Tetrachloroethene was observed to migrate deeper and
retain a smaller residual fraction when applied slowly over a small soil area as
opposed to a rapid application over a large soil area. Kuhlmeier and Sunderland
(1985) observed deep vertical movement (24.4 m) of BTX with almost no lateral
spread, even though textural discontinuities were present in the lacustrine and
fluvial deposits. They postulated that rapid leaks of underground storage tanks
caused the strongly vertical movement by NAPL advection.
39
-------
SECTION 5
MODELING THE MOVEMENT OF SOIL VOCS
Modeling VOC vapor transport in unsaturated soil is a recent phenomenon
for environmental scientists (primarily put forth by civil engineers, soil physicists,
and hydrologists). Early research on soil gases, such as radon, oxygen, carbon
dioxide, and pesticides, although basic to the recent surge of interest, did not
involve extensive modeling and did not address factors peculiar to VOCs such as
the nonaqueous-phase liquids characteristic of petroleum spills or vapor-density
gradients arising from dense organic solvents. Multiphase, multicomponent flow
models exist in petroleum engineering, however, and this knowledge has been
recognized and used to formulate environmental applications (Finder and Abriola,
1986). Much of the current environmental interest follows the advent of "soil
venting" or "vapor extraction" as a means of remediating vadose zone
contamination.
Table 3 compares some characteristics of 14 models that are entirely or partly
designed to simulate VOC movement in soil or porous media. The models selected
are representative of the major features and processes that researchers incorporate
in soil-VOC models. The list includes some relatively simple screening/
management models (Jury et al., 1983; Silka, 1988; Falta, et al., 1989; Shoemaker et al.,
1990), laboratory soil-column descriptions of vapor movement (Gierke et al., 1990;
Gierke et al., 1992; Brusseau, 1991), one field-scale description accompanied by field
data (Metcalfe and Farquhar, 1987), and many field-scale models that define theory
and evaluate model approaches to VOC movement in porous media (Abriola and
Finder, 1985; Corapcioglu and Baehr, 1987; Sleep and Sykes, 1989; Mendoza and
McAlary, 1990; Katyal et al., 1991; Massmann and Farrier, 1992). All of the models
embrace vapor diffusion and most of the models incorporate aqueous-vapor
partitioning.
The purpose of this discussion is to draw attention to the intent and extent
of these models, to indicate directions under development, and to outline some of
the knowledge gained from this form of analysis. The models are discussed in
three subsections screening/management models, laboratory soil column
simulations, and field-scale simulations.
40
-------
TABLE 3. COMPARATIVE FEATURES OF SOME VAPOR-TRANSPORT MODELS
Model Features
Dimensions
Aqueous-vapor distribution
Solid-aqueous distribution
Vapor-solid distribution
Degradation
Vapor diffusion
Vapor advection (P=pressure,
D=density)
Water flow (S=stcady state,
T=transient)
Immiscible phase
(S=sink/source term, F=flow)
Sorption nonequilibrium
Physical nonequilibrium
Multicomponent equation set
Application
(S=screening/management,
L=laboratory soil column,
F=field reseach)
Juryetal., Abriola& Corapcioglu& Metcalfe& Silka, 1988 Faltaetal,
1983 Pinder, Baehr, 1987 Farquhar, 1989
1985 1987
11 1 222
XXX XX
X X XX
X X
X XX X
P D
S T S
S,F S,F S
S F F F S S
Sleep &
Skyes, 1989
2
X
X
P,D
T
S
X
F
(continued)
-------
TABLE 3. (continued)
to
Model Features Gierke et Mendoza & Shoemaker Brusseau,
al, 1990 McAlary, etal, 1990 1991
1990
Dimensions 1 2 1,2 1
Aqueous-vapor distribution x x x x
Solid-aqueous distribution x x x x
Vapor-solid distribution x
Degradation x
Vapor diffusion x x x x
Vapor advection (P=pressure, P D P
D=density)
Water flow (S=steady state, S S
T=transient)
Immiscible phase
(S=sink/source term, F=flow)
Sorption nonequilibrium x
Physical nonequilibrium x x
Multicomponent equation set
Application L F S L
(S=screening/management,
L=laboratory soil column,
F=field research)
Katyal et
al, 1991
2
x
x
X
X
P,D
T
S,F
x
x
F
Gierke et Massmann
al, 1992 & Farrier,
1992
1 12
x
x
X
X X
P P
X
X
L F
Model can accomodate vapor sorptton u certain parameters are redefined.
-------
SCREENING/MANAGEMENT MODELS
Jury et al. (1983) described a model designed for comparing the soil behavior
of new and existing organic chemicals. Although the model was presented almost
10 years ago, the simplified approach with direct analytical solutions to equations
has bestowed the status of a "back-of-the-notebook" calculation for some of the
equations. The model assumes linear equilibrium distributions between vapor,
liquid, and adsorbed chemical, net first-order degradation rates, steady-state upward
or downward water flow, volatilization from the soil surface through a stagnant air
boundary layer, and homogeneous soil properties, all defined by user input
variables. Experimental evidence demonstrated that the model correctly predicted
the relative volatilization loss of five pesticides that were exposed to identical
conditions in the laboratory (Jury et al., 1984). The simulation of surface
volatilization of buried sources of VOCs was reported by Jury et al. (1990) and
discussed earlier in this report.
Silka (1988) developed a simple, two-dimensional vapor transport model as a
tool for interpreting soil-gas surveys. The model extends the total concentration
equations of Jury et al. (1983) to two dimensions, but omits the equations for water
movement and neglects biodegradation. As with many models, field
experimentation lags behind model development and no test of this model was
presented. Model runs demonstrated the importance of soil water to the design and
results of soil-gas surveys.
Falta et al. (1989) explored the simulated effect of gas-density gradients in
unsaturated porous media. Using thermodynamic properties of the compounds and
estimates of retardation by sorption, they screened 14 common ground-water
contaminants for the relative impact of density-driven flow as a function of media
permeability. Results showed the contaminants that were most likely to be
influenced by density-driven flow (reviewed previously in this report) and that
density-driven flow will be significant only if the permeability of the medium is at
least 10 um2.
Shoemaker et al. (1990) presented a screening model that encompasses vapor
sorption whenever soil moisture is dry enough for this process to occur. They
developed one- and two-dimensional solutions to the analytical model of Jury et al.
(1983). An "effective" or "two-phase" sorption coefficient is defined as the sum of
the solid-aqueous sorption coefficient and the vapor-solid coefficient. The vapor-
solid sorption is dependent on the soil water content, but the data currently are
limited to a few laboratory studies for a few soils and soil-like materials. The
43
-------
authors concluded that effects of vapor-phase sorption are greatest in soils of high
surface area that has an accompanying low water content. For soils that have high
water contents or low specific surface area, liquid-phase sorption models would be
adequate (quantification of these concepts was not provided). Shoemaker et al.
(1990) suggest that sorption be measured under field conditions to determine the
influence of vapor-phase sorption in situ.
LABORATORY SOIL COLUMN SIMULATIONS
In the screening models, all interphase transfer processes are facilitated by
assuming equilibrium with respect to interphase partitioning. Although these
assumptions simplify the transport analysis, they are rarely valid, and the
consequences of these assumptions are likely to be magnified in descriptions of
long-term or large-scale (field) transport phenomena. Asymmetrical breakthrough
curves representing laboratory soil column results provide evidence that the
supposition of equilibrium sorption is not generally valid (Brusseau and Rao, 1989).
Any of the interphase transfers that VOCs undergo (e.g., air-water, NAPL-water,
water-solid) may be rate-limited or exist out of equilibrium in soil. Laboratory soil
columns can be employed to estimate the effectiveness of techniques for describing
transport of VOCs in porous media.
Harmon et al. (1989) reviewed the modeling approaches for dealing with
nonequilibrium transport phenomena. They provided a concise summary of
chemical and physical nonequilibrium model approaches. Chemical nonequilibrium
models are generally two-site models. Sorption is described as rapidly reversible for
some fraction of sorption sites (the local equilibrium assumption) and characterized
by slow resorption kinetics for the remaining fraction of sorption sites. The
fraction of sites in each category and kinetic rate constants for forward and reverse
sorption are required input parameters. Physical nonequilibrium models are
generally based on a two-zone description of soil water (mobile/immobile) with a
mass transfer coefficient between the two zones. Diffusion-based physical models
assume that rapid chemical sorption occurs in the mobile water and that kinetically
limited sorption occurs in immobile regions. The transfer of water or contaminant
between these regions is modeled either by Pick's second law of diffusion (a second-
order differential equation taking the geometry of the immobile region into
account) or, more simply, by assuming a first-order rate transfer equation.
Gierke et al. (1990) developed a model to study the relative contributions of
gas and water advection, gas and water dispersion, mass transfer resistance,
diffusion in immobile water, sorption, and volatilization on the spreading and
44
-------
retardation of VOC breakthrough in soil columns. They found the rates of mass
transfer across the air-water and the mobile-immobile water interfaces to be fast
(mass transfer rates across the air-water interface and mobile-immobile interface
were 4.8 and 660 times the transport rate by advection in mobile water,
respectively) Both liquid dispersion and diffusion in immobile water were
important for describing TCE transport. Vapor dispersion was lower than
predicted by a factor of 10, and vapor diffusion was not an important transport
mechanism for TCE when the average pore water velocities were greater than
about 0.07 cm s'in sand, or greater than about 0.02 cm s'in aggregated porous "soil
material" (a fired clay used as an industrial insulator). Henry's Law constant for
TCE was almost twice as large as values in the literature and the authors surmise
that sorption onto packing and column materials produced this artifact.
Gierke et al. (1992) redesigned their experimental conditions to distinguish
between the impacts of different physical nonequilibrium mechanisms on organic
vapor transport. Gas flow rates were varied to simulate soil venting, and both
granular and aggregated soils were fabricated (as in previous work) under moist
and dry conditions with toluene as the test compound. The authors reevaluated
some of their conclusions from Gierke et al. (1990) and concluded that fingered or
preferential flow could explain the large liquid dispersion coefficients observed in
that study. Under dry conditions, toluene vapor transport in the granular and
aggregated "soil material" was affected only by gas advection, gas diffusion in the
mobile gas region, and vapor sorption. Gas-water mass-transfer was never
important. Nonequilibrium effects were observed in columns that consisted of
moist, uniformly sized fired-clay aggregate the effects were attributed to
intraaggregate diffusion in immobile water. The toluene breakthrough for moist
aggregated material took eight times longer than for moist sand, an effect
explained by the higher moisture content of the aggregated material and the
impact of intraaggregate diffusion.
Brusseau (1991) developed a model that accounts for both physical and
chemical nonequilibrium processes under forced gas advection and an immobile
liquid phase. The model used mass transfer coefficients for water-solid, air-water,
and mobile-immobile water. Brusseau tested the model using data from the
literature, including the data of Gierke et al. (1992). Values for all parameters,
independent of curve fitting, were obtained by a variety of estimation techniques.
Model predictions for each data set were shown to fit soil column breakthrough
curves better than when the local equilibrium concept was employed.
45
-------
FIELD-SCALE SIMULATIONS
The models discussed so far have simplified water flow, which, if
incorporated, is modeled as steady-state flow. In moist or fine-textured media, VOC
dissolution and transport in water are probably significant and must be included if
the model is intended to estimate actual field concentrations. Simplified
representations, such as those discussed above, have limited, specific conditions of
application and thus reduce the model complexity. Field-scale research models may
also possess limited application, e.g., Metcalfe and Farquhar (1987) and Massmann
and Farrier (1992) both depict vapor transport over short periods of time. The more
comprehensive field models that attempt to portray long-term transport in a
heterogeneous porous medium have exceeded our ability to obtain experimental
data. Both input and output data are extensive in these models. The complex
models are not designed to be used as general field tools, but to further our
understanding of processes and computing capabilities. Such models are often
verified numerically that is, the validity of the numeric computer code solution is
tested against a simplified version of the model that can be solved analytically.
Lack of data, however, precludes determining the validity of the model as a tool
for describing VOC transport in soil.
Abriola and Finder (1985) began developing a model to depict the multiphase
(solute, gas, and NAPL) migration of a petroleum spill in porous media. They
presented equations governed by mass conservation principles and volume
averaging theory to describe a contaminant composed of two distinct components,
one of which may be volatile and slightly water soluble and the other which is both
nonvolatile and insoluble in water. Finder and Abriola (1986) suggested that three
assumptions incorporated into their model are undergoing scrutiny (1) the use of
Darcy's Law to represent the convective flux of a fluid phase (neglects hysteresis
effects and the observation of preferential migration pathways), (2) the supposition
of immobile air (no vapor advection), and (3) the use of equilibrium partitioning.
Corapcioglu and Baehr (1987) devised a model for describing surface
petroleum spills that could be extended to include any number of reactive
constituents (such as BTX). The model utilizes mass conservation of oxygen as the
limiting factor (upper boundary) in the microbial degradation of petroleum
hydrocarbons. The general model can be divided into two subproblems to describe
an oil spill: (1) oil plume establishment in the unsaturated zone, and (2) solute and
vapor transport subsequent to immiscible plume establishment.
46
-------
To describe the pressure variation created during gas venting by a vacuum
withdrawal system, gaseous advection must be modeled in addition to gaseous
diffusion. Metcalfe and Farquhar (1987) use the equations for hydrodynamic
dispersion in porous media, as outlined in Bear (1979), to depict gaseous
advection/dispersion. Metcalfe and Farquhar (1987), rather than use mass
conservation equations, based their equations on molar quantities as a way to
handle the obvious density differences between air and VOCs (in this instance,
methane). They produced a conservative, two-dimensional model for vapor
movement in the unsaturated zone. The model was tested by simulating methane
migration from a landfill in Ontario over summer and winter boundary conditions
(assuming impermeable frozen soil during the winter). Using measured gas
concentrations as initial conditions, subsequent data (generally 1 month later) were
simulated by the model. The actual field data were spatially far more variable
than the model predictions, but model results were generally within 50% of
observed values.
Sleep and Sykes (1989) borrowed the concept of mass transfer approximation
from chemical engineering to represent nonequilibrium interphase transfers in a
field-scale model. Empirically determined mass-transfer coefficients are employed
to express the driving force between equilibrium and actual concentrations. Mass
transfer coefficients for air-water, NAPL-air, and NAPL-water are incorporated in
their model, although sorption is neglected. Model results simulating the
movement of a buried TCE source provided qualitative evidence for the
importance of the mass transfer coefficients. The model could not be verified
because comparable field data do not exist.
Mendoza and Me Alary (1990) and Mendoza and Frind (1990) employed a
radial coordinate system, density-driven flow, and equilibrium sorption estimates in
a model designed to simulate VOC gas transport over a few weeks. They
investigated the effect of surface boundary conditions and found that an
impermeable cover over the ground surface will increase the lateral migration of
the vapor plume. A permeable ground surface that allows natural venting of gases
will reduce the lateral extent of the vapor plume. Diffusion is the dominant vapor
transport mechanism for TCE in a deposit with a permeability of medium sand.
Density-driven vapor advection becomes important in coarse sands or gravels for
compounds with high vapor pressures and high molecular weights.
An EPA-sponsored model to simulate the two-dimensional flow and
transport of three fluid phases (water, NAPL, and gas) has been generated by
Katyal et al. (1991). The model allows the user to analyze flow only or coupled flow
47
-------
and contaminant transport. It can be used to analyze the two-phase flow of water
and NAPL, or three-phase flow of water, NAPL, and gas at variable pressure.
Transport of as many as five components can be described, assuming either local
equilibrium mass transfer or first-order, kinetically controlled mass transfer.
Required inputs are extensive, encompassing the air-water capillary retention
function, NAPL surface tension and interracial tension with water, NAPL viscosity,
maximum residual NAPL saturation, soil hydraulic conductivity, component
densities, mass transfer coefficients, and boundary condition data. Methods for
estimation of certain parameters are included. Resulting outputs are equally
extensive, including saturations, velocities, and concentrations for each phase at
every node at specified intervals. The computing is rendered more efficient by
incorporating time-lagged interphase mass-transfer rates and phase densities.
Massmann and Farrier (1992) have argued (as have others, including Brusseau
[1991] and Thorstenson and Pollock [1989]) that the validity of the classical single-
component advection-diffusion equation to describe vapor transport should be
bracketed by the material permeability and the total pressure gradient. They
evaluated the limits of the single-component advection-diffusion equation through
a set of fully coupled multicomponent equations. They concluded that the
advection-diffusion equation is adequate if media permeabilities are greater than
about 1010cm2. The equation significantly overestimates gas fluxes for low-
permeability materials, becoming egregious for permeabilities of 1012to 1013cm2.
Large atmospheric barometric pressure fluctuations can occur that create
significant subsurface gas pressure gradients and thus affect vapor movement.
These conditions may cause horizontal pressure gradients in heterogeneous soil,
depending on site geometry, material properties and the amplitude and period of
the barometric fluctuations. Massmann and Farrier (1992) did not address the
effects of gas density and soil water in their analysis.
SUMMARY OF MODELS
Mathematical modeling of soil-VOC movement has many objectives and
various strategies for meeting those objectives. Models have been developed to
screen for differences among compounds or to serve as management tools, such as
to guide the interpretation of soil gas surveys. Research on heterogeneous flow
domains, or transport-related nonequilibrium, is partly pursued by use of laboratory
soil columns and models to describe observations. Field-scale simulations to
describe flow and transport of contaminants created by spills of immiscible solvents
or petroleum, can expand our understanding of theory and our computing skills.
Research models are incorporating much more than the traditional homogeneous
48
-------
soil, equilibrium sorption, single contaminant, and isobaric conditions that
characterized earlier model efforts. Modeling research, however, runs far ahead of
the field data needed to calibrate and validate existing models. Accurate
quantitation of soil VOCs is crucial to model validation.
49
-------
SECTION 6
OBTAINING AND MAINTAINING VOC SAMPLES
Quantification and control of potential errors arising during sample
collection, handling, preparation, storage, and analysis are recognized as critical for
meeting data quality objectives of environmental samples (van Ee et al., 1990).
Specified procedures for obtaining soil VOC for analysis by SW-846 Methods
8240/8260 are minimal. For soil VOCs the need to standardize sampling procedures
has been recognized, but the methods for studying soil volatiles under controlled
conditions to verify sampling improvements have been troublesome. Issues of bulk
sample acquisition including recommendations for soil sampling devices and
subsampling techniques have been raised and reviewed by Lewis et al. (1991)
Volatile losses have been seen to contribute negative bias throughout each period
that the soil is exposed to air. Sampling and subsampling methods are driven by
the need to minimize exposed soil surface area, time of exposure, and soil
disaggregation to reduce negative sample bias (Lewis et al., 1991). Although the
natural or man-made variability that typify soils affects soil sampling for other
analytes as well, homogenization and observation techniques to manage variability
(e.g., mixing, compositing, and visual inspection of soil samples) are drastically
reduced in VOC sampling. Abbreviated sampling procedures exacerbate the
problem of securing representative samples for VOC analysis and certainly
contribute to the large variability reported for field soil-VOC concentrations
(Hewitt et al., 1992, Mitchell, et al., 1993).
This section first describes the general procedures used for VOC sampling.
Studies related to sampling methods are then discussed, separated into laboratory
and field investigations. Studies emphasizing the storage and preservation of soil-
VOC samples are presented in a separate subsection. Finally, a summary of
sampling and preservation results is provided.
CURRENT SAMPLING METHODS
Soil samples for VOC analysis are obtained by coring, augering, or scooping
devices. Coring is generally preferred because it disturbs the sample least. Lewis et
al. (1991) provide a detailed discussion of sampling devices for obtaining soils for
analysis of VOCs.
50
-------
The suggested container for the initial soil sample is a 125-mL wide-mouth
jar (SW-846 Method 8240), but the choice of sample containers is often determined
by site characteristics and the intended purpose of the sampling effort (Lewis et al.,
1991). Either 40-mL volatile organics analysis (VOA) vials are used, or wide-mouth
bottles (125-, 250-, or 500-mL) with Teflon-lined, foam-backed lids are used. Soil
typically fills the container, leaving as little headspace as practicable. Another
option is to collect 1- to 5-g soil samples in the field and place these into 40-mL
VOA vials. These can be capped in the field with a lid that can be connected
directly to a purge-and-trap sparger (Lewis et al., 1991). Alternatively, brass core
liners are capped and sealed in the field.
When transferring soil from a coring device to a wide-mouth bottle or vial,
the sampler's objectives are minimum disturbance to the sample and shortest
possible transfer time. Depending on the diameter of the soil core and the
diameter of the jar, soil can be extruded directly into the jar if the sample is fairly
cohesive, or subsampled with a subcorer. Plunger/barrel-style subcoring devices (10-
mL plastic syringes with the needle end cut off) are suggested for collecting and
extruding approximately 5 g of soil into a 40-mL VOA vial (Lewis et al., 1991)
Procedures that include soil compositing and mixing, either in the field or with
cold (4 °C) procedures in the laboratory, have not been specifically excluded,
although mixing procedures are known to create large losses of analyte.
The final subsampling and transfer steps are performed by personnel at an
analytical laboratory (excluding the 1- to 5-g samples placed in VOA vials in the
field). The standard procedure is to empty samples into an aluminum pan, briefly
"homogenize," and remove an aliquot into a sparging vessel for PT analysis. SW-846
Method 8240 requires a 1-g (wet weight) aliquot for analyte concentrations expected
in the 0.1 to 1.0 mg/kg range and a 5-g (wet weight) aliquot if concentrations less
than 0.1 mg/kg are expected. Recognizing the losses associated with the
subsampling step, SW-846 Method 5035 ("Determination of Volatile Organic
Compounds in Soils Using Equilibrium Headspace Analysis and Capillary Column
Gas Chromatography/Mass Spectrometry;" to be included in the Third Update to
SW-846) dictates the use of a soil sampler which delivers 5 g of soil to a 40-mL
VOA vial. The balance between sample representativeness (which would prescribe
ample homogenization) and preservation of volatile analytes is increasingly
disposed to sample preservation.
Described above are EPA procedures. The American Society for Testing
and Materials (ASTM) has independently voted acceptance of procedures for
sampling solid wastes; D 4547-91, "Standard Practice for Sampling Waste and Soils
51
-------
for Volatile Organics" (ASTM, 1991). The ASTM D 4547-91 procedures encompass
two methods of sampling loose granular materials and three methods to handle the
sample once it is collected. These procedures include (1) collecting a sample in a
metal ring and shipping the entire soil/ring sample to a laboratory, (2) subsampling
with a metal coring cylinder and placing the subsample in methanol, and (3)
subsampling as in (2) but placing the sample into a VOA vial that is capped with a
lid modified for direct connection to a PT sparger.
LABORATORY SAMPLING STUDIES
Opening of a vial or jar to take a subsample has been shown by many
researchers to cause large losses of VOCs. Amin and Narang (1985) showed that
chilled [sic.] clay that was spiked and mixed for 15 to 30 s on an ice-cold surface had
losses ranging from 7% for carbon tetrachloride to 36% for 1,1,2,2-tetrachloroethane.
Unchilled soil that was transferred from one sample tube to another showed losses
ranging from 14% to 53% (Amin and Narang, 1985). Maskarinec et al. (1988)
estimated an average 60% volatilization loss of VOCs during sample transfer steps;
this estimate was determined by comparing a pour, mix, and subsample procedure
with sample placed directly into vials that have lids which can be attached to a PT
sparger. Zarrabi et al. (1991) showed analyte losses of 20% to 80% from the
subsampling step, again using the modified-lid vials.
Siegrist and Jenssen (1990) have evaluated the effects of sample disturbance,
container headspace, and sample transfer steps on VOC measurements. Six VOCs
in aqueous solution were added to soil by saturated upward flow (15 pore volumes)
through a column of soil. The column was "desaturated under suction for less than
an hour. Subsequently, the column was sealed and stored overnight at 10 °C prior
to sampling. The sampling procedure consisted of concurrently inserting 10
stainless-steel sampling tubes, each 10 cm long with an aluminum-foil-covered top,
into the soil column through a sampling template. These tubes were removed
sequentially and placed into five container treatments. The procedure was
repeated, thus providing duplicates of the five container treatments. Undisturbed
samples were extruded directly into containers and disturbed samples were
generated by removing soil from the cores in 7 to 10 aliquots during the transfer
step. Disturbed samples were placed in jars or freezer bags that had little
headspace, but these treatments gave poor recoveries of VOCs Undisturbed
samples that had little container headspace gave better recoveries than those that
had large container headspace.
52
-------
Core extrusion into methanol yielded the highest VOC concentrations
(Siegrist and Jenssen, 1990). Improvement in VOC recovery that was attributed to
methanol preservation was 81% for TCE and approximately 10% for chlorobenzene.
VOC recoveries in the methanol treatment were 28 to 83% less than the predicted
concentrations based on calculated soil sorption estimates. The highly volatile
methylene chloride showed extremely high replicate variability and no significant
effect due to any treatment. The study did not use a PT preparation step in the
analysis it employed solvent extraction that was modified for the methanol
treatment.
Use of methanol as a vapor trap and as an extractant for alkylbenzenes
(ethylbenzene, toluene, and p-xylene) from oven-dry soil is described by Rhue et al.
(1988). This study compared the efficiency of methanol for trapping ethylbenzene
vapors with that of activated charcoal. Agreement between the two methods at
relative vapor pressures of 0.09 to 0.31 was quite good (methanol trapped 97% of the
amount trapped by activated charcoal). Rhue et al. (1988) then compared the
measurement of vapor-adsorbed alkylbenzenes by soils and clays using a
gravimetric method and methanol extracts. Gravimetric measurement has been
used by many researchers (e.geChiou and Shoup, 1985; Ong and Lion, 1991a), but
Rhue et al. (1988) preferred measuring vapor adsorption by methanol extraction of
the solids (allowing them to analyze either vapor or sorbed concentrations by UV-
Vis spectrophotometer or high pressure liquid chromatography). Methanol
extraction and gravimetric measurements of alkylbenzene vapor adsorption were in
good agreement (differences averaged less than 10%) for bentonite, kaolin, and a
sandy aquifer sample.
A vapor fortification method to spike dry soil with VOCs has been described
by Jenkins and Schumacher (1987) and has been further developed by Hewitt et al.
(1992) and Hewitt (1993). Reproducible soil-VOC contamination is achieved in dry
soil subsamples (1 to 2 g subsamples) by exposing soil to a vapor mixture of VOCs
in a closed desiccator. Relative standard deviations (RSD) of less than 9% were
observed for TCE, benzene, and toluene, and the RSD for trans-1,2-dichloroethene
was 14 to 23% (Hewitt et al., 1992). Maximum concentrations of soil VOCs are
achieved after 4 to 5 days of exposure (Hewitt, 1993). Dry spiked soil can then be
sealed in glass ampules (Hewitt, 1993) or capped in 40-mL vials with a modified lid
that is attached directly to the PT sparger (Hewitt et al., 1992). Soil concentrations
of the spiked, dry soil remain constant for at least 14 days. The ampulated soil
samples are suggested as blind performance evaluation materials and are also
useful in studies of sampling and analytical methods.
53
-------
FIELD SAMPLING STUDIES
Urban et al. (1989) recognized the potential for VOC losses during
transportation and storage from jars that may not seal well and the potential for
VOC losses during laboratory sample preparation. They compared data from
samples that were immersed in methanol in the field and analyzed by the medium-
level method (for samples containing contaminants at greater than 1 fig/g) with
data from samples that were analyzed by the standard, low-level soil method
specific in the EPA Contract Laboratory Program (CLP) Statement of Work (SOW,
7/87). Samples known to be contaminated with seven chlorinated solvents were
obtained with a split-spoon sampler. Material was added to fill either an empty 40-
mL vial or a preweighed 500-mL wide-mouth jar that contained 250 mL of
methanol and a mixture of three surrogate compounds. Results showed excellent
recoveries of surrogate compounds in the methanol-preserved samples. The
samples preserved in methanol always had greater concentrations of each VOC
than the corresponding 40-mL vial samples. For example, the results showed 2 to 50
times more TCE and 15 to 100 times more 1,1,-dichloroethene in the methanol-
immersed samples than in the corresponding 40-mL vial samples. A systematic bias
was introduced by the protocol of collecting the methanol sample before the
conventional sample in this study. The results, however, have been confirmed in
other field investigations (Hewitt, 1992; J. Smith, personal communication).
Conventional sample collection and analysis were compared with limited
disruption sampling and a field headspace method in a study by Hewitt (1992). For
the conventional sample, a 40-mL VOA vial was filled with soil, shipped and stored
at 4 °C, subsampled in the laboratory, and analyzed by PT/GC/MS within 14 days of
collection. The limited disruptive method consisted of taking a subsample with a
subcoring device in the field, placing it in either an empty vial that had a lid
modified for direct attachment to a sparger or a vial containing methanol, and
analyzing the sample by PT/GC/MS within 14 days of collection. (A hand-held
VOC Photo Vac probe was used to determine whether the sample concentration
was "low" or "high" and thus to be analyzed by the low-level or the high-level
procedure.) The field method involved placing a subsample in a vial containing 30
mL of water the air above the soil/water was analyzed by direct injection in a GC
within 2 days of collection. Results are shown in Figure 3. Log-log plots of
collocated samples showed 1 to 2 orders of magnitude more TCE measured by the
limited disruptive method as compared with conventional sample collection and
analysis (Figure 3a). The headspace analysis and limited disruptive method gave
very similar results (Figure 3b). Collocated headspace measurements were 1 to 3
orders of magnitude greater than measurements by the conventional procedures
54
-------
10*
CL
3
M'i'iTi I'l'irj i'I'm I'lTrj I'i'a
=- 1000/1 -
,i,i,iJxi , i.ul i ,iii.d i ,MJ,I i .1.1
10
10
3 e
fcl 'I'lTJ I 'ITFj I 'ITq I 'I-IT| I Tr
CONVENTIONAL PT/GC/MS
10"* 10"* 10"' 10° 10' 10*
LD PT/GC/MS
I0g
10'
10'
U
X
10"
M'l'ITj I 'I'lTJ I'l'ITJ I 'ITfl I'I'lJJ
.lil.iil i.i.ul I.I.I.L! i .I.I.L! i .i.i.h
10"* 10"* 10"
10° 101
CONVENTIONAL PT/GC/MS
Figure 3. Collocated soil samples analyzed by conventional PT/GC/MS, limited
disruption (LD) PT/GC/MS, and headspace (HS) GC. Soil concentrations in mg
TCE/Kg soil (after Hewitt, 1992).
55
-------
(Figure 3c). To determine whether degradation of TCE could be responsible for
the discrepancies, seven headspace samples were set aside, stored inverted (soil on
Teflon-lined septa) at room temperature, and monitored by repeated readings for 21
to 25 days. Little or no decrease in the TCE concentrations was observed. Loss of
analyte during the sample transfer step (shown by Jenkins and Schumacher, 1987;
Maskarinec et al., 1988; Siegrist and Jenssen, 1990) and throughout the storage time
in the vial prior to conventional analysis are implicated in this study.
Three sampling procedures for TCE-contaminated soil were investigated by
Slater et al. (1983). (1) Two subsamples (approximately 5 g each) were taken from
the ends of freshly collected soil cores and placed in three successive plastic freezer
bags. After cold storage, the soil was transferred to 5 mL of methanol. An aliquot
of the methanol was subsequently analyzed by PT/GC with a Hall detector. (2)
Brass liners, with four sampling ports drilled into the sides and covered with Teflon
tape, were capped with rubber, placed in freezer bags, and stored cold until analysis.
Four subsamples were removed in the laboratory, placed in methanol, and analyzed
as in Treatment 1. (3) Entire 15 x 2.5 cm soil cores were slid into glass jars, sealed,
placed in three successive plastic freezer bags, and stored cold until analysis. The
analysis consisted of a 20-h heated nitrogen purge of the entire jar and a charcoal
trap collection of VOCs, followed by carbon disulfide extraction of the charcoal,
then by gas chromatographic analysis of the extract. Treatments 1 and 2 were not
adequately sealed, and TCE was detected in the plastic bags prior to soil analyses.
Treatment 3 also had samples that were not properly sealed, although the
concentrations in the plastic bags were less than occurred for treatments 1 and 2
Treatment 3 showed TCE levels 1 to 2 orders of magnitude greater than
neighboring samples in the other two treatments (statistically significant at the 5%
level). Coefficients of variation ranging from 0 to 190% were found on replicate
subsamples by the first two treatments. (No coefficient of variation could be
reported where the entire 800-g sample was analyzed in Treatment 3.) Slater et al.
(1983) concluded that the differences in TCE concentrations resulted from natural
soil heterogeneity. Variability in the 5-g samples resulted in a statistical analysis
which concluded that eleven 5-g samples would have to be analyzed from a single
15 x 2.5-cm core section to be 60% confident that the estimated value for TCE was
within 100 mg/kg of the true value.
The effect of sample size on soil-VOC concentrations, noted by Slater et al.
(1983) above has been mentioned by other researchers as well. Bone (1988) suggested
that sample size influenced results of VOC sediment samples that were sent to two
different laboratories as a blind split. Acid and base-neutral extracts obtained by
using standard priority pollutant protocols were in good agreement, but the
56
-------
Eynon and Rushneck (1988) examined a curious phenomenon arising from
the comparison of compounds detected by the Toxicity Characteristic Leaching
Procedure (TCLP) and corresponding concentrations found by direct analyses.
Many compounds were detected by the TCLP test without being detected by direct
analysis, even though leaching procedures were not as analytically rigorous as
direct analysis procedures. In this comparison of 112 samples, volatile compounds at
or above the detection limit obtained by the TCLP test but below detection limits
by direct analysis (Method 1624, USEPA, 1982) in three or more samples included
methylene chloride (probably laboratory contamination), 2-propanone, toluene, 1,2-
dichloroethane, ethylbenzene, 2-butanone, 1,1,1-trichloroethane, and chloroform. It
was suggested that the relative sample size for the TCLP procedure is larger than
that for direct analysis and that the larger sample size could explain the greater
sensitivity of the TCLP.
Poulsen and Kueper (1992) contributed further to the determination of
appropriate sample size. They demonstrated the effects of mesostructure (stratified
sands) on the advection of a pure phase NAPL (tetrachlorethene). Within a few
days after release of tetrachloroethene onto a field soil, the compound was
distributed in distinct stringers occupying sand laminations that were separated by
a few to several centimeters. Estimates of the appropriate sample size to recover
pure stringers were on the order of cubic millimeters. Samples much larger than a
few milliliters in volume would be necessary to obtain an estimate of the mean
concentration in this soil. The authors used 50 g (or approximately 36 mL) samples
to estimate bulk properties.
Spatial Variability
An arbitrary increase in the VOC sample size beyond the current 1- to 5-g
size is likely insufficient to resolve VOC sampling problems. Sampling design must
address the long- and short-range spatial variability of VOCs in the field. Mitchell
et al. (1993) measured soil VOCs at a land treatment unit used for disposal of waste
oils and solvents. Samples (176) were collected from 21 borings over approximately
650 m2(0.7 acre). Borings were collected to a depth of 6.4 m using a hydraulic probe
and samples were obtained at approximately 1-m intervals. A field GC and heated
headspace technique (see "Field Methods for Determining Soil Gas and Soil VOCs)
was used to analyze 10- to 20-g subsamples for seven chlorinated aliphatic
compounds. Data obtained clearly demonstrated the extreme spatial heterogeneity
of soil VOCs at the site. The total soil-VOC concentrations varied from 6 [ig/kg to
154,000 fig/kg, and 90% of the sample VOC concentrations were between 94 ng/kg
and 20,100 fig/kg. The authors suggest that very high sample densities are needed to
57
-------
estimate the total mass of VOC within contaminated soil or to visualize a three-
dimensional soil-VOC distribution.
SAMPLE STORAGE AND PRESERVATION
Constraints on the Container Material
The ideal material for sample storage will maintain the sample integrity
without degradation of container material. Sample integrity can be
compromised by either sorption or leaching of organic contaminants of interest.
Leaching may be resorption of sorbed compounds or a release of "free" plasticizers
present in the bulk polymer.
Ten materials, including three metals, six synthetic polymers, and borosilicate
glass, were evaluated for halocarbon loss in laboratory experiments (Reynolds et al.,
1990). Borosilicate glass was the only material that did not diminish the halocarbon
concentrations. Loss rates for stainless steel were negligible for all compounds
tested except bromoform and hexachloroethane, which had losses amounting to 70%
after 5 weeks. The more halogenated compounds were generally removed before
the less halogenated compounds in solutions exposed to metals. Of the polymers,
the rigid polymers polytetrafluoroethylene (PTFE, Teflon®; E.I. Du Pont De
Nemours, Inc., Wilmington, DE) and rigid polyvinylchloride (PVC), were the best.
At low concentrations of halocarbons in water, these polymers showed little
adsorption during the 5-week experiment. At high concentrations, however, the
hydrocarbon solvents cause compound-specific swelling of polymers. Sorption rates
at high activities are dependent on concentration and this makes them difficult to
estimate. Therefore, if methanol or other solvent preservatives are used,
borosilicate glass is the best choice for storing samples. Also, after a polymer
container is used, the containers can become an important source of contamination
and should not be reused.
Use of stainless steel and polymer materials for suitability as well casing
materials was reviewed by Parker (1992). Stainless steel is susceptible to corrosion
in some soils (low pH, presence of H2S, and high salt content). Sample integrity of
metals is affected by stainless steel, but organics will not be affected by stainless
steel as long as the soil is not corrosive and adequate decontamination procedures
are followed. PTFE is highly resistant to chemical attack. However, the rate and
extent of sorption of chlorinated alkenes and chlorinated aromatics from aqueous
solution is greater for PTFE than for rigid PVC. For example, Parker et al. (1990)
found that loss of TCE was 10% after 8 h for PTFE, but losses for rigid PVC were
58
-------
only 6% after 1 week. Rigid PVC is resistant to chemical attack unless exposed to a
nearly saturated solution of a PVC solvent. PTFE leaches very little
contamination, but rigid PVC was a very close second. Parker (1992) concluded that
rigid PVC and stainless steel were the best choices for sampling organics, but that
PVC should be used if sampling for both organics and metals. Only threaded PVC,
PTFE, and stainless steel should be used because solvent bonding introduces
additional contaminants.
Studies of VOC-Spiked Storage Times
Zarrabi et al. (1991) reported on a batch approach to achieve "homogeneous"
VOC-spiked soils for use in studies of sampling methodology and potentially for
use as performance evaluation samples. Methanol solutions of VOCs were added to
moist soil in glass jars and mixed by tumbling for 120 seconds. Same-day results
showed benzene recoveries of approximately 40% for one soil and less than 10% for
another soil; chlorobenzene recoveries were 70% to 85% and 29% to 59%,
respectively, for the same two soils. Coefficients of variation for benzene and
chlorobenzene determinations ranged from 3% to 30%. Statistically significant
VOC losses occurred within 3 days of spiking, and recoveries and variability in the
data were "completely unacceptable" after 8 days.
Studies on storage of VOCs in water have shown that no leakage through
septum seals will occur if vials (40-mL VOA) are properly filled and sealed
(Maskarinec et al., 1990). The highly volatile chloromethane showed no loss over 56
days when stored in glass vials with Teflon-lined caps. In water, loss of 1,1,2,2-
tetrachloroethane with a concomitant increase in TCE occurred after 28 days at
4 °C. Trichloroethane appeared to degrade to dichloroethene; changes were noted
only after 56 days at 4 °C. Degradation of aromatic volatiles, especially styrene and
ethylbenzene, became apparent after storage for 28 days at 4 °C, but it could be
stopped by adding any of three acid preservatives tested (hydrochloric acid, sodium
bisulfate, and ascorbic acid). Degradation rates were faster at 25 °C. Within 28
days at 25 °C the disappearance of tetrachloroethane was accompanied by a
concomitant rise in the concentration of TCE, and decreasing levels of
trichloroethane were accompanied by increased concentrations of dichloroethene.
Storage of soil VOC-spiked standards, obtained by using the same water-
spike mixture and the same compounds as above (stock solution was mixed and
stored in a Tedlar bag; Maskarinec et al., 1990), has been less successful than the
storage of VOC-spiked water samples. Maskarinec et al. (1988) followed the
concentrations of a mixture of 15 volatile compounds in three soils for 56 days. The
59
-------
authors concluded at that time that 14 days should be the maximum holding time
prior to soil-VOC analysis, although there is no rigorous data analysis to support
the statement. In the same study, Maskarinec et al. (1989) showed very good initial
recoveries of the highly volatile compounds bromomethane (66% mean recovery)
and chloroethane (86% mean recovery) in three soils. The readily degradable
compound, styrene, had very low recoveries in two of the soils, producing a mean
recovery of 25%. Day-zero recoveries of 17 VOCs from three soils varied from 14 to
118% (mean recovery of 67% for all compounds) with standard deviations less than
25%.
Data on soil-VOC concentrations after storage of these samples were
presented by Jenkins et al. (1993). Data were expressed as the maximum holding
times (MHT) for the three soils, stored at 4, -20, or -70 °C. The sandy loam soil lost
VOCs rapidly under all storage conditions; MHT values were 0 to 3 days for most
of the compounds tested. The silt loam soil had MHTs of 0 to 14 days after storage
at 4 °C (mostly less than 3 days). At -70 °C the MHTs for the silt loam were also
low (mostly 0 to 1 days). At -20 °C, the silt loam MHTs were fairly high MHT
values were 18 days or more for all compounds (but bromomethane) as calculated
by the American Society of Testing and Materials (ASTM) method and MHT
values of 3 to 56 as calculated by the alternate method. The third soil, an
undescribed U.S. Army Toxic and Hazardous Materials Agency reference soil was
presterilized by an unspecified method. This soil exhibited MHTs of more than 100
days (calculated by either method) for 10 of the compounds when stored at -20 °C.
Tetrachloroethane was the only compound that had an MHT of less than 14 days at
-20 °C.
Achieving a good sample seal after soil has touched the sealing surfaces of a
vial is very difficult. This assertion was demonstrated by observing the loss of
methanol from vials that had been smeared with soil and then wiped with a gloved
hand before capping (Hewitt, 1992). Seven of ten soiled VOA vials showed
continuous weight loss, but no weight loss was observed in stored VOA vials that
had not been smeared with soil. The use of a syringe body with the nose removed
to subcore soil in the split-spoon sampler is recommended because the syringe can
be wiped free of soil particles, avoiding the problem of particles contacting the vial
sealing surfaces.
Addition of polymer absorbents (molecular sieve 5A and florisil) to reduce
sample volatilization losses during sample transfer steps increased spiked recoveries
(significant at p=O.05) for 42 of 60 soil/compound combinations tested (Zarrabi et al.,
1991). Still, recoveries in the presence of the solid absorbents were less than 50% for
60
-------
two of the three soils. In a 95% sand sample, the addition of the polymer
absorbents and traditional sample transfer steps showed increased VOC recoveries
over the modified sample lid/sparging procedure.
Amin and Narang (1985) reported that spiked frozen (-5 °C) sediment samples
can be stored as long as 7 days without significant loss of volatiles. Sediment
samples were sealed in vials, then spiked with a combination of 16 VOCs,
immediately frozen, stored, and connected to a purging system without opening the
vials. Losses were negligible on days 2, 3, 4, and 7, but losses of as much as 50%
occurred between 14 and 60 days of storage. Addition of a small amount of
methanol (1 mL methanol and 1.5 mL water per 5-g soil sample) preserved the
volatiles exceptionally well; negligible losses occurred over a period of 90 days.
Recoveries for 11 VOCs in samples preserved with methanol ranged from 57 to 99%
on day 1 and from 28 to 95% on day 90. Information on how the GC data were
quantified and when the internal standard (fluorobenzene) was added were not
given.
A study of five preservation methods for gasoline-contaminated soil
evaluated the following storage treatments (1) rubber-capped brass tubes, room
temperature, (2) rubber-capped brass tubes, 4 °C, (3) rubber-capped brass tubes on
dry ice, (4) 40-mL VGA vial with 5 mL methanol at room temperature, and (5)40-
mL VGA vial with 5 mL methanol at 4 °C (King, 1993). Separate batches of
laboratory-contaminated soil (at approximately 100 to 200 mg/kg as total petroleum
hydrocarbons, TPHs) were used for each treatment. Six samples from each batch
were analyzed for TPH concentration at 0, 3, 6, 10, and 14 days after sample mixing.
Treatment 1 (tubes that had not been refrigerated) showed a mean concentration of
91 mg/kg on day O decreasing to less than 20 mg/kg by day 6. Treatment 2 (tubes
with refrigeration) showed an initial concentration of 120 mg/kg decreasing to 34
mg/kg by day 6. Treatments 3, 4, and 5 did not show any deterioration exceeding
the "precision of the analytical methods" (statistics not provided) during the 14-day
holding time.
SUMMARY OF SAMPLING AND PRESERVATION METHODS
Lewis et al. (1991) reviewed sample design, selection of sampling devices,
sample collection procedures, and shipping considerations for soil VOCs The topics
covered by this review include sample preparation procedures and storage and
preservation techniques.
61
-------
Laboratory subsampling prior to analysis has been shown to create large
losses of VOCs (Amin and Narang, 1985; Maskarinec et al., 1988; Siegrist and Jenssen,
1990; Zarrabi et al., 1991; Hewitt, 1992). For example, losses during subsampling and
weighing were reported at 20% to 80% (Zarrabi et al., 1991) and averaging 60%
(Maskarinec et al., 1988). Losses from chilled soil ranged from 7% to 36% as
compared to losses ranging from 14% to 53% for nonchilled soil (Amin and Narang,
1985). Homogenizing the soil prior to weighing a sample creates higher losses than
samples transferred without homogenization (Siegrist and Jenssen, 1990). Samples
stored in containers with a large headspace lose more VOCs upon transfer than
samples stored in containers with little headspace (Siegrist and Jenssen, 1990). The
general trends outlined above mask the large variability of losses among
compounds on the same soil and the reported variability in losses of a single
compound among soil types.
Storage of soil samples for VOC analyses presents more challenges than
storage of water samples because prevention of volatile and degradative losses is
more complicated to accomplish and to verify. Soil samples are much more
difficult to seal than water samples due to the difficulty of removing all particles
that stick on the top rim of the jar. Slater et al. (1983) found that 31 of 54 jars of
soil leaked VOCs during 3 to 5 weeks of storage, shown by the presence of VOCs in
bags used to enclose the samples. Hewitt (1993) reported 7 of 10 vials leaked when
sealed after carefully removing soil from the rims of the vials. The addition of
polymer absorbents to prevent volatile losses may be helpful, especially in sandy
soils, but a separate step to extract VOCs from the absorbents may be required
(Zarrabi et al., 1991; Pignatello, 1990a). Soil placed in jars or vials via a subcorer
should largely eliminate the leaky seal problem (Hewitt, 1993).
Contract Laboratory Program (CLP) storage procedures currently allow soil
samples to be refrigerated (4 °C) for as long as 10 days after receipt at the
laboratory (USEPA 1991). Studies generally show, however, large losses of analyte
within 3 days when contaminated soils are stored at 4 °C (Maskarinec et al., 1989,
Zarrabi et al., 1991; King, 1993).
Degradation in refrigerated water samples was successfully halted by the
addition of acids, including sodium bisulfate (Maskarinec et al., 1990). In contrast,
soils are much harder to acidify. If soils are acidified, artifacts arising from
changes in the conformation of organic matter and changes in clay surface charges
may affect soil VOCs. Biocides such as sodium azide (Pignatello, 1990a) or mercuric
chloride (Zarrabi et al., 1991) can be used as preservatives, but these biocides are not
easy to use and dispose of. The preservation of gasoline-contaminated, fine-grained
62
-------
sand by keeping samples on dry ice was demonstrated by King (1993); however, the
results were reported as total petroleum hydrocarbons, so no compound-specific
preservation can be distinguished. Jenkins et al. (1993) found that -20 °C was
superior to -70 °C for preserving two soils, a silt loam and a presterilized soil
(method unspecified). However, in a third soil, a sandy loam, significant losses of
VOCs occurred at each temperature, -70 °C, -20 °C, and 4 °C (Jenkins et al., 1993).
The preservation of soil VOCs by adding the soil sample to a preweighed jar
of methanol in the field has been recommended by many researchers (Jenkins and
Schumacher, 1987; Bone 1988; Urban et al., 1989; Siegrist and Jenssen, 1990; Lewis et
al., 1991; Hewitt et al., 1992; King, 1993) and accepted as a standard procedure by the
ASTM (D 4547-91, "Standard Practice for .Sampling Waste and Soils for Volatile
Organic"). Methanol acts as a biocide and prevents loss by volatilization. The
amount of methanol necessary to prevent biodegradation and volatilization has not
been determined however, Amin and Narang (1985) found that 1 mL added to 5 g
of soil prevents loss of VOCs in frozen soil. Methanol provided a vapor trap that
was as effective as activated carbon for retaining ethylbenzene in a closed system
(Rhue et al., 1988). The analytical complications in a methanol-preserved sample are
addressed in the next section.
63
-------
SECTION 7
ANALYTICAL METHODOLOGY
This section describes (1) the SW-846 analytical methods (PT/GC/MS) used to
measure soil VOCs when extremely accurate and sensitive data are essential and (2)
studies of modifications that might improve the PT/GC/MS procedures. Methanol
has been shown to be a useful preservative, and therefore, studies of the analytical
sensitivity and precision that can be achieved with methanol extracts are also
discussed. Finally, studies that describe analytical procedures for measuring the
nonequilibrium or entrapped fraction of VOCs are presented.
SW-846 METHOD 8240 AND RELATED METHODS
SW-846 Method 8240 (USEPA, 1986, 1990) outlines the analysis of volatiles
from solid waste matrices by PT extraction and packed column gas GC/MS
detection. Proposed SW-846 Method 8260 (USEPA, 1990) is analogous to SW-846
Method 8240, except that a capillary gas chromatography column is used in place of
the packed column and the method is accompanied by an expanded list of analytes.
Related SW-846 methods (USEPA, 1990) include extractions, preparations, and
screening methods Method 5030, purge and trap; Method 3580, waste dilution
(methanol); Method 3810 (formerly Method 5020), screening by headspace analysis
and Method 3820, screening by hexadecane extraction. Analysis of specific groups
of analytes may be performed by other GC methods as follows: Method 8010,
Halogenated Volatile Organics, by halogen specific detector; Method 8011, 1,2-
Dibromoethane and l,2-Dibromo-3-chloropropane; Method 8015, Nonhalogenated
Volatile Organics, by flame ionization detector; Method 8020, Aromatic Volatile
Organics, by photometric ionization detector (ketones and ethyl ether); Method
8021, Halogenated and Aromatic Volatiles, by electrolytic and conductivity
detectors in series, capillary column; and Method 8030, Acrolein, Acrylonitrile, and
Acetonitrile, by flame ionization detector.
SW-846 Method 8240 suggests screening samples by hexadecane extraction or
by headspace analysis prior to GC/MS analysis to allow for the estimation of
appropriate sample size (1 g, 5 g, or methanol extraction) Samples that are likely
to contain greater than 1 mg/kg of an analyte are highly contaminated and must be
first extracted into methanol (1 or 4 g wet soil in 10 mL methanol). An aliquot of
the methanol is placed in water before purging to avoid overloading the gas
chromatagraph/mass spectrometer.
64
-------
Methods for the analysis of soil VOCs used in the Superfund Contract
Laboratory Program (CLP) are essentially SW-846 8240 and 8260 (USEPA, 1990) with
additional QA/QC data requirements. Both packed and capillary column methods
are included as described in the Statement of Work for Organics Analysis, Exhibit
D, Sections I through IV (USEPA, 1991). Optional sample screening by hexadecane
extraction is described and a medium-level soil method by methanol extraction,
analogous to the RCRA high-level soil method, is given in the CLP Statement of
Work (USEPA, 1991). Similarly, methods for the analysis of VOCs in water and
wastewater, "Purgeables"' Method 624 (EPA Environmental Monitoring Systems
Laboratory-Cincinnati) and "... isotope dilution GC/MS Method 1624 (EPA Office
of Water Regulations and Standards, Industrial Technology Division), also use PT
devices with GC/MS detection.
MODIFICATIONS OFFERED TO IMPROVE SOIL PURGE-AND-TRAP
ANALYSIS
PT analysis, as originally developed for the EPA, involved a 5-mL water
sample, an 11-min purge with an inert gas at 40 °C to a Tenax-GC trap, and a 4-min
resorption to a packed GC column (USEPA, 1982). Solid-sample matrices are ran in
like manner by adding 5 mL of water to approximately 5 g of solids, often
producing a foam, which makes stripping difficult, clogs the equipment, and
requires cleaning between samples (Amin and Narang, 1985). Historically, soils and
other solid samples analyzed with the PT extraction have low recoveries and
erratic precision (Lesnik, 1993).
Modifications to the PT technique that are designed to improve recovery of
VOCs from water are briefly mentioned. Whole column cryotrapping was
suggested by Pankow and Rosen (1988) as an alternative to the Tenax-GC traps.
Cryotraps placed after the Tenax trap are another option; many laboratories add it
to remove water before introducing the sample into the GC (Westendorf, 1992)
Increasing the purge temperature has also been suggested to improve purge
efficiency, and recent work suggests an alternative to replace oil or water baths
with mantle-style heaters (Jiang and Westendorf, 1992).
As compared with water samples, soil and sediments show poor spike
recoveries and high detection limits (Hiatt, 1981; Charles and Simmons, 1987). The
spike recoveries that are reported in accordance with SW-846 Method 8240/8260
procedures are in fact relative recoveries that do not reflect the absolute recoveries
or "efficiencies" of compound recovery. Recoveries are calculated by the relative
response factor of surrogate compounds, that is the peak area ratio of a surrogate
65
-------
to an internal standard compound. Both internal standards and surrogate additions
are added just before the sample is sparged. The internal standards and surrogates
are subject to the same matrix effect, and the response ratio does not quantify that
matrix effect.
Ward (1991) has suggested that the internal standard be introduced in a
microvolume sample loop placed between the sample vial and the GC inlet. A
valve would allow the internal standard to be swept out by the carrier gas during
the purge step. In this configuration, the internal standard would be used to
monitor the GC/MS system, and the surrogates would portray the matrix effects.
Techniques to improve extraction efficiencies from soils and solid matrices
over that achieved during the conventional purge (designed to extract VOCs from
water) have been developed. Cryogenic vacuum extraction was developed by Hiatt
(1981) for extraction of VOCs from sediment and fish tissues. Method 5032, to be
included in the Third Update to SW-846, is a vacuum distillation/cryogenic trap
procedure followed by GC/MS, applicable for extraction of a wide array of organics
from a variety of solid matrices (Lesnik, 1993).
Closed-loop stripping into a cryotrap with steam distillation of the volatiles
onto a Porapak N column was described by Amin and Narang (1985). VOCs were
then extracted from the absorbent column with methanol and the eluate was
analyzed by GC. Recovery of VOCs from spiked, frozen soil samples ranged from
60 to 100%.
Although altering the conductivity of the solution has been reported to
increase recovery of VOCs from water samples (attributed to salting out),
recoveries from soil samples generally do not increase with the conductivity of the
solution. Charles and Simmons (1987) investigated the effect of sediment type and
conductivity of the desorbing solution (0.01 M and 0.1 M KH2PO4) on the recovery
of PT analyses. The salt content of the desorbing solution did not affect VOC
recoveries from any of the three soils. Zarrabi et al. (1991) tested saturated NaCl
solution as a purging solution and found improved recovery of ketones but slightly
lower recoveries for the majority of VOCs.
The largest single improvement to soil-VOCs analysis by PT is arguably the
modified lids for 40-mL VOA vials that can be attached directly to a sparging
apparatus to avoid the laboratory subsampling step (Maskarinec et al., 1989; Zarrabi
et al., 1991; Hewitt et al., 1992). Teflon couplings (valves and fittings) are added in
place of the standard Teflon-faced septum and screw cap that also allow the
66
-------
addition of water and internal spiking solution to the sample before sparging. This
modified lid configuration has been discussed in the section "Obtaining and
Maintaining VOC Samples."
ANALYTICAL SENSITIVITY OF SOLVENT EXTRACTS
Analytically, soil methanol extracts are constrained to have lower sensitivity
than the conventional water PT extraction. The soil is effectively diluted when it
is placed in methanol in any proportion except equal parts of soil and methanol.
Furthermore, the addition of methanol to water in the sparger decreases the
recovery of at least some compounds by decreasing the aqueous-vapor distribution
coefficient. Kiang and Grob (1986) showed that headspace concentrations of eight
VOCs decreased as methanol in water increased from 1 to 20%. The effect was
most pronounced for the dichlorobenzenes. Control values with no methanol
addition were not presented because the VOC spike solution was in methanol.
Zarrabi et al. (1991) reported that the addition of methanol to soil in the
sparger at 1 and 10%o just before purging gave mixed results depending on the soil.
Differences were not significant at the 5% level as compared with the control (no
methanol). Trends in the data showed that 1% methanol had no effect or improved
the VOC recovery for a soil consisting of 95% sand and 0.14% organic carbon, but
10% methanol tended to decrease the VOC recovery in the same soil.
The CLP SOW methods (USEPA, 1991) and SW-846 Methods 8240, 8260,and
3580 (USEPA, 1990) use methanol to dilute samples that are too high in one or more
analyte. Methanol is added to soil (1:10 soil:methanol) and the methanol extract is
diluted into water (1:5 methanol:water, but this may vary) to quantify soil VOCs at
levels of 500 to 1,000 fig/kg. Using the high-level methanol extraction method (SW-
846 Method 8240), Hewitt et al. (1992) reported a detection limit on the order of 1000
ug/kg. The relative standard deviation of laboratory vapor-fortified soil by PT of
the methanol extract ranged from 0.5 to 38%, with a mean value of 12% at high-
level fortification and 16% at low-level fortification (Hewitt et al., 1992). Urban et
al. (1989) reported a detection limit of about 250 ^ig/kg for five VOCs (100 to 200 g
wet soil in 250 mL methanol) using the medium-level methodology in the CLP
SOW for Organics Analysis (7/87). Slater et al. (1983) used approximately equal
parts of wet soil and methanol and reported a quantification limit of 1,000 fig/kg
for TCE using a GC with a Hall detector.
Urban et al. (1989) showed that methanol-preserved samples contained as
much as an order of magnitude more VOCs than samples collected and analyzed by
67
-------
low-level methods. Duplicate analyses of methanol-preserved samples were
examined as well as duplicates generated by spiked soil/methanol samples left
overnight to equilibrate. Analyses of TCE, present in high concentrations in the
samples (at 1,000 to 11,000 fig/kg), gave relative percent differences of 0 to 6% in
duplicates. Spiked duplicate studies using five compounds (1,1-dichloroethene, TCE,
benzene, toluene, and chlorobenzene) showed recoveries well within the CLP
acceptance criteria (which vary by compound, but generally fall between 60 and
140%) and relative percent recoveries less than 9% (CLP acceptance values of 21 to
24%) for all compounds, except 1,1-dichloroethene. Recovery of 1,1-dichloroethene
was inconsistent; two of the four samples showed 86 to 96% duplicate recoveries of
1,1-dichloroethene with 0.3 to 1.0 relative percent differences; two other samples
showed 50 to 80% recoveries and 28 to 36 relative percent differences for this
compound.
A comparison of methanol and tetraethylene glycol dimethyl ether
(tetraglyme) for extraction of soil VOCs was reported by Jenkins and Schumacher
(1987). They found that methanol was better than tetraglyme at recovering four
volatiles from soil. Three vapor-contaminated soils were studied. Methanol was
found to extract 28.4% higher amounts of chloroform from a soil high in organic
matter (6.7% organic matter). Methanol was also easier to handle than tetraglyme
because it is less viscous and does not foam during the PT analysis. Tetraglyme,
like other ethers, is susceptible to the formation of peroxides, which can be
dangerous.
EXHAUSTIVE EXTRACTIONS TO RECOVER SORBED VOCs
Sawhney et al. (1988) investigated methodologies for recovering 1,2-
dibromoethane (EDB) from field-contaminated soil samples. They noted that most
methods for analyzing organics in soil are predicated on short-term spike-recovery
results (usually less than 24 h equilibration time). Soil samples that had years to
equilibrate with EDB in the field were tested for residual levels by PT, thermal
desorption, and solvent extraction techniques. Results for EDB recovered in their
Cheshire fine sandy loam soil by PT and by using various solvent extractions are
compared in Table 4. Hot methanol (75 °C for two 24-h periods) was used by the
authors as a base to compare extraction efficiencies. Purging as recommended by
SW-846 Method 8240 (11 min at 40 °C) resulted in 1.3% of the total recovered by
using hot methanol (90 ug EDB/kg soil). A series of repeated purges for longer
time periods, including two purges at 80 °C, amassed 26.1% of the EDB recovered by
hot methanol. Acetonitrile and acetone, heated to 75 °C, led to slightly higher EDB
recoveries than by hot methanol. Room temperature (20 °C) extractions using
68
-------
methanol, acetone, and acetonitrile for 24 h resulted in 60, 65, and 78% recoveries of
the hot methanol extraction, respectively. Using a ratio of 5 g soil to 25 mL
methanol, and a GC with an electron capture detector, their detection limit was 1.8
ug/kg. The concentration found in soils by hot solvent extraction ranged from 30
to 200 [ig/kg; PT recovered 1-2 fig/kg.
Sawhney et al. (1988) also tried thermal resorption on the EDB-contaminated
soil. Temperatures of 100 °C to 200 °C were investigated. They recovered
essentially no EDB from field-contaminated soil, but subsequent extraction with a
solvent showed that the EDB concentration remaining in the soil declined from
79% to 3.6% as the temperature increased, leading to the assumption that EDB was
destroyed by the high temperature necessary for thermal desorption. In contrast to
the field-contaminated soil, freshly spiked soil was completely and efficiently
thermally desorbed at 120 °C and at 200 °C if desorbed immediately yet freshly
spiked soil, allowed to sit for 18 h prior to thermal desorption, yielded only 49% in
thermal desorption. Apparently, surface exposed EDB is desorbed rapidly, but
entrapped EDB (which must diffuse from distant microsites) degrades before
desorbing.
Pavlostathis and Jaglal (1991) investigated the water-induced resorption of
TCE from a soil (silty clay texture, 0.13% organic carbon) that had been
contaminated for at least 18 years. Their objective was to produce a laboratory-
scale model of possible remediation techniques The effect of pH on TCE
resorption was negligible. The ionic strength of the aqueous desorbing solution
had no effect until the ionic strength exceeded 0.1 M, which produced a slightly
decreased desorption. The authors characterized the "total" TCE in the field-
contaminated soil as that obtained from a 13-h methanol extraction at 20 °C.
Continuous leaching in a soil column (approximately 24,000 pore volumes)
extracted 72% of the methanol-derived "total" soil TCE. Total soil TCE found by
extraction with cold methanol was estimated at approximately 2000 fig/kg.
Steinberg (1992) investigated the residual vapor-phase sorption of 1,1,1-
trichloroethrme, TCE, benzene, toluene, and ethylbenzene at spike concentrations
of several parts-per-thousand. An oven-dried calcareous soil (9.8%) carbonate as CO2,
0.16% organic carbon) was incubated for 2 to 87 h at temperatures of 5 °C, 25 °C,
and 45 °C. The labile fraction was then allowed to evaporate at 32 °C for 24 h. The
residual fraction was analyzed by (1) hot solvent extract, methanol slurry at 65 °C
for 24 h; (2) PT as described in SW-846 Method 8010 (1,1,1-trichloroethane only); and
(3) a "field extraction procedure" of hexane:water, 25 (volume basis). Sparging
(Method 8010) recovered approximately 10% of the 1,1,1-trichloroethane, and the
69
-------
TABLE 4. COMPARISON OF PURGE-AND-TRAP VERSUS SOLVENT EXTRACTION FOR ANALYSIS OF AGED,
EDb-CONTAMINATED SOILS*
Extraction Method Successive
Purge"
Purge and trap 1
2
3
4
5
Solvent N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
Solvent
NIA
N/A
N/A
N/A
N/A
methanol
methanol
methanol
methanol
methanol
methanol
acetonitrile
acetonitrile
acetone
Time
(hr)
0.18
0.18
0.5
0.5
0.5
24 (2X)
24
16
4
48
4
24
24
24
Temperature
(°C)
40
40
40
80
80
75
75
75
75
20
20
75
20
75
Relative Recovery
Per Purge (%)
1.3
1.0
3.6
9.6
10.9
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
Total Relative
Revocery (%)**
26.1
100
94
93
69
68
28
114
78
113
Abbreviation : EDB = 1,2-dibromoethane; N/A = not applicable
* Data compiled from Sawhney et al., 1988.
+ Consectutive runs/purges on the same sample
** Based on recovery of two 24-hour extraction
-------
quick hexane extraction recovered approximately 40% of the residual concentration
found by the hot solvent extraction. The increase in the residual fraction with
incubation time was shown to be a pseudo first-order process. The effect of
temperature showed a substantial activation energy for entrance of VOCs into the
residual fraction.
Supercritical fluid extraction (SFE) is an innovative extraction technology
that is most appropriate for semivolatiles research but may prove useful for
volatiles. SFE may provide a rapid technology for extraction of semivolatiles.
One of the unsolved problems for extraction of volatiles is the appropriate
extractant. Solvents often produce large peaks that mask the VOC peaks of
interest. Hawthorne et al. (1992) reported on SFE of semivolatiles and some
relatively volatile organics, including n-octane. They noted that native analytes
generally extract more slowly than spiked analytes from the same matrix and that
recoveries are highly matrix dependent. An exhaustive SFE, if found, could greatly
decrease the extraction time for slowly desorbing VOCs.
SUMMARY OF ANALYTICAL METHODOLOGY
PT methods that were developed for extracting VOCs from water have poor
recovery and low precision when applied to soils. Surrogate recoveries (reported as
the peak area ratio of surrogate/internal standard) do not reflect the extraction
efficiency or actual recovery from a soil matrix because surrogate compounds and
internal standards are introduced to the sample essentially simultaneously and
therefore, are subject to the same matrix effects. The addition of a micro volume
sample loop to add internal standards directly into the GC, bypassing contact with
the soil, has been suggested to alleviate this problem (Ward, 1991).
The addition of salts to the sparging solution to produce a "salting out" of
organics has not been effective in tests with soil (Charles and Simmons, 1987;
Zarrabi et al., 1991). Vacuum extraction with a cryogenic trap recovers VOCs and
other organics (including nonconventional VOCs) more efficiently than PT and
will be offered as Method 5032 in the Third Update to SW-846 (Hiatt, 1981; Lesnik,
1993).
The use of methanol to immerse field samples as a preservative technique
produces an analytical sensitivity of approximately 250 to 1,000 fig/kg when
analyzed by PT/GC/MS (Slater et al., 1983 Urban et al., 1989; Hewitt et al., 1992).
This loss of sensitivity results from the dilution of the soil in the methanol and the
decreased vapor pressure of VOCs in methanol. As the amount of methanol in the
71
-------
sparger increases, the purging efficiency decreases. Using direct injection but more
specific detectors than MS, detection limits can be decreased (e.g., 1.8 ug EDB/kg soil
with an electrolytic conductivity detector, Sawhney et al. 1988). Methanol is
generally superior to tetraglyme as a VOC extractant because recovery of VOCs
from tetraglyme was poor from a soil high in organic matter (Jenkins and
Schumacher, 1987).
More rigorous extraction techniques such as hot methanol or pulverization
of the soil (discussed under "Nonequilibrium Sorption" in section "Interphase
Transfers") have shown that some fraction of the soil VOCs are entrapped in
microsites in the soil or diffuse from them too slowly to be solubilized during the
PT extraction. Levels of EDB in ground water exceeding health standards were
traced to a slow diffusion of EDB from soils that had concentrations of 30 to 200 ug
EDB/kg soil when extracted by hot methanol and 1 to 2 gg EDB/kg soil when
extracted by PT (Sawhney et al. 1988). Thermal resorption was observed to destroy
the entrapped EDB, but quantitatively recover fresh (less than 18 hr) additions of
EDB (Sawhney et al., 1988).
Cold methanol was used to extract a soil that was contaminated with TCE 18
years prior to analysis (Pavlostathis and Jaglal, 1991). Continuous leaching (24,000
pore volumes) with water extracted 76% of the 2000 fig/kg found by extraction with
cold methanol. A soil that was contaminated in the laboratory by exposure to a
mixture of VOC vapors was extracted by hot methanol, hexane/water, and PT
(Steinberg, 1992). PT and hexane extracted approximately 10% and 40%,
respectively, of that which was extracted by hot methanol.
72
-------
SECTION 8
FIELD METHODS FOR DETERMINING SOIL GAS AND SOIL VOCS
Field methods for assessing soil VOCs offer advantages under many
circumstances. Literature that describes the uses of soil-VOC field methods is
briefly reviewed below, followed by soil-gas studies, soil headspace methods, and
finally, advanced field analysis techniques. Throughout this section, studies that
compare field data with laboratory data were targeted.
JUSTIFICATION FOR FIELD METHODS
The EPA has encouraged development of field technologies for on-site
analysis of samples at hazardous waste sites (Wesolowski and Alwan, 1991; Fribush
and Fisk, 1992). This impetus has resulted in improved efficiency for many short-
term environmental projects that are delayed by time-consuming, expensive, and
cumbersome laboratory procedures (Spittler, 1992; Wesolowski and Alwan, 1991).
Instrumentation that provides qualitative VOC screening includes total organic
analyzers with photoionization (PID) or flame ionization (FID) detectors. These
detectors have been available for approximately 20 years. Numerous brands and
types of field-deployable GCS have been and continue to be developed which
permit identification and quantification of volatiles with increasing precision and
sensitivity. Field methods are no longer limited to screening methods. Mobile
laboratory and field-deployable instrumentation is beginning to rival the
identification and low detection limits of SW-846 Method 8240.
Field methods, calibrated by relating field data to laboratory results, can
offer considerable reductions in cost and time (e.g., Cornell, 1992). Duplicates of a
representative portion of the field samples (10% of the field samples is suggested)
may be sent to the laboratory for analysis. Calibration curves of field versus
laboratory data can then be generated. The greatest uncertainty in data derives
from nonanalytical factors: site heterogeneity, sample-matrix variability, sample
collection procedures, and sample handling, transport, storage, and preparation
procedures. Cornell (1992) concludes that the analysis method will not seriously
compromise data integrity as long as proper calibration of laboratory and field
methods is observed.
Field measurements are particularly attractive for analyses of soil volatiles
because VOC losses during storage are eliminated with field analyses. On the other
73
-------
hand, the relatively simple headspace analyses by hand-held total organic carbon
detectors gives highly variable data that are subject to many interferences.
Correlations between headspace analysis with total organic carbon detectors and
laboratory PT analysis is poor (Siegrist, 1992). Total organic carbon analyses are
particularly inappropriate in soils of high, naturally occurring organic contents
(Nadeau and Tomaszewicz, 1988).
Field analyses for soil VOCs has been specifically recommended by Mitchell
et al.,(1993) for yet another reason. They examined the spatial heterogeneity of soil
VOCs at a small (650 m2) waste solvent landfill. The variability of VOC
concentrations measured in 176 samples demonstrated that an enormous number of
samples are needed to visualize the three-dimensional distribution of soil VOCs or
to estimate the total concentration of soil VOCs Mitchell et al. (1993) advised that
sampling designs ought to incorporate increased numbers of spatially separate
measurements, to be achieved by using less expensive, on-site analyses.
SOIL-GAS MEASUREMENTS
Soil-gas measurements have become an increasingly important analytical tool
at hazardous waste sites. They are a relatively rapid and cost-effective means to
(1) locate a contaminant source; (2) delineate a ground-water contaminant plume; (3)
plan monitoring-well, soil-boring, or vapor-extraction locations; (4) assess leakage of
underground tanks, lines, and subsequent vapor migration; (5) monitor migration of
gases from landfills, impoundments, industrial facilities; and (6) monitor the
progress of a cleanup (Devitt et al., 1987). The relative concentration of a
contaminant is measured, the spatial distribution is mapped, and correlations with
soil or water contaminant concentration(s) are developed site by site. Analytical
determinations of soil gas range from generalized organic-vapor sensors to field-
deployable or mobile laboratory GCS. Reviews of methods and available
technology for soil-gas sampling include Balfour et al. (1987), Devitt et al. (1987),
Jowise et al. (1987), and Marrin and Kerfoot (1988).
Properly executed, soil-gas analysis gives a subsurface reading of the vapor
phase concentration, and provides data that cannot be reproduced with confidence
once the soil is removed from its place in the field. The measurements are used as
relative comparisons at a given location because of the many soil, compound, and
environmental factors that influence the vapor phase concentration (Reisinger et
al. 1987; Marrin and Kerfoot, 1988). General correlations of soil gas and soil VOCs
are often poor (C. L Mayer and E. N. Amick, personal communications, 1992). Soil
gas concentrations may be unreliable if measured during periods of large
74
-------
barometric pressure changes, during or immediately after a rain, or across a site
with considerable variation in soil texture (either vertically or horizontally).
Golding et al. (1991) however, reported fairly good correlations of TPH in soil gas
versus soil headspace and between soil gas and soil TPH by an Iowa laboratory
procedure (Iowa Department of Natural Resources Method OA-1; methanol
extraction followed by PT modeled after SW-846 Method 8015). The data were
generated from 12 sites in Iowa near underground gasoline storage tanks Soil
samples were removed from the same drill hole as soil-gas samples (soil samples
collected subsequent to and deeper than gas samples). The data spanned 3 orders of
magnitude of laboratory soil TPH (5 x 103to 5 x 106[ig/kg) and logarithmic
transformations were analyzed. Correlation coefficients of 0.63 to 0.73 revealed
substantial scatter in the data, but statistically significant results at the 99%
confidence level.
Influence of temporal, environmental, and site factors on soil-gas
concentrations are reported by soil-gas researchers. Johnson and Perrott (1991) and
Smith et al. (1990) noted that soil-gas concentrations corresponded positively with
seasonal temperature fluctuations (increased soil-gas concentrations coincide with
increased temperatures). Yeates and Nielsen (1987) noted that differences between
winter and summer concentrations occur when the frozen soil acts as a "lid,"
creating higher soil gas concentrations during winter because release to the
atmosphere is inhibited. Jowise et al. (1987) reported that daily average variability
for soil gas samples taken from the same drill hole and depth were approximately
7% (they considered this analytical variability) and the relative standard deviation
of data from the same gas probes over 4 weeks was 40% (estimated to be 7%
analytical and 33% field variability). Many studies noted that the vapor
concentrations were lower in dense or fine-grained material and increased in
coarse-grained material (Kuhlmeier and Sunderland, 1983 Marrin and Thompson,
1987), Concomitant changes in soil moisture with soil texture may account for some
of this effect.
An understanding of the site and the fate and transfer properties of VOCs is
critical when interpreting soil-gas data (e.g., Yeates and Nielsen, 1987). The presence
and relative concentration of a VOC in the soil gas at the time of sampling will
depend on many factors, including the distance from the source of contamination;
physiochemical properties, age, and distribution of contaminants soil properties
including porosity, water content, permeability, texture, and composition, large-scale
geologic or anthropogenic barriers or conduits for diffusion; and environmental
factors including temperature, barometric pressure, relative humidity, and wind
velocity at the soil surface (Mehran et al., 1983). Given that the vapor phase is
75
-------
subject to short-term variability as a result of changes in temperature and soil
water content, soil gas data are best applied to rapid assessment of recent spills or
leaky tanks and to trace ground-water contamination under relatively coarse
materials (Mehran et al., 1983).
Marks et al. (1989) reported on the correlation between soil gas and ground-
water concentrations of benzene and toluene for 48 sites contaminated with
petroleum hydrocarbons. Observed soil-gas concentrations were poor predictors of
absolute levels of benzene and toluene in ground water. Soil-gas contamination,
however, was a relatively good predictor of presence or absence of ground-water
contamination based on Chi-square tests that included site-specific and pair-specific
factors. Sandy soils had the lowest incidence of false positives and false negatives,
and clayey soils had the highest incidence of false results. Where the distance
between soil-gas and ground-water samples was less than 1.5 m, the accuracy of
predictions was 90% for benzene and 75% for toluene. In contrast, where distances
were more than 3 m, the correlation between soil gas and ground water fell to 53%
for benzene and 63% for toluene.
A study used to design and implement soil clean-up at the Phoenix-Goodyear
Airport (Rosenbloom et al., in press) found long-term soil-gas monitoring provided
better estimates of the total mass of TCE than soil-VOC analyses. Soil-gas
estimates were 2 to 100 times greater than soil-VOC estimates of the total mass of
TCE (which may still have been lower than the actual quantity present). Soil
VOCs were measured after field immersion of samples in methanol and thus only
the highly contaminated soil samples contributed to the soil-VOC estimates of TCE
mass. Total mass estimates were based on an equilibrium, steady-state model, and
Kocwas used to estimate the sorption coefficient.
Smith et al. (1990) found good correlations between soil gas and shallow
ground-water concentrations of TCE at Picanniny Arsenal in New Jersey.
However, the concentrations of TCE in soil [measured by the hot methanol extract
method of Sawhney et al. (1988)] were 1 to 3 orders of magnitude greater than
predicted by soil-gas concentrations and aqueous-phase organic carbon partitioning.
The contribution of the aqueous phase (Henry's Law) was less than 3% of the total
mass of TCE sorbed. They suggested that the increased concentrations of TCE in
soil resulted from nonequilibrium, or slow resorption of TCE from soil organic
matter relative to the dissipation by degradation and diffusion of soil gas.
76
-------
Fiber-optic organic-vapor sensors are a potentially novel method of analyzing
soil gas. Lieberman et al. (1991) and Apitz et al. (1992) are using a pulsed-laser fiber-
optic fluorescence technique for evaluation of fuels in soil. The fluorescent
response of fuels varies with soil surface area of dry soils. Fluorescent response
was decreased by the addition of water to sand and increased by the addition of
water to clay (Apitz, et al., 1992). Barnard and Walt (1991) report on a fluorphor-
polymer combination that responds to the adsorption of organic vapors They
chose a system that is sensitive to benzene, toluene, ethylbenzene, xylene, and
unleaded gasoline. Calibration for field use was to be assessed in the next phase of
their research.
SOIL HEADSPACE METHODS
Soil headspace methods estimate soil VOCs by measuring the concentration
of VOCs in the vapor phase above a soil-water mixture. Soil samples are placed in
a suitable jar, vial, or plastic bag to which water is added or has already been
added, and the headspace concentration is measured generally within a few hours
after the sample was collected. The volume of water added is adjusted to create a
constant headspace volume for comparison of standards and samples. After being
shaken for a short time, a sample of the gas above the soil-water mixture is drawn
by syringe and analyzed using a portable or field-deployable GC. If the jar is
heated before a sample is drawn, it is called a heated headspace method.
Headspace methods assume rapid equilibrium between soil and water and
between water and air. However, soil resorption rates have been shown to depend
on VOC residence times after an initial rapid resorption phase and therefore,
headspace methods potentially underestimate contaminant concentrations and may
result in false negatives (Pavlostathis and Mathavan, 1992). Hewitt et al. (1992)
contend that lower values and false negatives are more likely with samples that
must be shipped, stored, and subsampled in the laboratory than with headspace
analysis that is performed within a day of sample collection. In either case, an
exhaustive resorption is not possible using water over a few days time. One may
deduce that dynamic (PT) headspace procedures using 40°C and 11 minutes will
extract essentially the same fraction of soil VOCs as headspace techniques and this
was demonstrated by the comparison of collocated limited disruption PT/GC/MS
and HS/GC samples (Hewitt, 1992) in the section "Obtaining and Maintaining VOC
Samples."
Static headspace is generally limited by the sensitivity of the procedure.
Increasing the temperature or equilibration time of headspace analysis would be
77
-------
expected to increase the concentration desorbed. Inescapably, the detection limits
of headspace analysis are limited by the volume of vapor that can be
accommodated by direct injection (as opposed to the PT preconcentration step.)
Crockett and DeHaan (1991) evaluated the effects of various headspace
procedures on (unidentified) GC peak heights in eight field-contaminated soils.
They showed that increasing the amount of soil in a vial, while holding the
headspace essentially constant, increased the vapor concentration of the VOCs
although the rate of increase in peak height was less than the rate of soil mass
added. Use of a saturated NaCl solution extractant in place of water produced
significantly larger peaks for one of the eight test soils, and gave no statistically
significant differences in the other soils. Methanol extraction (5 g soil in 5 mL
methanol) followed by headspace analysis of an aliquot of the methanol in water
showed variable recoveries by compound. As compared to water, methanol
appeared to be more efficient in extracting or partitioning the late eluting peaks,
but tended to be less efficient in extracting or partitioning the early eluting peaks.
The relative effects of methanol as an extractant and on the partitioning of the
VOCs between solution and vapor phases cannot be determined from this study.
Dry-soil heated headspace analysis was more effective at recovering VOCs than
water, although the authors reported analytical problems related to condensed
water in the GC as a drawback to heated headspace analysis.
Hewitt et al. (1992) compared PT analyses, using modified VOA vial lids for
low-level samples and the methanol extraction for high-level samples (per SW-846
Method 8240), with headspace analyses of laboratory- and field-contaminated soils.
Soil samples of 1 to 3 g were extracted in 5 mL water for the low-level PT method,
in 20 mL of methanol for the high-level PT method, and in 30 mL of water for the
headspace analysis. A field-portable Photo Vac GC with a packed column and a
photoionization detector that provided baseline resolution for benzene, toluene,
TCE, and trans-1,2-dichloroethene was used for headspace analysis. In laboratory-
fortified samples, the correlation between PT analysis and headspace analysis was
essentially unity with a slope of 0.948 for a soil that had 1.45% organic carbon. A
second soil with 6.69% organic carbon, gave statistically significant (0.5% level)
higher concentrations by PT analysis for 19 of 28 compound/spike combinations. A
statistically higher value (0.5% level) was found by headspace analysis only once.
The higher concentrations measured by PT analysis were reported using both the
high-level (methanol) PT procedure and the low-level (with modified sparger lids)
PT procedure. The differences were greatest for the two compounds with the
highest octanol-water partition coefficients, TCE, and toluene. Benzene and trans-
1,2-dichloroethene results showed little difference between headspace and PT
78
-------
analysis at a high spike level. The PT analysis produced statistically higher values
of these compounds at low spike concentrations. Analytical precision was better
overall with the headspace method.
Results from field-contaminated soil samples showed no significant
differences between using PT and headspace analyses when samples were collected
by subsampling 2 to 3 g of soil directly from the split-spoon sampler (Hewitt et al.
1992). Field spatial variability created large standard deviations in the data for both
methods of analysis and no differences due to the analysis method were observed.
A second field-contaminated soil was sampled from a well-mixed bulk sample. It
had been contaminated with TCE for at least 18 years and had been shown to
exhibit slow aqueous resorption of TCE (soil donated from study of Pavlostathis
and Jaglal, 1991). Good sample precision and statistically different means (0.5%
level) between the two analysis methods were found. Methanol PT analyses (high-
level method) always extracted more TCE than the headspace analysis. The
method differences were not attributed to organic carbon levels (only 0.13% organic
carbon in this soil), but rather to the slow resorption kinetics. Supporting data
demonstrated increased concentrations of TCE in headspace samples as the
extraction time (shaking) was increased by successive 10 min intervals.
Zoeller et al. (1992) compared laboratory GC/MS by SW-846 Method 8240
with a field static headspace method that incorporated a portable PT concentrator
to lower the detection limits. They reported that for soil contamination of less
than 5 ppm, GC/MS results were generally more reproducible than field methods,
whereas at higher contaminant levels, field VOC analyses were comparable to
GC/MS analyses. The methanol extraction for highly contaminated soil by Method
8240 reduced the precision and raised the detection limits thus generating field and
laboratory data of similar quality.
ADVANCED FIELD EXTRACTION AND ANALYSIS METHODS
The concept of on-site, real-time analysis for VOCs and other pollutants
using field-deployable GCs is enormously attractive. As a consequence, field
analytical capabilities are constantly improving. The positive identification of
compounds at low parts per billion soil concentrations is obtained by PT GC/MS.
Other GC detectors may meet or exceed the sensitivity of the mass spectrometer,
but lack the positive compound identification of the mass spectrometer. Field-
deployable mass spectrometers are being developed (e.g., Trainor and Laukien, 1988;
Eckenrode and Drew, 1991; Leibman et al., 1991; Wise et al., 1991).
79
-------
Wise et al. (1991) described investigations of two different types of direct
sampling mass spectrometers for use as rapid screening tools for VOCs in soil,
water, and air. For water and soil samples, volatiles are purged from the sample
without a preconcentration trap. Quantification is accomplished by integrating the
area of a reconstructed purge profile for the ions corresponding to the target
analytes. Purge efficiencies of 20 to 90% for benzene, TCE, and tetrachloroethene
in five different soil matrices were reported relative to a Ph 7 water purge. (These
efficiencies were achieved after spiked soils were left undisturbed for at least 1 h
prior to analysis and are typical of vacuum-stripping spike efficiencies from soil).
Continued research on the development of methods for the identification and
quantification of compounds in complex mixtures was suggested.
Two other field-hardy instruments include PT systems. Trainor and Laukien
(1988) described a complete GC/MS system that can be operated from a four-wheel
drive vehicle. It is based on a quadruple analyzer and an electron-impact
ionization source that has both full scan and selected ion monitoring capabilities.
A field-portable GC/MS that has three operating modes, including a purge and
thermal resorption mode and cryofocusing for light volatiles, has been marketed
and discussed at scientific conferences (Eckenrode and Drew, 1991). Both
instruments have been described as facilitating screening operations.
Leibman et al. (1991) discussed a transportable GC/MS designed to meet the
procedures and QC criteria outlined in SW-846 Method 8260. They reported that
the qualitative and quantitative analysis of 68 target compounds and associated
internal standards and surrogates can be accomplished in the field in an automated
sequence executed every 25 minutes. If only screening is required, a steeper GC
oven temperature ramp can reduce the time per sample.
SUMMARY OF FIELD METHODS
Field methods for soil VOCs have the specific advantage of avoiding analyte
loss during sample transport and storage. As with other types of analytes,
efficiency of time and cost are gained by field analyses. With soil VOCs however,
the gains in time and efficiency may need to be converted to increased numbers of
sample analyses to provide sufficiently accurate data over spatially variable sites
(Mitchell et al., 1993).
Soil gas measurements have many functions, but the uses rely mainly on
relative vapor concentrations and site-specific correlations. The vapor phase
concentrations are influenced by site, soil, compound, and environmental factors.
80
-------
Vapors of TCE and gasoline components increase with increasing soil temperature
(Smith et al., 1990; Johnson and Perrott, 1991). Higher vapor concentrations of TCE
and BTX have been reported to occur in coarse-grained material as opposed to fine-
grained material (Kuhlmeier and Sunderland, 1985; Marrin and Thompson, 1987). In
general, correlations of soil-gas and ground-water contamination are best in coarse-
grained soils and when the soil gas measurement is taken within 15 m of ground
water (Marks et al., 1989). Vapor-phase fiber optic sensors are being developed that
are expected to facilitate rapid measurement of soil gas for many VOCs (Barnard
and Walt, 1991; Lieberman et al, 1991; Apitz et al., 1992).
Soil (static) headspace measurements lack the concentrating step that lowers
the detection limit of PT, but have been, widely used for soil-VOC field analyses.
Correlations of static soil headspace with soil-VOC measurements have been shown
for soils with parts-per-million level contamination (Hewitt et al., 1992; Zoeller et
al., 1992), but the soil headspace data may be suppressed in soils high in organic
carbon (Hewitt et al., 1992). Also, concentrations of VOCs in soils with slowly
desorbing compounds may be underestimated by soil headspace techniques (Hewitt
et al., 1992). Addition of NaCl or methanol to the water in a static headspace
measurement created mixed results in the recoveries of various VOCs from eight
different soils (Crockett and DeHaan, 1991). Dry-soil heated headspace was more
effective at recovering VOCs than water, but analytical problems were
encountered related to condensation of water in the GC (Crockett and DeHaan,
1991).
Field extraction and analysis techniques that are utilizing detectors of high
specificity have been reported. Field-deployable direct sampling mass
spectrometers (Wise et al., 1991) and field-deployable PT/GC/MS systems based on a
quadruple MS (Trainor and Laukien, 1988; Eckenrode and Drew, 1991) or ion trap
detector MS (Liebman et al., 1991) have been undergoing field testing and use. Field
instrumentation that can achieve the sensitivity and specificity of SW-846 Method
8240/8260 is eminently expected within a few years.
81
-------
SECTION 9
CONCLUSIONS
FATE AND TRANSPORT OF SOIL VOCS
A schematic of the generalized fate of soil VOCs is presented in Figure 4.
The four phases of soil VOCs are shown in ovals. Processes that affect the total
concentration of soil VOCs are in rectangular boxes. Ultimately, losses from the
vadose zone occur through volatilization, degradation, or movement to ground
water. Nonequilibrium sorption of VOCs, that is, slow sorption or diffusion into
microsites, creates an "entrapped" fraction of VOCs ("entrapped" preferred term of
Travis and Maclnnis [1992]). The quantity of NAPL present in a soil is affected by
the amount of NAPL released, the rate of the release, and the soil porosity.
Equilibrium among the phases in the field is not presumed. Measurement of
one phase, such as the vapor phase, does not generally provide estimates of the
total VOC concentration in the soil (Marrin and Thompson, 1987 Smith et al., 1990;
Rosenbloom et al., in press). Both errors in measurement and erroneous
assumptions of equilibria contribute to poor predictions.
Estimates of soil sorption of nonpolar VOCs are based on hydrophobic
partitioning (Koc values). Such estimates do not account for: (1) sorption by soils
low in organic matter (<0.2% organic matter), or (2) long-term nonequilibrium
sorption (Smith et al., 1990). Sorption tends to correlate with soil texture or
permeability in soils low in organic matter; this was reported in deep deposits in
California where high VOC concentrations associated with clay deposits, low VOC
concentrations associated with sand deposits (Kuhlmeier and Sunderland, 1985).
Residual NAPL saturation may augment the measured soil concentrations at some
sites, resulting in an artificially inflated estimate of the sorbed concentration for
use in models.
Adsorption of VOCs by dry soil minerals is 2 to 4 orders of magnitude
greater than VOC sorption by wet soils (Chiou and Shoup, 1985; Ong and Lion,
199la). Adsorbed nonpolar VOCs are readily displaced by water molecules, and
therefore as the relative humidity approaches 100%, the physically adsorbed VOCs
are dislodged (Chiou and Shoup, 1985; Ong and Lion, 1991b). Vapor sorption has not
been found under natural field conditions, but the requirement for humid air in
vapor extraction systems can be attributed to vapor sorption.
82
-------
00
Volatilization ]
Migration to/from
Groundwater
Degradation
Figure 4. Fate of Soil VOCs
-------
Degradation of naturally occurring VOCs is generally rapid in aerobic soils,
mediated by soil microorganisms, and limited by the availability of oxygen
(Alexander, 1977; Barker et al., 1987). Halogenated aromatics also biodegrade under
aerobic conditions, although the rate is generally slower than that of the
unhalogenated analogs (Kobayashi and Rittman, 1982; Wilson et al., 1983).
Halogenated aliphatic compounds degrade anaerobically by biotic and abiotic
processes, but at rates much slower than observed for the unhalogenated analogs
(Smith and Dragun, 1984; Parsons et al., 1985). Evidence for anaerobic co-metabolism
of chlorinated solvents, that is, anaerobic degradation of TCE, and similar
chlorinated solvents in the ample presence of more easily degraded carbon
compounds, has been shown (Cline and Viste, 1985; Wilson and Wilson, 1985).
Movement of VOCs in soil occurs by vapor diffusion, vapor advection,
NAPL advection, aqueous diffusion, and aqueous convection. Vapor diffusion is
more rapid than liquid diffusion. The rates of vapor diffusion in soil are
influenced by chemical concentration gradients, soil permeability, moisture content,
temperature, and physiochemical properties of the VOC. Rates of advection are
influenced by soil permeability, moisture content, pressure gradients, thermal
gradients, gravity, and physiochemical properties of the VOC or the VOC mixture.
Estimates of soil-VOC movement have traditionally been based on vapor
diffusion and convection with soil solution. Presumably, vapor diffusion is
responsible for the appearance of two fluorocarbons in deep Texas sediments 44 m
deep, 40 years after manufacture of the compounds (Weeks et al., 1982). Another
example was the appearance of TCE near the soil surface at a site in Arizona; here
upward diffusion from contaminated ground water occurred over approximately 40
m in less than 30 years (Marrin and Thompson, 1987). Movement of NAPL spills
and leaks in highly permeable soils, however, is probably dominated by NAPL
advection. For example, carbon tetrachloride contaminating ground water 177 m
deep (time of travel unknown) is believed to have moved by density-driven vapor
advection (Falta et al., 1989). At a different site, benzene, toluene, and xylene
(solvents less dense than water) were found to have traveled 24.4 m in less than 7
years, presumably moving by gravity-driven NAPL advection (Kuhlmeier and
Sunderland, 1985).
Mathematical models of VOC movement are predominantly research tools.
Few field data exist with which to calibrate or validate field-scale models.
Differences in soil-gas and soil-VOC data generally cannot be resolved by current
models. Accurate measurements of soil-VOC concentrations are necessary to
support site characterization, remediation design, and monitoring efforts.
84
-------
SAMPLE SIZE
Soil sample collection for VOCs is confined by the need for minimal soil
perturbation and complete avoidance of soil homogenization procedures in an
effort to minimize loss of volatile analytes (Hewitt, 1992). Combine this with the
constraint of only 1 to 5 g soil used for PT/GC/MS soil-VOC analysis (SW-846
Method 8240/8260, USEPA, 1986, 1990), and a severe sampling problem results.
Approximately 3 mL of unhomogenized soil is collected to represent a core that
might be 2.5 x 15 cm in size or almost 300 mL in volume. Many researchers have
noted that the small sample size used in a PT/GC/MS analysis aggravates the basic
objective of characterizing soil-VOC levels at a field scale (Slater et al., 1983; Bone,
1988; Eynon and Rushneck, 1989; Poulsen and Kueper, 1992; Hewitt et al., 1992). An
analysis that utilizes much larger soil samples, or homogenization/subsampling
procedures that maintain the VOCs, should receive a high priority in the selection
of soil-VOC sampling and analysis methods.
SAMPLE PRESERVATION AND ANALYSIS
Sample preservation for VOCs is critical. There is ample evidence that 20%
to 100% of some VOCs are lost from soil stored cold (4 °C) in sealed vials or jars
(Maskarinec et al., 1988; Urban et al., 1989; Siegrist and Jenssen, 1990; Zarrabi et al.,
1991; Hewitt, 1992; King, 1993). A large part of the losses have been attributed to the
laboratory subsampling step (Maskarinec et al., 1988; Zarrabi et al., 1991). Part of the
reported losses are likely to arise from volatilization during storage because of the
difficulty in ensuring adequate seals of sample jars or vials. Use of a subcorer that
has been wiped free of particles to place soil in jars or vials has been shown to
overcome the problem of poor seals (Hewitt, 1993). Losses caused by biodegradation,
most commonly biodegradation of petroleum compounds, certainly occur in some
soils. Degradation can be inferred in samples when the loss of certain compounds
are accompanied by a concomitant increase in daughter products during storage
(Maskarinec et al., 1989; Jenkins et al., 1993). Halogenated aliphanes, although
refractory in aerobic culture studies, may degrade in some seemingly aerobic soils
by co-metabolism in the presence of a suitable carbon source (Cline and Viste, 1985).
Preservation against degradation may not be necessary for all compounds or in all
soils, but refrigeration at 4 °C is not sufficient to halt losses in samples prone to
degradation.
The procedures that have been used to quantify soil-VOC losses furnish
possible preservation methods: (1) sample collection and storage in vials with lids
modified to attach directly to the PT spargeq; (2) use of the modified lids as in
85
-------
option (1) with the addition of a biocide such as mercuric chloride; (3) sample
storage on dry ice;
(4) addition of polymer adsorbent beads, such as Tenax, florisil, or molecular sieve
5A to the soil sample; and (5) placing the sample in methanol in the field.
Sample collection into VOA vials with modified lids, option (1), fails to deal
with the sample size problem and may still incur biodegradation losses, especially in
samples contaminated by petroleum products. The addition of a biocide, option (2),
may reduce biodegradation, but no studies were found that could demonstrate the
effectiveness of biocides, and the disposal of biocide-contaminated soil is not trivial.
The effectiveness of dry ice as a preservative, option (3), is described in only
one study and data were for TPH, not specific compounds (King, 1993). More data
are needed to develop and evaluate preservation by dry ice. Similarly, the use of
polymer adsorbent beads, option (4), has been attempted in only one study (Zarrabi
et al., 1991). The use of adsorbent beads such as Tenax in a removable trap has not
been tried.
Of the preservation methods identified, the immersion of a sample in
methanol (5), has the most data on which to base an evaluation. The use of
methanol will increase the method detection limits over those attainable by low-
level SW-846 Methods 8240/8260. Low-level PT/GC/MS (SW-846 Methods 8240/8260)
can detect low ng/kg levels of most VOCs, but the losses that occur before analysis
have been observed at 1 to 3 orders of magnitude (Urban et al., 1989; Hewitt, 1992).
Therefore, when subsampling steps are included, low-level PT/GC/MS is no more
sensitive than methanol immersion or field headspace techniques and losses of the
most volatile compounds by conventional PT/GC/MS (e.g., vinyl chloride) are
practically guaranteed. No studies have been found comparing low-level
PT/GC/MS without subsampling and high-level PT/GC/MS procedures.
PT is not an efficient extraction for soil VOCs That is, spike recoveries are
erratic, varying with soil type, compound, and the time of contact between the soil
and VOC. The addition of salts or methanol to the purging solution has been
investigated, but results are uneven and no overall improvement is noted. Vacuum
extraction with a cryogenic trap may be more efficient than PT, but data directly
comparing the two procedures are lacking. The addition of surrogates into soil
samples and internal standards directly into the GC has been suggested (Ward, 1991).
These procedures would expedite monitoring GC/MS performance with the
internal standards and establish the use of surrogates to measure the matrix effect.
86
-------
FIELD METHODS
Factors cited in support of field analyses include: (1) improved time and cost
efficiency (Wesolowski and Alwan, 1991; Fribush and Fisk, 1992; Cornell, 1992); (2)
large analyte losses during sample transport and storage are avoided (Siegrist, 1992),
and (3) the large number of samples necessary to accurately characterize a
heterogeneous site are only feasible using on-site techniques (Mitchell et al., 1993).
The decision to rely on field methods or to corroborate field methods with
laboratory methods should be based on data quality objectives, the relative
importance of site-specific nonanalytical factors (such as, extreme site or soil
heterogeneity), and the stability of the VOCs of interest (e.g., highly volatile vinyl
chloride, readily degradable aromatic compounds). The development of
performance evaluation materials (PEMs) is critical to the QA/QC of field methods.
No soil-VOC methods can be adequately evaluated without accurate reporting of
the efficiency of soil spike recoveries and the use of soil PEMs.
Correlations between soil-gas and soil-VOC concentrations are site-specific
and often weak. Attempts to predict soil gas from soil-VOC concentrations or soil
VOC from soil gas concentrations differ by orders of magnitude (Smith et al., 1990;
Rosenbloom et al., in press). These data indicate at least one of the following: (1)
assumptions of equilibria among sorbed, vapor, and aqueous phases do not
correspond to field conditions, or (2) measurements of either soil-gas or soil-VOC
concentrations are inaccurate or not sufficiently representative of the area over
which the estimate is applied.
Soil-gas measurements are useful for rapid assessment of localized "hot spots"
of contamination or to determine the relative spatial distribution of contaminant
concentrations at a site. Soil-gas concentrations have been seen to increase when
soil temperatures increase (Smith et al., 1990; Johnson and Perrott, 1991) Erratic soil
gas behavior is predicted to occur during periods of rapidly changing barometric
pressure as would be associated with large thunderstorms (Massman and Farrier,
1992).
Soil static headspace techniques are essentially PT techniques minus the
preconcentration step. Static headspace measurements can be readily performed in
the field and the correlation with laboratory PT analyses can be good, if the
laboratory subsampling step is avoided (Hewitt et al., 1992). Two examples of soils
that had lower recoveries of TCE measured by an (unheated) headspace method
than by SW-846 Method 8240 have been reported: (1) soil high in organic matter
(6.7% organic carbon) and, (2) soil that exhibited slow resorption of TCE (Hewitt et
87
-------
al., 1992). Therefore, corroboration with laboratory procedures are still necessary. Dry-
soil heated headspace has been shown to be a more rigorous extraction procedure than
water-heated headspace, but analytical complications of water condensing in the GC
were reported (Crockett and DeHaan, 1991).
88
-------
REFERENCES
Abriola, L. M., and G.F. Finder. 1985. A multiphase approach to the modeling of porous
media contamination by organic compounds 1. Equation development. Water Resour.
Res. 21:11-18.
Alexander, M. 1977. Introduction to Soil Microbiology. John Wiley & Sons, Inc., New
York
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.
Anderson, T.A, J.J. Beauchamp, and B.T. Walton. 1991. Fate of volatile and semivolatile
organic chemicals in soils: Abiotic versus biotic losses. J. Environ. Qual. 20:420-424.
Apitz, S.E., G.A.. Theriault, and S.H. Lieberman. 1992. Optimization of the optical
characteristics of a fiber-optic guided laser fluorescence technique for the in situ
evaluation of fuels in soils. In, Proceedings of The International Society for Optical
Engineering, Vol 1637. January 22, 1992, Los Angeles, CA.
ASTM, 1991. Standard Practice for Sampling Waste and Soils for Volatile Organics, in,
1992 Annual Book of ASTM Standards, Volume 11.04, p. 108-111.
Balfour, W.D., C.E. Schmidt, and B.M. Eklund. 1987. Sampling approaches for the
measurement of volatile compounds at hazardous waste sites. J. Hazardous Mater.
14:135-148.
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.
Barker, J.F., G.C. Patrick and D. Major. 1987. Natural attenuation of aromatic
hydrocarbons in a shallow sand aquifer. Ground Water Monitor. Rev. 7(Winter):64-71.
89
-------
Barnard, S. M., and D.R. Walt. 1991. Fiber-optic organic vapor sensor. Environ.
Sci.Technol. 25:1301-1304.
Bear, J. 1979. Hydraulics of Groundwater. McGraw-Hill, New York.
Boethling R.S., and M. Alexander. 1979. Microbial degradation of organic compounds at
trace levels. Environ. Sci. Technol. 13:989-991.
Bone, LI. 1988. Preservation techniques for samples of solids, sludges, and nonaqueous
liquids, pp. 409-414. In, L.H. Keith, ed., Principles of Environmental Sampling.
American Chemical Society, Salem MA.
Bouchard, B.C., S.C. Mravik, and G.B. Smith. 1990. Benzene and naphthalene sorption
on soil contaminated with high molecular weight residual hydrocarbons from unleaded
gasoline. Chemosphere 21:975-989.
Boyd, S.A., and S. Sun. 1990. Residual petroleum and polychlorobiphenyl oils as sorptive
phases for organic contaminants in soils. Environ. Sci. Technol. 24:142-144.
Broholm, K, J.A. Cherry, and S. Feenstra. 1992. Dissolution of heterogeneously
distributed solvents residuals. Subsurface Restoration Conference, Dallas, TX, June 21-
24, 1992. Environmental Protection Agency, R.S. Kerr Environmental Research
Laboratory, Ada, OK.
Brusseau, M.L 1991. Transport of organic chemicals by gas advection in structured or
heterogeneous porous media Development of a model and application to column
experiments. Water Resour. Res. 27:3189-3199.
Brusseau, M.L, R.E. Jessup, and P.S.C. Rao. 1991. Nonequilibriurn sorption of organic
chemicals: Elucidation of rate-limiting processes. Environ. Sci. Technol. 25:134-142.
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.
Charles, M.J., and M.S. Simmons. 1987. Recovery studies of volatile organics in
sediments using purge/trap methods. Anal. Chem. 59:1217-1221.
90
-------
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., L.J. Peters, and V.H. Freed. 1979. A physical concept of soil-water
equilibria for nonionic organic compounds. Science 206:831-832.
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.
Cline, P.V., and D.R. Viste. 1985. Migration and degradation patterns of volatile organic
compounds. Waste Management Res. 3:351-360.
Corapcioglu, M.Y., and A.L. Baehr. 1987. A compositional multiphase model for
groundwater contamination by petroleum products 1. Theoretical considerations. Water
Resour. Res. 23:191-200.
Cornell, F.W. 1992. Site investigations: The role of field screening and analysis devices.
In, Proceedings of R&D '92, National Research & Development Conference on the
Control of Hazardous Material, February 4-6, 1992, San Francisco, CA. Hazardous
Material Control Research Institute, Greenbelt, MD.
Crockett, A.B., and M.S. DeHaan. 1991. Field screening procedures for determining the
presence of volatile organic compounds in soil. In, Field Screening Methods for
Hazardous Wastes and Toxic Chemicals, February 12-14, 1991, Las Vegas, NV. U.S.
Environmental Protection Agency, Las Vegas, NV.
Devitt, D.A,, R.B. Evans, W.A. Jury, T.H. Starks, B. Eklund, and A. Gholson. 1987. Soil
Gas Sensing for Detection and Mapping of Volatile Organics. National Well Water
Association, Dublin OH. 270 pp.
Dilling, W.L., M.B. Tefertiller, and G.J. Kallos. 1975. Evaporation rates and reactivities
of methylene chloride, chloroform, 1,1,1-trichloroethane, trichloroethylene,
tetrachloroethylene, and other chlorinated compounds in dilute aqueous solutions.
Environ. Sci. Technol. 9:833-838.
Dragun, J. 1988. The Soil Chemistry of Hazardous Materials. Hazardous Materials
Control Research Institute, Silver Spring, MD. 458 pp.
91
-------
Eckenrode, B.A, and R.C. Drew. 1991. An evaluation of field processing and detection
limits for volatiles and semi-volatiles using a portable GC/MS system. Pittsburgh
Conference, March 4, 1991, Chicago, IL.
Eynon, B.P., and D. Rushneck. 1988. Comparison of results of TCLP and direct analyses
of environmental samples, pp. 406-419. In, Proceedings of Eleventh "Annual Analytical
Symposium. May 11-12, 1988, Norfolk, VA.
Falta, R.W., I. Javandel, K Pruess, and P.A. Witherspoon. 1989. Density-driven flow of
gas in the evaporation of volatile organic compounds. Water Resour. Res. 25:2159-2169.
Fribush, H. M., and J.F. Fisk. 1992. Field analytical methods for Superfund.
Environmental Lab 4:3 6-41.
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.
Gierke, J.S., N.J. Hutzler, and J.C. Crittenden. 1990. Modeling the movement of volatile
organic chemicals in columns of unsaturated soil. Water Resour. Res. 26:1529-1547.
Gierke, J.S., N.J. Hutzler, and D.B. McKenzie. 1992. Vapor transport in unsaturated soil
column: Implications for vapor extraction. Water Resour. Res. 28:323-335.
Golding, R.D., M. Favero, and G. Thompson. 1991. Comparison of field headspace
versus field soil gas analysis versus standard method analysis of volatile petroleum
hydrocarbons in water and soil. In, Field Screening Methods for Hazardous Wastes and
Toxic Chemicals, Second International Symposium, February 12-14, 1991, Las Vegas,
NV. U.S. Environmental Protection Agency, Las Vegas, NV.
Hamaker, J.W., and J.M. Thompson. 1972. Adsorption, pp. 49-143. In, Goring C.A.I, and
J.W. Hamaker, eds., Organic Chemicals in the Soil Environment. Vol I. Marcel Dekker,
New York.
Harmon, T.C., W.P. Ball, and P.V. Roberts. 1989. Nonequilibrium transport of organic
contaminants in groundwater. pp. 405-437. In, Sawhney B.L. and K Brown, eds.,
Reactions and Movement of Organic Chemicals in Soils. SSSA Spec. Publ. 22. Soil
Science Society of America, Inc. Madison, WI.
92
-------
Hartmans, S., and J.A.M. de Bent. 1992. Aerobic vinyl chloride metabolism in
Mycobacterium aurum LI. Applied Environ. Microb. 58:1220-1226.
Hassett, J.J. and W.L. Banwart. 1989. The sorption of nonpolar organics by soils and
sediments. In, Sawhney, B.L., and K Brow eds., Reactions and Movement of Organic
Chemicals in Soils. SSSA Special Publication Number 22. Soil Science Society of
America, Inc. Madison, WI.
Hassett, J.J., W.L Banwart, and R.A. Griffin. 1983. Correlation of compound properties
with sorption characteristics of nonpolar compounds by soils and sediments: Concepts
and limitations, p. 161-178. In, C.W. Francis and S.I. Auerbach eds., Environment and
Solid Wastes. Butterworths, Boston.
Hawthorne, S.B., D.J. Miller, J.J. Langenfeld, and M.D. Burford. 1992. Physiochemical
and instrumental factors that can control SFE recoveries of neutral and ionic pollutants
from environmental samples. Abstract No 679. Pittsburgh Conference Book of Abstracts.
PittCon'92. March 9-12, 1992. New Orleans, LA.
Hewitt, A.D. 1992. Review of current and potential future sampling practices for volatile
organic compounds in soils. Proceedings from the 16th Annual Army Environmental
R&D Symposium, 23-25 June 1992, Williamsburg, VA.
Hewitt, A.D. 1993. Vapor fortification A method to prepare quality assurance soil
samples for the analysis of volatile organic compounds. Abstract from the National
Symposium on Measuring and Interpreting VOCs in soils: State of the art and research
needs. January 12-14, 1993, Las Vegas, NV. U.S. Environmental Protection Agency, Las
Vegas, NV.
Hewitt, AD., 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.
Hiatt, M.H. 1981. Analysis offish and sediment for volatile priority pollutants. Anal.
Chem. 53:1541-1543.
Hillel, D. 1971. Soil and Water. Academic Press, Inc., New York. 288 pp.
Holtz, R.D. and W.D. Kovacs. 1981. An Introduction to Geotechnical Engineering.
Prentice-Hall Inc. Englewood Cliffs, NJ.
93
-------
Howard, P.H. 1990. Handbook of Environmental Fate and Exposure Data for Organic
Chemicals. Vols I, II, and III. Lewis Publishers, Inc. Chelsea, MI. 574 pp. Vol I; 684
pp. Vol II; 546 pp. Vol III.
Jenkins, R.A., C.K Bayne, M.P. Maskarinec, L.H. Johnson, S.K Holladay, and B.A.
Tomkins. 1993. Experimental determination of pre-analytical holding times for volatile
organics in selected soils. Abstract from the National Symposium on Measuring and
Interpreting VOCs in soils: State of the art and research needs. January 12-14, 1993, Las
Vegas, NV. U.S. Environmental Protection Agency, Las Vegas, NV.
Jenkins, T.F. and P.W. Schumacher. 1987. Comparison of methanol and tetraglyme as
extraction solvents for determination of volatile organics in soil. Special Report 87-22.
US Army Corps of Engineers, Cold Regions Research & Engineering Laboratory.
Hanover, NH.
Jiang, K, and R.G. Westendorf. 1992. A comparison of sample heating techniques used
in purge and trap gas chromatography. Abstract no. 895. Pittsburgh Conference Book of
Abstracts. PittCon'92. March 9-12, 1992, New Orleans, LA.
Johnson, R.L., and M. Perrott. 1991. Gasoline vapor transport through a high-water-
content soil. J. Contaminant Hydol. 8:317-334.
Jowise, P.P., J.D. Villnow, LI. Gorelik, and J.M. Ryding. 1987. Comparative analysis of
soil gas sampling techniques. In, Proceeding of the National Conference on Hazardous
Wastes and Hazardous Materials. Washington DC, U.S. Environmental Protection
Agency, Washington DC.
Jurinak, J.J., and D.H. Vohnan. 1957. Application of Brunauer, Emmett and Teller
equation to ethylene dibromide adsorption by soils. Soil Sci. 83:487-496.
Jury, W.A., D. Russo, G. Streile, and H. El Abd. 1990. Evaluation of volatilization by
organic chemicals residing below the soil surface. Water Resour. Res. 26:13-20.
Jury, W.A., D. Russo, G. Streile, and H. El Abd. 1992. Correction to "Evaluation of
volatilization by organic chemicals residing below the soil surface." Water Resour. Res.
28:607-608.
Jury, WA., W.F. Spencer, and W.J. Farmer. 1983. Behavior assessment model for trace
organics in soil: I. Model description. J. Environ. Qual. 12:558-564.
94
-------
Jury, W.A., W.F. Spencer, and W.J. Farmer. 1984. Behavior assessment model for trace
organics in soil: IV. Review of experimental evidence. J. Environ. Qual. 13:580-586.
Karickhoff, S.W. 1984. Organic pollutant sorption in aquatic systems. J. Hydraulic Eng.
110:707-735.
Karickhoff, S.W., D.S. Brow and T.A. Scott. 1979. Sorption of hydrophobic pollutants
on natural sediments. Water Res. 13:241-248.
Katyal, A.K, J.J. Kaluarachchi, and J.C. Parker. 1991. MOFAT: A Two-Dimensional
Finite Element Program for Multiphase Flow and Multicomponent Transport. Program
Documentation and User's Guide. EPA/600/2-91/020. U.S. Environmental Protection
Agency, Ada, OK
Kiang, P. H., and R.L. Grob. 1986. A headspace technique for the determination of
volatile compounds in soil. J. Environ. Sci. Health A21:71-1OO.
King, P.H. 1993. Evaluation of sample holding times and preservation methods for
gasoline in fine-grained sand. Abstract from the National Symposium on Measuring and
Interpreting VOCs in soils: State of the art and research needs. January 12-14, 1993, Las
Vegas, NV. U.S. Environmental Protection Agency, Las Vegas, NV.
Kobayashi, H., and B.E. Rittmann. 1982. Microbial removal of hazardous organic
compounds. Environ. Sci. Technol. 16: 170A-183A.
Kreamer, D.K, E.P. Weeks, and G.M. Thompson. 1988. A field technique to measure
the tortuosity and sorption-affected porosity for gaseous diffusion of materials in the
unsaturated zone with experimental results from near Barnwell, South Carolina. Water
Resour. Res. 24:331-341.
Kuhlrneier, P.O., and G.L. Sunderland. 1985. Distribution of volatile aromatics in deep
unsaturated sediments. In, Proceedings of the Conference on Characterization and
Monitoring of the Vadose (Unsaturated) Zone, November 19-21, Stapleton Plaza Hotel,
Denver, CO. National Water Well Association, Dublin, OH.
Lappala, E. G., and G.M. Thompson. 1983. Detection of groundwater contamination by
shallow soil gas sampling in the vadose zone. In, Proceedings of the Conference on
Characterization and Monitoring of the Vadose (Unsaturated) Zone. Dec. 8-10, 1983.
Las Vegas, NV. National Well Water Association, Dublin, OH.
95
-------
Leibman, C.P., D. Dogruel, and P.P. Vandemeer. 1991. Transportable GC/ion trap mass
spectrometry for trace field analysis of organic compounds. Field Screening Methods for
Hazardous Wastes and Toxic chemicals, February 12-14, 1991, Las Vegas, NV. U.S.
Environmental Protection Agency, Las Vegas, NV.
Lesage, S., and R.E. Jackson. 1992. Practical organic analytical chemistry for hazardous
waste site investigations. In, S. Lesage and R.E. Jackson (eds.), Groundwater
Contamination and Analysis at Hazardous Waste Sites. Marcel Dekker, Inc., New York
NY.545 pp.
Lesnik, B. 1993. New SW-846 methods for the analysis of conventional and non-
conventional volatile organics in solid matrices. Abstract from the National Symposium
on Measuring and Interpreting VOCs in soils: State of the art and research needs.
January 12-14, 1993, Las Vegas, NV. U.S. Environmental Protection Agency, Las Vegas,
NV.
Lewis, T.E., A.B. Crockett, R.L. Siegrist, and K Zarrabi. 1991. Soil sampling and analysis
for volatile organic compounds. EPA/540/4-91/001. U.S. Environmental Protection
Agency, Cincinnati, OH. 24 pp.
Lieberman, S.H., G.A. Theriault, S.S. Cooper, P.G. Malone, R.S. Olsen, and P.W. Lurk.
1991. Rapid, subsurface, in situ field screening of petroleum hydrocarbon contamination
using laser induced fluorescence over optical fibers. In Proceeding of Field Screening
Methods for Hazardous Wastes and Toxic Chemicals, February 12-14, 1991, Las Vegas,
NV.
Lyman, W.J., W.F. Reehl, and D.H. Rosenblatt. 1990. Handbook of Chemical Property
Estimation Methods. American Chemical Society. Washington DC.
Mabey, W., and T. Mill. 1978. Critical review of hydrolysis of organic compounds in
water under environmental conditions. J. Phys. Chem Ref. Data 7:383-415.
Marks, B.J., D.A Selby, R.E. Hinchee, and G. Davitt. 1989. Soil gas and groundwater
levels of benzene and toluene-Qualitative and quantitative relationships. In, Proceedings
of the Conference on Petroleum Hydrocarbons and Organic Chemicals in Ground
Water: Prevention, Detection and Restoration. National Water Well Association, Nov
15-17, 1989, Houston TX.
96
-------
Marrin, D.L., and H.B. Kerfoot. 1988. Soil-gas surveying techniques. Envirom Sci.
Technol. 22:740-745.
Marrin, D.L., and G.M. Thompson. 1987. Gaseous behavior of TCE overlying a
contaminated aquifer. Ground Water 25:21-27.
Maskarinec, M.P., LH. Johnson, and C.K Bayne. 1989. Preparation of reference water
and soil samples for performance evaluation of volatile organic analysis. J. Assoc. Off.
Anal. Chem. 72:823-827.
Maskarinec, M.P., L.H. Johnsou and S.K Holladay. 1988. Recommendations for holding
times of environmental samples. In, Proceedings Waste Testing and Quality Assurance.
Vol II. July 11-15, 1988. The Westin Hotel, Washington DC.
Maskarinec, M.P., LH. Johnson, S.K Holladay, R.L Moody, C.K Bayne, and RA
Jenkins. 1990. Stability of volatile organic compounds in environmental water samples
during transport and storage. Environ. Sci. Technol. 24:1665-1670.
Massmann, J., and D.F. Farrier. 1992. Effects of atmospheric pressures on gas transport
in the vadose zone. Water Resour. Res. 28:777-791.
Mehran, M., MJ. Nimmons, and E.B. Sirota, 1983. Delineation of underground
hydrocarbon leaks by organic vapor detection. In, Management of Uncontrolled
Hazardous Waste Sites, October 31-November 2, 1983, Washington DC, Hazardous
Materials Control Research Institute, Silver Spring, MD.
Mendoza, CA, and E.O. Frind. 1990. Advective-dispersive transport of dense vapors in
the unsaturated zone 1. Model development. Water Resour. Res. 26:379-387.
Mendoza, CA, and T.A. McAlary. 1990. Modeling of ground-water contamination
caused by organic solvent vapors. Ground Water 28:199-206.
Metcalfe, D. E., and G.J. Farquhar. 1987. Modeling gas migration through unsaturated
soils from waste disposal sites. Water, Air, and Soil Pollution 32:247-259.
Millington, RJ., and J.P. Quirk. 1961. Permeability of porous solids. Trans. Faraday Soc.
57:1200-1207.
97
-------
Mingelgrin, U., and Z. Gerstl. 1983. Reevaluation of partitioning as a mechanism of
nonionic chemicals adsorption in soils. J. Environ. Qual. 12:1-11.
Munz, C., and P.V. Roberts. 1987. Air-water phase equilibria of volatile organic solutes.
Journal Am. Water Works Assoc. 79:62-69.
Nadeau, R.J., and F. Tomaszewicz. 1988. Influence of naturally occurring volatile
compounds on soil gas results. In, Proceedings of Field Screening Methods for
Hazardous Waste Site Investigations, October 11-13, 1988. U.S. Environmental
Protection Agency, Las Vegas, NV.
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 LW. Lion. 1991b. Trichloroethylene vapor sorption onto soil minerals.
Soil Sci. Soc. Am. J. 55:1559-1568.
Pankow, J.P., and M.E. Rosen. 1988. Determination of volatile compounds in water by
purging directly to a capillary column with whole column cryotrapping. Environ. Sci.
Technol. 22:398-405.
Parker, L.V., AD. Hewitt and T.E. Jenkins. 1990. Influence of casing materials on trace-
level chemicals in well water. Ground Water Monitor. Rev. 10(Spring):146-156.
Parker, LV. 1992. Suggested guidelines for the use of PTFE, PVC, and stainless steel in
samplers and well casings. In, Nielsen D.M. and M.N. Sara, eds., Current Practices in
Ground Water and Vadose Zone Investigations. ASTM STP 1118. American Society for
Testing and Materials, Philadelphia, PA. 431 pp.
Parsons, F., G.B. Lage, and R. Rice. 1985. Biotransformation of chlorinated organic
solvents in static microcosms. Environ. Toxic. Chem. 4:739-742.
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. Resorption kinetics of selected volatile
organic compounds from field contaminated soils. Environ. Sci. Technol. 26:532-538.
98
-------
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. Envirom Toxic. Chem. 9:1117-1126.
Finder, G.F., and LM. Abriola. 1986. On the simulation of nonaqueous phase organic
compounds in the subsurface. Water Resour. Res. 22:109S-119S.
Piwoni. M.D., and P. Banerjee. 1989. Sorption of volatile organic solvents from aqueous
solution onto subsurface solids. J. Contain. Hydrol. 4:162-179.
Plumb, R.H. 1991. The occurrence of Appendix IX organic constituents in disposal site
ground water. Ground Water Monitoring Rev. ll(Spring):157-164.
Poe, S.H., KT. 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.
Poulsen, M.M., and B.H. Kueper. 1992. A field experiment to study the behavior of
tetrachloroethylene in unsaturated porous media. Environ. Sci. Technol. 26:889-895.
Reisinger, H.J., D.R. Burris, LR. Cessar, and G.D. McCleary. 1987. Factors affecting the
utility of soil vapor assessment data. In, Proceedings of the First National Outdoor
Action Conference on Aquifer Restoration, Ground Water Monitoring and Geophysical
Methods, May 18-21, 1987, Las Vegas, NV. National Water Well Association, Dublin,
OH.
Reynolds, G.W., J.T. Hoff, and R.W. Gillham. 1990. Sampling bias caused by materials
used to monitor halocarbons in groundwater. Environ. Sci. Technol. 24:135-142.
Rhue, R.D., P.S.C. Rao, and R.E. Smith. 1988. Vapor-phase adsorption of alkylbenzenes
and water on soil and clays. Chemosphere 17:727-741.
Rosenbloom, J., P. Mock P. Lawson, J. Brown, and H.J. Turin, in press. Site-specific soil
clean-up levels for volatile organic compounds. Groundwater.
Roy, W.A., and R.A Griffin. 1990. Vapor-phase interactions and diffusion of organic
solvents in the unsaturated zone. Environ. Geol. Water Sci. 15:101-110.
99
-------
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., JJ. 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.
Schwarzenbach, R.P., and J. Westall. 1981. Transport of nonpolar organic compounds
from surface water to groundwater. Laboratory sorption studies. Environ. Sci. Technol.
15:1360-1367.
Shoemaker, C.A., T.B. Culver, L.W. Lion, and M.G. Peterson. 1990. Analytical models of
impact of two-phase sorption on subsurface transport of volatile chemicals. Water
Resour. Res. 26:745-758.
Siegrist, R.L 1993. VOC measurement in soils: The nature and validity of the process.
Abstract from the National Symposium on Measuring and Interpreting VOCs in soils:
State of the art and research needs. January 12-14, 1993, Las Vegas, NV. U.S.
Environmental Protection Agency, Las Vegas, NV.
Siegrist, R.L. 1992. Volatile organic compounds in contaminated soils: The nature and
validity of the measurement process. J. Hazard. Mat. 29:3-15.
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.
Silka, L.R. 1988. Simulation of vapor transport through the unsaturated zone-
Interpretation of soil-gas surveys. Ground Water Monitor. Rev. 8 (Spring):! 15-123.
Sims, R.C. 1990. Soil remediation techniques at uncontrolled hazardous waste sites. A
critical review. J. Air Waste Manage. Assoc. 40:704-732.
Slater, J.P., F.R. McLaren, D. Christenson, and D. Dineen. 1983. Sampling and analysis
of soil for volatile organic compounds: I. Methodology Development. In, Proceedings of
the Conference on Characterization and Monitoring of the Vadose (Unsaturated) Zone.
Dec. 8-10, 1983. Las Vegas, NV. National Well Water Association, Dublin, OH.
100
-------
Sleep, B.E., and J.F. Sykes. 1989. Modeling the transport of volatile organics in variably
saturated media. Water Resour. Res. 25:81-92.
Smith, J.A., C.T. Chiou, J.A. Kammer, and D.E. Kile. 1990. Effect of soil moisture on
the sorption of trichloroethene vapor to vadose-zone soil at Picatinny Arsenal, New
Jersey. Environ. Sci. Technol. 24:676-683.
Smith, L.R., and J. Dragun. 1984. Degradation of volatile chlorinated aliphatic priority
pollutants in groundwater. Environ. Int. 10:291-298.
Spencer, W.F., and M.M. Cliath. 1970. Resorption of lindane from soil as related to
vapor density. Soil Sci. Soc. Am. Proc. 34:574-578.
Spittler, T. 1992. Field analysis. An interview with Dr. Thomas Spittler. Analytical
Consumer. March 1992, pp. 9-10.
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.
Sulfita, J.M. 1989. Microbial ecology and pollutant biodegradation in subsurface
ecosystems. Chapter 7. In, Transport and fate of contaminants in the subsurface. Seminar
Publication. EPA/625/4-89/019. U.S. Environmental Protection Agency, Cincinnati, OH.
Swallow, KC. 1992. Nonpriority pollutant analysis and interpretation. In, S. Lesage and
R.E. Jackson (eds.), Groundwater Contamination and Analysis at Hazardous Waste
Sites. Marcel Dekker, Inc., New York, NY.545 pp.
Taylor, S.A., and G.L Ashcroft. 1972. Physical Edaphology. The physics of irrigated and
nonirrigated soils. W.H. Freeman and Company, San Francisco, CA 533 pp.
Thorstenson, D.C., and D.W. Pollock. 1989. Gas transport in unsaturated zones:
Multicomponent systems and the adequacy of Pick's Laws. Water Resour. Res. 25:477-
507.
101
-------
Trainor, T.M., and F.H. Laukien. 1988. Design and performance of a mobile mass
spectrometer developed for environmental field investigations. In, Field Screening
Methods for Hazardous Waste Site Investigations, October 11-13, 1988, Las Vegas, NV.
U.S. Environmental Protection Agency, Las Vegas, NV.
Travis, C.C., and J.M. Maclnm's. 1992. Vapor extraction of organics from subsurface soils,
Is it effective? Environ. Sci. Technol. 26:1885-1887.
Urban, M.J., J.S. Smith, E.K Schultz, and R.K Dickinson. 1989. Volatile organic analysis
for a soil, sediment or waste sample. In, Proceedings of the Fifth Annual Waste Testing
and Quality Assurance Symposium, July 24-28, 1989, Washington DC. U.S.
Environmental Protection Agency, Cincinnati, OH.
USEPA. 1982. Methods for Organic Chemical Analysis of Municipal and Industrial
Wastes. Test Method 624, Purgeables. EPA-600/4-82-057. U.S. Environmental Protection
Agency, Cincinnati, OH.
USEPA. 1986. Test Methods for Evaluating Solid Waste, SW-846. 3rd Edition. Office of
Solid Waste and Emergency Response. U.S. Environmental Protection Agency,
Washington DC.
USEPA. 1990. Update II to SW-846. Methods Section. Office of Solid Waste. U.S.
Environmental Protection Agency, Washington DC.
USEPA. 1991. Statement of Work for Organics Analysis. USEPA Contract Laboratory
Program. Document Number OLMO1.8, Revision 8, Augusg 1991. U.S. Environmental
Protection Agency, Research Triangle Park NC.
van Ee, J.J., L.J. Blume, and T.H. Starks. 1990. A Rationale for the Assessment of
Errors in the Sampling of Soils. EPA/600/4-90/013. Office of Research and
Development, Environmental Monitoring Systems Laboratory, Las Vegas, NV.
Vogel, T.M., C.S. Criddle, and P.L. McCarty. 1987. Transformations of halogenated
aliphatic compounds. Environ. Sci. Technol. 21:722-736.
Voudrias, E.A., and M. Reinhard. 1986. Abiotic organic reactions at mineral surfaces. In,
Davis, J.A. and KF. Hayes, eds., Geochemical Processes at Mineral Surfaces. American
Chemical Society, Washington, DC. pp. 462-486.
102
-------
Ward, S.E. 1991. GC/MS VOC method improvements. Environmental Lab 3:42-45.
Weeks, E.P., D.E. Earp, and G.M. Thompson. 1982. Use of atmospheric fluorocarbons
F-ll and F-12 to determine the diffusion parameters of the unsaturated zone in the
southern high plains of Texas. Water Resour. Res. 18:1365-1378.
Wesolowski, D., and A. Alwan. 1991. Field measurement of organic compounds by gas
chromatography. In, Simmons, M.S. ed., Hazardous Waste Measurements. Lewis
Publishers, Chelsea, MI. 315 pp.
Westendorf, R.G. 1992. Theory and operation of water management methods in purge
and trap gas chromatography. Abstract no. 893. Pittsburgh Conference Book of
Abstracts. PittCon'92. March 9-12, 1992. New Orleans, LA.
Wilson, J.T., and J.F. McNabb. 1983. Biological transformation of organic pollutants in
groundwater. EOS 64(33).
Wilson, J.T., J.F. McNabb, D.L. Balkwill, and W.C. Ghiorse. 1983. Enumeration and
characterization of bacteria indigenous to a shallow water-table aquifer. Groundwater
21:134-142.
Wilson, J.T., and B.H. Wilson. 1985. Biotransformation of trichloroethylene in soil.
Applied and Environ. Microb. 49:242-243.
Wise, M.B., G.B. Hunt C.V. Thompson, M.V. Buchanan, and M.R. Guerin. 1991.
Screening volatile organics by direct sampling ion trap and glow discharge mass
spectrometry. In, Field Screening Methods for Hazardous Wastes and Toxic Chemicals,
February 12-14, 1991, Las Vegas, NV. U.S. Environmental Protection Agency, Las
Vegas, NV.
Yeates, G.L., and D.M. Nielsen. 1987. Design and implementation of an effective soil gas
monitoring program for four-dimensional monitoring of volatile organics in the
subsurface. In Proceedings of the NWWA Focus Conference on Ground Water Issues,
April 21-23, 1987, Indianapolis, IN.
Zarrabi, K, C.K Fitzsimmons, S. Ward, S. Butala, T.E. Lewis, J.R. Parolini, V.G. King,
and T.W. Nail. 1991. Investigation of sample and sample handing techniques for the
measurement of volatile organic compounds in soil. (In preparation for submission to
journals)
103
-------
Zoeller, A., X. Zhang, and E. Overton. 1992. VOC analyses of contaminated sediments
using a field GC system and conventional GC/MS: How good are the results? Abstract
No. 144P. In, Pittsburgh Conference Book of Abstracts, PittCon'92. March 9-12, 1992.
New Orleans, LA.
1 r,, IU.S. GOVERNMENT PRINTING OFFICE: 1993 - 750-002/80278
------- |